diff options
Diffstat (limited to 'drivers/gpu/drm/amd')
273 files changed, 5916 insertions, 3912 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index 87080c06e5fc..930de203d533 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -66,7 +66,7 @@ amdgpu-y += amdgpu_device.o amdgpu_doorbell_mgr.o amdgpu_kms.o \ amdgpu_fw_attestation.o amdgpu_securedisplay.o \ amdgpu_eeprom.o amdgpu_mca.o amdgpu_psp_ta.o amdgpu_lsdma.o \ amdgpu_ring_mux.o amdgpu_xcp.o amdgpu_seq64.o amdgpu_aca.o amdgpu_dev_coredump.o \ - amdgpu_cper.o amdgpu_userq_fence.o amdgpu_eviction_fence.o + amdgpu_cper.o amdgpu_userq_fence.o amdgpu_eviction_fence.o amdgpu_ip.o amdgpu-$(CONFIG_PROC_FS) += amdgpu_fdinfo.o diff --git a/drivers/gpu/drm/amd/amdgpu/aldebaran.c b/drivers/gpu/drm/amd/amdgpu/aldebaran.c index e13fbd974141..9569dc16dd3d 100644 --- a/drivers/gpu/drm/amd/amdgpu/aldebaran.c +++ b/drivers/gpu/drm/amd/amdgpu/aldebaran.c @@ -71,18 +71,29 @@ aldebaran_get_reset_handler(struct amdgpu_reset_control *reset_ctl, return NULL; } +static inline uint32_t aldebaran_get_ip_block_mask(struct amdgpu_device *adev) +{ + uint32_t ip_block_mask = BIT(AMD_IP_BLOCK_TYPE_GFX) | + BIT(AMD_IP_BLOCK_TYPE_SDMA); + + if (adev->aid_mask) + ip_block_mask |= BIT(AMD_IP_BLOCK_TYPE_IH); + + return ip_block_mask; +} + static int aldebaran_mode2_suspend_ip(struct amdgpu_device *adev) { + uint32_t ip_block_mask = aldebaran_get_ip_block_mask(adev); + uint32_t ip_block; int r, i; amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE); amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE); for (i = adev->num_ip_blocks - 1; i >= 0; i--) { - if (!(adev->ip_blocks[i].version->type == - AMD_IP_BLOCK_TYPE_GFX || - adev->ip_blocks[i].version->type == - AMD_IP_BLOCK_TYPE_SDMA)) + ip_block = BIT(adev->ip_blocks[i].version->type); + if (!(ip_block_mask & ip_block)) continue; r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]); @@ -200,8 +211,10 @@ aldebaran_mode2_perform_reset(struct amdgpu_reset_control *reset_ctl, static int aldebaran_mode2_restore_ip(struct amdgpu_device *adev) { struct amdgpu_firmware_info *ucode_list[AMDGPU_UCODE_ID_MAXIMUM]; + uint32_t ip_block_mask = aldebaran_get_ip_block_mask(adev); struct amdgpu_firmware_info *ucode; struct amdgpu_ip_block *cmn_block; + struct amdgpu_ip_block *ih_block; int ucode_count = 0; int i, r; @@ -243,6 +256,18 @@ static int aldebaran_mode2_restore_ip(struct amdgpu_device *adev) if (r) return r; + if (ip_block_mask & BIT(AMD_IP_BLOCK_TYPE_IH)) { + ih_block = amdgpu_device_ip_get_ip_block(adev, + AMD_IP_BLOCK_TYPE_IH); + if (unlikely(!ih_block)) { + dev_err(adev->dev, "Failed to get IH handle\n"); + return -EINVAL; + } + r = amdgpu_ip_block_resume(ih_block); + if (r) + return r; + } + /* Reinit GFXHUB */ adev->gfxhub.funcs->init(adev); r = adev->gfxhub.funcs->gart_enable(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index a5ccd0ada16a..7edb7ba91786 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -470,9 +470,6 @@ struct amdgpu_sa_manager { void *cpu_ptr; }; -int amdgpu_fence_slab_init(void); -void amdgpu_fence_slab_fini(void); - /* * IRQS. */ @@ -1282,6 +1279,7 @@ struct amdgpu_device { bool debug_exp_resets; bool debug_disable_gpu_ring_reset; bool debug_vm_userptr; + bool debug_disable_ce_logs; /* Protection for the following isolation structure */ struct mutex enforce_isolation_mutex; @@ -1336,6 +1334,11 @@ static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_device *bdev) return container_of(bdev, struct amdgpu_device, mman.bdev); } +static inline bool amdgpu_is_multi_aid(struct amdgpu_device *adev) +{ + return !!adev->aid_mask; +} + int amdgpu_device_init(struct amdgpu_device *adev, uint32_t flags); void amdgpu_device_fini_hw(struct amdgpu_device *adev); @@ -1387,7 +1390,8 @@ void amdgpu_device_indirect_wreg64(struct amdgpu_device *adev, void amdgpu_device_indirect_wreg64_ext(struct amdgpu_device *adev, u64 reg_addr, u64 reg_data); u32 amdgpu_device_get_rev_id(struct amdgpu_device *adev); -bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type); +bool amdgpu_device_asic_has_dc_support(struct pci_dev *pdev, + enum amd_asic_type asic_type); bool amdgpu_device_has_dc_support(struct amdgpu_device *adev); void amdgpu_device_set_sriov_virtual_display(struct amdgpu_device *adev); @@ -1619,6 +1623,7 @@ void amdgpu_driver_release_kms(struct drm_device *dev); int amdgpu_device_ip_suspend(struct amdgpu_device *adev); int amdgpu_device_prepare(struct drm_device *dev); +void amdgpu_device_complete(struct drm_device *dev); int amdgpu_device_suspend(struct drm_device *dev, bool fbcon); int amdgpu_device_resume(struct drm_device *dev, bool fbcon); u32 amdgpu_get_vblank_counter_kms(struct drm_crtc *crtc); @@ -1760,4 +1765,19 @@ extern const struct attribute_group amdgpu_flash_attr_group; void amdgpu_set_init_level(struct amdgpu_device *adev, enum amdgpu_init_lvl_id lvl); + +static inline int amdgpu_device_bus_status_check(struct amdgpu_device *adev) +{ + u32 status; + int r; + + r = pci_read_config_dword(adev->pdev, PCI_COMMAND, &status); + if (r || PCI_POSSIBLE_ERROR(status)) { + dev_err(adev->dev, "device lost from bus!"); + return -ENODEV; + } + + return 0; +} + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c index 3835f2592914..cbc40cad581b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c @@ -115,6 +115,11 @@ static void aca_smu_bank_dump(struct amdgpu_device *adev, int idx, int total, st u64 event_id = qctx ? qctx->evid.event_id : RAS_EVENT_INVALID_ID; int i; + if (adev->debug_disable_ce_logs && + bank->smu_err_type == ACA_SMU_TYPE_CE && + !ACA_BANK_ERR_IS_DEFFERED(bank)) + return; + RAS_EVENT_LOG(adev, event_id, HW_ERR "Accelerator Check Architecture events logged\n"); /* plus 1 for output format, e.g: ACA[08/08]: xxxx */ for (i = 0; i < ARRAY_SIZE(aca_regs); i++) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c index 55161e5cdc30..fbe7616555c8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c @@ -248,18 +248,34 @@ void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev, kgd2kfd_interrupt(adev->kfd.dev, ih_ring_entry); } -void amdgpu_amdkfd_suspend(struct amdgpu_device *adev, bool run_pm) +void amdgpu_amdkfd_suspend(struct amdgpu_device *adev, bool suspend_proc) { if (adev->kfd.dev) - kgd2kfd_suspend(adev->kfd.dev, run_pm); + kgd2kfd_suspend(adev->kfd.dev, suspend_proc); } -int amdgpu_amdkfd_resume(struct amdgpu_device *adev, bool run_pm) +int amdgpu_amdkfd_resume(struct amdgpu_device *adev, bool resume_proc) { int r = 0; if (adev->kfd.dev) - r = kgd2kfd_resume(adev->kfd.dev, run_pm); + r = kgd2kfd_resume(adev->kfd.dev, resume_proc); + + return r; +} + +void amdgpu_amdkfd_suspend_process(struct amdgpu_device *adev) +{ + if (adev->kfd.dev) + kgd2kfd_suspend_process(adev->kfd.dev); +} + +int amdgpu_amdkfd_resume_process(struct amdgpu_device *adev) +{ + int r = 0; + + if (adev->kfd.dev) + r = kgd2kfd_resume_process(adev->kfd.dev); return r; } @@ -749,12 +765,12 @@ int amdgpu_amdkfd_send_close_event_drain_irq(struct amdgpu_device *adev, int amdgpu_amdkfd_check_and_lock_kfd(struct amdgpu_device *adev) { - return kgd2kfd_check_and_lock_kfd(); + return kgd2kfd_check_and_lock_kfd(adev->kfd.dev); } void amdgpu_amdkfd_unlock_kfd(struct amdgpu_device *adev) { - kgd2kfd_unlock_kfd(); + kgd2kfd_unlock_kfd(adev->kfd.dev); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h index b6ca41859b53..33eb4826b58b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h @@ -154,8 +154,10 @@ struct amdkfd_process_info { int amdgpu_amdkfd_init(void); void amdgpu_amdkfd_fini(void); -void amdgpu_amdkfd_suspend(struct amdgpu_device *adev, bool run_pm); -int amdgpu_amdkfd_resume(struct amdgpu_device *adev, bool run_pm); +void amdgpu_amdkfd_suspend(struct amdgpu_device *adev, bool suspend_proc); +int amdgpu_amdkfd_resume(struct amdgpu_device *adev, bool resume_proc); +void amdgpu_amdkfd_suspend_process(struct amdgpu_device *adev); +int amdgpu_amdkfd_resume_process(struct amdgpu_device *adev); void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev, const void *ih_ring_entry); void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev); @@ -411,16 +413,18 @@ struct kfd_dev *kgd2kfd_probe(struct amdgpu_device *adev, bool vf); bool kgd2kfd_device_init(struct kfd_dev *kfd, const struct kgd2kfd_shared_resources *gpu_resources); void kgd2kfd_device_exit(struct kfd_dev *kfd); -void kgd2kfd_suspend(struct kfd_dev *kfd, bool run_pm); -int kgd2kfd_resume(struct kfd_dev *kfd, bool run_pm); +void kgd2kfd_suspend(struct kfd_dev *kfd, bool suspend_proc); +int kgd2kfd_resume(struct kfd_dev *kfd, bool resume_proc); +void kgd2kfd_suspend_process(struct kfd_dev *kfd); +int kgd2kfd_resume_process(struct kfd_dev *kfd); int kgd2kfd_pre_reset(struct kfd_dev *kfd, struct amdgpu_reset_context *reset_context); int kgd2kfd_post_reset(struct kfd_dev *kfd); void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry); void kgd2kfd_set_sram_ecc_flag(struct kfd_dev *kfd); void kgd2kfd_smi_event_throttle(struct kfd_dev *kfd, uint64_t throttle_bitmask); -int kgd2kfd_check_and_lock_kfd(void); -void kgd2kfd_unlock_kfd(void); +int kgd2kfd_check_and_lock_kfd(struct kfd_dev *kfd); +void kgd2kfd_unlock_kfd(struct kfd_dev *kfd); int kgd2kfd_start_sched(struct kfd_dev *kfd, uint32_t node_id); int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id); bool kgd2kfd_compute_active(struct kfd_dev *kfd, uint32_t node_id); @@ -454,11 +458,20 @@ static inline void kgd2kfd_device_exit(struct kfd_dev *kfd) { } -static inline void kgd2kfd_suspend(struct kfd_dev *kfd, bool run_pm) +static inline void kgd2kfd_suspend(struct kfd_dev *kfd, bool suspend_proc) { } -static inline int kgd2kfd_resume(struct kfd_dev *kfd, bool run_pm) +static inline int kgd2kfd_resume(struct kfd_dev *kfd, bool resume_proc) +{ + return 0; +} + +static inline void kgd2kfd_suspend_process(struct kfd_dev *kfd) +{ +} + +static inline int kgd2kfd_resume_process(struct kfd_dev *kfd) { return 0; } @@ -489,12 +502,12 @@ void kgd2kfd_smi_event_throttle(struct kfd_dev *kfd, uint64_t throttle_bitmask) { } -static inline int kgd2kfd_check_and_lock_kfd(void) +static inline int kgd2kfd_check_and_lock_kfd(struct kfd_dev *kfd) { return 0; } -static inline void kgd2kfd_unlock_kfd(void) +static inline void kgd2kfd_unlock_kfd(struct kfd_dev *kfd) { } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c index ffbaa8bc5eea..1105a09e55dc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c @@ -320,7 +320,7 @@ static void set_barrier_auto_waitcnt(struct amdgpu_device *adev, bool enable_wai if (!down_read_trylock(&adev->reset_domain->sem)) return; - amdgpu_amdkfd_suspend(adev, false); + amdgpu_amdkfd_suspend(adev, true); if (suspend_resume_compute_scheduler(adev, true)) goto out; @@ -333,7 +333,7 @@ static void set_barrier_auto_waitcnt(struct amdgpu_device *adev, bool enable_wai out: suspend_resume_compute_scheduler(adev, false); - amdgpu_amdkfd_resume(adev, false); + amdgpu_amdkfd_resume(adev, true); up_read(&adev->reset_domain->sem); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c index ca4a6b82817f..df77558e03ef 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c @@ -561,6 +561,13 @@ static uint32_t read_vmid_from_vmfault_reg(struct amdgpu_device *adev) return REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID); } +static uint32_t kgd_hqd_sdma_get_doorbell(struct amdgpu_device *adev, + int engine, int queue) + +{ + return 0; +} + const struct kfd2kgd_calls gfx_v7_kfd2kgd = { .program_sh_mem_settings = kgd_program_sh_mem_settings, .set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping, @@ -578,4 +585,5 @@ const struct kfd2kgd_calls gfx_v7_kfd2kgd = { .set_scratch_backing_va = set_scratch_backing_va, .set_vm_context_page_table_base = set_vm_context_page_table_base, .read_vmid_from_vmfault_reg = read_vmid_from_vmfault_reg, + .hqd_sdma_get_doorbell = kgd_hqd_sdma_get_doorbell, }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c index 0f3e2944edd7..e68c0fa8d751 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c @@ -582,6 +582,13 @@ static void set_vm_context_page_table_base(struct amdgpu_device *adev, lower_32_bits(page_table_base)); } +static uint32_t kgd_hqd_sdma_get_doorbell(struct amdgpu_device *adev, + int engine, int queue) + +{ + return 0; +} + const struct kfd2kgd_calls gfx_v8_kfd2kgd = { .program_sh_mem_settings = kgd_program_sh_mem_settings, .set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping, @@ -599,4 +606,5 @@ const struct kfd2kgd_calls gfx_v8_kfd2kgd = { get_atc_vmid_pasid_mapping_info, .set_scratch_backing_va = set_scratch_backing_va, .set_vm_context_page_table_base = set_vm_context_page_table_base, + .hqd_sdma_get_doorbell = kgd_hqd_sdma_get_doorbell, }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c index 5a234eadae8b..15dde1f50328 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c @@ -212,7 +212,7 @@ int amdgpu_cper_entry_fill_bad_page_threshold_section(struct amdgpu_device *adev NONSTD_SEC_OFFSET(hdr->sec_cnt, idx)); amdgpu_cper_entry_fill_section_desc(adev, section_desc, true, false, - CPER_SEV_NUM, RUNTIME, NONSTD_SEC_LEN, + CPER_SEV_FATAL, RUNTIME, NONSTD_SEC_LEN, NONSTD_SEC_OFFSET(hdr->sec_cnt, idx)); section->hdr.valid_bits.err_info_cnt = 1; @@ -326,7 +326,9 @@ int amdgpu_cper_generate_bp_threshold_record(struct amdgpu_device *adev) return -ENOMEM; } - amdgpu_cper_entry_fill_hdr(adev, bp_threshold, AMDGPU_CPER_TYPE_BP_THRESHOLD, CPER_SEV_NUM); + amdgpu_cper_entry_fill_hdr(adev, bp_threshold, + AMDGPU_CPER_TYPE_BP_THRESHOLD, + CPER_SEV_FATAL); ret = amdgpu_cper_entry_fill_bad_page_threshold_section(adev, bp_threshold, 0); if (ret) return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c index 85567d0d9545..f5d5c45ddc0d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c @@ -944,6 +944,7 @@ static void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr) drm_sched_entity_fini(entity); } } + kref_put(&ctx->refcount, amdgpu_ctx_fini); } } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c index e49890a23ef6..2353455d4aff 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c @@ -1902,7 +1902,7 @@ no_preempt: continue; } job = to_amdgpu_job(s_job); - if (preempted && (&job->hw_fence) == fence) + if (preempted && (&job->hw_fence.base) == fence) /* mark the job as preempted */ job->preemption_status |= AMDGPU_IB_PREEMPTED; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index a59f194e3360..d282c0753b14 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -232,7 +232,7 @@ static int amdgpu_device_attr_sysfs_init(struct amdgpu_device *adev) { int ret = 0; - if (!amdgpu_sriov_vf(adev)) + if (amdgpu_nbio_is_replay_cnt_supported(adev)) ret = sysfs_create_file(&adev->dev->kobj, &dev_attr_pcie_replay_count.attr); @@ -241,7 +241,7 @@ static int amdgpu_device_attr_sysfs_init(struct amdgpu_device *adev) static void amdgpu_device_attr_sysfs_fini(struct amdgpu_device *adev) { - if (!amdgpu_sriov_vf(adev)) + if (amdgpu_nbio_is_replay_cnt_supported(adev)) sysfs_remove_file(&adev->dev->kobj, &dev_attr_pcie_replay_count.attr); } @@ -1288,14 +1288,14 @@ u32 amdgpu_device_get_rev_id(struct amdgpu_device *adev) */ static uint32_t amdgpu_invalid_rreg(struct amdgpu_device *adev, uint32_t reg) { - DRM_ERROR("Invalid callback to read register 0x%04X\n", reg); + dev_err(adev->dev, "Invalid callback to read register 0x%04X\n", reg); BUG(); return 0; } static uint32_t amdgpu_invalid_rreg_ext(struct amdgpu_device *adev, uint64_t reg) { - DRM_ERROR("Invalid callback to read register 0x%llX\n", reg); + dev_err(adev->dev, "Invalid callback to read register 0x%llX\n", reg); BUG(); return 0; } @@ -1312,15 +1312,17 @@ static uint32_t amdgpu_invalid_rreg_ext(struct amdgpu_device *adev, uint64_t reg */ static void amdgpu_invalid_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v) { - DRM_ERROR("Invalid callback to write register 0x%04X with 0x%08X\n", - reg, v); + dev_err(adev->dev, + "Invalid callback to write register 0x%04X with 0x%08X\n", reg, + v); BUG(); } static void amdgpu_invalid_wreg_ext(struct amdgpu_device *adev, uint64_t reg, uint32_t v) { - DRM_ERROR("Invalid callback to write register 0x%llX with 0x%08X\n", - reg, v); + dev_err(adev->dev, + "Invalid callback to write register 0x%llX with 0x%08X\n", reg, + v); BUG(); } @@ -1336,14 +1338,15 @@ static void amdgpu_invalid_wreg_ext(struct amdgpu_device *adev, uint64_t reg, ui */ static uint64_t amdgpu_invalid_rreg64(struct amdgpu_device *adev, uint32_t reg) { - DRM_ERROR("Invalid callback to read 64 bit register 0x%04X\n", reg); + dev_err(adev->dev, "Invalid callback to read 64 bit register 0x%04X\n", + reg); BUG(); return 0; } static uint64_t amdgpu_invalid_rreg64_ext(struct amdgpu_device *adev, uint64_t reg) { - DRM_ERROR("Invalid callback to read register 0x%llX\n", reg); + dev_err(adev->dev, "Invalid callback to read register 0x%llX\n", reg); BUG(); return 0; } @@ -1360,15 +1363,17 @@ static uint64_t amdgpu_invalid_rreg64_ext(struct amdgpu_device *adev, uint64_t r */ static void amdgpu_invalid_wreg64(struct amdgpu_device *adev, uint32_t reg, uint64_t v) { - DRM_ERROR("Invalid callback to write 64 bit register 0x%04X with 0x%08llX\n", - reg, v); + dev_err(adev->dev, + "Invalid callback to write 64 bit register 0x%04X with 0x%08llX\n", + reg, v); BUG(); } static void amdgpu_invalid_wreg64_ext(struct amdgpu_device *adev, uint64_t reg, uint64_t v) { - DRM_ERROR("Invalid callback to write 64 bit register 0x%llX with 0x%08llX\n", - reg, v); + dev_err(adev->dev, + "Invalid callback to write 64 bit register 0x%llX with 0x%08llX\n", + reg, v); BUG(); } @@ -1386,8 +1391,9 @@ static void amdgpu_invalid_wreg64_ext(struct amdgpu_device *adev, uint64_t reg, static uint32_t amdgpu_block_invalid_rreg(struct amdgpu_device *adev, uint32_t block, uint32_t reg) { - DRM_ERROR("Invalid callback to read register 0x%04X in block 0x%04X\n", - reg, block); + dev_err(adev->dev, + "Invalid callback to read register 0x%04X in block 0x%04X\n", + reg, block); BUG(); return 0; } @@ -1407,8 +1413,9 @@ static void amdgpu_block_invalid_wreg(struct amdgpu_device *adev, uint32_t block, uint32_t reg, uint32_t v) { - DRM_ERROR("Invalid block callback to write register 0x%04X in block 0x%04X with 0x%08X\n", - reg, block, v); + dev_err(adev->dev, + "Invalid block callback to write register 0x%04X in block 0x%04X with 0x%08X\n", + reg, block, v); BUG(); } @@ -1694,7 +1701,9 @@ int amdgpu_device_resize_fb_bar(struct amdgpu_device *adev) /* PCI_EXT_CAP_ID_VNDR extended capability is located at 0x100 */ if (!pci_find_ext_capability(adev->pdev, PCI_EXT_CAP_ID_VNDR)) - DRM_WARN("System can't access extended configuration space, please check!!\n"); + dev_warn( + adev->dev, + "System can't access extended configuration space, please check!!\n"); /* skip if the bios has already enabled large BAR */ if (adev->gmc.real_vram_size && @@ -1734,9 +1743,10 @@ int amdgpu_device_resize_fb_bar(struct amdgpu_device *adev) r = pci_resize_resource(adev->pdev, 0, rbar_size); if (r == -ENOSPC) - DRM_INFO("Not enough PCI address space for a large BAR."); + dev_info(adev->dev, + "Not enough PCI address space for a large BAR."); else if (r && r != -ENOTSUPP) - DRM_ERROR("Problem resizing BAR0 (%d).", r); + dev_err(adev->dev, "Problem resizing BAR0 (%d).", r); pci_assign_unassigned_bus_resources(adev->pdev->bus); @@ -1838,8 +1848,8 @@ bool amdgpu_device_seamless_boot_supported(struct amdgpu_device *adev) case 0: return false; default: - DRM_ERROR("Invalid value for amdgpu.seamless: %d\n", - amdgpu_seamless); + dev_err(adev->dev, "Invalid value for amdgpu.seamless: %d\n", + amdgpu_seamless); return false; } @@ -2015,7 +2025,7 @@ static void amdgpu_device_check_smu_prv_buffer_size(struct amdgpu_device *adev) return; if (!is_os_64) { - DRM_WARN("Not 64-bit OS, feature not supported\n"); + dev_warn(adev->dev, "Not 64-bit OS, feature not supported\n"); goto def_value; } si_meminfo(&si); @@ -2030,7 +2040,7 @@ static void amdgpu_device_check_smu_prv_buffer_size(struct amdgpu_device *adev) if (total_memory < dram_size_seven_GB) goto def_value1; } else { - DRM_WARN("Smu memory pool size not supported\n"); + dev_warn(adev->dev, "Smu memory pool size not supported\n"); goto def_value; } adev->pm.smu_prv_buffer_size = amdgpu_smu_memory_pool_size << 28; @@ -2038,7 +2048,7 @@ static void amdgpu_device_check_smu_prv_buffer_size(struct amdgpu_device *adev) return; def_value1: - DRM_WARN("No enough system memory\n"); + dev_warn(adev->dev, "No enough system memory\n"); def_value: adev->pm.smu_prv_buffer_size = 0; } @@ -2202,12 +2212,13 @@ static void amdgpu_switcheroo_set_state(struct pci_dev *pdev, amdgpu_device_load_pci_state(pdev); r = pci_enable_device(pdev); if (r) - DRM_WARN("pci_enable_device failed (%d)\n", r); + dev_warn(&pdev->dev, "pci_enable_device failed (%d)\n", + r); amdgpu_device_resume(dev, true); dev->switch_power_state = DRM_SWITCH_POWER_ON; } else { - pr_info("switched off\n"); + dev_info(&pdev->dev, "switched off\n"); dev->switch_power_state = DRM_SWITCH_POWER_CHANGING; amdgpu_device_prepare(dev); amdgpu_device_suspend(dev, true); @@ -2274,8 +2285,9 @@ int amdgpu_device_ip_set_clockgating_state(void *dev, r = adev->ip_blocks[i].version->funcs->set_clockgating_state( &adev->ip_blocks[i], state); if (r) - DRM_ERROR("set_clockgating_state of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + dev_err(adev->dev, + "set_clockgating_state of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, r); } return r; } @@ -2308,8 +2320,9 @@ int amdgpu_device_ip_set_powergating_state(void *dev, r = adev->ip_blocks[i].version->funcs->set_powergating_state( &adev->ip_blocks[i], state); if (r) - DRM_ERROR("set_powergating_state of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + dev_err(adev->dev, + "set_powergating_state of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, r); } return r; } @@ -2525,9 +2538,11 @@ static void amdgpu_device_enable_virtual_display(struct amdgpu_device *adev) } } - DRM_INFO("virtual display string:%s, %s:virtual_display:%d, num_crtc:%d\n", - amdgpu_virtual_display, pci_address_name, - adev->enable_virtual_display, adev->mode_info.num_crtc); + dev_info( + adev->dev, + "virtual display string:%s, %s:virtual_display:%d, num_crtc:%d\n", + amdgpu_virtual_display, pci_address_name, + adev->enable_virtual_display, adev->mode_info.num_crtc); kfree(pciaddstr); } @@ -2538,8 +2553,9 @@ void amdgpu_device_set_sriov_virtual_display(struct amdgpu_device *adev) if (amdgpu_sriov_vf(adev) && !adev->enable_virtual_display) { adev->mode_info.num_crtc = 1; adev->enable_virtual_display = true; - DRM_INFO("virtual_display:%d, num_crtc:%d\n", - adev->enable_virtual_display, adev->mode_info.num_crtc); + dev_info(adev->dev, "virtual_display:%d, num_crtc:%d\n", + adev->enable_virtual_display, + adev->mode_info.num_crtc); } } @@ -2773,21 +2789,29 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev) if (!amdgpu_device_pcie_dynamic_switching_supported(adev)) adev->pm.pp_feature &= ~PP_PCIE_DPM_MASK; + adev->virt.is_xgmi_node_migrate_enabled = false; + if (amdgpu_sriov_vf(adev)) { + adev->virt.is_xgmi_node_migrate_enabled = + amdgpu_ip_version((adev), GC_HWIP, 0) == IP_VERSION(9, 4, 4); + } + total = true; for (i = 0; i < adev->num_ip_blocks; i++) { ip_block = &adev->ip_blocks[i]; if ((amdgpu_ip_block_mask & (1 << i)) == 0) { - DRM_WARN("disabled ip block: %d <%s>\n", - i, adev->ip_blocks[i].version->funcs->name); + dev_warn(adev->dev, "disabled ip block: %d <%s>\n", i, + adev->ip_blocks[i].version->funcs->name); adev->ip_blocks[i].status.valid = false; } else if (ip_block->version->funcs->early_init) { r = ip_block->version->funcs->early_init(ip_block); if (r == -ENOENT) { adev->ip_blocks[i].status.valid = false; } else if (r) { - DRM_ERROR("early_init of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + dev_err(adev->dev, + "early_init of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, + r); total = false; } else { adev->ip_blocks[i].status.valid = true; @@ -2868,8 +2892,10 @@ static int amdgpu_device_ip_hw_init_phase1(struct amdgpu_device *adev) adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH) { r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]); if (r) { - DRM_ERROR("hw_init of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + dev_err(adev->dev, + "hw_init of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, + r); return r; } adev->ip_blocks[i].status.hw = true; @@ -2893,8 +2919,9 @@ static int amdgpu_device_ip_hw_init_phase2(struct amdgpu_device *adev) continue; r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]); if (r) { - DRM_ERROR("hw_init of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + dev_err(adev->dev, + "hw_init of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, r); return r; } adev->ip_blocks[i].status.hw = true; @@ -2932,8 +2959,11 @@ static int amdgpu_device_fw_loading(struct amdgpu_device *adev) } else { r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]); if (r) { - DRM_ERROR("hw_init of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + dev_err(adev->dev, + "hw_init of IP block <%s> failed %d\n", + adev->ip_blocks[i] + .version->funcs->name, + r); return r; } adev->ip_blocks[i].status.hw = true; @@ -2988,25 +3018,29 @@ static int amdgpu_device_init_schedulers(struct amdgpu_device *adev) r = drm_sched_init(&ring->sched, &args); if (r) { - DRM_ERROR("Failed to create scheduler on ring %s.\n", - ring->name); + dev_err(adev->dev, + "Failed to create scheduler on ring %s.\n", + ring->name); return r; } r = amdgpu_uvd_entity_init(adev, ring); if (r) { - DRM_ERROR("Failed to create UVD scheduling entity on ring %s.\n", - ring->name); + dev_err(adev->dev, + "Failed to create UVD scheduling entity on ring %s.\n", + ring->name); return r; } r = amdgpu_vce_entity_init(adev, ring); if (r) { - DRM_ERROR("Failed to create VCE scheduling entity on ring %s.\n", - ring->name); + dev_err(adev->dev, + "Failed to create VCE scheduling entity on ring %s.\n", + ring->name); return r; } } - amdgpu_xcp_update_partition_sched_list(adev); + if (adev->xcp_mgr) + amdgpu_xcp_update_partition_sched_list(adev); return 0; } @@ -3038,8 +3072,10 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev) if (adev->ip_blocks[i].version->funcs->sw_init) { r = adev->ip_blocks[i].version->funcs->sw_init(&adev->ip_blocks[i]); if (r) { - DRM_ERROR("sw_init of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + dev_err(adev->dev, + "sw_init of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, + r); goto init_failed; } } @@ -3053,7 +3089,8 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev) /* need to do common hw init early so everything is set up for gmc */ r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]); if (r) { - DRM_ERROR("hw_init %d failed %d\n", i, r); + dev_err(adev->dev, "hw_init %d failed %d\n", i, + r); goto init_failed; } adev->ip_blocks[i].status.hw = true; @@ -3065,17 +3102,21 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev) r = amdgpu_device_mem_scratch_init(adev); if (r) { - DRM_ERROR("amdgpu_mem_scratch_init failed %d\n", r); + dev_err(adev->dev, + "amdgpu_mem_scratch_init failed %d\n", + r); goto init_failed; } r = adev->ip_blocks[i].version->funcs->hw_init(&adev->ip_blocks[i]); if (r) { - DRM_ERROR("hw_init %d failed %d\n", i, r); + dev_err(adev->dev, "hw_init %d failed %d\n", i, + r); goto init_failed; } r = amdgpu_device_wb_init(adev); if (r) { - DRM_ERROR("amdgpu_device_wb_init failed %d\n", r); + dev_err(adev->dev, + "amdgpu_device_wb_init failed %d\n", r); goto init_failed; } adev->ip_blocks[i].status.hw = true; @@ -3087,14 +3128,16 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev) AMDGPU_GEM_DOMAIN_GTT, AMDGPU_CSA_SIZE); if (r) { - DRM_ERROR("allocate CSA failed %d\n", r); + dev_err(adev->dev, + "allocate CSA failed %d\n", r); goto init_failed; } } r = amdgpu_seq64_init(adev); if (r) { - DRM_ERROR("allocate seq64 failed %d\n", r); + dev_err(adev->dev, "allocate seq64 failed %d\n", + r); goto init_failed; } } @@ -3284,8 +3327,10 @@ int amdgpu_device_set_cg_state(struct amdgpu_device *adev, r = adev->ip_blocks[i].version->funcs->set_clockgating_state(&adev->ip_blocks[i], state); if (r) { - DRM_ERROR("set_clockgating_state(gate) of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + dev_err(adev->dev, + "set_clockgating_state(gate) of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, + r); return r; } } @@ -3321,8 +3366,10 @@ int amdgpu_device_set_pg_state(struct amdgpu_device *adev, r = adev->ip_blocks[i].version->funcs->set_powergating_state(&adev->ip_blocks[i], state); if (r) { - DRM_ERROR("set_powergating_state(gate) of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + dev_err(adev->dev, + "set_powergating_state(gate) of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, + r); return r; } } @@ -3388,8 +3435,10 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev) if (adev->ip_blocks[i].version->funcs->late_init) { r = adev->ip_blocks[i].version->funcs->late_init(&adev->ip_blocks[i]); if (r) { - DRM_ERROR("late_init of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + dev_err(adev->dev, + "late_init of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, + r); return r; } } @@ -3398,7 +3447,7 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev) r = amdgpu_ras_late_init(adev); if (r) { - DRM_ERROR("amdgpu_ras_late_init failed %d", r); + dev_err(adev->dev, "amdgpu_ras_late_init failed %d", r); return r; } @@ -3412,7 +3461,7 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev) r = amdgpu_device_enable_mgpu_fan_boost(); if (r) - DRM_ERROR("enable mgpu fan boost failed (%d).\n", r); + dev_err(adev->dev, "enable mgpu fan boost failed (%d).\n", r); /* For passthrough configuration on arcturus and aldebaran, enable special handling SBR */ if (amdgpu_passthrough(adev) && @@ -3445,7 +3494,9 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev) r = amdgpu_xgmi_set_pstate(gpu_instance->adev, AMDGPU_XGMI_PSTATE_MIN); if (r) { - DRM_ERROR("pstate setting failed (%d).\n", r); + dev_err(adev->dev, + "pstate setting failed (%d).\n", + r); break; } } @@ -3459,17 +3510,19 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev) static void amdgpu_ip_block_hw_fini(struct amdgpu_ip_block *ip_block) { + struct amdgpu_device *adev = ip_block->adev; int r; if (!ip_block->version->funcs->hw_fini) { - DRM_ERROR("hw_fini of IP block <%s> not defined\n", - ip_block->version->funcs->name); + dev_err(adev->dev, "hw_fini of IP block <%s> not defined\n", + ip_block->version->funcs->name); } else { r = ip_block->version->funcs->hw_fini(ip_block); /* XXX handle errors */ if (r) { - DRM_DEBUG("hw_fini of IP block <%s> failed %d\n", - ip_block->version->funcs->name, r); + dev_dbg(adev->dev, + "hw_fini of IP block <%s> failed %d\n", + ip_block->version->funcs->name, r); } } @@ -3510,15 +3563,16 @@ static int amdgpu_device_ip_fini_early(struct amdgpu_device *adev) r = adev->ip_blocks[i].version->funcs->early_fini(&adev->ip_blocks[i]); if (r) { - DRM_DEBUG("early_fini of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + dev_dbg(adev->dev, + "early_fini of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, r); } } amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE); amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE); - amdgpu_amdkfd_suspend(adev, false); + amdgpu_amdkfd_suspend(adev, true); amdgpu_userq_suspend(adev); /* Workaround for ASICs need to disable SMC first */ @@ -3533,7 +3587,8 @@ static int amdgpu_device_ip_fini_early(struct amdgpu_device *adev) if (amdgpu_sriov_vf(adev)) { if (amdgpu_virt_release_full_gpu(adev, false)) - DRM_ERROR("failed to release exclusive mode on fini\n"); + dev_err(adev->dev, + "failed to release exclusive mode on fini\n"); } return 0; @@ -3581,8 +3636,10 @@ static int amdgpu_device_ip_fini(struct amdgpu_device *adev) r = adev->ip_blocks[i].version->funcs->sw_fini(&adev->ip_blocks[i]); /* XXX handle errors */ if (r) { - DRM_DEBUG("sw_fini of IP block <%s> failed %d\n", - adev->ip_blocks[i].version->funcs->name, r); + dev_dbg(adev->dev, + "sw_fini of IP block <%s> failed %d\n", + adev->ip_blocks[i].version->funcs->name, + r); } } adev->ip_blocks[i].status.sw = false; @@ -3615,7 +3672,7 @@ static void amdgpu_device_delayed_init_work_handler(struct work_struct *work) r = amdgpu_ib_ring_tests(adev); if (r) - DRM_ERROR("ib ring test failed (%d).\n", r); + dev_err(adev->dev, "ib ring test failed (%d).\n", r); } static void amdgpu_device_delay_enable_gfx_off(struct work_struct *work) @@ -3756,8 +3813,9 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC) { r = amdgpu_dpm_set_mp1_state(adev, adev->mp1_state); if (r) { - DRM_ERROR("SMC failed to set mp1 state %d, %d\n", - adev->mp1_state, r); + dev_err(adev->dev, + "SMC failed to set mp1 state %d, %d\n", + adev->mp1_state, r); return r; } } @@ -4041,12 +4099,14 @@ static void amdgpu_device_detect_sriov_bios(struct amdgpu_device *adev) /** * amdgpu_device_asic_has_dc_support - determine if DC supports the asic * + * @pdev : pci device context * @asic_type: AMD asic type * * Check if there is DC (new modesetting infrastructre) support for an asic. * returns true if DC has support, false if not. */ -bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type) +bool amdgpu_device_asic_has_dc_support(struct pci_dev *pdev, + enum amd_asic_type asic_type) { switch (asic_type) { #ifdef CONFIG_DRM_AMDGPU_SI @@ -4089,7 +4149,9 @@ bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type) #else default: if (amdgpu_dc > 0) - DRM_INFO_ONCE("Display Core has been requested via kernel parameter but isn't supported by ASIC, ignoring\n"); + dev_info_once( + &pdev->dev, + "Display Core has been requested via kernel parameter but isn't supported by ASIC, ignoring\n"); return false; #endif } @@ -4108,7 +4170,7 @@ bool amdgpu_device_has_dc_support(struct amdgpu_device *adev) (adev->harvest_ip_mask & AMD_HARVEST_IP_DMU_MASK)) return false; - return amdgpu_device_asic_has_dc_support(adev->asic_type); + return amdgpu_device_asic_has_dc_support(adev->pdev, adev->asic_type); } static void amdgpu_device_xgmi_reset_func(struct work_struct *__work) @@ -4150,7 +4212,8 @@ static void amdgpu_device_xgmi_reset_func(struct work_struct *__work) fail: if (adev->asic_reset_res) - DRM_WARN("ASIC reset failed with error, %d for drm dev, %s", + dev_warn(adev->dev, + "ASIC reset failed with error, %d for drm dev, %s", adev->asic_reset_res, adev_to_drm(adev)->unique); amdgpu_put_xgmi_hive(hive); } @@ -4274,7 +4337,7 @@ static void amdgpu_device_set_mcbp(struct amdgpu_device *adev) adev->gfx.mcbp = true; if (adev->gfx.mcbp) - DRM_INFO("MCBP is enabled\n"); + dev_info(adev->dev, "MCBP is enabled\n"); } /** @@ -4342,9 +4405,11 @@ int amdgpu_device_init(struct amdgpu_device *adev, adev->audio_endpt_rreg = &amdgpu_block_invalid_rreg; adev->audio_endpt_wreg = &amdgpu_block_invalid_wreg; - DRM_INFO("initializing kernel modesetting (%s 0x%04X:0x%04X 0x%04X:0x%04X 0x%02X).\n", - amdgpu_asic_name[adev->asic_type], pdev->vendor, pdev->device, - pdev->subsystem_vendor, pdev->subsystem_device, pdev->revision); + dev_info( + adev->dev, + "initializing kernel modesetting (%s 0x%04X:0x%04X 0x%04X:0x%04X 0x%02X).\n", + amdgpu_asic_name[adev->asic_type], pdev->vendor, pdev->device, + pdev->subsystem_vendor, pdev->subsystem_device, pdev->revision); /* mutex initialization are all done here so we * can recall function without having locking issues @@ -4461,8 +4526,10 @@ int amdgpu_device_init(struct amdgpu_device *adev, if (!adev->rmmio) return -ENOMEM; - DRM_INFO("register mmio base: 0x%08X\n", (uint32_t)adev->rmmio_base); - DRM_INFO("register mmio size: %u\n", (unsigned int)adev->rmmio_size); + dev_info(adev->dev, "register mmio base: 0x%08X\n", + (uint32_t)adev->rmmio_base); + dev_info(adev->dev, "register mmio size: %u\n", + (unsigned int)adev->rmmio_size); /* * Reset domain needs to be present early, before XGMI hive discovered @@ -4599,7 +4666,7 @@ int amdgpu_device_init(struct amdgpu_device *adev, r = -EINVAL; goto failed; } - DRM_INFO("GPU posting now...\n"); + dev_info(adev->dev, "GPU posting now...\n"); r = amdgpu_device_asic_init(adev); if (r) { dev_err(adev->dev, "gpu post error!\n"); @@ -4709,12 +4776,12 @@ fence_driver_init: r = amdgpu_pm_sysfs_init(adev); if (r) - DRM_ERROR("registering pm sysfs failed (%d).\n", r); + dev_err(adev->dev, "registering pm sysfs failed (%d).\n", r); r = amdgpu_ucode_sysfs_init(adev); if (r) { adev->ucode_sysfs_en = false; - DRM_ERROR("Creating firmware sysfs failed (%d).\n", r); + dev_err(adev->dev, "Creating firmware sysfs failed (%d).\n", r); } else adev->ucode_sysfs_en = true; @@ -4963,7 +5030,7 @@ static int amdgpu_device_evict_resources(struct amdgpu_device *adev) ret = amdgpu_ttm_evict_resources(adev, TTM_PL_VRAM); if (ret) - DRM_WARN("evicting device resources failed\n"); + dev_warn(adev->dev, "evicting device resources failed\n"); return ret; } @@ -5035,6 +5102,28 @@ int amdgpu_device_prepare(struct drm_device *dev) } /** + * amdgpu_device_complete - complete power state transition + * + * @dev: drm dev pointer + * + * Undo the changes from amdgpu_device_prepare. This will be + * called on all resume transitions, including those that failed. + */ +void amdgpu_device_complete(struct drm_device *dev) +{ + struct amdgpu_device *adev = drm_to_adev(dev); + int i; + + for (i = 0; i < adev->num_ip_blocks; i++) { + if (!adev->ip_blocks[i].status.valid) + continue; + if (!adev->ip_blocks[i].version->funcs->complete) + continue; + adev->ip_blocks[i].version->funcs->complete(&adev->ip_blocks[i]); + } +} + +/** * amdgpu_device_suspend - initiate device suspend * * @dev: drm dev pointer @@ -5055,6 +5144,8 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients) adev->in_suspend = true; if (amdgpu_sriov_vf(adev)) { + if (!adev->in_s0ix && !adev->in_runpm) + amdgpu_amdkfd_suspend_process(adev); amdgpu_virt_fini_data_exchange(adev); r = amdgpu_virt_request_full_gpu(adev, false); if (r) @@ -5062,7 +5153,7 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients) } if (amdgpu_acpi_smart_shift_update(dev, AMDGPU_SS_DEV_D3)) - DRM_WARN("smart shift update failed\n"); + dev_warn(adev->dev, "smart shift update failed\n"); if (notify_clients) drm_client_dev_suspend(adev_to_drm(adev), false); @@ -5074,7 +5165,7 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients) amdgpu_device_ip_suspend_phase1(adev); if (!adev->in_s0ix) { - amdgpu_amdkfd_suspend(adev, adev->in_runpm); + amdgpu_amdkfd_suspend(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm); amdgpu_userq_suspend(adev); } @@ -5098,6 +5189,32 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients) return 0; } +static inline int amdgpu_virt_resume(struct amdgpu_device *adev) +{ + int r; + unsigned int prev_physical_node_id = adev->gmc.xgmi.physical_node_id; + + /* During VM resume, QEMU programming of VF MSIX table (register GFXMSIX_VECT0_ADDR_LO) + * may not work. The access could be blocked by nBIF protection as VF isn't in + * exclusive access mode. Exclusive access is enabled now, disable/enable MSIX + * so that QEMU reprograms MSIX table. + */ + amdgpu_restore_msix(adev); + + r = adev->gfxhub.funcs->get_xgmi_info(adev); + if (r) + return r; + + dev_info(adev->dev, "xgmi node, old id %d, new id %d\n", + prev_physical_node_id, adev->gmc.xgmi.physical_node_id); + + adev->vm_manager.vram_base_offset = adev->gfxhub.funcs->get_mc_fb_offset(adev); + adev->vm_manager.vram_base_offset += + adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size; + + return 0; +} + /** * amdgpu_device_resume - initiate device resume * @@ -5119,6 +5236,12 @@ int amdgpu_device_resume(struct drm_device *dev, bool notify_clients) return r; } + if (amdgpu_virt_xgmi_migrate_enabled(adev)) { + r = amdgpu_virt_resume(adev); + if (r) + goto exit; + } + if (dev->switch_power_state == DRM_SWITCH_POWER_OFF) return 0; @@ -5140,7 +5263,7 @@ int amdgpu_device_resume(struct drm_device *dev, bool notify_clients) } if (!adev->in_s0ix) { - r = amdgpu_amdkfd_resume(adev, adev->in_runpm); + r = amdgpu_amdkfd_resume(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm); if (r) goto exit; @@ -5159,6 +5282,9 @@ exit: if (amdgpu_sriov_vf(adev)) { amdgpu_virt_init_data_exchange(adev); amdgpu_virt_release_full_gpu(adev, true); + + if (!adev->in_s0ix && !r && !adev->in_runpm) + r = amdgpu_amdkfd_resume_process(adev); } if (r) @@ -5196,7 +5322,7 @@ exit: adev->in_suspend = false; if (amdgpu_acpi_smart_shift_update(dev, AMDGPU_SS_DEV_D0)) - DRM_WARN("smart shift update failed\n"); + dev_warn(adev->dev, "smart shift update failed\n"); return 0; } @@ -5727,7 +5853,9 @@ int amdgpu_device_reinit_after_reset(struct amdgpu_reset_context *reset_context) amdgpu_coredump(tmp_adev, false, vram_lost, reset_context->job); if (vram_lost) { - DRM_INFO("VRAM is lost due to GPU reset!\n"); + dev_info( + tmp_adev->dev, + "VRAM is lost due to GPU reset!\n"); amdgpu_inc_vram_lost(tmp_adev); } @@ -6006,29 +6134,20 @@ static int amdgpu_device_health_check(struct list_head *device_list_handle) { struct amdgpu_device *tmp_adev; int ret = 0; - u32 status; list_for_each_entry(tmp_adev, device_list_handle, reset_list) { - pci_read_config_dword(tmp_adev->pdev, PCI_COMMAND, &status); - if (PCI_POSSIBLE_ERROR(status)) { - dev_err(tmp_adev->dev, "device lost from bus!"); - ret = -ENODEV; - } + ret |= amdgpu_device_bus_status_check(tmp_adev); } return ret; } -static int amdgpu_device_halt_activities(struct amdgpu_device *adev, - struct amdgpu_job *job, - struct amdgpu_reset_context *reset_context, - struct list_head *device_list, - struct amdgpu_hive_info *hive, - bool need_emergency_restart) +static int amdgpu_device_recovery_prepare(struct amdgpu_device *adev, + struct list_head *device_list, + struct amdgpu_hive_info *hive) { - struct list_head *device_list_handle = NULL; struct amdgpu_device *tmp_adev = NULL; - int i, r = 0; + int r; /* * Build list of devices to reset. @@ -6045,26 +6164,55 @@ static int amdgpu_device_halt_activities(struct amdgpu_device *adev, } if (!list_is_first(&adev->reset_list, device_list)) list_rotate_to_front(&adev->reset_list, device_list); - device_list_handle = device_list; } else { list_add_tail(&adev->reset_list, device_list); - device_list_handle = device_list; } if (!amdgpu_sriov_vf(adev) && (!adev->pcie_reset_ctx.occurs_dpc)) { - r = amdgpu_device_health_check(device_list_handle); + r = amdgpu_device_health_check(device_list); if (r) return r; } - /* We need to lock reset domain only once both for XGMI and single device */ - tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device, - reset_list); + return 0; +} + +static void amdgpu_device_recovery_get_reset_lock(struct amdgpu_device *adev, + struct list_head *device_list) +{ + struct amdgpu_device *tmp_adev = NULL; + + if (list_empty(device_list)) + return; + tmp_adev = + list_first_entry(device_list, struct amdgpu_device, reset_list); amdgpu_device_lock_reset_domain(tmp_adev->reset_domain); +} - /* block all schedulers and reset given job's ring */ - list_for_each_entry(tmp_adev, device_list_handle, reset_list) { +static void amdgpu_device_recovery_put_reset_lock(struct amdgpu_device *adev, + struct list_head *device_list) +{ + struct amdgpu_device *tmp_adev = NULL; + + if (list_empty(device_list)) + return; + tmp_adev = + list_first_entry(device_list, struct amdgpu_device, reset_list); + amdgpu_device_unlock_reset_domain(tmp_adev->reset_domain); +} +static void amdgpu_device_halt_activities(struct amdgpu_device *adev, + struct amdgpu_job *job, + struct amdgpu_reset_context *reset_context, + struct list_head *device_list, + struct amdgpu_hive_info *hive, + bool need_emergency_restart) +{ + struct amdgpu_device *tmp_adev = NULL; + int i; + + /* block all schedulers and reset given job's ring */ + list_for_each_entry(tmp_adev, device_list, reset_list) { amdgpu_device_set_mp1_state(tmp_adev); /* @@ -6113,8 +6261,6 @@ static int amdgpu_device_halt_activities(struct amdgpu_device *adev, } atomic_inc(&tmp_adev->gpu_reset_counter); } - - return r; } static int amdgpu_device_asic_reset(struct amdgpu_device *adev, @@ -6220,7 +6366,8 @@ static int amdgpu_device_sched_resume(struct list_head *device_list, } else { dev_info(tmp_adev->dev, "GPU reset(%d) succeeded!\n", atomic_read(&tmp_adev->gpu_reset_counter)); if (amdgpu_acpi_smart_shift_update(adev_to_drm(tmp_adev), AMDGPU_SS_DEV_D0)) - DRM_WARN("smart shift update failed\n"); + dev_warn(tmp_adev->dev, + "smart shift update failed\n"); } } @@ -6252,11 +6399,6 @@ static void amdgpu_device_gpu_resume(struct amdgpu_device *adev, amdgpu_ras_set_error_query_ready(tmp_adev, true); } - - tmp_adev = list_first_entry(device_list, struct amdgpu_device, - reset_list); - amdgpu_device_unlock_reset_domain(tmp_adev->reset_domain); - } @@ -6306,7 +6448,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, */ if (need_emergency_restart && amdgpu_ras_get_context(adev) && amdgpu_ras_get_context(adev)->reboot) { - DRM_WARN("Emergency reboot."); + dev_warn(adev->dev, "Emergency reboot."); ksys_sync_helper(); emergency_restart(); @@ -6324,11 +6466,14 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, reset_context->hive = hive; INIT_LIST_HEAD(&device_list); - r = amdgpu_device_halt_activities(adev, job, reset_context, &device_list, - hive, need_emergency_restart); - if (r) + if (amdgpu_device_recovery_prepare(adev, &device_list, hive)) goto end_reset; + /* We need to lock reset domain only once both for XGMI and single device */ + amdgpu_device_recovery_get_reset_lock(adev, &device_list); + + amdgpu_device_halt_activities(adev, job, reset_context, &device_list, + hive, need_emergency_restart); if (need_emergency_restart) goto skip_sched_resume; /* @@ -6337,7 +6482,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, * * job->base holds a reference to parent fence */ - if (job && dma_fence_is_signaled(&job->hw_fence)) { + if (job && dma_fence_is_signaled(&job->hw_fence.base)) { job_signaled = true; dev_info(adev->dev, "Guilty job already signaled, skipping HW reset"); goto skip_hw_reset; @@ -6345,13 +6490,15 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, r = amdgpu_device_asic_reset(adev, &device_list, reset_context); if (r) - goto end_reset; + goto reset_unlock; skip_hw_reset: r = amdgpu_device_sched_resume(&device_list, reset_context, job_signaled); if (r) - goto end_reset; + goto reset_unlock; skip_sched_resume: amdgpu_device_gpu_resume(adev, &device_list, need_emergency_restart); +reset_unlock: + amdgpu_device_recovery_put_reset_lock(adev, &device_list); end_reset: if (hive) { mutex_unlock(&hive->hive_lock); @@ -6747,7 +6894,6 @@ pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev, pci_channel_sta struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(adev); struct amdgpu_reset_context reset_context; struct list_head device_list; - int r = 0; dev_info(adev->dev, "PCI error: detected callback!!\n"); @@ -6772,14 +6918,14 @@ pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev, pci_channel_sta memset(&reset_context, 0, sizeof(reset_context)); INIT_LIST_HEAD(&device_list); - r = amdgpu_device_halt_activities(adev, NULL, &reset_context, &device_list, - hive, false); + amdgpu_device_recovery_prepare(adev, &device_list, hive); + amdgpu_device_recovery_get_reset_lock(adev, &device_list); + amdgpu_device_halt_activities(adev, NULL, &reset_context, &device_list, + hive, false); if (hive) { mutex_unlock(&hive->hive_lock); amdgpu_put_xgmi_hive(hive); } - if (r) - return PCI_ERS_RESULT_DISCONNECT; return PCI_ERS_RESULT_NEED_RESET; case pci_channel_io_perm_failure: /* Permanent error, prepare for device removal */ @@ -6889,8 +7035,8 @@ out: if (hive) { list_for_each_entry(tmp_adev, &device_list, reset_list) amdgpu_device_unset_mp1_state(tmp_adev); - amdgpu_device_unlock_reset_domain(adev->reset_domain); } + amdgpu_device_recovery_put_reset_lock(adev, &device_list); } if (hive) { @@ -6936,6 +7082,7 @@ void amdgpu_pci_resume(struct pci_dev *pdev) amdgpu_device_sched_resume(&device_list, NULL, NULL); amdgpu_device_gpu_resume(adev, &device_list, false); + amdgpu_device_recovery_put_reset_lock(adev, &device_list); adev->pcie_reset_ctx.occurs_dpc = false; if (hive) { @@ -6960,11 +7107,11 @@ bool amdgpu_device_cache_pci_state(struct pci_dev *pdev) adev->pci_state = pci_store_saved_state(pdev); if (!adev->pci_state) { - DRM_ERROR("Failed to store PCI saved state"); + dev_err(adev->dev, "Failed to store PCI saved state"); return false; } } else { - DRM_WARN("Failed to save PCI state, err:%d\n", r); + dev_warn(adev->dev, "Failed to save PCI state, err:%d\n", r); return false; } @@ -6985,7 +7132,7 @@ bool amdgpu_device_load_pci_state(struct pci_dev *pdev) if (!r) { pci_restore_state(pdev); } else { - DRM_WARN("Failed to load PCI state, err:%d\n", r); + dev_warn(adev->dev, "Failed to load PCI state, err:%d\n", r); return false; } @@ -7231,7 +7378,7 @@ struct dma_fence *amdgpu_device_enforce_isolation(struct amdgpu_device *adev, dep = amdgpu_sync_peek_fence(&isolation->prev, ring); r = amdgpu_sync_fence(&isolation->active, &f->finished, GFP_NOWAIT); if (r) - DRM_WARN("OOM tracking isolation\n"); + dev_warn(adev->dev, "OOM tracking isolation\n"); out_grab_ref: dma_fence_get(dep); @@ -7299,9 +7446,11 @@ uint32_t amdgpu_device_wait_on_rreg(struct amdgpu_device *adev, tmp_ = RREG32(reg_addr); loop--; if (!loop) { - DRM_WARN("Register(%d) [%s] failed to reach value 0x%08x != 0x%08xn", - inst, reg_name, (uint32_t)expected_value, - (uint32_t)(tmp_ & (mask))); + dev_warn( + adev->dev, + "Register(%d) [%s] failed to reach value 0x%08x != 0x%08xn", + inst, reg_name, (uint32_t)expected_value, + (uint32_t)(tmp_ & (mask))); ret = -ETIMEDOUT; break; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c index a0e9bf9b2710..81b3443c8d7f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c @@ -321,10 +321,12 @@ static int amdgpu_discovery_read_binary_from_file(struct amdgpu_device *adev, const struct firmware *fw; int r; - r = request_firmware(&fw, fw_name, adev->dev); + r = firmware_request_nowarn(&fw, fw_name, adev->dev); if (r) { - dev_err(adev->dev, "can't load firmware \"%s\"\n", - fw_name); + if (amdgpu_discovery == 2) + dev_err(adev->dev, "can't load firmware \"%s\"\n", fw_name); + else + drm_info(&adev->ddev, "Optional firmware \"%s\" was not found\n", fw_name); return r; } @@ -459,16 +461,12 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) /* Read from file if it is the preferred option */ fw_name = amdgpu_discovery_get_fw_name(adev); if (fw_name != NULL) { - dev_info(adev->dev, "use ip discovery information from file"); + drm_dbg(&adev->ddev, "use ip discovery information from file"); r = amdgpu_discovery_read_binary_from_file(adev, adev->mman.discovery_bin, fw_name); - - if (r) { - dev_err(adev->dev, "failed to read ip discovery binary from file\n"); - r = -EINVAL; + if (r) goto out; - } - } else { + drm_dbg(&adev->ddev, "use ip discovery information from memory"); r = amdgpu_discovery_read_binary_from_mem( adev, adev->mman.discovery_bin); if (r) @@ -1338,10 +1336,8 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev) int r; r = amdgpu_discovery_init(adev); - if (r) { - DRM_ERROR("amdgpu_discovery_init failed\n"); + if (r) return r; - } wafl_ver = 0; adev->gfx.xcc_mask = 0; @@ -2579,8 +2575,10 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) break; default: r = amdgpu_discovery_reg_base_init(adev); - if (r) - return -EINVAL; + if (r) { + drm_err(&adev->ddev, "discovery failed: %d\n", r); + return r; + } amdgpu_discovery_harvest_ip(adev); amdgpu_discovery_get_gfx_info(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c index 35c778426a7c..9e463d3ee927 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c @@ -1317,7 +1317,7 @@ amdgpu_display_user_framebuffer_create(struct drm_device *dev, /* Handle is imported dma-buf, so cannot be migrated to VRAM for scanout */ bo = gem_to_amdgpu_bo(obj); domains = amdgpu_display_supported_domains(drm_to_adev(dev), bo->flags); - if (obj->import_attach && !(domains & AMDGPU_GEM_DOMAIN_GTT)) { + if (drm_gem_is_imported(obj) && !(domains & AMDGPU_GEM_DOMAIN_GTT)) { drm_dbg_kms(dev, "Cannot create framebuffer from imported dma_buf\n"); drm_gem_object_put(obj); return ERR_PTR(-EINVAL); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c index 44e120f9f764..ff98c87b2e0b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c @@ -513,8 +513,8 @@ bool amdgpu_dmabuf_is_xgmi_accessible(struct amdgpu_device *adev, if (!adev) return false; - if (obj->import_attach) { - struct dma_buf *dma_buf = obj->import_attach->dmabuf; + if (drm_gem_is_imported(obj)) { + struct dma_buf *dma_buf = obj->dma_buf; if (dma_buf->ops != &amdgpu_dmabuf_ops) /* No XGMI with non AMD GPUs */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell_mgr.c index 3f3662e8b871..3040437d99c2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell_mgr.c @@ -41,7 +41,8 @@ u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index) if (index < adev->doorbell.num_kernel_doorbells) return readl(adev->doorbell.cpu_addr + index); - DRM_ERROR("reading beyond doorbell aperture: 0x%08x!\n", index); + dev_err(adev->dev, "reading beyond doorbell aperture: 0x%08x!\n", + index); return 0; } @@ -63,7 +64,8 @@ void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v) if (index < adev->doorbell.num_kernel_doorbells) writel(v, adev->doorbell.cpu_addr + index); else - DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index); + dev_err(adev->dev, + "writing beyond doorbell aperture: 0x%08x!\n", index); } /** @@ -83,7 +85,8 @@ u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index) if (index < adev->doorbell.num_kernel_doorbells) return atomic64_read((atomic64_t *)(adev->doorbell.cpu_addr + index)); - DRM_ERROR("reading beyond doorbell aperture: 0x%08x!\n", index); + dev_err(adev->dev, "reading beyond doorbell aperture: 0x%08x!\n", + index); return 0; } @@ -105,7 +108,8 @@ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v) if (index < adev->doorbell.num_kernel_doorbells) atomic64_set((atomic64_t *)(adev->doorbell.cpu_addr + index), v); else - DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index); + dev_err(adev->dev, + "writing beyond doorbell aperture: 0x%08x!\n", index); } /** @@ -166,7 +170,8 @@ int amdgpu_doorbell_create_kernel_doorbells(struct amdgpu_device *adev) NULL, (void **)&adev->doorbell.cpu_addr); if (r) { - DRM_ERROR("Failed to allocate kernel doorbells, err=%d\n", r); + dev_err(adev->dev, + "Failed to allocate kernel doorbells, err=%d\n", r); return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 4db92e0a60da..b299e15bb5e5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -144,6 +144,7 @@ enum AMDGPU_DEBUG_MASK { AMDGPU_DEBUG_DISABLE_GPU_RING_RESET = BIT(6), AMDGPU_DEBUG_SMU_POOL = BIT(7), AMDGPU_DEBUG_VM_USERPTR = BIT(8), + AMDGPU_DEBUG_DISABLE_RAS_CE_LOG = BIT(9) }; unsigned int amdgpu_vram_limit = UINT_MAX; @@ -2278,6 +2279,11 @@ static void amdgpu_init_debug_options(struct amdgpu_device *adev) pr_info("debug: VM mode debug for userptr is enabled\n"); adev->debug_vm_userptr = true; } + + if (amdgpu_debug_mask & AMDGPU_DEBUG_DISABLE_RAS_CE_LOG) { + pr_info("debug: disable kernel logs of correctable errors\n"); + adev->debug_disable_ce_logs = true; + } } static unsigned long amdgpu_fix_asic_type(struct pci_dev *pdev, unsigned long flags) @@ -2321,7 +2327,7 @@ static int amdgpu_pci_probe(struct pci_dev *pdev, amdgpu_aspm = 0; if (amdgpu_virtual_display || - amdgpu_device_asic_has_dc_support(flags & AMD_ASIC_MASK)) + amdgpu_device_asic_has_dc_support(pdev, flags & AMD_ASIC_MASK)) supports_atomic = true; if ((flags & AMD_EXP_HW_SUPPORT) && !amdgpu_exp_hw_support) { @@ -2570,7 +2576,7 @@ static int amdgpu_pmops_prepare(struct device *dev) static void amdgpu_pmops_complete(struct device *dev) { - /* nothing to do */ + amdgpu_device_complete(dev_get_drvdata(dev)); } static int amdgpu_pmops_suspend(struct device *dev) @@ -2906,20 +2912,6 @@ done: return ret; } -static int amdgpu_drm_release(struct inode *inode, struct file *filp) -{ - struct drm_file *file_priv = filp->private_data; - struct amdgpu_fpriv *fpriv = file_priv->driver_priv; - - if (fpriv) { - fpriv->evf_mgr.fd_closing = true; - amdgpu_eviction_fence_destroy(&fpriv->evf_mgr); - amdgpu_userq_mgr_fini(&fpriv->userq_mgr); - } - - return drm_release(inode, filp); -} - long amdgpu_drm_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) { @@ -2971,7 +2963,7 @@ static const struct file_operations amdgpu_driver_kms_fops = { .owner = THIS_MODULE, .open = drm_open, .flush = amdgpu_flush, - .release = amdgpu_drm_release, + .release = drm_release, .unlocked_ioctl = amdgpu_drm_ioctl, .mmap = drm_gem_mmap, .poll = drm_poll, @@ -3107,10 +3099,6 @@ static int __init amdgpu_init(void) if (r) goto error_sync; - r = amdgpu_fence_slab_init(); - if (r) - goto error_fence; - r = amdgpu_userq_fence_slab_init(); if (r) goto error_fence; @@ -3145,7 +3133,6 @@ static void __exit amdgpu_exit(void) amdgpu_unregister_atpx_handler(); amdgpu_acpi_release(); amdgpu_sync_fini(); - amdgpu_fence_slab_fini(); amdgpu_userq_fence_slab_fini(); mmu_notifier_synchronize(); amdgpu_xcp_drv_release(); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c index 8cecf25996ed..2c3547f4cea4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c @@ -42,37 +42,6 @@ #include "amdgpu_reset.h" /* - * Fences mark an event in the GPUs pipeline and are used - * for GPU/CPU synchronization. When the fence is written, - * it is expected that all buffers associated with that fence - * are no longer in use by the associated ring on the GPU and - * that the relevant GPU caches have been flushed. - */ - -struct amdgpu_fence { - struct dma_fence base; - - /* RB, DMA, etc. */ - struct amdgpu_ring *ring; - ktime_t start_timestamp; -}; - -static struct kmem_cache *amdgpu_fence_slab; - -int amdgpu_fence_slab_init(void) -{ - amdgpu_fence_slab = KMEM_CACHE(amdgpu_fence, SLAB_HWCACHE_ALIGN); - if (!amdgpu_fence_slab) - return -ENOMEM; - return 0; -} - -void amdgpu_fence_slab_fini(void) -{ - rcu_barrier(); - kmem_cache_destroy(amdgpu_fence_slab); -} -/* * Cast helper */ static const struct dma_fence_ops amdgpu_fence_ops; @@ -130,14 +99,14 @@ static u32 amdgpu_fence_read(struct amdgpu_ring *ring) * * @ring: ring the fence is associated with * @f: resulting fence object - * @job: job the fence is embedded in + * @af: amdgpu fence input * @flags: flags to pass into the subordinate .emit_fence() call * * Emits a fence command on the requested ring (all asics). * Returns 0 on success, -ENOMEM on failure. */ -int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f, struct amdgpu_job *job, - unsigned int flags) +int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f, + struct amdgpu_fence *af, unsigned int flags) { struct amdgpu_device *adev = ring->adev; struct dma_fence *fence; @@ -146,36 +115,28 @@ int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f, struct amd uint32_t seq; int r; - if (job == NULL) { - /* create a sperate hw fence */ - am_fence = kmem_cache_alloc(amdgpu_fence_slab, GFP_ATOMIC); - if (am_fence == NULL) + if (!af) { + /* create a separate hw fence */ + am_fence = kzalloc(sizeof(*am_fence), GFP_KERNEL); + if (!am_fence) return -ENOMEM; - fence = &am_fence->base; - am_fence->ring = ring; } else { - /* take use of job-embedded fence */ - fence = &job->hw_fence; + am_fence = af; } + fence = &am_fence->base; + am_fence->ring = ring; seq = ++ring->fence_drv.sync_seq; - if (job && job->job_run_counter) { - /* reinit seq for resubmitted jobs */ - fence->seqno = seq; - /* TO be inline with external fence creation and other drivers */ + if (af) { + dma_fence_init(fence, &amdgpu_job_fence_ops, + &ring->fence_drv.lock, + adev->fence_context + ring->idx, seq); + /* Against remove in amdgpu_job_{free, free_cb} */ dma_fence_get(fence); } else { - if (job) { - dma_fence_init(fence, &amdgpu_job_fence_ops, - &ring->fence_drv.lock, - adev->fence_context + ring->idx, seq); - /* Against remove in amdgpu_job_{free, free_cb} */ - dma_fence_get(fence); - } else { - dma_fence_init(fence, &amdgpu_fence_ops, - &ring->fence_drv.lock, - adev->fence_context + ring->idx, seq); - } + dma_fence_init(fence, &amdgpu_fence_ops, + &ring->fence_drv.lock, + adev->fence_context + ring->idx, seq); } amdgpu_ring_emit_fence(ring, ring->fence_drv.gpu_addr, @@ -326,7 +287,9 @@ static void amdgpu_fence_fallback(struct timer_list *t) fence_drv.fallback_timer); if (amdgpu_fence_process(ring)) - DRM_WARN("Fence fallback timer expired on ring %s\n", ring->name); + dev_warn(ring->adev->dev, + "Fence fallback timer expired on ring %s\n", + ring->name); } /** @@ -718,7 +681,7 @@ void amdgpu_fence_driver_clear_job_fences(struct amdgpu_ring *ring) * it right here or we won't be able to track them in fence_drv * and they will remain unsignaled during sa_bo free. */ - job = container_of(old, struct amdgpu_job, hw_fence); + job = container_of(old, struct amdgpu_job, hw_fence.base); if (!job->base.s_fence && !dma_fence_is_signaled(old)) dma_fence_signal(old); RCU_INIT_POINTER(*ptr, NULL); @@ -780,7 +743,7 @@ static const char *amdgpu_fence_get_timeline_name(struct dma_fence *f) static const char *amdgpu_job_fence_get_timeline_name(struct dma_fence *f) { - struct amdgpu_job *job = container_of(f, struct amdgpu_job, hw_fence); + struct amdgpu_job *job = container_of(f, struct amdgpu_job, hw_fence.base); return (const char *)to_amdgpu_ring(job->base.sched)->name; } @@ -810,7 +773,7 @@ static bool amdgpu_fence_enable_signaling(struct dma_fence *f) */ static bool amdgpu_job_fence_enable_signaling(struct dma_fence *f) { - struct amdgpu_job *job = container_of(f, struct amdgpu_job, hw_fence); + struct amdgpu_job *job = container_of(f, struct amdgpu_job, hw_fence.base); if (!timer_pending(&to_amdgpu_ring(job->base.sched)->fence_drv.fallback_timer)) amdgpu_fence_schedule_fallback(to_amdgpu_ring(job->base.sched)); @@ -830,7 +793,7 @@ static void amdgpu_fence_free(struct rcu_head *rcu) struct dma_fence *f = container_of(rcu, struct dma_fence, rcu); /* free fence_slab if it's separated fence*/ - kmem_cache_free(amdgpu_fence_slab, to_amdgpu_fence(f)); + kfree(to_amdgpu_fence(f)); } /** @@ -845,7 +808,7 @@ static void amdgpu_job_fence_free(struct rcu_head *rcu) struct dma_fence *f = container_of(rcu, struct dma_fence, rcu); /* free job if fence has a parent job */ - kfree(container_of(f, struct amdgpu_job, hw_fence)); + kfree(container_of(f, struct amdgpu_job, hw_fence.base)); } /** diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c index 1ae88c459da5..b0082aa7f3c6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c @@ -144,7 +144,8 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev) /* If algo exists, it means that the i2c_adapter's initialized */ if (!adev->pm.fru_eeprom_i2c_bus || !adev->pm.fru_eeprom_i2c_bus->algo) { - DRM_WARN("Cannot access FRU, EEPROM accessor not initialized"); + dev_warn(adev->dev, + "Cannot access FRU, EEPROM accessor not initialized"); return -ENODEV; } @@ -152,19 +153,22 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev) len = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, fru_addr, buf, sizeof(buf)); if (len != 8) { - DRM_ERROR("Couldn't read the IPMI Common Header: %d", len); + dev_err(adev->dev, "Couldn't read the IPMI Common Header: %d", + len); return len < 0 ? len : -EIO; } if (buf[0] != 1) { - DRM_ERROR("Bad IPMI Common Header version: 0x%02x", buf[0]); + dev_err(adev->dev, "Bad IPMI Common Header version: 0x%02x", + buf[0]); return -EIO; } for (csum = 0; len > 0; len--) csum += buf[len - 1]; if (csum) { - DRM_ERROR("Bad IPMI Common Header checksum: 0x%02x", csum); + dev_err(adev->dev, "Bad IPMI Common Header checksum: 0x%02x", + csum); return -EIO; } @@ -179,12 +183,14 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev) /* Read the header of the PIA. */ len = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, addr, buf, 3); if (len != 3) { - DRM_ERROR("Couldn't read the Product Info Area header: %d", len); + dev_err(adev->dev, + "Couldn't read the Product Info Area header: %d", len); return len < 0 ? len : -EIO; } if (buf[0] != 1) { - DRM_ERROR("Bad IPMI Product Info Area version: 0x%02x", buf[0]); + dev_err(adev->dev, "Bad IPMI Product Info Area version: 0x%02x", + buf[0]); return -EIO; } @@ -197,14 +203,16 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev) len = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, addr, pia, size); if (len != size) { kfree(pia); - DRM_ERROR("Couldn't read the Product Info Area: %d", len); + dev_err(adev->dev, "Couldn't read the Product Info Area: %d", + len); return len < 0 ? len : -EIO; } for (csum = 0; size > 0; size--) csum += pia[size - 1]; if (csum) { - DRM_ERROR("Bad Product Info Area checksum: 0x%02x", csum); + dev_err(adev->dev, "Bad Product Info Area checksum: 0x%02x", + csum); kfree(pia); return -EIO; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index e5e33a68d935..6626a6e64ff5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -317,8 +317,7 @@ static int amdgpu_gem_object_open(struct drm_gem_object *obj, */ if (!vm->is_compute_context || !vm->process_info) return 0; - if (!obj->import_attach || - !dma_buf_is_dynamic(obj->import_attach->dmabuf)) + if (!drm_gem_is_imported(obj) || !dma_buf_is_dynamic(obj->dma_buf)) return 0; mutex_lock_nested(&vm->process_info->lock, 1); if (!WARN_ON(!vm->process_info->eviction_fence)) { @@ -1024,7 +1023,7 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data, break; } case AMDGPU_GEM_OP_SET_PLACEMENT: - if (robj->tbo.base.import_attach && + if (drm_gem_is_imported(&robj->tbo.base) && args->value & AMDGPU_GEM_DOMAIN_VRAM) { r = -EINVAL; amdgpu_bo_unreserve(robj); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index c5646af055ab..c80c8f543532 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -149,7 +149,7 @@ static bool amdgpu_gfx_is_graphics_multipipe_capable(struct amdgpu_device *adev) static bool amdgpu_gfx_is_compute_multipipe_capable(struct amdgpu_device *adev) { if (amdgpu_compute_multipipe != -1) { - DRM_INFO("amdgpu: forcing compute pipe policy %d\n", + dev_info(adev->dev, "amdgpu: forcing compute pipe policy %d\n", amdgpu_compute_multipipe); return amdgpu_compute_multipipe == 1; } @@ -674,7 +674,7 @@ int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev, int xcc_id) * generation exposes more than 64 queues. If so, the * definition of queue_mask needs updating */ if (WARN_ON(i > (sizeof(queue_mask)*8))) { - DRM_ERROR("Invalid KCQ enabled: %d\n", i); + dev_err(adev->dev, "Invalid KCQ enabled: %d\n", i); break; } @@ -683,15 +683,15 @@ int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev, int xcc_id) amdgpu_device_flush_hdp(adev, NULL); - DRM_INFO("kiq ring mec %d pipe %d q %d\n", kiq_ring->me, kiq_ring->pipe, - kiq_ring->queue); + dev_info(adev->dev, "kiq ring mec %d pipe %d q %d\n", kiq_ring->me, + kiq_ring->pipe, kiq_ring->queue); spin_lock(&kiq->ring_lock); r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size * adev->gfx.num_compute_rings + kiq->pmf->set_resources_size); if (r) { - DRM_ERROR("Failed to lock KIQ (%d).\n", r); + dev_err(adev->dev, "Failed to lock KIQ (%d).\n", r); spin_unlock(&kiq->ring_lock); return r; } @@ -712,7 +712,7 @@ int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev, int xcc_id) r = amdgpu_ring_test_helper(kiq_ring); spin_unlock(&kiq->ring_lock); if (r) - DRM_ERROR("KCQ enable failed\n"); + dev_err(adev->dev, "KCQ enable failed\n"); return r; } @@ -734,7 +734,7 @@ int amdgpu_gfx_enable_kgq(struct amdgpu_device *adev, int xcc_id) r = amdgpu_mes_map_legacy_queue(adev, &adev->gfx.gfx_ring[j]); if (r) { - DRM_ERROR("failed to map gfx queue\n"); + dev_err(adev->dev, "failed to map gfx queue\n"); return r; } } @@ -748,7 +748,7 @@ int amdgpu_gfx_enable_kgq(struct amdgpu_device *adev, int xcc_id) r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size * adev->gfx.num_gfx_rings); if (r) { - DRM_ERROR("Failed to lock KIQ (%d).\n", r); + dev_err(adev->dev, "Failed to lock KIQ (%d).\n", r); spin_unlock(&kiq->ring_lock); return r; } @@ -769,7 +769,7 @@ int amdgpu_gfx_enable_kgq(struct amdgpu_device *adev, int xcc_id) r = amdgpu_ring_test_helper(kiq_ring); spin_unlock(&kiq->ring_lock); if (r) - DRM_ERROR("KGQ enable failed\n"); + dev_err(adev->dev, "KGQ enable failed\n"); return r; } @@ -1030,7 +1030,7 @@ int amdgpu_gfx_cp_ecc_error_irq(struct amdgpu_device *adev, ih_data.head = *ras_if; - DRM_ERROR("CP ECC ERROR IRQ\n"); + dev_err(adev->dev, "CP ECC ERROR IRQ\n"); amdgpu_ras_interrupt_dispatch(adev, &ih_data); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index 6b0fbbb91e57..97b562a79ea8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -38,6 +38,13 @@ #include <drm/drm_drv.h> #include <drm/ttm/ttm_tt.h> +static const u64 four_gb = 0x100000000ULL; + +bool amdgpu_gmc_is_pdb0_enabled(struct amdgpu_device *adev) +{ + return adev->gmc.xgmi.connected_to_cpu || amdgpu_virt_xgmi_migrate_enabled(adev); +} + /** * amdgpu_gmc_pdb0_alloc - allocate vram for pdb0 * @@ -251,10 +258,20 @@ void amdgpu_gmc_sysvm_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc u64 hive_vram_end = mc->xgmi.node_segment_size * mc->xgmi.num_physical_nodes - 1; mc->vram_start = mc->xgmi.node_segment_size * mc->xgmi.physical_node_id; mc->vram_end = mc->vram_start + mc->xgmi.node_segment_size - 1; - mc->gart_start = hive_vram_end + 1; + /* node_segment_size may not 4GB aligned on SRIOV, align up is needed. */ + mc->gart_start = ALIGN(hive_vram_end + 1, four_gb); mc->gart_end = mc->gart_start + mc->gart_size - 1; - mc->fb_start = hive_vram_start; - mc->fb_end = hive_vram_end; + if (amdgpu_virt_xgmi_migrate_enabled(adev)) { + /* set mc->vram_start to 0 to switch the returned GPU address of + * amdgpu_bo_create_reserved() from FB aperture to GART aperture. + */ + mc->vram_start = 0; + mc->vram_end = mc->vram_start + mc->mc_vram_size - 1; + mc->visible_vram_size = min(mc->visible_vram_size, mc->real_vram_size); + } else { + mc->fb_start = hive_vram_start; + mc->fb_end = hive_vram_end; + } dev_info(adev->dev, "VRAM: %lluM 0x%016llX - 0x%016llX (%lluM used)\n", mc->mc_vram_size >> 20, mc->vram_start, mc->vram_end, mc->real_vram_size >> 20); @@ -276,7 +293,6 @@ void amdgpu_gmc_sysvm_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc void amdgpu_gmc_gart_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc, enum amdgpu_gart_placement gart_placement) { - const uint64_t four_gb = 0x100000000ULL; u64 size_af, size_bf; /*To avoid the hole, limit the max mc address to AMDGPU_GMC_HOLE_START*/ u64 max_mc_address = min(adev->gmc.mc_mask, AMDGPU_GMC_HOLE_START - 1); @@ -1041,9 +1057,7 @@ void amdgpu_gmc_init_pdb0(struct amdgpu_device *adev) */ u64 vram_size = adev->gmc.xgmi.node_segment_size * adev->gmc.xgmi.num_physical_nodes; u64 pde0_page_size = (1ULL<<adev->gmc.vmid0_page_table_block_size)<<21; - u64 vram_addr = adev->vm_manager.vram_base_offset - - adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size; - u64 vram_end = vram_addr + vram_size; + u64 vram_addr, vram_end; u64 gart_ptb_gpu_pa = amdgpu_gmc_vram_pa(adev, adev->gart.bo); int idx; @@ -1056,6 +1070,11 @@ void amdgpu_gmc_init_pdb0(struct amdgpu_device *adev) flags |= AMDGPU_PTE_FRAG((adev->gmc.vmid0_page_table_block_size + 9*1)); flags |= AMDGPU_PDE_PTE_FLAG(adev); + vram_addr = adev->vm_manager.vram_base_offset; + if (!amdgpu_virt_xgmi_migrate_enabled(adev)) + vram_addr -= adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size; + vram_end = vram_addr + vram_size; + /* The first n PDE0 entries are used as PTE, * pointing to vram */ @@ -1429,3 +1448,232 @@ bool amdgpu_gmc_need_reset_on_init(struct amdgpu_device *adev) return false; } + +enum amdgpu_memory_partition +amdgpu_gmc_get_vf_memory_partition(struct amdgpu_device *adev) +{ + switch (adev->gmc.num_mem_partitions) { + case 0: + return UNKNOWN_MEMORY_PARTITION_MODE; + case 1: + return AMDGPU_NPS1_PARTITION_MODE; + case 2: + return AMDGPU_NPS2_PARTITION_MODE; + case 4: + return AMDGPU_NPS4_PARTITION_MODE; + case 8: + return AMDGPU_NPS8_PARTITION_MODE; + default: + return AMDGPU_NPS1_PARTITION_MODE; + } +} + +enum amdgpu_memory_partition +amdgpu_gmc_get_memory_partition(struct amdgpu_device *adev, u32 *supp_modes) +{ + enum amdgpu_memory_partition mode = UNKNOWN_MEMORY_PARTITION_MODE; + + if (adev->nbio.funcs && + adev->nbio.funcs->get_memory_partition_mode) + mode = adev->nbio.funcs->get_memory_partition_mode(adev, + supp_modes); + else + dev_warn(adev->dev, "memory partition mode query is not supported\n"); + + return mode; +} + +enum amdgpu_memory_partition +amdgpu_gmc_query_memory_partition(struct amdgpu_device *adev) +{ + if (amdgpu_sriov_vf(adev)) + return amdgpu_gmc_get_vf_memory_partition(adev); + else + return amdgpu_gmc_get_memory_partition(adev, NULL); +} + +static bool amdgpu_gmc_validate_partition_info(struct amdgpu_device *adev) +{ + enum amdgpu_memory_partition mode; + u32 supp_modes; + bool valid; + + mode = amdgpu_gmc_get_memory_partition(adev, &supp_modes); + + /* Mode detected by hardware not present in supported modes */ + if ((mode != UNKNOWN_MEMORY_PARTITION_MODE) && + !(BIT(mode - 1) & supp_modes)) + return false; + + switch (mode) { + case UNKNOWN_MEMORY_PARTITION_MODE: + case AMDGPU_NPS1_PARTITION_MODE: + valid = (adev->gmc.num_mem_partitions == 1); + break; + case AMDGPU_NPS2_PARTITION_MODE: + valid = (adev->gmc.num_mem_partitions == 2); + break; + case AMDGPU_NPS4_PARTITION_MODE: + valid = (adev->gmc.num_mem_partitions == 3 || + adev->gmc.num_mem_partitions == 4); + break; + case AMDGPU_NPS8_PARTITION_MODE: + valid = (adev->gmc.num_mem_partitions == 8); + break; + default: + valid = false; + } + + return valid; +} + +static bool amdgpu_gmc_is_node_present(int *node_ids, int num_ids, int nid) +{ + int i; + + /* Check if node with id 'nid' is present in 'node_ids' array */ + for (i = 0; i < num_ids; ++i) + if (node_ids[i] == nid) + return true; + + return false; +} + +static void +amdgpu_gmc_init_acpi_mem_ranges(struct amdgpu_device *adev, + struct amdgpu_mem_partition_info *mem_ranges) +{ + struct amdgpu_numa_info numa_info; + int node_ids[AMDGPU_MAX_MEM_RANGES]; + int num_ranges = 0, ret; + int num_xcc, xcc_id; + uint32_t xcc_mask; + + num_xcc = NUM_XCC(adev->gfx.xcc_mask); + xcc_mask = (1U << num_xcc) - 1; + + for_each_inst(xcc_id, xcc_mask) { + ret = amdgpu_acpi_get_mem_info(adev, xcc_id, &numa_info); + if (ret) + continue; + + if (numa_info.nid == NUMA_NO_NODE) { + mem_ranges[0].size = numa_info.size; + mem_ranges[0].numa.node = numa_info.nid; + num_ranges = 1; + break; + } + + if (amdgpu_gmc_is_node_present(node_ids, num_ranges, + numa_info.nid)) + continue; + + node_ids[num_ranges] = numa_info.nid; + mem_ranges[num_ranges].numa.node = numa_info.nid; + mem_ranges[num_ranges].size = numa_info.size; + ++num_ranges; + } + + adev->gmc.num_mem_partitions = num_ranges; +} + +void amdgpu_gmc_init_sw_mem_ranges(struct amdgpu_device *adev, + struct amdgpu_mem_partition_info *mem_ranges) +{ + enum amdgpu_memory_partition mode; + u32 start_addr = 0, size; + int i, r, l; + + mode = amdgpu_gmc_query_memory_partition(adev); + + switch (mode) { + case UNKNOWN_MEMORY_PARTITION_MODE: + adev->gmc.num_mem_partitions = 0; + break; + case AMDGPU_NPS1_PARTITION_MODE: + adev->gmc.num_mem_partitions = 1; + break; + case AMDGPU_NPS2_PARTITION_MODE: + adev->gmc.num_mem_partitions = 2; + break; + case AMDGPU_NPS4_PARTITION_MODE: + if (adev->flags & AMD_IS_APU) + adev->gmc.num_mem_partitions = 3; + else + adev->gmc.num_mem_partitions = 4; + break; + case AMDGPU_NPS8_PARTITION_MODE: + adev->gmc.num_mem_partitions = 8; + break; + default: + adev->gmc.num_mem_partitions = 1; + break; + } + + /* Use NPS range info, if populated */ + r = amdgpu_gmc_get_nps_memranges(adev, mem_ranges, + &adev->gmc.num_mem_partitions); + if (!r) { + l = 0; + for (i = 1; i < adev->gmc.num_mem_partitions; ++i) { + if (mem_ranges[i].range.lpfn > + mem_ranges[i - 1].range.lpfn) + l = i; + } + + } else { + if (!adev->gmc.num_mem_partitions) { + dev_warn(adev->dev, + "Not able to detect NPS mode, fall back to NPS1\n"); + adev->gmc.num_mem_partitions = 1; + } + /* Fallback to sw based calculation */ + size = (adev->gmc.real_vram_size + SZ_16M) >> AMDGPU_GPU_PAGE_SHIFT; + size /= adev->gmc.num_mem_partitions; + + for (i = 0; i < adev->gmc.num_mem_partitions; ++i) { + mem_ranges[i].range.fpfn = start_addr; + mem_ranges[i].size = + ((u64)size << AMDGPU_GPU_PAGE_SHIFT); + mem_ranges[i].range.lpfn = start_addr + size - 1; + start_addr += size; + } + + l = adev->gmc.num_mem_partitions - 1; + } + + /* Adjust the last one */ + mem_ranges[l].range.lpfn = + (adev->gmc.real_vram_size >> AMDGPU_GPU_PAGE_SHIFT) - 1; + mem_ranges[l].size = + adev->gmc.real_vram_size - + ((u64)mem_ranges[l].range.fpfn << AMDGPU_GPU_PAGE_SHIFT); +} + +int amdgpu_gmc_init_mem_ranges(struct amdgpu_device *adev) +{ + bool valid; + + adev->gmc.mem_partitions = kcalloc(AMDGPU_MAX_MEM_RANGES, + sizeof(struct amdgpu_mem_partition_info), + GFP_KERNEL); + if (!adev->gmc.mem_partitions) + return -ENOMEM; + + if (adev->gmc.is_app_apu) + amdgpu_gmc_init_acpi_mem_ranges(adev, adev->gmc.mem_partitions); + else + amdgpu_gmc_init_sw_mem_ranges(adev, adev->gmc.mem_partitions); + + if (amdgpu_sriov_vf(adev)) + valid = true; + else + valid = amdgpu_gmc_validate_partition_info(adev); + if (!valid) { + /* TODO: handle invalid case */ + dev_warn(adev->dev, + "Mem ranges not matching with hardware config\n"); + } + + return 0; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h index 80fa29c26e9e..397c6ccdb903 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h @@ -84,6 +84,8 @@ enum amdgpu_memory_partition { #define AMDGPU_GMC_INIT_RESET_NPS BIT(0) +#define AMDGPU_MAX_MEM_RANGES 8 + /* * GMC page fault information */ @@ -394,6 +396,7 @@ static inline uint64_t amdgpu_gmc_sign_extend(uint64_t addr) return addr; } +bool amdgpu_gmc_is_pdb0_enabled(struct amdgpu_device *adev); int amdgpu_gmc_pdb0_alloc(struct amdgpu_device *adev); void amdgpu_gmc_get_pde_for_bo(struct amdgpu_bo *bo, int level, uint64_t *addr, uint64_t *flags); @@ -455,5 +458,13 @@ int amdgpu_gmc_request_memory_partition(struct amdgpu_device *adev, int nps_mode); void amdgpu_gmc_prepare_nps_mode_change(struct amdgpu_device *adev); bool amdgpu_gmc_need_reset_on_init(struct amdgpu_device *adev); - +enum amdgpu_memory_partition +amdgpu_gmc_get_vf_memory_partition(struct amdgpu_device *adev); +enum amdgpu_memory_partition +amdgpu_gmc_get_memory_partition(struct amdgpu_device *adev, u32 *supp_modes); +enum amdgpu_memory_partition +amdgpu_gmc_query_memory_partition(struct amdgpu_device *adev); +int amdgpu_gmc_init_mem_ranges(struct amdgpu_device *adev); +void amdgpu_gmc_init_sw_mem_ranges(struct amdgpu_device *adev, + struct amdgpu_mem_partition_info *mem_ranges); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c index 8179d0814db9..57101d24422f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c @@ -24,7 +24,6 @@ * Alex Deucher */ -#include <linux/export.h> #include <linux/pci.h> #include <drm/drm_edid.h> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c index 802743efa3b3..206b70acb29a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c @@ -128,6 +128,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs, struct amdgpu_device *adev = ring->adev; struct amdgpu_ib *ib = &ibs[0]; struct dma_fence *tmp = NULL; + struct amdgpu_fence *af; bool need_ctx_switch; struct amdgpu_vm *vm; uint64_t fence_ctx; @@ -154,6 +155,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs, csa_va = job->csa_va; gds_va = job->gds_va; init_shadow = job->init_shadow; + af = &job->hw_fence; } else { vm = NULL; fence_ctx = 0; @@ -161,6 +163,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs, csa_va = 0; gds_va = 0; init_shadow = false; + af = NULL; } if (!ring->sched.ready) { @@ -282,7 +285,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs, amdgpu_ring_init_cond_exec(ring, ring->cond_exe_gpu_addr); } - r = amdgpu_fence_emit(ring, f, job, fence_flags); + r = amdgpu_fence_emit(ring, f, af, fence_flags); if (r) { dev_err(adev->dev, "failed to emit fence (%d)\n", r); if (job && job->vmid) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c index 30f16968b578..a6419246e9c2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c @@ -218,7 +218,7 @@ int amdgpu_ih_process(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih) restart_ih: count = AMDGPU_IH_MAX_NUM_IVS; - DRM_DEBUG("%s: rptr %d, wptr %d\n", __func__, ih->rptr, wptr); + dev_dbg(adev->dev, "%s: rptr %d, wptr %d\n", __func__, ih->rptr, wptr); /* Order reading of wptr vs. reading of IH ring data */ rmb(); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.c new file mode 100644 index 000000000000..99e1cf4fc955 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.c @@ -0,0 +1,96 @@ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "amdgpu.h" +#include "amdgpu_ip.h" + +static int8_t amdgpu_logical_to_dev_inst(struct amdgpu_device *adev, + enum amd_hw_ip_block_type block, + int8_t inst) +{ + int8_t dev_inst; + + switch (block) { + case GC_HWIP: + case SDMA0_HWIP: + /* Both JPEG and VCN as JPEG is only alias of VCN */ + case VCN_HWIP: + dev_inst = adev->ip_map.dev_inst[block][inst]; + break; + default: + /* For rest of the IPs, no look up required. + * Assume 'logical instance == physical instance' for all configs. */ + dev_inst = inst; + break; + } + + return dev_inst; +} + +static uint32_t amdgpu_logical_to_dev_mask(struct amdgpu_device *adev, + enum amd_hw_ip_block_type block, + uint32_t mask) +{ + uint32_t dev_mask = 0; + int8_t log_inst, dev_inst; + + while (mask) { + log_inst = ffs(mask) - 1; + dev_inst = amdgpu_logical_to_dev_inst(adev, block, log_inst); + dev_mask |= (1 << dev_inst); + mask &= ~(1 << log_inst); + } + + return dev_mask; +} + +static void amdgpu_populate_ip_map(struct amdgpu_device *adev, + enum amd_hw_ip_block_type ip_block, + uint32_t inst_mask) +{ + int l = 0, i; + + while (inst_mask) { + i = ffs(inst_mask) - 1; + adev->ip_map.dev_inst[ip_block][l++] = i; + inst_mask &= ~(1 << i); + } + for (; l < HWIP_MAX_INSTANCE; l++) + adev->ip_map.dev_inst[ip_block][l] = -1; +} + +void amdgpu_ip_map_init(struct amdgpu_device *adev) +{ + u32 ip_map[][2] = { + { GC_HWIP, adev->gfx.xcc_mask }, + { SDMA0_HWIP, adev->sdma.sdma_mask }, + { VCN_HWIP, adev->vcn.inst_mask }, + }; + int i; + + for (i = 0; i < ARRAY_SIZE(ip_map); ++i) + amdgpu_populate_ip_map(adev, ip_map[i][0], ip_map[i][1]); + + adev->ip_map.logical_to_dev_inst = amdgpu_logical_to_dev_inst; + adev->ip_map.logical_to_dev_mask = amdgpu_logical_to_dev_mask; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.h new file mode 100644 index 000000000000..2490fd322aec --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ip.h @@ -0,0 +1,29 @@ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __AMDGPU_IP_H__ +#define __AMDGPU_IP_H__ + +void amdgpu_ip_map_init(struct amdgpu_device *adev); + +#endif /* __AMDGPU_IP_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c index 13c60cac4261..183fa33c2434 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c @@ -142,8 +142,9 @@ void amdgpu_irq_disable_all(struct amdgpu_device *adev) r = src->funcs->set(adev, src, k, AMDGPU_IRQ_STATE_DISABLE); if (r) - DRM_ERROR("error disabling interrupt (%d)\n", - r); + dev_err(adev->dev, + "error disabling interrupt (%d)\n", + r); } } } @@ -242,7 +243,7 @@ static bool amdgpu_msi_ok(struct amdgpu_device *adev) return true; } -static void amdgpu_restore_msix(struct amdgpu_device *adev) +void amdgpu_restore_msix(struct amdgpu_device *adev) { u16 ctrl; @@ -315,7 +316,7 @@ int amdgpu_irq_init(struct amdgpu_device *adev) adev->irq.irq = irq; adev_to_drm(adev)->max_vblank_count = 0x00ffffff; - DRM_DEBUG("amdgpu: irq initialized.\n"); + dev_dbg(adev->dev, "amdgpu: irq initialized.\n"); return 0; free_vectors: @@ -461,10 +462,10 @@ void amdgpu_irq_dispatch(struct amdgpu_device *adev, src_id = entry.src_id; if (client_id >= AMDGPU_IRQ_CLIENTID_MAX) { - DRM_DEBUG("Invalid client_id in IV: %d\n", client_id); + dev_dbg(adev->dev, "Invalid client_id in IV: %d\n", client_id); } else if (src_id >= AMDGPU_MAX_IRQ_SRC_ID) { - DRM_DEBUG("Invalid src_id in IV: %d\n", src_id); + dev_dbg(adev->dev, "Invalid src_id in IV: %d\n", src_id); } else if (((client_id == AMDGPU_IRQ_CLIENTID_LEGACY) || (client_id == SOC15_IH_CLIENTID_ISP)) && @@ -472,18 +473,21 @@ void amdgpu_irq_dispatch(struct amdgpu_device *adev, generic_handle_domain_irq(adev->irq.domain, src_id); } else if (!adev->irq.client[client_id].sources) { - DRM_DEBUG("Unregistered interrupt client_id: %d src_id: %d\n", - client_id, src_id); + dev_dbg(adev->dev, + "Unregistered interrupt client_id: %d src_id: %d\n", + client_id, src_id); } else if ((src = adev->irq.client[client_id].sources[src_id])) { r = src->funcs->process(adev, src, &entry); if (r < 0) - DRM_ERROR("error processing interrupt (%d)\n", r); + dev_err(adev->dev, "error processing interrupt (%d)\n", + r); else if (r) handled = true; } else { - DRM_DEBUG("Unregistered interrupt src_id: %d of client_id:%d\n", + dev_dbg(adev->dev, + "Unregistered interrupt src_id: %d of client_id:%d\n", src_id, client_id); } @@ -732,7 +736,7 @@ int amdgpu_irq_add_domain(struct amdgpu_device *adev) adev->irq.domain = irq_domain_create_linear(NULL, AMDGPU_MAX_IRQ_SRC_ID, &amdgpu_hw_irqdomain_ops, adev); if (!adev->irq.domain) { - DRM_ERROR("GPU irq add domain failed\n"); + dev_err(adev->dev, "GPU irq add domain failed\n"); return -ENODEV; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h index 04c0b4fa17a4..9f0417456abd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h @@ -146,5 +146,6 @@ void amdgpu_irq_gpu_reset_resume_helper(struct amdgpu_device *adev); int amdgpu_irq_add_domain(struct amdgpu_device *adev); void amdgpu_irq_remove_domain(struct amdgpu_device *adev); unsigned amdgpu_irq_create_mapping(struct amdgpu_device *adev, unsigned src_id); +void amdgpu_restore_msix(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h index 4f3b7b5d9c1f..1d1c4b1ec7e7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h @@ -28,6 +28,8 @@ #ifndef __AMDGPU_ISP_H__ #define __AMDGPU_ISP_H__ +#include <linux/pm_domain.h> + #define ISP_REGS_OFFSET_END 0x629A4 struct amdgpu_isp; @@ -54,6 +56,7 @@ struct amdgpu_isp { struct isp_platform_data *isp_pdata; unsigned int harvest_config; const struct firmware *fw; + struct generic_pm_domain ispgpd; }; extern const struct amdgpu_ip_block_version isp_v4_1_0_ip_block; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c index 1e24590ae144..2b58e353cca1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c @@ -92,8 +92,7 @@ static enum drm_gpu_sched_stat amdgpu_job_timedout(struct drm_sched_job *s_job) struct drm_wedge_task_info *info = NULL; struct amdgpu_task_info *ti; struct amdgpu_device *adev = ring->adev; - int idx; - int r; + int idx, r; if (!drm_dev_enter(adev_to_drm(adev), &idx)) { dev_info(adev->dev, "%s - device unplugged skipping recovery on scheduler:%s", @@ -133,43 +132,20 @@ static enum drm_gpu_sched_stat amdgpu_job_timedout(struct drm_sched_job *s_job) if (unlikely(adev->debug_disable_gpu_ring_reset)) { dev_err(adev->dev, "Ring reset disabled by debug mask\n"); } else if (amdgpu_gpu_recovery && ring->funcs->reset) { - bool is_guilty; - - dev_err(adev->dev, "Starting %s ring reset\n", s_job->sched->name); - /* stop the scheduler, but don't mess with the - * bad job yet because if ring reset fails - * we'll fall back to full GPU reset. - */ - drm_sched_wqueue_stop(&ring->sched); - - /* for engine resets, we need to reset the engine, - * but individual queues may be unaffected. - * check here to make sure the accounting is correct. - */ - if (ring->funcs->is_guilty) - is_guilty = ring->funcs->is_guilty(ring); - else - is_guilty = true; - - if (is_guilty) - dma_fence_set_error(&s_job->s_fence->finished, -ETIME); - - r = amdgpu_ring_reset(ring, job->vmid); + dev_err(adev->dev, "Starting %s ring reset\n", + s_job->sched->name); + r = amdgpu_ring_reset(ring, job->vmid, NULL); if (!r) { - if (amdgpu_ring_sched_ready(ring)) - drm_sched_stop(&ring->sched, s_job); - if (is_guilty) { - atomic_inc(&ring->adev->gpu_reset_counter); - amdgpu_fence_driver_force_completion(ring); - } - if (amdgpu_ring_sched_ready(ring)) - drm_sched_start(&ring->sched, 0); - dev_err(adev->dev, "Ring %s reset succeeded\n", ring->sched.name); - drm_dev_wedged_event(adev_to_drm(adev), DRM_WEDGE_RECOVERY_NONE, info); + atomic_inc(&ring->adev->gpu_reset_counter); + dev_err(adev->dev, "Ring %s reset succeeded\n", + ring->sched.name); + drm_dev_wedged_event(adev_to_drm(adev), + DRM_WEDGE_RECOVERY_NONE, info); goto exit; } - dev_err(adev->dev, "Ring %s reset failure\n", ring->sched.name); + dev_err(adev->dev, "Ring %s reset failed\n", ring->sched.name); } + dma_fence_set_error(&s_job->s_fence->finished, -ETIME); amdgpu_vm_put_task_info(ti); @@ -275,8 +251,8 @@ void amdgpu_job_free_resources(struct amdgpu_job *job) /* Check if any fences where initialized */ if (job->base.s_fence && job->base.s_fence->finished.ops) f = &job->base.s_fence->finished; - else if (job->hw_fence.ops) - f = &job->hw_fence; + else if (job->hw_fence.base.ops) + f = &job->hw_fence.base; else f = NULL; @@ -293,10 +269,10 @@ static void amdgpu_job_free_cb(struct drm_sched_job *s_job) amdgpu_sync_free(&job->explicit_sync); /* only put the hw fence if has embedded fence */ - if (!job->hw_fence.ops) + if (!job->hw_fence.base.ops) kfree(job); else - dma_fence_put(&job->hw_fence); + dma_fence_put(&job->hw_fence.base); } void amdgpu_job_set_gang_leader(struct amdgpu_job *job, @@ -325,10 +301,10 @@ void amdgpu_job_free(struct amdgpu_job *job) if (job->gang_submit != &job->base.s_fence->scheduled) dma_fence_put(job->gang_submit); - if (!job->hw_fence.ops) + if (!job->hw_fence.base.ops) kfree(job); else - dma_fence_put(&job->hw_fence); + dma_fence_put(&job->hw_fence.base); } struct dma_fence *amdgpu_job_submit(struct amdgpu_job *job) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h index 529045d60412..2f302266662b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h @@ -48,7 +48,7 @@ struct amdgpu_job { struct drm_sched_job base; struct amdgpu_vm *vm; struct amdgpu_sync explicit_sync; - struct dma_fence hw_fence; + struct amdgpu_fence hw_fence; struct dma_fence *gang_submit; uint32_t preamble_status; uint32_t preemption_status; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c index dda29132dfb2..82d58ac7afb0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c @@ -463,7 +463,8 @@ int amdgpu_jpeg_reg_dump_init(struct amdgpu_device *adev, adev->jpeg.ip_dump = kcalloc(adev->jpeg.num_jpeg_inst * count, sizeof(uint32_t), GFP_KERNEL); if (!adev->jpeg.ip_dump) { - DRM_ERROR("Failed to allocate memory for JPEG IP Dump\n"); + dev_err(adev->dev, + "Failed to allocate memory for JPEG IP Dump\n"); return -ENOMEM; } adev->jpeg.reg_list = reg; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index d555853c5717..4c072451a8a1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -1503,6 +1503,9 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev, amdgpu_vm_bo_del(adev, fpriv->prt_va); amdgpu_bo_unreserve(pd); } + fpriv->evf_mgr.fd_closing = true; + amdgpu_eviction_fence_destroy(&fpriv->evf_mgr); + amdgpu_userq_mgr_fini(&fpriv->userq_mgr); amdgpu_ctx_mgr_fini(&fpriv->ctx_mgr); amdgpu_vm_fini(adev, &fpriv->vm); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c index 6fa9fa11c8f3..135598502c8d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c @@ -47,7 +47,7 @@ static int amdgpu_mes_doorbell_init(struct amdgpu_device *adev) /* Bitmap for dynamic allocation of kernel doorbells */ mes->doorbell_bitmap = bitmap_zalloc(PAGE_SIZE / sizeof(u32), GFP_KERNEL); if (!mes->doorbell_bitmap) { - DRM_ERROR("Failed to allocate MES doorbell bitmap\n"); + dev_err(adev->dev, "Failed to allocate MES doorbell bitmap\n"); return -ENOMEM; } @@ -256,7 +256,7 @@ int amdgpu_mes_suspend(struct amdgpu_device *adev) r = adev->mes.funcs->suspend_gang(&adev->mes, &input); amdgpu_mes_unlock(&adev->mes); if (r) - DRM_ERROR("failed to suspend all gangs"); + dev_err(adev->dev, "failed to suspend all gangs"); return r; } @@ -280,7 +280,7 @@ int amdgpu_mes_resume(struct amdgpu_device *adev) r = adev->mes.funcs->resume_gang(&adev->mes, &input); amdgpu_mes_unlock(&adev->mes); if (r) - DRM_ERROR("failed to resume all gangs"); + dev_err(adev->dev, "failed to resume all gangs"); return r; } @@ -304,7 +304,7 @@ int amdgpu_mes_map_legacy_queue(struct amdgpu_device *adev, r = adev->mes.funcs->map_legacy_queue(&adev->mes, &queue_input); amdgpu_mes_unlock(&adev->mes); if (r) - DRM_ERROR("failed to map legacy queue\n"); + dev_err(adev->dev, "failed to map legacy queue\n"); return r; } @@ -329,7 +329,7 @@ int amdgpu_mes_unmap_legacy_queue(struct amdgpu_device *adev, r = adev->mes.funcs->unmap_legacy_queue(&adev->mes, &queue_input); amdgpu_mes_unlock(&adev->mes); if (r) - DRM_ERROR("failed to unmap legacy queue\n"); + dev_err(adev->dev, "failed to unmap legacy queue\n"); return r; } @@ -361,7 +361,7 @@ int amdgpu_mes_reset_legacy_queue(struct amdgpu_device *adev, r = adev->mes.funcs->reset_hw_queue(&adev->mes, &queue_input); amdgpu_mes_unlock(&adev->mes); if (r) - DRM_ERROR("failed to reset legacy queue\n"); + dev_err(adev->dev, "failed to reset legacy queue\n"); return r; } @@ -469,7 +469,8 @@ int amdgpu_mes_set_shader_debugger(struct amdgpu_device *adev, int r; if (!adev->mes.funcs->misc_op) { - DRM_ERROR("mes set shader debugger is not supported!\n"); + dev_err(adev->dev, + "mes set shader debugger is not supported!\n"); return -EINVAL; } @@ -493,7 +494,7 @@ int amdgpu_mes_set_shader_debugger(struct amdgpu_device *adev, r = adev->mes.funcs->misc_op(&adev->mes, &op_input); if (r) - DRM_ERROR("failed to set_shader_debugger\n"); + dev_err(adev->dev, "failed to set_shader_debugger\n"); amdgpu_mes_unlock(&adev->mes); @@ -507,7 +508,8 @@ int amdgpu_mes_flush_shader_debugger(struct amdgpu_device *adev, int r; if (!adev->mes.funcs->misc_op) { - DRM_ERROR("mes flush shader debugger is not supported!\n"); + dev_err(adev->dev, + "mes flush shader debugger is not supported!\n"); return -EINVAL; } @@ -519,7 +521,7 @@ int amdgpu_mes_flush_shader_debugger(struct amdgpu_device *adev, r = adev->mes.funcs->misc_op(&adev->mes, &op_input); if (r) - DRM_ERROR("failed to set_shader_debugger\n"); + dev_err(adev->dev, "failed to set_shader_debugger\n"); amdgpu_mes_unlock(&adev->mes); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c index d085687a47ea..e56ba93a8df6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c @@ -53,6 +53,15 @@ u64 amdgpu_nbio_get_pcie_replay_count(struct amdgpu_device *adev) return 0; } +bool amdgpu_nbio_is_replay_cnt_supported(struct amdgpu_device *adev) +{ + if (amdgpu_sriov_vf(adev) || !adev->asic_funcs->get_pcie_replay_count || + (!adev->nbio.funcs || !adev->nbio.funcs->get_pcie_replay_count)) + return false; + + return true; +} + int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block) { int r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h index 79c2f807b9fe..b528de6a01f6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h @@ -119,4 +119,6 @@ int amdgpu_nbio_ras_sw_init(struct amdgpu_device *adev); int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block); u64 amdgpu_nbio_get_pcie_replay_count(struct amdgpu_device *adev); +bool amdgpu_nbio_is_replay_cnt_supported(struct amdgpu_device *adev); + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c index 73403744331a..c5fda18967c8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c @@ -32,6 +32,7 @@ #include <linux/list.h> #include <linux/slab.h> #include <linux/dma-buf.h> +#include <linux/export.h> #include <drm/drm_drv.h> #include <drm/amdgpu_drm.h> @@ -62,7 +63,7 @@ static void amdgpu_bo_destroy(struct ttm_buffer_object *tbo) amdgpu_bo_kunmap(bo); - if (bo->tbo.base.import_attach) + if (drm_gem_is_imported(&bo->tbo.base)) drm_prime_gem_destroy(&bo->tbo.base, bo->tbo.sg); drm_gem_object_release(&bo->tbo.base); amdgpu_bo_unref(&bo->parent); @@ -939,7 +940,7 @@ int amdgpu_bo_pin(struct amdgpu_bo *bo, u32 domain) domain = bo->preferred_domains & domain; /* A shared bo cannot be migrated to VRAM */ - if (bo->tbo.base.import_attach) { + if (drm_gem_is_imported(&bo->tbo.base)) { if (domain & AMDGPU_GEM_DOMAIN_GTT) domain = AMDGPU_GEM_DOMAIN_GTT; else @@ -967,7 +968,7 @@ int amdgpu_bo_pin(struct amdgpu_bo *bo, u32 domain) */ domain = amdgpu_bo_get_preferred_domain(adev, domain); - if (bo->tbo.base.import_attach) + if (drm_gem_is_imported(&bo->tbo.base)) dma_buf_pin(bo->tbo.base.import_attach); /* force to pin into visible video ram */ @@ -1018,7 +1019,7 @@ void amdgpu_bo_unpin(struct amdgpu_bo *bo) if (bo->tbo.pin_count) return; - if (bo->tbo.base.import_attach) + if (drm_gem_is_imported(&bo->tbo.base)) dma_buf_unpin(bo->tbo.base.import_attach); if (bo->tbo.resource->mem_type == TTM_PL_VRAM) { @@ -1263,7 +1264,7 @@ void amdgpu_bo_move_notify(struct ttm_buffer_object *bo, amdgpu_bo_kunmap(abo); - if (abo->tbo.base.dma_buf && !abo->tbo.base.import_attach && + if (abo->tbo.base.dma_buf && !drm_gem_is_imported(&abo->tbo.base) && old_mem && old_mem->mem_type != TTM_PL_SYSTEM) dma_buf_move_notify(abo->tbo.base.dma_buf); @@ -1473,6 +1474,26 @@ u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo) } /** + * amdgpu_bo_fb_aper_addr - return FB aperture GPU offset of the VRAM bo + * @bo: amdgpu VRAM buffer object for which we query the offset + * + * Returns: + * current FB aperture GPU offset of the object. + */ +u64 amdgpu_bo_fb_aper_addr(struct amdgpu_bo *bo) +{ + struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); + uint64_t offset, fb_base; + + WARN_ON_ONCE(bo->tbo.resource->mem_type != TTM_PL_VRAM); + + fb_base = adev->gmc.fb_start; + fb_base += adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size; + offset = (bo->tbo.resource->start << PAGE_SHIFT) + fb_base; + return amdgpu_gmc_sign_extend(offset); +} + +/** * amdgpu_bo_gpu_offset_no_check - return GPU offset of bo * @bo: amdgpu object for which we query the offset * diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h index 375448627f7b..c316920f3450 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h @@ -304,6 +304,7 @@ int amdgpu_bo_sync_wait_resv(struct amdgpu_device *adev, struct dma_resv *resv, bool intr); int amdgpu_bo_sync_wait(struct amdgpu_bo *bo, void *owner, bool intr); u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo); +u64 amdgpu_bo_fb_aper_addr(struct amdgpu_bo *bo); u64 amdgpu_bo_gpu_offset_no_check(struct amdgpu_bo *bo); uint32_t amdgpu_bo_mem_stats_placement(struct amdgpu_bo *bo); uint32_t amdgpu_bo_get_preferred_domain(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index e6f0b035e20b..a4bede28df17 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -252,6 +252,7 @@ static int psp_early_init(struct amdgpu_ip_block *ip_block) break; case IP_VERSION(14, 0, 2): case IP_VERSION(14, 0, 3): + adev->psp.sup_ifwi_up = !amdgpu_sriov_vf(adev); psp_v14_0_set_psp_funcs(psp); break; case IP_VERSION(14, 0, 5): @@ -596,6 +597,10 @@ int psp_wait_for(struct psp_context *psp, uint32_t reg_index, udelay(1); } + dev_err(adev->dev, + "psp reg (0x%x) wait timed out, mask: %x, read: %x exp: %x", + reg_index, mask, val, reg_val); + return -ETIME; } @@ -654,6 +659,10 @@ static const char *psp_gfx_cmd_name(enum psp_gfx_cmd_id cmd_id) return "BOOT_CFG"; case GFX_CMD_ID_CONFIG_SQ_PERFMON: return "CONFIG_SQ_PERFMON"; + case GFX_CMD_ID_FB_FW_RESERV_ADDR: + return "FB_FW_RESERV_ADDR"; + case GFX_CMD_ID_FB_FW_RESERV_EXT_ADDR: + return "FB_FW_RESERV_EXT_ADDR"; default: return "UNKNOWN CMD"; } @@ -871,6 +880,8 @@ static int psp_tmr_init(struct psp_context *psp) &psp->tmr_bo, &psp->tmr_mc_addr, pptr); } + if (amdgpu_virt_xgmi_migrate_enabled(psp->adev) && psp->tmr_bo) + psp->tmr_mc_addr = amdgpu_bo_fb_aper_addr(psp->tmr_bo); return ret; } @@ -984,6 +995,93 @@ int psp_get_fw_attestation_records_addr(struct psp_context *psp, return ret; } +static int psp_get_fw_reservation_info(struct psp_context *psp, + uint32_t cmd_id, + uint64_t *addr, + uint32_t *size) +{ + int ret; + uint32_t status; + struct psp_gfx_cmd_resp *cmd; + + cmd = acquire_psp_cmd_buf(psp); + + cmd->cmd_id = cmd_id; + + ret = psp_cmd_submit_buf(psp, NULL, cmd, + psp->fence_buf_mc_addr); + if (ret) { + release_psp_cmd_buf(psp); + return ret; + } + + status = cmd->resp.status; + if (status == PSP_ERR_UNKNOWN_COMMAND) { + release_psp_cmd_buf(psp); + *addr = 0; + *size = 0; + return 0; + } + + *addr = (uint64_t)cmd->resp.uresp.fw_reserve_info.reserve_base_address_hi << 32 | + cmd->resp.uresp.fw_reserve_info.reserve_base_address_lo; + *size = cmd->resp.uresp.fw_reserve_info.reserve_size; + + release_psp_cmd_buf(psp); + + return 0; +} + +int psp_update_fw_reservation(struct psp_context *psp) +{ + int ret; + uint64_t reserv_addr, reserv_addr_ext; + uint32_t reserv_size, reserv_size_ext; + struct amdgpu_device *adev = psp->adev; + + if (amdgpu_sriov_vf(psp->adev)) + return 0; + + if ((amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(14, 0, 2)) && + (amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(14, 0, 3))) + return 0; + + ret = psp_get_fw_reservation_info(psp, GFX_CMD_ID_FB_FW_RESERV_ADDR, &reserv_addr, &reserv_size); + if (ret) + return ret; + ret = psp_get_fw_reservation_info(psp, GFX_CMD_ID_FB_FW_RESERV_EXT_ADDR, &reserv_addr_ext, &reserv_size_ext); + if (ret) + return ret; + + if (reserv_addr != adev->gmc.real_vram_size - reserv_size) { + dev_warn(adev->dev, "reserve fw region is not valid!\n"); + return 0; + } + + amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL, NULL); + + reserv_size = roundup(reserv_size, SZ_1M); + + ret = amdgpu_bo_create_kernel_at(adev, reserv_addr, reserv_size, &adev->mman.fw_reserved_memory, NULL); + if (ret) { + dev_err(adev->dev, "reserve fw region failed(%d)!\n", ret); + amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL, NULL); + return ret; + } + + reserv_size_ext = roundup(reserv_size_ext, SZ_1M); + + ret = amdgpu_bo_create_kernel_at(adev, reserv_addr_ext, reserv_size_ext, + &adev->mman.fw_reserved_memory_extend, NULL); + if (ret) { + dev_err(adev->dev, "reserve extend fw region failed(%d)!\n", ret); + amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory_extend, NULL, NULL); + return ret; + } + + return 0; +} + static int psp_boot_config_get(struct amdgpu_device *adev, uint32_t *boot_cfg) { struct psp_context *psp = &adev->psp; @@ -1270,6 +1368,11 @@ int psp_ta_load(struct psp_context *psp, struct ta_context *context) psp_copy_fw(psp, context->bin_desc.start_addr, context->bin_desc.size_bytes); + if (amdgpu_virt_xgmi_migrate_enabled(psp->adev) && + context->mem_context.shared_bo) + context->mem_context.shared_mc_addr = + amdgpu_bo_fb_aper_addr(context->mem_context.shared_bo); + psp_prep_ta_load_cmd_buf(cmd, psp->fw_pri_mc_addr, context); ret = psp_cmd_submit_buf(psp, NULL, cmd, @@ -2337,11 +2440,27 @@ bool amdgpu_psp_tos_reload_needed(struct amdgpu_device *adev) return false; } +static void psp_update_gpu_addresses(struct amdgpu_device *adev) +{ + struct psp_context *psp = &adev->psp; + + if (psp->cmd_buf_bo && psp->cmd_buf_mem) { + psp->fw_pri_mc_addr = amdgpu_bo_fb_aper_addr(psp->fw_pri_bo); + psp->fence_buf_mc_addr = amdgpu_bo_fb_aper_addr(psp->fence_buf_bo); + psp->cmd_buf_mc_addr = amdgpu_bo_fb_aper_addr(psp->cmd_buf_bo); + } + if (adev->firmware.rbuf && psp->km_ring.ring_mem) + psp->km_ring.ring_mem_mc_addr = amdgpu_bo_fb_aper_addr(adev->firmware.rbuf); +} + static int psp_hw_start(struct psp_context *psp) { struct amdgpu_device *adev = psp->adev; int ret; + if (amdgpu_virt_xgmi_migrate_enabled(adev)) + psp_update_gpu_addresses(adev); + if (!amdgpu_sriov_vf(adev)) { if ((is_psp_fw_valid(psp->kdb)) && (psp->funcs->bootloader_load_kdb != NULL)) { @@ -2440,6 +2559,14 @@ static int psp_hw_start(struct psp_context *psp) return ret; } + if (!amdgpu_in_reset(adev) && !adev->in_suspend) { + ret = psp_update_fw_reservation(psp); + if (ret) { + dev_err(adev->dev, "update fw reservation failed!\n"); + return ret; + } + } + if (amdgpu_sriov_vf(adev) && amdgpu_in_reset(adev)) goto skip_pin_bo; @@ -3522,8 +3649,12 @@ int psp_init_sos_microcode(struct psp_context *psp, const char *chip_name) uint8_t *ucode_array_start_addr; int err = 0; - err = amdgpu_ucode_request(adev, &adev->psp.sos_fw, AMDGPU_UCODE_REQUIRED, - "amdgpu/%s_sos.bin", chip_name); + if (amdgpu_is_kicker_fw(adev)) + err = amdgpu_ucode_request(adev, &adev->psp.sos_fw, AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_sos_kicker.bin", chip_name); + else + err = amdgpu_ucode_request(adev, &adev->psp.sos_fw, AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_sos.bin", chip_name); if (err) goto out; @@ -3799,8 +3930,12 @@ int psp_init_ta_microcode(struct psp_context *psp, const char *chip_name) struct amdgpu_device *adev = psp->adev; int err; - err = amdgpu_ucode_request(adev, &adev->psp.ta_fw, AMDGPU_UCODE_REQUIRED, - "amdgpu/%s_ta.bin", chip_name); + if (amdgpu_is_kicker_fw(adev)) + err = amdgpu_ucode_request(adev, &adev->psp.ta_fw, AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_ta_kicker.bin", chip_name); + else + err = amdgpu_ucode_request(adev, &adev->psp.ta_fw, AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_ta.bin", chip_name); if (err) return err; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h index 428adc7f741d..4bc0ec49d2e9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h @@ -51,6 +51,17 @@ #define C2PMSG_CMD_SPI_GET_ROM_IMAGE_ADDR_HI 0x10 #define C2PMSG_CMD_SPI_GET_FLASH_IMAGE 0x11 +/* Command register bit 31 set to indicate readiness */ +#define MBOX_TOS_READY_FLAG (GFX_FLAG_RESPONSE) +#define MBOX_TOS_READY_MASK (GFX_CMD_RESPONSE_MASK | GFX_CMD_STATUS_MASK) + +/* Values to check for a successful GFX_CMD response wait. Check against + * both status bits and response state - helps to detect a command failure + * or other unexpected cases like a device drop reading all 0xFFs + */ +#define MBOX_TOS_RESP_FLAG (GFX_FLAG_RESPONSE) +#define MBOX_TOS_RESP_MASK (GFX_CMD_RESPONSE_MASK | GFX_CMD_STATUS_MASK) + extern const struct attribute_group amdgpu_flash_attr_group; enum psp_shared_mem_size { @@ -588,7 +599,7 @@ int psp_init_cap_microcode(struct psp_context *psp, const char *chip_name); int psp_get_fw_attestation_records_addr(struct psp_context *psp, uint64_t *output_ptr); - +int psp_update_fw_reservation(struct psp_context *psp); int psp_load_fw_list(struct psp_context *psp, struct amdgpu_firmware_info **ucode_list, int ucode_count); void psp_copy_fw(struct psp_context *psp, uint8_t *start_addr, uint32_t bin_size); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index de0944947eaf..c508697a3412 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1107,6 +1107,9 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev, err_info->de_count, blk_name); } } else { + if (adev->debug_disable_ce_logs) + return; + for_each_ras_error(err_node, err_data) { err_info = &err_node->err_info; mcm_info = &err_info->mcm_info; @@ -3003,6 +3006,15 @@ int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev, return 0; } + if (!con->eeprom_control.is_eeprom_valid) { + dev_warn(adev->dev, + "Failed to save EEPROM table data because of EEPROM data corruption!"); + if (new_cnt) + *new_cnt = 0; + + return 0; + } + mutex_lock(&con->recovery_lock); control = &con->eeprom_control; data = con->eh_data; @@ -3294,7 +3306,6 @@ static int amdgpu_ras_poison_creation_handler(struct amdgpu_device *adev, uint64_t de_queried_count; uint32_t new_detect_count, total_detect_count; uint32_t need_query_count = poison_creation_count; - bool query_data_timeout = false; enum ras_event_type type = RAS_EVENT_TYPE_POISON_CREATION; memset(&info, 0, sizeof(info)); @@ -3323,21 +3334,13 @@ static int amdgpu_ras_poison_creation_handler(struct amdgpu_device *adev, timeout = MAX_UMC_POISON_POLLING_TIME_ASYNC; if (timeout) { - if (!--timeout) { - query_data_timeout = true; + if (!--timeout) break; - } msleep(1); } } } while (total_detect_count < need_query_count); - if (query_data_timeout) { - dev_warn(adev->dev, "Can't find deferred error! count: %u\n", - (need_query_count - total_detect_count)); - return -ENOENT; - } - if (total_detect_count) schedule_delayed_work(&ras->page_retirement_dwork, 0); @@ -3488,8 +3491,7 @@ int amdgpu_ras_init_badpage_info(struct amdgpu_device *adev) control = &con->eeprom_control; ret = amdgpu_ras_eeprom_init(control); - if (ret) - return ret; + control->is_eeprom_valid = !ret; if (!adev->umc.ras || !adev->umc.ras->convert_ras_err_addr) control->ras_num_pa_recs = control->ras_num_recs; @@ -3498,10 +3500,12 @@ int amdgpu_ras_init_badpage_info(struct amdgpu_device *adev) adev->umc.ras->get_retire_flip_bits) adev->umc.ras->get_retire_flip_bits(adev); - if (control->ras_num_recs) { + if (control->ras_num_recs && control->is_eeprom_valid) { ret = amdgpu_ras_load_bad_pages(adev); - if (ret) - return ret; + if (ret) { + control->is_eeprom_valid = false; + return 0; + } amdgpu_dpm_send_hbm_bad_pages_num( adev, control->ras_num_bad_pages); @@ -3520,7 +3524,7 @@ int amdgpu_ras_init_badpage_info(struct amdgpu_device *adev) dev_warn(adev->dev, "Failed to format RAS EEPROM data in V3 version!\n"); } - return ret; + return 0; } int amdgpu_ras_recovery_init(struct amdgpu_device *adev, bool init_bp_info) @@ -4414,8 +4418,10 @@ void amdgpu_ras_clear_err_state(struct amdgpu_device *adev) struct amdgpu_ras *ras; ras = amdgpu_ras_get_context(adev); - if (ras) + if (ras) { ras->ras_err_state = 0; + ras->gpu_reset_flags = 0; + } } void amdgpu_ras_set_err_poison(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c index 2c58e09e56f9..54838746f97d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c @@ -277,10 +277,11 @@ static int __write_table_header(struct amdgpu_ras_eeprom_control *control) up_read(&adev->reset_domain->sem); if (res < 0) { - DRM_ERROR("Failed to write EEPROM table header:%d", res); + dev_err(adev->dev, "Failed to write EEPROM table header:%d", + res); } else if (res < RAS_TABLE_HEADER_SIZE) { - DRM_ERROR("Short write:%d out of %d\n", - res, RAS_TABLE_HEADER_SIZE); + dev_err(adev->dev, "Short write:%d out of %d\n", res, + RAS_TABLE_HEADER_SIZE); res = -EIO; } else { res = 0; @@ -323,7 +324,8 @@ static int __write_table_ras_info(struct amdgpu_ras_eeprom_control *control) buf = kzalloc(RAS_TABLE_V2_1_INFO_SIZE, GFP_KERNEL); if (!buf) { - DRM_ERROR("Failed to alloc buf to write table ras info\n"); + dev_err(adev->dev, + "Failed to alloc buf to write table ras info\n"); return -ENOMEM; } @@ -338,10 +340,11 @@ static int __write_table_ras_info(struct amdgpu_ras_eeprom_control *control) up_read(&adev->reset_domain->sem); if (res < 0) { - DRM_ERROR("Failed to write EEPROM table ras info:%d", res); + dev_err(adev->dev, "Failed to write EEPROM table ras info:%d", + res); } else if (res < RAS_TABLE_V2_1_INFO_SIZE) { - DRM_ERROR("Short write:%d out of %d\n", - res, RAS_TABLE_V2_1_INFO_SIZE); + dev_err(adev->dev, "Short write:%d out of %d\n", res, + RAS_TABLE_V2_1_INFO_SIZE); res = -EIO; } else { res = 0; @@ -476,6 +479,8 @@ int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control) control->ras_num_recs = 0; control->ras_num_bad_pages = 0; + control->ras_num_mca_recs = 0; + control->ras_num_pa_recs = 0; control->ras_fri = 0; amdgpu_dpm_send_hbm_bad_pages_num(adev, control->ras_num_bad_pages); @@ -607,13 +612,13 @@ static int __amdgpu_ras_eeprom_write(struct amdgpu_ras_eeprom_control *control, buf, buf_size); up_read(&adev->reset_domain->sem); if (res < 0) { - DRM_ERROR("Writing %d EEPROM table records error:%d", - num, res); + dev_err(adev->dev, "Writing %d EEPROM table records error:%d", + num, res); } else if (res < buf_size) { /* Short write, return error. */ - DRM_ERROR("Wrote %d records out of %d", - res / RAS_TABLE_RECORD_SIZE, num); + dev_err(adev->dev, "Wrote %d records out of %d", + res / RAS_TABLE_RECORD_SIZE, num); res = -EIO; } else { res = 0; @@ -761,18 +766,17 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control) dev_warn(adev->dev, "Saved bad pages %d reaches threshold value %d\n", control->ras_num_bad_pages, ras->bad_page_cnt_threshold); - control->tbl_hdr.header = RAS_TABLE_HDR_BAD; - if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1) { - control->tbl_rai.rma_status = GPU_RETIRED__ECC_REACH_THRESHOLD; - control->tbl_rai.health_percent = 0; - } - if ((amdgpu_bad_page_threshold != -1) && - (amdgpu_bad_page_threshold != -2)) + (amdgpu_bad_page_threshold != -2)) { + control->tbl_hdr.header = RAS_TABLE_HDR_BAD; + if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1) { + control->tbl_rai.rma_status = GPU_RETIRED__ECC_REACH_THRESHOLD; + control->tbl_rai.health_percent = 0; + } ras->is_rma = true; - - /* ignore the -ENOTSUPP return value */ - amdgpu_dpm_send_rma_reason(adev); + /* ignore the -ENOTSUPP return value */ + amdgpu_dpm_send_rma_reason(adev); + } } if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1) @@ -787,8 +791,9 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control) buf_size = control->ras_num_recs * RAS_TABLE_RECORD_SIZE; buf = kcalloc(control->ras_num_recs, RAS_TABLE_RECORD_SIZE, GFP_KERNEL); if (!buf) { - DRM_ERROR("allocating memory for table of size %d bytes failed\n", - control->tbl_hdr.tbl_size); + dev_err(adev->dev, + "allocating memory for table of size %d bytes failed\n", + control->tbl_hdr.tbl_size); res = -ENOMEM; goto Out; } @@ -800,12 +805,11 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control) buf, buf_size); up_read(&adev->reset_domain->sem); if (res < 0) { - DRM_ERROR("EEPROM failed reading records:%d\n", - res); + dev_err(adev->dev, "EEPROM failed reading records:%d\n", res); goto Out; } else if (res < buf_size) { - DRM_ERROR("EEPROM read %d out of %d bytes\n", - res, buf_size); + dev_err(adev->dev, "EEPROM read %d out of %d bytes\n", res, + buf_size); res = -EIO; goto Out; } @@ -866,11 +870,12 @@ int amdgpu_ras_eeprom_append(struct amdgpu_ras_eeprom_control *control, return 0; if (num == 0) { - DRM_ERROR("will not append 0 records\n"); + dev_err(adev->dev, "will not append 0 records\n"); return -EINVAL; } else if (num > control->ras_max_record_count) { - DRM_ERROR("cannot append %d records than the size of table %d\n", - num, control->ras_max_record_count); + dev_err(adev->dev, + "cannot append %d records than the size of table %d\n", + num, control->ras_max_record_count); return -EINVAL; } @@ -924,13 +929,13 @@ static int __amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control, buf, buf_size); up_read(&adev->reset_domain->sem); if (res < 0) { - DRM_ERROR("Reading %d EEPROM table records error:%d", - num, res); + dev_err(adev->dev, "Reading %d EEPROM table records error:%d", + num, res); } else if (res < buf_size) { /* Short read, return error. */ - DRM_ERROR("Read %d records out of %d", - res / RAS_TABLE_RECORD_SIZE, num); + dev_err(adev->dev, "Read %d records out of %d", + res / RAS_TABLE_RECORD_SIZE, num); res = -EIO; } else { res = 0; @@ -964,11 +969,11 @@ int amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control, return 0; if (num == 0) { - DRM_ERROR("will not read 0 records\n"); + dev_err(adev->dev, "will not read 0 records\n"); return -EINVAL; } else if (num > control->ras_num_recs) { - DRM_ERROR("too many records to read:%d available:%d\n", - num, control->ras_num_recs); + dev_err(adev->dev, "too many records to read:%d available:%d\n", + num, control->ras_num_recs); return -EINVAL; } @@ -1300,7 +1305,8 @@ static int __verify_ras_table_checksum(struct amdgpu_ras_eeprom_control *control buf = kzalloc(buf_size, GFP_KERNEL); if (!buf) { - DRM_ERROR("Out of memory checking RAS table checksum.\n"); + dev_err(adev->dev, + "Out of memory checking RAS table checksum.\n"); return -ENOMEM; } @@ -1309,7 +1315,7 @@ static int __verify_ras_table_checksum(struct amdgpu_ras_eeprom_control *control control->ras_header_offset, buf, buf_size); if (res < buf_size) { - DRM_ERROR("Partial read for checksum, res:%d\n", res); + dev_err(adev->dev, "Partial read for checksum, res:%d\n", res); /* On partial reads, return -EIO. */ if (res >= 0) @@ -1334,7 +1340,8 @@ static int __read_table_ras_info(struct amdgpu_ras_eeprom_control *control) buf = kzalloc(RAS_TABLE_V2_1_INFO_SIZE, GFP_KERNEL); if (!buf) { - DRM_ERROR("Failed to alloc buf to read EEPROM table ras info\n"); + dev_err(adev->dev, + "Failed to alloc buf to read EEPROM table ras info\n"); return -ENOMEM; } @@ -1346,7 +1353,8 @@ static int __read_table_ras_info(struct amdgpu_ras_eeprom_control *control) control->i2c_address + control->ras_info_offset, buf, RAS_TABLE_V2_1_INFO_SIZE); if (res < RAS_TABLE_V2_1_INFO_SIZE) { - DRM_ERROR("Failed to read EEPROM table ras info, res:%d", res); + dev_err(adev->dev, + "Failed to read EEPROM table ras info, res:%d", res); res = res >= 0 ? -EIO : res; goto Out; } @@ -1387,7 +1395,8 @@ int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control) control->i2c_address + control->ras_header_offset, buf, RAS_TABLE_HEADER_SIZE); if (res < RAS_TABLE_HEADER_SIZE) { - DRM_ERROR("Failed to read EEPROM table header, res:%d", res); + dev_err(adev->dev, "Failed to read EEPROM table header, res:%d", + res); return res >= 0 ? -EIO : res; } @@ -1452,8 +1461,9 @@ int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control) control->ras_num_mca_recs * adev->umc.retire_unit; if (hdr->header == RAS_TABLE_HDR_VAL) { - DRM_DEBUG_DRIVER("Found existing EEPROM table with %d records", - control->ras_num_bad_pages); + dev_dbg(adev->dev, + "Found existing EEPROM table with %d records", + control->ras_num_bad_pages); if (hdr->version >= RAS_TABLE_VER_V2_1) { res = __read_table_ras_info(control); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h index ec6d7ea37ad0..35c69ac3dbeb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h @@ -114,6 +114,8 @@ struct amdgpu_ras_eeprom_control { /* Record channel info which occurred bad pages */ u32 bad_channel_bitmap; + + bool is_eeprom_valid; }; /* diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h index b95b47110769..784ba2ec354c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h @@ -127,6 +127,22 @@ struct amdgpu_fence_driver { struct dma_fence **fences; }; +/* + * Fences mark an event in the GPUs pipeline and are used + * for GPU/CPU synchronization. When the fence is written, + * it is expected that all buffers associated with that fence + * are no longer in use by the associated ring on the GPU and + * that the relevant GPU caches have been flushed. + */ + +struct amdgpu_fence { + struct dma_fence base; + + /* RB, DMA, etc. */ + struct amdgpu_ring *ring; + ktime_t start_timestamp; +}; + extern const struct drm_sched_backend_ops amdgpu_sched_ops; void amdgpu_fence_driver_clear_job_fences(struct amdgpu_ring *ring); @@ -141,8 +157,8 @@ void amdgpu_fence_driver_hw_init(struct amdgpu_device *adev); void amdgpu_fence_driver_hw_fini(struct amdgpu_device *adev); int amdgpu_fence_driver_sw_init(struct amdgpu_device *adev); void amdgpu_fence_driver_sw_fini(struct amdgpu_device *adev); -int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **fence, struct amdgpu_job *job, - unsigned flags); +int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f, + struct amdgpu_fence *af, unsigned int flags); int amdgpu_fence_emit_polling(struct amdgpu_ring *ring, uint32_t *s, uint32_t timeout); bool amdgpu_fence_process(struct amdgpu_ring *ring); @@ -252,9 +268,9 @@ struct amdgpu_ring_funcs { void (*patch_cntl)(struct amdgpu_ring *ring, unsigned offset); void (*patch_ce)(struct amdgpu_ring *ring, unsigned offset); void (*patch_de)(struct amdgpu_ring *ring, unsigned offset); - int (*reset)(struct amdgpu_ring *ring, unsigned int vmid); + int (*reset)(struct amdgpu_ring *ring, unsigned int vmid, + struct amdgpu_fence *timedout_fence); void (*emit_cleaner_shader)(struct amdgpu_ring *ring); - bool (*is_guilty)(struct amdgpu_ring *ring); }; /** @@ -409,7 +425,7 @@ struct amdgpu_ring { #define amdgpu_ring_patch_cntl(r, o) ((r)->funcs->patch_cntl((r), (o))) #define amdgpu_ring_patch_ce(r, o) ((r)->funcs->patch_ce((r), (o))) #define amdgpu_ring_patch_de(r, o) ((r)->funcs->patch_de((r), (o))) -#define amdgpu_ring_reset(r, v) (r)->funcs->reset((r), (v)) +#define amdgpu_ring_reset(r, v, f) (r)->funcs->reset((r), (v), (f)) unsigned int amdgpu_ring_max_ibs(enum amdgpu_ring_type type); int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c index 6716ac281c49..7e26a44dcc1f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c @@ -534,41 +534,17 @@ bool amdgpu_sdma_is_shared_inv_eng(struct amdgpu_device *adev, struct amdgpu_rin static int amdgpu_sdma_soft_reset(struct amdgpu_device *adev, u32 instance_id) { struct amdgpu_sdma_instance *sdma_instance = &adev->sdma.instance[instance_id]; - int r = -EOPNOTSUPP; - switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { - case IP_VERSION(4, 4, 2): - case IP_VERSION(4, 4, 4): - case IP_VERSION(4, 4, 5): - /* For SDMA 4.x, use the existing DPM interface for backward compatibility */ - r = amdgpu_dpm_reset_sdma(adev, 1 << instance_id); - break; - case IP_VERSION(5, 0, 0): - case IP_VERSION(5, 0, 1): - case IP_VERSION(5, 0, 2): - case IP_VERSION(5, 0, 5): - case IP_VERSION(5, 2, 0): - case IP_VERSION(5, 2, 2): - case IP_VERSION(5, 2, 4): - case IP_VERSION(5, 2, 5): - case IP_VERSION(5, 2, 6): - case IP_VERSION(5, 2, 3): - case IP_VERSION(5, 2, 1): - case IP_VERSION(5, 2, 7): - if (sdma_instance->funcs->soft_reset_kernel_queue) - r = sdma_instance->funcs->soft_reset_kernel_queue(adev, instance_id); - break; - default: - break; - } + if (sdma_instance->funcs->soft_reset_kernel_queue) + return sdma_instance->funcs->soft_reset_kernel_queue(adev, instance_id); - return r; + return -EOPNOTSUPP; } /** * amdgpu_sdma_reset_engine - Reset a specific SDMA engine * @adev: Pointer to the AMDGPU device - * @instance_id: ID of the SDMA engine instance to reset + * @instance_id: Logical ID of the SDMA engine instance to reset * * Returns: 0 on success, or a negative error code on failure. */ @@ -578,35 +554,35 @@ int amdgpu_sdma_reset_engine(struct amdgpu_device *adev, uint32_t instance_id) struct amdgpu_sdma_instance *sdma_instance = &adev->sdma.instance[instance_id]; struct amdgpu_ring *gfx_ring = &sdma_instance->ring; struct amdgpu_ring *page_ring = &sdma_instance->page; - bool gfx_sched_stopped = false, page_sched_stopped = false; mutex_lock(&sdma_instance->engine_reset_mutex); /* Stop the scheduler's work queue for the GFX and page rings if they are running. * This ensures that no new tasks are submitted to the queues while * the reset is in progress. */ - if (!amdgpu_ring_sched_ready(gfx_ring)) { - drm_sched_wqueue_stop(&gfx_ring->sched); - gfx_sched_stopped = true; - } + drm_sched_wqueue_stop(&gfx_ring->sched); - if (adev->sdma.has_page_queue && !amdgpu_ring_sched_ready(page_ring)) { + if (adev->sdma.has_page_queue) drm_sched_wqueue_stop(&page_ring->sched); - page_sched_stopped = true; - } - if (sdma_instance->funcs->stop_kernel_queue) + if (sdma_instance->funcs->stop_kernel_queue) { sdma_instance->funcs->stop_kernel_queue(gfx_ring); + if (adev->sdma.has_page_queue) + sdma_instance->funcs->stop_kernel_queue(page_ring); + } /* Perform the SDMA reset for the specified instance */ ret = amdgpu_sdma_soft_reset(adev, instance_id); if (ret) { - dev_err(adev->dev, "Failed to reset SDMA instance %u\n", instance_id); + dev_err(adev->dev, "Failed to reset SDMA logical instance %u\n", instance_id); goto exit; } - if (sdma_instance->funcs->start_kernel_queue) + if (sdma_instance->funcs->start_kernel_queue) { sdma_instance->funcs->start_kernel_queue(gfx_ring); + if (adev->sdma.has_page_queue) + sdma_instance->funcs->start_kernel_queue(page_ring); + } exit: /* Restart the scheduler's work queue for the GFX and page rings @@ -614,12 +590,9 @@ exit: * to be submitted to the queues after the reset is complete. */ if (!ret) { - if (gfx_sched_stopped && amdgpu_ring_sched_ready(gfx_ring)) { - drm_sched_wqueue_start(&gfx_ring->sched); - } - if (page_sched_stopped && amdgpu_ring_sched_ready(page_ring)) { + drm_sched_wqueue_start(&gfx_ring->sched); + if (adev->sdma.has_page_queue) drm_sched_wqueue_start(&page_ring->sched); - } } mutex_unlock(&sdma_instance->engine_reset_mutex); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index 9c5df35f05b7..27ab4e754b2a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -299,7 +299,8 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, struct amdgpu_bo *abo_src, *abo_dst; if (!adev->mman.buffer_funcs_enabled) { - DRM_ERROR("Trying to move memory with ring turned off.\n"); + dev_err(adev->dev, + "Trying to move memory with ring turned off.\n"); return -EINVAL; } @@ -934,7 +935,7 @@ static int amdgpu_ttm_backend_bind(struct ttm_device *bdev, if (gtt->userptr) { r = amdgpu_ttm_tt_pin_userptr(bdev, ttm); if (r) { - DRM_ERROR("failed to pin userptr\n"); + dev_err(adev->dev, "failed to pin userptr\n"); return r; } } else if (ttm->page_flags & TTM_TT_FLAG_EXTERNAL) { @@ -1060,7 +1061,7 @@ static void amdgpu_ttm_backend_unbind(struct ttm_device *bdev, /* if the pages have userptr pinning then clear that first */ if (gtt->userptr) { amdgpu_ttm_tt_unpin_userptr(bdev, ttm); - } else if (ttm->sg && gtt->gobj->import_attach) { + } else if (ttm->sg && drm_gem_is_imported(gtt->gobj)) { struct dma_buf_attachment *attach; attach = gtt->gobj->import_attach; @@ -1781,7 +1782,7 @@ static int amdgpu_ttm_reserve_tmr(struct amdgpu_device *adev) &ctx->c2p_bo, NULL); if (ret) { - DRM_ERROR("alloc c2p_bo failed(%d)!\n", ret); + dev_err(adev->dev, "alloc c2p_bo failed(%d)!\n", ret); amdgpu_ttm_training_reserve_vram_fini(adev); return ret; } @@ -1793,7 +1794,7 @@ static int amdgpu_ttm_reserve_tmr(struct amdgpu_device *adev) adev, adev->gmc.real_vram_size - reserve_size, reserve_size, &adev->mman.fw_reserved_memory, NULL); if (ret) { - DRM_ERROR("alloc tmr failed(%d)!\n", ret); + dev_err(adev->dev, "alloc tmr failed(%d)!\n", ret); amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL, NULL); return ret; @@ -1864,13 +1865,14 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) adev->need_swiotlb, dma_addressing_limited(adev->dev)); if (r) { - DRM_ERROR("failed initializing buffer object driver(%d).\n", r); + dev_err(adev->dev, + "failed initializing buffer object driver(%d).\n", r); return r; } r = amdgpu_ttm_pools_init(adev); if (r) { - DRM_ERROR("failed to init ttm pools(%d).\n", r); + dev_err(adev->dev, "failed to init ttm pools(%d).\n", r); return r; } adev->mman.initialized = true; @@ -1878,7 +1880,7 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) /* Initialize VRAM pool with all of VRAM divided into pages */ r = amdgpu_vram_mgr_init(adev); if (r) { - DRM_ERROR("Failed initializing VRAM heap.\n"); + dev_err(adev->dev, "Failed initializing VRAM heap.\n"); return r; } @@ -1958,7 +1960,7 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) DRM_DEBUG_DRIVER("Skipped stolen memory reservation\n"); } - DRM_INFO("amdgpu: %uM of VRAM memory ready\n", + dev_info(adev->dev, "amdgpu: %uM of VRAM memory ready\n", (unsigned int)(adev->gmc.real_vram_size / (1024 * 1024))); /* Compute GTT size, either based on TTM limit @@ -1981,10 +1983,10 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) /* Initialize GTT memory pool */ r = amdgpu_gtt_mgr_init(adev, gtt_size); if (r) { - DRM_ERROR("Failed initializing GTT heap.\n"); + dev_err(adev->dev, "Failed initializing GTT heap.\n"); return r; } - DRM_INFO("amdgpu: %uM of GTT memory ready.\n", + dev_info(adev->dev, "amdgpu: %uM of GTT memory ready.\n", (unsigned int)(gtt_size / (1024 * 1024))); if (adev->flags & AMD_IS_APU) { @@ -1995,40 +1997,40 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) /* Initialize doorbell pool on PCI BAR */ r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_DOORBELL, adev->doorbell.size / PAGE_SIZE); if (r) { - DRM_ERROR("Failed initializing doorbell heap.\n"); + dev_err(adev->dev, "Failed initializing doorbell heap.\n"); return r; } /* Create a boorbell page for kernel usages */ r = amdgpu_doorbell_create_kernel_doorbells(adev); if (r) { - DRM_ERROR("Failed to initialize kernel doorbells.\n"); + dev_err(adev->dev, "Failed to initialize kernel doorbells.\n"); return r; } /* Initialize preemptible memory pool */ r = amdgpu_preempt_mgr_init(adev); if (r) { - DRM_ERROR("Failed initializing PREEMPT heap.\n"); + dev_err(adev->dev, "Failed initializing PREEMPT heap.\n"); return r; } /* Initialize various on-chip memory pools */ r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_GDS, adev->gds.gds_size); if (r) { - DRM_ERROR("Failed initializing GDS heap.\n"); + dev_err(adev->dev, "Failed initializing GDS heap.\n"); return r; } r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_GWS, adev->gds.gws_size); if (r) { - DRM_ERROR("Failed initializing gws heap.\n"); + dev_err(adev->dev, "Failed initializing gws heap.\n"); return r; } r = amdgpu_ttm_init_on_chip(adev, AMDGPU_PL_OA, adev->gds.oa_size); if (r) { - DRM_ERROR("Failed initializing oa heap.\n"); + dev_err(adev->dev, "Failed initializing oa heap.\n"); return r; } if (amdgpu_bo_create_kernel(adev, PAGE_SIZE, PAGE_SIZE, @@ -2060,6 +2062,8 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev) /* return the FW reserved memory back to VRAM */ amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL, NULL); + amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory_extend, NULL, + NULL); if (adev->mman.stolen_reserved_size) amdgpu_bo_free_kernel(&adev->mman.stolen_reserved_memory, NULL, NULL); @@ -2089,7 +2093,7 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev) ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_DOORBELL); ttm_device_fini(&adev->mman.bdev); adev->mman.initialized = false; - DRM_INFO("amdgpu: ttm finalized\n"); + dev_info(adev->dev, "amdgpu: ttm finalized\n"); } /** @@ -2121,8 +2125,9 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable) DRM_SCHED_PRIORITY_KERNEL, &sched, 1, NULL); if (r) { - DRM_ERROR("Failed setting up TTM BO move entity (%d)\n", - r); + dev_err(adev->dev, + "Failed setting up TTM BO move entity (%d)\n", + r); return; } @@ -2130,8 +2135,9 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable) DRM_SCHED_PRIORITY_NORMAL, &sched, 1, NULL); if (r) { - DRM_ERROR("Failed setting up TTM BO move entity (%d)\n", - r); + dev_err(adev->dev, + "Failed setting up TTM BO move entity (%d)\n", + r); goto error_free_entity; } } else { @@ -2202,7 +2208,8 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset, int r; if (!direct_submit && !ring->sched.ready) { - DRM_ERROR("Trying to move memory with ring turned off.\n"); + dev_err(adev->dev, + "Trying to move memory with ring turned off.\n"); return -EINVAL; } @@ -2237,7 +2244,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset, error_free: amdgpu_job_free(job); - DRM_ERROR("Error scheduling IBs (%d)\n", r); + dev_err(adev->dev, "Error scheduling IBs (%d)\n", r); return r; } @@ -2356,7 +2363,8 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo, int r; if (!adev->mman.buffer_funcs_enabled) { - DRM_ERROR("Trying to clear memory with ring turned off.\n"); + dev_err(adev->dev, + "Trying to clear memory with ring turned off.\n"); return -EINVAL; } @@ -2416,7 +2424,7 @@ int amdgpu_ttm_evict_resources(struct amdgpu_device *adev, int mem_type) man = ttm_manager_type(&adev->mman.bdev, mem_type); break; default: - DRM_ERROR("Trying to evict invalid memory type\n"); + dev_err(adev->dev, "Trying to evict invalid memory type\n"); return -EINVAL; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h index 208b7d1d8a27..215c198e4aff 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h @@ -86,6 +86,7 @@ struct amdgpu_mman { uint32_t discovery_tmr_size; /* fw reserved memory */ struct amdgpu_bo *fw_reserved_memory; + struct amdgpu_bo *fw_reserved_memory_extend; /* firmware VRAM reservation */ u64 fw_vram_usage_start_offset; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c index 2505c46a9c3d..a0b50a8ac9c4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c @@ -30,6 +30,10 @@ #define AMDGPU_UCODE_NAME_MAX (128) +static const struct kicker_device kicker_device_list[] = { + {0x744B, 0x00}, +}; + static void amdgpu_ucode_print_common_hdr(const struct common_firmware_header *hdr) { DRM_DEBUG("size_bytes: %u\n", le32_to_cpu(hdr->size_bytes)); @@ -1155,6 +1159,9 @@ int amdgpu_ucode_init_bo(struct amdgpu_device *adev) adev->firmware.max_ucodes = AMDGPU_UCODE_ID_MAXIMUM; } + if (amdgpu_virt_xgmi_migrate_enabled(adev) && adev->firmware.fw_buf) + adev->firmware.fw_buf_mc = amdgpu_bo_fb_aper_addr(adev->firmware.fw_buf); + for (i = 0; i < adev->firmware.max_ucodes; i++) { ucode = &adev->firmware.ucode[i]; if (ucode->fw) { @@ -1387,6 +1394,19 @@ static const char *amdgpu_ucode_legacy_naming(struct amdgpu_device *adev, int bl return NULL; } +bool amdgpu_is_kicker_fw(struct amdgpu_device *adev) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(kicker_device_list); i++) { + if (adev->pdev->device == kicker_device_list[i].device && + adev->pdev->revision == kicker_device_list[i].revision) + return true; + } + + return false; +} + void amdgpu_ucode_ip_version_decode(struct amdgpu_device *adev, int block_type, char *ucode_prefix, int len) { int maj, min, rev; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h index 9e89c3487be5..6349aad6da35 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h @@ -605,6 +605,11 @@ struct amdgpu_firmware { uint32_t pldm_version; }; +struct kicker_device{ + unsigned short device; + u8 revision; +}; + void amdgpu_ucode_print_mc_hdr(const struct common_firmware_header *hdr); void amdgpu_ucode_print_smc_hdr(const struct common_firmware_header *hdr); void amdgpu_ucode_print_imu_hdr(const struct common_firmware_header *hdr); @@ -632,5 +637,6 @@ amdgpu_ucode_get_load_type(struct amdgpu_device *adev, int load_type); const char *amdgpu_ucode_name(enum AMDGPU_UCODE_ID ucode_id); void amdgpu_ucode_ip_version_decode(struct amdgpu_device *adev, int block_type, char *ucode_prefix, int len); +bool amdgpu_is_kicker_fw(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h index 577c6194db78..9320461bb486 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h @@ -301,6 +301,9 @@ struct amdgpu_virt { union amd_sriov_ras_caps ras_telemetry_en_caps; struct amdgpu_virt_ras ras; struct amd_sriov_ras_telemetry_error_count count_cache; + + /* hibernate and resume with different VF feature for xgmi enabled system */ + bool is_xgmi_node_migrate_enabled; }; struct amdgpu_video_codec_info; @@ -386,6 +389,10 @@ static inline bool is_virtual_machine(void) ((adev)->virt.gim_feature & AMDGIM_FEATURE_VCN_RB_DECOUPLE) #define amdgpu_sriov_is_mes_info_enable(adev) \ ((adev)->virt.gim_feature & AMDGIM_FEATURE_MES_INFO_ENABLE) + +#define amdgpu_virt_xgmi_migrate_enabled(adev) \ + ((adev)->virt.is_xgmi_node_migrate_enabled && (adev)->gmc.xgmi.node_segment_size != 0) + bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev); void amdgpu_virt_init_setting(struct amdgpu_device *adev); int amdgpu_virt_request_full_gpu(struct amdgpu_device *adev, bool init); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 0ff95a56c2ce..af0f655dfd5b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -1271,8 +1271,8 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va, } else { struct drm_gem_object *obj = &bo->tbo.base; - if (obj->import_attach && bo_va->is_xgmi) { - struct dma_buf *dma_buf = obj->import_attach->dmabuf; + if (drm_gem_is_imported(obj) && bo_va->is_xgmi) { + struct dma_buf *dma_buf = obj->dma_buf; struct drm_gem_object *gobj = dma_buf->priv; struct amdgpu_bo *abo = gem_to_amdgpu_bo(gobj); @@ -1631,7 +1631,7 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev, * validation */ if (vm->is_compute_context && - bo_va->base.bo->tbo.base.import_attach && + drm_gem_is_imported(&bo_va->base.bo->tbo.base) && (!bo_va->base.bo->tbo.resource || bo_va->base.bo->tbo.resource->mem_type == TTM_PL_SYSTEM)) amdgpu_vm_bo_evicted_user(&bo_va->base); @@ -2395,10 +2395,11 @@ void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint32_t min_vm_size, else adev->vm_manager.fragment_size = amdgpu_vm_fragment_size; - DRM_INFO("vm size is %u GB, %u levels, block size is %u-bit, fragment size is %u-bit\n", - vm_size, adev->vm_manager.num_level + 1, - adev->vm_manager.block_size, - adev->vm_manager.fragment_size); + dev_info( + adev->dev, + "vm size is %u GB, %u levels, block size is %u-bit, fragment size is %u-bit\n", + vm_size, adev->vm_manager.num_level + 1, + adev->vm_manager.block_size, adev->vm_manager.fragment_size); } /** @@ -2565,8 +2566,8 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode & AMDGPU_VM_USE_CPU_FOR_GFX); - DRM_DEBUG_DRIVER("VM update mode is %s\n", - vm->use_cpu_for_update ? "CPU" : "SDMA"); + dev_dbg(adev->dev, "VM update mode is %s\n", + vm->use_cpu_for_update ? "CPU" : "SDMA"); WARN_ONCE((vm->use_cpu_for_update && !amdgpu_gmc_vram_full_visible(&adev->gmc)), "CPU update of VM recommended only for large BAR system\n"); @@ -2608,7 +2609,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, r = amdgpu_vm_create_task_info(vm); if (r) - DRM_DEBUG("Failed to create task info for VM\n"); + dev_dbg(adev->dev, "Failed to create task info for VM\n"); amdgpu_bo_unreserve(vm->root.bo); amdgpu_bo_unref(&root_bo); @@ -2659,8 +2660,8 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm) /* Update VM state */ vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode & AMDGPU_VM_USE_CPU_FOR_COMPUTE); - DRM_DEBUG_DRIVER("VM update mode is %s\n", - vm->use_cpu_for_update ? "CPU" : "SDMA"); + dev_dbg(adev->dev, "VM update mode is %s\n", + vm->use_cpu_for_update ? "CPU" : "SDMA"); WARN_ONCE((vm->use_cpu_for_update && !amdgpu_gmc_vram_full_visible(&adev->gmc)), "CPU update of VM recommended only for large BAR system\n"); @@ -2983,7 +2984,7 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid, error_unlock: amdgpu_bo_unreserve(root); if (r < 0) - DRM_ERROR("Can't handle page fault (%d)\n", r); + dev_err(adev->dev, "Can't handle page fault (%d)\n", r); error_unref: amdgpu_bo_unref(&root); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c index 322816805bfb..c8fcafeb6864 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c @@ -445,6 +445,222 @@ void amdgpu_xcp_release_sched(struct amdgpu_device *adev, } } +int amdgpu_xcp_select_scheds(struct amdgpu_device *adev, + u32 hw_ip, u32 hw_prio, + struct amdgpu_fpriv *fpriv, + unsigned int *num_scheds, + struct drm_gpu_scheduler ***scheds) +{ + u32 sel_xcp_id; + int i; + struct amdgpu_xcp_mgr *xcp_mgr = adev->xcp_mgr; + + if (fpriv->xcp_id == AMDGPU_XCP_NO_PARTITION) { + u32 least_ref_cnt = ~0; + + fpriv->xcp_id = 0; + for (i = 0; i < xcp_mgr->num_xcps; i++) { + u32 total_ref_cnt; + + total_ref_cnt = atomic_read(&xcp_mgr->xcp[i].ref_cnt); + if (total_ref_cnt < least_ref_cnt) { + fpriv->xcp_id = i; + least_ref_cnt = total_ref_cnt; + } + } + } + sel_xcp_id = fpriv->xcp_id; + + if (xcp_mgr->xcp[sel_xcp_id].gpu_sched[hw_ip][hw_prio].num_scheds) { + *num_scheds = + xcp_mgr->xcp[fpriv->xcp_id].gpu_sched[hw_ip][hw_prio].num_scheds; + *scheds = + xcp_mgr->xcp[fpriv->xcp_id].gpu_sched[hw_ip][hw_prio].sched; + atomic_inc(&adev->xcp_mgr->xcp[sel_xcp_id].ref_cnt); + dev_dbg(adev->dev, "Selected partition #%d", sel_xcp_id); + } else { + dev_err(adev->dev, "Failed to schedule partition #%d.", sel_xcp_id); + return -ENOENT; + } + + return 0; +} + +static void amdgpu_set_xcp_id(struct amdgpu_device *adev, + uint32_t inst_idx, + struct amdgpu_ring *ring) +{ + int xcp_id; + enum AMDGPU_XCP_IP_BLOCK ip_blk; + uint32_t inst_mask; + + ring->xcp_id = AMDGPU_XCP_NO_PARTITION; + if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) + adev->gfx.enforce_isolation[0].xcp_id = ring->xcp_id; + if ((adev->xcp_mgr->mode == AMDGPU_XCP_MODE_NONE) || + (ring->funcs->type == AMDGPU_RING_TYPE_CPER)) + return; + + inst_mask = 1 << inst_idx; + + switch (ring->funcs->type) { + case AMDGPU_HW_IP_GFX: + case AMDGPU_RING_TYPE_COMPUTE: + case AMDGPU_RING_TYPE_KIQ: + ip_blk = AMDGPU_XCP_GFX; + break; + case AMDGPU_RING_TYPE_SDMA: + ip_blk = AMDGPU_XCP_SDMA; + break; + case AMDGPU_RING_TYPE_VCN_ENC: + case AMDGPU_RING_TYPE_VCN_JPEG: + ip_blk = AMDGPU_XCP_VCN; + break; + default: + dev_err(adev->dev, "Not support ring type %d!", ring->funcs->type); + return; + } + + for (xcp_id = 0; xcp_id < adev->xcp_mgr->num_xcps; xcp_id++) { + if (adev->xcp_mgr->xcp[xcp_id].ip[ip_blk].inst_mask & inst_mask) { + ring->xcp_id = xcp_id; + dev_dbg(adev->dev, "ring:%s xcp_id :%u", ring->name, + ring->xcp_id); + if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) + adev->gfx.enforce_isolation[xcp_id].xcp_id = xcp_id; + break; + } + } +} + +static void amdgpu_xcp_gpu_sched_update(struct amdgpu_device *adev, + struct amdgpu_ring *ring, + unsigned int sel_xcp_id) +{ + unsigned int *num_gpu_sched; + + num_gpu_sched = &adev->xcp_mgr->xcp[sel_xcp_id] + .gpu_sched[ring->funcs->type][ring->hw_prio].num_scheds; + adev->xcp_mgr->xcp[sel_xcp_id].gpu_sched[ring->funcs->type][ring->hw_prio] + .sched[(*num_gpu_sched)++] = &ring->sched; + dev_dbg(adev->dev, "%s :[%d] gpu_sched[%d][%d] = %d", + ring->name, sel_xcp_id, ring->funcs->type, + ring->hw_prio, *num_gpu_sched); +} + +static int amdgpu_xcp_sched_list_update(struct amdgpu_device *adev) +{ + struct amdgpu_ring *ring; + int i; + + for (i = 0; i < MAX_XCP; i++) { + atomic_set(&adev->xcp_mgr->xcp[i].ref_cnt, 0); + memset(adev->xcp_mgr->xcp[i].gpu_sched, 0, sizeof(adev->xcp_mgr->xcp->gpu_sched)); + } + + if (adev->xcp_mgr->mode == AMDGPU_XCP_MODE_NONE) + return 0; + + for (i = 0; i < AMDGPU_MAX_RINGS; i++) { + ring = adev->rings[i]; + if (!ring || !ring->sched.ready || ring->no_scheduler) + continue; + + amdgpu_xcp_gpu_sched_update(adev, ring, ring->xcp_id); + + /* VCN may be shared by two partitions under CPX MODE in certain + * configs. + */ + if ((ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC || + ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG) && + (adev->xcp_mgr->num_xcps > adev->vcn.num_vcn_inst)) + amdgpu_xcp_gpu_sched_update(adev, ring, ring->xcp_id + 1); + } + + return 0; +} + +int amdgpu_xcp_update_partition_sched_list(struct amdgpu_device *adev) +{ + int i; + + for (i = 0; i < adev->num_rings; i++) { + struct amdgpu_ring *ring = adev->rings[i]; + + if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE || + ring->funcs->type == AMDGPU_RING_TYPE_KIQ) + amdgpu_set_xcp_id(adev, ring->xcc_id, ring); + else + amdgpu_set_xcp_id(adev, ring->me, ring); + } + + return amdgpu_xcp_sched_list_update(adev); +} + +void amdgpu_xcp_update_supported_modes(struct amdgpu_xcp_mgr *xcp_mgr) +{ + struct amdgpu_device *adev = xcp_mgr->adev; + + xcp_mgr->supp_xcp_modes = 0; + + switch (NUM_XCC(adev->gfx.xcc_mask)) { + case 8: + xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) | + BIT(AMDGPU_DPX_PARTITION_MODE) | + BIT(AMDGPU_QPX_PARTITION_MODE) | + BIT(AMDGPU_CPX_PARTITION_MODE); + break; + case 6: + xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) | + BIT(AMDGPU_TPX_PARTITION_MODE) | + BIT(AMDGPU_CPX_PARTITION_MODE); + break; + case 4: + xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) | + BIT(AMDGPU_DPX_PARTITION_MODE) | + BIT(AMDGPU_CPX_PARTITION_MODE); + break; + case 2: + xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) | + BIT(AMDGPU_CPX_PARTITION_MODE); + break; + case 1: + xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) | + BIT(AMDGPU_CPX_PARTITION_MODE); + break; + + default: + break; + } +} + +int amdgpu_xcp_pre_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags) +{ + /* TODO: + * Stop user queues and threads, and make sure GPU is empty of work. + */ + + if (flags & AMDGPU_XCP_OPS_KFD) + amdgpu_amdkfd_device_fini_sw(xcp_mgr->adev); + + return 0; +} + +int amdgpu_xcp_post_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags) +{ + int ret = 0; + + if (flags & AMDGPU_XCP_OPS_KFD) { + amdgpu_amdkfd_device_probe(xcp_mgr->adev); + amdgpu_amdkfd_device_init(xcp_mgr->adev); + /* If KFD init failed, return failure */ + if (!xcp_mgr->adev->kfd.init_complete) + ret = -EIO; + } + + return ret; +} + /*====================== xcp sysfs - configuration ======================*/ #define XCP_CFG_SYSFS_RES_ATTR_SHOW(_name) \ static ssize_t amdgpu_xcp_res_sysfs_##_name##_show( \ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h index 454b33f889fb..70a0f8400b57 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h @@ -39,6 +39,8 @@ #define AMDGPU_XCP_NO_PARTITION (~0) +#define AMDGPU_XCP_OPS_KFD (1 << 0) + struct amdgpu_fpriv; enum AMDGPU_XCP_IP_BLOCK { @@ -144,10 +146,6 @@ struct amdgpu_xcp_mgr_funcs { int (*suspend)(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id); int (*prepare_resume)(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id); int (*resume)(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id); - int (*select_scheds)(struct amdgpu_device *adev, - u32 hw_ip, u32 hw_prio, struct amdgpu_fpriv *fpriv, - unsigned int *num_scheds, struct drm_gpu_scheduler ***scheds); - int (*update_partition_sched_list)(struct amdgpu_device *adev); }; int amdgpu_xcp_prepare_suspend(struct amdgpu_xcp_mgr *xcp_mgr, int xcp_id); @@ -176,19 +174,18 @@ int amdgpu_xcp_open_device(struct amdgpu_device *adev, struct drm_file *file_priv); void amdgpu_xcp_release_sched(struct amdgpu_device *adev, struct amdgpu_ctx_entity *entity); - +int amdgpu_xcp_select_scheds(struct amdgpu_device *adev, + u32 hw_ip, u32 hw_prio, + struct amdgpu_fpriv *fpriv, + unsigned int *num_scheds, + struct drm_gpu_scheduler ***scheds); +void amdgpu_xcp_update_supported_modes(struct amdgpu_xcp_mgr *xcp_mgr); +int amdgpu_xcp_update_partition_sched_list(struct amdgpu_device *adev); +int amdgpu_xcp_pre_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags); +int amdgpu_xcp_post_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags); void amdgpu_xcp_sysfs_init(struct amdgpu_device *adev); void amdgpu_xcp_sysfs_fini(struct amdgpu_device *adev); -#define amdgpu_xcp_select_scheds(adev, e, c, d, x, y) \ - ((adev)->xcp_mgr && (adev)->xcp_mgr->funcs && \ - (adev)->xcp_mgr->funcs->select_scheds ? \ - (adev)->xcp_mgr->funcs->select_scheds((adev), (e), (c), (d), (x), (y)) : -ENOENT) -#define amdgpu_xcp_update_partition_sched_list(adev) \ - ((adev)->xcp_mgr && (adev)->xcp_mgr->funcs && \ - (adev)->xcp_mgr->funcs->update_partition_sched_list ? \ - (adev)->xcp_mgr->funcs->update_partition_sched_list(adev) : 0) - static inline int amdgpu_xcp_get_num_xcp(struct amdgpu_xcp_mgr *xcp_mgr) { if (!xcp_mgr) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index d9ad37711c3e..1ede308a7c67 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -1771,16 +1771,25 @@ void amdgpu_xgmi_early_init(struct amdgpu_device *adev) case IP_VERSION(9, 4, 0): case IP_VERSION(9, 4, 1): case IP_VERSION(9, 4, 2): - adev->gmc.xgmi.max_speed = XGMI_SPEED_25GT; + /* 25 GT/s */ + adev->gmc.xgmi.max_speed = 25; adev->gmc.xgmi.max_width = 16; break; case IP_VERSION(9, 4, 3): case IP_VERSION(9, 4, 4): case IP_VERSION(9, 5, 0): - adev->gmc.xgmi.max_speed = XGMI_SPEED_32GT; + /* 32 GT/s */ + adev->gmc.xgmi.max_speed = 32; adev->gmc.xgmi.max_width = 16; break; default: break; } } + +void amgpu_xgmi_set_max_speed_width(struct amdgpu_device *adev, + uint16_t max_speed, uint8_t max_width) +{ + adev->gmc.xgmi.max_speed = max_speed; + adev->gmc.xgmi.max_width = max_width; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h index f994be985f42..bba0b26fee8f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h @@ -25,12 +25,6 @@ #include <drm/task_barrier.h> #include "amdgpu_ras.h" -enum amdgpu_xgmi_link_speed { - XGMI_SPEED_16GT = 16, - XGMI_SPEED_25GT = 25, - XGMI_SPEED_32GT = 32 -}; - struct amdgpu_hive_info { struct kobject kobj; uint64_t hive_id; @@ -97,7 +91,7 @@ struct amdgpu_xgmi { struct ras_common_if *ras_if; bool connected_to_cpu; struct amdgpu_xgmi_ras *ras; - enum amdgpu_xgmi_link_speed max_speed; + uint16_t max_speed; uint8_t max_width; }; @@ -130,4 +124,6 @@ int amdgpu_xgmi_get_ext_link(struct amdgpu_device *adev, int link_num); void amdgpu_xgmi_early_init(struct amdgpu_device *adev); uint32_t amdgpu_xgmi_get_max_bandwidth(struct amdgpu_device *adev); +void amgpu_xgmi_set_max_speed_width(struct amdgpu_device *adev, + uint16_t max_speed, uint8_t max_width); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c b/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c index 1c083304ae77..914cf4bfb033 100644 --- a/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c +++ b/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c @@ -29,12 +29,11 @@ #include "gfx_v9_4_3.h" #include "gfxhub_v1_2.h" #include "sdma_v4_4_2.h" +#include "amdgpu_ip.h" #define XCP_INST_MASK(num_inst, xcp_id) \ (num_inst ? GENMASK(num_inst - 1, 0) << (xcp_id * num_inst) : 0) -#define AMDGPU_XCP_OPS_KFD (1 << 0) - void aqua_vanjaram_doorbell_index_init(struct amdgpu_device *adev) { int i; @@ -62,234 +61,6 @@ void aqua_vanjaram_doorbell_index_init(struct amdgpu_device *adev) adev->doorbell_index.max_assignment = AMDGPU_DOORBELL_LAYOUT1_MAX_ASSIGNMENT << 1; } -static bool aqua_vanjaram_xcp_vcn_shared(struct amdgpu_device *adev) -{ - return (adev->xcp_mgr->num_xcps > adev->vcn.num_vcn_inst); -} - -static void aqua_vanjaram_set_xcp_id(struct amdgpu_device *adev, - uint32_t inst_idx, struct amdgpu_ring *ring) -{ - int xcp_id; - enum AMDGPU_XCP_IP_BLOCK ip_blk; - uint32_t inst_mask; - - ring->xcp_id = AMDGPU_XCP_NO_PARTITION; - if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) - adev->gfx.enforce_isolation[0].xcp_id = ring->xcp_id; - if ((adev->xcp_mgr->mode == AMDGPU_XCP_MODE_NONE) || - (ring->funcs->type == AMDGPU_RING_TYPE_CPER)) - return; - - inst_mask = 1 << inst_idx; - - switch (ring->funcs->type) { - case AMDGPU_HW_IP_GFX: - case AMDGPU_RING_TYPE_COMPUTE: - case AMDGPU_RING_TYPE_KIQ: - ip_blk = AMDGPU_XCP_GFX; - break; - case AMDGPU_RING_TYPE_SDMA: - ip_blk = AMDGPU_XCP_SDMA; - break; - case AMDGPU_RING_TYPE_VCN_ENC: - case AMDGPU_RING_TYPE_VCN_JPEG: - ip_blk = AMDGPU_XCP_VCN; - break; - default: - DRM_ERROR("Not support ring type %d!", ring->funcs->type); - return; - } - - for (xcp_id = 0; xcp_id < adev->xcp_mgr->num_xcps; xcp_id++) { - if (adev->xcp_mgr->xcp[xcp_id].ip[ip_blk].inst_mask & inst_mask) { - ring->xcp_id = xcp_id; - dev_dbg(adev->dev, "ring:%s xcp_id :%u", ring->name, - ring->xcp_id); - if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE) - adev->gfx.enforce_isolation[xcp_id].xcp_id = xcp_id; - break; - } - } -} - -static void aqua_vanjaram_xcp_gpu_sched_update( - struct amdgpu_device *adev, - struct amdgpu_ring *ring, - unsigned int sel_xcp_id) -{ - unsigned int *num_gpu_sched; - - num_gpu_sched = &adev->xcp_mgr->xcp[sel_xcp_id] - .gpu_sched[ring->funcs->type][ring->hw_prio].num_scheds; - adev->xcp_mgr->xcp[sel_xcp_id].gpu_sched[ring->funcs->type][ring->hw_prio] - .sched[(*num_gpu_sched)++] = &ring->sched; - DRM_DEBUG("%s :[%d] gpu_sched[%d][%d] = %d", ring->name, - sel_xcp_id, ring->funcs->type, - ring->hw_prio, *num_gpu_sched); -} - -static int aqua_vanjaram_xcp_sched_list_update( - struct amdgpu_device *adev) -{ - struct amdgpu_ring *ring; - int i; - - for (i = 0; i < MAX_XCP; i++) { - atomic_set(&adev->xcp_mgr->xcp[i].ref_cnt, 0); - memset(adev->xcp_mgr->xcp[i].gpu_sched, 0, sizeof(adev->xcp_mgr->xcp->gpu_sched)); - } - - if (adev->xcp_mgr->mode == AMDGPU_XCP_MODE_NONE) - return 0; - - for (i = 0; i < AMDGPU_MAX_RINGS; i++) { - ring = adev->rings[i]; - if (!ring || !ring->sched.ready || ring->no_scheduler) - continue; - - aqua_vanjaram_xcp_gpu_sched_update(adev, ring, ring->xcp_id); - - /* VCN may be shared by two partitions under CPX MODE in certain - * configs. - */ - if ((ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC || - ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG) && - aqua_vanjaram_xcp_vcn_shared(adev)) - aqua_vanjaram_xcp_gpu_sched_update(adev, ring, ring->xcp_id + 1); - } - - return 0; -} - -static int aqua_vanjaram_update_partition_sched_list(struct amdgpu_device *adev) -{ - int i; - - for (i = 0; i < adev->num_rings; i++) { - struct amdgpu_ring *ring = adev->rings[i]; - - if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE || - ring->funcs->type == AMDGPU_RING_TYPE_KIQ) - aqua_vanjaram_set_xcp_id(adev, ring->xcc_id, ring); - else - aqua_vanjaram_set_xcp_id(adev, ring->me, ring); - } - - return aqua_vanjaram_xcp_sched_list_update(adev); -} - -static int aqua_vanjaram_select_scheds( - struct amdgpu_device *adev, - u32 hw_ip, - u32 hw_prio, - struct amdgpu_fpriv *fpriv, - unsigned int *num_scheds, - struct drm_gpu_scheduler ***scheds) -{ - u32 sel_xcp_id; - int i; - - if (fpriv->xcp_id == AMDGPU_XCP_NO_PARTITION) { - u32 least_ref_cnt = ~0; - - fpriv->xcp_id = 0; - for (i = 0; i < adev->xcp_mgr->num_xcps; i++) { - u32 total_ref_cnt; - - total_ref_cnt = atomic_read(&adev->xcp_mgr->xcp[i].ref_cnt); - if (total_ref_cnt < least_ref_cnt) { - fpriv->xcp_id = i; - least_ref_cnt = total_ref_cnt; - } - } - } - sel_xcp_id = fpriv->xcp_id; - - if (adev->xcp_mgr->xcp[sel_xcp_id].gpu_sched[hw_ip][hw_prio].num_scheds) { - *num_scheds = adev->xcp_mgr->xcp[fpriv->xcp_id].gpu_sched[hw_ip][hw_prio].num_scheds; - *scheds = adev->xcp_mgr->xcp[fpriv->xcp_id].gpu_sched[hw_ip][hw_prio].sched; - atomic_inc(&adev->xcp_mgr->xcp[sel_xcp_id].ref_cnt); - DRM_DEBUG("Selected partition #%d", sel_xcp_id); - } else { - DRM_ERROR("Failed to schedule partition #%d.", sel_xcp_id); - return -ENOENT; - } - - return 0; -} - -static int8_t aqua_vanjaram_logical_to_dev_inst(struct amdgpu_device *adev, - enum amd_hw_ip_block_type block, - int8_t inst) -{ - int8_t dev_inst; - - switch (block) { - case GC_HWIP: - case SDMA0_HWIP: - /* Both JPEG and VCN as JPEG is only alias of VCN */ - case VCN_HWIP: - dev_inst = adev->ip_map.dev_inst[block][inst]; - break; - default: - /* For rest of the IPs, no look up required. - * Assume 'logical instance == physical instance' for all configs. */ - dev_inst = inst; - break; - } - - return dev_inst; -} - -static uint32_t aqua_vanjaram_logical_to_dev_mask(struct amdgpu_device *adev, - enum amd_hw_ip_block_type block, - uint32_t mask) -{ - uint32_t dev_mask = 0; - int8_t log_inst, dev_inst; - - while (mask) { - log_inst = ffs(mask) - 1; - dev_inst = aqua_vanjaram_logical_to_dev_inst(adev, block, log_inst); - dev_mask |= (1 << dev_inst); - mask &= ~(1 << log_inst); - } - - return dev_mask; -} - -static void aqua_vanjaram_populate_ip_map(struct amdgpu_device *adev, - enum amd_hw_ip_block_type ip_block, - uint32_t inst_mask) -{ - int l = 0, i; - - while (inst_mask) { - i = ffs(inst_mask) - 1; - adev->ip_map.dev_inst[ip_block][l++] = i; - inst_mask &= ~(1 << i); - } - for (; l < HWIP_MAX_INSTANCE; l++) - adev->ip_map.dev_inst[ip_block][l] = -1; -} - -void aqua_vanjaram_ip_map_init(struct amdgpu_device *adev) -{ - u32 ip_map[][2] = { - { GC_HWIP, adev->gfx.xcc_mask }, - { SDMA0_HWIP, adev->sdma.sdma_mask }, - { VCN_HWIP, adev->vcn.inst_mask }, - }; - int i; - - for (i = 0; i < ARRAY_SIZE(ip_map); ++i) - aqua_vanjaram_populate_ip_map(adev, ip_map[i][0], ip_map[i][1]); - - adev->ip_map.logical_to_dev_inst = aqua_vanjaram_logical_to_dev_inst; - adev->ip_map.logical_to_dev_mask = aqua_vanjaram_logical_to_dev_mask; -} - /* Fixed pattern for smn addressing on different AIDs: * bit[34]: indicate cross AID access * bit[33:32]: indicate target AID id @@ -353,11 +124,14 @@ static int aqua_vanjaram_query_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr) if (adev->nbio.funcs->get_compute_partition_mode) { mode = adev->nbio.funcs->get_compute_partition_mode(adev); - if (mode != derv_mode) + if (mode != derv_mode) { dev_warn( adev->dev, "Mismatch in compute partition mode - reported : %d derived : %d", mode, derv_mode); + if (derv_mode == AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE) + amdgpu_device_bus_status_check(adev); + } } return mode; @@ -593,72 +367,6 @@ static bool __aqua_vanjaram_is_valid_mode(struct amdgpu_xcp_mgr *xcp_mgr, return false; } -static int __aqua_vanjaram_pre_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags) -{ - /* TODO: - * Stop user queues and threads, and make sure GPU is empty of work. - */ - - if (flags & AMDGPU_XCP_OPS_KFD) - amdgpu_amdkfd_device_fini_sw(xcp_mgr->adev); - - return 0; -} - -static int __aqua_vanjaram_post_partition_switch(struct amdgpu_xcp_mgr *xcp_mgr, u32 flags) -{ - int ret = 0; - - if (flags & AMDGPU_XCP_OPS_KFD) { - amdgpu_amdkfd_device_probe(xcp_mgr->adev); - amdgpu_amdkfd_device_init(xcp_mgr->adev); - /* If KFD init failed, return failure */ - if (!xcp_mgr->adev->kfd.init_complete) - ret = -EIO; - } - - return ret; -} - -static void -__aqua_vanjaram_update_supported_modes(struct amdgpu_xcp_mgr *xcp_mgr) -{ - struct amdgpu_device *adev = xcp_mgr->adev; - - xcp_mgr->supp_xcp_modes = 0; - - switch (NUM_XCC(adev->gfx.xcc_mask)) { - case 8: - xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) | - BIT(AMDGPU_DPX_PARTITION_MODE) | - BIT(AMDGPU_QPX_PARTITION_MODE) | - BIT(AMDGPU_CPX_PARTITION_MODE); - break; - case 6: - xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) | - BIT(AMDGPU_TPX_PARTITION_MODE) | - BIT(AMDGPU_CPX_PARTITION_MODE); - break; - case 4: - xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) | - BIT(AMDGPU_DPX_PARTITION_MODE) | - BIT(AMDGPU_CPX_PARTITION_MODE); - break; - /* this seems only existing in emulation phase */ - case 2: - xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) | - BIT(AMDGPU_CPX_PARTITION_MODE); - break; - case 1: - xcp_mgr->supp_xcp_modes = BIT(AMDGPU_SPX_PARTITION_MODE) | - BIT(AMDGPU_CPX_PARTITION_MODE); - break; - - default: - break; - } -} - static void __aqua_vanjaram_update_available_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr) { int mode; @@ -705,7 +413,7 @@ static int aqua_vanjaram_switch_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr, goto out; } - ret = __aqua_vanjaram_pre_partition_switch(xcp_mgr, flags); + ret = amdgpu_xcp_pre_partition_switch(xcp_mgr, flags); if (ret) goto unlock; @@ -718,7 +426,7 @@ static int aqua_vanjaram_switch_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr, *num_xcps = num_xcc / num_xcc_per_xcp; amdgpu_xcp_init(xcp_mgr, *num_xcps, mode); - ret = __aqua_vanjaram_post_partition_switch(xcp_mgr, flags); + ret = amdgpu_xcp_post_partition_switch(xcp_mgr, flags); if (!ret) __aqua_vanjaram_update_available_partition_mode(xcp_mgr); unlock: @@ -801,9 +509,6 @@ struct amdgpu_xcp_mgr_funcs aqua_vanjaram_xcp_funcs = { .get_ip_details = &aqua_vanjaram_get_xcp_ip_details, .get_xcp_res_info = &aqua_vanjaram_get_xcp_res_info, .get_xcp_mem_id = &aqua_vanjaram_get_xcp_mem_id, - .select_scheds = &aqua_vanjaram_select_scheds, - .update_partition_sched_list = - &aqua_vanjaram_update_partition_sched_list }; static int aqua_vanjaram_xcp_mgr_init(struct amdgpu_device *adev) @@ -818,7 +523,7 @@ static int aqua_vanjaram_xcp_mgr_init(struct amdgpu_device *adev) if (ret) return ret; - __aqua_vanjaram_update_supported_modes(adev->xcp_mgr); + amdgpu_xcp_update_supported_modes(adev->xcp_mgr); /* TODO: Default memory node affinity init */ return ret; @@ -858,7 +563,7 @@ int aqua_vanjaram_init_soc_config(struct amdgpu_device *adev) if (ret) return ret; - aqua_vanjaram_ip_map_init(adev); + amdgpu_ip_map_init(adev); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 75ea071744eb..8c377ecbb8a7 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -9522,7 +9522,9 @@ static void gfx_v10_ring_insert_nop(struct amdgpu_ring *ring, uint32_t num_nop) amdgpu_ring_insert_nop(ring, num_nop - 1); } -static int gfx_v10_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid) +static int gfx_v10_0_reset_kgq(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; struct amdgpu_kiq *kiq = &adev->gfx.kiq[0]; @@ -9538,6 +9540,8 @@ static int gfx_v10_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid) if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues) return -EINVAL; + drm_sched_wqueue_stop(&ring->sched); + spin_lock_irqsave(&kiq->ring_lock, flags); if (amdgpu_ring_alloc(kiq_ring, 5 + 7 + 7 + kiq->pmf->map_queues_size)) { @@ -9575,11 +9579,17 @@ static int gfx_v10_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid) return r; } - return amdgpu_ring_test_ring(ring); + r = amdgpu_ring_test_ring(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static int gfx_v10_0_reset_kcq(struct amdgpu_ring *ring, - unsigned int vmid) + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; struct amdgpu_kiq *kiq = &adev->gfx.kiq[0]; @@ -9593,6 +9603,8 @@ static int gfx_v10_0_reset_kcq(struct amdgpu_ring *ring, if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues) return -EINVAL; + drm_sched_wqueue_stop(&ring->sched); + spin_lock_irqsave(&kiq->ring_lock, flags); if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) { @@ -9647,7 +9659,12 @@ static int gfx_v10_0_reset_kcq(struct amdgpu_ring *ring, if (r) return r; - return amdgpu_ring_test_ring(ring); + r = amdgpu_ring_test_ring(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static void gfx_v10_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printer *p) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c index afd6d59164bf..37dcec2d0784 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c @@ -85,6 +85,7 @@ MODULE_FIRMWARE("amdgpu/gc_11_0_0_pfp.bin"); MODULE_FIRMWARE("amdgpu/gc_11_0_0_me.bin"); MODULE_FIRMWARE("amdgpu/gc_11_0_0_mec.bin"); MODULE_FIRMWARE("amdgpu/gc_11_0_0_rlc.bin"); +MODULE_FIRMWARE("amdgpu/gc_11_0_0_rlc_kicker.bin"); MODULE_FIRMWARE("amdgpu/gc_11_0_0_rlc_1.bin"); MODULE_FIRMWARE("amdgpu/gc_11_0_0_toc.bin"); MODULE_FIRMWARE("amdgpu/gc_11_0_1_pfp.bin"); @@ -759,6 +760,10 @@ static int gfx_v11_0_init_microcode(struct amdgpu_device *adev) err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw, AMDGPU_UCODE_REQUIRED, "amdgpu/gc_11_0_0_rlc_1.bin"); + else if (amdgpu_is_kicker_fw(adev)) + err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw, + AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_rlc_kicker.bin", ucode_prefix); else err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw, AMDGPU_UCODE_REQUIRED, @@ -6806,7 +6811,9 @@ static int gfx_v11_reset_gfx_pipe(struct amdgpu_ring *ring) return 0; } -static int gfx_v11_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid) +static int gfx_v11_0_reset_kgq(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; int r; @@ -6814,6 +6821,8 @@ static int gfx_v11_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid) if (amdgpu_sriov_vf(adev)) return -EINVAL; + drm_sched_wqueue_stop(&ring->sched); + r = amdgpu_mes_reset_legacy_queue(ring->adev, ring, vmid, false); if (r) { @@ -6835,7 +6844,12 @@ static int gfx_v11_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid) return r; } - return amdgpu_ring_test_ring(ring); + r = amdgpu_ring_test_ring(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static int gfx_v11_0_reset_compute_pipe(struct amdgpu_ring *ring) @@ -6968,7 +6982,9 @@ static int gfx_v11_0_reset_compute_pipe(struct amdgpu_ring *ring) return 0; } -static int gfx_v11_0_reset_kcq(struct amdgpu_ring *ring, unsigned int vmid) +static int gfx_v11_0_reset_kcq(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; int r = 0; @@ -6976,6 +6992,8 @@ static int gfx_v11_0_reset_kcq(struct amdgpu_ring *ring, unsigned int vmid) if (amdgpu_sriov_vf(adev)) return -EINVAL; + drm_sched_wqueue_stop(&ring->sched); + r = amdgpu_mes_reset_legacy_queue(ring->adev, ring, vmid, true); if (r) { dev_warn(adev->dev, "fail(%d) to reset kcq and try pipe reset\n", r); @@ -6995,7 +7013,12 @@ static int gfx_v11_0_reset_kcq(struct amdgpu_ring *ring, unsigned int vmid) return r; } - return amdgpu_ring_test_ring(ring); + r = amdgpu_ring_test_ring(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static void gfx_v11_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printer *p) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c index 1234c8d64e20..e4fc42470cf3 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c @@ -5307,7 +5307,9 @@ static int gfx_v12_reset_gfx_pipe(struct amdgpu_ring *ring) return 0; } -static int gfx_v12_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid) +static int gfx_v12_0_reset_kgq(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; int r; @@ -5315,6 +5317,8 @@ static int gfx_v12_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid) if (amdgpu_sriov_vf(adev)) return -EINVAL; + drm_sched_wqueue_stop(&ring->sched); + r = amdgpu_mes_reset_legacy_queue(ring->adev, ring, vmid, false); if (r) { dev_warn(adev->dev, "reset via MES failed and try pipe reset %d\n", r); @@ -5335,7 +5339,12 @@ static int gfx_v12_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid) return r; } - return amdgpu_ring_test_ring(ring); + r = amdgpu_ring_test_ring(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static int gfx_v12_0_reset_compute_pipe(struct amdgpu_ring *ring) @@ -5421,7 +5430,9 @@ static int gfx_v12_0_reset_compute_pipe(struct amdgpu_ring *ring) return 0; } -static int gfx_v12_0_reset_kcq(struct amdgpu_ring *ring, unsigned int vmid) +static int gfx_v12_0_reset_kcq(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; int r; @@ -5429,6 +5440,8 @@ static int gfx_v12_0_reset_kcq(struct amdgpu_ring *ring, unsigned int vmid) if (amdgpu_sriov_vf(adev)) return -EINVAL; + drm_sched_wqueue_stop(&ring->sched); + r = amdgpu_mes_reset_legacy_queue(ring->adev, ring, vmid, true); if (r) { dev_warn(adev->dev, "fail(%d) to reset kcq and try pipe reset\n", r); @@ -5448,7 +5461,12 @@ static int gfx_v12_0_reset_kcq(struct amdgpu_ring *ring, unsigned int vmid) return r; } - return amdgpu_ring_test_ring(ring); + r = amdgpu_ring_test_ring(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static void gfx_v12_0_ring_begin_use(struct amdgpu_ring *ring) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c index da0534ff1271..2aa323dab34e 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c @@ -4884,76 +4884,6 @@ static void gfx_v7_0_emit_mem_sync_compute(struct amdgpu_ring *ring) amdgpu_ring_write(ring, 0x0000000A); /* poll interval */ } -static void gfx_v7_0_wait_reg_mem(struct amdgpu_ring *ring, int eng_sel, - int mem_space, int opt, uint32_t addr0, - uint32_t addr1, uint32_t ref, uint32_t mask, - uint32_t inv) -{ - amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5)); - amdgpu_ring_write(ring, - /* memory (1) or register (0) */ - (WAIT_REG_MEM_MEM_SPACE(mem_space) | - WAIT_REG_MEM_OPERATION(opt) | /* wait */ - WAIT_REG_MEM_FUNCTION(3) | /* equal */ - WAIT_REG_MEM_ENGINE(eng_sel))); - - if (mem_space) - BUG_ON(addr0 & 0x3); /* Dword align */ - amdgpu_ring_write(ring, addr0); - amdgpu_ring_write(ring, addr1); - amdgpu_ring_write(ring, ref); - amdgpu_ring_write(ring, mask); - amdgpu_ring_write(ring, inv); /* poll interval */ -} - -static void gfx_v7_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg, - uint32_t val, uint32_t mask) -{ - gfx_v7_0_wait_reg_mem(ring, 0, 0, 0, reg, 0, val, mask, 0x20); -} - -static int gfx_v7_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid) -{ - struct amdgpu_device *adev = ring->adev; - struct amdgpu_kiq *kiq = &adev->gfx.kiq[0]; - struct amdgpu_ring *kiq_ring = &kiq->ring; - unsigned long flags; - u32 tmp; - int r; - - if (amdgpu_sriov_vf(adev)) - return -EINVAL; - - if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues) - return -EINVAL; - - spin_lock_irqsave(&kiq->ring_lock, flags); - - if (amdgpu_ring_alloc(kiq_ring, 5)) { - spin_unlock_irqrestore(&kiq->ring_lock, flags); - return -ENOMEM; - } - - tmp = REG_SET_FIELD(0, CP_VMID_RESET, RESET_REQUEST, 1 << vmid); - gfx_v7_0_ring_emit_wreg(kiq_ring, mmCP_VMID_RESET, tmp); - amdgpu_ring_commit(kiq_ring); - - spin_unlock_irqrestore(&kiq->ring_lock, flags); - - r = amdgpu_ring_test_ring(kiq_ring); - if (r) - return r; - - if (amdgpu_ring_alloc(ring, 7 + 12 + 5)) - return -ENOMEM; - gfx_v7_0_ring_emit_fence_gfx(ring, ring->fence_drv.gpu_addr, - ring->fence_drv.sync_seq, AMDGPU_FENCE_FLAG_EXEC); - gfx_v7_0_ring_emit_reg_wait(ring, mmCP_VMID_RESET, 0, 0xffff); - gfx_v7_0_ring_emit_wreg(ring, mmCP_VMID_RESET, 0); - - return amdgpu_ring_test_ring(ring); -} - static const struct amd_ip_funcs gfx_v7_0_ip_funcs = { .name = "gfx_v7_0", .early_init = gfx_v7_0_early_init, @@ -5003,7 +4933,6 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = { .emit_wreg = gfx_v7_0_ring_emit_wreg, .soft_recovery = gfx_v7_0_ring_soft_recovery, .emit_mem_sync = gfx_v7_0_emit_mem_sync, - .reset = gfx_v7_0_reset_kgq, }; static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = { diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 5ee2237d8ee8..68c401ecb3ec 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -6339,34 +6339,6 @@ static void gfx_v8_0_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, amdgpu_ring_write(ring, val); } -static void gfx_v8_0_wait_reg_mem(struct amdgpu_ring *ring, int eng_sel, - int mem_space, int opt, uint32_t addr0, - uint32_t addr1, uint32_t ref, uint32_t mask, - uint32_t inv) -{ - amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5)); - amdgpu_ring_write(ring, - /* memory (1) or register (0) */ - (WAIT_REG_MEM_MEM_SPACE(mem_space) | - WAIT_REG_MEM_OPERATION(opt) | /* wait */ - WAIT_REG_MEM_FUNCTION(3) | /* equal */ - WAIT_REG_MEM_ENGINE(eng_sel))); - - if (mem_space) - BUG_ON(addr0 & 0x3); /* Dword align */ - amdgpu_ring_write(ring, addr0); - amdgpu_ring_write(ring, addr1); - amdgpu_ring_write(ring, ref); - amdgpu_ring_write(ring, mask); - amdgpu_ring_write(ring, inv); /* poll interval */ -} - -static void gfx_v8_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg, - uint32_t val, uint32_t mask) -{ - gfx_v8_0_wait_reg_mem(ring, 0, 0, 0, reg, 0, val, mask, 0x20); -} - static void gfx_v8_0_ring_soft_recovery(struct amdgpu_ring *ring, unsigned vmid) { struct amdgpu_device *adev = ring->adev; @@ -6843,48 +6815,6 @@ static void gfx_v8_0_emit_wave_limit(struct amdgpu_ring *ring, bool enable) } -static int gfx_v8_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid) -{ - struct amdgpu_device *adev = ring->adev; - struct amdgpu_kiq *kiq = &adev->gfx.kiq[0]; - struct amdgpu_ring *kiq_ring = &kiq->ring; - unsigned long flags; - u32 tmp; - int r; - - if (amdgpu_sriov_vf(adev)) - return -EINVAL; - - if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues) - return -EINVAL; - - spin_lock_irqsave(&kiq->ring_lock, flags); - - if (amdgpu_ring_alloc(kiq_ring, 5)) { - spin_unlock_irqrestore(&kiq->ring_lock, flags); - return -ENOMEM; - } - - tmp = REG_SET_FIELD(0, CP_VMID_RESET, RESET_REQUEST, 1 << vmid); - gfx_v8_0_ring_emit_wreg(kiq_ring, mmCP_VMID_RESET, tmp); - amdgpu_ring_commit(kiq_ring); - - spin_unlock_irqrestore(&kiq->ring_lock, flags); - - r = amdgpu_ring_test_ring(kiq_ring); - if (r) - return r; - - if (amdgpu_ring_alloc(ring, 7 + 12 + 5)) - return -ENOMEM; - gfx_v8_0_ring_emit_fence_gfx(ring, ring->fence_drv.gpu_addr, - ring->fence_drv.sync_seq, AMDGPU_FENCE_FLAG_EXEC); - gfx_v8_0_ring_emit_reg_wait(ring, mmCP_VMID_RESET, 0, 0xffff); - gfx_v8_0_ring_emit_wreg(ring, mmCP_VMID_RESET, 0); - - return amdgpu_ring_test_ring(ring); -} - static const struct amd_ip_funcs gfx_v8_0_ip_funcs = { .name = "gfx_v8_0", .early_init = gfx_v8_0_early_init, @@ -6950,7 +6880,6 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_gfx = { .emit_wreg = gfx_v8_0_ring_emit_wreg, .soft_recovery = gfx_v8_0_ring_soft_recovery, .emit_mem_sync = gfx_v8_0_emit_mem_sync, - .reset = gfx_v8_0_reset_kgq, }; static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_compute = { diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index d377a7c57d5e..76ba664efecb 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -2235,6 +2235,25 @@ static int gfx_v9_0_sw_init(struct amdgpu_ip_block *ip_block) } switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { + case IP_VERSION(9, 0, 1): + case IP_VERSION(9, 2, 1): + case IP_VERSION(9, 4, 0): + case IP_VERSION(9, 2, 2): + case IP_VERSION(9, 1, 0): + case IP_VERSION(9, 3, 0): + adev->gfx.cleaner_shader_ptr = gfx_9_4_2_cleaner_shader_hex; + adev->gfx.cleaner_shader_size = sizeof(gfx_9_4_2_cleaner_shader_hex); + if (adev->gfx.me_fw_version >= 167 && + adev->gfx.pfp_fw_version >= 196 && + adev->gfx.mec_fw_version >= 474) { + adev->gfx.enable_cleaner_shader = true; + r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size); + if (r) { + adev->gfx.enable_cleaner_shader = false; + dev_err(adev->dev, "Failed to initialize cleaner shader\n"); + } + } + break; case IP_VERSION(9, 4, 2): adev->gfx.cleaner_shader_ptr = gfx_9_4_2_cleaner_shader_hex; adev->gfx.cleaner_shader_size = sizeof(gfx_9_4_2_cleaner_shader_hex); @@ -7152,53 +7171,9 @@ static void gfx_v9_ring_insert_nop(struct amdgpu_ring *ring, uint32_t num_nop) amdgpu_ring_insert_nop(ring, num_nop - 1); } -static int gfx_v9_0_reset_kgq(struct amdgpu_ring *ring, unsigned int vmid) -{ - struct amdgpu_device *adev = ring->adev; - struct amdgpu_kiq *kiq = &adev->gfx.kiq[0]; - struct amdgpu_ring *kiq_ring = &kiq->ring; - unsigned long flags; - u32 tmp; - int r; - - if (amdgpu_sriov_vf(adev)) - return -EINVAL; - - if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues) - return -EINVAL; - - spin_lock_irqsave(&kiq->ring_lock, flags); - - if (amdgpu_ring_alloc(kiq_ring, 5)) { - spin_unlock_irqrestore(&kiq->ring_lock, flags); - return -ENOMEM; - } - - tmp = REG_SET_FIELD(0, CP_VMID_RESET, RESET_REQUEST, 1 << vmid); - gfx_v9_0_ring_emit_wreg(kiq_ring, - SOC15_REG_OFFSET(GC, 0, mmCP_VMID_RESET), tmp); - amdgpu_ring_commit(kiq_ring); - - spin_unlock_irqrestore(&kiq->ring_lock, flags); - - r = amdgpu_ring_test_ring(kiq_ring); - if (r) - return r; - - if (amdgpu_ring_alloc(ring, 7 + 7 + 5)) - return -ENOMEM; - gfx_v9_0_ring_emit_fence(ring, ring->fence_drv.gpu_addr, - ring->fence_drv.sync_seq, AMDGPU_FENCE_FLAG_EXEC); - gfx_v9_0_ring_emit_reg_wait(ring, - SOC15_REG_OFFSET(GC, 0, mmCP_VMID_RESET), 0, 0xffff); - gfx_v9_0_ring_emit_wreg(ring, - SOC15_REG_OFFSET(GC, 0, mmCP_VMID_RESET), 0); - - return amdgpu_ring_test_ring(ring); -} - static int gfx_v9_0_reset_kcq(struct amdgpu_ring *ring, - unsigned int vmid) + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; struct amdgpu_kiq *kiq = &adev->gfx.kiq[0]; @@ -7212,6 +7187,8 @@ static int gfx_v9_0_reset_kcq(struct amdgpu_ring *ring, if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues) return -EINVAL; + drm_sched_wqueue_stop(&ring->sched); + spin_lock_irqsave(&kiq->ring_lock, flags); if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) { @@ -7267,7 +7244,13 @@ static int gfx_v9_0_reset_kcq(struct amdgpu_ring *ring, DRM_ERROR("fail to remap queue\n"); return r; } - return amdgpu_ring_test_ring(ring); + + r = amdgpu_ring_test_ring(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static void gfx_v9_ip_print(struct amdgpu_ip_block *ip_block, struct drm_printer *p) @@ -7477,7 +7460,6 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_gfx = { .emit_reg_write_reg_wait = gfx_v9_0_ring_emit_reg_write_reg_wait, .soft_recovery = gfx_v9_0_ring_soft_recovery, .emit_mem_sync = gfx_v9_0_emit_mem_sync, - .reset = gfx_v9_0_reset_kgq, .emit_cleaner_shader = gfx_v9_0_ring_emit_cleaner_shader, .begin_use = amdgpu_gfx_enforce_isolation_ring_begin_use, .end_use = amdgpu_gfx_enforce_isolation_ring_end_use, diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c index c233edf60569..daed0f187bda 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c @@ -3552,7 +3552,8 @@ static int gfx_v9_4_3_reset_hw_pipe(struct amdgpu_ring *ring) } static int gfx_v9_4_3_reset_kcq(struct amdgpu_ring *ring, - unsigned int vmid) + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; struct amdgpu_kiq *kiq = &adev->gfx.kiq[ring->xcc_id]; @@ -3566,6 +3567,8 @@ static int gfx_v9_4_3_reset_kcq(struct amdgpu_ring *ring, if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues) return -EINVAL; + drm_sched_wqueue_stop(&ring->sched); + spin_lock_irqsave(&kiq->ring_lock, flags); if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size)) { @@ -3619,7 +3622,13 @@ pipe_reset: dev_err(adev->dev, "fail to remap queue\n"); return r; } - return amdgpu_ring_test_ring(ring); + + r = amdgpu_ring_test_ring(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } enum amdgpu_gfx_cp_ras_mem_id { diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c index cb25f7f0dfc1..6c03bf9f1ae8 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c @@ -74,6 +74,8 @@ static void gfxhub_v1_2_setup_vm_pt_regs(struct amdgpu_device *adev, static void gfxhub_v1_2_xcc_init_gart_aperture_regs(struct amdgpu_device *adev, uint32_t xcc_mask) { + uint64_t gart_start = amdgpu_virt_xgmi_migrate_enabled(adev) ? + adev->gmc.vram_start : adev->gmc.fb_start; uint64_t pt_base; int i; @@ -91,10 +93,10 @@ static void gfxhub_v1_2_xcc_init_gart_aperture_regs(struct amdgpu_device *adev, if (adev->gmc.pdb0_bo) { WREG32_SOC15(GC, GET_INST(GC, i), regVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32, - (u32)(adev->gmc.fb_start >> 12)); + (u32)(gart_start >> 12)); WREG32_SOC15(GC, GET_INST(GC, i), regVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32, - (u32)(adev->gmc.fb_start >> 44)); + (u32)(gart_start >> 44)); WREG32_SOC15(GC, GET_INST(GC, i), regVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32, @@ -180,7 +182,7 @@ gfxhub_v1_2_xcc_init_system_aperture_regs(struct amdgpu_device *adev, /* In the case squeezing vram into GART aperture, we don't use * FB aperture and AGP aperture. Disable them. */ - if (adev->gmc.pdb0_bo) { + if (adev->gmc.pdb0_bo && adev->gmc.xgmi.connected_to_cpu) { WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_FB_LOCATION_TOP, 0); WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_FB_LOCATION_BASE, 0x00FFFFFF); WREG32_SOC15(GC, GET_INST(GC, i), regMC_VM_AGP_TOP, 0); diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 78f65aea03f8..f73da518a6e8 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -78,8 +78,6 @@ #define mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION_DCN2 0x05ea #define mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION_DCN2_BASE_IDX 2 -#define MAX_MEM_RANGES 8 - static const char * const gfxhub_client_ids[] = { "CB", "DB", @@ -411,11 +409,6 @@ static const uint32_t ecc_umc_mcumc_ctrl_mask_addrs[] = { (0x001d43e0 + 0x00001800), }; -static inline bool gmc_v9_0_is_multi_chiplet(struct amdgpu_device *adev) -{ - return !!adev->aid_mask; -} - static int gmc_v9_0_ecc_interrupt_state(struct amdgpu_device *adev, struct amdgpu_irq_src *src, unsigned int type, @@ -649,7 +642,7 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, addr, entry->client_id, soc15_ih_clientid_name[entry->client_id]); - if (gmc_v9_0_is_multi_chiplet(adev)) + if (amdgpu_is_multi_aid(adev)) dev_err(adev->dev, " cookie node_id %d fault from die %s%d%s\n", node_id, node_id % 4 == 3 ? "RSV" : "AID", node_id / 4, node_id % 4 == 1 ? ".XCD0" : node_id % 4 == 2 ? ".XCD1" : ""); @@ -798,7 +791,7 @@ static bool gmc_v9_0_use_invalidate_semaphore(struct amdgpu_device *adev, uint32_t vmhub) { if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) || - gmc_v9_0_is_multi_chiplet(adev)) + amdgpu_is_multi_aid(adev)) return false; return ((vmhub == AMDGPU_MMHUB0(0) || @@ -1382,46 +1375,6 @@ static unsigned int gmc_v9_0_get_vbios_fb_size(struct amdgpu_device *adev) return size; } -static enum amdgpu_memory_partition -gmc_v9_0_get_memory_partition(struct amdgpu_device *adev, u32 *supp_modes) -{ - enum amdgpu_memory_partition mode = UNKNOWN_MEMORY_PARTITION_MODE; - - if (adev->nbio.funcs->get_memory_partition_mode) - mode = adev->nbio.funcs->get_memory_partition_mode(adev, - supp_modes); - - return mode; -} - -static enum amdgpu_memory_partition -gmc_v9_0_query_vf_memory_partition(struct amdgpu_device *adev) -{ - switch (adev->gmc.num_mem_partitions) { - case 0: - return UNKNOWN_MEMORY_PARTITION_MODE; - case 1: - return AMDGPU_NPS1_PARTITION_MODE; - case 2: - return AMDGPU_NPS2_PARTITION_MODE; - case 4: - return AMDGPU_NPS4_PARTITION_MODE; - default: - return AMDGPU_NPS1_PARTITION_MODE; - } - - return AMDGPU_NPS1_PARTITION_MODE; -} - -static enum amdgpu_memory_partition -gmc_v9_0_query_memory_partition(struct amdgpu_device *adev) -{ - if (amdgpu_sriov_vf(adev)) - return gmc_v9_0_query_vf_memory_partition(adev); - - return gmc_v9_0_get_memory_partition(adev, NULL); -} - static bool gmc_v9_0_need_reset_on_init(struct amdgpu_device *adev) { if (adev->nbio.funcs && adev->nbio.funcs->is_nps_switch_requested && @@ -1443,7 +1396,7 @@ static const struct amdgpu_gmc_funcs gmc_v9_0_gmc_funcs = { .get_vm_pte = gmc_v9_0_get_vm_pte, .override_vm_pte_flags = gmc_v9_0_override_vm_pte_flags, .get_vbios_fb_size = gmc_v9_0_get_vbios_fb_size, - .query_mem_partition_mode = &gmc_v9_0_query_memory_partition, + .query_mem_partition_mode = &amdgpu_gmc_query_memory_partition, .request_mem_partition_mode = &amdgpu_gmc_request_memory_partition, .need_reset_on_init = &gmc_v9_0_need_reset_on_init, }; @@ -1550,7 +1503,7 @@ static void gmc_v9_0_set_mmhub_ras_funcs(struct amdgpu_device *adev) static void gmc_v9_0_set_gfxhub_funcs(struct amdgpu_device *adev) { - if (gmc_v9_0_is_multi_chiplet(adev)) + if (amdgpu_is_multi_aid(adev)) adev->gfxhub.funcs = &gfxhub_v1_2_funcs; else adev->gfxhub.funcs = &gfxhub_v1_0_funcs; @@ -1596,7 +1549,7 @@ static void gmc_v9_0_init_nps_details(struct amdgpu_device *adev) if (amdgpu_sriov_vf(adev) || (adev->flags & AMD_IS_APU)) return; - mode = gmc_v9_0_get_memory_partition(adev, &supp_modes); + mode = amdgpu_gmc_get_memory_partition(adev, &supp_modes); /* Mode detected by hardware and supported modes available */ if ((mode != UNKNOWN_MEMORY_PARTITION_MODE) && supp_modes) { @@ -1632,7 +1585,7 @@ static int gmc_v9_0_early_init(struct amdgpu_ip_block *ip_block) */ if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 0) || amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1) || - gmc_v9_0_is_multi_chiplet(adev)) + amdgpu_is_multi_aid(adev)) adev->gmc.xgmi.supported = true; if (amdgpu_ip_version(adev, XGMI_HWIP, 0) == IP_VERSION(6, 1, 0)) { @@ -1719,7 +1672,7 @@ static void gmc_v9_0_vram_gtt_location(struct amdgpu_device *adev, /* add the xgmi offset of the physical node */ base += adev->gmc.xgmi.physical_node_id * adev->gmc.xgmi.node_segment_size; - if (adev->gmc.xgmi.connected_to_cpu) { + if (amdgpu_gmc_is_pdb0_enabled(adev)) { amdgpu_gmc_sysvm_location(adev, mc); } else { amdgpu_gmc_vram_location(adev, mc, base); @@ -1834,7 +1787,7 @@ static int gmc_v9_0_gart_init(struct amdgpu_device *adev) return 0; } - if (adev->gmc.xgmi.connected_to_cpu) { + if (amdgpu_gmc_is_pdb0_enabled(adev)) { adev->gmc.vmid0_page_table_depth = 1; adev->gmc.vmid0_page_table_block_size = 12; } else { @@ -1860,7 +1813,7 @@ static int gmc_v9_0_gart_init(struct amdgpu_device *adev) if (r) return r; - if (adev->gmc.xgmi.connected_to_cpu) + if (amdgpu_gmc_is_pdb0_enabled(adev)) r = amdgpu_gmc_pdb0_alloc(adev); } @@ -1882,188 +1835,6 @@ static void gmc_v9_0_save_registers(struct amdgpu_device *adev) adev->gmc.sdpif_register = RREG32_SOC15(DCE, 0, mmDCHUBBUB_SDPIF_MMIO_CNTRL_0); } -static bool gmc_v9_0_validate_partition_info(struct amdgpu_device *adev) -{ - enum amdgpu_memory_partition mode; - u32 supp_modes; - bool valid; - - mode = gmc_v9_0_get_memory_partition(adev, &supp_modes); - - /* Mode detected by hardware not present in supported modes */ - if ((mode != UNKNOWN_MEMORY_PARTITION_MODE) && - !(BIT(mode - 1) & supp_modes)) - return false; - - switch (mode) { - case UNKNOWN_MEMORY_PARTITION_MODE: - case AMDGPU_NPS1_PARTITION_MODE: - valid = (adev->gmc.num_mem_partitions == 1); - break; - case AMDGPU_NPS2_PARTITION_MODE: - valid = (adev->gmc.num_mem_partitions == 2); - break; - case AMDGPU_NPS4_PARTITION_MODE: - valid = (adev->gmc.num_mem_partitions == 3 || - adev->gmc.num_mem_partitions == 4); - break; - default: - valid = false; - } - - return valid; -} - -static bool gmc_v9_0_is_node_present(int *node_ids, int num_ids, int nid) -{ - int i; - - /* Check if node with id 'nid' is present in 'node_ids' array */ - for (i = 0; i < num_ids; ++i) - if (node_ids[i] == nid) - return true; - - return false; -} - -static void -gmc_v9_0_init_acpi_mem_ranges(struct amdgpu_device *adev, - struct amdgpu_mem_partition_info *mem_ranges) -{ - struct amdgpu_numa_info numa_info; - int node_ids[MAX_MEM_RANGES]; - int num_ranges = 0, ret; - int num_xcc, xcc_id; - uint32_t xcc_mask; - - num_xcc = NUM_XCC(adev->gfx.xcc_mask); - xcc_mask = (1U << num_xcc) - 1; - - for_each_inst(xcc_id, xcc_mask) { - ret = amdgpu_acpi_get_mem_info(adev, xcc_id, &numa_info); - if (ret) - continue; - - if (numa_info.nid == NUMA_NO_NODE) { - mem_ranges[0].size = numa_info.size; - mem_ranges[0].numa.node = numa_info.nid; - num_ranges = 1; - break; - } - - if (gmc_v9_0_is_node_present(node_ids, num_ranges, - numa_info.nid)) - continue; - - node_ids[num_ranges] = numa_info.nid; - mem_ranges[num_ranges].numa.node = numa_info.nid; - mem_ranges[num_ranges].size = numa_info.size; - ++num_ranges; - } - - adev->gmc.num_mem_partitions = num_ranges; -} - -static void -gmc_v9_0_init_sw_mem_ranges(struct amdgpu_device *adev, - struct amdgpu_mem_partition_info *mem_ranges) -{ - enum amdgpu_memory_partition mode; - u32 start_addr = 0, size; - int i, r, l; - - mode = gmc_v9_0_query_memory_partition(adev); - - switch (mode) { - case UNKNOWN_MEMORY_PARTITION_MODE: - adev->gmc.num_mem_partitions = 0; - break; - case AMDGPU_NPS1_PARTITION_MODE: - adev->gmc.num_mem_partitions = 1; - break; - case AMDGPU_NPS2_PARTITION_MODE: - adev->gmc.num_mem_partitions = 2; - break; - case AMDGPU_NPS4_PARTITION_MODE: - if (adev->flags & AMD_IS_APU) - adev->gmc.num_mem_partitions = 3; - else - adev->gmc.num_mem_partitions = 4; - break; - default: - adev->gmc.num_mem_partitions = 1; - break; - } - - /* Use NPS range info, if populated */ - r = amdgpu_gmc_get_nps_memranges(adev, mem_ranges, - &adev->gmc.num_mem_partitions); - if (!r) { - l = 0; - for (i = 1; i < adev->gmc.num_mem_partitions; ++i) { - if (mem_ranges[i].range.lpfn > - mem_ranges[i - 1].range.lpfn) - l = i; - } - - } else { - if (!adev->gmc.num_mem_partitions) { - dev_err(adev->dev, - "Not able to detect NPS mode, fall back to NPS1"); - adev->gmc.num_mem_partitions = 1; - } - /* Fallback to sw based calculation */ - size = (adev->gmc.real_vram_size + SZ_16M) >> AMDGPU_GPU_PAGE_SHIFT; - size /= adev->gmc.num_mem_partitions; - - for (i = 0; i < adev->gmc.num_mem_partitions; ++i) { - mem_ranges[i].range.fpfn = start_addr; - mem_ranges[i].size = - ((u64)size << AMDGPU_GPU_PAGE_SHIFT); - mem_ranges[i].range.lpfn = start_addr + size - 1; - start_addr += size; - } - - l = adev->gmc.num_mem_partitions - 1; - } - - /* Adjust the last one */ - mem_ranges[l].range.lpfn = - (adev->gmc.real_vram_size >> AMDGPU_GPU_PAGE_SHIFT) - 1; - mem_ranges[l].size = - adev->gmc.real_vram_size - - ((u64)mem_ranges[l].range.fpfn << AMDGPU_GPU_PAGE_SHIFT); -} - -static int gmc_v9_0_init_mem_ranges(struct amdgpu_device *adev) -{ - bool valid; - - adev->gmc.mem_partitions = kcalloc(MAX_MEM_RANGES, - sizeof(struct amdgpu_mem_partition_info), - GFP_KERNEL); - if (!adev->gmc.mem_partitions) - return -ENOMEM; - - /* TODO : Get the range from PSP/Discovery for dGPU */ - if (adev->gmc.is_app_apu) - gmc_v9_0_init_acpi_mem_ranges(adev, adev->gmc.mem_partitions); - else - gmc_v9_0_init_sw_mem_ranges(adev, adev->gmc.mem_partitions); - - if (amdgpu_sriov_vf(adev)) - valid = true; - else - valid = gmc_v9_0_validate_partition_info(adev); - if (!valid) { - /* TODO: handle invalid case */ - dev_WARN(adev->dev, - "Mem ranges not matching with hardware config"); - } - - return 0; -} - static void gmc_v9_4_3_init_vram_info(struct amdgpu_device *adev) { adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM; @@ -2085,7 +1856,7 @@ static int gmc_v9_0_sw_init(struct amdgpu_ip_block *ip_block) spin_lock_init(&adev->gmc.invalidate_lock); - if (gmc_v9_0_is_multi_chiplet(adev)) { + if (amdgpu_is_multi_aid(adev)) { gmc_v9_4_3_init_vram_info(adev); } else if (!adev->bios) { if (adev->flags & AMD_IS_APU) { @@ -2235,8 +2006,8 @@ static int gmc_v9_0_sw_init(struct amdgpu_ip_block *ip_block) amdgpu_gmc_get_vbios_allocations(adev); - if (gmc_v9_0_is_multi_chiplet(adev)) { - r = gmc_v9_0_init_mem_ranges(adev); + if (amdgpu_is_multi_aid(adev)) { + r = amdgpu_gmc_init_mem_ranges(adev); if (r) return r; } @@ -2264,7 +2035,7 @@ static int gmc_v9_0_sw_init(struct amdgpu_ip_block *ip_block) adev->vm_manager.first_kfd_vmid = (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1) || amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) || - gmc_v9_0_is_multi_chiplet(adev)) ? + amdgpu_is_multi_aid(adev)) ? 3 : 8; @@ -2276,7 +2047,7 @@ static int gmc_v9_0_sw_init(struct amdgpu_ip_block *ip_block) if (r) return r; - if (gmc_v9_0_is_multi_chiplet(adev)) + if (amdgpu_is_multi_aid(adev)) amdgpu_gmc_sysfs_init(adev); return 0; @@ -2286,7 +2057,7 @@ static int gmc_v9_0_sw_fini(struct amdgpu_ip_block *ip_block) { struct amdgpu_device *adev = ip_block->adev; - if (gmc_v9_0_is_multi_chiplet(adev)) + if (amdgpu_is_multi_aid(adev)) amdgpu_gmc_sysfs_fini(adev); amdgpu_gmc_ras_fini(adev); @@ -2360,7 +2131,7 @@ static int gmc_v9_0_gart_enable(struct amdgpu_device *adev) { int r; - if (adev->gmc.xgmi.connected_to_cpu) + if (amdgpu_gmc_is_pdb0_enabled(adev)) amdgpu_gmc_init_pdb0(adev); if (adev->gart.bo == NULL) { @@ -2518,7 +2289,7 @@ static int gmc_v9_0_resume(struct amdgpu_ip_block *ip_block) * information again. */ if (adev->gmc.reset_flags & AMDGPU_GMC_INIT_RESET_NPS) { - gmc_v9_0_init_sw_mem_ranges(adev, adev->gmc.mem_partitions); + amdgpu_gmc_init_sw_mem_ranges(adev, adev->gmc.mem_partitions); adev->gmc.reset_flags &= ~AMDGPU_GMC_INIT_RESET_NPS; } diff --git a/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c b/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c index cfa91d709d49..cc626036ed9c 100644 --- a/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c @@ -32,6 +32,7 @@ #include "gc/gc_11_0_0_sh_mask.h" MODULE_FIRMWARE("amdgpu/gc_11_0_0_imu.bin"); +MODULE_FIRMWARE("amdgpu/gc_11_0_0_imu_kicker.bin"); MODULE_FIRMWARE("amdgpu/gc_11_0_1_imu.bin"); MODULE_FIRMWARE("amdgpu/gc_11_0_2_imu.bin"); MODULE_FIRMWARE("amdgpu/gc_11_0_3_imu.bin"); @@ -51,8 +52,12 @@ static int imu_v11_0_init_microcode(struct amdgpu_device *adev) DRM_DEBUG("\n"); amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix)); - err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED, - "amdgpu/%s_imu.bin", ucode_prefix); + if (amdgpu_is_kicker_fw(adev)) + err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_imu_kicker.bin", ucode_prefix); + else + err = amdgpu_ucode_request(adev, &adev->gfx.imu_fw, AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_imu.bin", ucode_prefix); if (err) goto out; diff --git a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c index 574880d67009..f857796f0297 100644 --- a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c +++ b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c @@ -29,6 +29,12 @@ #include "amdgpu.h" #include "isp_v4_1_1.h" +#define ISP_PERFORMANCE_STATE_LOW 0 +#define ISP_PERFORMANCE_STATE_HIGH 1 + +#define ISP_HIGH_PERFORMANC_XCLK 788 +#define ISP_HIGH_PERFORMANC_ICLK 788 + static const unsigned int isp_4_1_1_int_srcid[MAX_ISP411_INT_SRC] = { ISP_4_1__SRCID__ISP_RINGBUFFER_WPT9, ISP_4_1__SRCID__ISP_RINGBUFFER_WPT10, @@ -56,6 +62,125 @@ static struct gpiod_lookup_table isp_sensor_gpio_table = { }, }; +static int isp_poweroff(struct generic_pm_domain *genpd) +{ + struct amdgpu_isp *isp = container_of(genpd, struct amdgpu_isp, ispgpd); + struct amdgpu_device *adev = isp->adev; + + return amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_ISP, true, 0); +} + +static int isp_poweron(struct generic_pm_domain *genpd) +{ + struct amdgpu_isp *isp = container_of(genpd, struct amdgpu_isp, ispgpd); + struct amdgpu_device *adev = isp->adev; + + return amdgpu_dpm_set_powergating_by_smu(adev, AMD_IP_BLOCK_TYPE_ISP, false, 0); +} + +static int isp_set_performance_state(struct generic_pm_domain *genpd, + unsigned int state) +{ + struct amdgpu_isp *isp = container_of(genpd, struct amdgpu_isp, ispgpd); + struct amdgpu_device *adev = isp->adev; + u32 iclk, xclk; + int ret; + + switch (state) { + case ISP_PERFORMANCE_STATE_HIGH: + xclk = ISP_HIGH_PERFORMANC_XCLK; + iclk = ISP_HIGH_PERFORMANC_ICLK; + break; + case ISP_PERFORMANCE_STATE_LOW: + /* isp runs at default lowest clock-rate on power-on, do nothing */ + return 0; + default: + return -EINVAL; + } + + ret = amdgpu_dpm_set_soft_freq_range(adev, PP_ISPXCLK, xclk, 0); + if (ret) { + drm_err(&adev->ddev, "failed to set xclk %u to %u: %d\n", + xclk, state, ret); + return ret; + } + + ret = amdgpu_dpm_set_soft_freq_range(adev, PP_ISPICLK, iclk, 0); + if (ret) { + drm_err(&adev->ddev, "failed to set iclk %u to %u: %d\n", + iclk, state, ret); + return ret; + } + + return 0; +} + +static int isp_genpd_add_device(struct device *dev, void *data) +{ + struct generic_pm_domain *gpd = data; + struct platform_device *pdev = container_of(dev, struct platform_device, dev); + struct amdgpu_isp *isp = container_of(gpd, struct amdgpu_isp, ispgpd); + struct amdgpu_device *adev = isp->adev; + int ret; + + if (!pdev) + return -EINVAL; + + if (!dev->type->name) { + drm_dbg(&adev->ddev, "Invalid device type to add\n"); + goto exit; + } + + if (strcmp(dev->type->name, "mfd_device")) { + drm_dbg(&adev->ddev, "Invalid isp mfd device %s to add\n", pdev->mfd_cell->name); + goto exit; + } + + ret = pm_genpd_add_device(gpd, dev); + if (ret) { + drm_err(&adev->ddev, "Failed to add dev %s to genpd %d\n", + pdev->mfd_cell->name, ret); + return -ENODEV; + } + +exit: + /* Continue to add */ + return 0; +} + +static int isp_genpd_remove_device(struct device *dev, void *data) +{ + struct generic_pm_domain *gpd = data; + struct platform_device *pdev = container_of(dev, struct platform_device, dev); + struct amdgpu_isp *isp = container_of(gpd, struct amdgpu_isp, ispgpd); + struct amdgpu_device *adev = isp->adev; + int ret; + + if (!pdev) + return -EINVAL; + + if (!dev->type->name) { + drm_dbg(&adev->ddev, "Invalid device type to remove\n"); + goto exit; + } + + if (strcmp(dev->type->name, "mfd_device")) { + drm_dbg(&adev->ddev, "Invalid isp mfd device %s to remove\n", + pdev->mfd_cell->name); + goto exit; + } + + ret = pm_genpd_remove_device(dev); + if (ret) { + drm_err(&adev->ddev, "Failed to remove dev from genpd %d\n", ret); + return -ENODEV; + } + +exit: + /* Continue to remove */ + return 0; +} + static int isp_v4_1_1_hw_init(struct amdgpu_isp *isp) { struct amdgpu_device *adev = isp->adev; @@ -81,11 +206,21 @@ static int isp_v4_1_1_hw_init(struct amdgpu_isp *isp) isp_base = adev->rmmio_base; + isp->ispgpd.name = "ISP_v_4_1_1"; + isp->ispgpd.power_off = isp_poweroff; + isp->ispgpd.power_on = isp_poweron; + isp->ispgpd.set_performance_state = isp_set_performance_state; + + r = pm_genpd_init(&isp->ispgpd, NULL, true); + if (r) { + drm_err(&adev->ddev, "failed to initialize genpd (%d)\n", r); + return -EINVAL; + } + isp->isp_cell = kcalloc(3, sizeof(struct mfd_cell), GFP_KERNEL); if (!isp->isp_cell) { r = -ENOMEM; - drm_err(&adev->ddev, - "%s: isp mfd cell alloc failed\n", __func__); + drm_err(&adev->ddev, "isp mfd cell alloc failed (%d)\n", r); goto failure; } @@ -95,16 +230,14 @@ static int isp_v4_1_1_hw_init(struct amdgpu_isp *isp) GFP_KERNEL); if (!isp->isp_res) { r = -ENOMEM; - drm_err(&adev->ddev, - "%s: isp mfd res alloc failed\n", __func__); + drm_err(&adev->ddev, "isp mfd resource alloc failed (%d)\n", r); goto failure; } isp->isp_pdata = kzalloc(sizeof(*isp->isp_pdata), GFP_KERNEL); if (!isp->isp_pdata) { r = -ENOMEM; - drm_err(&adev->ddev, - "%s: isp platform data alloc failed\n", __func__); + drm_err(&adev->ddev, "isp platform data alloc failed (%d)\n", r); goto failure; } @@ -142,8 +275,7 @@ static int isp_v4_1_1_hw_init(struct amdgpu_isp *isp) isp->isp_i2c_res = kcalloc(1, sizeof(struct resource), GFP_KERNEL); if (!isp->isp_i2c_res) { r = -ENOMEM; - drm_err(&adev->ddev, - "%s: isp mfd res alloc failed\n", __func__); + drm_err(&adev->ddev, "isp mfd res alloc failed (%d)\n", r); goto failure; } @@ -162,8 +294,7 @@ static int isp_v4_1_1_hw_init(struct amdgpu_isp *isp) isp->isp_gpio_res = kcalloc(1, sizeof(struct resource), GFP_KERNEL); if (!isp->isp_gpio_res) { r = -ENOMEM; - drm_err(&adev->ddev, - "%s: isp gpio res alloc failed\n", __func__); + drm_err(&adev->ddev, "isp gpio resource alloc failed (%d)\n", r); goto failure; } @@ -179,10 +310,23 @@ static int isp_v4_1_1_hw_init(struct amdgpu_isp *isp) isp->isp_cell[2].platform_data = isp->isp_pdata; isp->isp_cell[2].pdata_size = sizeof(struct isp_platform_data); - r = mfd_add_hotplug_devices(isp->parent, isp->isp_cell, 3); + /* add only amd_isp_capture and amd_isp_i2c_designware to genpd */ + r = mfd_add_hotplug_devices(isp->parent, isp->isp_cell, 2); if (r) { - drm_err(&adev->ddev, - "%s: add mfd hotplug device failed\n", __func__); + drm_err(&adev->ddev, "add mfd hotplug device failed (%d)\n", r); + goto failure; + } + + r = device_for_each_child(isp->parent, &isp->ispgpd, + isp_genpd_add_device); + if (r) { + drm_err(&adev->ddev, "failed to add devices to genpd (%d)\n", r); + goto failure; + } + + r = mfd_add_hotplug_devices(isp->parent, &isp->isp_cell[2], 1); + if (r) { + drm_err(&adev->ddev, "add pinctl hotplug device failed (%d)\n", r); goto failure; } @@ -201,6 +345,9 @@ failure: static int isp_v4_1_1_hw_fini(struct amdgpu_isp *isp) { + device_for_each_child(isp->parent, NULL, + isp_genpd_remove_device); + mfd_remove_devices(isp->parent); kfree(isp->isp_res); diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c index 4cde8a8bcc83..781a5a8a8361 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c @@ -764,11 +764,21 @@ static int jpeg_v2_0_process_interrupt(struct amdgpu_device *adev, return 0; } -static int jpeg_v2_0_ring_reset(struct amdgpu_ring *ring, unsigned int vmid) +static int jpeg_v2_0_ring_reset(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { + int r; + + drm_sched_wqueue_stop(&ring->sched); jpeg_v2_0_stop(ring->adev); jpeg_v2_0_start(ring->adev); - return amdgpu_ring_test_helper(ring); + r = amdgpu_ring_test_helper(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static const struct amd_ip_funcs jpeg_v2_0_ip_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c index 8b39e114f3be..5be9cdcae32c 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c @@ -643,11 +643,21 @@ static int jpeg_v2_5_process_interrupt(struct amdgpu_device *adev, return 0; } -static int jpeg_v2_5_ring_reset(struct amdgpu_ring *ring, unsigned int vmid) +static int jpeg_v2_5_ring_reset(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { + int r; + + drm_sched_wqueue_stop(&ring->sched); jpeg_v2_5_stop_inst(ring->adev, ring->me); jpeg_v2_5_start_inst(ring->adev, ring->me); - return amdgpu_ring_test_helper(ring); + r = amdgpu_ring_test_helper(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static const struct amd_ip_funcs jpeg_v2_5_ip_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c index 2f8510c2986b..a24bd833d644 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c @@ -555,11 +555,21 @@ static int jpeg_v3_0_process_interrupt(struct amdgpu_device *adev, return 0; } -static int jpeg_v3_0_ring_reset(struct amdgpu_ring *ring, unsigned int vmid) +static int jpeg_v3_0_ring_reset(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { + int r; + + drm_sched_wqueue_stop(&ring->sched); jpeg_v3_0_stop(ring->adev); jpeg_v3_0_start(ring->adev); - return amdgpu_ring_test_helper(ring); + r = amdgpu_ring_test_helper(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static const struct amd_ip_funcs jpeg_v3_0_ip_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c index f17ec5414fd6..1d4edd77837d 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c @@ -720,14 +720,24 @@ static int jpeg_v4_0_process_interrupt(struct amdgpu_device *adev, return 0; } -static int jpeg_v4_0_ring_reset(struct amdgpu_ring *ring, unsigned int vmid) +static int jpeg_v4_0_ring_reset(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { + int r; + if (amdgpu_sriov_vf(ring->adev)) return -EINVAL; + drm_sched_wqueue_stop(&ring->sched); jpeg_v4_0_stop(ring->adev); jpeg_v4_0_start(ring->adev); - return amdgpu_ring_test_helper(ring); + r = amdgpu_ring_test_helper(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static const struct amd_ip_funcs jpeg_v4_0_ip_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c index 79e342d5ab28..78441f8fce97 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c @@ -1143,14 +1143,24 @@ static void jpeg_v4_0_3_core_stall_reset(struct amdgpu_ring *ring) WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CORE_RST_CTRL, 0x00); } -static int jpeg_v4_0_3_ring_reset(struct amdgpu_ring *ring, unsigned int vmid) +static int jpeg_v4_0_3_ring_reset(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { + int r; + if (amdgpu_sriov_vf(ring->adev)) return -EOPNOTSUPP; + drm_sched_wqueue_stop(&ring->sched); jpeg_v4_0_3_core_stall_reset(ring); jpeg_v4_0_3_start_jrbc(ring); - return amdgpu_ring_test_helper(ring); + r = amdgpu_ring_test_helper(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static const struct amd_ip_funcs jpeg_v4_0_3_ip_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c index 3b6f65a25646..6f8a16da9d60 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c @@ -834,14 +834,24 @@ static void jpeg_v5_0_1_core_stall_reset(struct amdgpu_ring *ring) WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CORE_RST_CTRL, 0x00); } -static int jpeg_v5_0_1_ring_reset(struct amdgpu_ring *ring, unsigned int vmid) +static int jpeg_v5_0_1_ring_reset(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { + int r; + if (amdgpu_sriov_vf(ring->adev)) return -EOPNOTSUPP; + drm_sched_wqueue_stop(&ring->sched); jpeg_v5_0_1_core_stall_reset(ring); jpeg_v5_0_1_init_jrbc(ring); - return amdgpu_ring_test_helper(ring); + r = amdgpu_ring_test_helper(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static const struct amd_ip_funcs jpeg_v5_0_1_ip_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c index c9eba537de09..28eb846280dd 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c @@ -1630,10 +1630,12 @@ static int mes_v11_0_hw_init(struct amdgpu_ip_block *ip_block) if (r) goto failure; - r = mes_v11_0_set_hw_resources_1(&adev->mes); - if (r) { - DRM_ERROR("failed mes_v11_0_set_hw_resources_1, r=%d\n", r); - goto failure; + if ((adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x50) { + r = mes_v11_0_set_hw_resources_1(&adev->mes); + if (r) { + DRM_ERROR("failed mes_v11_0_set_hw_resources_1, r=%d\n", r); + goto failure; + } } r = mes_v11_0_query_sched_status(&adev->mes); diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c index b4f17332d466..6b222630f3fa 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c @@ -1742,7 +1742,8 @@ static int mes_v12_0_hw_init(struct amdgpu_ip_block *ip_block) if (r) goto failure; - mes_v12_0_set_hw_resources_1(&adev->mes, AMDGPU_MES_SCHED_PIPE); + if ((adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x4b) + mes_v12_0_set_hw_resources_1(&adev->mes, AMDGPU_MES_SCHED_PIPE); mes_v12_0_init_aggregated_doorbell(&adev->mes); diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c index 76167fadb292..cc688ae79e84 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c @@ -76,6 +76,8 @@ static void mmhub_v1_8_setup_vm_pt_regs(struct amdgpu_device *adev, uint32_t vmi static void mmhub_v1_8_init_gart_aperture_regs(struct amdgpu_device *adev) { + uint64_t gart_start = amdgpu_virt_xgmi_migrate_enabled(adev) ? + adev->gmc.vram_start : adev->gmc.fb_start; uint64_t pt_base; u32 inst_mask; int i; @@ -95,10 +97,10 @@ static void mmhub_v1_8_init_gart_aperture_regs(struct amdgpu_device *adev) if (adev->gmc.pdb0_bo) { WREG32_SOC15(MMHUB, i, regVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32, - (u32)(adev->gmc.fb_start >> 12)); + (u32)(gart_start >> 12)); WREG32_SOC15(MMHUB, i, regVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32, - (u32)(adev->gmc.fb_start >> 44)); + (u32)(gart_start >> 44)); WREG32_SOC15(MMHUB, i, regVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32, diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c index a376f072700d..1c22bc11c1f8 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c @@ -31,9 +31,6 @@ #define NPS_MODE_MASK 0x000000FFL -/* Core 0 Port 0 counter */ -#define smnPCIEP_NAK_COUNTER 0x1A340218 - static void nbio_v7_9_remap_hdp_registers(struct amdgpu_device *adev) { WREG32_SOC15(NBIO, 0, regBIF_BX0_REMAP_HDP_MEM_FLUSH_CNTL, @@ -467,22 +464,6 @@ static void nbio_v7_9_init_registers(struct amdgpu_device *adev) } } -static u64 nbio_v7_9_get_pcie_replay_count(struct amdgpu_device *adev) -{ - u32 val, nak_r, nak_g; - - if (adev->flags & AMD_IS_APU) - return 0; - - /* Get the number of NAKs received and generated */ - val = RREG32_PCIE(smnPCIEP_NAK_COUNTER); - nak_r = val & 0xFFFF; - nak_g = val >> 16; - - /* Add the total number of NAKs, i.e the number of replays */ - return (nak_r + nak_g); -} - #define MMIO_REG_HOLE_OFFSET 0x1A000 static void nbio_v7_9_set_reg_remap(struct amdgpu_device *adev) @@ -524,7 +505,6 @@ const struct amdgpu_nbio_funcs nbio_v7_9_funcs = { .get_memory_partition_mode = nbio_v7_9_get_memory_partition_mode, .is_nps_switch_requested = nbio_v7_9_is_nps_switch_requested, .init_registers = nbio_v7_9_init_registers, - .get_pcie_replay_count = nbio_v7_9_get_pcie_replay_count, .set_reg_remap = nbio_v7_9_set_reg_remap, }; diff --git a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h index f4a91b126c73..73f87131a7e9 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h +++ b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h @@ -106,7 +106,9 @@ enum psp_gfx_cmd_id /*IDs of performance monitoring/profiling*/ GFX_CMD_ID_CONFIG_SQ_PERFMON = 0x00000046, /* Config CGTT_SQ_CLK_CTRL */ /* Dynamic memory partitioninig (NPS mode change)*/ - GFX_CMD_ID_FB_NPS_MODE = 0x00000048, /* Configure memory partitioning mode */ + GFX_CMD_ID_FB_NPS_MODE = 0x00000048, /* Configure memory partitioning mode */ + GFX_CMD_ID_FB_FW_RESERV_ADDR = 0x00000050, /* Query FW reservation addr */ + GFX_CMD_ID_FB_FW_RESERV_EXT_ADDR = 0x00000051, /* Query FW reservation extended addr */ }; /* PSP boot config sub-commands */ @@ -404,11 +406,19 @@ struct psp_gfx_uresp_bootcfg { uint32_t boot_cfg; /* boot config data */ }; +/* Command-specific response for fw reserve info */ +struct psp_gfx_uresp_fw_reserve_info { + uint32_t reserve_base_address_hi; + uint32_t reserve_base_address_lo; + uint32_t reserve_size; +}; + /* Union of command-specific responses for GPCOM ring. */ union psp_gfx_uresp { struct psp_gfx_uresp_reserved reserved; struct psp_gfx_uresp_bootcfg boot_cfg; struct psp_gfx_uresp_fwar_db_info fwar_db_info; + struct psp_gfx_uresp_fw_reserve_info fw_reserve_info; }; /* Structure of GFX Response buffer. diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c index 145186a1e48f..2c4ebd98927f 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c @@ -94,7 +94,7 @@ static int psp_v10_0_ring_create(struct psp_context *psp, /* Wait for response flag (bit 31) in C2PMSG_64 */ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), - 0x80000000, 0x8000FFFF, false); + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); return ret; } @@ -115,7 +115,7 @@ static int psp_v10_0_ring_stop(struct psp_context *psp, /* Wait for response flag (bit 31) in C2PMSG_64 */ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), - 0x80000000, 0x80000000, false); + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); return ret; } diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c index 215543575f47..1a4a26e6ffd2 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c @@ -277,11 +277,13 @@ static int psp_v11_0_ring_stop(struct psp_context *psp, /* Wait for response flag (bit 31) */ if (amdgpu_sriov_vf(adev)) - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); else - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); return ret; } @@ -317,13 +319,15 @@ static int psp_v11_0_ring_create(struct psp_context *psp, mdelay(20); /* Wait for response flag (bit 31) in C2PMSG_101 */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101), - 0x80000000, 0x8000FFFF, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } else { /* Wait for sOS ready for ring creation */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), + MBOX_TOS_READY_FLAG, MBOX_TOS_READY_MASK, false); if (ret) { DRM_ERROR("Failed to wait for sOS ready for ring creation\n"); return ret; @@ -347,8 +351,9 @@ static int psp_v11_0_ring_create(struct psp_context *psp, mdelay(20); /* Wait for response flag (bit 31) in C2PMSG_64 */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), - 0x80000000, 0x8000FFFF, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } return ret; @@ -381,7 +386,8 @@ static int psp_v11_0_mode1_reset(struct psp_context *psp) offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64); - ret = psp_wait_for(psp, offset, 0x80000000, 0x8000FFFF, false); + ret = psp_wait_for(psp, offset, MBOX_TOS_READY_FLAG, + MBOX_TOS_READY_MASK, false); if (ret) { DRM_INFO("psp is not working correctly before mode1 reset!\n"); @@ -395,7 +401,8 @@ static int psp_v11_0_mode1_reset(struct psp_context *psp) offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_33); - ret = psp_wait_for(psp, offset, 0x80000000, 0x80000000, false); + ret = psp_wait_for(psp, offset, MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, + false); if (ret) { DRM_INFO("psp mode 1 reset failed!\n"); diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.c index 5697760a819b..338d015c0f2e 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.c @@ -41,8 +41,9 @@ static int psp_v11_0_8_ring_stop(struct psp_context *psp, /* there might be handshake issue with hardware which needs delay */ mdelay(20); /* Wait for response flag (bit 31) */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } else { /* Write the ring destroy command*/ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, @@ -50,8 +51,9 @@ static int psp_v11_0_8_ring_stop(struct psp_context *psp, /* there might be handshake issue with hardware which needs delay */ mdelay(20); /* Wait for response flag (bit 31) */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } return ret; @@ -87,13 +89,15 @@ static int psp_v11_0_8_ring_create(struct psp_context *psp, mdelay(20); /* Wait for response flag (bit 31) in C2PMSG_101 */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101), - 0x80000000, 0x8000FFFF, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } else { /* Wait for sOS ready for ring creation */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), + MBOX_TOS_READY_FLAG, MBOX_TOS_READY_MASK, false); if (ret) { DRM_ERROR("Failed to wait for trust OS ready for ring creation\n"); return ret; @@ -117,8 +121,9 @@ static int psp_v11_0_8_ring_create(struct psp_context *psp, mdelay(20); /* Wait for response flag (bit 31) in C2PMSG_64 */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), - 0x80000000, 0x8000FFFF, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c index 80153f837470..d54b3e0fabaf 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c @@ -163,7 +163,7 @@ static int psp_v12_0_ring_create(struct psp_context *psp, /* Wait for response flag (bit 31) in C2PMSG_64 */ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), - 0x80000000, 0x8000FFFF, false); + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); return ret; } @@ -184,11 +184,13 @@ static int psp_v12_0_ring_stop(struct psp_context *psp, /* Wait for response flag (bit 31) */ if (amdgpu_sriov_vf(adev)) - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); else - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); return ret; } @@ -219,7 +221,8 @@ static int psp_v12_0_mode1_reset(struct psp_context *psp) offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64); - ret = psp_wait_for(psp, offset, 0x80000000, 0x8000FFFF, false); + ret = psp_wait_for(psp, offset, MBOX_TOS_READY_FLAG, + MBOX_TOS_READY_MASK, false); if (ret) { DRM_INFO("psp is not working correctly before mode1 reset!\n"); @@ -233,7 +236,8 @@ static int psp_v12_0_mode1_reset(struct psp_context *psp) offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_33); - ret = psp_wait_for(psp, offset, 0x80000000, 0x80000000, false); + ret = psp_wait_for(psp, offset, MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, + false); if (ret) { DRM_INFO("psp mode 1 reset failed!\n"); diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c index df612fd9cc50..58b6b64dcd68 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c @@ -42,7 +42,9 @@ MODULE_FIRMWARE("amdgpu/psp_13_0_5_ta.bin"); MODULE_FIRMWARE("amdgpu/psp_13_0_8_toc.bin"); MODULE_FIRMWARE("amdgpu/psp_13_0_8_ta.bin"); MODULE_FIRMWARE("amdgpu/psp_13_0_0_sos.bin"); +MODULE_FIRMWARE("amdgpu/psp_13_0_0_sos_kicker.bin"); MODULE_FIRMWARE("amdgpu/psp_13_0_0_ta.bin"); +MODULE_FIRMWARE("amdgpu/psp_13_0_0_ta_kicker.bin"); MODULE_FIRMWARE("amdgpu/psp_13_0_7_sos.bin"); MODULE_FIRMWARE("amdgpu/psp_13_0_7_ta.bin"); MODULE_FIRMWARE("amdgpu/psp_13_0_10_sos.bin"); @@ -382,8 +384,9 @@ static int psp_v13_0_ring_stop(struct psp_context *psp, /* there might be handshake issue with hardware which needs delay */ mdelay(20); /* Wait for response flag (bit 31) */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } else { /* Write the ring destroy command*/ WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_64, @@ -391,8 +394,9 @@ static int psp_v13_0_ring_stop(struct psp_context *psp, /* there might be handshake issue with hardware which needs delay */ mdelay(20); /* Wait for response flag (bit 31) */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } return ret; @@ -428,13 +432,15 @@ static int psp_v13_0_ring_create(struct psp_context *psp, mdelay(20); /* Wait for response flag (bit 31) in C2PMSG_101 */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101), - 0x80000000, 0x8000FFFF, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } else { /* Wait for sOS ready for ring creation */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), + MBOX_TOS_READY_FLAG, MBOX_TOS_READY_MASK, false); if (ret) { DRM_ERROR("Failed to wait for trust OS ready for ring creation\n"); return ret; @@ -458,8 +464,9 @@ static int psp_v13_0_ring_create(struct psp_context *psp, mdelay(20); /* Wait for response flag (bit 31) in C2PMSG_64 */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), - 0x80000000, 0x8000FFFF, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c index eaa5512a21da..f65af52c1c19 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c @@ -204,8 +204,9 @@ static int psp_v13_0_4_ring_stop(struct psp_context *psp, /* there might be handshake issue with hardware which needs delay */ mdelay(20); /* Wait for response flag (bit 31) */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } else { /* Write the ring destroy command*/ WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_64, @@ -213,8 +214,9 @@ static int psp_v13_0_4_ring_stop(struct psp_context *psp, /* there might be handshake issue with hardware which needs delay */ mdelay(20); /* Wait for response flag (bit 31) */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } return ret; @@ -250,13 +252,15 @@ static int psp_v13_0_4_ring_create(struct psp_context *psp, mdelay(20); /* Wait for response flag (bit 31) in C2PMSG_101 */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101), - 0x80000000, 0x8000FFFF, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_101), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } else { /* Wait for sOS ready for ring creation */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), + MBOX_TOS_READY_FLAG, MBOX_TOS_READY_MASK, false); if (ret) { DRM_ERROR("Failed to wait for trust OS ready for ring creation\n"); return ret; @@ -280,8 +284,9 @@ static int psp_v13_0_4_ring_create(struct psp_context *psp, mdelay(20); /* Wait for response flag (bit 31) in C2PMSG_64 */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), - 0x80000000, 0x8000FFFF, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_64), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c index 256288c6cd78..ffa47c7d24c9 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c @@ -248,8 +248,9 @@ static int psp_v14_0_ring_stop(struct psp_context *psp, /* there might be handshake issue with hardware which needs delay */ mdelay(20); /* Wait for response flag (bit 31) */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_101), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_101), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } else { /* Write the ring destroy command*/ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_64, @@ -257,8 +258,9 @@ static int psp_v14_0_ring_stop(struct psp_context *psp, /* there might be handshake issue with hardware which needs delay */ mdelay(20); /* Wait for response flag (bit 31) */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_64), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_64), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } return ret; @@ -294,13 +296,15 @@ static int psp_v14_0_ring_create(struct psp_context *psp, mdelay(20); /* Wait for response flag (bit 31) in C2PMSG_101 */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_101), - 0x80000000, 0x8000FFFF, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_101), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } else { /* Wait for sOS ready for ring creation */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_64), - 0x80000000, 0x80000000, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_64), + MBOX_TOS_READY_FLAG, MBOX_TOS_READY_MASK, false); if (ret) { DRM_ERROR("Failed to wait for trust OS ready for ring creation\n"); return ret; @@ -324,8 +328,9 @@ static int psp_v14_0_ring_create(struct psp_context *psp, mdelay(20); /* Wait for response flag (bit 31) in C2PMSG_64 */ - ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_64), - 0x80000000, 0x8000FFFF, false); + ret = psp_wait_for( + psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_64), + MBOX_TOS_RESP_FLAG, MBOX_TOS_RESP_MASK, false); } return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c index bcde34e4e0a1..c05f3c1f50db 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c @@ -45,6 +45,7 @@ #include "amdgpu_ras.h" MODULE_FIRMWARE("amdgpu/sdma_4_4_2.bin"); +MODULE_FIRMWARE("amdgpu/sdma_4_4_4.bin"); MODULE_FIRMWARE("amdgpu/sdma_4_4_5.bin"); static const struct amdgpu_hwip_reg_entry sdma_reg_list_4_4_2[] = { @@ -109,6 +110,8 @@ static void sdma_v4_4_2_set_ras_funcs(struct amdgpu_device *adev); static void sdma_v4_4_2_update_reset_mask(struct amdgpu_device *adev); static int sdma_v4_4_2_stop_queue(struct amdgpu_ring *ring); static int sdma_v4_4_2_restore_queue(struct amdgpu_ring *ring); +static int sdma_v4_4_2_soft_reset_engine(struct amdgpu_device *adev, + u32 instance_id); static u32 sdma_v4_4_2_get_reg_offset(struct amdgpu_device *adev, u32 instance, u32 offset) @@ -490,7 +493,7 @@ static void sdma_v4_4_2_inst_gfx_stop(struct amdgpu_device *adev, { struct amdgpu_ring *sdma[AMDGPU_MAX_SDMA_INSTANCES]; u32 doorbell_offset, doorbell; - u32 rb_cntl, ib_cntl; + u32 rb_cntl, ib_cntl, sdma_cntl; int i; for_each_inst(i, inst_mask) { @@ -502,6 +505,9 @@ static void sdma_v4_4_2_inst_gfx_stop(struct amdgpu_device *adev, ib_cntl = RREG32_SDMA(i, regSDMA_GFX_IB_CNTL); ib_cntl = REG_SET_FIELD(ib_cntl, SDMA_GFX_IB_CNTL, IB_ENABLE, 0); WREG32_SDMA(i, regSDMA_GFX_IB_CNTL, ib_cntl); + sdma_cntl = RREG32_SDMA(i, regSDMA_CNTL); + sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA_CNTL, UTC_L1_ENABLE, 0); + WREG32_SDMA(i, regSDMA_CNTL, sdma_cntl); if (sdma[i]->use_doorbell) { doorbell = RREG32_SDMA(i, regSDMA_GFX_DOORBELL); @@ -995,6 +1001,7 @@ static int sdma_v4_4_2_inst_start(struct amdgpu_device *adev, /* set utc l1 enable flag always to 1 */ temp = RREG32_SDMA(i, regSDMA_CNTL); temp = REG_SET_FIELD(temp, SDMA_CNTL, UTC_L1_ENABLE, 1); + WREG32_SDMA(i, regSDMA_CNTL, temp); if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) < IP_VERSION(4, 4, 5)) { /* enable context empty interrupt during initialization */ @@ -1337,6 +1344,7 @@ static bool sdma_v4_4_2_fw_support_paging_queue(struct amdgpu_device *adev) static const struct amdgpu_sdma_funcs sdma_v4_4_2_sdma_funcs = { .stop_kernel_queue = &sdma_v4_4_2_stop_queue, .start_kernel_queue = &sdma_v4_4_2_restore_queue, + .soft_reset_kernel_queue = &sdma_v4_4_2_soft_reset_engine, }; static int sdma_v4_4_2_early_init(struct amdgpu_ip_block *ip_block) @@ -1648,45 +1656,27 @@ static bool sdma_v4_4_2_is_queue_selected(struct amdgpu_device *adev, uint32_t i return (context_status & SDMA_GFX_CONTEXT_STATUS__SELECTED_MASK) != 0; } -static bool sdma_v4_4_2_ring_is_guilty(struct amdgpu_ring *ring) -{ - struct amdgpu_device *adev = ring->adev; - uint32_t instance_id = ring->me; - - return sdma_v4_4_2_is_queue_selected(adev, instance_id, false); -} - -static bool sdma_v4_4_2_page_ring_is_guilty(struct amdgpu_ring *ring) -{ - struct amdgpu_device *adev = ring->adev; - uint32_t instance_id = ring->me; - - if (!adev->sdma.has_page_queue) - return false; - - return sdma_v4_4_2_is_queue_selected(adev, instance_id, true); -} - -static int sdma_v4_4_2_reset_queue(struct amdgpu_ring *ring, unsigned int vmid) +static int sdma_v4_4_2_reset_queue(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; - u32 id = GET_INST(SDMA0, ring->me); + u32 id = ring->me; int r; if (!(adev->sdma.supported_reset & AMDGPU_RESET_TYPE_PER_QUEUE)) return -EOPNOTSUPP; - amdgpu_amdkfd_suspend(adev, false); + amdgpu_amdkfd_suspend(adev, true); r = amdgpu_sdma_reset_engine(adev, id); - amdgpu_amdkfd_resume(adev, false); - + amdgpu_amdkfd_resume(adev, true); return r; } static int sdma_v4_4_2_stop_queue(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; - u32 instance_id = GET_INST(SDMA0, ring->me); + u32 instance_id = ring->me; u32 inst_mask; uint64_t rptr; @@ -1724,8 +1714,8 @@ static int sdma_v4_4_2_stop_queue(struct amdgpu_ring *ring) static int sdma_v4_4_2_restore_queue(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; - u32 inst_mask; - int i; + u32 inst_mask, tmp_mask; + int i, r; inst_mask = 1 << ring->me; udelay(50); @@ -1742,7 +1732,33 @@ static int sdma_v4_4_2_restore_queue(struct amdgpu_ring *ring) return -ETIMEDOUT; } - return sdma_v4_4_2_inst_start(adev, inst_mask, true); + r = sdma_v4_4_2_inst_start(adev, inst_mask, true); + if (r) + return r; + + tmp_mask = inst_mask; + for_each_inst(i, tmp_mask) { + ring = &adev->sdma.instance[i].ring; + + amdgpu_fence_driver_force_completion(ring); + + if (adev->sdma.has_page_queue) { + struct amdgpu_ring *page = &adev->sdma.instance[i].page; + + amdgpu_fence_driver_force_completion(page); + } + } + + return r; +} + +static int sdma_v4_4_2_soft_reset_engine(struct amdgpu_device *adev, + u32 instance_id) +{ + /* For SDMA 4.x, use the existing DPM interface for backward compatibility + * we need to convert the logical instance ID to physical instance ID before reset. + */ + return amdgpu_dpm_reset_sdma(adev, 1 << GET_INST(SDMA0, instance_id)); } static int sdma_v4_4_2_set_trap_irq_state(struct amdgpu_device *adev, @@ -2139,7 +2155,6 @@ static const struct amdgpu_ring_funcs sdma_v4_4_2_ring_funcs = { .emit_reg_wait = sdma_v4_4_2_ring_emit_reg_wait, .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, .reset = sdma_v4_4_2_reset_queue, - .is_guilty = sdma_v4_4_2_ring_is_guilty, }; static const struct amdgpu_ring_funcs sdma_v4_4_2_page_ring_funcs = { @@ -2172,7 +2187,6 @@ static const struct amdgpu_ring_funcs sdma_v4_4_2_page_ring_funcs = { .emit_reg_wait = sdma_v4_4_2_ring_emit_reg_wait, .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, .reset = sdma_v4_4_2_reset_queue, - .is_guilty = sdma_v4_4_2_page_ring_is_guilty, }; static void sdma_v4_4_2_set_ring_funcs(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c index 9505ae96fbec..4d72b085b3dd 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c @@ -1399,6 +1399,7 @@ static int sdma_v5_0_sw_init(struct amdgpu_ip_block *ip_block) return r; for (i = 0; i < adev->sdma.num_instances; i++) { + mutex_init(&adev->sdma.instance[i].engine_reset_mutex); adev->sdma.instance[i].funcs = &sdma_v5_0_sdma_funcs; ring = &adev->sdma.instance[i].ring; ring->ring_obj = NULL; @@ -1538,12 +1539,19 @@ static int sdma_v5_0_soft_reset(struct amdgpu_ip_block *ip_block) return 0; } -static int sdma_v5_0_reset_queue(struct amdgpu_ring *ring, unsigned int vmid) +static int sdma_v5_0_reset_queue(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; u32 inst_id = ring->me; + int r; + + amdgpu_amdkfd_suspend(adev, true); + r = amdgpu_sdma_reset_engine(adev, inst_id); + amdgpu_amdkfd_resume(adev, true); - return amdgpu_sdma_reset_engine(adev, inst_id); + return r; } static int sdma_v5_0_stop_queue(struct amdgpu_ring *ring) @@ -1610,7 +1618,10 @@ static int sdma_v5_0_restore_queue(struct amdgpu_ring *ring) r = sdma_v5_0_gfx_resume_instance(adev, inst_id, true); amdgpu_gfx_rlc_exit_safe_mode(adev, 0); - return r; + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + return 0; } static int sdma_v5_0_ring_preempt_ib(struct amdgpu_ring *ring) diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c index a6e612b4a892..42a25150f83a 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c @@ -1318,6 +1318,7 @@ static int sdma_v5_2_sw_init(struct amdgpu_ip_block *ip_block) } for (i = 0; i < adev->sdma.num_instances; i++) { + mutex_init(&adev->sdma.instance[i].engine_reset_mutex); adev->sdma.instance[i].funcs = &sdma_v5_2_sdma_funcs; ring = &adev->sdma.instance[i].ring; ring->ring_obj = NULL; @@ -1451,12 +1452,19 @@ static int sdma_v5_2_wait_for_idle(struct amdgpu_ip_block *ip_block) return -ETIMEDOUT; } -static int sdma_v5_2_reset_queue(struct amdgpu_ring *ring, unsigned int vmid) +static int sdma_v5_2_reset_queue(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; u32 inst_id = ring->me; + int r; + + amdgpu_amdkfd_suspend(adev, true); + r = amdgpu_sdma_reset_engine(adev, inst_id); + amdgpu_amdkfd_resume(adev, true); - return amdgpu_sdma_reset_engine(adev, inst_id); + return r; } static int sdma_v5_2_stop_queue(struct amdgpu_ring *ring) @@ -1526,7 +1534,10 @@ static int sdma_v5_2_restore_queue(struct amdgpu_ring *ring) r = sdma_v5_2_gfx_resume_instance(adev, inst_id, true); amdgpu_gfx_rlc_exit_safe_mode(adev, 0); - return r; + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + return 0; } static int sdma_v5_2_ring_preempt_ib(struct amdgpu_ring *ring) diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c index 5a70ae17be04..d2effa531817 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c @@ -1374,9 +1374,42 @@ static int sdma_v6_0_sw_init(struct amdgpu_ip_block *ip_block) else DRM_ERROR("Failed to allocated memory for SDMA IP Dump\n"); - /* add firmware version checks here */ - if (0 && !adev->sdma.disable_uq) - adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { + case IP_VERSION(6, 0, 0): + if ((adev->sdma.instance[0].fw_version >= 24) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 0, 1): + if ((adev->sdma.instance[0].fw_version >= 18) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 0, 2): + if ((adev->sdma.instance[0].fw_version >= 21) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 0, 3): + if ((adev->sdma.instance[0].fw_version >= 25) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 1, 0): + if ((adev->sdma.instance[0].fw_version >= 14) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 1, 1): + if ((adev->sdma.instance[0].fw_version >= 17) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 1, 2): + if ((adev->sdma.instance[0].fw_version >= 15) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 1, 3): + if ((adev->sdma.instance[0].fw_version >= 10) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + default: + break; + } r = amdgpu_sdma_sysfs_reset_mask_init(adev); if (r) @@ -1537,7 +1570,9 @@ static int sdma_v6_0_ring_preempt_ib(struct amdgpu_ring *ring) return r; } -static int sdma_v6_0_reset_queue(struct amdgpu_ring *ring, unsigned int vmid) +static int sdma_v6_0_reset_queue(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; int i, r; @@ -1555,11 +1590,18 @@ static int sdma_v6_0_reset_queue(struct amdgpu_ring *ring, unsigned int vmid) return -EINVAL; } + drm_sched_wqueue_stop(&ring->sched); + r = amdgpu_mes_reset_legacy_queue(adev, ring, vmid, true); if (r) return r; - return sdma_v6_0_gfx_resume_instance(adev, i, true); + r = sdma_v6_0_gfx_resume_instance(adev, i, true); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static int sdma_v6_0_set_trap_irq_state(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c index ad47d0bdf777..99a080bad2a3 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c @@ -802,7 +802,9 @@ static bool sdma_v7_0_check_soft_reset(struct amdgpu_ip_block *ip_block) return false; } -static int sdma_v7_0_reset_queue(struct amdgpu_ring *ring, unsigned int vmid) +static int sdma_v7_0_reset_queue(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; int i, r; @@ -820,11 +822,18 @@ static int sdma_v7_0_reset_queue(struct amdgpu_ring *ring, unsigned int vmid) return -EINVAL; } + drm_sched_wqueue_stop(&ring->sched); + r = amdgpu_mes_reset_legacy_queue(adev, ring, vmid, true); if (r) return r; - return sdma_v7_0_gfx_resume_instance(adev, i, true); + r = sdma_v7_0_gfx_resume_instance(adev, i, true); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } /** @@ -1349,9 +1358,15 @@ static int sdma_v7_0_sw_init(struct amdgpu_ip_block *ip_block) else DRM_ERROR("Failed to allocated memory for SDMA IP Dump\n"); - /* add firmware version checks here */ - if (0 && !adev->sdma.disable_uq) - adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { + case IP_VERSION(7, 0, 0): + case IP_VERSION(7, 0, 1): + if ((adev->sdma.instance[0].fw_version >= 7836028) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + default: + break; + } return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.h b/drivers/gpu/drm/amd/amdgpu/soc15.h index ef7c603b50ae..c8ac11a9cdef 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.h +++ b/drivers/gpu/drm/amd/amdgpu/soc15.h @@ -118,7 +118,6 @@ int vega10_reg_base_init(struct amdgpu_device *adev); int vega20_reg_base_init(struct amdgpu_device *adev); int arct_reg_base_init(struct amdgpu_device *adev); int aldebaran_reg_base_init(struct amdgpu_device *adev); -void aqua_vanjaram_ip_map_init(struct amdgpu_device *adev); u64 aqua_vanjaram_encode_ext_smn_addressing(int ext_id); int aqua_vanjaram_init_soc_config(struct amdgpu_device *adev); ssize_t aqua_vanjaram_get_reg_state(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c index b5071f77f78d..eec9133e1b2c 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c @@ -1967,18 +1967,27 @@ static int vcn_v4_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p, return 0; } -static int vcn_v4_0_ring_reset(struct amdgpu_ring *ring, unsigned int vmid) +static int vcn_v4_0_ring_reset(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[ring->me]; + int r; if (!(adev->vcn.supported_reset & AMDGPU_RESET_TYPE_PER_QUEUE)) return -EOPNOTSUPP; + drm_sched_wqueue_stop(&ring->sched); vcn_v4_0_stop(vinst); vcn_v4_0_start(vinst); - return amdgpu_ring_test_helper(ring); + r = amdgpu_ring_test_helper(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static struct amdgpu_ring_funcs vcn_v4_0_unified_ring_vm_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c index 5a33140f5723..d8fd32c1e38e 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c @@ -1594,7 +1594,9 @@ static void vcn_v4_0_3_unified_ring_set_wptr(struct amdgpu_ring *ring) } } -static int vcn_v4_0_3_ring_reset(struct amdgpu_ring *ring, unsigned int vmid) +static int vcn_v4_0_3_ring_reset(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { int r = 0; int vcn_inst; @@ -1607,6 +1609,8 @@ static int vcn_v4_0_3_ring_reset(struct amdgpu_ring *ring, unsigned int vmid) if (!(adev->vcn.supported_reset & AMDGPU_RESET_TYPE_PER_QUEUE)) return -EOPNOTSUPP; + drm_sched_wqueue_stop(&ring->sched); + vcn_inst = GET_INST(VCN, ring->me); r = amdgpu_dpm_reset_vcn(adev, 1 << vcn_inst); @@ -1621,8 +1625,11 @@ static int vcn_v4_0_3_ring_reset(struct amdgpu_ring *ring, unsigned int vmid) vcn_v4_0_3_hw_init_inst(vinst); vcn_v4_0_3_start_dpg_mode(vinst, adev->vcn.inst[ring->me].indirect_sram); r = amdgpu_ring_test_helper(ring); - - return r; + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static const struct amdgpu_ring_funcs vcn_v4_0_3_unified_ring_vm_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c index 16ade84facc7..7e37ddea6355 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c @@ -1465,18 +1465,27 @@ static void vcn_v4_0_5_unified_ring_set_wptr(struct amdgpu_ring *ring) } } -static int vcn_v4_0_5_ring_reset(struct amdgpu_ring *ring, unsigned int vmid) +static int vcn_v4_0_5_ring_reset(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[ring->me]; + int r; if (!(adev->vcn.supported_reset & AMDGPU_RESET_TYPE_PER_QUEUE)) return -EOPNOTSUPP; + drm_sched_wqueue_stop(&ring->sched); vcn_v4_0_5_stop(vinst); vcn_v4_0_5_start(vinst); - return amdgpu_ring_test_helper(ring); + r = amdgpu_ring_test_helper(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static struct amdgpu_ring_funcs vcn_v4_0_5_unified_ring_vm_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c index f8e3f0b882da..47c0bcc9e7d8 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c @@ -1192,18 +1192,27 @@ static void vcn_v5_0_0_unified_ring_set_wptr(struct amdgpu_ring *ring) } } -static int vcn_v5_0_0_ring_reset(struct amdgpu_ring *ring, unsigned int vmid) +static int vcn_v5_0_0_ring_reset(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) { struct amdgpu_device *adev = ring->adev; struct amdgpu_vcn_inst *vinst = &adev->vcn.inst[ring->me]; + int r; if (!(adev->vcn.supported_reset & AMDGPU_RESET_TYPE_PER_QUEUE)) return -EOPNOTSUPP; + drm_sched_wqueue_stop(&ring->sched); vcn_v5_0_0_stop(vinst); vcn_v5_0_0_start(vinst); - return amdgpu_ring_test_helper(ring); + r = amdgpu_ring_test_helper(ring); + if (r) + return r; + amdgpu_fence_driver_force_completion(ring); + drm_sched_wqueue_start(&ring->sched); + return 0; } static const struct amdgpu_ring_funcs vcn_v5_0_0_unified_ring_vm_funcs = { diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c index 338cf43c45fe..cdefd7fcb0da 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c @@ -669,6 +669,9 @@ static int vcn_v5_0_1_start_dpg_mode(struct amdgpu_vcn_inst *vinst, if (indirect) amdgpu_vcn_psp_update_sram(adev, inst_idx, AMDGPU_UCODE_ID_VCN0_RAM); + /* resetting ring, fw should not check RB ring */ + fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET; + /* Pause dpg */ vcn_v5_0_1_pause_dpg_mode(vinst, &state); @@ -681,7 +684,7 @@ static int vcn_v5_0_1_start_dpg_mode(struct amdgpu_vcn_inst *vinst, tmp = RREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE); tmp &= ~(VCN_RB_ENABLE__RB1_EN_MASK); WREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE, tmp); - fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET; + WREG32_SOC15(VCN, vcn_inst, regUVD_RB_RPTR, 0); WREG32_SOC15(VCN, vcn_inst, regUVD_RB_WPTR, 0); @@ -692,6 +695,7 @@ static int vcn_v5_0_1_start_dpg_mode(struct amdgpu_vcn_inst *vinst, tmp = RREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE); tmp |= VCN_RB_ENABLE__RB1_EN_MASK; WREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE, tmp); + /* resetting done, fw can check RB ring */ fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF); WREG32_SOC15(VCN, vcn_inst, regVCN_RB1_DB_CTRL, diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c index a2149afa5803..828a9ceef1e7 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c @@ -22,7 +22,6 @@ */ #include <linux/device.h> -#include <linux/export.h> #include <linux/err.h> #include <linux/fs.h> #include <linux/file.h> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c index bf0854bd5555..7e749f9b6d69 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c @@ -971,7 +971,7 @@ int kgd2kfd_pre_reset(struct kfd_dev *kfd, kfd_smi_event_update_gpu_reset(node, false, reset_context); } - kgd2kfd_suspend(kfd, false); + kgd2kfd_suspend(kfd, true); for (i = 0; i < kfd->num_nodes; i++) kfd_signal_reset_event(kfd->nodes[i]); @@ -1013,13 +1013,33 @@ int kgd2kfd_post_reset(struct kfd_dev *kfd) return 0; } -bool kfd_is_locked(void) +bool kfd_is_locked(struct kfd_dev *kfd) { + uint8_t id = 0; + struct kfd_node *dev; + lockdep_assert_held(&kfd_processes_mutex); - return (kfd_locked > 0); + + /* check reset/suspend lock */ + if (kfd_locked > 0) + return true; + + if (kfd) + return kfd->kfd_dev_lock > 0; + + /* check lock on all cgroup accessible devices */ + while (kfd_topology_enum_kfd_devices(id++, &dev) == 0) { + if (!dev || kfd_devcgroup_check_permission(dev)) + continue; + + if (dev->kfd->kfd_dev_lock > 0) + return true; + } + + return false; } -void kgd2kfd_suspend(struct kfd_dev *kfd, bool run_pm) +void kgd2kfd_suspend(struct kfd_dev *kfd, bool suspend_proc) { struct kfd_node *node; int i; @@ -1027,14 +1047,8 @@ void kgd2kfd_suspend(struct kfd_dev *kfd, bool run_pm) if (!kfd->init_complete) return; - /* for runtime suspend, skip locking kfd */ - if (!run_pm) { - mutex_lock(&kfd_processes_mutex); - /* For first KFD device suspend all the KFD processes */ - if (++kfd_locked == 1) - kfd_suspend_all_processes(); - mutex_unlock(&kfd_processes_mutex); - } + if (suspend_proc) + kgd2kfd_suspend_process(kfd); for (i = 0; i < kfd->num_nodes; i++) { node = kfd->nodes[i]; @@ -1042,7 +1056,7 @@ void kgd2kfd_suspend(struct kfd_dev *kfd, bool run_pm) } } -int kgd2kfd_resume(struct kfd_dev *kfd, bool run_pm) +int kgd2kfd_resume(struct kfd_dev *kfd, bool resume_proc) { int ret, i; @@ -1055,14 +1069,36 @@ int kgd2kfd_resume(struct kfd_dev *kfd, bool run_pm) return ret; } - /* for runtime resume, skip unlocking kfd */ - if (!run_pm) { - mutex_lock(&kfd_processes_mutex); - if (--kfd_locked == 0) - ret = kfd_resume_all_processes(); - WARN_ONCE(kfd_locked < 0, "KFD suspend / resume ref. error"); - mutex_unlock(&kfd_processes_mutex); - } + if (resume_proc) + ret = kgd2kfd_resume_process(kfd); + + return ret; +} + +void kgd2kfd_suspend_process(struct kfd_dev *kfd) +{ + if (!kfd->init_complete) + return; + + mutex_lock(&kfd_processes_mutex); + /* For first KFD device suspend all the KFD processes */ + if (++kfd_locked == 1) + kfd_suspend_all_processes(); + mutex_unlock(&kfd_processes_mutex); +} + +int kgd2kfd_resume_process(struct kfd_dev *kfd) +{ + int ret = 0; + + if (!kfd->init_complete) + return 0; + + mutex_lock(&kfd_processes_mutex); + if (--kfd_locked == 0) + ret = kfd_resume_all_processes(); + WARN_ONCE(kfd_locked < 0, "KFD suspend / resume ref. error"); + mutex_unlock(&kfd_processes_mutex); return ret; } @@ -1442,24 +1478,53 @@ unsigned int kfd_get_num_xgmi_sdma_engines(struct kfd_node *node) kfd_get_num_sdma_engines(node); } -int kgd2kfd_check_and_lock_kfd(void) +int kgd2kfd_check_and_lock_kfd(struct kfd_dev *kfd) { + struct kfd_process *p; + int r = 0, temp, idx; + mutex_lock(&kfd_processes_mutex); - if (!hash_empty(kfd_processes_table) || kfd_is_locked()) { - mutex_unlock(&kfd_processes_mutex); - return -EBUSY; + + if (hash_empty(kfd_processes_table) && !kfd_is_locked(kfd)) + goto out; + + /* fail under system reset/resume or kfd device is partition switching. */ + if (kfd_is_locked(kfd)) { + r = -EBUSY; + goto out; + } + + /* + * ensure all running processes are cgroup excluded from device before mode switch. + * i.e. no pdd was created on the process socket. + */ + idx = srcu_read_lock(&kfd_processes_srcu); + hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) { + int i; + + for (i = 0; i < p->n_pdds; i++) { + if (p->pdds[i]->dev->kfd != kfd) + continue; + + r = -EBUSY; + goto proc_check_unlock; + } } - ++kfd_locked; +proc_check_unlock: + srcu_read_unlock(&kfd_processes_srcu, idx); +out: + if (!r) + ++kfd->kfd_dev_lock; mutex_unlock(&kfd_processes_mutex); - return 0; + return r; } -void kgd2kfd_unlock_kfd(void) +void kgd2kfd_unlock_kfd(struct kfd_dev *kfd) { mutex_lock(&kfd_processes_mutex); - --kfd_locked; + --kfd->kfd_dev_lock; mutex_unlock(&kfd_processes_mutex); } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c b/drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c index dbcb60eb54b2..1d170dc50df3 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c @@ -23,7 +23,6 @@ */ #include <linux/device.h> -#include <linux/export.h> #include <linux/err.h> #include <linux/fs.h> #include <linux/sched.h> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c index 8fa6489b6f5d..505036968a77 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c @@ -240,7 +240,7 @@ static int pm_map_queues_v9(struct packet_manager *pm, uint32_t *buffer, packet->bitfields2.engine_sel = engine_sel__mes_map_queues__compute_vi; - packet->bitfields2.gws_control_queue = q->gws ? 1 : 0; + packet->bitfields2.gws_control_queue = q->properties.is_gws ? 1 : 0; packet->bitfields2.extended_engine_sel = extended_engine_sel__mes_map_queues__legacy_engine_sel; packet->bitfields2.queue_type = diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h index d221c58dccc3..67694bcd9464 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h @@ -372,6 +372,9 @@ struct kfd_dev { /* bitmap for dynamic doorbell allocation from doorbell object */ unsigned long *doorbell_bitmap; + + /* for dynamic partitioning */ + int kfd_dev_lock; }; enum kfd_mempool { @@ -1536,7 +1539,7 @@ static inline bool kfd_flush_tlb_after_unmap(struct kfd_dev *dev) int kfd_send_exception_to_runtime(struct kfd_process *p, unsigned int queue_id, uint64_t error_reason); -bool kfd_is_locked(void); +bool kfd_is_locked(struct kfd_dev *kfd); /* Compute profile */ void kfd_inc_compute_active(struct kfd_node *dev); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c index 722ac1662bdc..5be28c6c4f6a 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c @@ -854,7 +854,7 @@ struct kfd_process *kfd_create_process(struct task_struct *thread) */ mutex_lock(&kfd_processes_mutex); - if (kfd_is_locked()) { + if (kfd_is_locked(NULL)) { pr_debug("KFD is locked! Cannot create process"); process = ERR_PTR(-EINVAL); goto out; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c index 865dca2547de..a0f22ea6d15a 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c @@ -1171,13 +1171,12 @@ svm_range_split_head(struct svm_range *prange, uint64_t new_start, } static void -svm_range_add_child(struct svm_range *prange, struct mm_struct *mm, - struct svm_range *pchild, enum svm_work_list_ops op) +svm_range_add_child(struct svm_range *prange, struct svm_range *pchild, enum svm_work_list_ops op) { pr_debug("add child 0x%p [0x%lx 0x%lx] to prange 0x%p child list %d\n", pchild, pchild->start, pchild->last, prange, op); - pchild->work_item.mm = mm; + pchild->work_item.mm = NULL; pchild->work_item.op = op; list_add_tail(&pchild->child_list, &prange->child_list); } @@ -1278,7 +1277,7 @@ svm_range_get_pte_flags(struct kfd_node *node, mapping_flags |= ext_coherent ? AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC; /* system memory accessed by the dGPU */ } else { - if (gc_ip_version < IP_VERSION(9, 5, 0)) + if (gc_ip_version < IP_VERSION(9, 5, 0) || ext_coherent) mapping_flags |= AMDGPU_VM_MTYPE_UC; else mapping_flags |= AMDGPU_VM_MTYPE_NC; @@ -2394,15 +2393,17 @@ svm_range_add_list_work(struct svm_range_list *svms, struct svm_range *prange, prange->work_item.op != SVM_OP_UNMAP_RANGE) prange->work_item.op = op; } else { - prange->work_item.op = op; - - /* Pairs with mmput in deferred_list_work */ - mmget(mm); - prange->work_item.mm = mm; - list_add_tail(&prange->deferred_list, - &prange->svms->deferred_range_list); - pr_debug("add prange 0x%p [0x%lx 0x%lx] to work list op %d\n", - prange, prange->start, prange->last, op); + /* Pairs with mmput in deferred_list_work. + * If process is exiting and mm is gone, don't update mmu notifier. + */ + if (mmget_not_zero(mm)) { + prange->work_item.mm = mm; + prange->work_item.op = op; + list_add_tail(&prange->deferred_list, + &prange->svms->deferred_range_list); + pr_debug("add prange 0x%p [0x%lx 0x%lx] to work list op %d\n", + prange, prange->start, prange->last, op); + } } spin_unlock(&svms->deferred_list_lock); } @@ -2416,8 +2417,7 @@ void schedule_deferred_list_work(struct svm_range_list *svms) } static void -svm_range_unmap_split(struct mm_struct *mm, struct svm_range *parent, - struct svm_range *prange, unsigned long start, +svm_range_unmap_split(struct svm_range *parent, struct svm_range *prange, unsigned long start, unsigned long last) { struct svm_range *head; @@ -2438,12 +2438,12 @@ svm_range_unmap_split(struct mm_struct *mm, struct svm_range *parent, svm_range_split(tail, last + 1, tail->last, &head); if (head != prange && tail != prange) { - svm_range_add_child(parent, mm, head, SVM_OP_UNMAP_RANGE); - svm_range_add_child(parent, mm, tail, SVM_OP_ADD_RANGE); + svm_range_add_child(parent, head, SVM_OP_UNMAP_RANGE); + svm_range_add_child(parent, tail, SVM_OP_ADD_RANGE); } else if (tail != prange) { - svm_range_add_child(parent, mm, tail, SVM_OP_UNMAP_RANGE); + svm_range_add_child(parent, tail, SVM_OP_UNMAP_RANGE); } else if (head != prange) { - svm_range_add_child(parent, mm, head, SVM_OP_UNMAP_RANGE); + svm_range_add_child(parent, head, SVM_OP_UNMAP_RANGE); } else if (parent != prange) { prange->work_item.op = SVM_OP_UNMAP_RANGE; } @@ -2520,14 +2520,14 @@ svm_range_unmap_from_cpu(struct mm_struct *mm, struct svm_range *prange, l = min(last, pchild->last); if (l >= s) svm_range_unmap_from_gpus(pchild, s, l, trigger); - svm_range_unmap_split(mm, prange, pchild, start, last); + svm_range_unmap_split(prange, pchild, start, last); mutex_unlock(&pchild->lock); } s = max(start, prange->start); l = min(last, prange->last); if (l >= s) svm_range_unmap_from_gpus(prange, s, l, trigger); - svm_range_unmap_split(mm, prange, prange, start, last); + svm_range_unmap_split(prange, prange, start, last); if (unmap_parent) svm_range_add_list_work(svms, prange, mm, SVM_OP_UNMAP_RANGE); @@ -2570,8 +2570,6 @@ svm_range_cpu_invalidate_pagetables(struct mmu_interval_notifier *mni, if (range->event == MMU_NOTIFY_RELEASE) return true; - if (!mmget_not_zero(mni->mm)) - return true; start = mni->interval_tree.start; last = mni->interval_tree.last; @@ -2598,7 +2596,6 @@ svm_range_cpu_invalidate_pagetables(struct mmu_interval_notifier *mni, } svm_range_unlock(prange); - mmput(mni->mm); return true; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c index baa2374acdeb..4ec73f33535e 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c @@ -510,6 +510,10 @@ static ssize_t node_show(struct kobject *kobj, struct attribute *attr, dev->node_props.capability |= HSA_CAP_AQL_QUEUE_DOUBLE_MAP; + if (KFD_GC_VERSION(dev->gpu) < IP_VERSION(10, 0, 0) && + (dev->gpu->adev->sdma.supported_reset & AMDGPU_RESET_TYPE_PER_QUEUE)) + dev->node_props.capability2 |= HSA_CAP2_PER_SDMA_QUEUE_RESET_SUPPORTED; + sysfs_show_32bit_prop(buffer, offs, "max_engine_clk_fcompute", dev->node_props.max_engine_clk_fcompute); @@ -2008,8 +2012,6 @@ static void kfd_topology_set_capabilities(struct kfd_topology_device *dev) if (!amdgpu_sriov_vf(dev->gpu->adev)) dev->node_props.capability |= HSA_CAP_PER_QUEUE_RESET_SUPPORTED; - if (dev->gpu->adev->sdma.supported_reset & AMDGPU_RESET_TYPE_PER_QUEUE) - dev->node_props.capability2 |= HSA_CAP2_PER_SDMA_QUEUE_RESET_SUPPORTED; } else { dev->node_props.debug_prop |= HSA_DBG_WATCH_ADDR_MASK_LO_BIT_GFX10 | HSA_DBG_WATCH_ADDR_MASK_HI_BIT; diff --git a/drivers/gpu/drm/amd/amdxcp/amdgpu_xcp_drv.c b/drivers/gpu/drm/amd/amdxcp/amdgpu_xcp_drv.c index faed84172dd4..8bc36f04b1b7 100644 --- a/drivers/gpu/drm/amd/amdxcp/amdgpu_xcp_drv.c +++ b/drivers/gpu/drm/amd/amdxcp/amdgpu_xcp_drv.c @@ -21,6 +21,7 @@ * */ +#include <linux/export.h> #include <linux/init.h> #include <linux/module.h> #include <linux/platform_device.h> diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index d3100f641ac6..3dd4f9e9931d 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -1758,10 +1758,11 @@ dm_dmub_send_vbios_gpint_command(struct amdgpu_device *adev, return DMUB_STATUS_TIMEOUT; } -static struct dml2_soc_bb *dm_dmub_get_vbios_bounding_box(struct amdgpu_device *adev) +static void *dm_dmub_get_vbios_bounding_box(struct amdgpu_device *adev) { - struct dml2_soc_bb *bb; + void *bb; long long addr; + unsigned int bb_size; int i = 0; uint16_t chunk; enum dmub_gpint_command send_addrs[] = { @@ -1774,6 +1775,7 @@ static struct dml2_soc_bb *dm_dmub_get_vbios_bounding_box(struct amdgpu_device * switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(4, 0, 1): + bb_size = sizeof(struct dml2_soc_bb); break; default: return NULL; @@ -1781,7 +1783,7 @@ static struct dml2_soc_bb *dm_dmub_get_vbios_bounding_box(struct amdgpu_device * bb = dm_allocate_gpu_mem(adev, DC_MEM_ALLOC_TYPE_GART, - sizeof(struct dml2_soc_bb), + bb_size, &addr); if (!bb) return NULL; @@ -1847,7 +1849,7 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) mutex_init(&adev->dm.audio_lock); if (amdgpu_dm_irq_init(adev)) { - drm_err(adev_to_drm(adev), "amdgpu: failed to initialize DM IRQ support.\n"); + drm_err(adev_to_drm(adev), "failed to initialize DM IRQ support.\n"); goto error; } @@ -2037,7 +2039,7 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) adev->dm.hpd_rx_offload_wq = hpd_rx_irq_create_workqueue(adev); if (!adev->dm.hpd_rx_offload_wq) { - drm_err(adev_to_drm(adev), "amdgpu: failed to create hpd rx offload workqueue.\n"); + drm_err(adev_to_drm(adev), "failed to create hpd rx offload workqueue.\n"); goto error; } @@ -2053,7 +2055,7 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) adev->dm.freesync_module = mod_freesync_create(adev->dm.dc); if (!adev->dm.freesync_module) { drm_err(adev_to_drm(adev), - "amdgpu: failed to initialize freesync_module.\n"); + "failed to initialize freesync_module.\n"); } else drm_dbg_driver(adev_to_drm(adev), "amdgpu: freesync_module init done %p.\n", adev->dm.freesync_module); @@ -2064,7 +2066,7 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) adev->dm.vblank_control_workqueue = create_singlethread_workqueue("dm_vblank_control_workqueue"); if (!adev->dm.vblank_control_workqueue) - drm_err(adev_to_drm(adev), "amdgpu: failed to initialize vblank_workqueue.\n"); + drm_err(adev_to_drm(adev), "failed to initialize vblank_workqueue.\n"); } if (adev->dm.dc->caps.ips_support && @@ -2075,7 +2077,7 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) adev->dm.hdcp_workqueue = hdcp_create_workqueue(adev, &init_params.cp_psp, adev->dm.dc); if (!adev->dm.hdcp_workqueue) - drm_err(adev_to_drm(adev), "amdgpu: failed to initialize hdcp_workqueue.\n"); + drm_err(adev_to_drm(adev), "failed to initialize hdcp_workqueue.\n"); else drm_dbg_driver(adev_to_drm(adev), "amdgpu: hdcp_workqueue init done %p.\n", adev->dm.hdcp_workqueue); @@ -2085,20 +2087,20 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) init_completion(&adev->dm.dmub_aux_transfer_done); adev->dm.dmub_notify = kzalloc(sizeof(struct dmub_notification), GFP_KERNEL); if (!adev->dm.dmub_notify) { - drm_info(adev_to_drm(adev), "amdgpu: fail to allocate adev->dm.dmub_notify"); + drm_info(adev_to_drm(adev), "fail to allocate adev->dm.dmub_notify"); goto error; } adev->dm.delayed_hpd_wq = create_singlethread_workqueue("amdgpu_dm_hpd_wq"); if (!adev->dm.delayed_hpd_wq) { - drm_err(adev_to_drm(adev), "amdgpu: failed to create hpd offload workqueue.\n"); + drm_err(adev_to_drm(adev), "failed to create hpd offload workqueue.\n"); goto error; } amdgpu_dm_outbox_init(adev); if (!register_dmub_notify_callback(adev, DMUB_NOTIFICATION_AUX_REPLY, dmub_aux_setconfig_callback, false)) { - drm_err(adev_to_drm(adev), "amdgpu: fail to register dmub aux callback"); + drm_err(adev_to_drm(adev), "fail to register dmub aux callback"); goto error; } @@ -2107,7 +2109,7 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) if (!register_dmub_notify_callback(adev, DMUB_NOTIFICATION_FUSED_IO, dmub_aux_fused_io_callback, false)) { - drm_err(adev_to_drm(adev), "amdgpu: fail to register dmub fused io callback"); + drm_err(adev_to_drm(adev), "fail to register dmub fused io callback"); goto error; } /* Enable outbox notification only after IRQ handlers are registered and DMUB is alive. @@ -2125,7 +2127,7 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) if (amdgpu_dm_initialize_drm_device(adev)) { drm_err(adev_to_drm(adev), - "amdgpu: failed to initialize sw for display support.\n"); + "failed to initialize sw for display support.\n"); goto error; } @@ -2140,14 +2142,14 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) if (drm_vblank_init(adev_to_drm(adev), adev->dm.display_indexes_num)) { drm_err(adev_to_drm(adev), - "amdgpu: failed to initialize sw for display support.\n"); + "failed to initialize sw for display support.\n"); goto error; } #if defined(CONFIG_DRM_AMD_SECURE_DISPLAY) amdgpu_dm_crtc_secure_display_create_contexts(adev); if (!adev->dm.secure_display_ctx.crtc_ctx) - drm_err(adev_to_drm(adev), "amdgpu: failed to initialize secure display contexts.\n"); + drm_err(adev_to_drm(adev), "failed to initialize secure display contexts.\n"); if (amdgpu_ip_version(adev, DCE_HWIP, 0) >= IP_VERSION(4, 0, 1)) adev->dm.secure_display_ctx.support_mul_roi = true; @@ -2404,6 +2406,7 @@ static int dm_dmub_sw_init(struct amdgpu_device *adev) DMUB_WINDOW_MEMORY_TYPE_FB, //DMUB_WINDOW_5_TRACEBUFF DMUB_WINDOW_MEMORY_TYPE_FB, //DMUB_WINDOW_6_FW_STATE DMUB_WINDOW_MEMORY_TYPE_FB, //DMUB_WINDOW_7_SCRATCH_MEM + DMUB_WINDOW_MEMORY_TYPE_FB, //DMUB_WINDOW_IB_MEM DMUB_WINDOW_MEMORY_TYPE_FB, //DMUB_WINDOW_SHARED_STATE }; int r; @@ -2570,7 +2573,7 @@ static int dm_sw_init(struct amdgpu_ip_block *ip_block) adev->dm.cgs_device = amdgpu_cgs_create_device(adev); if (!adev->dm.cgs_device) { - drm_err(adev_to_drm(adev), "amdgpu: failed to create cgs device.\n"); + drm_err(adev_to_drm(adev), "failed to create cgs device.\n"); return -EINVAL; } @@ -3060,6 +3063,77 @@ static void hpd_rx_irq_work_suspend(struct amdgpu_display_manager *dm) } } +static int dm_cache_state(struct amdgpu_device *adev) +{ + int r; + + adev->dm.cached_state = drm_atomic_helper_suspend(adev_to_drm(adev)); + if (IS_ERR(adev->dm.cached_state)) { + r = PTR_ERR(adev->dm.cached_state); + adev->dm.cached_state = NULL; + } + + return adev->dm.cached_state ? 0 : r; +} + +static void dm_destroy_cached_state(struct amdgpu_device *adev) +{ + struct amdgpu_display_manager *dm = &adev->dm; + struct drm_device *ddev = adev_to_drm(adev); + struct dm_plane_state *dm_new_plane_state; + struct drm_plane_state *new_plane_state; + struct dm_crtc_state *dm_new_crtc_state; + struct drm_crtc_state *new_crtc_state; + struct drm_plane *plane; + struct drm_crtc *crtc; + int i; + + if (!dm->cached_state) + return; + + /* Force mode set in atomic commit */ + for_each_new_crtc_in_state(dm->cached_state, crtc, new_crtc_state, i) { + new_crtc_state->active_changed = true; + dm_new_crtc_state = to_dm_crtc_state(new_crtc_state); + reset_freesync_config_for_crtc(dm_new_crtc_state); + } + + /* + * atomic_check is expected to create the dc states. We need to release + * them here, since they were duplicated as part of the suspend + * procedure. + */ + for_each_new_crtc_in_state(dm->cached_state, crtc, new_crtc_state, i) { + dm_new_crtc_state = to_dm_crtc_state(new_crtc_state); + if (dm_new_crtc_state->stream) { + WARN_ON(kref_read(&dm_new_crtc_state->stream->refcount) > 1); + dc_stream_release(dm_new_crtc_state->stream); + dm_new_crtc_state->stream = NULL; + } + dm_new_crtc_state->base.color_mgmt_changed = true; + } + + for_each_new_plane_in_state(dm->cached_state, plane, new_plane_state, i) { + dm_new_plane_state = to_dm_plane_state(new_plane_state); + if (dm_new_plane_state->dc_state) { + WARN_ON(kref_read(&dm_new_plane_state->dc_state->refcount) > 1); + dc_plane_state_release(dm_new_plane_state->dc_state); + dm_new_plane_state->dc_state = NULL; + } + } + + drm_atomic_helper_resume(ddev, dm->cached_state); + + dm->cached_state = NULL; +} + +static void dm_complete(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + + dm_destroy_cached_state(adev); +} + static int dm_prepare_suspend(struct amdgpu_ip_block *ip_block) { struct amdgpu_device *adev = ip_block->adev; @@ -3068,11 +3142,8 @@ static int dm_prepare_suspend(struct amdgpu_ip_block *ip_block) return 0; WARN_ON(adev->dm.cached_state); - adev->dm.cached_state = drm_atomic_helper_suspend(adev_to_drm(adev)); - if (IS_ERR(adev->dm.cached_state)) - return PTR_ERR(adev->dm.cached_state); - return 0; + return dm_cache_state(adev); } static int dm_suspend(struct amdgpu_ip_block *ip_block) @@ -3106,9 +3177,10 @@ static int dm_suspend(struct amdgpu_ip_block *ip_block) } if (!adev->dm.cached_state) { - adev->dm.cached_state = drm_atomic_helper_suspend(adev_to_drm(adev)); - if (IS_ERR(adev->dm.cached_state)) - return PTR_ERR(adev->dm.cached_state); + int r = dm_cache_state(adev); + + if (r) + return r; } s3_handle_hdmi_cec(adev_to_drm(adev), true); @@ -3295,12 +3367,6 @@ static int dm_resume(struct amdgpu_ip_block *ip_block) struct amdgpu_dm_connector *aconnector; struct drm_connector *connector; struct drm_connector_list_iter iter; - struct drm_crtc *crtc; - struct drm_crtc_state *new_crtc_state; - struct dm_crtc_state *dm_new_crtc_state; - struct drm_plane *plane; - struct drm_plane_state *new_plane_state; - struct dm_plane_state *dm_new_plane_state; struct dm_atomic_state *dm_state = to_dm_atomic_state(dm->atomic_obj.state); enum dc_connection_type new_connection_type = dc_connection_none; struct dc_state *dc_state; @@ -3457,40 +3523,7 @@ static int dm_resume(struct amdgpu_ip_block *ip_block) } drm_connector_list_iter_end(&iter); - /* Force mode set in atomic commit */ - for_each_new_crtc_in_state(dm->cached_state, crtc, new_crtc_state, i) { - new_crtc_state->active_changed = true; - dm_new_crtc_state = to_dm_crtc_state(new_crtc_state); - reset_freesync_config_for_crtc(dm_new_crtc_state); - } - - /* - * atomic_check is expected to create the dc states. We need to release - * them here, since they were duplicated as part of the suspend - * procedure. - */ - for_each_new_crtc_in_state(dm->cached_state, crtc, new_crtc_state, i) { - dm_new_crtc_state = to_dm_crtc_state(new_crtc_state); - if (dm_new_crtc_state->stream) { - WARN_ON(kref_read(&dm_new_crtc_state->stream->refcount) > 1); - dc_stream_release(dm_new_crtc_state->stream); - dm_new_crtc_state->stream = NULL; - } - dm_new_crtc_state->base.color_mgmt_changed = true; - } - - for_each_new_plane_in_state(dm->cached_state, plane, new_plane_state, i) { - dm_new_plane_state = to_dm_plane_state(new_plane_state); - if (dm_new_plane_state->dc_state) { - WARN_ON(kref_read(&dm_new_plane_state->dc_state->refcount) > 1); - dc_plane_state_release(dm_new_plane_state->dc_state); - dm_new_plane_state->dc_state = NULL; - } - } - - drm_atomic_helper_resume(ddev, dm->cached_state); - - dm->cached_state = NULL; + dm_destroy_cached_state(adev); /* Do mst topology probing after resuming cached state*/ drm_connector_list_iter_begin(ddev, &iter); @@ -3539,6 +3572,7 @@ static const struct amd_ip_funcs amdgpu_dm_funcs = { .prepare_suspend = dm_prepare_suspend, .suspend = dm_suspend, .resume = dm_resume, + .complete = dm_complete, .is_idle = dm_is_idle, .wait_for_idle = dm_wait_for_idle, .check_soft_reset = dm_check_soft_reset, @@ -3610,13 +3644,15 @@ static void update_connector_ext_caps(struct amdgpu_dm_connector *aconnector) luminance_range = &conn_base->display_info.luminance_range; - if (luminance_range->max_luminance) { - caps->aux_min_input_signal = luminance_range->min_luminance; + if (luminance_range->max_luminance) caps->aux_max_input_signal = luminance_range->max_luminance; - } else { - caps->aux_min_input_signal = 0; + else caps->aux_max_input_signal = 512; - } + + if (luminance_range->min_luminance) + caps->aux_min_input_signal = luminance_range->min_luminance; + else + caps->aux_min_input_signal = 1; min_input_signal_override = drm_get_panel_min_brightness_quirk(aconnector->drm_edid); if (min_input_signal_override >= 0) @@ -4001,19 +4037,19 @@ static int register_hpd_handlers(struct amdgpu_device *adev) if (dc_is_dmub_outbox_supported(adev->dm.dc)) { if (!register_dmub_notify_callback(adev, DMUB_NOTIFICATION_HPD, dmub_hpd_callback, true)) { - drm_err(adev_to_drm(adev), "amdgpu: fail to register dmub hpd callback"); + drm_err(adev_to_drm(adev), "fail to register dmub hpd callback"); return -EINVAL; } if (!register_dmub_notify_callback(adev, DMUB_NOTIFICATION_HPD_IRQ, dmub_hpd_callback, true)) { - drm_err(adev_to_drm(adev), "amdgpu: fail to register dmub hpd callback"); + drm_err(adev_to_drm(adev), "fail to register dmub hpd callback"); return -EINVAL; } if (!register_dmub_notify_callback(adev, DMUB_NOTIFICATION_HPD_SENSE_NOTIFY, dmub_hpd_sense_callback, true)) { - drm_err(adev_to_drm(adev), "amdgpu: fail to register dmub hpd sense callback"); + drm_err(adev_to_drm(adev), "fail to register dmub hpd sense callback"); return -EINVAL; } } @@ -4718,9 +4754,23 @@ static int get_brightness_range(const struct amdgpu_dm_backlight_caps *caps, return 1; } +/* Rescale from [min..max] to [0..MAX_BACKLIGHT_LEVEL] */ +static inline u32 scale_input_to_fw(int min, int max, u64 input) +{ + return DIV_ROUND_CLOSEST_ULL(input * MAX_BACKLIGHT_LEVEL, max - min); +} + +/* Rescale from [0..MAX_BACKLIGHT_LEVEL] to [min..max] */ +static inline u32 scale_fw_to_input(int min, int max, u64 input) +{ + return min + DIV_ROUND_CLOSEST_ULL(input * (max - min), MAX_BACKLIGHT_LEVEL); +} + static void convert_custom_brightness(const struct amdgpu_dm_backlight_caps *caps, - uint32_t *brightness) + unsigned int min, unsigned int max, + uint32_t *user_brightness) { + u32 brightness = scale_input_to_fw(min, max, *user_brightness); u8 prev_signal = 0, prev_lum = 0; int i = 0; @@ -4731,7 +4781,7 @@ static void convert_custom_brightness(const struct amdgpu_dm_backlight_caps *cap return; /* choose start to run less interpolation steps */ - if (caps->luminance_data[caps->data_points/2].input_signal > *brightness) + if (caps->luminance_data[caps->data_points/2].input_signal > brightness) i = caps->data_points/2; do { u8 signal = caps->luminance_data[i].input_signal; @@ -4742,17 +4792,18 @@ static void convert_custom_brightness(const struct amdgpu_dm_backlight_caps *cap * brightness < signal: interpolate between previous and current luminance numerator * brightness > signal: find next data point */ - if (*brightness > signal) { + if (brightness > signal) { prev_signal = signal; prev_lum = lum; i++; continue; } - if (*brightness < signal) + if (brightness < signal) lum = prev_lum + DIV_ROUND_CLOSEST((lum - prev_lum) * - (*brightness - prev_signal), + (brightness - prev_signal), signal - prev_signal); - *brightness = DIV_ROUND_CLOSEST(lum * *brightness, 101); + *user_brightness = scale_fw_to_input(min, max, + DIV_ROUND_CLOSEST(lum * brightness, 101)); return; } while (i < caps->data_points); } @@ -4765,11 +4816,10 @@ static u32 convert_brightness_from_user(const struct amdgpu_dm_backlight_caps *c if (!get_brightness_range(caps, &min, &max)) return brightness; - convert_custom_brightness(caps, &brightness); + convert_custom_brightness(caps, min, max, &brightness); - // Rescale 0..255 to min..max - return min + DIV_ROUND_CLOSEST((max - min) * brightness, - AMDGPU_MAX_BL_LEVEL); + // Rescale 0..max to min..max + return min + DIV_ROUND_CLOSEST_ULL((u64)(max - min) * brightness, max); } static u32 convert_brightness_to_user(const struct amdgpu_dm_backlight_caps *caps, @@ -4782,8 +4832,8 @@ static u32 convert_brightness_to_user(const struct amdgpu_dm_backlight_caps *cap if (brightness < min) return 0; - // Rescale min..max to 0..255 - return DIV_ROUND_CLOSEST(AMDGPU_MAX_BL_LEVEL * (brightness - min), + // Rescale min..max to 0..max + return DIV_ROUND_CLOSEST_ULL((u64)max * (brightness - min), max - min); } @@ -4813,6 +4863,14 @@ static void amdgpu_dm_backlight_set_level(struct amdgpu_display_manager *dm, reallow_idle = true; } + if (trace_amdgpu_dm_brightness_enabled()) { + trace_amdgpu_dm_brightness(__builtin_return_address(0), + user_brightness, + brightness, + caps->aux_support, + power_supply_is_system_supplied() > 0); + } + if (caps->aux_support) { rc = dc_link_set_backlight_level_nits(link, true, brightness, AUX_BL_DEFAULT_TRANSITION_TIME_MS); @@ -4908,7 +4966,7 @@ amdgpu_dm_register_backlight_device(struct amdgpu_dm_connector *aconnector) struct drm_device *drm = aconnector->base.dev; struct amdgpu_display_manager *dm = &drm_to_adev(drm)->dm; struct backlight_properties props = { 0 }; - struct amdgpu_dm_backlight_caps caps = { 0 }; + struct amdgpu_dm_backlight_caps *caps; char bl_name[16]; int min, max; @@ -4922,22 +4980,21 @@ amdgpu_dm_register_backlight_device(struct amdgpu_dm_connector *aconnector) return; } - amdgpu_acpi_get_backlight_caps(&caps); - if (caps.caps_valid && get_brightness_range(&caps, &min, &max)) { + caps = &dm->backlight_caps[aconnector->bl_idx]; + if (get_brightness_range(caps, &min, &max)) { if (power_supply_is_system_supplied() > 0) - props.brightness = (max - min) * DIV_ROUND_CLOSEST(caps.ac_level, 100); + props.brightness = (max - min) * DIV_ROUND_CLOSEST(caps->ac_level, 100); else - props.brightness = (max - min) * DIV_ROUND_CLOSEST(caps.dc_level, 100); + props.brightness = (max - min) * DIV_ROUND_CLOSEST(caps->dc_level, 100); /* min is zero, so max needs to be adjusted */ props.max_brightness = max - min; drm_dbg(drm, "Backlight caps: min: %d, max: %d, ac %d, dc %d\n", min, max, - caps.ac_level, caps.dc_level); + caps->ac_level, caps->dc_level); } else - props.brightness = AMDGPU_MAX_BL_LEVEL; + props.brightness = props.max_brightness = MAX_BACKLIGHT_LEVEL; - if (caps.data_points && !(amdgpu_dc_debug_mask & DC_DISABLE_CUSTOM_BRIGHTNESS_CURVE)) + if (caps->data_points && !(amdgpu_dc_debug_mask & DC_DISABLE_CUSTOM_BRIGHTNESS_CURVE)) drm_info(drm, "Using custom brightness curve\n"); - props.max_brightness = AMDGPU_MAX_BL_LEVEL; props.type = BACKLIGHT_RAW; snprintf(bl_name, sizeof(bl_name), "amdgpu_bl%d", @@ -7519,7 +7576,7 @@ static enum dc_status dm_validate_stream_and_context(struct dc *dc, dc_result = DC_FAIL_ATTACH_SURFACES; if (dc_result == DC_OK) - dc_result = dc_validate_global_state(dc, dc_state, true); + dc_result = dc_validate_global_state(dc, dc_state, DC_VALIDATE_MODE_ONLY); cleanup: if (dc_state) @@ -7577,7 +7634,7 @@ create_validate_stream_for_sink(struct drm_connector *connector, dc_result = dm_validate_stream_and_context(adev->dm.dc, stream); if (dc_result != DC_OK) { - DRM_DEBUG_KMS("Mode %dx%d (clk %d) pixel_encoding:%s color_depth:%s failed validation -- %s\n", + DRM_DEBUG_KMS("Pruned mode %d x %d (clk %d) %s %s -- %s\n", drm_mode->hdisplay, drm_mode->vdisplay, drm_mode->clock, @@ -7844,6 +7901,22 @@ static int dm_encoder_helper_atomic_check(struct drm_encoder *encoder, int clock, bpp = 0; bool is_y420 = false; + if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) { + struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder); + struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode; + enum drm_mode_status result; + + result = drm_crtc_helper_mode_valid_fixed(encoder->crtc, adjusted_mode, native_mode); + if (result != MODE_OK && dm_new_connector_state->scaling == RMX_OFF) { + drm_dbg_driver(encoder->dev, + "mode %dx%d@%dHz is not native, enabling scaling\n", + adjusted_mode->hdisplay, adjusted_mode->vdisplay, + drm_mode_vrefresh(adjusted_mode)); + dm_new_connector_state->scaling = RMX_FULL; + } + return 0; + } + if (!aconnector->mst_output_port) return 0; @@ -8301,7 +8374,7 @@ static int amdgpu_dm_connector_get_modes(struct drm_connector *connector) drm_add_modes_noedid(connector, 1920, 1080); } else { amdgpu_dm_connector_ddc_get_modes(connector, drm_edid); - if (encoder) + if (encoder && connector->connector_type != DRM_MODE_CONNECTOR_eDP) amdgpu_dm_connector_add_common_modes(encoder, connector); amdgpu_dm_connector_add_freesync_modes(connector, drm_edid); } @@ -12141,7 +12214,7 @@ static int amdgpu_dm_atomic_check(struct drm_device *dev, drm_dbg_atomic(dev, "MST drm_dp_mst_atomic_check() failed\n"); goto fail; } - status = dc_validate_global_state(dc, dm_state->context, true); + status = dc_validate_global_state(dc, dm_state->context, DC_VALIDATE_MODE_ONLY); if (status != DC_OK) { drm_dbg_atomic(dev, "DC global validation failure: %s (%d)", dc_status_to_str(status), status); diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h index d7d92f9911e4..b937da0a4e4a 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h @@ -636,8 +636,9 @@ struct amdgpu_display_manager { * @bb_from_dmub: * * Bounding box data read from dmub during early initialization for DCN4+ + * Data is stored as a byte array that should be casted to the appropriate bb struct */ - struct dml2_soc_bb *bb_from_dmub; + void *bb_from_dmub; /** * @oem_i2c: diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c index d4395b92fb85..9e3e51a2dc49 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c @@ -1029,6 +1029,10 @@ enum dc_edid_status dm_helpers_read_local_edid( return EDID_NO_RESPONSE; edid = drm_edid_raw(drm_edid); // FIXME: Get rid of drm_edid_raw() + if (!edid || + edid->extensions >= sizeof(sink->dc_edid.raw_edid) / EDID_LENGTH) + return EDID_BAD_INPUT; + sink->dc_edid.length = EDID_LENGTH * (edid->extensions + 1); memmove(sink->dc_edid.raw_edid, (uint8_t *)edid, sink->dc_edid.length); diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c index 25e8befbcc47..7187d5aedf0a 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c @@ -107,7 +107,7 @@ static ssize_t dm_dp_aux_transfer(struct drm_dp_aux *aux, if (payload.write && result >= 0) { if (result) { /*one byte indicating partially written bytes*/ - drm_dbg_dp(adev_to_drm(adev), "amdgpu: AUX partially written\n"); + drm_dbg_dp(adev_to_drm(adev), "AUX partially written\n"); result = payload.data[0]; } else if (!payload.reply[0]) /*I2C_ACK|AUX_ACK*/ @@ -133,11 +133,11 @@ static ssize_t dm_dp_aux_transfer(struct drm_dp_aux *aux, break; } - drm_dbg_dp(adev_to_drm(adev), "amdgpu: DP AUX transfer fail:%d\n", operation_result); + drm_dbg_dp(adev_to_drm(adev), "DP AUX transfer fail:%d\n", operation_result); } if (payload.reply[0]) - drm_dbg_dp(adev_to_drm(adev), "amdgpu: AUX reply command not ACK: 0x%02x.", + drm_dbg_dp(adev_to_drm(adev), "AUX reply command not ACK: 0x%02x.", payload.reply[0]); return result; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_trace.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_trace.h index 4686d4b0cbad..95f890fda8aa 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_trace.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_trace.h @@ -726,6 +726,32 @@ TRACE_EVENT(dcn_optc_lock_unlock_state, ) ); +TRACE_EVENT(amdgpu_dm_brightness, + TP_PROTO(void *function, u32 user_brightness, u32 converted_brightness, bool aux, bool ac), + TP_ARGS(function, user_brightness, converted_brightness, aux, ac), + TP_STRUCT__entry( + __field(void *, function) + __field(u32, user_brightness) + __field(u32, converted_brightness) + __field(bool, aux) + __field(bool, ac) + ), + TP_fast_assign( + __entry->function = function; + __entry->user_brightness = user_brightness; + __entry->converted_brightness = converted_brightness; + __entry->aux = aux; + __entry->ac = ac; + ), + TP_printk("%ps: brightness requested=%u converted=%u aux=%s power=%s", + (void *)__entry->function, + (u32)__entry->user_brightness, + (u32)__entry->converted_brightness, + (__entry->aux) ? "true" : "false", + (__entry->ac) ? "AC" : "DC" + ) +); + #endif /* _AMDGPU_DM_TRACE_H_ */ #undef TRACE_INCLUDE_PATH diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table2.c b/drivers/gpu/drm/amd/display/dc/bios/command_table2.c index 2c645dffec18..f2b1720a6a66 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/command_table2.c +++ b/drivers/gpu/drm/amd/display/dc/bios/command_table2.c @@ -396,6 +396,7 @@ static enum bp_result transmitter_control_v1_7( process_phy_transition_init_params.display_port_link_rate = link->cur_link_settings.link_rate; process_phy_transition_init_params.transition_bitmask = link->phy_transition_bitmask; } + dig_v1_7.skip_phy_ssc_reduction = link->wa_flags.skip_phy_ssc_reduction; } // Handle PRE_OFF_TO_ON: Process ACPI PHY Transition Interlock diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.c index a3b8e3d4a429..514a5efda102 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.c @@ -22,8 +22,6 @@ #include "dcn/dcn_4_1_0_offset.h" #include "dcn/dcn_4_1_0_sh_mask.h" -#include "dml/dcn401/dcn401_fpu.h" - #define DCN_BASE__INST0_SEG1 0x000000C0 #define mmCLK01_CLK0_CLK_PLL_REQ 0x16E37 @@ -183,43 +181,36 @@ static void dcn401_init_single_clock(struct clk_mgr_internal *clk_mgr, PPCLK_e c static void dcn401_build_wm_range_table(struct clk_mgr *clk_mgr) { - /* legacy */ - DC_FP_START(); - dcn401_build_wm_range_table_fpu(clk_mgr); - DC_FP_END(); - - if (clk_mgr->ctx->dc->debug.using_dml21) { - /* For min clocks use as reported by PM FW and report those as min */ - uint16_t min_uclk_mhz = clk_mgr->bw_params->clk_table.entries[0].memclk_mhz; - uint16_t min_dcfclk_mhz = clk_mgr->bw_params->clk_table.entries[0].dcfclk_mhz; - - /* Set A - Normal - default values */ - clk_mgr->bw_params->wm_table.nv_entries[WM_A].valid = true; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.wm_type = WATERMARKS_CLOCK_RANGE; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.min_dcfclk = min_dcfclk_mhz; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.max_dcfclk = 0xFFFF; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.min_uclk = min_uclk_mhz; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.max_uclk = 0xFFFF; + /* For min clocks use as reported by PM FW and report those as min */ + uint16_t min_uclk_mhz = clk_mgr->bw_params->clk_table.entries[0].memclk_mhz; + uint16_t min_dcfclk_mhz = clk_mgr->bw_params->clk_table.entries[0].dcfclk_mhz; - /* Set B - Unused on dcn4 */ - clk_mgr->bw_params->wm_table.nv_entries[WM_B].valid = false; + /* Set A - Normal - default values */ + clk_mgr->bw_params->wm_table.nv_entries[WM_A].valid = true; + clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.wm_type = WATERMARKS_CLOCK_RANGE; + clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.min_dcfclk = min_dcfclk_mhz; + clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.max_dcfclk = 0xFFFF; + clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.min_uclk = min_uclk_mhz; + clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.max_uclk = 0xFFFF; - /* Set 1A - Dummy P-State - P-State latency set to "dummy p-state" value */ - /* 'DalDummyClockChangeLatencyNs' registry key option set to 0x7FFFFFFF can be used to disable Set C for dummy p-state */ - if (clk_mgr->ctx->dc->bb_overrides.dummy_clock_change_latency_ns != 0x7FFFFFFF) { - clk_mgr->bw_params->wm_table.nv_entries[WM_1A].valid = true; - clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.wm_type = WATERMARKS_DUMMY_PSTATE; - clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.min_dcfclk = min_dcfclk_mhz; - clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.max_dcfclk = 0xFFFF; - clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.min_uclk = min_uclk_mhz; - clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.max_uclk = 0xFFFF; - } else { - clk_mgr->bw_params->wm_table.nv_entries[WM_1A].valid = false; - } + /* Set B - Unused on dcn4 */ + clk_mgr->bw_params->wm_table.nv_entries[WM_B].valid = false; - /* Set 1B - Unused on dcn4 */ - clk_mgr->bw_params->wm_table.nv_entries[WM_1B].valid = false; + /* Set 1A - Dummy P-State - P-State latency set to "dummy p-state" value */ + /* 'DalDummyClockChangeLatencyNs' registry key option set to 0x7FFFFFFF can be used to disable Set C for dummy p-state */ + if (clk_mgr->ctx->dc->bb_overrides.dummy_clock_change_latency_ns != 0x7FFFFFFF) { + clk_mgr->bw_params->wm_table.nv_entries[WM_1A].valid = true; + clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.wm_type = WATERMARKS_DUMMY_PSTATE; + clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.min_dcfclk = min_dcfclk_mhz; + clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.max_dcfclk = 0xFFFF; + clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.min_uclk = min_uclk_mhz; + clk_mgr->bw_params->wm_table.nv_entries[WM_1A].pmfw_breakdown.max_uclk = 0xFFFF; + } else { + clk_mgr->bw_params->wm_table.nv_entries[WM_1A].valid = false; } + + /* Set 1B - Unused on dcn4 */ + clk_mgr->bw_params->wm_table.nv_entries[WM_1B].valid = false; } void dcn401_init_clocks(struct clk_mgr *clk_mgr_base) @@ -320,6 +311,25 @@ void dcn401_init_clocks(struct clk_mgr *clk_mgr_base) dcn401_build_wm_range_table(clk_mgr_base); } +bool dcn401_is_dc_mode_present(struct clk_mgr *clk_mgr_base) +{ + struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base); + + return clk_mgr->smu_present && clk_mgr->dpm_present && + ((clk_mgr_base->bw_params->clk_table.num_entries_per_clk.num_dcfclk_levels && + clk_mgr_base->bw_params->dc_mode_limit.dcfclk_mhz) || + (clk_mgr_base->bw_params->clk_table.num_entries_per_clk.num_dispclk_levels && + clk_mgr_base->bw_params->dc_mode_limit.dispclk_mhz) || + (clk_mgr_base->bw_params->clk_table.num_entries_per_clk.num_dtbclk_levels && + clk_mgr_base->bw_params->dc_mode_limit.dtbclk_mhz) || + (clk_mgr_base->bw_params->clk_table.num_entries_per_clk.num_fclk_levels && + clk_mgr_base->bw_params->dc_mode_limit.fclk_mhz) || + (clk_mgr_base->bw_params->clk_table.num_entries_per_clk.num_memclk_levels && + clk_mgr_base->bw_params->dc_mode_limit.memclk_mhz) || + (clk_mgr_base->bw_params->clk_table.num_entries_per_clk.num_socclk_levels && + clk_mgr_base->bw_params->dc_mode_limit.socclk_mhz)); +} + static void dcn401_dump_clk_registers(struct clk_state_registers_and_bypass *regs_and_bypass, struct clk_mgr *clk_mgr_base, struct clk_log_info *log_info) { @@ -1505,6 +1515,7 @@ static struct clk_mgr_funcs dcn401_funcs = { .get_dispclk_from_dentist = dcn401_get_dispclk_from_dentist, .get_hard_min_memclk = dcn401_get_hard_min_memclk, .get_hard_min_fclk = dcn401_get_hard_min_fclk, + .is_dc_mode_present = dcn401_is_dc_mode_present, }; struct clk_mgr_internal *dcn401_clk_mgr_construct( diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.h index 6c9ae5ca2c7e..616e964df96d 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.h +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.h @@ -105,6 +105,7 @@ struct dcn401_clk_mgr { }; void dcn401_init_clocks(struct clk_mgr *clk_mgr_base); +bool dcn401_is_dc_mode_present(struct clk_mgr *clk_mgr_base); struct clk_mgr_internal *dcn401_clk_mgr_construct(struct dc_context *ctx, struct dccg *dccg); diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 56d011a1323c..c744aa9d830f 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -241,6 +241,7 @@ static bool create_links( DC_LOG_DC("BIOS object table - end"); /* Create a link for each usb4 dpia port */ + dc->lowest_dpia_link_index = MAX_LINKS; for (i = 0; i < dc->res_pool->usb4_dpia_count; i++) { struct link_init_data link_init_params = {0}; struct dc_link *link; @@ -253,6 +254,9 @@ static bool create_links( link = dc->link_srv->create_link(&link_init_params); if (link) { + if (dc->lowest_dpia_link_index > dc->link_count) + dc->lowest_dpia_link_index = dc->link_count; + dc->links[dc->link_count] = link; link->dc = dc; ++dc->link_count; @@ -2377,7 +2381,7 @@ enum dc_status dc_commit_streams(struct dc *dc, struct dc_commit_streams_params context->power_source = params->power_source; - res = dc_validate_with_context(dc, set, params->stream_count, context, false); + res = dc_validate_with_context(dc, set, params->stream_count, context, DC_VALIDATE_MODE_AND_PROGRAMMING); /* * Only update link encoder to stream assignment after bandwidth validation passed. @@ -3300,7 +3304,8 @@ static void copy_stream_update_to_stream(struct dc *dc, if (dsc_validate_context) { stream->timing.dsc_cfg = *update->dsc_config; stream->timing.flags.DSC = enable_dsc; - if (dc->res_pool->funcs->validate_bandwidth(dc, dsc_validate_context, true) != DC_OK) { + if (dc->res_pool->funcs->validate_bandwidth(dc, dsc_validate_context, + DC_VALIDATE_MODE_ONLY) != DC_OK) { stream->timing.dsc_cfg = old_dsc_cfg; stream->timing.flags.DSC = old_dsc_enabled; update->dsc_config = NULL; @@ -3522,7 +3527,7 @@ static bool update_planes_and_stream_state(struct dc *dc, } if (update_type == UPDATE_TYPE_FULL) { - if (dc->res_pool->funcs->validate_bandwidth(dc, context, false) != DC_OK) { + if (dc->res_pool->funcs->validate_bandwidth(dc, context, DC_VALIDATE_MODE_AND_PROGRAMMING) != DC_OK) { BREAK_TO_DEBUGGER(); goto fail; } @@ -4628,7 +4633,8 @@ static struct dc_state *create_minimal_transition_state(struct dc *dc, backup_and_set_minimal_pipe_split_policy(dc, base_context, policy); /* commit minimal state */ - if (dc->res_pool->funcs->validate_bandwidth(dc, minimal_transition_context, false) == DC_OK) { + if (dc->res_pool->funcs->validate_bandwidth(dc, minimal_transition_context, + DC_VALIDATE_MODE_AND_PROGRAMMING) == DC_OK) { /* prevent underflow and corruption when reconfiguring pipes */ force_vsync_flip_in_minimal_transition_context(minimal_transition_context); } else { @@ -5151,7 +5157,7 @@ static bool update_planes_and_stream_v1(struct dc *dc, copy_stream_update_to_stream(dc, context, stream, stream_update); if (update_type >= UPDATE_TYPE_FULL) { - if (dc->res_pool->funcs->validate_bandwidth(dc, context, false) != DC_OK) { + if (dc->res_pool->funcs->validate_bandwidth(dc, context, DC_VALIDATE_MODE_AND_PROGRAMMING) != DC_OK) { DC_ERROR("Mode validation failed for stream update!\n"); dc_state_release(context); return false; @@ -6337,13 +6343,14 @@ void dc_set_edp_power(const struct dc *dc, struct dc_link *edp_link, edp_link->dc->link_srv->edp_set_panel_power(edp_link, powerOn); } -/* - ***************************************************************************** +/** * dc_get_power_profile_for_dc_state() - extracts power profile from dc state * * Called when DM wants to make power policy decisions based on dc_state * - ***************************************************************************** + * @context: Pointer to the dc_state from which the power profile is extracted. + * + * Return: The power profile structure containing the power level information. */ struct dc_power_profile dc_get_power_profile_for_dc_state(const struct dc_state *context) { @@ -6359,13 +6366,14 @@ struct dc_power_profile dc_get_power_profile_for_dc_state(const struct dc_state return profile; } -/* - ********************************************************************************** +/** * dc_get_det_buffer_size_from_state() - extracts detile buffer size from dc state * - * Called when DM wants to log detile buffer size from dc_state + * This function is called to log the detile buffer size from the dc_state. + * + * @context: a pointer to the dc_state from which the detile buffer size is extracted. * - ********************************************************************************** + * Return: the size of the detile buffer, or 0 if not available. */ unsigned int dc_get_det_buffer_size_from_state(const struct dc_state *context) { @@ -6377,6 +6385,36 @@ unsigned int dc_get_det_buffer_size_from_state(const struct dc_state *context) return 0; } +/** + * dc_get_host_router_index: Get index of host router from a dpia link + * + * This function return a host router index of the target link. If the target link is dpia link. + * + * @link: Pointer to the target link (input) + * @host_router_index: Pointer to store the host router index of the target link (output). + * + * Return: true if the host router index is found and valid. + * + */ +bool dc_get_host_router_index(const struct dc_link *link, unsigned int *host_router_index) +{ + struct dc *dc; + + if (!link || !host_router_index || link->ep_type != DISPLAY_ENDPOINT_USB4_DPIA) + return false; + + dc = link->ctx->dc; + + if (link->link_index < dc->lowest_dpia_link_index) + return false; + + *host_router_index = (link->link_index - dc->lowest_dpia_link_index) / dc->caps.num_of_dpias_per_host_router; + if (*host_router_index < dc->caps.num_of_host_routers) + return true; + else + return false; +} + bool dc_is_cursor_limit_pending(struct dc *dc) { uint32_t i; diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_debug.c b/drivers/gpu/drm/amd/display/dc/core/dc_debug.c index 7551d0a3fe82..bbce751b485f 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_debug.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_debug.c @@ -268,6 +268,8 @@ char *dc_status_to_str(enum dc_status status) return "Insufficient DP link bandwidth"; case DC_FAIL_HW_CURSOR_SUPPORT: return "HW Cursor not supported"; + case DC_FAIL_DP_TUNNEL_BW_VALIDATE: + return "Fail DP Tunnel BW validation"; case DC_ERROR_UNEXPECTED: return "Unexpected error"; } diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_exports.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_exports.c index 71e15da4bb69..130455f2802a 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link_exports.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_exports.c @@ -515,7 +515,8 @@ void dc_link_enable_hpd_filter(struct dc_link *link, bool enable) link->dc->link_srv->enable_hpd_filter(link, enable); } -bool dc_link_dp_dpia_validate(struct dc *dc, const struct dc_stream_state *streams, const unsigned int count) +enum dc_status dc_link_validate_dp_tunneling_bandwidth(const struct dc *dc, const struct dc_state *new_ctx) { - return dc->link_srv->validate_dpia_bandwidth(streams, count); + return dc->link_srv->validate_dp_tunnel_bandwidth(dc, new_ctx); } + diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c index 3da25bd8b578..854fc51f159c 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c @@ -4053,7 +4053,7 @@ static bool add_all_planes_for_stream( * @set: An array of dc_validation_set with all the current streams reference * @set_count: Total of streams * @context: New context - * @fast_validate: Enable or disable fast validation + * @validate_mode: identify the validation mode * * This function updates the potential new stream in the context object. It * creates multiple lists for the add, remove, and unchanged streams. In @@ -4068,7 +4068,7 @@ enum dc_status dc_validate_with_context(struct dc *dc, const struct dc_validation_set set[], int set_count, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { struct dc_stream_state *unchanged_streams[MAX_PIPES] = { 0 }; struct dc_stream_state *del_streams[MAX_PIPES] = { 0 }; @@ -4242,7 +4242,7 @@ enum dc_status dc_validate_with_context(struct dc *dc, dc_state_set_stream_subvp_cursor_limit(context->streams[i], context, false); } - res = dc_validate_global_state(dc, context, fast_validate); + res = dc_validate_global_state(dc, context, validate_mode); /* calculate pixel rate divider after deciding pxiel clock & odm combine */ if ((dc->hwss.calculate_pix_rate_divider) && (res == DC_OK)) { @@ -4299,7 +4299,7 @@ static void decide_hblank_borrow(struct pipe_ctx *pipe_ctx) * * @dc: dc struct for this driver * @new_ctx: state to be validated - * @fast_validate: set to true if only yes/no to support matters + * @validate_mode: identify the validation mode * * Checks hardware resource availability and bandwidth requirement. * @@ -4309,7 +4309,7 @@ static void decide_hblank_borrow(struct pipe_ctx *pipe_ctx) enum dc_status dc_validate_global_state( struct dc *dc, struct dc_state *new_ctx, - bool fast_validate) + enum dc_validate_mode validate_mode) { enum dc_status result = DC_ERROR_UNEXPECTED; int i, j; @@ -4368,7 +4368,7 @@ enum dc_status dc_validate_global_state( result = resource_build_scaling_params_for_context(dc, new_ctx); if (result == DC_OK) - result = dc->res_pool->funcs->validate_bandwidth(dc, new_ctx, fast_validate); + result = dc->res_pool->funcs->validate_bandwidth(dc, new_ctx, validate_mode); return result; } diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_state.c b/drivers/gpu/drm/amd/display/dc/core/dc_state.c index 4db7383720fd..47712a4aec55 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_state.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_state.c @@ -194,11 +194,6 @@ static void init_state(struct dc *dc, struct dc_state *state) struct dc_state *dc_state_create(struct dc *dc, struct dc_state_create_params *params) { struct dc_state *state; -#ifdef CONFIG_DRM_AMD_DC_FP - struct dml2_configuration_options *dml2_opt = &dc->dml2_tmp; - - memcpy(dml2_opt, &dc->dml2_options, sizeof(dc->dml2_options)); -#endif state = kvzalloc(sizeof(struct dc_state), GFP_KERNEL); @@ -211,14 +206,12 @@ struct dc_state *dc_state_create(struct dc *dc, struct dc_state_create_params *p #ifdef CONFIG_DRM_AMD_DC_FP if (dc->debug.using_dml2) { - dml2_opt->use_clock_dc_limits = false; - if (!dml2_create(dc, dml2_opt, &state->bw_ctx.dml2)) { + if (!dml2_create(dc, &dc->dml2_options, &state->bw_ctx.dml2)) { dc_state_release(state); return NULL; } - dml2_opt->use_clock_dc_limits = true; - if (!dml2_create(dc, dml2_opt, &state->bw_ctx.dml2_dc_power_source)) { + if (!dml2_create(dc, &dc->dml2_dc_power_options, &state->bw_ctx.dml2_dc_power_source)) { dc_state_release(state); return NULL; } diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 1d917be36fc4..ce1957c7862f 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -46,6 +46,8 @@ #include "dmub/inc/dmub_cmd.h" +#include "sspl/dc_spl_types.h" + struct abm_save_restore; /* forward declaration */ @@ -53,7 +55,7 @@ struct aux_payload; struct set_config_cmd_payload; struct dmub_notification; -#define DC_VER "3.2.334" +#define DC_VER "3.2.340" /** * MAX_SURFACES - representative of the upper bound of surfaces that can be piped to a single CRTC @@ -66,7 +68,11 @@ struct dmub_notification; #define MAX_STREAMS 6 #define MIN_VIEWPORT_SIZE 12 #define MAX_NUM_EDP 2 -#define MAX_HOST_ROUTERS_NUM 2 +#define MAX_SUPPORTED_FORMATS 7 + +#define MAX_HOST_ROUTERS_NUM 3 +#define MAX_DPIA_PER_HOST_ROUTER 3 +#define MAX_DPIA_NUM (MAX_HOST_ROUTERS_NUM * MAX_DPIA_PER_HOST_ROUTER) /* Display Core Interfaces */ struct dc_versions { @@ -192,6 +198,34 @@ struct dpp_color_caps { struct rom_curve_caps ogam_rom_caps; }; +/* Below structure is to describe the HW support for mem layout, extend support + range to match what OS could handle in the roadmap */ +struct lut3d_caps { + uint32_t dma_3d_lut : 1; /*< DMA mode support for 3D LUT */ + struct { + uint32_t swizzle_3d_rgb : 1; + uint32_t swizzle_3d_bgr : 1; + uint32_t linear_1d : 1; + } mem_layout_support; + struct { + uint32_t unorm_12msb : 1; + uint32_t unorm_12lsb : 1; + uint32_t float_fp1_5_10 : 1; + } mem_format_support; + struct { + uint32_t order_rgba : 1; + uint32_t order_bgra : 1; + } mem_pixel_order_support; + /*< size options are 9, 17, 33, 45, 65 */ + struct { + uint32_t dim_9 : 1; /* 3D LUT support for 9x9x9 */ + uint32_t dim_17 : 1; /* 3D LUT support for 17x17x17 */ + uint32_t dim_33 : 1; /* 3D LUT support for 33x33x33 */ + uint32_t dim_45 : 1; /* 3D LUT support for 45x45x45 */ + uint32_t dim_65 : 1; /* 3D LUT support for 65x65x65 */ + } lut_dim_caps; +}; + /** * struct mpc_color_caps - color pipeline capabilities for multiple pipe and * plane combined blocks @@ -203,6 +237,9 @@ struct dpp_color_caps { * @shared_3d_lut: shared 3D LUT flag. Can be either DPP or MPC, but single * instance * @ogam_rom_caps: pre-definied curve caps for regamma 1D LUT + * @mcm_3d_lut_caps: HW support cap for MCM LUT memory + * @rmcm_3d_lut_caps: HW support cap for RMCM LUT memory + * @preblend: whether color manager supports preblend with MPC */ struct mpc_color_caps { uint16_t gamut_remap : 1; @@ -211,6 +248,9 @@ struct mpc_color_caps { uint16_t num_3dluts : 3; uint16_t shared_3d_lut:1; struct rom_curve_caps ogam_rom_caps; + struct lut3d_caps mcm_3d_lut_caps; + struct lut3d_caps rmcm_3d_lut_caps; + bool preblend; }; /** @@ -305,6 +345,8 @@ struct dc_caps { /* Conservative limit for DCC cases which require ODM4:1 to support*/ uint32_t dcc_plane_width_limit; struct dc_scl_caps scl_caps; + uint8_t num_of_host_routers; + uint8_t num_of_dpias_per_host_router; }; struct dc_bug_wa { @@ -481,6 +523,8 @@ struct dc_config { bool set_pipe_unlock_order; bool enable_dpia_pre_training; bool unify_link_enc_assignment; + struct spl_sharpness_range dcn_sharpness_range; + struct spl_sharpness_range dcn_override_sharpness_range; }; enum visual_confirm { @@ -786,10 +830,8 @@ union dpia_debug_options { uint32_t disable_mst_dsc_work_around:1; /* bit 3 */ uint32_t enable_force_tbt3_work_around:1; /* bit 4 */ uint32_t disable_usb4_pm_support:1; /* bit 5 */ - uint32_t enable_consolidated_dpia_dp_lt:1; /* bit 6 */ - uint32_t enable_dpia_pre_training:1; /* bit 7 */ - uint32_t unify_link_enc_assignment:1; /* bit 8 */ - uint32_t reserved:24; + uint32_t enable_usb4_bw_zero_alloc_patch:1; /* bit 6 */ + uint32_t reserved:25; } bits; uint32_t raw; }; @@ -1151,7 +1193,7 @@ struct dc_init_data { uint32_t *dcn_reg_offsets; uint32_t *nbio_reg_offsets; uint32_t *clk_reg_offsets; - struct dml2_soc_bb *bb_from_dmub; + void *bb_from_dmub; }; struct dc_callback_init { @@ -1389,6 +1431,8 @@ struct dc_plane_state { int sharpness_level; enum linear_light_scaling linear_light_scaling; unsigned int sdr_white_level_nits; + struct spl_sharpness_range sharpness_range; + enum sharpness_range_source sharpness_source; }; struct dc_plane_info { @@ -1570,6 +1614,7 @@ struct dc_scratch_space { bool blank_stream_on_ocs_change; bool read_dpcd204h_on_irq_hpd; bool force_dp_ffe_preset; + bool skip_phy_ssc_reduction; } wa_flags; union dc_dp_ffe_preset forced_dp_ffe_preset; struct link_mst_stream_allocation_table mst_stream_alloc_table; @@ -1603,6 +1648,7 @@ struct dc { uint8_t link_count; struct dc_link *links[MAX_LINKS]; + uint8_t lowest_dpia_link_index; struct link_service *link_srv; struct dc_state *current_state; @@ -1662,7 +1708,7 @@ struct dc { } scratch; struct dml2_configuration_options dml2_options; - struct dml2_configuration_options dml2_tmp; + struct dml2_configuration_options dml2_dc_power_options; enum dc_acpi_cm_power_state power_state; }; @@ -1767,19 +1813,15 @@ enum dc_status dc_validate_with_context(struct dc *dc, const struct dc_validation_set set[], int set_count, struct dc_state *context, - bool fast_validate); + enum dc_validate_mode validate_mode); bool dc_set_generic_gpio_for_stereo(bool enable, struct gpio_service *gpio_service); -/* - * fast_validate: we return after determining if we can support the new state, - * but before we populate the programming info - */ enum dc_status dc_validate_global_state( struct dc *dc, struct dc_state *new_ctx, - bool fast_validate); + enum dc_validate_mode validate_mode); bool dc_acquire_release_mpc_3dlut( struct dc *dc, bool acquire, @@ -2375,17 +2417,12 @@ void dc_link_dp_dpia_handle_usb4_bandwidth_allocation_for_link( struct dc_link *link, int peak_bw); /* - * Validate the BW of all the valid DPIA links to make sure it doesn't exceed - * available BW for each host router - * - * @dc: pointer to dc struct - * @stream: pointer to all possible streams - * @count: number of valid DPIA streams + * Calculates the DP tunneling bandwidth required for the stream timing + * and aggregates the stream bandwidth for the respective DP tunneling link * - * return: TRUE if bw used by DPIAs doesn't exceed available BW else return FALSE + * return: dc_status */ -bool dc_link_dp_dpia_validate(struct dc *dc, const struct dc_stream_state *streams, - const unsigned int count); +enum dc_status dc_link_validate_dp_tunneling_bandwidth(const struct dc *dc, const struct dc_state *new_ctx); /* Sink Interfaces - A sink corresponds to a display output device */ @@ -2595,6 +2632,8 @@ struct dc_power_profile dc_get_power_profile_for_dc_state(const struct dc_state unsigned int dc_get_det_buffer_size_from_state(const struct dc_state *context); +bool dc_get_host_router_index(const struct dc_link *link, unsigned int *host_router_index); + /* DSC Interfaces */ #include "dc_dsc.h" diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c index afbcf866520e..0a47d1a3515b 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c +++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c @@ -1656,7 +1656,7 @@ bool dc_wake_and_execute_gpint(const struct dc_context *ctx, enum dmub_gpint_com return result; } -void dc_dmub_srv_fams2_update_config(struct dc *dc, +static void dc_dmub_srv_rb_based_fams2_update_config(struct dc *dc, struct dc_state *context, bool enable) { @@ -1722,6 +1722,63 @@ void dc_dmub_srv_fams2_update_config(struct dc *dc, dm_execute_dmub_cmd_list(dc->ctx, num_cmds, cmd, DM_DMUB_WAIT_TYPE_WAIT); } +static void dc_dmub_srv_ib_based_fams2_update_config(struct dc *dc, + struct dc_state *context, + bool enable) +{ + struct dmub_fams2_config_v2 *config = (struct dmub_fams2_config_v2 *)dc->ctx->dmub_srv->dmub->ib_mem_gart.cpu_addr; + union dmub_rb_cmd cmd; + uint32_t i; + + memset(config, 0, sizeof(*config)); + memset(&cmd, 0, sizeof(cmd)); + + cmd.ib_fams2_config.header.type = DMUB_CMD__FW_ASSISTED_MCLK_SWITCH; + cmd.ib_fams2_config.header.sub_type = DMUB_CMD__FAMS2_IB_CONFIG; + + cmd.ib_fams2_config.ib_data.src.quad_part = dc->ctx->dmub_srv->dmub->ib_mem_gart.gpu_addr; + cmd.ib_fams2_config.ib_data.size = sizeof(*config); + + if (enable && context->bw_ctx.bw.dcn.fams2_global_config.features.bits.enable) { + /* copy static feature configuration overrides */ + config->global.features.bits.enable_stall_recovery = dc->debug.fams2_config.bits.enable_stall_recovery; + config->global.features.bits.enable_offload_flip = dc->debug.fams2_config.bits.enable_offload_flip; + config->global.features.bits.enable_debug = dc->debug.fams2_config.bits.enable_debug; + + /* send global configuration parameters */ + memcpy(&config->global, &context->bw_ctx.bw.dcn.fams2_global_config, + sizeof(struct dmub_cmd_fams2_global_config)); + + /* construct per-stream configs */ + for (i = 0; i < context->bw_ctx.bw.dcn.fams2_global_config.num_streams; i++) { + /* copy stream static base state */ + memcpy(&config->stream_v1[i].base, + &context->bw_ctx.bw.dcn.fams2_stream_base_params[i], + sizeof(config->stream_v1[i].base)); + + /* copy stream static sub-state */ + memcpy(&config->stream_v1[i].sub_state, + &context->bw_ctx.bw.dcn.fams2_stream_sub_params_v2[i], + sizeof(config->stream_v1[i].sub_state)); + } + } + + config->global.features.bits.enable_visual_confirm = dc->debug.visual_confirm == VISUAL_CONFIRM_FAMS2; + config->global.features.bits.enable = enable; + + dm_execute_dmub_cmd_list(dc->ctx, 1, &cmd, DM_DMUB_WAIT_TYPE_WAIT); +} + +void dc_dmub_srv_fams2_update_config(struct dc *dc, + struct dc_state *context, + bool enable) +{ + if (dc->debug.fams_version.major == 2) + dc_dmub_srv_rb_based_fams2_update_config(dc, context, enable); + if (dc->debug.fams_version.major == 3) + dc_dmub_srv_ib_based_fams2_update_config(dc, context, enable); +} + void dc_dmub_srv_fams2_drr_update(struct dc *dc, uint32_t tg_inst, uint32_t vtotal_min, @@ -1847,83 +1904,250 @@ void dc_dmub_srv_fams2_passthrough_flip( } } -bool dc_dmub_srv_ips_residency_cntl(struct dc_dmub_srv *dc_dmub_srv, bool start_measurement) + +bool dc_dmub_srv_ips_residency_cntl(const struct dc_context *ctx, uint8_t panel_inst, bool start_measurement) { - bool result; + union dmub_rb_cmd cmd; - if (!dc_dmub_srv || !dc_dmub_srv->dmub) + memset(&cmd, 0, sizeof(cmd)); + + cmd.ips_residency_cntl.header.type = DMUB_CMD__IPS; + cmd.ips_residency_cntl.header.sub_type = DMUB_CMD__IPS_RESIDENCY_CNTL; + cmd.ips_residency_cntl.header.payload_bytes = sizeof(struct dmub_cmd_ips_residency_cntl_data); + + // only panel_inst=0 is supported at the moment + cmd.ips_residency_cntl.cntl_data.panel_inst = panel_inst; + cmd.ips_residency_cntl.cntl_data.start_measurement = start_measurement; + + if (!dc_wake_and_execute_dmub_cmd(ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY)) return false; - result = dc_wake_and_execute_gpint(dc_dmub_srv->ctx, DMUB_GPINT__IPS_RESIDENCY, - start_measurement, NULL, DM_DMUB_WAIT_TYPE_WAIT); + return true; +} + +bool dc_dmub_srv_ips_query_residency_info(const struct dc_context *ctx, uint8_t panel_inst, struct dmub_ips_residency_info *driver_info, + enum ips_residency_mode ips_mode) +{ + union dmub_rb_cmd cmd; + uint32_t bytes = sizeof(struct dmub_ips_residency_info); + + dmub_flush_buffer_mem(&ctx->dmub_srv->dmub->scratch_mem_fb); + memset(&cmd, 0, sizeof(cmd)); + + cmd.ips_query_residency_info.header.type = DMUB_CMD__IPS; + cmd.ips_query_residency_info.header.sub_type = DMUB_CMD__IPS_QUERY_RESIDENCY_INFO; + cmd.ips_query_residency_info.header.payload_bytes = sizeof(struct dmub_cmd_ips_query_residency_info_data); + + cmd.ips_query_residency_info.info_data.dest.quad_part = ctx->dmub_srv->dmub->scratch_mem_fb.gpu_addr; + cmd.ips_query_residency_info.info_data.size = bytes; + cmd.ips_query_residency_info.info_data.panel_inst = panel_inst; + cmd.ips_query_residency_info.info_data.ips_mode = (uint32_t)ips_mode; + + if (!dc_wake_and_execute_dmub_cmd(ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY) || + cmd.ips_query_residency_info.header.ret_status == 0) + return false; + + // copy the result to the output since ret_status != 0 means the command returned data + memcpy(driver_info, ctx->dmub_srv->dmub->scratch_mem_fb.cpu_addr, bytes); + + return true; +} + +bool dmub_lsdma_init(struct dc_dmub_srv *dc_dmub_srv) +{ + struct dc_context *dc_ctx = dc_dmub_srv->ctx; + union dmub_rb_cmd cmd; + enum dm_dmub_wait_type wait_type; + struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data; + bool result; + + memset(&cmd, 0, sizeof(cmd)); + + cmd.cmd_common.header.type = DMUB_CMD__LSDMA; + cmd.cmd_common.header.sub_type = DMUB_CMD__LSDMA_INIT_CONFIG; + wait_type = DM_DMUB_WAIT_TYPE_NO_WAIT; + + lsdma_data->u.init_data.gpu_addr_base.quad_part = dc_ctx->dmub_srv->dmub->lsdma_rb_fb.gpu_addr; + lsdma_data->u.init_data.ring_size = dc_ctx->dmub_srv->dmub->lsdma_rb_fb.size; + + result = dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, wait_type); + + if (!result) + DC_ERROR("LSDMA Init failed in DMUB"); return result; } -void dc_dmub_srv_ips_query_residency_info(struct dc_dmub_srv *dc_dmub_srv, struct ips_residency_info *output) +bool dmub_lsdma_send_linear_copy_packet( + struct dc_dmub_srv *dc_dmub_srv, + uint64_t src_addr, + uint64_t dst_addr, + uint32_t count) { - uint32_t i; - enum dmub_gpint_command command_code; + struct dc_context *dc_ctx = dc_dmub_srv->ctx; + union dmub_rb_cmd cmd; + enum dm_dmub_wait_type wait_type; + struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data; + bool result; - if (!dc_dmub_srv || !dc_dmub_srv->dmub) - return; + memset(&cmd, 0, sizeof(cmd)); - switch (output->ips_mode) { - case DMUB_IPS_MODE_IPS1_MAX: - command_code = DMUB_GPINT__GET_IPS1_HISTOGRAM_COUNTER; - break; - case DMUB_IPS_MODE_IPS2: - command_code = DMUB_GPINT__GET_IPS2_HISTOGRAM_COUNTER; - break; - case DMUB_IPS_MODE_IPS1_RCG: - command_code = DMUB_GPINT__GET_IPS1_RCG_HISTOGRAM_COUNTER; - break; - case DMUB_IPS_MODE_IPS1_ONO2_ON: - command_code = DMUB_GPINT__GET_IPS1_ONO2_ON_HISTOGRAM_COUNTER; - break; - default: - command_code = DMUB_GPINT__INVALID_COMMAND; - break; - } + cmd.cmd_common.header.type = DMUB_CMD__LSDMA; + cmd.cmd_common.header.sub_type = DMUB_CMD__LSDMA_LINEAR_COPY; + wait_type = DM_DMUB_WAIT_TYPE_NO_WAIT; - if (command_code == DMUB_GPINT__INVALID_COMMAND) - return; + lsdma_data->u.linear_copy_data.count = count - 1; // LSDMA controller expects bytes to copy -1 + lsdma_data->u.linear_copy_data.src_lo = src_addr & 0xFFFFFFFF; + lsdma_data->u.linear_copy_data.src_hi = (src_addr >> 32) & 0xFFFFFFFF; + lsdma_data->u.linear_copy_data.dst_lo = dst_addr & 0xFFFFFFFF; + lsdma_data->u.linear_copy_data.dst_hi = (dst_addr >> 32) & 0xFFFFFFFF; - for (i = 0; i < GPINT_RETRY_NUM; i++) { - // false could mean GPINT timeout, in which case we should retry - if (dc_wake_and_execute_gpint(dc_dmub_srv->ctx, DMUB_GPINT__GET_IPS_RESIDENCY_PERCENT, - (uint16_t)(output->ips_mode), &output->residency_percent, - DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY)) - break; - udelay(100); - } + result = dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, wait_type); + + if (!result) + DC_ERROR("LSDMA Linear Copy failed in DMUB"); + + return result; +} + +bool dmub_lsdma_send_tiled_to_tiled_copy_command( + struct dc_dmub_srv *dc_dmub_srv, + struct lsdma_send_tiled_to_tiled_copy_command_params params) +{ + struct dc_context *dc_ctx = dc_dmub_srv->ctx; + union dmub_rb_cmd cmd; + enum dm_dmub_wait_type wait_type; + struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data; + bool result; + + memset(&cmd, 0, sizeof(cmd)); + + cmd.cmd_common.header.type = DMUB_CMD__LSDMA; + cmd.cmd_common.header.sub_type = DMUB_CMD__LSDMA_TILED_TO_TILED_COPY; + wait_type = DM_DMUB_WAIT_TYPE_NO_WAIT; + + lsdma_data->u.tiled_copy_data.src_addr_lo = params.src_addr & 0xFFFFFFFF; + lsdma_data->u.tiled_copy_data.src_addr_hi = (params.src_addr >> 32) & 0xFFFFFFFF; + lsdma_data->u.tiled_copy_data.dst_addr_lo = params.dst_addr & 0xFFFFFFFF; + lsdma_data->u.tiled_copy_data.dst_addr_hi = (params.dst_addr >> 32) & 0xFFFFFFFF; + lsdma_data->u.tiled_copy_data.src_x = params.src_x; + lsdma_data->u.tiled_copy_data.src_y = params.src_y; + lsdma_data->u.tiled_copy_data.dst_x = params.dst_x; + lsdma_data->u.tiled_copy_data.dst_y = params.dst_y; + lsdma_data->u.tiled_copy_data.src_width = params.src_width - 1; // LSDMA controller expects width -1 + lsdma_data->u.tiled_copy_data.dst_width = params.dst_width - 1; // LSDMA controller expects width -1 + lsdma_data->u.tiled_copy_data.src_swizzle_mode = params.swizzle_mode; + lsdma_data->u.tiled_copy_data.dst_swizzle_mode = params.swizzle_mode; + lsdma_data->u.tiled_copy_data.src_element_size = params.element_size; + lsdma_data->u.tiled_copy_data.dst_element_size = params.element_size; + lsdma_data->u.tiled_copy_data.rect_x = params.rect_x; + lsdma_data->u.tiled_copy_data.rect_y = params.rect_y; + lsdma_data->u.tiled_copy_data.dcc = params.dcc; + lsdma_data->u.tiled_copy_data.tmz = params.tmz; + lsdma_data->u.tiled_copy_data.read_compress = params.read_compress; + lsdma_data->u.tiled_copy_data.write_compress = params.write_compress; + lsdma_data->u.tiled_copy_data.src_height = params.src_height - 1; // LSDMA controller expects height -1 + lsdma_data->u.tiled_copy_data.dst_height = params.dst_height - 1; // LSDMA controller expects height -1 + lsdma_data->u.tiled_copy_data.data_format = params.data_format; + lsdma_data->u.tiled_copy_data.max_com = params.max_com; + lsdma_data->u.tiled_copy_data.max_uncom = params.max_uncom; + + result = dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, wait_type); + + if (!result) + DC_ERROR("LSDMA Tiled to Tiled Copy failed in DMUB"); + + return result; +} + +bool dmub_lsdma_send_pio_copy_command( + struct dc_dmub_srv *dc_dmub_srv, + uint64_t src_addr, + uint64_t dst_addr, + uint32_t byte_count, + uint32_t overlap_disable) +{ + struct dc_context *dc_ctx = dc_dmub_srv->ctx; + union dmub_rb_cmd cmd; + enum dm_dmub_wait_type wait_type; + struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data; + bool result; - if (!dc_wake_and_execute_gpint(dc_dmub_srv->ctx, DMUB_GPINT__GET_IPS_RESIDENCY_ENTRY_COUNTER, - (uint16_t)(output->ips_mode), - &output->entry_counter, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY)) - output->entry_counter = 0; + memset(&cmd, 0, sizeof(cmd)); + + cmd.cmd_common.header.type = DMUB_CMD__LSDMA; + cmd.cmd_common.header.sub_type = DMUB_CMD__LSDMA_PIO_COPY; + wait_type = DM_DMUB_WAIT_TYPE_NO_WAIT; + + lsdma_data->u.pio_copy_data.packet.fields.byte_count = byte_count; + lsdma_data->u.pio_copy_data.packet.fields.overlap_disable = overlap_disable; + lsdma_data->u.pio_copy_data.src_lo = src_addr & 0xFFFFFFFF; + lsdma_data->u.pio_copy_data.src_hi = (src_addr >> 32) & 0xFFFFFFFF; + lsdma_data->u.pio_copy_data.dst_lo = dst_addr & 0xFFFFFFFF; + lsdma_data->u.pio_copy_data.dst_hi = (dst_addr >> 32) & 0xFFFFFFFF; - if (!dc_wake_and_execute_gpint(dc_dmub_srv->ctx, DMUB_GPINT__GET_IPS_RESIDENCY_DURATION_US_LO, - (uint16_t)(output->ips_mode), - &output->total_active_time_us[0], DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY)) - output->total_active_time_us[0] = 0; - if (!dc_wake_and_execute_gpint(dc_dmub_srv->ctx, DMUB_GPINT__GET_IPS_RESIDENCY_DURATION_US_HI, - (uint16_t)(output->ips_mode), - &output->total_active_time_us[1], DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY)) - output->total_active_time_us[1] = 0; + result = dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, wait_type); - if (!dc_wake_and_execute_gpint(dc_dmub_srv->ctx, DMUB_GPINT__GET_IPS_INACTIVE_RESIDENCY_DURATION_US_LO, - (uint16_t)(output->ips_mode), - &output->total_inactive_time_us[0], DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY)) - output->total_inactive_time_us[0] = 0; - if (!dc_wake_and_execute_gpint(dc_dmub_srv->ctx, DMUB_GPINT__GET_IPS_INACTIVE_RESIDENCY_DURATION_US_HI, - (uint16_t)(output->ips_mode), - &output->total_inactive_time_us[1], DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY)) - output->total_inactive_time_us[1] = 0; + if (!result) + DC_ERROR("LSDMA PIO Copy failed in DMUB"); - // NUM_IPS_HISTOGRAM_BUCKETS = 16 - for (i = 0; i < 16; i++) - if (!dc_wake_and_execute_gpint(dc_dmub_srv->ctx, command_code, i, &output->histogram[i], - DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY)) - output->histogram[i] = 0; + return result; } + +bool dmub_lsdma_send_pio_constfill_command( + struct dc_dmub_srv *dc_dmub_srv, + uint64_t dst_addr, + uint32_t byte_count, + uint32_t data) +{ + struct dc_context *dc_ctx = dc_dmub_srv->ctx; + union dmub_rb_cmd cmd; + enum dm_dmub_wait_type wait_type; + struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data; + bool result; + + memset(&cmd, 0, sizeof(cmd)); + + cmd.cmd_common.header.type = DMUB_CMD__LSDMA; + cmd.cmd_common.header.sub_type = DMUB_CMD__LSDMA_PIO_CONSTFILL; + wait_type = DM_DMUB_WAIT_TYPE_NO_WAIT; + + lsdma_data->u.pio_constfill_data.packet.fields.constant_fill = 1; + lsdma_data->u.pio_constfill_data.packet.fields.byte_count = byte_count; + lsdma_data->u.pio_constfill_data.dst_lo = dst_addr & 0xFFFFFFFF; + lsdma_data->u.pio_constfill_data.dst_hi = (dst_addr >> 32) & 0xFFFFFFFF; + lsdma_data->u.pio_constfill_data.data = data; + + result = dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, wait_type); + + if (!result) + DC_ERROR("LSDMA PIO Constfill failed in DMUB"); + + return result; +} + +bool dmub_lsdma_send_poll_reg_write_command(struct dc_dmub_srv *dc_dmub_srv, uint32_t reg_addr, uint32_t reg_data) +{ + struct dc_context *dc_ctx = dc_dmub_srv->ctx; + union dmub_rb_cmd cmd; + enum dm_dmub_wait_type wait_type; + struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data; + bool result; + + memset(&cmd, 0, sizeof(cmd)); + + cmd.cmd_common.header.type = DMUB_CMD__LSDMA; + cmd.cmd_common.header.sub_type = DMUB_CMD__LSDMA_POLL_REG_WRITE; + wait_type = DM_DMUB_WAIT_TYPE_NO_WAIT; + + lsdma_data->u.reg_write_data.reg_addr = reg_addr; + lsdma_data->u.reg_write_data.reg_data = reg_data; + + result = dc_wake_and_execute_dmub_cmd(dc_ctx, &cmd, wait_type); + + if (!result) + DC_ERROR("LSDMA Poll Reg failed in DMUB"); + + return result; +} + diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h index ada5c2fb2db3..1f1c155be30e 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h +++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h @@ -210,6 +210,60 @@ void dc_dmub_srv_fams2_passthrough_flip( struct dc_surface_update *srf_updates, int surface_count); +bool dmub_lsdma_init(struct dc_dmub_srv *dc_dmub_srv); +bool dmub_lsdma_send_linear_copy_packet( + struct dc_dmub_srv *dc_dmub_srv, + uint64_t src_addr, + uint64_t dst_addr, + uint32_t count); +bool dmub_lsdma_send_pio_copy_command( + struct dc_dmub_srv *dc_dmub_srv, + uint64_t src_addr, + uint64_t dst_addr, + uint32_t byte_count, + uint32_t overlap_disable); +bool dmub_lsdma_send_pio_constfill_command( + struct dc_dmub_srv *dc_dmub_srv, + uint64_t dst_addr, + uint32_t byte_count, + uint32_t data); + +struct lsdma_send_tiled_to_tiled_copy_command_params { + uint64_t src_addr; + uint64_t dst_addr; + + uint32_t src_x : 16; + uint32_t src_y : 16; + + uint32_t dst_x : 16; + uint32_t dst_y : 16; + + uint32_t src_width : 16; + uint32_t dst_width : 16; + + uint32_t rect_x : 16; + uint32_t rect_y : 16; + + uint32_t src_height : 16; + uint32_t dst_height : 16; + + uint32_t data_format : 6; + uint32_t swizzle_mode : 5; + uint32_t element_size : 3; + uint32_t dcc : 1; + uint32_t tmz : 1; + uint32_t read_compress : 2; + uint32_t write_compress : 2; + uint32_t max_com : 2; + uint32_t max_uncom : 1; + uint32_t padding : 9; +}; + +bool dmub_lsdma_send_tiled_to_tiled_copy_command( + struct dc_dmub_srv *dc_dmub_srv, + struct lsdma_send_tiled_to_tiled_copy_command_params params); +bool dmub_lsdma_send_poll_reg_write_command(struct dc_dmub_srv *dc_dmub_srv, uint32_t reg_addr, uint32_t reg_data); + /** * struct ips_residency_info - struct containing info from dmub_ips_residency_stats * @@ -223,7 +277,7 @@ void dc_dmub_srv_fams2_passthrough_flip( * @histogram: Histogram of given IPS state durations - bucket definitions in dmub_ips.c */ struct ips_residency_info { - enum dmub_ips_mode ips_mode; + enum ips_residency_mode ips_mode; unsigned int residency_percent; unsigned int entry_counter; unsigned int total_active_time_us[2]; @@ -231,21 +285,10 @@ struct ips_residency_info { unsigned int histogram[16]; }; -/** - * bool dc_dmub_srv_ips_residency_cntl() - Controls IPS residency measurement status - * - * @dc_dmub_srv: The DC DMUB service pointer - * @start_measurement: Describes whether to start or stop measurement - * - * Return: true if GPINT was sent successfully, false otherwise - */ -bool dc_dmub_srv_ips_residency_cntl(struct dc_dmub_srv *dc_dmub_srv, bool start_measurement); +bool dc_dmub_srv_ips_residency_cntl(const struct dc_context *ctx, uint8_t panel_inst, bool start_measurement); + +bool dc_dmub_srv_ips_query_residency_info(const struct dc_context *ctx, uint8_t panel_inst, + struct dmub_ips_residency_info *driver_info, + enum ips_residency_mode ips_mode); -/** - * bool dc_dmub_srv_ips_query_residency_info() - Queries DMCUB for residency info - * - * @dc_dmub_srv: The DC DMUB service pointer - * @output: Output struct to copy the the residency info to - */ -void dc_dmub_srv_ips_query_residency_info(struct dc_dmub_srv *dc_dmub_srv, struct ips_residency_info *output); #endif /* _DMUB_DC_SRV_H_ */ diff --git a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h index 0bad8304ccf6..5ce1be362534 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h @@ -162,6 +162,11 @@ struct dc_link_settings { struct dc_tunnel_settings { bool should_enable_dp_tunneling; bool should_use_dp_bw_allocation; + uint8_t cm_id; + uint8_t group_id; + uint32_t bw_granularity; + uint32_t estimated_bw; + uint32_t allocated_bw; }; union dc_dp_ffe_preset { @@ -957,11 +962,21 @@ union usb4_driver_bw_cap { uint8_t raw; }; +/* DPCD[0xE0021] DP_IN_ADAPTER_TUNNEL_INFORMATION register. */ +union dpia_tunnel_info { + struct { + uint8_t group_id :3; + uint8_t rsvd :5; + } bits; + uint8_t raw; +}; + /* DP Tunneling over USB4 */ struct dpcd_usb4_dp_tunneling_info { union dp_tun_cap_support dp_tun_cap; union dpia_info dpia_info; union usb4_driver_bw_cap driver_bw_cap; + union dpia_tunnel_info dpia_tunnel_info; uint8_t usb4_driver_id; uint8_t usb4_topology_id[DPCD_USB4_TOPOLOGY_ID_LEN]; }; @@ -1172,8 +1187,8 @@ struct dc_lttpr_caps { union dp_128b_132b_supported_lttpr_link_rates supported_128b_132b_rates; union dp_alpm_lttpr_cap alpm; uint8_t aux_rd_interval[MAX_REPEATER_CNT - 1]; - uint8_t lttpr_ieee_oui[3]; - uint8_t lttpr_device_id[6]; + uint8_t lttpr_ieee_oui[3]; // Always read from closest LTTPR to host + uint8_t lttpr_device_id[6]; // Always read from closest LTTPR to host }; struct dc_dongle_dfp_cap_ext { diff --git a/drivers/gpu/drm/amd/display/dc/dc_hw_types.h b/drivers/gpu/drm/amd/display/dc/dc_hw_types.h index d562ddeca512..667852517246 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_hw_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_hw_types.h @@ -68,7 +68,7 @@ enum dc_plane_addr_type { struct dc_plane_address { enum dc_plane_addr_type type; - bool tmz_surface; + uint8_t tmz_surface; union { struct{ PHYSICAL_ADDRESS_LOC addr; @@ -974,6 +974,7 @@ struct dc_crtc_timing { uint32_t pix_clk_100hz; uint32_t min_refresh_in_uhz; + uint32_t max_refresh_in_uhz; uint32_t vic; uint32_t hdmi_vic; @@ -1103,7 +1104,8 @@ enum mpcc_gamut_remap_mode_select { enum mpcc_gamut_remap_id { MPCC_OGAM_GAMUT_REMAP, MPCC_MCM_FIRST_GAMUT_REMAP, - MPCC_MCM_SECOND_GAMUT_REMAP + MPCC_MCM_SECOND_GAMUT_REMAP, + MPCC_RMCM_GAMUT_REMAP, }; enum cursor_matrix_mode { diff --git a/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c b/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c index e3a8283b4098..7f57661433eb 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c +++ b/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c @@ -156,15 +156,16 @@ void translate_SPL_in_params_from_pipe_ctx(struct pipe_ctx *pipe_ctx, struct spl spl_in->adaptive_sharpness.enable = true; spl_in->adaptive_sharpness.sharpness_level = 0; } else if (sharpness_setting == SHARPNESS_CUSTOM) { - spl_in->adaptive_sharpness.sharpness_range.sdr_rgb_min = 0; - spl_in->adaptive_sharpness.sharpness_range.sdr_rgb_max = 1750; - spl_in->adaptive_sharpness.sharpness_range.sdr_rgb_mid = 750; - spl_in->adaptive_sharpness.sharpness_range.sdr_yuv_min = 0; - spl_in->adaptive_sharpness.sharpness_range.sdr_yuv_max = 3500; - spl_in->adaptive_sharpness.sharpness_range.sdr_yuv_mid = 1500; - spl_in->adaptive_sharpness.sharpness_range.hdr_rgb_min = 0; - spl_in->adaptive_sharpness.sharpness_range.hdr_rgb_max = 2750; - spl_in->adaptive_sharpness.sharpness_range.hdr_rgb_mid = 1500; + /* SAT: read harpness_range from dc_plane_state */ + spl_in->adaptive_sharpness.sharpness_range.sdr_rgb_min = plane_state->sharpness_range.sdr_rgb_min; + spl_in->adaptive_sharpness.sharpness_range.sdr_rgb_max = plane_state->sharpness_range.sdr_rgb_max; + spl_in->adaptive_sharpness.sharpness_range.sdr_rgb_mid = plane_state->sharpness_range.sdr_rgb_mid; + spl_in->adaptive_sharpness.sharpness_range.sdr_yuv_min = plane_state->sharpness_range.sdr_yuv_min; + spl_in->adaptive_sharpness.sharpness_range.sdr_yuv_max = plane_state->sharpness_range.sdr_yuv_max; + spl_in->adaptive_sharpness.sharpness_range.sdr_yuv_mid = plane_state->sharpness_range.sdr_yuv_mid; + spl_in->adaptive_sharpness.sharpness_range.hdr_rgb_min = plane_state->sharpness_range.hdr_rgb_min; + spl_in->adaptive_sharpness.sharpness_range.hdr_rgb_max = plane_state->sharpness_range.hdr_rgb_max; + spl_in->adaptive_sharpness.sharpness_range.hdr_rgb_mid = plane_state->sharpness_range.hdr_rgb_mid; if (force_sharpness_level > 0) { if (force_sharpness_level > 10) diff --git a/drivers/gpu/drm/amd/display/dc/dc_types.h b/drivers/gpu/drm/amd/display/dc/dc_types.h index a4cd0eb39a3a..b203ed020cd5 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_types.h @@ -1255,7 +1255,6 @@ enum dc_cm2_gpu_mem_layout { enum dc_cm2_gpu_mem_pixel_component_order { DC_CM2_GPU_MEM_PIXEL_COMPONENT_ORDER_RGBA, - DC_CM2_GPU_MEM_PIXEL_COMPONENT_ORDER_BGRA }; enum dc_cm2_gpu_mem_format { @@ -1277,7 +1276,6 @@ struct dc_cm2_gpu_mem_format_parameters { enum dc_cm2_gpu_mem_size { DC_CM2_GPU_MEM_SIZE_171717, - DC_CM2_GPU_MEM_SIZE_333333, DC_CM2_GPU_MEM_SIZE_TRANSFORMED, }; @@ -1372,4 +1370,19 @@ struct set_backlight_level_params { uint8_t aux_inst; }; +enum dc_validate_mode { + /* validate the mode and program HW */ + DC_VALIDATE_MODE_AND_PROGRAMMING = 0, + /* only validate the mode */ + DC_VALIDATE_MODE_ONLY = 1, + /* validate the mode and get the max state (voltage level) */ + DC_VALIDATE_MODE_AND_STATE_INDEX = 2, +}; + +struct dc_validation_dpia_set { + const struct dc_link *link; + const struct dc_tunnel_settings *tunnel_settings; + uint32_t required_bw; +}; + #endif /* DC_TYPES_H_ */ diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c index ffd172231fdf..668ee2d405fd 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c @@ -727,7 +727,7 @@ void dccg401_init(struct dccg *dccg) } } -void dccg401_set_dto_dscclk(struct dccg *dccg, uint32_t inst) +void dccg401_set_dto_dscclk(struct dccg *dccg, uint32_t inst, uint32_t num_slices_h) { struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg); diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.h index 55e8718aad22..5947a35363aa 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.h +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.h @@ -209,7 +209,7 @@ void dccg401_disable_symclk32_le( struct dccg *dccg, int hpo_le_inst); void dccg401_disable_dpstreamclk(struct dccg *dccg, int dp_hpo_inst); -void dccg401_set_dto_dscclk(struct dccg *dccg, uint32_t inst); +void dccg401_set_dto_dscclk(struct dccg *dccg, uint32_t inst, uint32_t num_slices_h); void dccg401_set_ref_dscclk(struct dccg *dccg, uint32_t dsc_inst); void dccg401_set_src_sel( @@ -230,7 +230,6 @@ void dccg401_set_dp_dto( const struct dp_dto_params *params); void dccg401_enable_symclk_se(struct dccg *dccg, uint32_t stream_enc_inst, uint32_t link_enc_inst); void dccg401_disable_symclk_se(struct dccg *dccg, uint32_t stream_enc_inst, uint32_t link_enc_inst); -void dccg401_set_dto_dscclk(struct dccg *dccg, uint32_t inst); void dccg401_set_dtbclk_p_src( struct dccg *dccg, enum streamclk_source src, diff --git a/drivers/gpu/drm/amd/display/dc/dml/Makefile b/drivers/gpu/drm/amd/display/dc/dml/Makefile index e1d500633dfa..b357683b4255 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dml/Makefile @@ -114,9 +114,6 @@ CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/calcs/dcn_calcs.o := $(dml_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/calcs/dcn_calc_auto.o := $(dml_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/calcs/dcn_calc_math.o := $(dml_rcflags) -CFLAGS_$(AMDDALPATH)/dc/dml/dcn401/dcn401_fpu.o := $(dml_ccflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn401/dcn401_fpu.o := $(dml_rcflags) - ifdef CONFIG_DRM_AMD_DC_FP DML += display_mode_lib.o display_rq_dlg_helpers.o dml1_display_rq_dlg_calc.o DML += dcn10/dcn10_fpu.o @@ -137,7 +134,6 @@ DML += dcn303/dcn303_fpu.o DML += dcn314/dcn314_fpu.o DML += dcn35/dcn35_fpu.o DML += dcn351/dcn351_fpu.o -DML += dcn401/dcn401_fpu.o DML += dsc/rc_calc_fpu.o DML += calcs/dcn_calcs.o calcs/dcn_calc_math.o calcs/dcn_calc_auto.o endif diff --git a/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calcs.c b/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calcs.c index f1235bf9a596..74962791302f 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calcs.c +++ b/drivers/gpu/drm/amd/display/dc/dml/calcs/dcn_calcs.c @@ -748,7 +748,7 @@ static unsigned int get_highest_allowed_voltage_level(bool is_vmin_only_asic) bool dcn_validate_bandwidth( struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { /* * we want a breakdown of the various stages of validation, which the @@ -1119,7 +1119,7 @@ bool dcn_validate_bandwidth( BW_VAL_TRACE_END_VOLTAGE_LEVEL(); - if (v->voltage_level != number_of_states_plus_one && !fast_validate) { + if (v->voltage_level != number_of_states_plus_one && validate_mode == DC_VALIDATE_MODE_AND_PROGRAMMING) { float bw_consumed = v->total_bandwidth_consumed_gbyte_per_second; if (bw_consumed < v->fabric_and_dram_bandwidth_vmin0p65) @@ -1286,7 +1286,7 @@ bool dcn_validate_bandwidth( } } else if (v->voltage_level == number_of_states_plus_one) { BW_VAL_TRACE_SKIP(fail); - } else if (fast_validate) { + } else if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) { BW_VAL_TRACE_SKIP(fast); } diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c index e9fea9c2162e..2a2eaf6adf26 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c @@ -1315,7 +1315,7 @@ static void swizzle_to_dml_params( int dcn20_populate_dml_pipes_from_context(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate) + enum dc_validate_mode validate_mode) { int pipe_cnt, i; bool synchronized_vblank = true; @@ -1733,7 +1733,7 @@ void dcn20_calculate_wm(struct dc *dc, struct dc_state *context, int *out_pipe_cnt, int *pipe_split_from, int vlevel, - bool fast_validate) + enum dc_validate_mode validate_mode) { int pipe_cnt, i, pipe_idx; @@ -1780,10 +1780,10 @@ void dcn20_calculate_wm(struct dc *dc, struct dc_state *context, if (pipe_cnt != pipe_idx) { if (dc->res_pool->funcs->populate_dml_pipes) pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, - context, pipes, fast_validate); + context, pipes, validate_mode); else pipe_cnt = dcn20_populate_dml_pipes_from_context(dc, - context, pipes, fast_validate); + context, pipes, validate_mode); } *out_pipe_cnt = pipe_cnt; @@ -2027,7 +2027,7 @@ void dcn20_patch_bounding_box(struct dc *dc, struct _vcs_dpi_soc_bounding_box_st } static bool dcn20_validate_bandwidth_internal(struct dc *dc, struct dc_state *context, - bool fast_validate, display_e2e_pipe_params_st *pipes) + enum dc_validate_mode validate_mode, display_e2e_pipe_params_st *pipes) { bool out = false; @@ -2040,7 +2040,7 @@ static bool dcn20_validate_bandwidth_internal(struct dc *dc, struct dc_state *co BW_VAL_TRACE_COUNT(); - out = dcn20_fast_validate_bw(dc, context, pipes, &pipe_cnt, pipe_split_from, &vlevel, fast_validate); + out = dcn20_fast_validate_bw(dc, context, pipes, &pipe_cnt, pipe_split_from, &vlevel, validate_mode); if (pipe_cnt == 0) goto validate_out; @@ -2050,12 +2050,12 @@ static bool dcn20_validate_bandwidth_internal(struct dc *dc, struct dc_state *co BW_VAL_TRACE_END_VOLTAGE_LEVEL(); - if (fast_validate) { + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) { BW_VAL_TRACE_SKIP(fast); goto validate_out; } - dcn20_calculate_wm(dc, context, pipes, &pipe_cnt, pipe_split_from, vlevel, fast_validate); + dcn20_calculate_wm(dc, context, pipes, &pipe_cnt, pipe_split_from, vlevel, validate_mode); dcn20_calculate_dlg_params(dc, context, pipes, pipe_cnt, vlevel); BW_VAL_TRACE_END_WATERMARKS(); @@ -2077,7 +2077,7 @@ validate_out: } bool dcn20_validate_bandwidth_fp(struct dc *dc, struct dc_state *context, - bool fast_validate, display_e2e_pipe_params_st *pipes) + enum dc_validate_mode validate_mode, display_e2e_pipe_params_st *pipes) { bool voltage_supported = false; bool full_pstate_supported = false; @@ -2095,12 +2095,11 @@ bool dcn20_validate_bandwidth_fp(struct dc *dc, struct dc_state *context, /*Unsafe due to current pipe merge and split logic*/ ASSERT(context != dc->current_state); - if (fast_validate) { - return dcn20_validate_bandwidth_internal(dc, context, true, pipes); - } + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) + return dcn20_validate_bandwidth_internal(dc, context, validate_mode, pipes); // Best case, we support full UCLK switch latency - voltage_supported = dcn20_validate_bandwidth_internal(dc, context, false, pipes); + voltage_supported = dcn20_validate_bandwidth_internal(dc, context, DC_VALIDATE_MODE_AND_PROGRAMMING, pipes); full_pstate_supported = context->bw_ctx.bw.dcn.clk.p_state_change_support; if (context->bw_ctx.dml.soc.dummy_pstate_latency_us == 0 || @@ -2113,7 +2112,7 @@ bool dcn20_validate_bandwidth_fp(struct dc *dc, struct dc_state *context, context->bw_ctx.dml.soc.dram_clock_change_latency_us = context->bw_ctx.dml.soc.dummy_pstate_latency_us; memset(pipes, 0, dc->res_pool->pipe_count * sizeof(display_e2e_pipe_params_st)); - voltage_supported = dcn20_validate_bandwidth_internal(dc, context, false, pipes); + voltage_supported = dcn20_validate_bandwidth_internal(dc, context, DC_VALIDATE_MODE_AND_PROGRAMMING, pipes); dummy_pstate_supported = context->bw_ctx.bw.dcn.clk.p_state_change_support; if (voltage_supported && (dummy_pstate_supported || !(context->stream_count))) { @@ -2156,14 +2155,14 @@ void dcn20_fpu_adjust_dppclk(struct vba_vars_st *v, int dcn21_populate_dml_pipes_from_context(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate) + enum dc_validate_mode validate_mode) { uint32_t pipe_cnt; int i; dc_assert_fp_enabled(); - pipe_cnt = dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); + pipe_cnt = dcn20_populate_dml_pipes_from_context(dc, context, pipes, validate_mode); for (i = 0; i < pipe_cnt; i++) { @@ -2239,7 +2238,7 @@ static void dcn21_calculate_wm(struct dc *dc, struct dc_state *context, int *out_pipe_cnt, int *pipe_split_from, int vlevel_req, - bool fast_validate) + enum dc_validate_mode validate_mode) { int pipe_cnt, i, pipe_idx; int vlevel, vlevel_max; @@ -2281,10 +2280,10 @@ static void dcn21_calculate_wm(struct dc *dc, struct dc_state *context, if (pipe_cnt != pipe_idx) { if (dc->res_pool->funcs->populate_dml_pipes) pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, - context, pipes, fast_validate); + context, pipes, validate_mode); else pipe_cnt = dcn21_populate_dml_pipes_from_context(dc, - context, pipes, fast_validate); + context, pipes, validate_mode); } *out_pipe_cnt = pipe_cnt; @@ -2319,7 +2318,7 @@ static void dcn21_calculate_wm(struct dc *dc, struct dc_state *context, } bool dcn21_validate_bandwidth_fp(struct dc *dc, struct dc_state *context, - bool fast_validate, display_e2e_pipe_params_st *pipes) + enum dc_validate_mode validate_mode, display_e2e_pipe_params_st *pipes) { bool out = false; @@ -2337,7 +2336,7 @@ bool dcn21_validate_bandwidth_fp(struct dc *dc, struct dc_state *context, /*Unsafe due to current pipe merge and split logic*/ ASSERT(context != dc->current_state); - out = dcn21_fast_validate_bw(dc, context, pipes, &pipe_cnt, pipe_split_from, &vlevel, fast_validate); + out = dcn21_fast_validate_bw(dc, context, pipes, &pipe_cnt, pipe_split_from, &vlevel, validate_mode); if (pipe_cnt == 0) goto validate_out; @@ -2347,12 +2346,12 @@ bool dcn21_validate_bandwidth_fp(struct dc *dc, struct dc_state *context, BW_VAL_TRACE_END_VOLTAGE_LEVEL(); - if (fast_validate) { + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) { BW_VAL_TRACE_SKIP(fast); goto validate_out; } - dcn21_calculate_wm(dc, context, pipes, &pipe_cnt, pipe_split_from, vlevel, fast_validate); + dcn21_calculate_wm(dc, context, pipes, &pipe_cnt, pipe_split_from, vlevel, validate_mode); dcn20_calculate_dlg_params(dc, context, pipes, pipe_cnt, vlevel); BW_VAL_TRACE_END_WATERMARKS(); diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.h index b6c34198ddc8..aed00039ca62 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.h @@ -44,14 +44,14 @@ void dcn20_calculate_dlg_params(struct dc *dc, int dcn20_populate_dml_pipes_from_context(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate); + enum dc_validate_mode validate_mode); void dcn20_calculate_wm(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, int *out_pipe_cnt, int *pipe_split_from, int vlevel, - bool fast_validate); + enum dc_validate_mode validate_mode); void dcn20_cap_soc_clocks(struct _vcs_dpi_soc_bounding_box_st *bb, struct pp_smu_nv_clock_table max_clocks); void dcn20_update_bounding_box(struct dc *dc, @@ -62,7 +62,7 @@ void dcn20_update_bounding_box(struct dc *dc, void dcn20_patch_bounding_box(struct dc *dc, struct _vcs_dpi_soc_bounding_box_st *bb); bool dcn20_validate_bandwidth_fp(struct dc *dc, struct dc_state *context, - bool fast_validate, display_e2e_pipe_params_st *pipes); + enum dc_validate_mode validate_mode, display_e2e_pipe_params_st *pipes); void dcn20_fpu_set_wm_ranges(int i, struct pp_smu_wm_range_sets *ranges, struct _vcs_dpi_soc_bounding_box_st *loaded_bb); @@ -75,9 +75,9 @@ void dcn20_fpu_adjust_dppclk(struct vba_vars_st *v, int dcn21_populate_dml_pipes_from_context(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate); -bool dcn21_validate_bandwidth_fp(struct dc *dc, struct dc_state *context, bool - fast_validate, display_e2e_pipe_params_st *pipes); + enum dc_validate_mode validate_mode); +bool dcn21_validate_bandwidth_fp(struct dc *dc, struct dc_state *context, enum + dc_validate_mode, display_e2e_pipe_params_st *pipes); void dcn21_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params); void dcn21_clk_mgr_set_bw_params_wm_table(struct clk_bw_params *bw_params); diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn30/dcn30_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn30/dcn30_fpu.c index 88789987bdbc..e5f5c0663750 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn30/dcn30_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn30/dcn30_fpu.c @@ -339,7 +339,8 @@ void dcn30_fpu_calculate_wm_and_dlg( * newly found dummy_latency_index */ context->bw_ctx.dml.soc.dram_clock_change_latency_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.pstate_latency_us; - dcn30_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, false, true); + dcn30_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, + DC_VALIDATE_MODE_AND_PROGRAMMING, true); maxMpcComb = context->bw_ctx.dml.vba.maxMpcComb; dcfclk = context->bw_ctx.dml.vba.DCFCLKState[vlevel][context->bw_ctx.dml.vba.maxMpcComb]; pstate_en = context->bw_ctx.dml.vba.DRAMClockChangeSupport[vlevel][maxMpcComb] != dm_dram_clock_change_unsupported; @@ -630,7 +631,8 @@ int dcn30_find_dummy_latency_index_for_fw_based_mclk_switch(struct dc *dc, while (dummy_latency_index < max_latency_table_entries) { context->bw_ctx.dml.soc.dram_clock_change_latency_us = dc->clk_mgr->bw_params->dummy_pstate_table[dummy_latency_index].dummy_pstate_latency_us; - dcn30_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, false, true); + dcn30_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, + DC_VALIDATE_MODE_AND_PROGRAMMING, true); if (context->bw_ctx.dml.soc.allow_dram_self_refresh_or_dram_clock_change_in_vblank == dm_allow_self_refresh_and_mclk_switch) diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.h index d2ae43a82ba5..dfcc5d50071e 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.h @@ -55,5 +55,5 @@ int dcn_get_approx_det_segs_required_for_pstate( int dcn31x_populate_dml_pipes_from_context(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate); + enum dc_validate_mode validate_mode); #endif /* __DCN31_FPU_H__*/ diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c index 5ed117e11aa2..df9d50b9b57c 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c @@ -306,7 +306,7 @@ static unsigned int get_vertical_back_porch(struct dc_crtc_timing *timing) int dcn314_populate_dml_pipes_from_context_fpu(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate) + enum dc_validate_mode validate_mode) { int i, pipe_cnt; struct resource_context *res_ctx = &context->res_ctx; @@ -316,7 +316,7 @@ int dcn314_populate_dml_pipes_from_context_fpu(struct dc *dc, struct dc_state *c dc_assert_fp_enabled(); - dcn31x_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); + dcn31x_populate_dml_pipes_from_context(dc, context, pipes, validate_mode); for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) { struct dc_crtc_timing *timing; diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.h index d32c5bb99f4c..362ac79184ea 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.h @@ -35,6 +35,6 @@ void dcn314_update_bw_bounding_box_fpu(struct dc *dc, struct clk_bw_params *bw_params); int dcn314_populate_dml_pipes_from_context_fpu(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate); + enum dc_validate_mode validate_mode); #endif diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c index b0fc1fd20208..6160952245b4 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c @@ -290,7 +290,7 @@ int dcn32_find_dummy_latency_index_for_fw_based_mclk_switch(struct dc *dc, vba->DRAMClockChangeSupport[vlevel][context->bw_ctx.dml.vba.maxMpcComb] = temp_clock_change_support; context->bw_ctx.dml.soc.dram_clock_change_latency_us = dc->clk_mgr->bw_params->dummy_pstate_table[dummy_latency_index].dummy_pstate_latency_us; - dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, false); + dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, DC_VALIDATE_MODE_AND_PROGRAMMING); /* for subvp + DRR case, if subvp pipes are still present we support pstate */ if (vba->DRAMClockChangeSupport[vlevel][vba->maxMpcComb] == dm_dram_clock_change_unsupported && @@ -1479,7 +1479,7 @@ static bool dcn32_full_validate_bw_helper(struct dc *dc, /* Conditions for setting up phantom pipes for SubVP: * 1. Not force disable SubVP - * 2. Full update (i.e. !fast_validate) + * 2. Full update (i.e. DC_VALIDATE_MODE_AND_PROGRAMMING) * 3. Enough pipes are available to support SubVP (TODO: Which pipes will use VACTIVE / VBLANK / SUBVP?) * 4. Display configuration passes validation * 5. (Config doesn't support MCLK in VACTIVE/VBLANK || dc->debug.force_subvp_mclk_switch) @@ -1517,7 +1517,8 @@ static bool dcn32_full_validate_bw_helper(struct dc *dc, dc->res_pool->funcs->add_phantom_pipes(dc, context, pipes, *pipe_cnt, dc_pipe_idx); - *pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, false); + *pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, + DC_VALIDATE_MODE_AND_PROGRAMMING); // Populate dppclk to trigger a recalculate in dml_get_voltage_level // so the phantom pipe DLG params can be assigned correctly. pipes[0].clks_cfg.dppclk_mhz = get_dppclk_calculated(&context->bw_ctx.dml, pipes, *pipe_cnt, 0); @@ -1560,7 +1561,8 @@ static bool dcn32_full_validate_bw_helper(struct dc *dc, dc_state_remove_phantom_streams_and_planes(dc, context); dc_state_release_phantom_streams_and_planes(dc, context); vba->DRAMClockChangeSupport[*vlevel][vba->maxMpcComb] = dm_dram_clock_change_unsupported; - *pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, false); + *pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, + DC_VALIDATE_MODE_AND_PROGRAMMING); *vlevel = dml_get_voltage_level(&context->bw_ctx.dml, pipes, *pipe_cnt); /* This may adjust vlevel and maxMpcComb */ @@ -2138,7 +2140,7 @@ bool dcn32_internal_validate_bw(struct dc *dc, display_e2e_pipe_params_st *pipes, int *pipe_cnt_out, int *vlevel_out, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool out = false; bool repopulate_pipes = false; @@ -2162,7 +2164,7 @@ bool dcn32_internal_validate_bw(struct dc *dc, for (i = 0; i < context->stream_count; i++) resource_update_pipes_for_stream_with_slice_count(context, dc->current_state, dc->res_pool, context->streams[i], 1); - pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, fast_validate); + pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, validate_mode); if (!pipe_cnt) { out = true; @@ -2172,13 +2174,13 @@ bool dcn32_internal_validate_bw(struct dc *dc, dml_log_pipe_params(&context->bw_ctx.dml, pipes, pipe_cnt); context->bw_ctx.dml.soc.max_vratio_pre = dcn32_determine_max_vratio_prefetch(dc, context); - if (!fast_validate) { + if (validate_mode == DC_VALIDATE_MODE_AND_PROGRAMMING) { if (!dcn32_full_validate_bw_helper(dc, context, pipes, &vlevel, split, merge, &pipe_cnt, &repopulate_pipes)) goto validate_fail; } - if (fast_validate || + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING || (dc->debug.dml_disallow_alternate_prefetch_modes && (vlevel == context->bw_ctx.dml.soc.num_states || vba->DRAMClockChangeSupport[vlevel][vba->maxMpcComb] == dm_dram_clock_change_unsupported))) { @@ -2195,7 +2197,7 @@ bool dcn32_internal_validate_bw(struct dc *dc, context->bw_ctx.dml.soc.allow_for_pstate_or_stutter_in_vblank_final = dm_prefetch_support_none; - context->bw_ctx.dml.validate_max_state = fast_validate; + context->bw_ctx.dml.validate_max_state = (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING); vlevel = dml_get_voltage_level(&context->bw_ctx.dml, pipes, pipe_cnt); context->bw_ctx.dml.validate_max_state = false; @@ -2247,7 +2249,7 @@ bool dcn32_internal_validate_bw(struct dc *dc, int flag_vlevel = vlevel; int i; - pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, fast_validate); + pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, validate_mode); if (!dc->config.enable_windowed_mpo_odm) dcn32_update_dml_pipes_odm_policy_based_on_context(dc, context, pipes); @@ -2343,7 +2345,7 @@ void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context, } context->bw_ctx.dml.soc.dram_clock_change_latency_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.pstate_latency_us; - dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, false); + dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, DC_VALIDATE_MODE_AND_PROGRAMMING); maxMpcComb = context->bw_ctx.dml.vba.maxMpcComb; if (is_subvp_p_drr) { context->bw_ctx.dml.vba.DRAMClockChangeSupport[vlevel][maxMpcComb] = dm_dram_clock_change_vblank_w_mall_sub_vp; @@ -2389,7 +2391,8 @@ void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context, context->bw_ctx.dml.soc.fclk_change_latency_us = dc->clk_mgr->bw_params->dummy_pstate_table[dummy_latency_index].dummy_pstate_latency_us; } - dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel_temp, false); + dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel_temp, + DC_VALIDATE_MODE_AND_PROGRAMMING); if (vlevel_temp < vlevel) { vlevel = vlevel_temp; maxMpcComb = context->bw_ctx.dml.vba.maxMpcComb; @@ -2410,7 +2413,8 @@ void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context, stream_status->fpo_in_use = false; } context->bw_ctx.dml.soc.fclk_change_latency_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.fclk_change_latency_us; - dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, false); + dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, + DC_VALIDATE_MODE_AND_PROGRAMMING); } } } diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.h index 276e90e4e0ce..273d2bd79d85 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.h @@ -49,7 +49,7 @@ bool dcn32_internal_validate_bw(struct dc *dc, display_e2e_pipe_params_st *pipes, int *pipe_cnt_out, int *vlevel_out, - bool fast_validate); + enum dc_validate_mode validate_mode); void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c index 92f0a099d089..5d73efa2f0c9 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c @@ -437,7 +437,7 @@ static unsigned int get_vertical_back_porch(struct dc_crtc_timing *timing) int dcn35_populate_dml_pipes_from_context_fpu(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate) + enum dc_validate_mode validate_mode) { int i, pipe_cnt; struct resource_context *res_ctx = &context->res_ctx; @@ -446,7 +446,7 @@ int dcn35_populate_dml_pipes_from_context_fpu(struct dc *dc, const unsigned int max_allowed_vblank_nom = 1023; dcn31_populate_dml_pipes_from_context(dc, context, pipes, - fast_validate); + validate_mode); for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) { struct dc_crtc_timing *timing; diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.h index 067480fc3691..d121c5afce71 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.h @@ -37,7 +37,7 @@ void dcn35_update_bw_bounding_box_fpu(struct dc *dc, int dcn35_populate_dml_pipes_from_context_fpu(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate); + enum dc_validate_mode validate_mode); void dcn35_decide_zstate_support(struct dc *dc, struct dc_state *context); diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn351/dcn351_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn351/dcn351_fpu.c index 17d0b4923b0c..6f516af82956 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn351/dcn351_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn351/dcn351_fpu.c @@ -470,7 +470,7 @@ static unsigned int get_vertical_back_porch(struct dc_crtc_timing *timing) int dcn351_populate_dml_pipes_from_context_fpu(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate) + enum dc_validate_mode validate_mode) { int i, pipe_cnt; struct resource_context *res_ctx = &context->res_ctx; @@ -479,7 +479,7 @@ int dcn351_populate_dml_pipes_from_context_fpu(struct dc *dc, const unsigned int max_allowed_vblank_nom = 1023; dcn31_populate_dml_pipes_from_context(dc, context, pipes, - fast_validate); + validate_mode); for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) { struct dc_crtc_timing *timing; diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn351/dcn351_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn351/dcn351_fpu.h index f93efab9a668..f71d9d8d0759 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn351/dcn351_fpu.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn351/dcn351_fpu.h @@ -12,7 +12,7 @@ void dcn351_update_bw_bounding_box_fpu(struct dc *dc, int dcn351_populate_dml_pipes_from_context_fpu(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate); + enum dc_validate_mode validate_mode); void dcn351_decide_zstate_support(struct dc *dc, struct dc_state *context); diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn401/dcn401_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn401/dcn401_fpu.c deleted file mode 100644 index 4fbecb5ff349..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn401/dcn401_fpu.c +++ /dev/null @@ -1,239 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dcn401_fpu.h" -#include "dcn401/dcn401_resource.h" -// We need this includes for WATERMARKS_* defines -#include "clk_mgr/dcn401/dcn401_smu14_driver_if.h" -#include "link.h" - -#define DC_LOGGER_INIT(logger) - -void dcn401_build_wm_range_table_fpu(struct clk_mgr *clk_mgr) -{ - /* defaults */ - double pstate_latency_us = clk_mgr->ctx->dc->dml.soc.dram_clock_change_latency_us; - double fclk_change_latency_us = clk_mgr->ctx->dc->dml.soc.fclk_change_latency_us; - double sr_exit_time_us = clk_mgr->ctx->dc->dml.soc.sr_exit_time_us; - double sr_enter_plus_exit_time_us = clk_mgr->ctx->dc->dml.soc.sr_enter_plus_exit_time_us; - /* For min clocks use as reported by PM FW and report those as min */ - uint16_t min_uclk_mhz = clk_mgr->bw_params->clk_table.entries[0].memclk_mhz; - uint16_t min_dcfclk_mhz = clk_mgr->bw_params->clk_table.entries[0].dcfclk_mhz; - uint16_t setb_min_uclk_mhz = min_uclk_mhz; - uint16_t dcfclk_mhz_for_the_second_state = clk_mgr->ctx->dc->dml.soc.clock_limits[2].dcfclk_mhz; - - dc_assert_fp_enabled(); - - /* For Set B ranges use min clocks state 2 when available, and report those to PM FW */ - if (dcfclk_mhz_for_the_second_state) - clk_mgr->bw_params->wm_table.nv_entries[WM_B].pmfw_breakdown.min_dcfclk = dcfclk_mhz_for_the_second_state; - else - clk_mgr->bw_params->wm_table.nv_entries[WM_B].pmfw_breakdown.min_dcfclk = clk_mgr->bw_params->clk_table.entries[0].dcfclk_mhz; - - if (clk_mgr->bw_params->clk_table.entries[2].memclk_mhz) - setb_min_uclk_mhz = clk_mgr->bw_params->clk_table.entries[2].memclk_mhz; - - /* Set A - Normal - default values */ - clk_mgr->bw_params->wm_table.nv_entries[WM_A].valid = true; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.pstate_latency_us = pstate_latency_us; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.fclk_change_latency_us = fclk_change_latency_us; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.sr_exit_time_us = sr_exit_time_us; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.sr_enter_plus_exit_time_us = sr_enter_plus_exit_time_us; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.wm_type = WATERMARKS_CLOCK_RANGE; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.min_dcfclk = min_dcfclk_mhz; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.max_dcfclk = 0xFFFF; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.min_uclk = min_uclk_mhz; - clk_mgr->bw_params->wm_table.nv_entries[WM_A].pmfw_breakdown.max_uclk = 0xFFFF; - - /* Set B - Performance - higher clocks, using DPM[2] DCFCLK and UCLK */ - clk_mgr->bw_params->wm_table.nv_entries[WM_B].valid = true; - clk_mgr->bw_params->wm_table.nv_entries[WM_B].dml_input.pstate_latency_us = pstate_latency_us; - clk_mgr->bw_params->wm_table.nv_entries[WM_B].dml_input.fclk_change_latency_us = fclk_change_latency_us; - clk_mgr->bw_params->wm_table.nv_entries[WM_B].dml_input.sr_exit_time_us = sr_exit_time_us; - clk_mgr->bw_params->wm_table.nv_entries[WM_B].dml_input.sr_enter_plus_exit_time_us = sr_enter_plus_exit_time_us; - clk_mgr->bw_params->wm_table.nv_entries[WM_B].pmfw_breakdown.wm_type = WATERMARKS_CLOCK_RANGE; - clk_mgr->bw_params->wm_table.nv_entries[WM_B].pmfw_breakdown.max_dcfclk = 0xFFFF; - clk_mgr->bw_params->wm_table.nv_entries[WM_B].pmfw_breakdown.min_uclk = setb_min_uclk_mhz; - clk_mgr->bw_params->wm_table.nv_entries[WM_B].pmfw_breakdown.max_uclk = 0xFFFF; - - /* Set C - Dummy P-State - P-State latency set to "dummy p-state" value */ - /* 'DalDummyClockChangeLatencyNs' registry key option set to 0x7FFFFFFF can be used to disable Set C for dummy p-state */ - if (clk_mgr->ctx->dc->bb_overrides.dummy_clock_change_latency_ns != 0x7FFFFFFF) { - clk_mgr->bw_params->wm_table.nv_entries[WM_C].valid = true; - clk_mgr->bw_params->wm_table.nv_entries[WM_C].dml_input.pstate_latency_us = 50; - clk_mgr->bw_params->wm_table.nv_entries[WM_C].dml_input.fclk_change_latency_us = fclk_change_latency_us; - clk_mgr->bw_params->wm_table.nv_entries[WM_C].dml_input.sr_exit_time_us = sr_exit_time_us; - clk_mgr->bw_params->wm_table.nv_entries[WM_C].dml_input.sr_enter_plus_exit_time_us = sr_enter_plus_exit_time_us; - clk_mgr->bw_params->wm_table.nv_entries[WM_C].pmfw_breakdown.wm_type = WATERMARKS_DUMMY_PSTATE; - clk_mgr->bw_params->wm_table.nv_entries[WM_C].pmfw_breakdown.min_dcfclk = min_dcfclk_mhz; - clk_mgr->bw_params->wm_table.nv_entries[WM_C].pmfw_breakdown.max_dcfclk = 0xFFFF; - clk_mgr->bw_params->wm_table.nv_entries[WM_C].pmfw_breakdown.min_uclk = min_uclk_mhz; - clk_mgr->bw_params->wm_table.nv_entries[WM_C].pmfw_breakdown.max_uclk = 0xFFFF; - clk_mgr->bw_params->dummy_pstate_table[0].dram_speed_mts = clk_mgr->bw_params->clk_table.entries[0].memclk_mhz * 16; - clk_mgr->bw_params->dummy_pstate_table[0].dummy_pstate_latency_us = 50; - clk_mgr->bw_params->dummy_pstate_table[1].dram_speed_mts = clk_mgr->bw_params->clk_table.entries[1].memclk_mhz * 16; - clk_mgr->bw_params->dummy_pstate_table[1].dummy_pstate_latency_us = 9; - clk_mgr->bw_params->dummy_pstate_table[2].dram_speed_mts = clk_mgr->bw_params->clk_table.entries[2].memclk_mhz * 16; - clk_mgr->bw_params->dummy_pstate_table[2].dummy_pstate_latency_us = 8; - clk_mgr->bw_params->dummy_pstate_table[3].dram_speed_mts = clk_mgr->bw_params->clk_table.entries[3].memclk_mhz * 16; - clk_mgr->bw_params->dummy_pstate_table[3].dummy_pstate_latency_us = 5; - } - /* Set D - MALL - SR enter and exit time specific to MALL, TBD after bringup or later phase for now use DRAM values / 2 */ - /* For MALL DRAM clock change latency is N/A, for watermak calculations use lowest value dummy P state latency */ - clk_mgr->bw_params->wm_table.nv_entries[WM_D].valid = true; - clk_mgr->bw_params->wm_table.nv_entries[WM_D].dml_input.pstate_latency_us = clk_mgr->bw_params->dummy_pstate_table[3].dummy_pstate_latency_us; - clk_mgr->bw_params->wm_table.nv_entries[WM_D].dml_input.fclk_change_latency_us = fclk_change_latency_us; - clk_mgr->bw_params->wm_table.nv_entries[WM_D].dml_input.sr_exit_time_us = sr_exit_time_us / 2; // TBD - clk_mgr->bw_params->wm_table.nv_entries[WM_D].dml_input.sr_enter_plus_exit_time_us = sr_enter_plus_exit_time_us / 2; // TBD - clk_mgr->bw_params->wm_table.nv_entries[WM_D].pmfw_breakdown.wm_type = WATERMARKS_MALL; - clk_mgr->bw_params->wm_table.nv_entries[WM_D].pmfw_breakdown.min_dcfclk = min_dcfclk_mhz; - clk_mgr->bw_params->wm_table.nv_entries[WM_D].pmfw_breakdown.max_dcfclk = 0xFFFF; - clk_mgr->bw_params->wm_table.nv_entries[WM_D].pmfw_breakdown.min_uclk = min_uclk_mhz; - clk_mgr->bw_params->wm_table.nv_entries[WM_D].pmfw_breakdown.max_uclk = 0xFFFF; -} - -/* - * dcn401_update_bw_bounding_box - * - * This would override some dcn4_01 ip_or_soc initial parameters hardcoded from - * spreadsheet with actual values as per dGPU SKU: - * - with passed few options from dc->config - * - with dentist_vco_frequency from Clk Mgr (currently hardcoded, but might - * need to get it from PM FW) - * - with passed latency values (passed in ns units) in dc-> bb override for - * debugging purposes - * - with passed latencies from VBIOS (in 100_ns units) if available for - * certain dGPU SKU - * - with number of DRAM channels from VBIOS (which differ for certain dGPU SKU - * of the same ASIC) - * - clocks levels with passed clk_table entries from Clk Mgr as reported by PM - * FW for different clocks (which might differ for certain dGPU SKU of the - * same ASIC) - */ -void dcn401_update_bw_bounding_box_fpu(struct dc *dc, struct clk_bw_params *bw_params) -{ - dc_assert_fp_enabled(); - - /* Override from passed dc->bb_overrides if available*/ - if (dc->bb_overrides.sr_exit_time_ns) - dc->dml2_options.bbox_overrides.sr_exit_latency_us = - dc->bb_overrides.sr_exit_time_ns / 1000.0; - - if (dc->bb_overrides.sr_enter_plus_exit_time_ns) - dc->dml2_options.bbox_overrides.sr_enter_plus_exit_latency_us = - dc->bb_overrides.sr_enter_plus_exit_time_ns / 1000.0; - - if (dc->bb_overrides.urgent_latency_ns) - dc->dml2_options.bbox_overrides.urgent_latency_us = - dc->bb_overrides.urgent_latency_ns / 1000.0; - - if (dc->bb_overrides.dram_clock_change_latency_ns) - dc->dml2_options.bbox_overrides.dram_clock_change_latency_us = - dc->bb_overrides.dram_clock_change_latency_ns / 1000.0; - - if (dc->bb_overrides.fclk_clock_change_latency_ns) - dc->dml2_options.bbox_overrides.fclk_change_latency_us = - dc->bb_overrides.fclk_clock_change_latency_ns / 1000; - - /* Override from VBIOS if VBIOS bb_info available */ - if (dc->ctx->dc_bios->funcs->get_soc_bb_info) { - struct bp_soc_bb_info bb_info = {0}; - if (dc->ctx->dc_bios->funcs->get_soc_bb_info(dc->ctx->dc_bios, &bb_info) == BP_RESULT_OK) { - if (bb_info.dram_clock_change_latency_100ns > 0) - dc->dml2_options.bbox_overrides.dram_clock_change_latency_us = - bb_info.dram_clock_change_latency_100ns * 10; - - if (bb_info.dram_sr_enter_exit_latency_100ns > 0) - dc->dml2_options.bbox_overrides.sr_enter_plus_exit_latency_us = - bb_info.dram_sr_enter_exit_latency_100ns * 10; - - if (bb_info.dram_sr_exit_latency_100ns > 0) - dc->dml2_options.bbox_overrides.sr_exit_latency_us = - bb_info.dram_sr_exit_latency_100ns * 10; - } - } - - /* Override from VBIOS for num_chan */ - if (dc->ctx->dc_bios->vram_info.num_chans) { - dc->dml2_options.bbox_overrides.dram_num_chan = - dc->ctx->dc_bios->vram_info.num_chans; - - } - - if (dc->ctx->dc_bios->vram_info.dram_channel_width_bytes) - dc->dml2_options.bbox_overrides.dram_chanel_width_bytes = - dc->ctx->dc_bios->vram_info.dram_channel_width_bytes; - - dc->dml2_options.bbox_overrides.disp_pll_vco_speed_mhz = dc->clk_mgr->dentist_vco_freq_khz / 1000.0; - dc->dml2_options.bbox_overrides.xtalclk_mhz = dc->ctx->dc_bios->fw_info.pll_info.crystal_frequency / 1000.0; - dc->dml2_options.bbox_overrides.dchub_refclk_mhz = dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000.0; - dc->dml2_options.bbox_overrides.dprefclk_mhz = dc->clk_mgr->dprefclk_khz / 1000.0; - - if (dc->clk_mgr->bw_params->clk_table.num_entries > 1) { - unsigned int i = 0; - - dc->dml2_options.bbox_overrides.clks_table.num_states = dc->clk_mgr->bw_params->clk_table.num_entries; - - dc->dml2_options.bbox_overrides.clks_table.num_entries_per_clk.num_dcfclk_levels = - dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_dcfclk_levels; - - dc->dml2_options.bbox_overrides.clks_table.num_entries_per_clk.num_fclk_levels = - dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_fclk_levels; - - dc->dml2_options.bbox_overrides.clks_table.num_entries_per_clk.num_memclk_levels = - dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_memclk_levels; - - dc->dml2_options.bbox_overrides.clks_table.num_entries_per_clk.num_socclk_levels = - dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_socclk_levels; - - dc->dml2_options.bbox_overrides.clks_table.num_entries_per_clk.num_dtbclk_levels = - dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_dtbclk_levels; - - dc->dml2_options.bbox_overrides.clks_table.num_entries_per_clk.num_dispclk_levels = - dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_dispclk_levels; - - dc->dml2_options.bbox_overrides.clks_table.num_entries_per_clk.num_dppclk_levels = - dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_dppclk_levels; - - for (i = 0; i < dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_dcfclk_levels; i++) { - if (dc->clk_mgr->bw_params->clk_table.entries[i].dcfclk_mhz) - dc->dml2_options.bbox_overrides.clks_table.clk_entries[i].dcfclk_mhz = - dc->clk_mgr->bw_params->clk_table.entries[i].dcfclk_mhz; - } - - for (i = 0; i < dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_fclk_levels; i++) { - if (dc->clk_mgr->bw_params->clk_table.entries[i].fclk_mhz) - dc->dml2_options.bbox_overrides.clks_table.clk_entries[i].fclk_mhz = - dc->clk_mgr->bw_params->clk_table.entries[i].fclk_mhz; - } - - for (i = 0; i < dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_memclk_levels; i++) { - if (dc->clk_mgr->bw_params->clk_table.entries[i].memclk_mhz) - dc->dml2_options.bbox_overrides.clks_table.clk_entries[i].memclk_mhz = - dc->clk_mgr->bw_params->clk_table.entries[i].memclk_mhz; - } - - for (i = 0; i < dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_socclk_levels; i++) { - if (dc->clk_mgr->bw_params->clk_table.entries[i].socclk_mhz) - dc->dml2_options.bbox_overrides.clks_table.clk_entries[i].socclk_mhz = - dc->clk_mgr->bw_params->clk_table.entries[i].socclk_mhz; - } - - for (i = 0; i < dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_dtbclk_levels; i++) { - if (dc->clk_mgr->bw_params->clk_table.entries[i].dtbclk_mhz) - dc->dml2_options.bbox_overrides.clks_table.clk_entries[i].dtbclk_mhz = - dc->clk_mgr->bw_params->clk_table.entries[i].dtbclk_mhz; - } - - for (i = 0; i < dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_dispclk_levels; i++) { - if (dc->clk_mgr->bw_params->clk_table.entries[i].dispclk_mhz) { - dc->dml2_options.bbox_overrides.clks_table.clk_entries[i].dispclk_mhz = - dc->clk_mgr->bw_params->clk_table.entries[i].dispclk_mhz; - dc->dml2_options.bbox_overrides.clks_table.clk_entries[i].dppclk_mhz = - dc->clk_mgr->bw_params->clk_table.entries[i].dispclk_mhz; - } - } - } -} - diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn401/dcn401_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn401/dcn401_fpu.h deleted file mode 100644 index 329f1788843c..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn401/dcn401_fpu.h +++ /dev/null @@ -1,14 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DCN401_FPU_H__ -#define __DCN401_FPU_H__ - -#include "clk_mgr.h" - -void dcn401_build_wm_range_table_fpu(struct clk_mgr *clk_mgr); - -void dcn401_update_bw_bounding_box_fpu(struct dc *dc, struct clk_bw_params *bw_params); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/Makefile b/drivers/gpu/drm/amd/display/dc/dml2/Makefile index 157ecf008d6c..4c21ce42054c 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dml2/Makefile @@ -81,10 +81,11 @@ AMD_DAL_DML2 = $(addprefix $(AMDDALPATH)/dc/dml2/,$(DML2)) AMD_DISPLAY_FILES += $(AMD_DAL_DML2) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.o := $(dml2_ccflags) $(frame_warn_flag) +CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_utils.o := $(dml2_ccflags) $(frame_warn_flag) +CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_factory.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.o := $(dml2_ccflags) @@ -94,17 +95,16 @@ CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.o := $(dml2_ccflag CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.o := $(dml2_ccflags) - - CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml21_wrapper.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/dml21_translation_helper.o := $(dml2_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/dml21_utils.o := $(dml2_ccflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_factory.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_utils.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.o := $(dml2_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.o := $(dml2_rcflags) @@ -120,6 +120,7 @@ CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/dml21_utils.o := $(dml2_rcflags) DML21 := src/dml2_top/dml2_top_interfaces.o DML21 += src/dml2_top/dml2_top_soc15.o DML21 += src/dml2_core/dml2_core_dcn4.o +DML21 += src/dml2_core/dml2_core_utils.o DML21 += src/dml2_core/dml2_core_factory.o DML21 += src/dml2_core/dml2_core_dcn4_calcs.o DML21 += src/dml2_dpmm/dml2_dpmm_dcn4.o diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c index 7ae9c0ba0c9e..715f9019a33e 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c @@ -10189,7 +10189,7 @@ dml_uint_t dml_mode_support_ex(struct dml_mode_support_ex_params_st *in_out_para result = mode_support_pwr_states(&in_out_params->out_lowest_state_idx, in_out_params->mode_lib, in_out_params->in_display_cfg, - 0, + in_out_params->in_start_state_idx, in_out_params->mode_lib->states.num_states - 1); if (result) diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h index 0670e4dc4fd9..dbeb08466092 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h @@ -1917,6 +1917,7 @@ struct display_mode_lib_st { struct dml_mode_support_ex_params_st { struct display_mode_lib_st *mode_lib; const struct dml_display_cfg_st *in_display_cfg; + dml_uint_t in_start_state_idx; dml_uint_t out_lowest_state_idx; struct dml_mode_support_info_st *out_evaluation_info; }; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c index d47cacfdb695..a06217a9eef6 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c @@ -2,7 +2,6 @@ // // Copyright 2024 Advanced Micro Devices, Inc. - #include "dml21_wrapper.h" #include "dml2_core_dcn4_calcs.h" #include "dml2_internal_shared_types.h" @@ -11,277 +10,263 @@ #include "dml21_translation_helper.h" #include "bounding_boxes/dcn4_soc_bb.h" -static void dml21_init_socbb_params(struct dml2_initialize_instance_in_out *dml_init, - const struct dml2_configuration_options *config, - const struct dc *in_dc) +static void dml21_populate_pmo_options(struct dml2_pmo_options *pmo_options, + const struct dc *in_dc, + const struct dml2_configuration_options *config) { - const struct dml2_soc_bb *soc_bb; - const struct dml2_soc_qos_parameters *qos_params; - - switch (in_dc->ctx->dce_version) { - case DCN_VERSION_4_01: - default: - if (config->bb_from_dmub) - soc_bb = config->bb_from_dmub; - else - soc_bb = &dml2_socbb_dcn401; + bool disable_fams2 = !in_dc->debug.fams2_config.bits.enable; - qos_params = &dml_dcn4_variant_a_soc_qos_params; - } - - /* patch soc bb */ - memcpy(&dml_init->soc_bb, soc_bb, sizeof(struct dml2_soc_bb)); + /* ODM options */ + pmo_options->disable_dyn_odm = !config->minimize_dispclk_using_odm; + pmo_options->disable_dyn_odm_for_multi_stream = true; + pmo_options->disable_dyn_odm_for_stream_with_svp = true; - /* patch qos params */ - memcpy(&dml_init->soc_bb.qos_parameters, qos_params, sizeof(struct dml2_soc_qos_parameters)); -} + pmo_options->disable_vblank = ((in_dc->debug.dml21_disable_pstate_method_mask >> 1) & 1); -static void dml21_external_socbb_params(struct dml2_initialize_instance_in_out *dml_init, - const struct dml2_configuration_options *config) -{ - memcpy(&dml_init->soc_bb, &config->external_socbb_ip_params->soc_bb, sizeof(struct dml2_soc_bb)); -} + /* NOTE: DRR and SubVP Require FAMS2 */ + pmo_options->disable_svp = ((in_dc->debug.dml21_disable_pstate_method_mask >> 2) & 1) || + in_dc->debug.force_disable_subvp || + disable_fams2; + pmo_options->disable_drr_clamped = ((in_dc->debug.dml21_disable_pstate_method_mask >> 3) & 1) || + disable_fams2; + pmo_options->disable_drr_var = ((in_dc->debug.dml21_disable_pstate_method_mask >> 4) & 1) || + disable_fams2; + pmo_options->disable_fams2 = disable_fams2; -static void dml21_external_ip_params(struct dml2_initialize_instance_in_out *dml_init, - const struct dml2_configuration_options *config) -{ - memcpy(&dml_init->ip_caps, &config->external_socbb_ip_params->ip_params, sizeof(struct dml2_ip_capabilities)); + pmo_options->disable_drr_var_when_var_active = in_dc->debug.disable_fams_gaming == INGAME_FAMS_DISABLE || + in_dc->debug.disable_fams_gaming == INGAME_FAMS_MULTI_DISP_CLAMPED_ONLY; + pmo_options->disable_drr_clamped_when_var_active = in_dc->debug.disable_fams_gaming == INGAME_FAMS_DISABLE; } -static void dml21_init_ip_params(struct dml2_initialize_instance_in_out *dml_init, +/* + * Populate dml_init based on default static values in soc bb. The default + * values are for reference and support at least minimal operation of current + * SoC and DCN hardware. The values could be modifed by subsequent override + * functions to reflect our true hardware capability. + */ +static void populate_default_dml_init_params(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc) { - const struct dml2_ip_capabilities *ip_caps; - switch (in_dc->ctx->dce_version) { case DCN_VERSION_4_01: + dml_init->options.project_id = dml2_project_dcn4x_stage2_auto_drr_svp; + dml21_populate_pmo_options(&dml_init->options.pmo_options, in_dc, config); + dml_init->soc_bb = dml2_socbb_dcn401; + dml_init->soc_bb.qos_parameters = dml_dcn4_variant_a_soc_qos_params; + dml_init->ip_caps = dml2_dcn401_max_ip_caps; + break; default: - ip_caps = &dml2_dcn401_max_ip_caps; + memset(dml_init, 0, sizeof(*dml_init)); + DC_ERR("unsupported dcn version for DML21!"); + return; } - - memcpy(&dml_init->ip_caps, ip_caps, sizeof(struct dml2_ip_capabilities)); } -void dml21_initialize_soc_bb_params(struct dml2_initialize_instance_in_out *dml_init, +static void override_dml_init_with_values_from_hardware_default(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc) { - if (config->use_native_soc_bb_construction) - dml21_init_socbb_params(dml_init, config, in_dc); - else - dml21_external_socbb_params(dml_init, config); + dml_init->soc_bb.dchub_refclk_mhz = in_dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000; + dml_init->soc_bb.dprefclk_mhz = in_dc->clk_mgr->dprefclk_khz / 1000; + dml_init->soc_bb.dispclk_dppclk_vco_speed_mhz = in_dc->clk_mgr->dentist_vco_freq_khz / 1000.0; } -void dml21_initialize_ip_params(struct dml2_initialize_instance_in_out *dml_init, +/* + * SMU stands for System Management Unit. It is a power management processor. + * It owns the initialization of dc's clock table and programming of clock values + * based on dc's requests. + * Our clock values in base soc bb is a dummy placeholder. The real clock values + * are retrieved from SMU firmware to dc clock table at runtime. + * This function overrides our dummy placeholder values with real values in dc + * clock table. + */ +static void override_dml_init_with_values_from_smu( + struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc) { - if (config->use_native_soc_bb_construction) - dml21_init_ip_params(dml_init, config, in_dc); - else - dml21_external_ip_params(dml_init, config); -} - -void dml21_apply_soc_bb_overrides(struct dml2_initialize_instance_in_out *dml_init, - const struct dml2_configuration_options *config, const struct dc *in_dc) -{ int i; - const struct clk_bw_params *dc_bw_params = in_dc->clk_mgr->bw_params; const struct clk_limit_table *dc_clk_table = &dc_bw_params->clk_table; - struct dml2_soc_bb *dml_soc_bb = &dml_init->soc_bb; - struct dml2_soc_state_table *dml_clk_table = &dml_soc_bb->clk_table; + struct dml2_soc_state_table *dml_clk_table = &dml_init->soc_bb.clk_table; - /* override clocks if smu is present */ - if (in_dc->clk_mgr->funcs->is_smu_present && in_dc->clk_mgr->funcs->is_smu_present(in_dc->clk_mgr)) { - /* dcfclk */ - if (dc_clk_table->num_entries_per_clk.num_dcfclk_levels) { - dml_clk_table->dcfclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dcfclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->dcfclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dcfclk_mhz && - dc_clk_table->entries[i].dcfclk_mhz > dc_bw_params->dc_mode_limit.dcfclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].dcfclk_mhz < dc_bw_params->dc_mode_limit.dcfclk_mhz) { - dml_clk_table->dcfclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dcfclk_mhz * 1000; - dml_clk_table->dcfclk.num_clk_values = i + 1; - } else { - dml_clk_table->dcfclk.clk_values_khz[i] = 0; - dml_clk_table->dcfclk.num_clk_values = i; - } + if (!in_dc->clk_mgr->funcs->is_smu_present || + !in_dc->clk_mgr->funcs->is_smu_present(in_dc->clk_mgr)) + /* skip if smu is not present */ + return; + + /* dcfclk */ + if (dc_clk_table->num_entries_per_clk.num_dcfclk_levels) { + dml_clk_table->dcfclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dcfclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->dcfclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dcfclk_mhz && + dc_clk_table->entries[i].dcfclk_mhz > dc_bw_params->dc_mode_limit.dcfclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].dcfclk_mhz < dc_bw_params->dc_mode_limit.dcfclk_mhz) { + dml_clk_table->dcfclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dcfclk_mhz * 1000; + dml_clk_table->dcfclk.num_clk_values = i + 1; } else { - dml_clk_table->dcfclk.clk_values_khz[i] = dc_clk_table->entries[i].dcfclk_mhz * 1000; + dml_clk_table->dcfclk.clk_values_khz[i] = 0; + dml_clk_table->dcfclk.num_clk_values = i; } } else { - dml_clk_table->dcfclk.clk_values_khz[i] = 0; + dml_clk_table->dcfclk.clk_values_khz[i] = dc_clk_table->entries[i].dcfclk_mhz * 1000; } + } else { + dml_clk_table->dcfclk.clk_values_khz[i] = 0; } } + } - /* fclk */ - if (dc_clk_table->num_entries_per_clk.num_fclk_levels) { - dml_clk_table->fclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_fclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->fclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.fclk_mhz && - dc_clk_table->entries[i].fclk_mhz > dc_bw_params->dc_mode_limit.fclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].fclk_mhz < dc_bw_params->dc_mode_limit.fclk_mhz) { - dml_clk_table->fclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.fclk_mhz * 1000; - dml_clk_table->fclk.num_clk_values = i + 1; - } else { - dml_clk_table->fclk.clk_values_khz[i] = 0; - dml_clk_table->fclk.num_clk_values = i; - } + /* fclk */ + if (dc_clk_table->num_entries_per_clk.num_fclk_levels) { + dml_clk_table->fclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_fclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->fclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.fclk_mhz && + dc_clk_table->entries[i].fclk_mhz > dc_bw_params->dc_mode_limit.fclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].fclk_mhz < dc_bw_params->dc_mode_limit.fclk_mhz) { + dml_clk_table->fclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.fclk_mhz * 1000; + dml_clk_table->fclk.num_clk_values = i + 1; } else { - dml_clk_table->fclk.clk_values_khz[i] = dc_clk_table->entries[i].fclk_mhz * 1000; + dml_clk_table->fclk.clk_values_khz[i] = 0; + dml_clk_table->fclk.num_clk_values = i; } } else { - dml_clk_table->fclk.clk_values_khz[i] = 0; + dml_clk_table->fclk.clk_values_khz[i] = dc_clk_table->entries[i].fclk_mhz * 1000; } + } else { + dml_clk_table->fclk.clk_values_khz[i] = 0; } } + } - /* uclk */ - if (dc_clk_table->num_entries_per_clk.num_memclk_levels) { - dml_clk_table->uclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_memclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->uclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.memclk_mhz && - dc_clk_table->entries[i].memclk_mhz > dc_bw_params->dc_mode_limit.memclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].memclk_mhz < dc_bw_params->dc_mode_limit.memclk_mhz) { - dml_clk_table->uclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.memclk_mhz * 1000; - dml_clk_table->uclk.num_clk_values = i + 1; - } else { - dml_clk_table->uclk.clk_values_khz[i] = 0; - dml_clk_table->uclk.num_clk_values = i; - } + /* uclk */ + if (dc_clk_table->num_entries_per_clk.num_memclk_levels) { + dml_clk_table->uclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_memclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->uclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.memclk_mhz && + dc_clk_table->entries[i].memclk_mhz > dc_bw_params->dc_mode_limit.memclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].memclk_mhz < dc_bw_params->dc_mode_limit.memclk_mhz) { + dml_clk_table->uclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.memclk_mhz * 1000; + dml_clk_table->uclk.num_clk_values = i + 1; } else { - dml_clk_table->uclk.clk_values_khz[i] = dc_clk_table->entries[i].memclk_mhz * 1000; + dml_clk_table->uclk.clk_values_khz[i] = 0; + dml_clk_table->uclk.num_clk_values = i; } } else { - dml_clk_table->uclk.clk_values_khz[i] = 0; + dml_clk_table->uclk.clk_values_khz[i] = dc_clk_table->entries[i].memclk_mhz * 1000; } + } else { + dml_clk_table->uclk.clk_values_khz[i] = 0; } } + } - /* dispclk */ - if (dc_clk_table->num_entries_per_clk.num_dispclk_levels) { - dml_clk_table->dispclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dispclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->dispclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dispclk_mhz && - dc_clk_table->entries[i].dispclk_mhz > dc_bw_params->dc_mode_limit.dispclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].dispclk_mhz < dc_bw_params->dc_mode_limit.dispclk_mhz) { - dml_clk_table->dispclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dispclk_mhz * 1000; - dml_clk_table->dispclk.num_clk_values = i + 1; - } else { - dml_clk_table->dispclk.clk_values_khz[i] = 0; - dml_clk_table->dispclk.num_clk_values = i; - } + /* dispclk */ + if (dc_clk_table->num_entries_per_clk.num_dispclk_levels) { + dml_clk_table->dispclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dispclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->dispclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dispclk_mhz && + dc_clk_table->entries[i].dispclk_mhz > dc_bw_params->dc_mode_limit.dispclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].dispclk_mhz < dc_bw_params->dc_mode_limit.dispclk_mhz) { + dml_clk_table->dispclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dispclk_mhz * 1000; + dml_clk_table->dispclk.num_clk_values = i + 1; } else { - dml_clk_table->dispclk.clk_values_khz[i] = dc_clk_table->entries[i].dispclk_mhz * 1000; + dml_clk_table->dispclk.clk_values_khz[i] = 0; + dml_clk_table->dispclk.num_clk_values = i; } } else { - dml_clk_table->dispclk.clk_values_khz[i] = 0; + dml_clk_table->dispclk.clk_values_khz[i] = dc_clk_table->entries[i].dispclk_mhz * 1000; } + } else { + dml_clk_table->dispclk.clk_values_khz[i] = 0; } } + } - /* dppclk */ - if (dc_clk_table->num_entries_per_clk.num_dppclk_levels) { - dml_clk_table->dppclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dppclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->dppclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dppclk_mhz && - dc_clk_table->entries[i].dppclk_mhz > dc_bw_params->dc_mode_limit.dppclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].dppclk_mhz < dc_bw_params->dc_mode_limit.dppclk_mhz) { - dml_clk_table->dppclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dppclk_mhz * 1000; - dml_clk_table->dppclk.num_clk_values = i + 1; - } else { - dml_clk_table->dppclk.clk_values_khz[i] = 0; - dml_clk_table->dppclk.num_clk_values = i; - } + /* dppclk */ + if (dc_clk_table->num_entries_per_clk.num_dppclk_levels) { + dml_clk_table->dppclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dppclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->dppclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dppclk_mhz && + dc_clk_table->entries[i].dppclk_mhz > dc_bw_params->dc_mode_limit.dppclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].dppclk_mhz < dc_bw_params->dc_mode_limit.dppclk_mhz) { + dml_clk_table->dppclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dppclk_mhz * 1000; + dml_clk_table->dppclk.num_clk_values = i + 1; } else { - dml_clk_table->dppclk.clk_values_khz[i] = dc_clk_table->entries[i].dppclk_mhz * 1000; + dml_clk_table->dppclk.clk_values_khz[i] = 0; + dml_clk_table->dppclk.num_clk_values = i; } } else { - dml_clk_table->dppclk.clk_values_khz[i] = 0; + dml_clk_table->dppclk.clk_values_khz[i] = dc_clk_table->entries[i].dppclk_mhz * 1000; } + } else { + dml_clk_table->dppclk.clk_values_khz[i] = 0; } } + } - /* dtbclk */ - if (dc_clk_table->num_entries_per_clk.num_dtbclk_levels) { - dml_clk_table->dtbclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dtbclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->dtbclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dtbclk_mhz && - dc_clk_table->entries[i].dtbclk_mhz > dc_bw_params->dc_mode_limit.dtbclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].dtbclk_mhz < dc_bw_params->dc_mode_limit.dtbclk_mhz) { - dml_clk_table->dtbclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dtbclk_mhz * 1000; - dml_clk_table->dtbclk.num_clk_values = i + 1; - } else { - dml_clk_table->dtbclk.clk_values_khz[i] = 0; - dml_clk_table->dtbclk.num_clk_values = i; - } + /* dtbclk */ + if (dc_clk_table->num_entries_per_clk.num_dtbclk_levels) { + dml_clk_table->dtbclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_dtbclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->dtbclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.dtbclk_mhz && + dc_clk_table->entries[i].dtbclk_mhz > dc_bw_params->dc_mode_limit.dtbclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].dtbclk_mhz < dc_bw_params->dc_mode_limit.dtbclk_mhz) { + dml_clk_table->dtbclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.dtbclk_mhz * 1000; + dml_clk_table->dtbclk.num_clk_values = i + 1; } else { - dml_clk_table->dtbclk.clk_values_khz[i] = dc_clk_table->entries[i].dtbclk_mhz * 1000; + dml_clk_table->dtbclk.clk_values_khz[i] = 0; + dml_clk_table->dtbclk.num_clk_values = i; } } else { - dml_clk_table->dtbclk.clk_values_khz[i] = 0; + dml_clk_table->dtbclk.clk_values_khz[i] = dc_clk_table->entries[i].dtbclk_mhz * 1000; } + } else { + dml_clk_table->dtbclk.clk_values_khz[i] = 0; } } + } - /* socclk */ - if (dc_clk_table->num_entries_per_clk.num_socclk_levels) { - dml_clk_table->socclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_socclk_levels; - for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { - if (i < dml_clk_table->socclk.num_clk_values) { - if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.socclk_mhz && - dc_clk_table->entries[i].socclk_mhz > dc_bw_params->dc_mode_limit.socclk_mhz) { - if (i == 0 || dc_clk_table->entries[i-1].socclk_mhz < dc_bw_params->dc_mode_limit.socclk_mhz) { - dml_clk_table->socclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.socclk_mhz * 1000; - dml_clk_table->socclk.num_clk_values = i + 1; - } else { - dml_clk_table->socclk.clk_values_khz[i] = 0; - dml_clk_table->socclk.num_clk_values = i; - } + /* socclk */ + if (dc_clk_table->num_entries_per_clk.num_socclk_levels) { + dml_clk_table->socclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_socclk_levels; + for (i = 0; i < min(DML_MAX_CLK_TABLE_SIZE, MAX_NUM_DPM_LVL); i++) { + if (i < dml_clk_table->socclk.num_clk_values) { + if (config->use_clock_dc_limits && dc_bw_params->dc_mode_limit.socclk_mhz && + dc_clk_table->entries[i].socclk_mhz > dc_bw_params->dc_mode_limit.socclk_mhz) { + if (i == 0 || dc_clk_table->entries[i-1].socclk_mhz < dc_bw_params->dc_mode_limit.socclk_mhz) { + dml_clk_table->socclk.clk_values_khz[i] = dc_bw_params->dc_mode_limit.socclk_mhz * 1000; + dml_clk_table->socclk.num_clk_values = i + 1; } else { - dml_clk_table->socclk.clk_values_khz[i] = dc_clk_table->entries[i].socclk_mhz * 1000; + dml_clk_table->socclk.clk_values_khz[i] = 0; + dml_clk_table->socclk.num_clk_values = i; } } else { - dml_clk_table->socclk.clk_values_khz[i] = 0; + dml_clk_table->socclk.clk_values_khz[i] = dc_clk_table->entries[i].socclk_mhz * 1000; } + } else { + dml_clk_table->socclk.clk_values_khz[i] = 0; } } - - /* do not override phyclks for now */ - /* phyclk */ - // dml_clk_table->phyclk.num_clk_values = dc_clk_table->num_entries_per_clk.num_phyclk_levels; - // for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { - // dml_clk_table->phyclk.clk_values_khz[i] = dc_clk_table->entries[i].phyclk_mhz * 1000; - // } - - /* phyclk_d18 */ - // dml_clk_table->phyclk_d18.num_clk_values = dc_clk_table->num_entries_per_clk.num_phyclk_d18_levels; - // for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { - // dml_clk_table->phyclk_d18.clk_values_khz[i] = dc_clk_table->entries[i].phyclk_d18_mhz * 1000; - // } - - /* phyclk_d32 */ - // dml_clk_table->phyclk_d32.num_clk_values = dc_clk_table->num_entries_per_clk.num_phyclk_d32_levels; - // for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { - // dml_clk_table->phyclk_d32.clk_values_khz[i] = dc_clk_table->entries[i].phyclk_d32_mhz * 1000; - // } } +} - dml_soc_bb->dchub_refclk_mhz = in_dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000; - dml_soc_bb->dprefclk_mhz = in_dc->clk_mgr->dprefclk_khz / 1000; - dml_soc_bb->xtalclk_mhz = in_dc->ctx->dc_bios->fw_info.pll_info.crystal_frequency / 1000; - dml_soc_bb->dispclk_dppclk_vco_speed_mhz = in_dc->clk_mgr->dentist_vco_freq_khz / 1000.0; +static void override_dml_init_with_values_from_vbios( + struct dml2_initialize_instance_in_out *dml_init, + const struct dml2_configuration_options *config, + const struct dc *in_dc) +{ + const struct clk_bw_params *dc_bw_params = in_dc->clk_mgr->bw_params; + struct dml2_soc_bb *dml_soc_bb = &dml_init->soc_bb; + struct dml2_soc_state_table *dml_clk_table = &dml_init->soc_bb.clk_table; - /* override bounding box paramters from VBIOS */ if (in_dc->ctx->dc_bios->bb_info.dram_clock_change_latency_100ns > 0) dml_soc_bb->power_management_parameters.dram_clk_change_blackout_us = (in_dc->ctx->dc_bios->bb_info.dram_clock_change_latency_100ns + 9) / 10; @@ -308,32 +293,120 @@ void dml21_apply_soc_bb_overrides(struct dml2_initialize_instance_in_out *dml_in dml_clk_table->dram_config.channel_width_bytes = in_dc->ctx->dc_bios->vram_info.dram_channel_width_bytes; } - /* override bounding box paramters from DC config */ - if (in_dc->bb_overrides.sr_exit_time_ns) { - dml_soc_bb->power_management_parameters.stutter_exit_latency_us = - in_dc->bb_overrides.sr_exit_time_ns / 1000.0; + dml_init->soc_bb.xtalclk_mhz = in_dc->ctx->dc_bios->fw_info.pll_info.crystal_frequency / 1000; +} + + +static void override_dml_init_with_values_from_dmub(struct dml2_initialize_instance_in_out *dml_init, + const struct dml2_configuration_options *config, + const struct dc *in_dc) +{ + /* + * TODO - There seems to be overlaps between the values overriden from + * dmub and vbios. Investigate and identify the values that DMUB needs + * to own. + */ +// const struct dmub_soc_bb_params *dmub_bb_params = +// (const struct dmub_soc_bb_params *)config->bb_from_dmub; + +// if (dmub_bb_params == NULL) +// return; + +// if (dmub_bb_params->dram_clk_change_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.dram_clk_change_blackout_us = +// (double) dmub_bb_params->dram_clk_change_blackout_ns / 1000.0; +// if (dmub_bb_params->dram_clk_change_read_only_ns > 0) +// dml_init->soc_bb.power_management_parameters.dram_clk_change_read_only_us = +// (double) dmub_bb_params->dram_clk_change_read_only_ns / 1000.0; +// if (dmub_bb_params->dram_clk_change_write_only_ns > 0) +// dml_init->soc_bb.power_management_parameters.dram_clk_change_write_only_us = +// (double) dmub_bb_params->dram_clk_change_write_only_ns / 1000.0; +// if (dmub_bb_params->fclk_change_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.fclk_change_blackout_us = +// (double) dmub_bb_params->fclk_change_blackout_ns / 1000.0; +// if (dmub_bb_params->g7_ppt_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.g7_ppt_blackout_us = +// (double) dmub_bb_params->g7_ppt_blackout_ns / 1000.0; +// if (dmub_bb_params->stutter_enter_plus_exit_latency_ns > 0) +// dml_init->soc_bb.power_management_parameters.stutter_enter_plus_exit_latency_us = +// (double) dmub_bb_params->stutter_enter_plus_exit_latency_ns / 1000.0; +// if (dmub_bb_params->stutter_exit_latency_ns > 0) +// dml_init->soc_bb.power_management_parameters.stutter_exit_latency_us = +// (double) dmub_bb_params->stutter_exit_latency_ns / 1000.0; +// if (dmub_bb_params->z8_stutter_enter_plus_exit_latency_ns > 0) +// dml_init->soc_bb.power_management_parameters.z8_stutter_enter_plus_exit_latency_us = +// (double) dmub_bb_params->z8_stutter_enter_plus_exit_latency_ns / 1000.0; +// if (dmub_bb_params->z8_stutter_exit_latency_ns > 0) +// dml_init->soc_bb.power_management_parameters.z8_stutter_exit_latency_us = +// (double) dmub_bb_params->z8_stutter_exit_latency_ns / 1000.0; +// if (dmub_bb_params->z8_min_idle_time_ns > 0) +// dml_init->soc_bb.power_management_parameters.z8_min_idle_time = +// (double) dmub_bb_params->z8_min_idle_time_ns / 1000.0; +// #ifndef TRIM_DML2_DCN6B_IP_SENSITIVE +// if (dmub_bb_params->type_b_dram_clk_change_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.lpddr5_dram_clk_change_blackout_us = +// (double) dmub_bb_params->type_b_dram_clk_change_blackout_ns / 1000.0; +// if (dmub_bb_params->type_b_ppt_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.lpddr5_ppt_blackout_us = +// (double) dmub_bb_params->type_b_ppt_blackout_ns / 1000.0; +// #else +// if (dmub_bb_params->type_b_dram_clk_change_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.type_b_dram_clk_change_blackout_us = +// (double) dmub_bb_params->type_b_dram_clk_change_blackout_ns / 1000.0; +// if (dmub_bb_params->type_b_ppt_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.type_b_ppt_blackout_us = +// (double) dmub_bb_params->type_b_ppt_blackout_ns / 1000.0; +// #endif +// if (dmub_bb_params->vmin_limit_dispclk_khz > 0) +// dml_init->soc_bb.vmin_limit.dispclk_khz = dmub_bb_params->vmin_limit_dispclk_khz; +// if (dmub_bb_params->vmin_limit_dcfclk_khz > 0) +// dml_init->soc_bb.vmin_limit.dcfclk_khz = dmub_bb_params->vmin_limit_dcfclk_khz; +// if (dmub_bb_params->g7_temperature_read_blackout_ns > 0) +// dml_init->soc_bb.power_management_parameters.g7_temperature_read_blackout_us = +// (double) dmub_bb_params->g7_temperature_read_blackout_ns / 1000.0; +} + +static void override_dml_init_with_values_from_software_policy(struct dml2_initialize_instance_in_out *dml_init, + const struct dml2_configuration_options *config, + const struct dc *in_dc) +{ + if (!config->use_native_soc_bb_construction) { + dml_init->soc_bb = config->external_socbb_ip_params->soc_bb; + dml_init->ip_caps = config->external_socbb_ip_params->ip_params; } - if (in_dc->bb_overrides.sr_enter_plus_exit_time_ns) { - dml_soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us = + if (in_dc->bb_overrides.sr_exit_time_ns) + dml_init->soc_bb.power_management_parameters.stutter_exit_latency_us = + in_dc->bb_overrides.sr_exit_time_ns / 1000.0; + + if (in_dc->bb_overrides.sr_enter_plus_exit_time_ns) + dml_init->soc_bb.power_management_parameters.stutter_enter_plus_exit_latency_us = in_dc->bb_overrides.sr_enter_plus_exit_time_ns / 1000.0; - } - if (in_dc->bb_overrides.dram_clock_change_latency_ns) { - dml_soc_bb->power_management_parameters.dram_clk_change_blackout_us = + if (in_dc->bb_overrides.dram_clock_change_latency_ns) + dml_init->soc_bb.power_management_parameters.dram_clk_change_blackout_us = in_dc->bb_overrides.dram_clock_change_latency_ns / 1000.0; - } - if (in_dc->bb_overrides.fclk_clock_change_latency_ns) { - dml_soc_bb->power_management_parameters.fclk_change_blackout_us = + if (in_dc->bb_overrides.fclk_clock_change_latency_ns) + dml_init->soc_bb.power_management_parameters.fclk_change_blackout_us = in_dc->bb_overrides.fclk_clock_change_latency_ns / 1000.0; - } +} - //TODO - // if (in_dc->bb_overrides.dummy_clock_change_latency_ns) { - // dml_soc_bb->power_management_parameters.dram_clk_change_blackout_us = - // in_dc->bb_overrides.dram_clock_change_latency_ns / 1000.0; - // } +void dml21_populate_dml_init_params(struct dml2_initialize_instance_in_out *dml_init, + const struct dml2_configuration_options *config, + const struct dc *in_dc) +{ + populate_default_dml_init_params(dml_init, config, in_dc); + + override_dml_init_with_values_from_hardware_default(dml_init, config, in_dc); + + override_dml_init_with_values_from_smu(dml_init, config, in_dc); + + override_dml_init_with_values_from_vbios(dml_init, config, in_dc); + + override_dml_init_with_values_from_dmub(dml_init, config, in_dc); + + override_dml_init_with_values_from_software_policy(dml_init, config, in_dc); } static unsigned int calc_max_hardware_v_total(const struct dc_stream_state *stream) @@ -726,7 +799,6 @@ static void populate_dml21_surface_config_from_plane_state( switch (plane_state->tiling_info.gfxversion) { case DcGfxVersion7: case DcGfxVersion8: - // Placeholder for programming the array_mode break; case DcGfxVersion9: case DcGfxVersion10: @@ -788,6 +860,7 @@ static void populate_dml21_plane_config_from_plane_state(struct dml2_context *dm plane->pixel_format = dml2_420_10; break; case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616: + case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616: case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F: case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F: plane->pixel_format = dml2_444_64; @@ -888,10 +961,8 @@ static void populate_dml21_plane_config_from_plane_state(struct dml2_context *dm case DC_CM2_GPU_MEM_SIZE_171717: plane->tdlut.tdlut_width_mode = dml2_tdlut_width_17_cube; break; - case DC_CM2_GPU_MEM_SIZE_333333: - plane->tdlut.tdlut_width_mode = dml2_tdlut_width_33_cube; - break; case DC_CM2_GPU_MEM_SIZE_TRANSFORMED: + default: //plane->tdlut.tdlut_width_mode = dml2_tdlut_width_flatten; // dml2_tdlut_width_flatten undefined break; } diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h index 73a013be1e48..9880d3e0398e 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h @@ -17,9 +17,7 @@ struct dml2_context; struct dml2_configuration_options; struct dml2_initialize_instance_in_out; -void dml21_apply_soc_bb_overrides(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc); -void dml21_initialize_soc_bb_params(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc); -void dml21_initialize_ip_params(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc); +void dml21_populate_dml_init_params(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc); bool dml21_map_dc_state_into_dml_display_cfg(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx); void dml21_copy_clocks_to_dc_state(struct dml2_context *in_ctx, struct dc_state *context); void dml21_extract_watermark_sets(const struct dc *in_dc, union dcn_watermark_set *watermarks, struct dml2_context *in_ctx); diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c index 930e86cdb88a..ee721606b883 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c @@ -384,6 +384,7 @@ void dml21_build_fams2_programming(const struct dc *dc, /* reset fams2 data */ memset(&context->bw_ctx.bw.dcn.fams2_stream_base_params, 0, sizeof(union dmub_cmd_fams2_config) * DML2_MAX_PLANES); memset(&context->bw_ctx.bw.dcn.fams2_stream_sub_params, 0, sizeof(union dmub_cmd_fams2_config) * DML2_MAX_PLANES); + memset(&context->bw_ctx.bw.dcn.fams2_stream_sub_params_v2, 0, sizeof(union dmub_fams2_stream_static_sub_state_v2) * DML2_MAX_PLANES); memset(&context->bw_ctx.bw.dcn.fams2_global_config, 0, sizeof(struct dmub_cmd_fams2_global_config)); if (dml_ctx->v21.mode_programming.programming->fams2_required) { @@ -414,9 +415,16 @@ void dml21_build_fams2_programming(const struct dc *dc, memcpy(static_base_state, &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_base_params, sizeof(union dmub_cmd_fams2_config)); - memcpy(static_sub_state, - &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_sub_params, - sizeof(union dmub_cmd_fams2_config)); + + if (dc->debug.fams_version.major == 3) { + memcpy(&context->bw_ctx.bw.dcn.fams2_stream_sub_params_v2[num_fams2_streams], + &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_sub_params_v2, + sizeof(union dmub_fams2_stream_static_sub_state_v2)); + } else { + memcpy(static_sub_state, + &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_sub_params, + sizeof(union dmub_cmd_fams2_config)); + } switch (dc->debug.fams_version.minor) { case 1: diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c index 208d3651b6ba..03de3cf06ae5 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c @@ -2,8 +2,6 @@ // // Copyright 2024 Advanced Micro Devices, Inc. -#include <linux/vmalloc.h> - #include "dml2_internal_types.h" #include "dml_top.h" #include "dml2_core_dcn4_calcs.h" @@ -37,15 +35,11 @@ static bool dml21_allocate_memory(struct dml2_context **dml_ctx) return true; } -static void dml21_apply_debug_options(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config) +static void dml21_populate_configuration_options(const struct dc *in_dc, + struct dml2_context *dml_ctx, + const struct dml2_configuration_options *config) { - bool disable_fams2; - struct dml2_pmo_options *pmo_options = &dml_ctx->v21.dml_init.options.pmo_options; - - /* ODM options */ - pmo_options->disable_dyn_odm = !config->minimize_dispclk_using_odm; - pmo_options->disable_dyn_odm_for_multi_stream = true; - pmo_options->disable_dyn_odm_for_stream_with_svp = true; + dml_ctx->config = *config; /* UCLK P-State options */ if (in_dc->debug.dml21_force_pstate_method) { @@ -55,52 +49,20 @@ static void dml21_apply_debug_options(const struct dc *in_dc, struct dml2_contex } else { dml_ctx->config.pmo.force_pstate_method_enable = false; } - - pmo_options->disable_vblank = ((in_dc->debug.dml21_disable_pstate_method_mask >> 1) & 1); - - /* NOTE: DRR and SubVP Require FAMS2 */ - disable_fams2 = !in_dc->debug.fams2_config.bits.enable; - pmo_options->disable_svp = ((in_dc->debug.dml21_disable_pstate_method_mask >> 2) & 1) || - in_dc->debug.force_disable_subvp || - disable_fams2; - pmo_options->disable_drr_clamped = ((in_dc->debug.dml21_disable_pstate_method_mask >> 3) & 1) || - disable_fams2; - pmo_options->disable_drr_var = ((in_dc->debug.dml21_disable_pstate_method_mask >> 4) & 1) || - disable_fams2; - pmo_options->disable_fams2 = disable_fams2; - - pmo_options->disable_drr_var_when_var_active = in_dc->debug.disable_fams_gaming == INGAME_FAMS_DISABLE || - in_dc->debug.disable_fams_gaming == INGAME_FAMS_MULTI_DISP_CLAMPED_ONLY; - pmo_options->disable_drr_clamped_when_var_active = in_dc->debug.disable_fams_gaming == INGAME_FAMS_DISABLE; } -static void dml21_init(const struct dc *in_dc, struct dml2_context **dml_ctx, const struct dml2_configuration_options *config) +static void dml21_init(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config) { - switch (in_dc->ctx->dce_version) { - case DCN_VERSION_4_01: - (*dml_ctx)->v21.dml_init.options.project_id = dml2_project_dcn4x_stage2_auto_drr_svp; - break; - default: - (*dml_ctx)->v21.dml_init.options.project_id = dml2_project_invalid; - } - (*dml_ctx)->architecture = dml2_architecture_21; + dml_ctx->architecture = dml2_architecture_21; - /* Store configuration options */ - (*dml_ctx)->config = *config; + dml21_populate_configuration_options(in_dc, dml_ctx, config); DC_FP_START(); - /*Initialize SOCBB and DCNIP params */ - dml21_initialize_soc_bb_params(&(*dml_ctx)->v21.dml_init, config, in_dc); - dml21_initialize_ip_params(&(*dml_ctx)->v21.dml_init, config, in_dc); - dml21_apply_soc_bb_overrides(&(*dml_ctx)->v21.dml_init, config, in_dc); + dml21_populate_dml_init_params(&dml_ctx->v21.dml_init, config, in_dc); - /* apply debug overrides */ - dml21_apply_debug_options(in_dc, *dml_ctx, config); - - /*Initialize DML21 instance */ - dml2_initialize_instance(&(*dml_ctx)->v21.dml_init); + dml2_initialize_instance(&dml_ctx->v21.dml_init); DC_FP_END(); } @@ -111,7 +73,7 @@ bool dml21_create(const struct dc *in_dc, struct dml2_context **dml_ctx, const s if (!dml21_allocate_memory(dml_ctx)) return false; - dml21_init(in_dc, dml_ctx, config); + dml21_init(in_dc, *dml_ctx, config); return true; } @@ -328,12 +290,13 @@ static bool dml21_check_mode_support(const struct dc *in_dc, struct dc_state *co return true; } -bool dml21_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx, bool fast_validate) +bool dml21_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx, + enum dc_validate_mode validate_mode) { bool out = false; - /* Use dml_validate_only for fast_validate path */ - if (fast_validate) + /* Use dml21_check_mode_support for DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX path */ + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) out = dml21_check_mode_support(in_dc, context, dml_ctx); else out = dml21_mode_check_and_programming(in_dc, context, dml_ctx); @@ -496,7 +459,7 @@ bool dml21_create_copy(struct dml2_context **dst_dml_ctx, return true; } -void dml21_reinit(const struct dc *in_dc, struct dml2_context **dml_ctx, const struct dml2_configuration_options *config) +void dml21_reinit(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config) { dml21_init(in_dc, dml_ctx, config); } diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h index 42e715024bc9..15f92029d2e5 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h @@ -14,6 +14,7 @@ struct dc; struct dc_state; struct dml2_configuration_options; struct dml2_context; +enum dc_validate_mode; /** * dml2_create - Creates dml21_context. @@ -33,22 +34,23 @@ void dml21_copy(struct dml2_context *dst_dml_ctx, struct dml2_context *src_dml_ctx); bool dml21_create_copy(struct dml2_context **dst_dml_ctx, struct dml2_context *src_dml_ctx); -void dml21_reinit(const struct dc *in_dc, struct dml2_context **dml_ctx, const struct dml2_configuration_options *config); +void dml21_reinit(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config); /** * dml21_validate - Determines if a display configuration is supported or not. * @in_dc: dc. * @context: dc_state to be validated. - * @fast_validate: Fast validate will not populate context.res_ctx. + * @validate_mode: DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX + * will not populate context.res_ctx. * * Based on fast_validate option internally would call: * - * -dml21_mode_check_and_programming - for non fast_validate option + * -dml21_mode_check_and_programming - for DC_VALIDATE_MODE_AND_PROGRAMMING option * Calculates if dc_state can be supported on the input display * configuration. If supported, generates the necessary HW * programming for the new dc_state. * - * -dml21_check_mode_support - for fast_validate option + * -dml21_check_mode_support - for DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX option * Calculates if dc_state can be supported for the input display * config. @@ -56,7 +58,8 @@ void dml21_reinit(const struct dc *in_dc, struct dml2_context **dml_ctx, const s * separate dc_states for validation. * Return: True if mode is supported, false otherwise. */ -bool dml21_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx, bool fast_validate); +bool dml21_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx, + enum dc_validate_mode validate_mode); /* Prepare hubp mcache_regs for hubp mcache ID and split coordinate programming */ void dml21_prepare_mcache_programming(struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx); diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top.h index c047d56527c4..a64ec4dcf11a 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top.h @@ -43,5 +43,4 @@ bool dml2_build_mode_programming(struct dml2_build_mode_programming_in_out *in_o */ bool dml2_build_mcache_programming(struct dml2_build_mcache_programming_in_out *in_out); - #endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h index 84c90050668c..b05030926ce8 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h @@ -46,6 +46,7 @@ struct dml2_display_dlg_regs { uint32_t dst_y_delta_drq_limit; uint32_t refcyc_per_vm_dmdata; uint32_t dmdata_dl_delta; + uint32_t dst_y_svp_drq_limit; // MRQ uint32_t refcyc_per_meta_chunk_vblank_l; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h index 255f05de362c..e8dc6471c0be 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h @@ -222,6 +222,7 @@ struct dml2_composition_cfg { struct { bool enabled; + bool upsp_enabled; struct { double h_ratio; double v_ratio; @@ -426,6 +427,7 @@ struct dml2_stream_parameters { struct dml2_display_cfg { bool gpuvm_enable; + bool ffbm_enable; bool hostvm_enable; // Allocate DET proportionally between streams based on pixel rate diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h index 5f0bc42d1d2f..8c9f414aa6bf 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h @@ -93,12 +93,15 @@ struct dml2_soc_power_management_parameters { double dram_clk_change_write_only_us; double fclk_change_blackout_us; double g7_ppt_blackout_us; + double g7_temperature_read_blackout_us; double stutter_enter_plus_exit_latency_us; double stutter_exit_latency_us; double z8_stutter_enter_plus_exit_latency_us; double z8_stutter_exit_latency_us; double z8_min_idle_time; double g6_temp_read_blackout_us[DML_MAX_CLK_TABLE_SIZE]; + double type_b_dram_clk_change_blackout_us; + double type_b_ppt_blackout_us; }; struct dml2_clk_table { @@ -130,6 +133,7 @@ struct dml2_soc_state_table { struct dml2_soc_vmin_clock_limits { unsigned long dispclk_khz; + unsigned long dcfclk_khz; }; struct dml2_soc_bb { @@ -138,6 +142,7 @@ struct dml2_soc_bb { struct dml2_soc_power_management_parameters power_management_parameters; struct dml2_soc_vmin_clock_limits vmin_limit; + double lower_bound_bandwidth_dchub; unsigned int dprefclk_mhz; unsigned int xtalclk_mhz; unsigned int pcie_refclk_mhz; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h index 0dbf886d8926..98c0234e2f47 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h @@ -53,7 +53,9 @@ enum dml2_output_type_and_rate__rate { dml2_output_rate_hdmi_rate_6x4 = 9, dml2_output_rate_hdmi_rate_8x4 = 10, dml2_output_rate_hdmi_rate_10x4 = 11, - dml2_output_rate_hdmi_rate_12x4 = 12 + dml2_output_rate_hdmi_rate_12x4 = 12, + dml2_output_rate_hdmi_rate_16x4 = 13, + dml2_output_rate_hdmi_rate_20x4 = 14 }; struct dml2_pmo_options { @@ -279,7 +281,10 @@ struct dml2_per_stream_programming { } phantom_stream; union dmub_cmd_fams2_config fams2_base_params; - union dmub_cmd_fams2_config fams2_sub_params; + union { + union dmub_cmd_fams2_config fams2_sub_params; + union dmub_fams2_stream_static_sub_state_v2 fams2_sub_params_v2; + }; }; //----------------- @@ -674,9 +679,14 @@ struct dml2_display_cfg_programming { // unlimited # of mcache struct dml2_mcache_surface_allocation non_optimized_mcache_allocation[DML2_MAX_PLANES]; + bool failed_prefetch; + bool failed_uclk_pstate; bool failed_mcache_validation; bool failed_dpmm; bool failed_mode_programming; + bool failed_mode_programming_dcfclk; + bool failed_mode_programming_prefetch; + bool failed_mode_programming_flip; bool failed_map_watermarks; } informative; }; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c index c4dad7164d31..b9cff2198511 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c @@ -4685,7 +4685,10 @@ static void calculate_tdlut_setting( //the tdlut is fetched during the 2 row times of prefetch. if (p->setup_for_tdlut) { *p->tdlut_groups_per_2row_ub = (unsigned int)math_ceil2((double) *p->tdlut_bytes_per_frame / *p->tdlut_bytes_per_group, 1); - *p->tdlut_opt_time = (*p->tdlut_bytes_per_frame - p->cursor_buffer_size * 1024) / tdlut_drain_rate; + if (*p->tdlut_bytes_per_frame > p->cursor_buffer_size * 1024) + *p->tdlut_opt_time = (*p->tdlut_bytes_per_frame - p->cursor_buffer_size * 1024) / tdlut_drain_rate; + else + *p->tdlut_opt_time = 0; *p->tdlut_drain_time = p->cursor_buffer_size * 1024 / tdlut_drain_rate; *p->tdlut_bytes_to_deliver = (unsigned int) (p->cursor_buffer_size * 1024.0); } @@ -4858,7 +4861,7 @@ static double get_urgent_bandwidth_required( double ReadBandwidthChroma[], double PrefetchBandwidthLuma[], double PrefetchBandwidthChroma[], - double PrefetchBandwidthOto[], + double PrefetchBandwidthMax[], double excess_vactive_fill_bw_l[], double excess_vactive_fill_bw_c[], double cursor_bw[], @@ -4922,9 +4925,9 @@ static double get_urgent_bandwidth_required( l->vm_row_bw = NumberOfDPP[k] * prefetch_vmrow_bw[k]; l->flip_and_active_bw = l->per_plane_flip_bw[k] + ReadBandwidthLuma[k] * l->adj_factor_p0 + ReadBandwidthChroma[k] * l->adj_factor_p1 + cursor_bw[k] * l->adj_factor_cur; l->flip_and_prefetch_bw = l->per_plane_flip_bw[k] + NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * l->adj_factor_p0_pre + PrefetchBandwidthChroma[k] * l->adj_factor_p1_pre) + prefetch_cursor_bw[k] * l->adj_factor_cur_pre; - l->flip_and_prefetch_bw_oto = l->per_plane_flip_bw[k] + NumberOfDPP[k] * (PrefetchBandwidthOto[k] * l->adj_factor_p0_pre + PrefetchBandwidthChroma[k] * l->adj_factor_p1_pre) + prefetch_cursor_bw[k] * l->adj_factor_cur_pre; + l->flip_and_prefetch_bw_max = l->per_plane_flip_bw[k] + NumberOfDPP[k] * (PrefetchBandwidthMax[k] * l->adj_factor_p0_pre + PrefetchBandwidthChroma[k] * l->adj_factor_p1_pre) + prefetch_cursor_bw[k] * l->adj_factor_cur_pre; l->active_and_excess_bw = (ReadBandwidthLuma[k] + excess_vactive_fill_bw_l[k]) * l->tmp_nom_adj_factor_p0 + (ReadBandwidthChroma[k] + excess_vactive_fill_bw_c[k]) * l->tmp_nom_adj_factor_p1 + dpte_row_bw[k] + meta_row_bw[k]; - surface_required_bw[k] = math_max5(l->vm_row_bw, l->flip_and_active_bw, l->flip_and_prefetch_bw, l->active_and_excess_bw, l->flip_and_prefetch_bw_oto); + surface_required_bw[k] = math_max5(l->vm_row_bw, l->flip_and_active_bw, l->flip_and_prefetch_bw, l->active_and_excess_bw, l->flip_and_prefetch_bw_max); /* export peak required bandwidth for the surface */ surface_peak_required_bw[k] = math_max2(surface_required_bw[k], surface_peak_required_bw[k]); @@ -5122,7 +5125,7 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch s->Tsw_est3 = 0.0; s->cursor_prefetch_bytes = 0; *p->prefetch_cursor_bw = 0; - *p->RequiredPrefetchBWOTO = 0.0; + *p->RequiredPrefetchBWMax = 0.0; dcc_mrq_enable = (p->dcc_enable && p->mrq_present); @@ -5353,7 +5356,7 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch * mp will fail if ms decides to use equ schedule and mp decides to use oto schedule * and the required bandwidth increases when going from ms to mp */ - *p->RequiredPrefetchBWOTO = s->prefetch_bw_oto; + *p->RequiredPrefetchBWMax = s->prefetch_bw_oto; #ifdef __DML_VBA_DEBUG__ DML_LOG_VERBOSE("DML::%s: vactive_sw_bw_l = %f\n", __func__, p->vactive_sw_bw_l); @@ -5715,8 +5718,14 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch s->TimeForFetchingVM = s->Tvm_equ; s->TimeForFetchingRowInVBlank = s->Tr0_equ; - *p->dst_y_per_vm_vblank = math_ceil2(4.0 * s->TimeForFetchingVM / s->LineTime, 1.0) / 4.0; - *p->dst_y_per_row_vblank = math_ceil2(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; + *p->dst_y_per_vm_vblank = math_ceil2(4.0 * s->TimeForFetchingVM / s->LineTime, 1.0) / 4.0; + *p->dst_y_per_row_vblank = math_ceil2(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; + + /* equ bw should be propagated so a ceiling of the equ bw is accounted for prior to mode programming. + * Overall bandwidth may be lower when going from mode support to mode programming but final pixel data + * bandwidth may end up higher than what was calculated in mode support. + */ + *p->RequiredPrefetchBWMax = math_max2(s->prefetch_bw_equ, *p->RequiredPrefetchBWMax); #ifdef __DML_VBA_DEBUG__ DML_LOG_VERBOSE("DML::%s: Using equ bw scheduling for prefetch\n", __func__); @@ -6112,7 +6121,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, l->zero_array, //PrefetchBandwidthLuma, l->zero_array, //PrefetchBandwidthChroma, - l->zero_array, //PrefetchBWOTO + l->zero_array, //PrefetchBWMax l->zero_array, l->zero_array, l->zero_array, @@ -6149,7 +6158,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, l->zero_array, //PrefetchBandwidthLuma, l->zero_array, //PrefetchBandwidthChroma, - l->zero_array, //PrefetchBWOTO + l->zero_array, //PrefetchBWMax p->excess_vactive_fill_bw_l, p->excess_vactive_fill_bw_c, p->cursor_bw, @@ -6186,7 +6195,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, p->prefetch_bandwidth_l, p->prefetch_bandwidth_c, - p->prefetch_bandwidth_oto, // to prevent ms/mp mismatch when oto bw > total vactive bw + p->prefetch_bandwidth_max, // to prevent ms/mp mismatches where mp prefetch bw > ms prefetch bw p->excess_vactive_fill_bw_l, p->excess_vactive_fill_bw_c, p->cursor_bw, @@ -6223,7 +6232,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, p->prefetch_bandwidth_l, p->prefetch_bandwidth_c, - p->prefetch_bandwidth_oto, // to prevent ms/mp mismatch when oto bw > total vactive bw + p->prefetch_bandwidth_max, // to prevent ms/mp mismatch where mp prefetch bw > ms prefetch bw p->excess_vactive_fill_bw_l, p->excess_vactive_fill_bw_c, p->cursor_bw, @@ -6260,7 +6269,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, p->prefetch_bandwidth_l, p->prefetch_bandwidth_c, - p->prefetch_bandwidth_oto, // to prevent ms/mp mismatch when oto bw > total vactive bw + p->prefetch_bandwidth_max, // to prevent ms/mp mismatches where mp prefetch bw > ms prefetch bw p->excess_vactive_fill_bw_l, p->excess_vactive_fill_bw_c, p->cursor_bw, @@ -7487,7 +7496,7 @@ static noinline_for_stack void dml_core_ms_prefetch_check(struct dml2_core_inter CalculatePrefetchSchedule_params->VRatioPrefetchC = &mode_lib->ms.VRatioPreC[k]; CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWLuma = &mode_lib->ms.RequiredPrefetchPixelDataBWLuma[k]; // prefetch_sw_bw_l CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWChroma = &mode_lib->ms.RequiredPrefetchPixelDataBWChroma[k]; // prefetch_sw_bw_c - CalculatePrefetchSchedule_params->RequiredPrefetchBWOTO = &mode_lib->ms.RequiredPrefetchBWOTO[k]; + CalculatePrefetchSchedule_params->RequiredPrefetchBWMax = &mode_lib->ms.RequiredPrefetchBWMax[k]; CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &mode_lib->ms.NoTimeForDynamicMetadata[k]; CalculatePrefetchSchedule_params->Tno_bw = &mode_lib->ms.Tno_bw[k]; CalculatePrefetchSchedule_params->Tno_bw_flip = &mode_lib->ms.Tno_bw_flip[k]; @@ -7632,7 +7641,7 @@ static noinline_for_stack void dml_core_ms_prefetch_check(struct dml2_core_inter calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->ms.vactive_sw_bw_c; calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->ms.RequiredPrefetchPixelDataBWLuma; calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->ms.RequiredPrefetchPixelDataBWChroma; - calculate_peak_bandwidth_params->prefetch_bandwidth_oto = mode_lib->ms.RequiredPrefetchBWOTO; + calculate_peak_bandwidth_params->prefetch_bandwidth_max = mode_lib->ms.RequiredPrefetchBWMax; calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->ms.excess_vactive_fill_bw_l; calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->ms.excess_vactive_fill_bw_c; calculate_peak_bandwidth_params->cursor_bw = mode_lib->ms.cursor_bw; @@ -7799,7 +7808,7 @@ static noinline_for_stack void dml_core_ms_prefetch_check(struct dml2_core_inter calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->ms.vactive_sw_bw_c; calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->ms.RequiredPrefetchPixelDataBWLuma; calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->ms.RequiredPrefetchPixelDataBWChroma; - calculate_peak_bandwidth_params->prefetch_bandwidth_oto = mode_lib->ms.RequiredPrefetchBWOTO; + calculate_peak_bandwidth_params->prefetch_bandwidth_max = mode_lib->ms.RequiredPrefetchBWMax; calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->ms.excess_vactive_fill_bw_l; calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->ms.excess_vactive_fill_bw_c; calculate_peak_bandwidth_params->cursor_bw = mode_lib->ms.cursor_bw; @@ -7905,6 +7914,7 @@ static noinline_for_stack void dml_core_ms_prefetch_check(struct dml2_core_inter } + static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out_params) { struct dml2_core_internal_display_mode_lib *mode_lib = in_out_params->mode_lib; @@ -11253,7 +11263,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex CalculatePrefetchSchedule_params->VRatioPrefetchC = &mode_lib->mp.VRatioPrefetchC[k]; CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWLuma = &mode_lib->mp.RequiredPrefetchPixelDataBWLuma[k]; CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWChroma = &mode_lib->mp.RequiredPrefetchPixelDataBWChroma[k]; - CalculatePrefetchSchedule_params->RequiredPrefetchBWOTO = &s->dummy_single_array[0][k]; + CalculatePrefetchSchedule_params->RequiredPrefetchBWMax = &s->dummy_single_array[0][k]; CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &mode_lib->mp.NotEnoughTimeForDynamicMetadata[k]; CalculatePrefetchSchedule_params->Tno_bw = &mode_lib->mp.Tno_bw[k]; CalculatePrefetchSchedule_params->Tno_bw_flip = &mode_lib->mp.Tno_bw_flip[k]; @@ -11396,7 +11406,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->mp.vactive_sw_bw_c; calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->mp.RequiredPrefetchPixelDataBWLuma; calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->mp.RequiredPrefetchPixelDataBWChroma; - calculate_peak_bandwidth_params->prefetch_bandwidth_oto = s->dummy_single_array[0]; + calculate_peak_bandwidth_params->prefetch_bandwidth_max = s->dummy_single_array[0]; calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->mp.excess_vactive_fill_bw_l; calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->mp.excess_vactive_fill_bw_c; calculate_peak_bandwidth_params->cursor_bw = mode_lib->mp.cursor_bw; @@ -11536,7 +11546,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex calculate_peak_bandwidth_params->meta_row_bw = mode_lib->mp.meta_row_bw; calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->mp.prefetch_cursor_bw; calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->mp.prefetch_vmrow_bw; - calculate_peak_bandwidth_params->prefetch_bandwidth_oto = s->dummy_single_array[0]; + calculate_peak_bandwidth_params->prefetch_bandwidth_max = s->dummy_single_array[0]; calculate_peak_bandwidth_params->flip_bw = mode_lib->mp.final_flip_bw; calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->mp.UrgentBurstFactorLuma; calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->mp.UrgentBurstFactorChroma; @@ -11880,7 +11890,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex } //Maximum Bandwidth Used - s->TotalWRBandwidth = 0; + mode_lib->mp.TotalWRBandwidth = 0; for (k = 0; k < display_cfg->num_streams; ++k) { s->WRBandwidth = 0; if (display_cfg->stream_descriptors[k].writeback.active_writebacks_per_stream > 0) { @@ -11889,7 +11899,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex (display_cfg->stream_descriptors[k].timing.h_total * display_cfg->stream_descriptors[k].writeback.writeback_stream[0].input_height / ((double)display_cfg->stream_descriptors[k].timing.pixel_clock_khz / 1000)) * (display_cfg->stream_descriptors[k].writeback.writeback_stream[0].pixel_format == dml2_444_32 ? 4.0 : 8.0); - s->TotalWRBandwidth = s->TotalWRBandwidth + s->WRBandwidth; + mode_lib->mp.TotalWRBandwidth = mode_lib->mp.TotalWRBandwidth + s->WRBandwidth; } } @@ -13059,6 +13069,10 @@ void dml2_core_calcs_get_informative(const struct dml2_core_internal_display_mod out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_10x4; else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_12x4) out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_12x4; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_16x4) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_16x4; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_20x4) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_20x4; out->informative.mode_support_info.AlignedYPitch[k] = mode_lib->ms.support.AlignedYPitch[k]; out->informative.mode_support_info.AlignedCPitch[k] = mode_lib->ms.support.AlignedCPitch[k]; @@ -13243,7 +13257,7 @@ void dml2_core_calcs_get_informative(const struct dml2_core_internal_display_mod out->informative.misc.DisplayPipeLineDeliveryTimeLumaPrefetch[k] = mode_lib->mp.DisplayPipeLineDeliveryTimeLumaPrefetch[k]; out->informative.misc.DisplayPipeLineDeliveryTimeChromaPrefetch[k] = mode_lib->mp.DisplayPipeLineDeliveryTimeChromaPrefetch[k]; - out->informative.misc.WritebackRequiredBandwidth = mode_lib->scratch.dml_core_mode_programming_locals.TotalWRBandwidth / 1000.0; + out->informative.misc.WritebackRequiredBandwidth = mode_lib->mp.TotalWRBandwidth / 1000.0; out->informative.misc.WritebackAllowDRAMClockChangeEndPosition[k] = mode_lib->mp.WritebackAllowDRAMClockChangeEndPosition[k]; out->informative.misc.WritebackAllowFCLKChangeEndPosition[k] = mode_lib->mp.WritebackAllowFCLKChangeEndPosition[k]; out->informative.misc.DSCCLK_calculated[k] = mode_lib->mp.DSCCLK[k]; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h index bdee6ad7bc59..28687565ac22 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h @@ -102,6 +102,7 @@ struct dml2_core_internal_DmlPipe { double DCFClkDeepSleep; unsigned int DPPPerSurface; bool ScalerEnabled; + bool UPSPEnabled; enum dml2_rotation_angle RotationAngle; bool mirrored; unsigned int ViewportHeight; @@ -186,7 +187,9 @@ enum dml2_core_internal_output_type_rate { dml2_core_internal_output_rate_hdmi_rate_6x4 = 9, dml2_core_internal_output_rate_hdmi_rate_8x4 = 10, dml2_core_internal_output_rate_hdmi_rate_10x4 = 11, - dml2_core_internal_output_rate_hdmi_rate_12x4 = 12 + dml2_core_internal_output_rate_hdmi_rate_12x4 = 12, + dml2_core_internal_output_rate_hdmi_rate_16x4 = 13, + dml2_core_internal_output_rate_hdmi_rate_20x4 = 14 }; struct dml2_core_internal_watermarks { @@ -260,12 +263,14 @@ struct dml2_core_internal_mode_support_info { bool AvgBandwidthSupport; bool UrgVactiveBandwidthSupport; bool EnoughUrgentLatencyHidingSupport; + bool PrefetchScheduleSupported; bool PrefetchSupported; bool PrefetchBandwidthSupported; bool DynamicMetadataSupported; bool VRatioInPrefetchSupported; bool DISPCLK_DPPCLK_Support; bool TotalAvailablePipesSupport; + bool ODMSupport; bool ModeSupport; bool ViewportSizeSupport; @@ -314,9 +319,7 @@ struct dml2_core_internal_mode_support_info { double non_urg_bandwidth_required[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]; // same as urg_bandwidth, except not scaled by urg burst factor double non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]; - bool avg_bandwidth_support_ok[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]; - double max_urgent_latency_us; double max_non_urgent_latency_us; double avg_non_urgent_latency_us; @@ -329,6 +332,8 @@ struct dml2_core_internal_mode_support_info { bool temp_read_or_ppt_support; struct dml2_core_internal_watermarks watermarks; + bool dcfclk_support; + bool qos_bandwidth_support; }; struct dml2_core_internal_mode_support { @@ -350,9 +355,11 @@ struct dml2_core_internal_mode_support { double SOCCLK; /// <brief Basically just the clock freq at the min (or given) state double DCFCLK; /// <brief Basically just the clock freq at the min (or given) state and max combine setting double GlobalDPPCLK; /// <brief the Max DPPCLK freq out of all pipes + double GlobalDTBCLK; /// <brief the Max DTBCLK freq out of all pipes double uclk_freq_mhz; double dram_bw_mbps; double max_dram_bw_mbps; + double min_available_urgent_bandwidth_MBps; /// <brief Minimum guaranteed available urgent return bandwidth in MBps double MaxFabricClock; /// <brief Basically just the clock freq at the min (or given) state double MaxDCFCLK; /// <brief Basically just the clock freq at the min (or given) state and max combine setting @@ -394,9 +401,13 @@ struct dml2_core_internal_mode_support { double TWait[DML2_MAX_PLANES]; bool UnboundedRequestEnabled; + unsigned int compbuf_reserved_space_64b; + bool hw_debug5; unsigned int CompressedBufferSizeInkByte; double VRatioPreY[DML2_MAX_PLANES]; double VRatioPreC[DML2_MAX_PLANES]; + unsigned int req_per_swath_ub_l[DML2_MAX_PLANES]; + unsigned int req_per_swath_ub_c[DML2_MAX_PLANES]; unsigned int swath_width_luma_ub[DML2_MAX_PLANES]; unsigned int swath_width_chroma_ub[DML2_MAX_PLANES]; unsigned int RequiredSlots[DML2_MAX_PLANES]; @@ -417,8 +428,8 @@ struct dml2_core_internal_mode_support { double dst_y_prefetch[DML2_MAX_PLANES]; double LinesForVM[DML2_MAX_PLANES]; double LinesForDPTERow[DML2_MAX_PLANES]; - double SwathWidthYSingleDPP[DML2_MAX_PLANES]; - double SwathWidthCSingleDPP[DML2_MAX_PLANES]; + unsigned int SwathWidthYSingleDPP[DML2_MAX_PLANES]; + unsigned int SwathWidthCSingleDPP[DML2_MAX_PLANES]; unsigned int BytePerPixelY[DML2_MAX_PLANES]; unsigned int BytePerPixelC[DML2_MAX_PLANES]; double BytePerPixelInDETY[DML2_MAX_PLANES]; @@ -469,13 +480,58 @@ struct dml2_core_internal_mode_support { double mall_prefetch_sdp_overhead_factor[DML2_MAX_PLANES]; // overhead to the imall or phantom pipe double mall_prefetch_dram_overhead_factor[DML2_MAX_PLANES]; + bool is_using_mall_for_ss[DML2_MAX_PLANES]; + unsigned int meta_row_width_chroma[DML2_MAX_PLANES]; + unsigned int PixelPTEReqHeightC[DML2_MAX_PLANES]; + bool PTE_BUFFER_MODE[DML2_MAX_PLANES]; + unsigned int meta_req_height_chroma[DML2_MAX_PLANES]; + unsigned int meta_pte_bytes_per_frame_ub_c[DML2_MAX_PLANES]; + unsigned int dpde0_bytes_per_frame_ub_c[DML2_MAX_PLANES]; + unsigned int dpte_row_width_luma_ub[DML2_MAX_PLANES]; + unsigned int meta_req_width[DML2_MAX_PLANES]; + unsigned int meta_row_width[DML2_MAX_PLANES]; + unsigned int PixelPTEReqWidthY[DML2_MAX_PLANES]; + unsigned int dpte_row_height_linear[DML2_MAX_PLANES]; + unsigned int PTERequestSizeY[DML2_MAX_PLANES]; + unsigned int dpte_row_width_chroma_ub[DML2_MAX_PLANES]; + unsigned int PixelPTEReqWidthC[DML2_MAX_PLANES]; + unsigned int meta_pte_bytes_per_frame_ub_l[DML2_MAX_PLANES]; + unsigned int dpte_row_height_linear_chroma[DML2_MAX_PLANES]; + unsigned int PTERequestSizeC[DML2_MAX_PLANES]; + unsigned int meta_req_height[DML2_MAX_PLANES]; + unsigned int dpde0_bytes_per_frame_ub_l[DML2_MAX_PLANES]; + unsigned int meta_req_width_chroma[DML2_MAX_PLANES]; + unsigned int PixelPTEReqHeightY[DML2_MAX_PLANES]; + unsigned int BIGK_FRAGMENT_SIZE[DML2_MAX_PLANES]; + unsigned int vm_group_bytes[DML2_MAX_PLANES]; + unsigned int VReadyOffsetPix[DML2_MAX_PLANES]; + unsigned int VUpdateOffsetPix[DML2_MAX_PLANES]; + unsigned int VUpdateWidthPix[DML2_MAX_PLANES]; + double TSetup[DML2_MAX_PLANES]; + double Tdmdl_vm_raw[DML2_MAX_PLANES]; + double Tdmdl_raw[DML2_MAX_PLANES]; + unsigned int VStartupMin[DML2_MAX_PLANES]; /// <brief Minimum vstartup to meet the prefetch schedule (i.e. the prefetch solution can be found at this vstartup time); not the actual global sync vstartup pos. + double MaxActiveDRAMClockChangeLatencySupported[DML2_MAX_PLANES]; + double MaxActiveFCLKChangeLatencySupported; + // Backend bool RequiresDSC[DML2_MAX_PLANES]; bool RequiresFEC[DML2_MAX_PLANES]; double OutputBpp[DML2_MAX_PLANES]; + double DesiredOutputBpp[DML2_MAX_PLANES]; + double PixelClockBackEnd[DML2_MAX_PLANES]; unsigned int DSCDelay[DML2_MAX_PLANES]; enum dml2_core_internal_output_type OutputType[DML2_MAX_PLANES]; enum dml2_core_internal_output_type_rate OutputRate[DML2_MAX_PLANES]; + bool TotalAvailablePipesSupportNoDSC; + bool TotalAvailablePipesSupportDSC; + unsigned int NumberOfDPPNoDSC; + unsigned int NumberOfDPPDSC; + enum dml2_odm_mode ODMModeNoDSC; + enum dml2_odm_mode ODMModeDSC; + double RequiredDISPCLKPerSurfaceNoDSC; + double RequiredDISPCLKPerSurfaceDSC; + unsigned int EstimatedNumberOfDSCSlices[DML2_MAX_PLANES]; // Bandwidth Related Info double BandwidthAvailableForImmediateFlip; @@ -484,8 +540,14 @@ struct dml2_core_internal_mode_support { double WriteBandwidth[DML2_MAX_PLANES][DML2_MAX_WRITEBACK]; double RequiredPrefetchPixelDataBWLuma[DML2_MAX_PLANES]; double RequiredPrefetchPixelDataBWChroma[DML2_MAX_PLANES]; - /* oto bw should also be considered when calculating peak urgent bw to avoid situations oto/equ mismatches between ms and mp */ - double RequiredPrefetchBWOTO[DML2_MAX_PLANES]; + /* Max bandwidth calculated from prefetch schedule should be considered in addition to the pixel data bw to avoid ms/mp mismatches. + * 1. oto bw should also be considered when calculating peak urgent bw to avoid situations oto/equ mismatches between ms and mp + * + * 2. equ bandwidth needs to be considered for calculating peak urgent bw when equ schedule is used in mode support. + * Some slight difference in variables may cause the pixel data bandwidth to be higher + * even though overall equ prefetch bandwidths can be lower going from ms to mp + */ + double RequiredPrefetchBWMax[DML2_MAX_PLANES]; double cursor_bw[DML2_MAX_PLANES]; double prefetch_cursor_bw[DML2_MAX_PLANES]; double prefetch_vmrow_bw[DML2_MAX_PLANES]; @@ -538,7 +600,44 @@ struct dml2_core_internal_mode_support { bool mall_comb_mcache_c[DML2_MAX_PLANES]; bool lc_comb_mcache[DML2_MAX_PLANES]; + unsigned int vmpg_width_y[DML2_MAX_PLANES]; + unsigned int vmpg_height_y[DML2_MAX_PLANES]; + unsigned int vmpg_width_c[DML2_MAX_PLANES]; + unsigned int vmpg_height_c[DML2_MAX_PLANES]; + + unsigned int meta_row_height_luma[DML2_MAX_PLANES]; + unsigned int meta_row_height_chroma[DML2_MAX_PLANES]; + unsigned int meta_row_bytes_per_row_ub_l[DML2_MAX_PLANES]; + unsigned int meta_row_bytes_per_row_ub_c[DML2_MAX_PLANES]; + unsigned int dpte_row_bytes_per_row_l[DML2_MAX_PLANES]; + unsigned int dpte_row_bytes_per_row_c[DML2_MAX_PLANES]; + + unsigned int pstate_bytes_required_l[DML2_MAX_PLANES]; + unsigned int pstate_bytes_required_c[DML2_MAX_PLANES]; + unsigned int cursor_bytes_per_chunk[DML2_MAX_PLANES]; + unsigned int cursor_bytes_per_line[DML2_MAX_PLANES]; + + unsigned int MaximumVStartup[DML2_MAX_PLANES]; + + double HostVMInefficiencyFactor; + double HostVMInefficiencyFactorPrefetch; + + unsigned int tdlut_pte_bytes_per_frame[DML2_MAX_PLANES]; + unsigned int tdlut_bytes_per_frame[DML2_MAX_PLANES]; + unsigned int tdlut_groups_per_2row_ub[DML2_MAX_PLANES]; + double tdlut_opt_time[DML2_MAX_PLANES]; + double tdlut_drain_time[DML2_MAX_PLANES]; + unsigned int tdlut_bytes_per_group[DML2_MAX_PLANES]; + + double Tvm_trips_flip[DML2_MAX_PLANES]; + double Tr0_trips_flip[DML2_MAX_PLANES]; + double Tvm_trips_flip_rounded[DML2_MAX_PLANES]; + double Tr0_trips_flip_rounded[DML2_MAX_PLANES]; + unsigned int DSTYAfterScaler[DML2_MAX_PLANES]; + unsigned int DSTXAfterScaler[DML2_MAX_PLANES]; + + enum dml2_pstate_method pstate_switch_modes[DML2_MAX_PLANES]; }; /// @brief A mega structure that houses various info for model programming step. @@ -548,6 +647,7 @@ struct dml2_core_internal_mode_program { double FabricClock; /// <brief Basically just the clock freq at the min (or given) state //double DCFCLK; /// <brief Basically just the clock freq at the min (or given) state and max combine setting double dram_bw_mbps; + double min_available_urgent_bandwidth_MBps; /// <brief Minimum guaranteed available urgent return bandwidth in MBps double uclk_freq_mhz; unsigned int NoOfDPP[DML2_MAX_PLANES]; enum dml2_odm_mode ODMMode[DML2_MAX_PLANES]; @@ -599,6 +699,8 @@ struct dml2_core_internal_mode_program { unsigned int MacroTileHeightC[DML2_MAX_PLANES]; unsigned int MacroTileWidthY[DML2_MAX_PLANES]; unsigned int MacroTileWidthC[DML2_MAX_PLANES]; + double MaximumSwathWidthLuma[DML2_MAX_PLANES]; + double MaximumSwathWidthChroma[DML2_MAX_PLANES]; bool surf_linear128_l[DML2_MAX_PLANES]; bool surf_linear128_c[DML2_MAX_PLANES]; @@ -631,6 +733,14 @@ struct dml2_core_internal_mode_program { double UrgentBurstFactorChroma[DML2_MAX_PLANES]; double UrgentBurstFactorChromaPre[DML2_MAX_PLANES]; + double MaximumSwathWidthInLineBufferLuma; + double MaximumSwathWidthInLineBufferChroma; + + unsigned int vmpg_width_y[DML2_MAX_PLANES]; + unsigned int vmpg_height_y[DML2_MAX_PLANES]; + unsigned int vmpg_width_c[DML2_MAX_PLANES]; + unsigned int vmpg_height_c[DML2_MAX_PLANES]; + double meta_row_bw[DML2_MAX_PLANES]; unsigned int meta_row_bytes[DML2_MAX_PLANES]; unsigned int meta_req_width[DML2_MAX_PLANES]; @@ -652,7 +762,9 @@ struct dml2_core_internal_mode_program { unsigned int PTERequestSizeC[DML2_MAX_PLANES]; double TWait[DML2_MAX_PLANES]; + double Tdmdl_vm_raw[DML2_MAX_PLANES]; double Tdmdl_vm[DML2_MAX_PLANES]; + double Tdmdl_raw[DML2_MAX_PLANES]; double Tdmdl[DML2_MAX_PLANES]; double TSetup[DML2_MAX_PLANES]; unsigned int dpde0_bytes_per_frame_ub_l[DML2_MAX_PLANES]; @@ -684,6 +796,38 @@ struct dml2_core_internal_mode_program { double TCalc; unsigned int TotImmediateFlipBytes; + unsigned int MaxTotalDETInKByte; + unsigned int NomDETInKByte; + unsigned int MinCompressedBufferSizeInKByte; + double PixelClockBackEnd[DML2_MAX_PLANES]; + double OutputBpp[DML2_MAX_PLANES]; + bool dsc_enable[DML2_MAX_PLANES]; + unsigned int num_dsc_slices[DML2_MAX_PLANES]; + unsigned int meta_row_bytes_per_row_ub_l[DML2_MAX_PLANES]; + unsigned int meta_row_bytes_per_row_ub_c[DML2_MAX_PLANES]; + unsigned int dpte_row_bytes_per_row_l[DML2_MAX_PLANES]; + unsigned int dpte_row_bytes_per_row_c[DML2_MAX_PLANES]; + unsigned int cursor_bytes_per_chunk[DML2_MAX_PLANES]; + unsigned int cursor_bytes_per_line[DML2_MAX_PLANES]; + unsigned int MaxVStartupLines[DML2_MAX_PLANES]; /// <brief more like vblank for the plane's OTG + double HostVMInefficiencyFactor; + double HostVMInefficiencyFactorPrefetch; + unsigned int tdlut_pte_bytes_per_frame[DML2_MAX_PLANES]; + unsigned int tdlut_bytes_per_frame[DML2_MAX_PLANES]; + unsigned int tdlut_groups_per_2row_ub[DML2_MAX_PLANES]; + double tdlut_opt_time[DML2_MAX_PLANES]; + double tdlut_drain_time[DML2_MAX_PLANES]; + unsigned int tdlut_bytes_per_group[DML2_MAX_PLANES]; + double Tvm_trips_flip[DML2_MAX_PLANES]; + double Tr0_trips_flip[DML2_MAX_PLANES]; + double Tvm_trips_flip_rounded[DML2_MAX_PLANES]; + double Tr0_trips_flip_rounded[DML2_MAX_PLANES]; + bool immediate_flip_required; // any pipes need immediate flip + double SOCCLK; /// <brief Basically just the clock freq at the min (or given) state + double TotalWRBandwidth; + double max_urgent_latency_us; + double df_response_time_us; + // ------------------- // Output // ------------------- @@ -694,9 +838,12 @@ struct dml2_core_internal_mode_program { // Support bool UrgVactiveBandwidthSupport; + bool PrefetchScheduleSupported; + bool UrgentBandwidthSupport; bool PrefetchModeSupported; // <brief Is the prefetch mode (bandwidth and latency) supported bool ImmediateFlipSupported; bool ImmediateFlipSupportedForPipe[DML2_MAX_PLANES]; + bool dcfclk_support; // Clock double Dcfclk; @@ -788,7 +935,7 @@ struct dml2_core_internal_mode_program { // RQ registers bool PTE_BUFFER_MODE[DML2_MAX_PLANES]; unsigned int BIGK_FRAGMENT_SIZE[DML2_MAX_PLANES]; - + double VActiveLatencyHidingUs[DML2_MAX_PLANES]; unsigned int SubViewportLinesNeededInMALL[DML2_MAX_PLANES]; bool is_using_mall_for_ss[DML2_MAX_PLANES]; @@ -1001,10 +1148,10 @@ struct dml2_core_calcs_mode_programming_locals { double dummy_bw[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]; double surface_dummy_bw[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max][DML2_MAX_PLANES]; double surface_dummy_bw0[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max][DML2_MAX_PLANES]; - unsigned int dummy_integer_array[2][DML2_MAX_PLANES]; + unsigned int dummy_integer_array[4][DML2_MAX_PLANES]; enum dml2_output_encoder_class dummy_output_encoder_array[DML2_MAX_PLANES]; double dummy_single_array[2][DML2_MAX_PLANES]; - unsigned int dummy_long_array[4][DML2_MAX_PLANES]; + unsigned int dummy_long_array[8][DML2_MAX_PLANES]; bool dummy_boolean_array[2][DML2_MAX_PLANES]; bool dummy_boolean[2]; double dummy_single[2]; @@ -1028,7 +1175,6 @@ struct dml2_core_calcs_mode_programming_locals { double dlg_vblank_start; double LSetup; double blank_lines_remaining; - double TotalWRBandwidth; double WRBandwidth; struct dml2_core_internal_DmlPipe myPipe; double PixelClockBackEndFactor; @@ -1153,6 +1299,7 @@ struct dml2_core_calcs_CalculateVMRowAndSwath_params { unsigned int HostVMMinPageSize; unsigned int DCCMetaBufferSizeBytes; bool mrq_present; + enum dml2_pstate_method pstate_switch_modes[DML2_MAX_PLANES]; // Output bool *PTEBufferSizeNotExceeded; @@ -1389,7 +1536,7 @@ struct dml2_core_shared_get_urgent_bandwidth_required_locals { double vm_row_bw; double flip_and_active_bw; double flip_and_prefetch_bw; - double flip_and_prefetch_bw_oto; + double flip_and_prefetch_bw_max; double active_and_excess_bw; }; @@ -1418,6 +1565,7 @@ struct dml2_core_shared_CalculateFlipSchedule_locals { struct dml2_core_shared_rq_dlg_get_dlg_reg_locals { unsigned int plane_idx; + unsigned int stream_idx; enum dml2_source_format_class source_format; const struct dml2_timing_cfg *timing; bool dual_plane; @@ -1625,6 +1773,9 @@ struct dml2_core_calcs_CalculateSwathAndDETConfiguration_params { double *BytePerPixDETC; unsigned int *DPPPerSurface; bool mrq_present; + unsigned int dummy[2][DML2_MAX_PLANES]; + unsigned int swath_width_luma_ub_single_dpp[DML2_MAX_PLANES]; + unsigned int swath_width_chroma_ub_single_dpp[DML2_MAX_PLANES]; // output unsigned int *req_per_swath_ub_l; @@ -1642,6 +1793,8 @@ struct dml2_core_calcs_CalculateSwathAndDETConfiguration_params { unsigned int *DETBufferSizeC; unsigned int *full_swath_bytes_l; unsigned int *full_swath_bytes_c; + unsigned int *full_swath_bytes_single_dpp_l; + unsigned int *full_swath_bytes_single_dpp_c; bool *UnboundedRequestEnabled; unsigned int *compbuf_reserved_space_64b; unsigned int *CompressedBufferSizeInkByte; @@ -1801,7 +1954,7 @@ struct dml2_core_calcs_CalculatePrefetchSchedule_params { double *VRatioPrefetchC; double *RequiredPrefetchPixelDataBWLuma; double *RequiredPrefetchPixelDataBWChroma; - double *RequiredPrefetchBWOTO; + double *RequiredPrefetchBWMax; bool *NotEnoughTimeForDynamicMetadata; double *Tno_bw; double *Tno_bw_flip; @@ -2038,7 +2191,7 @@ struct dml2_core_calcs_calculate_peak_bandwidth_required_params { double *surface_read_bandwidth_c; double *prefetch_bandwidth_l; double *prefetch_bandwidth_c; - double *prefetch_bandwidth_oto; + double *prefetch_bandwidth_max; double *excess_vactive_fill_bw_l; double *excess_vactive_fill_bw_c; double *cursor_bw; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.c index 7a220c0141c2..5f301befed16 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.c @@ -464,7 +464,7 @@ bool dml2_core_utils_get_segment_horizontal_contiguous(enum dml2_swizzle_mode sw bool dml2_core_utils_is_linear(enum dml2_swizzle_mode sw_mode) { - return (sw_mode == dml2_sw_linear || sw_mode == dml2_sw_linear_256b || sw_mode == dml2_linear_64elements); + return sw_mode == dml2_sw_linear; }; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c index f486b090bbfc..22969a533a7b 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c @@ -389,9 +389,6 @@ static bool map_min_clocks_to_dpm(const struct dml2_core_mode_support_result *mo if (result) result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.dispclk_khz, &state_table->dispclk); - if (result) - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.deepsleep_dcfclk_khz, &state_table->dcfclk); - for (i = 0; i < DML2_MAX_DCN_PIPES; i++) { if (result) result = round_up_to_next_dpm(&display_cfg->plane_programming[i].min_clocks.dcn4x.dppclk_khz, &state_table->dppclk); diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h index b226225103c3..611c80f4f1bf 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h @@ -10,15 +10,74 @@ #define DML_LOG_LEVEL_DEFAULT DML_LOG_LEVEL_WARN #define DML_LOG_INTERNAL(fmt, ...) dm_output_to_console(fmt, ## __VA_ARGS__) -/* ASSERT with message output */ -#define DML_ASSERT_MSG(condition, fmt, ...) \ - do { \ - if (!(condition)) { \ - DML_LOG_ERROR("DML ASSERT hit in %s line %d\n", __func__, __LINE__); \ - DML_LOG_ERROR(fmt, ## __VA_ARGS__); \ - DML_ASSERT(condition); \ - } \ - } while (0) +/* private helper macros */ +#define _BOOL_FORMAT(field) "%s", field ? "true" : "false" +#define _UINT_FORMAT(field) "%u", field +#define _INT_FORMAT(field) "%d", field +#define _DOUBLE_FORMAT(field) "%lf", field +#define _ELEMENT_FUNC "function" +#define _ELEMENT_COMP_IF "component_interface" +#define _ELEMENT_TOP_IF "top_interface" +#define _LOG_ENTRY(element) do { \ + DML_LOG_INTERNAL("<"element" name=\""); \ + DML_LOG_INTERNAL(__func__); \ + DML_LOG_INTERNAL("\">\n"); \ +} while (0) +#define _LOG_EXIT(element) DML_LOG_INTERNAL("</"element">\n") +#define _LOG_SCALAR(field, format) do { \ + DML_LOG_INTERNAL(#field" = "format(field)); \ + DML_LOG_INTERNAL("\n"); \ +} while (0) +#define _LOG_ARRAY(field, size, format) do { \ + DML_LOG_INTERNAL(#field " = ["); \ + for (int _i = 0; _i < (int) size; _i++) { \ + DML_LOG_INTERNAL(format(field[_i])); \ + if (_i + 1 == (int) size) \ + DML_LOG_INTERNAL("]\n"); \ + else \ + DML_LOG_INTERNAL(", "); \ +}} while (0) +#define _LOG_2D_ARRAY(field, size0, size1, format) do { \ + DML_LOG_INTERNAL(#field" = ["); \ + for (int _i = 0; _i < (int) size0; _i++) { \ + DML_LOG_INTERNAL("\n\t["); \ + for (int _j = 0; _j < (int) size1; _j++) { \ + DML_LOG_INTERNAL(format(field[_i][_j])); \ + if (_j + 1 == (int) size1) \ + DML_LOG_INTERNAL("]"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ + if (_i + 1 == (int) size0) \ + DML_LOG_INTERNAL("]\n"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ +} while (0) +#define _LOG_3D_ARRAY(field, size0, size1, size2, format) do { \ + DML_LOG_INTERNAL(#field" = ["); \ + for (int _i = 0; _i < (int) size0; _i++) { \ + DML_LOG_INTERNAL("\n\t["); \ + for (int _j = 0; _j < (int) size1; _j++) { \ + DML_LOG_INTERNAL("["); \ + for (int _k = 0; _k < (int) size2; _k++) { \ + DML_LOG_INTERNAL(format(field[_i][_j][_k])); \ + if (_k + 1 == (int) size2) \ + DML_LOG_INTERNAL("]"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ + if (_j + 1 == (int) size1) \ + DML_LOG_INTERNAL("]"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ + if (_i + 1 == (int) size0) \ + DML_LOG_INTERNAL("]\n"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ +} while (0) /* fatal errors for unrecoverable DML states until a full reset */ #define DML_LOG_LEVEL_FATAL 0 @@ -28,7 +87,7 @@ #define DML_LOG_LEVEL_WARN 2 /* high level tracing of DML interfaces */ #define DML_LOG_LEVEL_INFO 3 -/* detailed tracing of DML internal components */ +/* tracing of DML internal executions */ #define DML_LOG_LEVEL_DEBUG 4 /* detailed tracing of DML calculation procedure */ #define DML_LOG_LEVEL_VERBOSE 5 @@ -37,30 +96,94 @@ #define DML_LOG_LEVEL DML_LOG_LEVEL_DEFAULT #endif /* #ifndef DML_LOG_LEVEL */ +/* public macros for DML_LOG_LEVEL_FATAL and up */ #define DML_LOG_FATAL(fmt, ...) DML_LOG_INTERNAL("[DML FATAL] " fmt, ## __VA_ARGS__) + +/* public macros for DML_LOG_LEVEL_ERROR and up */ #if DML_LOG_LEVEL >= DML_LOG_LEVEL_ERROR #define DML_LOG_ERROR(fmt, ...) DML_LOG_INTERNAL("[DML ERROR] "fmt, ## __VA_ARGS__) +#define DML_ASSERT_MSG(condition, fmt, ...) \ + do { \ + if (!(condition)) { \ + DML_LOG_ERROR("ASSERT hit in %s line %d\n", __func__, __LINE__); \ + DML_LOG_ERROR(fmt, ## __VA_ARGS__); \ + DML_ASSERT(condition); \ + } \ + } while (0) #else #define DML_LOG_ERROR(fmt, ...) ((void)0) +#define DML_ASSERT_MSG(condition, fmt, ...) ((void)0) #endif + +/* public macros for DML_LOG_LEVEL_WARN and up */ #if DML_LOG_LEVEL >= DML_LOG_LEVEL_WARN #define DML_LOG_WARN(fmt, ...) DML_LOG_INTERNAL("[DML WARN] "fmt, ## __VA_ARGS__) #else #define DML_LOG_WARN(fmt, ...) ((void)0) #endif + +/* public macros for DML_LOG_LEVEL_INFO and up */ #if DML_LOG_LEVEL >= DML_LOG_LEVEL_INFO #define DML_LOG_INFO(fmt, ...) DML_LOG_INTERNAL("[DML INFO] "fmt, ## __VA_ARGS__) +#define DML_LOG_TOP_IF_ENTER() _LOG_ENTRY(_ELEMENT_TOP_IF) +#define DML_LOG_TOP_IF_EXIT() _LOG_EXIT(_ELEMENT_TOP_IF) #else #define DML_LOG_INFO(fmt, ...) ((void)0) +#define DML_LOG_TOP_IF_ENTER() ((void)0) +#define DML_LOG_TOP_IF_EXIT() ((void)0) #endif + +/* public macros for DML_LOG_LEVEL_DEBUG and up */ #if DML_LOG_LEVEL >= DML_LOG_LEVEL_DEBUG -#define DML_LOG_DEBUG(fmt, ...) DML_LOG_INTERNAL("[DML DEBUG] "fmt, ## __VA_ARGS__) +#define DML_LOG_DEBUG(fmt, ...) DML_LOG_INTERNAL(fmt, ## __VA_ARGS__) +#define DML_LOG_COMP_IF_ENTER() _LOG_ENTRY(_ELEMENT_COMP_IF) +#define DML_LOG_COMP_IF_EXIT() _LOG_EXIT(_ELEMENT_COMP_IF) +#define DML_LOG_FUNC_ENTER() _LOG_ENTRY(_ELEMENT_FUNC) +#define DML_LOG_FUNC_EXIT() _LOG_EXIT(_ELEMENT_FUNC) +#define DML_LOG_DEBUG_BOOL(field) _LOG_SCALAR(field, _BOOL_FORMAT) +#define DML_LOG_DEBUG_UINT(field) _LOG_SCALAR(field, _UINT_FORMAT) +#define DML_LOG_DEBUG_INT(field) _LOG_SCALAR(field, _INT_FORMAT) +#define DML_LOG_DEBUG_DOUBLE(field) _LOG_SCALAR(field, _DOUBLE_FORMAT) +#define DML_LOG_DEBUG_ARRAY_BOOL(field, size) _LOG_ARRAY(field, size, _BOOL_FORMAT) +#define DML_LOG_DEBUG_ARRAY_UINT(field, size) _LOG_ARRAY(field, size, _UINT_FORMAT) +#define DML_LOG_DEBUG_ARRAY_INT(field, size) _LOG_ARRAY(field, size, _INT_FORMAT) +#define DML_LOG_DEBUG_ARRAY_DOUBLE(field, size) _LOG_ARRAY(field, size, _DOUBLE_FORMAT) +#define DML_LOG_DEBUG_2D_ARRAY_BOOL(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _BOOL_FORMAT) +#define DML_LOG_DEBUG_2D_ARRAY_UINT(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _UINT_FORMAT) +#define DML_LOG_DEBUG_2D_ARRAY_INT(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _INT_FORMAT) +#define DML_LOG_DEBUG_2D_ARRAY_DOUBLE(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _DOUBLE_FORMAT) +#define DML_LOG_DEBUG_3D_ARRAY_BOOL(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _BOOL_FORMAT) +#define DML_LOG_DEBUG_3D_ARRAY_UINT(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _UINT_FORMAT) +#define DML_LOG_DEBUG_3D_ARRAY_INT(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _INT_FORMAT) +#define DML_LOG_DEBUG_3D_ARRAY_DOUBLE(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _DOUBLE_FORMAT) #else #define DML_LOG_DEBUG(fmt, ...) ((void)0) +#define DML_LOG_COMP_IF_ENTER() ((void)0) +#define DML_LOG_COMP_IF_EXIT() ((void)0) +#define DML_LOG_FUNC_ENTER() ((void)0) +#define DML_LOG_FUNC_EXIT() ((void)0) +#define DML_LOG_DEBUG_BOOL(field) ((void)0) +#define DML_LOG_DEBUG_UINT(field) ((void)0) +#define DML_LOG_DEBUG_INT(field) ((void)0) +#define DML_LOG_DEBUG_DOUBLE(field) ((void)0) +#define DML_LOG_DEBUG_ARRAY_BOOL(field, size) ((void)0) +#define DML_LOG_DEBUG_ARRAY_UINT(field, size) ((void)0) +#define DML_LOG_DEBUG_ARRAY_INT(field, size) ((void)0) +#define DML_LOG_DEBUG_ARRAY_DOUBLE(field, size) ((void)0) +#define DML_LOG_DEBUG_2D_ARRAY_BOOL(field, size0, size1) ((void)0) +#define DML_LOG_DEBUG_2D_ARRAY_UINT(field, size0, size1) ((void)0) +#define DML_LOG_DEBUG_2D_ARRAY_INT(field, size0, size1) ((void)0) +#define DML_LOG_DEBUG_2D_ARRAY_DOUBLE(field, size0, size1) ((void)0) +#define DML_LOG_DEBUG_3D_ARRAY_BOOL(field, size0, size1, size2) ((void)0) +#define DML_LOG_DEBUG_3D_ARRAY_UINT(field, size0, size1, size2) ((void)0) +#define DML_LOG_DEBUG_3D_ARRAY_INT(field, size0, size1, size2) ((void)0) +#define DML_LOG_DEBUG_3D_ARRAY_DOUBLE(field, size0, size1, size2) ((void)0) #endif + +/* public macros for DML_LOG_LEVEL_VERBOSE */ #if DML_LOG_LEVEL >= DML_LOG_LEVEL_VERBOSE -#define DML_LOG_VERBOSE(fmt, ...) DML_LOG_INTERNAL("[DML VERBOSE] "fmt, ## __VA_ARGS__) +#define DML_LOG_VERBOSE(fmt, ...) DML_LOG_INTERNAL(fmt, ## __VA_ARGS__) #else #define DML_LOG_VERBOSE(fmt, ...) ((void)0) -#endif +#endif /* #if DML_LOG_LEVEL >= DML_LOG_LEVEL_VERBOSE */ #endif /* __DML2_DEBUG_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h index 00688b9f1df4..d52aa82283b3 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h @@ -202,6 +202,8 @@ struct dml2_core_mode_support_result { } active; unsigned int dispclk_khz; + unsigned int dpprefclk_khz; + unsigned int dtbrefclk_khz; unsigned int dcfclk_deepsleep_khz; unsigned int socclk_khz; @@ -446,13 +448,17 @@ struct dml2_core_internal_state_intermediates { }; struct dml2_core_mode_support_locals { - struct dml2_core_calcs_mode_support_ex mode_support_ex_params; + union { + struct dml2_core_calcs_mode_support_ex mode_support_ex_params; + }; struct dml2_display_cfg svp_expanded_display_cfg; struct dml2_calculate_mcache_allocation_in_out calc_mcache_allocation_params; }; struct dml2_core_mode_programming_locals { - struct dml2_core_calcs_mode_programming_ex mode_programming_ex_params; + union { + struct dml2_core_calcs_mode_programming_ex mode_programming_ex_params; + }; struct dml2_display_cfg svp_expanded_display_cfg; }; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c index 6b3b8803e0ae..a56e75cdf712 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c @@ -868,7 +868,7 @@ bool dml2_svp_remove_all_phantom_pipes(struct dml2_context *ctx, struct dc_state /* Conditions for setting up phantom pipes for SubVP: * 1. Not force disable SubVP - * 2. Full update (i.e. !fast_validate) + * 2. Full update (i.e. DC_VALIDATE_MODE_AND_PROGRAMMING) * 3. Enough pipes are available to support SubVP (TODO: Which pipes will use VACTIVE / VBLANK / SUBVP?) * 4. Display configuration passes validation * 5. (Config doesn't support MCLK in VACTIVE/VBLANK || dc->debug.force_subvp_mclk_switch) diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c index 5de775fd8fce..3b866e876bf4 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c @@ -953,6 +953,7 @@ static void populate_dml_surface_cfg_from_plane_state(enum dml_project_id dml2_p out->SourcePixelFormat[location] = dml_420_10; break; case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616: + case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616: case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F: case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F: out->SourcePixelFormat[location] = dml_444_64; @@ -1188,22 +1189,6 @@ static unsigned int map_plane_to_dml_display_cfg(const struct dml2_context *dml2 return location; } -static void apply_legacy_svp_drr_settings(struct dml2_context *dml2, const struct dc_state *state, struct dml_display_cfg_st *dml_dispcfg) -{ - int i; - - if (state->bw_ctx.bw.dcn.clk.fw_based_mclk_switching) { - ASSERT(state->stream_count == 1); - dml_dispcfg->timing.DRRDisplay[0] = true; - } else if (state->bw_ctx.bw.dcn.legacy_svp_drr_stream_index_valid) { - - for (i = 0; i < dml_dispcfg->num_timings; i++) { - if (dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[i] == state->streams[state->bw_ctx.bw.dcn.legacy_svp_drr_stream_index]->stream_id) - dml_dispcfg->timing.DRRDisplay[i] = true; - } - } -} - static void dml2_populate_pipe_to_plane_index_mapping(struct dml2_context *dml2, struct dc_state *state) { unsigned int i; @@ -1436,9 +1421,6 @@ void map_dc_state_into_dml_display_cfg(struct dml2_context *dml2, struct dc_stat } } } - - if (!dml2->config.use_native_pstate_optimization) - apply_legacy_svp_drr_settings(dml2, context, dml_dispcfg); } void dml2_update_pipe_ctx_dchub_regs(struct _vcs_dpi_dml_display_rq_regs_st *rq_regs, diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c index 525b7d04bf84..0318260370ed 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c @@ -24,8 +24,6 @@ * */ -#include <linux/vmalloc.h> - #include "display_mode_core.h" #include "dml2_internal_types.h" #include "dml2_utils.h" @@ -95,12 +93,17 @@ static void map_hw_resources(struct dml2_context *dml2, static unsigned int pack_and_call_dml_mode_support_ex(struct dml2_context *dml2, const struct dml_display_cfg_st *display_cfg, - struct dml_mode_support_info_st *evaluation_info) + struct dml_mode_support_info_st *evaluation_info, + enum dc_validate_mode validate_mode) { struct dml2_wrapper_scratch *s = &dml2->v20.scratch; s->mode_support_params.mode_lib = &dml2->v20.dml_core_ctx; s->mode_support_params.in_display_cfg = display_cfg; + if (validate_mode == DC_VALIDATE_MODE_ONLY) + s->mode_support_params.in_start_state_idx = dml2->v20.dml_core_ctx.states.num_states - 1; + else + s->mode_support_params.in_start_state_idx = 0; s->mode_support_params.out_evaluation_info = evaluation_info; memset(evaluation_info, 0, sizeof(struct dml_mode_support_info_st)); @@ -112,10 +115,8 @@ static unsigned int pack_and_call_dml_mode_support_ex(struct dml2_context *dml2, static bool optimize_configuration(struct dml2_context *dml2, struct dml2_wrapper_optimize_configuration_params *p) { int unused_dpps = p->ip_params->max_num_dpp; - int i, j; - int odms_needed, refresh_rate_hz, dpps_needed, subvp_height, pstate_width_fw_delay_lines, surface_count; - int subvp_timing_to_add, new_timing_index, subvp_surface_to_add, new_surface_index; - float frame_time_sec, max_frame_time_sec; + int i; + int odms_needed; int largest_blend_and_timing = 0; bool optimization_done = false; @@ -130,79 +131,6 @@ static bool optimize_configuration(struct dml2_context *dml2, struct dml2_wrappe if (p->new_display_config != p->cur_display_config) *p->new_display_config = *p->cur_display_config; - // Optimize P-State Support - if (dml2->config.use_native_pstate_optimization) { - if (p->cur_mode_support_info->DRAMClockChangeSupport[0] == dml_dram_clock_change_unsupported) { - // Find a display with < 120Hz refresh rate with maximal refresh rate that's not already subvp - subvp_timing_to_add = -1; - subvp_surface_to_add = -1; - max_frame_time_sec = 0; - surface_count = 0; - for (i = 0; i < (int) p->cur_display_config->num_timings; i++) { - refresh_rate_hz = (int)div_u64((unsigned long long) p->cur_display_config->timing.PixelClock[i] * 1000 * 1000, - (p->cur_display_config->timing.HTotal[i] * p->cur_display_config->timing.VTotal[i])); - if (refresh_rate_hz < 120) { - // Check its upstream surfaces to see if this one could be converted to subvp. - dpps_needed = 0; - for (j = 0; j < (int) p->cur_display_config->num_surfaces; j++) { - if (p->cur_display_config->plane.BlendingAndTiming[j] == i && - p->cur_display_config->plane.UseMALLForPStateChange[j] == dml_use_mall_pstate_change_disable) { - dpps_needed += p->cur_mode_support_info->DPPPerSurface[j]; - subvp_surface_to_add = j; - surface_count++; - } - } - - if (surface_count == 1 && dpps_needed > 0 && dpps_needed <= unused_dpps) { - frame_time_sec = (float)1 / refresh_rate_hz; - if (frame_time_sec > max_frame_time_sec) { - max_frame_time_sec = frame_time_sec; - subvp_timing_to_add = i; - } - } - } - } - if (subvp_timing_to_add >= 0) { - new_timing_index = p->new_display_config->num_timings++; - new_surface_index = p->new_display_config->num_surfaces++; - // Add a phantom pipe reflecting the main pipe's timing - dml2_util_copy_dml_timing(&p->new_display_config->timing, new_timing_index, subvp_timing_to_add); - - pstate_width_fw_delay_lines = (int)(((double)(p->config->svp_pstate.subvp_fw_processing_delay_us + - p->config->svp_pstate.subvp_pstate_allow_width_us) / 1000000) * - (p->new_display_config->timing.PixelClock[subvp_timing_to_add] * 1000 * 1000) / - (double)p->new_display_config->timing.HTotal[subvp_timing_to_add]); - - subvp_height = p->cur_mode_support_info->SubViewportLinesNeededInMALL[subvp_timing_to_add] + pstate_width_fw_delay_lines; - - p->new_display_config->timing.VActive[new_timing_index] = subvp_height; - p->new_display_config->timing.VTotal[new_timing_index] = subvp_height + - p->new_display_config->timing.VTotal[subvp_timing_to_add] - p->new_display_config->timing.VActive[subvp_timing_to_add]; - - p->new_display_config->output.OutputDisabled[new_timing_index] = true; - - p->new_display_config->plane.UseMALLForPStateChange[subvp_surface_to_add] = dml_use_mall_pstate_change_sub_viewport; - - dml2_util_copy_dml_plane(&p->new_display_config->plane, new_surface_index, subvp_surface_to_add); - dml2_util_copy_dml_surface(&p->new_display_config->surface, new_surface_index, subvp_surface_to_add); - - p->new_display_config->plane.ViewportHeight[new_surface_index] = subvp_height; - p->new_display_config->plane.ViewportHeightChroma[new_surface_index] = subvp_height; - p->new_display_config->plane.ViewportStationary[new_surface_index] = false; - - p->new_display_config->plane.UseMALLForStaticScreen[new_surface_index] = dml_use_mall_static_screen_disable; - p->new_display_config->plane.UseMALLForPStateChange[new_surface_index] = dml_use_mall_pstate_change_phantom_pipe; - - p->new_display_config->plane.NumberOfCursors[new_surface_index] = 0; - - p->new_policy->ImmediateFlipRequirement[new_surface_index] = dml_immediate_flip_not_required; - - p->new_display_config->plane.BlendingAndTiming[new_surface_index] = new_timing_index; - - optimization_done = true; - } - } - } // Optimize Clocks if (!optimization_done) { @@ -226,7 +154,8 @@ static bool optimize_configuration(struct dml2_context *dml2, struct dml2_wrappe return optimization_done; } -static int calculate_lowest_supported_state_for_temp_read(struct dml2_context *dml2, struct dc_state *display_state) +static int calculate_lowest_supported_state_for_temp_read(struct dml2_context *dml2, struct dc_state *display_state, + enum dc_validate_mode validate_mode) { struct dml2_calculate_lowest_supported_state_for_temp_read_scratch *s = &dml2->v20.scratch.dml2_calculate_lowest_supported_state_for_temp_read_scratch; struct dml2_wrapper_scratch *s_global = &dml2->v20.scratch; @@ -268,7 +197,8 @@ static int calculate_lowest_supported_state_for_temp_read(struct dml2_context *d dml2->v20.dml_core_ctx.states.state_array[j].dram_clock_change_latency_us = s_global->dummy_pstate_table[i].dummy_pstate_latency_us; } - dml_result = pack_and_call_dml_mode_support_ex(dml2, &s->cur_display_config, &s->evaluation_info); + dml_result = pack_and_call_dml_mode_support_ex(dml2, &s->cur_display_config, &s->evaluation_info, + validate_mode); if (dml_result && s->evaluation_info.DRAMClockChangeSupport[0] == dml_dram_clock_change_vactive) { map_hw_resources(dml2, &s->cur_display_config, &s->evaluation_info); @@ -333,7 +263,8 @@ static bool does_configuration_meet_sw_policies(struct dml2_context *ctx, const } static bool dml_mode_support_wrapper(struct dml2_context *dml2, - struct dc_state *display_state) + struct dc_state *display_state, + enum dc_validate_mode validate_mode) { struct dml2_wrapper_scratch *s = &dml2->v20.scratch; unsigned int result = 0, i; @@ -369,7 +300,8 @@ static bool dml_mode_support_wrapper(struct dml2_context *dml2, result = pack_and_call_dml_mode_support_ex(dml2, &s->cur_display_config, - &s->mode_support_info); + &s->mode_support_info, + validate_mode); if (result) result = does_configuration_meet_sw_policies(dml2, &s->cur_display_config, &s->mode_support_info); @@ -390,7 +322,8 @@ static bool dml_mode_support_wrapper(struct dml2_context *dml2, dml2->v20.dml_core_ctx.policy = s->new_policy; optimized_result = pack_and_call_dml_mode_support_ex(dml2, &s->new_display_config, - &s->mode_support_info); + &s->mode_support_info, + validate_mode); if (optimized_result) optimized_result = does_configuration_meet_sw_policies(dml2, &s->new_display_config, &s->mode_support_info); @@ -409,7 +342,8 @@ static bool dml_mode_support_wrapper(struct dml2_context *dml2, if (!optimized_result) { result = pack_and_call_dml_mode_support_ex(dml2, &s->cur_display_config, - &s->mode_support_info); + &s->mode_support_info, + validate_mode); } } @@ -419,118 +353,7 @@ static bool dml_mode_support_wrapper(struct dml2_context *dml2, return result; } -static int find_drr_eligible_stream(struct dc_state *display_state) -{ - int i; - - for (i = 0; i < display_state->stream_count; i++) { - if (dc_state_get_stream_subvp_type(display_state, display_state->streams[i]) == SUBVP_NONE - && display_state->streams[i]->ignore_msa_timing_param) { - // Use ignore_msa_timing_param flag to identify as DRR - return i; - } - } - - return -1; -} - -static bool optimize_pstate_with_svp_and_drr(struct dml2_context *dml2, struct dc_state *display_state) -{ - struct dml2_wrapper_scratch *s = &dml2->v20.scratch; - bool pstate_optimization_done = false; - bool pstate_optimization_success = false; - bool result = false; - int drr_display_index = 0, non_svp_streams = 0; - bool force_svp = dml2->config.svp_pstate.force_enable_subvp; - - display_state->bw_ctx.bw.dcn.clk.fw_based_mclk_switching = false; - display_state->bw_ctx.bw.dcn.legacy_svp_drr_stream_index_valid = false; - - result = dml_mode_support_wrapper(dml2, display_state); - - if (!result) { - pstate_optimization_done = true; - } else if (s->mode_support_info.DRAMClockChangeSupport[0] != dml_dram_clock_change_unsupported && !force_svp) { - pstate_optimization_success = true; - pstate_optimization_done = true; - } - - if (display_state->stream_count == 1 && dml2->config.callbacks.can_support_mclk_switch_using_fw_based_vblank_stretch(dml2->config.callbacks.dc, display_state)) { - display_state->bw_ctx.bw.dcn.clk.fw_based_mclk_switching = true; - - result = dml_mode_support_wrapper(dml2, display_state); - } else { - non_svp_streams = display_state->stream_count; - - while (!pstate_optimization_done) { - result = dml_mode_programming(&dml2->v20.dml_core_ctx, s->mode_support_params.out_lowest_state_idx, &s->cur_display_config, true); - - // Always try adding SVP first - if (result) - result = dml2_svp_add_phantom_pipe_to_dc_state(dml2, display_state, &s->mode_support_info); - else - pstate_optimization_done = true; - - - if (result) { - result = dml_mode_support_wrapper(dml2, display_state); - } else { - pstate_optimization_done = true; - } - - if (result) { - non_svp_streams--; - - if (s->mode_support_info.DRAMClockChangeSupport[0] != dml_dram_clock_change_unsupported) { - if (dml2_svp_validate_static_schedulability(dml2, display_state, s->mode_support_info.DRAMClockChangeSupport[0])) { - pstate_optimization_success = true; - pstate_optimization_done = true; - } else { - pstate_optimization_success = false; - pstate_optimization_done = false; - } - } else { - drr_display_index = find_drr_eligible_stream(display_state); - - // If there is only 1 remaining non SubVP pipe that is DRR, check static - // schedulability for SubVP + DRR. - if (non_svp_streams == 1 && drr_display_index >= 0) { - if (dml2_svp_drr_schedulable(dml2, display_state, &display_state->streams[drr_display_index]->timing)) { - display_state->bw_ctx.bw.dcn.legacy_svp_drr_stream_index_valid = true; - display_state->bw_ctx.bw.dcn.legacy_svp_drr_stream_index = drr_display_index; - result = dml_mode_support_wrapper(dml2, display_state); - } - - if (result && s->mode_support_info.DRAMClockChangeSupport[0] != dml_dram_clock_change_unsupported) { - pstate_optimization_success = true; - pstate_optimization_done = true; - } else { - pstate_optimization_success = false; - pstate_optimization_done = false; - } - } - - if (pstate_optimization_success) { - pstate_optimization_done = true; - } else { - pstate_optimization_done = false; - } - } - } - } - } - - if (!pstate_optimization_success) { - dml2_svp_remove_all_phantom_pipes(dml2, display_state); - display_state->bw_ctx.bw.dcn.clk.fw_based_mclk_switching = false; - display_state->bw_ctx.bw.dcn.legacy_svp_drr_stream_index_valid = false; - result = dml_mode_support_wrapper(dml2, display_state); - } - - return result; -} - -static bool call_dml_mode_support_and_programming(struct dc_state *context) +static bool call_dml_mode_support_and_programming(struct dc_state *context, enum dc_validate_mode validate_mode) { unsigned int result = 0; unsigned int min_state = 0; @@ -544,16 +367,13 @@ static bool call_dml_mode_support_and_programming(struct dc_state *context) struct dml2_wrapper_scratch *s = &dml2->v20.scratch; if (!context->streams[0]->sink->link->dc->caps.is_apu) { - min_state_for_g6_temp_read = calculate_lowest_supported_state_for_temp_read(dml2, context); + min_state_for_g6_temp_read = calculate_lowest_supported_state_for_temp_read(dml2, context, + validate_mode); ASSERT(min_state_for_g6_temp_read >= 0); } - if (!dml2->config.use_native_pstate_optimization) { - result = optimize_pstate_with_svp_and_drr(dml2, context); - } else { - result = dml_mode_support_wrapper(dml2, context); - } + result = dml_mode_support_wrapper(dml2, context, validate_mode); /* Upon trying to sett certain frequencies in FRL, min_state_for_g6_temp_read is reported as -1. This leads to an invalid value of min_state causing crashes later on. * Use the default logic for min_state only when min_state_for_g6_temp_read is a valid value. In other cases, use the value calculated by the DML directly. @@ -575,7 +395,8 @@ static bool call_dml_mode_support_and_programming(struct dc_state *context) return result; } -static bool dml2_validate_and_build_resource(const struct dc *in_dc, struct dc_state *context) +static bool dml2_validate_and_build_resource(const struct dc *in_dc, struct dc_state *context, + enum dc_validate_mode validate_mode) { struct dml2_context *dml2 = context->bw_ctx.dml2; struct dml2_wrapper_scratch *s = &dml2->v20.scratch; @@ -611,7 +432,7 @@ static bool dml2_validate_and_build_resource(const struct dc *in_dc, struct dc_s copy_dummy_pstate_table(s->dummy_pstate_table, in_dc->clk_mgr->bw_params->dummy_pstate_table, 4); - result = call_dml_mode_support_and_programming(context); + result = call_dml_mode_support_and_programming(context, validate_mode); /* Call map dc pipes to map the pipes based on the DML output. For correctly determining if recalculation * is required or not, the resource context needs to correctly reflect the number of active pipes. We would * only know the correct number if active pipes after dml2_map_dc_pipes is called. @@ -628,7 +449,7 @@ static bool dml2_validate_and_build_resource(const struct dc *in_dc, struct dc_s need_recalculation = dml2_verify_det_buffer_configuration(dml2, context, &dml2->det_helper_scratch); if (need_recalculation) { /* Engage the DML again if recalculation is required. */ - call_dml_mode_support_and_programming(context); + call_dml_mode_support_and_programming(context, validate_mode); if (!dml2->config.skip_hw_state_mapping) { dml2_map_dc_pipes(dml2, context, &s->cur_display_config, &s->dml_to_dc_pipe_mapping, in_dc->current_state); } @@ -684,7 +505,7 @@ static bool dml2_validate_and_build_resource(const struct dc *in_dc, struct dc_s return result; } -static bool dml2_validate_only(struct dc_state *context) +static bool dml2_validate_only(struct dc_state *context, enum dc_validate_mode validate_mode) { struct dml2_context *dml2; unsigned int result = 0; @@ -708,7 +529,8 @@ static bool dml2_validate_only(struct dc_state *context) result = pack_and_call_dml_mode_support_ex(dml2, &dml2->v20.scratch.cur_display_config, - &dml2->v20.scratch.mode_support_info); + &dml2->v20.scratch.mode_support_info, + validate_mode); if (result) result = does_configuration_meet_sw_policies(dml2, &dml2->v20.scratch.cur_display_config, &dml2->v20.scratch.mode_support_info); @@ -723,7 +545,8 @@ static void dml2_apply_debug_options(const struct dc *dc, struct dml2_context *d } } -bool dml2_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml2, bool fast_validate) +bool dml2_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml2, + enum dc_validate_mode validate_mode) { bool out = false; @@ -733,17 +556,17 @@ bool dml2_validate(const struct dc *in_dc, struct dc_state *context, struct dml2 /* DML2.1 validation path */ if (dml2->architecture == dml2_architecture_21) { - out = dml21_validate(in_dc, context, dml2, fast_validate); + out = dml21_validate(in_dc, context, dml2, validate_mode); return out; } DC_FP_START(); - /* Use dml_validate_only for fast_validate path */ - if (fast_validate) - out = dml2_validate_only(context); + /* Use dml_validate_only for DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX path */ + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) + out = dml2_validate_only(context, validate_mode); else - out = dml2_validate_and_build_resource(in_dc, context); + out = dml2_validate_and_build_resource(in_dc, context, validate_mode); DC_FP_END(); @@ -757,8 +580,8 @@ static inline struct dml2_context *dml2_allocate_memory(void) static void dml2_init(const struct dc *in_dc, const struct dml2_configuration_options *config, struct dml2_context **dml2) { - if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version == DCN_VERSION_4_01)) { - dml21_reinit(in_dc, dml2, config); + if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version >= DCN_VERSION_4_01)) { + dml21_reinit(in_dc, *dml2, config); return; } @@ -803,9 +626,7 @@ static void dml2_init(const struct dc *in_dc, const struct dml2_configuration_op bool dml2_create(const struct dc *in_dc, const struct dml2_configuration_options *config, struct dml2_context **dml2) { // TODO : Temporarily add DCN_VERSION_3_2 for N-1 validation. Remove DCN_VERSION_3_2 after N-1 validation phase is complete. - if ((in_dc->debug.using_dml21) - && (in_dc->ctx->dce_version == DCN_VERSION_4_01 - )) + if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version >= DCN_VERSION_4_01)) return dml21_create(in_dc, dml2, config); // Allocate Mode Lib Ctx @@ -874,8 +695,8 @@ void dml2_reinit(const struct dc *in_dc, const struct dml2_configuration_options *config, struct dml2_context **dml2) { - if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version == DCN_VERSION_4_01)) { - dml21_reinit(in_dc, dml2, config); + if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version >= DCN_VERSION_4_01)) { + dml21_reinit(in_dc, *dml2, config); return; } diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h index 5100f269368e..c384e141cebc 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h @@ -240,7 +240,7 @@ struct dml2_configuration_options { bool use_clock_dc_limits; bool gpuvm_enable; bool force_tdlut_enable; - struct dml2_soc_bb *bb_from_dmub; + void *bb_from_dmub; }; /* @@ -272,7 +272,7 @@ void dml2_reinit(const struct dc *in_dc, * dml2_validate - Determines if a display configuration is supported or not. * @in_dc: dc. * @context: dc_state to be validated. - * @fast_validate: Fast validate will not populate context.res_ctx. + * @validate_mode: DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX will not populate context.res_ctx. * * DML1.0 compatible interface for validation. * @@ -295,7 +295,7 @@ void dml2_reinit(const struct dc *in_dc, bool dml2_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml2, - bool fast_validate); + enum dc_validate_mode validate_mode); /* * dml2_extract_dram_and_fclk_change_support - Extracts the FCLK and UCLK change support info. diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.c index 97bf26fa3573..36187f890d5d 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.c @@ -231,7 +231,7 @@ static struct dpp_funcs dcn401_dpp_funcs = { .dpp_program_regamma_pwl = NULL, .dpp_set_pre_degam = dpp3_set_pre_degam, .dpp_program_input_lut = NULL, - .dpp_full_bypass = dpp401_full_bypass, + .dpp_full_bypass = NULL, .dpp_setup = dpp401_dpp_setup, .dpp_program_degamma_pwl = NULL, .dpp_program_cm_dealpha = dpp3_program_cm_dealpha, diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.h b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.h index ecaa976e1f52..5a6a861402b3 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.h +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.h @@ -641,6 +641,7 @@ uint32_t ISHARP_DELTA_DATA; \ uint32_t ISHARP_DELTA_INDEX; \ uint32_t ISHARP_NLDELTA_SOFT_CLIP + struct dcn401_dpp_registers { DPP_REG_VARIABLE_LIST_DCN401; }; @@ -683,8 +684,6 @@ void dpp401_dscl_set_scaler_manual_scale( struct dpp *dpp_base, const struct scaler_data *scl_data); -void dpp401_full_bypass(struct dpp *dpp_base); - void dpp401_dpp_setup( struct dpp *dpp_base, enum surface_pixel_format format, diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c index 712aff7e17f7..7aab77b58869 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c @@ -88,30 +88,6 @@ enum dscl_mode_sel { DSCL_MODE_DSCL_BYPASS = 6 }; -void dpp401_full_bypass(struct dpp *dpp_base) -{ - struct dcn401_dpp *dpp = TO_DCN401_DPP(dpp_base); - - /* Input pixel format: ARGB8888 */ - REG_SET(CNVC_SURFACE_PIXEL_FORMAT, 0, - CNVC_SURFACE_PIXEL_FORMAT, 0x8); - - /* Zero expansion */ - REG_SET_3(FORMAT_CONTROL, 0, - CNVC_BYPASS, 0, - FORMAT_CONTROL__ALPHA_EN, 0, - FORMAT_EXPANSION_MODE, 0); - - /* COLOR_KEYER_CONTROL.COLOR_KEYER_EN = 0 this should be default */ - if (dpp->tf_mask->CM_BYPASS_EN) - REG_SET(CM_CONTROL, 0, CM_BYPASS_EN, 1); - else - REG_SET(CM_CONTROL, 0, CM_BYPASS, 1); - - /* Setting degamma bypass for now */ - REG_SET(CM_DGAM_CONTROL, 0, CM_DGAM_LUT_MODE, 0); -} - void dpp401_set_cursor_attributes( struct dpp *dpp_base, struct dc_cursor_attributes *cursor_attributes) diff --git a/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.c b/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.c index 4222679fd4c9..af5de564faec 100644 --- a/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.c +++ b/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.c @@ -17,8 +17,6 @@ static void dsc_write_to_registers(struct display_stream_compressor *dsc, const /* Object I/F functions */ //static void dsc401_get_enc_caps(struct dsc_enc_caps *dsc_enc_caps, int pixel_clock_100Hz); //static bool dsc401_get_packed_pps(struct display_stream_compressor *dsc, const struct dsc_config *dsc_cfg, uint8_t *dsc_packed_pps); -static void dsc401_wait_disconnect_pending_clear(struct display_stream_compressor *dsc); -static void dsc401_get_enc_caps(struct dsc_enc_caps *dsc_enc_caps, int pixel_clock_100Hz); static const struct dsc_funcs dcn401_dsc_funcs = { .dsc_get_enc_caps = dsc401_get_enc_caps, @@ -66,7 +64,7 @@ void dsc401_construct(struct dcn401_dsc *dsc, dsc->max_image_width = 5184; } -static void dsc401_get_enc_caps(struct dsc_enc_caps *dsc_enc_caps, int pixel_clock_100Hz) +void dsc401_get_enc_caps(struct dsc_enc_caps *dsc_enc_caps, int pixel_clock_100Hz) { int min_dsc_unit_required = (pixel_clock_100Hz + MAX_THROUGHPUT_PER_DSC_100HZ - 1) / MAX_THROUGHPUT_PER_DSC_100HZ; @@ -191,7 +189,7 @@ void dsc401_disable(struct display_stream_compressor *dsc) DSC_CLOCK_EN, 0); } -static void dsc401_wait_disconnect_pending_clear(struct display_stream_compressor *dsc) +void dsc401_wait_disconnect_pending_clear(struct display_stream_compressor *dsc) { struct dcn401_dsc *dsc401 = TO_DCN401_DSC(dsc); diff --git a/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.h b/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.h index e3ca70058e64..1d927d8e83bf 100644 --- a/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.h +++ b/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.h @@ -341,5 +341,7 @@ void dsc401_set_config(struct display_stream_compressor *dsc, const struct dsc_c void dsc401_enable(struct display_stream_compressor *dsc, int opp_pipe); void dsc401_disable(struct display_stream_compressor *dsc); void dsc401_disconnect(struct display_stream_compressor *dsc); +void dsc401_get_enc_caps(struct dsc_enc_caps *dsc_enc_caps, int pixel_clock_100Hz); +void dsc401_wait_disconnect_pending_clear(struct display_stream_compressor *dsc); #endif diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.h b/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.h index c7765e6f09e6..f8f991785d4f 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.h +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.h @@ -666,10 +666,29 @@ struct dcn_mi_mask { DCN_HUBP_REG_FIELD_LIST(uint32_t); }; +struct dcn_fl_regs_st { + uint32_t lut_enable; + uint32_t lut_done; + uint32_t lut_addr_mode; + uint32_t lut_width; + uint32_t lut_tmz; + uint32_t lut_crossbar_sel_r; + uint32_t lut_crossbar_sel_g; + uint32_t lut_crossbar_sel_b; + uint32_t lut_addr_hi; + uint32_t lut_addr_lo; + uint32_t refcyc_3dlut_group; + uint32_t lut_fl_bias; + uint32_t lut_fl_scale; + uint32_t lut_fl_mode; + uint32_t lut_fl_format; +}; + struct dcn_hubp_state { struct _vcs_dpi_display_dlg_regs_st dlg_attr; struct _vcs_dpi_display_ttu_regs_st ttu_attr; struct _vcs_dpi_display_rq_regs_st rq_regs; + struct dcn_fl_regs_st fl_regs; uint32_t pixel_format; uint32_t inuse_addr_hi; uint32_t inuse_addr_lo; diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c index e8730cc40edb..252e862449a2 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c @@ -1186,8 +1186,10 @@ void dce110_disable_stream(struct pipe_ctx *pipe_ctx) if (dccg) { dccg->funcs->disable_symclk32_se(dccg, dp_hpo_inst); dccg->funcs->set_dpstreamclk(dccg, REFCLK, tg->inst, dp_hpo_inst); - if (dccg && dccg->funcs->set_dtbclk_dto) - dccg->funcs->set_dtbclk_dto(dccg, &dto_params); + if (!(dc->ctx->dce_version >= DCN_VERSION_3_5)) { + if (dccg && dccg->funcs->set_dtbclk_dto) + dccg->funcs->set_dtbclk_dto(dccg, &dto_params); + } } } else if (dccg && dccg->funcs->disable_symclk_se) { dccg->funcs->disable_symclk_se(dccg, stream_enc->stream_enc_inst, @@ -1225,7 +1227,7 @@ void dce110_blank_stream(struct pipe_ctx *pipe_ctx) return; if (link->local_sink && link->local_sink->sink_signal == SIGNAL_TYPE_EDP) { - if (!link->skip_implict_edp_power_control) + if (!link->skip_implict_edp_power_control && hws) hws->funcs.edp_backlight_control(link, false); link->dc->hwss.set_abm_immediate_disable(pipe_ctx); } @@ -1379,7 +1381,7 @@ static void populate_audio_dp_link_info( } } -static void build_audio_output( +void build_audio_output( struct dc_state *state, const struct pipe_ctx *pipe_ctx, struct audio_output *audio_output) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.h b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.h index 06789ac3a224..7cd8c1576988 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.h @@ -110,5 +110,9 @@ void dce110_enable_dp_link_output( enum signal_type signal, enum clock_source_id clock_source, const struct dc_link_settings *link_settings); +void build_audio_output( + struct dc_state *state, + const struct pipe_ctx *pipe_ctx, + struct audio_output *audio_output); #endif /* __DC_HWSS_DCE110_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c index f9ee55998b6b..9ad29be925e2 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c @@ -327,6 +327,35 @@ static void dcn10_log_hubp_states(struct dc *dc, void *log_ctx) } } + DTN_INFO("\n=======HUBP FL======\n"); + DTN_INFO( + "HUBP FL: Enabled Done adr_mode width tmz xbar_sel_R xbar_sel_G xbar_sel_B adr_hi adr_low REFCYC Bias Scale Mode Format\n"); + for (i = 0; i < pool->pipe_count; i++) { + struct dcn_hubp_state *s = &(TO_DCN10_HUBP(pool->hubps[i])->state); + struct dcn_fl_regs_st *fl_regs = &s->fl_regs; + + if (!s->blank_en) { + DTN_INFO("[%2d]: %5xh %6xh %5d %6d %2xh %2xh %6xh %6d %8d %8d %7d %8xh %5x %5x %5x", + pool->hubps[i]->inst, + fl_regs->lut_enable, + fl_regs->lut_done, + fl_regs->lut_addr_mode, + fl_regs->lut_width, + fl_regs->lut_tmz, + fl_regs->lut_crossbar_sel_r, + fl_regs->lut_crossbar_sel_g, + fl_regs->lut_crossbar_sel_b, + fl_regs->lut_addr_hi, + fl_regs->lut_addr_lo, + fl_regs->refcyc_3dlut_group, + fl_regs->lut_fl_bias, + fl_regs->lut_fl_scale, + fl_regs->lut_fl_mode, + fl_regs->lut_fl_format); + DTN_INFO("\n"); + } + } + DTN_INFO("\n=========RQ========\n"); DTN_INFO("HUBP: drq_exp_m prq_exp_m mrq_exp_m crq_exp_m plane1_ba L:chunk_s min_chu_s meta_ch_s" " min_m_c_s dpte_gr_s mpte_gr_s swath_hei pte_row_h C:chunk_s min_chu_s meta_ch_s" @@ -511,6 +540,36 @@ static void dcn10_log_color_state(struct dc *dc, dc->caps.color.mpc.num_3dluts, dc->caps.color.mpc.ogam_ram, dc->caps.color.mpc.ocsc); + DTN_INFO("===== MPC RMCM 3DLUT =====\n"); + DTN_INFO("MPCC: SIZE MODE MODE_CUR RD_SEL 30BIT_EN WR_EN_MASK RAM_SEL OUT_NORM_FACTOR FL_SEL OUT_OFFSET OUT_SCALE FL_DONE SOFT_UNDERFLOW HARD_UNDERFLOW MEM_PWR_ST FORCE DIS MODE\n"); + for (i = 0; i < pool->mpcc_count; i++) { + struct mpcc_state s = {0}; + + pool->mpc->funcs->read_mpcc_state(pool->mpc, i, &s); + if (s.opp_id != 0xf) + DTN_INFO("[%2d]: %4xh %4xh %6xh %4x %4x %4x %4x %4x %4xh %4xh %6xh %4x %4x %4x %4x %4x %4x %4x\n", + i, s.rmcm_regs.rmcm_3dlut_size, s.rmcm_regs.rmcm_3dlut_mode, s.rmcm_regs.rmcm_3dlut_mode_cur, + s.rmcm_regs.rmcm_3dlut_read_sel, s.rmcm_regs.rmcm_3dlut_30bit_en, s.rmcm_regs.rmcm_3dlut_wr_en_mask, + s.rmcm_regs.rmcm_3dlut_ram_sel, s.rmcm_regs.rmcm_3dlut_out_norm_factor, s.rmcm_regs.rmcm_3dlut_fl_sel, + s.rmcm_regs.rmcm_3dlut_out_offset_r, s.rmcm_regs.rmcm_3dlut_out_scale_r, s.rmcm_regs.rmcm_3dlut_fl_done, + s.rmcm_regs.rmcm_3dlut_fl_soft_underflow, s.rmcm_regs.rmcm_3dlut_fl_hard_underflow, s.rmcm_regs.rmcm_3dlut_mem_pwr_state, + s.rmcm_regs.rmcm_3dlut_mem_pwr_force, s.rmcm_regs.rmcm_3dlut_mem_pwr_dis, s.rmcm_regs.rmcm_3dlut_mem_pwr_mode); + } + DTN_INFO("\n"); + DTN_INFO("===== MPC RMCM Shaper =====\n"); + DTN_INFO("MPCC: CNTL LUT_MODE MODE_CUR WR_EN_MASK WR_SEL OFFSET SCALE START_B START_SEG_B END_B END_BASE_B MEM_PWR_ST FORCE DIS MODE\n"); + for (i = 0; i < pool->mpcc_count; i++) { + struct mpcc_state s = {0}; + + pool->mpc->funcs->read_mpcc_state(pool->mpc, i, &s); + if (s.opp_id != 0xf) + DTN_INFO("[%2d]: %4xh %4xh %6xh %4x %4x %4x %4x %4x %4xh %4xh %6xh %4x %4x %4x %4x\n", + i, s.rmcm_regs.rmcm_cntl, s.rmcm_regs.rmcm_shaper_lut_mode, s.rmcm_regs.rmcm_shaper_mode_cur, + s.rmcm_regs.rmcm_shaper_lut_write_en_mask, s.rmcm_regs.rmcm_shaper_lut_write_sel, s.rmcm_regs.rmcm_shaper_offset_b, + s.rmcm_regs.rmcm_shaper_scale_b, s.rmcm_regs.rmcm_shaper_rama_exp_region_start_b, s.rmcm_regs.rmcm_shaper_rama_exp_region_start_seg_b, + s.rmcm_regs.rmcm_shaper_rama_exp_region_end_b, s.rmcm_regs.rmcm_shaper_rama_exp_region_end_base_b, s.rmcm_regs.rmcm_shaper_mem_pwr_state, + s.rmcm_regs.rmcm_shaper_mem_pwr_force, s.rmcm_regs.rmcm_shaper_mem_pwr_dis, s.rmcm_regs.rmcm_shaper_mem_pwr_mode); + } } void dcn10_log_hw_state(struct dc *dc, diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c index c277df12c817..3207addbd4eb 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c @@ -283,14 +283,13 @@ void dcn20_setup_gsl_group_as_lock( } /* at this point we want to program whether it's to enable or disable */ - if (pipe_ctx->stream_res.tg->funcs->set_gsl != NULL && - pipe_ctx->stream_res.tg->funcs->set_gsl_source_select != NULL) { + if (pipe_ctx->stream_res.tg->funcs->set_gsl != NULL) { pipe_ctx->stream_res.tg->funcs->set_gsl( pipe_ctx->stream_res.tg, &gsl); - - pipe_ctx->stream_res.tg->funcs->set_gsl_source_select( - pipe_ctx->stream_res.tg, group_idx, enable ? 4 : 0); + if (pipe_ctx->stream_res.tg->funcs->set_gsl_source_select != NULL) + pipe_ctx->stream_res.tg->funcs->set_gsl_source_select( + pipe_ctx->stream_res.tg, group_idx, enable ? 4 : 0); } else BREAK_TO_DEBUGGER(); } @@ -956,7 +955,7 @@ enum dc_status dcn20_enable_stream_timing( return DC_ERROR_UNEXPECTED; } - hws->funcs.wait_for_blank_complete(pipe_ctx->stream_res.opp); + fsleep(stream->timing.v_total * (stream->timing.h_total * 10000u / stream->timing.pix_clk_100hz)); params.vertical_total_min = stream->adjust.v_total_min; params.vertical_total_max = stream->adjust.v_total_max; @@ -1971,14 +1970,6 @@ static void dcn20_program_pipe( pipe_ctx->plane_state->update_flags.bits.hdr_mult)) hws->funcs.set_hdr_multiplier(pipe_ctx); - if (hws->funcs.populate_mcm_luts) { - if (pipe_ctx->plane_state) { - hws->funcs.populate_mcm_luts(dc, pipe_ctx, pipe_ctx->plane_state->mcm_luts, - pipe_ctx->plane_state->lut_bank_a); - pipe_ctx->plane_state->lut_bank_a = !pipe_ctx->plane_state->lut_bank_a; - } - } - if (pipe_ctx->plane_state && (pipe_ctx->plane_state->update_flags.bits.in_transfer_func_change || pipe_ctx->plane_state->update_flags.bits.gamma_change || @@ -2492,7 +2483,7 @@ bool dcn20_update_bandwidth( struct dce_hwseq *hws = dc->hwseq; /* recalculate DML parameters */ - if (dc->res_pool->funcs->validate_bandwidth(dc, context, false) != DC_OK) + if (dc->res_pool->funcs->validate_bandwidth(dc, context, DC_VALIDATE_MODE_AND_PROGRAMMING) != DC_OK) return false; /* apply updated bandwidth parameters */ @@ -2816,6 +2807,8 @@ void dcn20_reset_back_end_for_pipe( { struct dc_link *link = pipe_ctx->stream->link; const struct link_hwss *link_hwss = get_link_hwss(link, &pipe_ctx->link_res); + struct dccg *dccg = dc->res_pool->dccg; + struct dtbclk_dto_params dto_params = {0}; DC_LOGGER_INIT(dc->ctx->logger); if (pipe_ctx->stream_res.stream_enc == NULL) { @@ -2876,6 +2869,13 @@ void dcn20_reset_back_end_for_pipe( &pipe_ctx->link_res, pipe_ctx->stream->signal); link->phy_state.symclk_state = SYMCLK_OFF_TX_OFF; } + if (dc->link_srv->dp_is_128b_132b_signal(pipe_ctx) && dccg + && dc->ctx->dce_version >= DCN_VERSION_3_5) { + dto_params.otg_inst = pipe_ctx->stream_res.tg->inst; + dto_params.timing = &pipe_ctx->stream->timing; + if (dccg && dccg->funcs->set_dtbclk_dto) + dccg->funcs->set_dtbclk_dto(dccg, &dto_params); + } } /* diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_hwseq.c index e68f21fd5f0f..a40e119d8582 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_hwseq.c @@ -55,15 +55,15 @@ #include "dcn20/dcn20_optc.h" #include "dcn30/dcn30_cm_common.h" -#define DC_LOGGER_INIT(logger) +#define DC_LOGGER_INIT(logger) \ + struct dal_logger *dc_logger = logger #define CTX \ hws->ctx #define REG(reg)\ hws->regs->reg #define DC_LOGGER \ - stream->ctx->logger - + dc_logger #undef FN #define FN(reg_name, field_name) \ @@ -76,6 +76,8 @@ static void update_dsc_on_stream(struct pipe_ctx *pipe_ctx, bool enable) struct pipe_ctx *odm_pipe; int opp_cnt = 1; + DC_LOGGER_INIT(stream->ctx->logger); + ASSERT(dsc); for (odm_pipe = pipe_ctx->next_odm_pipe; odm_pipe; odm_pipe = odm_pipe->next_odm_pipe) opp_cnt++; @@ -528,3 +530,32 @@ void dcn314_disable_link_output(struct dc_link *link, apply_symclk_on_tx_off_wa(link); } + + +void dcn314_plane_atomic_power_down(struct dc *dc, + struct dpp *dpp, + struct hubp *hubp) +{ + struct dce_hwseq *hws = dc->hwseq; + DC_LOGGER_INIT(dc->ctx->logger); + + if (REG(DC_IP_REQUEST_CNTL)) { + REG_SET(DC_IP_REQUEST_CNTL, 0, IP_REQUEST_EN, 1); + + if (hws->funcs.dpp_pg_control) { + hws->funcs.dpp_pg_control(hws, dpp->inst, false); + dpp->funcs->dpp_reset(dpp); + } + + if (hws->funcs.hubp_pg_control) { + hws->funcs.hubp_pg_control(hws, hubp->inst, false); + hubp->funcs->hubp_reset(hubp); + } + + REG_SET(DC_IP_REQUEST_CNTL, 0, IP_REQUEST_EN, 0); + DC_LOG_DEBUG("Power gated front end %d\n", hubp->inst); + } + + if (hws->funcs.dpp_root_clock_control) + hws->funcs.dpp_root_clock_control(hws, dpp->inst, false); +} diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_hwseq.h b/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_hwseq.h index 2305ad282f21..12a57b79edfb 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_hwseq.h @@ -47,4 +47,6 @@ void dcn314_dpp_root_clock_control(struct dce_hwseq *hws, unsigned int dpp_inst, void dcn314_disable_link_output(struct dc_link *link, const struct link_resource *link_res, enum signal_type signal); +void dcn314_plane_atomic_power_down(struct dc *dc, struct dpp *dpp, struct hubp *hubp); + #endif /* __DC_HWSS_DCN314_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_init.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_init.c index f5112742edf9..6963d25608ac 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_init.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_init.c @@ -137,7 +137,7 @@ static const struct hwseq_private_funcs dcn314_private_funcs = { .disable_vga = dcn20_disable_vga, .bios_golden_init = dcn10_bios_golden_init, .plane_atomic_disable = dcn20_plane_atomic_disable, - .plane_atomic_power_down = dcn10_plane_atomic_power_down, + .plane_atomic_power_down = dcn314_plane_atomic_power_down, .enable_power_gating_plane = dcn314_enable_power_gating_plane, .dpp_root_clock_control = dcn314_dpp_root_clock_control, .hubp_pg_control = dcn31_hubp_pg_control, diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c index a0b05b9ef660..416b1dca3dac 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c @@ -1063,15 +1063,17 @@ void dcn32_update_dsc_on_stream(struct pipe_ctx *pipe_ctx, bool enable) dsc_cfg.dc_dsc_cfg.num_slices_h /= opp_cnt; if (should_use_dto_dscclk) - dccg->funcs->set_dto_dscclk(dccg, dsc->inst); + dccg->funcs->set_dto_dscclk(dccg, dsc->inst, dsc_cfg.dc_dsc_cfg.num_slices_h); dsc->funcs->dsc_set_config(dsc, &dsc_cfg, &dsc_optc_cfg); dsc->funcs->dsc_enable(dsc, pipe_ctx->stream_res.opp->inst); for (odm_pipe = pipe_ctx->next_odm_pipe; odm_pipe; odm_pipe = odm_pipe->next_odm_pipe) { struct display_stream_compressor *odm_dsc = odm_pipe->stream_res.dsc; ASSERT(odm_dsc); + if (!odm_dsc) + continue; if (should_use_dto_dscclk) - dccg->funcs->set_dto_dscclk(dccg, odm_dsc->inst); + dccg->funcs->set_dto_dscclk(dccg, odm_dsc->inst, dsc_cfg.dc_dsc_cfg.num_slices_h); odm_dsc->funcs->dsc_set_config(odm_dsc, &dsc_cfg, &dsc_optc_cfg); odm_dsc->funcs->dsc_enable(odm_dsc, odm_pipe->stream_res.opp->inst); } diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c index c814d957305a..a267f574b619 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c @@ -1047,6 +1047,15 @@ void dcn35_calc_blocks_to_gate(struct dc *dc, struct dc_state *context, if (dc->caps.sequential_ono) { update_state->pg_pipe_res_update[PG_HUBP][pipe_ctx->stream_res.dsc->inst] = false; update_state->pg_pipe_res_update[PG_DPP][pipe_ctx->stream_res.dsc->inst] = false; + + /* All HUBP/DPP instances must be powered if the DSC inst != HUBP inst */ + if (!pipe_ctx->top_pipe && pipe_ctx->plane_res.hubp && + pipe_ctx->plane_res.hubp->inst != pipe_ctx->stream_res.dsc->inst) { + for (j = 0; j < dc->res_pool->pipe_count; ++j) { + update_state->pg_pipe_res_update[PG_HUBP][j] = false; + update_state->pg_pipe_res_update[PG_DPP][j] = false; + } + } } } @@ -1193,6 +1202,25 @@ void dcn35_calc_blocks_to_ungate(struct dc *dc, struct dc_state *context, update_state->pg_pipe_res_update[PG_HDMISTREAM][0] = true; if (dc->caps.sequential_ono) { + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *new_pipe = &context->res_ctx.pipe_ctx[i]; + + if (new_pipe->stream_res.dsc && !new_pipe->top_pipe && + update_state->pg_pipe_res_update[PG_DSC][new_pipe->stream_res.dsc->inst]) { + update_state->pg_pipe_res_update[PG_HUBP][new_pipe->stream_res.dsc->inst] = true; + update_state->pg_pipe_res_update[PG_DPP][new_pipe->stream_res.dsc->inst] = true; + + /* All HUBP/DPP instances must be powered if the DSC inst != HUBP inst */ + if (new_pipe->plane_res.hubp && + new_pipe->plane_res.hubp->inst != new_pipe->stream_res.dsc->inst) { + for (j = 0; j < dc->res_pool->pipe_count; ++j) { + update_state->pg_pipe_res_update[PG_HUBP][j] = true; + update_state->pg_pipe_res_update[PG_DPP][j] = true; + } + } + } + } + for (i = dc->res_pool->pipe_count - 1; i >= 0; i--) { if (update_state->pg_pipe_res_update[PG_HUBP][i] && update_state->pg_pipe_res_update[PG_DPP][i]) { diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c index c4177a9a662f..a0d61df07f22 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c @@ -2,6 +2,8 @@ // // Copyright 2024 Advanced Micro Devices, Inc. + +#include "os_types.h" #include "dm_services.h" #include "basics/dc_common.h" #include "dm_helpers.h" @@ -143,13 +145,8 @@ void dcn401_init_hw(struct dc *dc) dc->clk_mgr->funcs->init_clocks(dc->clk_mgr); // mark dcmode limits present if any clock has distinct AC and DC values from SMU - dc->caps.dcmode_power_limits_present = - (dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_dcfclk_levels && dc->clk_mgr->bw_params->dc_mode_limit.dcfclk_mhz) || - (dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_dispclk_levels && dc->clk_mgr->bw_params->dc_mode_limit.dispclk_mhz) || - (dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_dtbclk_levels && dc->clk_mgr->bw_params->dc_mode_limit.dtbclk_mhz) || - (dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_fclk_levels && dc->clk_mgr->bw_params->dc_mode_limit.fclk_mhz) || - (dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_memclk_levels && dc->clk_mgr->bw_params->dc_mode_limit.memclk_mhz) || - (dc->clk_mgr->bw_params->clk_table.num_entries_per_clk.num_socclk_levels && dc->clk_mgr->bw_params->dc_mode_limit.socclk_mhz); + dc->caps.dcmode_power_limits_present = dc->clk_mgr->funcs->is_dc_mode_present && + dc->clk_mgr->funcs->is_dc_mode_present(dc->clk_mgr); } // Initialize the dccg @@ -396,249 +393,6 @@ static void dcn401_get_mcm_lut_xable_from_pipe_ctx(struct dc *dc, struct pipe_ct } } -static void dcn401_set_mcm_location_post_blend(struct dc *dc, struct pipe_ctx *pipe_ctx, bool bPostBlend) -{ - struct mpc *mpc = dc->res_pool->mpc; - int mpcc_id = pipe_ctx->plane_res.hubp->inst; - - if (!pipe_ctx->plane_state) - return; - - mpc->funcs->set_movable_cm_location(mpc, MPCC_MOVABLE_CM_LOCATION_BEFORE, mpcc_id); - pipe_ctx->plane_state->mcm_location = (bPostBlend) ? - MPCC_MOVABLE_CM_LOCATION_AFTER : - MPCC_MOVABLE_CM_LOCATION_BEFORE; -} - -static void dc_get_lut_mode( - enum dc_cm2_gpu_mem_layout layout, - enum hubp_3dlut_fl_mode *mode, - enum hubp_3dlut_fl_addressing_mode *addr_mode) -{ - switch (layout) { - case DC_CM2_GPU_MEM_LAYOUT_3D_SWIZZLE_LINEAR_RGB: - *mode = hubp_3dlut_fl_mode_native_1; - *addr_mode = hubp_3dlut_fl_addressing_mode_sw_linear; - break; - case DC_CM2_GPU_MEM_LAYOUT_3D_SWIZZLE_LINEAR_BGR: - *mode = hubp_3dlut_fl_mode_native_2; - *addr_mode = hubp_3dlut_fl_addressing_mode_sw_linear; - break; - case DC_CM2_GPU_MEM_LAYOUT_1D_PACKED_LINEAR: - *mode = hubp_3dlut_fl_mode_transform; - *addr_mode = hubp_3dlut_fl_addressing_mode_simple_linear; - break; - default: - *mode = hubp_3dlut_fl_mode_disable; - *addr_mode = hubp_3dlut_fl_addressing_mode_sw_linear; - break; - } -} - -static void dc_get_lut_format( - enum dc_cm2_gpu_mem_format dc_format, - enum hubp_3dlut_fl_format *format) -{ - switch (dc_format) { - case DC_CM2_GPU_MEM_FORMAT_16161616_UNORM_12MSB: - *format = hubp_3dlut_fl_format_unorm_12msb_bitslice; - break; - case DC_CM2_GPU_MEM_FORMAT_16161616_UNORM_12LSB: - *format = hubp_3dlut_fl_format_unorm_12lsb_bitslice; - break; - case DC_CM2_GPU_MEM_FORMAT_16161616_FLOAT_FP1_5_10: - *format = hubp_3dlut_fl_format_float_fp1_5_10; - break; - } -} - -static void dc_get_lut_xbar( - enum dc_cm2_gpu_mem_pixel_component_order order, - enum hubp_3dlut_fl_crossbar_bit_slice *cr_r, - enum hubp_3dlut_fl_crossbar_bit_slice *y_g, - enum hubp_3dlut_fl_crossbar_bit_slice *cb_b) -{ - switch (order) { - case DC_CM2_GPU_MEM_PIXEL_COMPONENT_ORDER_RGBA: - *cr_r = hubp_3dlut_fl_crossbar_bit_slice_32_47; - *y_g = hubp_3dlut_fl_crossbar_bit_slice_16_31; - *cb_b = hubp_3dlut_fl_crossbar_bit_slice_0_15; - break; - case DC_CM2_GPU_MEM_PIXEL_COMPONENT_ORDER_BGRA: - *cr_r = hubp_3dlut_fl_crossbar_bit_slice_0_15; - *y_g = hubp_3dlut_fl_crossbar_bit_slice_16_31; - *cb_b = hubp_3dlut_fl_crossbar_bit_slice_32_47; - break; - } -} - -static void dc_get_lut_width( - enum dc_cm2_gpu_mem_size size, - enum hubp_3dlut_fl_width *width) -{ - switch (size) { - case DC_CM2_GPU_MEM_SIZE_333333: - *width = hubp_3dlut_fl_width_33; - break; - case DC_CM2_GPU_MEM_SIZE_171717: - *width = hubp_3dlut_fl_width_17; - break; - case DC_CM2_GPU_MEM_SIZE_TRANSFORMED: - *width = hubp_3dlut_fl_width_transformed; - break; - } -} -static bool dc_is_rmcm_3dlut_supported(struct hubp *hubp, struct mpc *mpc) -{ - if (mpc->funcs->rmcm.update_3dlut_fast_load_select && - mpc->funcs->rmcm.program_lut_read_write_control && - hubp->funcs->hubp_program_3dlut_fl_addr && - mpc->funcs->rmcm.program_bit_depth && - hubp->funcs->hubp_program_3dlut_fl_mode && - hubp->funcs->hubp_program_3dlut_fl_addressing_mode && - hubp->funcs->hubp_program_3dlut_fl_format && - hubp->funcs->hubp_update_3dlut_fl_bias_scale && - mpc->funcs->rmcm.program_bias_scale && - hubp->funcs->hubp_program_3dlut_fl_crossbar && - hubp->funcs->hubp_program_3dlut_fl_width && - mpc->funcs->rmcm.update_3dlut_fast_load_select && - mpc->funcs->rmcm.populate_lut && - mpc->funcs->rmcm.program_lut_mode && - hubp->funcs->hubp_enable_3dlut_fl && - mpc->funcs->rmcm.enable_3dlut_fl) - return true; - - return false; -} - -bool dcn401_program_rmcm_luts( - struct hubp *hubp, - struct pipe_ctx *pipe_ctx, - enum dc_cm2_transfer_func_source lut3d_src, - struct dc_cm2_func_luts *mcm_luts, - struct mpc *mpc, - bool lut_bank_a, - int mpcc_id) -{ - struct dpp *dpp_base = pipe_ctx->plane_res.dpp; - union mcm_lut_params m_lut_params; - enum MCM_LUT_XABLE shaper_xable, lut3d_xable = MCM_LUT_DISABLE, lut1d_xable; - enum hubp_3dlut_fl_mode mode; - enum hubp_3dlut_fl_addressing_mode addr_mode; - enum hubp_3dlut_fl_format format = 0; - enum hubp_3dlut_fl_crossbar_bit_slice crossbar_bit_slice_y_g = 0; - enum hubp_3dlut_fl_crossbar_bit_slice crossbar_bit_slice_cb_b = 0; - enum hubp_3dlut_fl_crossbar_bit_slice crossbar_bit_slice_cr_r = 0; - enum hubp_3dlut_fl_width width = 0; - struct dc *dc = hubp->ctx->dc; - - bool bypass_rmcm_3dlut = false; - bool bypass_rmcm_shaper = false; - - dcn401_get_mcm_lut_xable_from_pipe_ctx(dc, pipe_ctx, &shaper_xable, &lut3d_xable, &lut1d_xable); - - /* 3DLUT */ - switch (lut3d_src) { - case DC_CM2_TRANSFER_FUNC_SOURCE_SYSMEM: - memset(&m_lut_params, 0, sizeof(m_lut_params)); - // Don't know what to do in this case. - //case DC_CM2_TRANSFER_FUNC_SOURCE_SYSMEM: - break; - case DC_CM2_TRANSFER_FUNC_SOURCE_VIDMEM: - dc_get_lut_width(mcm_luts->lut3d_data.gpu_mem_params.size, &width); - if (!dc_is_rmcm_3dlut_supported(hubp, mpc) || - !mpc->funcs->rmcm.is_config_supported(width)) - return false; - - //0. disable fl on mpc - mpc->funcs->update_3dlut_fast_load_select(mpc, mpcc_id, 0xF); - - //1. power down the block - mpc->funcs->rmcm.power_on_shaper_3dlut(mpc, mpcc_id, false); - - //2. program RMCM - //2a. 3dlut reg programming - mpc->funcs->rmcm.program_lut_read_write_control(mpc, MCM_LUT_3DLUT, lut_bank_a, - (!bypass_rmcm_3dlut) && lut3d_xable != MCM_LUT_DISABLE, mpcc_id); - - hubp->funcs->hubp_program_3dlut_fl_addr(hubp, - mcm_luts->lut3d_data.gpu_mem_params.addr); - - mpc->funcs->rmcm.program_bit_depth(mpc, - mcm_luts->lut3d_data.gpu_mem_params.bit_depth, mpcc_id); - - // setting native or transformed mode, - dc_get_lut_mode(mcm_luts->lut3d_data.gpu_mem_params.layout, &mode, &addr_mode); - - //these program the mcm 3dlut - hubp->funcs->hubp_program_3dlut_fl_mode(hubp, mode); - - hubp->funcs->hubp_program_3dlut_fl_addressing_mode(hubp, addr_mode); - - //seems to be only for the MCM - dc_get_lut_format(mcm_luts->lut3d_data.gpu_mem_params.format_params.format, &format); - hubp->funcs->hubp_program_3dlut_fl_format(hubp, format); - - mpc->funcs->rmcm.program_bias_scale(mpc, - mcm_luts->lut3d_data.gpu_mem_params.format_params.float_params.bias, - mcm_luts->lut3d_data.gpu_mem_params.format_params.float_params.scale, - mpcc_id); - hubp->funcs->hubp_update_3dlut_fl_bias_scale(hubp, - mcm_luts->lut3d_data.gpu_mem_params.format_params.float_params.bias, - mcm_luts->lut3d_data.gpu_mem_params.format_params.float_params.scale); - - dc_get_lut_xbar( - mcm_luts->lut3d_data.gpu_mem_params.component_order, - &crossbar_bit_slice_cr_r, - &crossbar_bit_slice_y_g, - &crossbar_bit_slice_cb_b); - - hubp->funcs->hubp_program_3dlut_fl_crossbar(hubp, - crossbar_bit_slice_cr_r, - crossbar_bit_slice_y_g, - crossbar_bit_slice_cb_b); - - mpc->funcs->rmcm.program_3dlut_size(mpc, width, mpcc_id); - - mpc->funcs->update_3dlut_fast_load_select(mpc, mpcc_id, hubp->inst); - - //2b. shaper reg programming - memset(&m_lut_params, 0, sizeof(m_lut_params)); - - if (mcm_luts->shaper->type == TF_TYPE_HWPWL) { - m_lut_params.pwl = &mcm_luts->shaper->pwl; - } else if (mcm_luts->shaper->type == TF_TYPE_DISTRIBUTED_POINTS) { - ASSERT(false); - cm_helper_translate_curve_to_hw_format( - dc->ctx, - mcm_luts->shaper, - &dpp_base->regamma_params, true); - m_lut_params.pwl = &dpp_base->regamma_params; - } - if (m_lut_params.pwl) { - mpc->funcs->rmcm.populate_lut(mpc, m_lut_params, lut_bank_a, mpcc_id); - mpc->funcs->rmcm.program_lut_mode(mpc, !bypass_rmcm_shaper, lut_bank_a, mpcc_id); - } else { - //RMCM 3dlut won't work without its shaper - return false; - } - - //3. Select the hubp connected to this RMCM - hubp->funcs->hubp_enable_3dlut_fl(hubp, true); - mpc->funcs->rmcm.enable_3dlut_fl(mpc, true, mpcc_id); - - //4. power on the block - if (m_lut_params.pwl) - mpc->funcs->rmcm.power_on_shaper_3dlut(mpc, mpcc_id, true); - - break; - default: - return false; - } - - return true; -} - void dcn401_populate_mcm_luts(struct dc *dc, struct pipe_ctx *pipe_ctx, struct dc_cm2_func_luts mcm_luts, @@ -664,25 +418,6 @@ void dcn401_populate_mcm_luts(struct dc *dc, dcn401_get_mcm_lut_xable_from_pipe_ctx(dc, pipe_ctx, &shaper_xable, &lut3d_xable, &lut1d_xable); - //MCM - setting its location (Before/After) blender - //set to post blend (true) - dcn401_set_mcm_location_post_blend( - dc, - pipe_ctx, - mcm_luts.lut3d_data.mpc_mcm_post_blend); - - //RMCM - 3dLUT+Shaper - if (mcm_luts.lut3d_data.rmcm_3dlut_enable) { - dcn401_program_rmcm_luts( - hubp, - pipe_ctx, - lut3d_src, - &mcm_luts, - mpc, - lut_bank_a, - mpcc_id); - } - /* 1D LUT */ if (mcm_luts.lut1d_func) { memset(&m_lut_params, 0, sizeof(m_lut_params)); @@ -740,15 +475,15 @@ void dcn401_populate_mcm_luts(struct dc *dc, break; case DC_CM2_TRANSFER_FUNC_SOURCE_VIDMEM: switch (mcm_luts.lut3d_data.gpu_mem_params.size) { - case DC_CM2_GPU_MEM_SIZE_333333: - width = hubp_3dlut_fl_width_33; - break; case DC_CM2_GPU_MEM_SIZE_171717: width = hubp_3dlut_fl_width_17; break; case DC_CM2_GPU_MEM_SIZE_TRANSFORMED: width = hubp_3dlut_fl_width_transformed; break; + default: + //TODO: handle default case + break; } //check for support @@ -817,11 +552,14 @@ void dcn401_populate_mcm_luts(struct dc *dc, //navi 4x has a bug and r and blue are swapped and need to be worked around here in //TODO: need to make a method for get_xbar per asic OR do the workaround in program_crossbar for 4x - dc_get_lut_xbar( - mcm_luts.lut3d_data.gpu_mem_params.component_order, - &crossbar_bit_slice_cr_r, - &crossbar_bit_slice_y_g, - &crossbar_bit_slice_cb_b); + switch (mcm_luts.lut3d_data.gpu_mem_params.component_order) { + case DC_CM2_GPU_MEM_PIXEL_COMPONENT_ORDER_RGBA: + default: + crossbar_bit_slice_cr_r = hubp_3dlut_fl_crossbar_bit_slice_0_15; + crossbar_bit_slice_y_g = hubp_3dlut_fl_crossbar_bit_slice_16_31; + crossbar_bit_slice_cb_b = hubp_3dlut_fl_crossbar_bit_slice_32_47; + break; + } if (hubp->funcs->hubp_program_3dlut_fl_crossbar) hubp->funcs->hubp_program_3dlut_fl_crossbar(hubp, @@ -2269,14 +2007,6 @@ void dcn401_program_pipe( pipe_ctx->plane_state->update_flags.bits.hdr_mult)) hws->funcs.set_hdr_multiplier(pipe_ctx); - if (hws->funcs.populate_mcm_luts) { - if (pipe_ctx->plane_state) { - hws->funcs.populate_mcm_luts(dc, pipe_ctx, pipe_ctx->plane_state->mcm_luts, - pipe_ctx->plane_state->lut_bank_a); - pipe_ctx->plane_state->lut_bank_a = !pipe_ctx->plane_state->lut_bank_a; - } - } - if (pipe_ctx->plane_state && (pipe_ctx->plane_state->update_flags.bits.in_transfer_func_change || pipe_ctx->plane_state->update_flags.bits.gamma_change || @@ -2651,7 +2381,7 @@ bool dcn401_update_bandwidth( struct dce_hwseq *hws = dc->hwseq; /* recalculate DML parameters */ - if (dc->res_pool->funcs->validate_bandwidth(dc, context, false) != DC_OK) + if (dc->res_pool->funcs->validate_bandwidth(dc, context, DC_VALIDATE_MODE_AND_PROGRAMMING) != DC_OK) return false; /* apply updated bandwidth parameters */ diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h index ce65b4f6c672..781cf0efccc6 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h @@ -109,12 +109,4 @@ void dcn401_detect_pipe_changes( void dcn401_plane_atomic_power_down(struct dc *dc, struct dpp *dpp, struct hubp *hubp); -bool dcn401_program_rmcm_luts( - struct hubp *hubp, - struct pipe_ctx *pipe_ctx, - enum dc_cm2_transfer_func_source lut3d_src, - struct dc_cm2_func_luts *mcm_luts, - struct mpc *mpc, - bool lut_bank_a, - int mpcc_id); #endif /* __DC_HWSS_DCN401_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/inc/core_status.h b/drivers/gpu/drm/amd/display/dc/inc/core_status.h index f3696143590c..82085d9c3f40 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/core_status.h +++ b/drivers/gpu/drm/amd/display/dc/inc/core_status.h @@ -59,6 +59,7 @@ enum dc_status { DC_FAIL_DP_PAYLOAD_ALLOCATION = 27, DC_FAIL_DP_LINK_BANDWIDTH = 28, DC_FAIL_HW_CURSOR_SUPPORT = 29, + DC_FAIL_DP_TUNNEL_BW_VALIDATE = 30, DC_ERROR_UNEXPECTED = -1 }; diff --git a/drivers/gpu/drm/amd/display/dc/inc/core_types.h b/drivers/gpu/drm/amd/display/dc/inc/core_types.h index 0cf349cafb3e..c9454fe1cd05 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/core_types.h +++ b/drivers/gpu/drm/amd/display/dc/inc/core_types.h @@ -82,7 +82,7 @@ struct resource_funcs { enum dc_status (*validate_bandwidth)( struct dc *dc, struct dc_state *context, - bool fast_validate); + enum dc_validate_mode validate_mode); void (*calculate_wm_and_dlg)( struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, @@ -107,7 +107,7 @@ struct resource_funcs { struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate); + enum dc_validate_mode validate_mode); /* * Algorithm for assigning available link encoders to links. @@ -223,6 +223,11 @@ struct resource_funcs { const struct dc_stream_state *stream); bool (*program_mcache_pipe_config)(struct dc_state *context, const struct dc_mcache_params *mcache_params); + enum dc_status (*update_dc_state_for_encoder_switch)(struct dc_link *link, + struct dc_link_settings *link_setting, + uint8_t pipe_count, + struct pipe_ctx *pipes, + struct audio_output *audio_output); }; struct audio_support{ @@ -556,7 +561,10 @@ struct dcn_bw_output { struct dml2_mcache_surface_allocation mcache_allocations[DML2_MAX_PLANES]; struct dmub_cmd_fams2_global_config fams2_global_config; union dmub_cmd_fams2_config fams2_stream_base_params[DML2_MAX_PLANES]; - union dmub_cmd_fams2_config fams2_stream_sub_params[DML2_MAX_PLANES]; + union { + union dmub_cmd_fams2_config fams2_stream_sub_params[DML2_MAX_PLANES]; + union dmub_fams2_stream_static_sub_state_v2 fams2_stream_sub_params_v2[DML2_MAX_PLANES]; + }; struct dml2_display_arb_regs arb_regs; }; diff --git a/drivers/gpu/drm/amd/display/dc/inc/dcn_calcs.h b/drivers/gpu/drm/amd/display/dc/inc/dcn_calcs.h index d19a595c2be4..134091d5842d 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/dcn_calcs.h +++ b/drivers/gpu/drm/amd/display/dc/inc/dcn_calcs.h @@ -622,7 +622,7 @@ extern const struct dcn_ip_params dcn10_ip_defaults; bool dcn_validate_bandwidth( struct dc *dc, struct dc_state *context, - bool fast_validate); + enum dc_validate_mode validate_mode); void dcn_get_soc_clks( struct dc *dc, diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h b/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h index c14d64687a3d..3b736f4687a6 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr.h @@ -324,6 +324,8 @@ struct clk_mgr_funcs { int (*get_dispclk_from_dentist)(struct clk_mgr *clk_mgr_base); + bool (*is_dc_mode_present)(struct clk_mgr *clk_mgr); + }; struct clk_mgr { diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h index e94e9ba60f55..61c4d2a7db1c 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h @@ -211,7 +211,7 @@ struct dccg_funcs { struct dccg *dccg, enum streamclk_source src, uint32_t otg_inst); - void (*set_dto_dscclk)(struct dccg *dccg, uint32_t dsc_inst); + void (*set_dto_dscclk)(struct dccg *dccg, uint32_t dsc_inst, uint32_t num_slices_h); void (*set_ref_dscclk)(struct dccg *dccg, uint32_t dsc_inst); void (*dccg_root_gate_disable_control)(struct dccg *dccg, uint32_t pipe_idx, uint32_t disable_clock_gating); }; diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h b/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h index 6e303b81bfb0..7641439f6ca0 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h @@ -190,6 +190,42 @@ struct mpc_grph_gamut_adjustment { enum mpcc_gamut_remap_id mpcc_gamut_remap_block_id; }; +struct mpc_rmcm_regs { + uint32_t rmcm_3dlut_mem_pwr_state; + uint32_t rmcm_3dlut_mem_pwr_force; + uint32_t rmcm_3dlut_mem_pwr_dis; + uint32_t rmcm_3dlut_mem_pwr_mode; + uint32_t rmcm_3dlut_size; + uint32_t rmcm_3dlut_mode; + uint32_t rmcm_3dlut_mode_cur; + uint32_t rmcm_3dlut_read_sel; + uint32_t rmcm_3dlut_30bit_en; + uint32_t rmcm_3dlut_wr_en_mask; + uint32_t rmcm_3dlut_ram_sel; + uint32_t rmcm_3dlut_out_norm_factor; + uint32_t rmcm_3dlut_fl_sel; + uint32_t rmcm_3dlut_out_offset_r; + uint32_t rmcm_3dlut_out_scale_r; + uint32_t rmcm_3dlut_fl_done; + uint32_t rmcm_3dlut_fl_soft_underflow; + uint32_t rmcm_3dlut_fl_hard_underflow; + uint32_t rmcm_cntl; + uint32_t rmcm_shaper_mem_pwr_state; + uint32_t rmcm_shaper_mem_pwr_force; + uint32_t rmcm_shaper_mem_pwr_dis; + uint32_t rmcm_shaper_mem_pwr_mode; + uint32_t rmcm_shaper_lut_mode; + uint32_t rmcm_shaper_mode_cur; + uint32_t rmcm_shaper_lut_write_en_mask; + uint32_t rmcm_shaper_lut_write_sel; + uint32_t rmcm_shaper_offset_b; + uint32_t rmcm_shaper_scale_b; + uint32_t rmcm_shaper_rama_exp_region_start_b; + uint32_t rmcm_shaper_rama_exp_region_start_seg_b; + uint32_t rmcm_shaper_rama_exp_region_end_b; + uint32_t rmcm_shaper_rama_exp_region_end_base_b; +}; + struct mpcc_sm_cfg { bool enable; /* 0-single plane,2-row subsampling,4-column subsampling,6-checkboard subsampling */ @@ -301,6 +337,7 @@ struct mpcc_state { uint32_t rgam_mode; uint32_t rgam_lut; struct mpc_grph_gamut_adjustment gamut_remap; + struct mpc_rmcm_regs rmcm_regs; }; /** @@ -1038,6 +1075,11 @@ struct mpc_funcs { */ void (*program_3dlut_size)(struct mpc *mpc, bool is_17x17x17, int mpcc_id); + /** + * @mcm: + * + * MPC MCM new HW sequential programming functions + */ struct { void (*program_3dlut_size)(struct mpc *mpc, uint32_t width, int mpcc_id); void (*program_bias_scale)(struct mpc *mpc, uint16_t bias, uint16_t scale, int mpcc_id); @@ -1050,6 +1092,11 @@ struct mpc_funcs { bool lut_bank_a, int mpcc_id); } mcm; + /** + * @rmcm: + * + * MPC RMCM new HW sequential programming functions + */ struct { void (*enable_3dlut_fl)(struct mpc *mpc, bool enable, int mpcc_id); void (*update_3dlut_fast_load_select)(struct mpc *mpc, int mpcc_id, int hubp_idx); diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/stream_encoder.h b/drivers/gpu/drm/amd/display/dc/inc/hw/stream_encoder.h index fe7f3137f228..27f950ae45ee 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/stream_encoder.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/stream_encoder.h @@ -117,6 +117,7 @@ struct stream_encoder { uint32_t stream_enc_inst; struct vpg *vpg; struct afmt *afmt; + struct apg *apg; }; struct enc_state { diff --git a/drivers/gpu/drm/amd/display/dc/inc/link.h b/drivers/gpu/drm/amd/display/dc/inc/link.h index 7d16351bba99..f2503402c10e 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/link.h +++ b/drivers/gpu/drm/amd/display/dc/inc/link.h @@ -144,9 +144,9 @@ struct link_service { uint32_t (*dp_link_bandwidth_kbps)( const struct dc_link *link, const struct dc_link_settings *link_settings); - bool (*validate_dpia_bandwidth)( - const struct dc_stream_state *stream, - const unsigned int num_streams); + enum dc_status (*validate_dp_tunnel_bandwidth)( + const struct dc *dc, + const struct dc_state *new_ctx); uint32_t (*dp_required_hblank_size_bytes)( const struct dc_link *link, diff --git a/drivers/gpu/drm/amd/display/dc/link/accessories/link_dp_cts.c b/drivers/gpu/drm/amd/display/dc/link/accessories/link_dp_cts.c index 96febabf464a..2956c2b3ad1a 100644 --- a/drivers/gpu/drm/amd/display/dc/link/accessories/link_dp_cts.c +++ b/drivers/gpu/drm/amd/display/dc/link/accessories/link_dp_cts.c @@ -34,6 +34,7 @@ #include "dm_helpers.h" #include "dc_dmub_srv.h" #include "dce/dmub_hw_lock_mgr.h" +#include "clk_mgr.h" #define DC_LOGGER \ link->ctx->logger @@ -67,10 +68,17 @@ static void dp_retrain_link_dp_test(struct dc_link *link, { struct pipe_ctx *pipes[MAX_PIPES]; struct dc_state *state = link->dc->current_state; + struct dc_stream_update stream_update = { 0 }; + bool dpms_off = false; + bool needs_divider_update = false; bool was_hpo_acquired = resource_is_hpo_acquired(link->dc->current_state); bool is_hpo_acquired; uint8_t count; int i; + struct audio_output audio_output[MAX_PIPES]; + + needs_divider_update = (link->dc->link_srv->dp_get_encoding_format(link_setting) != + link->dc->link_srv->dp_get_encoding_format((const struct dc_link_settings *) &link->cur_link_settings)); udelay(100); @@ -83,16 +91,59 @@ static void dp_retrain_link_dp_test(struct dc_link *link, link->dc, state, pipes[i]); + + // Disable OTG and re-enable after updating clocks + pipes[i]->stream_res.tg->funcs->disable_crtc(pipes[i]->stream_res.tg); } - if (link->dc->hwss.setup_hpo_hw_control) { - is_hpo_acquired = resource_is_hpo_acquired(state); - if (was_hpo_acquired != is_hpo_acquired) - link->dc->hwss.setup_hpo_hw_control(link->dc->hwseq, is_hpo_acquired); + if (needs_divider_update && link->dc->res_pool->funcs->update_dc_state_for_encoder_switch) { + link->dc->res_pool->funcs->update_dc_state_for_encoder_switch(link, + link_setting, count, + *pipes, &audio_output[0]); + for (i = 0; i < count; i++) { + pipes[i]->clock_source->funcs->program_pix_clk( + pipes[i]->clock_source, + &pipes[i]->stream_res.pix_clk_params, + link->dc->link_srv->dp_get_encoding_format(&pipes[i]->link_config.dp_link_settings), + &pipes[i]->pll_settings); + + if (pipes[i]->stream_res.audio != NULL) { + const struct link_hwss *link_hwss = get_link_hwss( + link, &pipes[i]->link_res); + + link_hwss->setup_audio_output(pipes[i], &audio_output[i], + pipes[i]->stream_res.audio->inst); + + pipes[i]->stream_res.audio->funcs->az_configure( + pipes[i]->stream_res.audio, + pipes[i]->stream->signal, + &audio_output[i].crtc_info, + &pipes[i]->stream->audio_info, + &audio_output[i].dp_link_info); + + if (link->dc->config.disable_hbr_audio_dp2 && + pipes[i]->stream_res.audio->funcs->az_disable_hbr_audio && + link->dc->link_srv->dp_is_128b_132b_signal(pipes[i])) + pipes[i]->stream_res.audio->funcs->az_disable_hbr_audio(pipes[i]->stream_res.audio); + } + } } - for (i = count-1; i >= 0; i--) - link_set_dpms_on(state, pipes[i]); + // Toggle on HPO I/O if necessary + is_hpo_acquired = resource_is_hpo_acquired(state); + if (was_hpo_acquired != is_hpo_acquired && link->dc->hwss.setup_hpo_hw_control) + link->dc->hwss.setup_hpo_hw_control(link->dc->hwseq, is_hpo_acquired); + + for (i = 0; i < count; i++) + pipes[i]->stream_res.tg->funcs->enable_crtc(pipes[i]->stream_res.tg); + + // Set DPMS on with stream update + for (i = 0; i < state->stream_count; i++) + if (state->streams[i] && state->streams[i]->link && state->streams[i]->link == link) { + stream_update.stream = state->streams[i]; + stream_update.dpms_off = &dpms_off; + dc_update_planes_and_stream(state->clk_mgr->ctx->dc, NULL, 0, state->streams[i], &stream_update); + } } static void dp_test_send_link_training(struct dc_link *link) diff --git a/drivers/gpu/drm/amd/display/dc/link/link_detection.c b/drivers/gpu/drm/amd/display/dc/link/link_detection.c index 9655e6fa53a4..827b630daf49 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_detection.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_detection.c @@ -593,8 +593,9 @@ static bool detect_dp(struct dc_link *link, if (sink_caps->transaction_type == DDC_TRANSACTION_TYPE_I2C_OVER_AUX) { sink_caps->signal = SIGNAL_TYPE_DISPLAY_PORT; - if (!detect_dp_sink_caps(link)) + if (!detect_dp_sink_caps(link)) { return false; + } if (is_dp_branch_device(link)) /* DP SST branch */ diff --git a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c index 273a3be6d593..bd51b279ad14 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c @@ -140,7 +140,8 @@ void link_blank_dp_stream(struct dc_link *link, bool hw_init) } } - if ((!link->wa_flags.dp_keep_receiver_powered) || hw_init) + if (((!link->wa_flags.dp_keep_receiver_powered) || hw_init) && + (link->type != dc_connection_none)) dpcd_write_rx_power_ctrl(link, false); } } @@ -842,14 +843,14 @@ void link_set_dsc_on_stream(struct pipe_ctx *pipe_ctx, bool enable) dsc_cfg.dc_dsc_cfg.num_slices_h /= opp_cnt; if (should_use_dto_dscclk) - dccg->funcs->set_dto_dscclk(dccg, dsc->inst); + dccg->funcs->set_dto_dscclk(dccg, dsc->inst, dsc_cfg.dc_dsc_cfg.num_slices_h); dsc->funcs->dsc_set_config(dsc, &dsc_cfg, &dsc_optc_cfg); dsc->funcs->dsc_enable(dsc, pipe_ctx->stream_res.opp->inst); for (odm_pipe = pipe_ctx->next_odm_pipe; odm_pipe; odm_pipe = odm_pipe->next_odm_pipe) { struct display_stream_compressor *odm_dsc = odm_pipe->stream_res.dsc; if (should_use_dto_dscclk) - dccg->funcs->set_dto_dscclk(dccg, odm_dsc->inst); + dccg->funcs->set_dto_dscclk(dccg, odm_dsc->inst, dsc_cfg.dc_dsc_cfg.num_slices_h); odm_dsc->funcs->dsc_set_config(odm_dsc, &dsc_cfg, &dsc_optc_cfg); odm_dsc->funcs->dsc_enable(odm_dsc, odm_pipe->stream_res.opp->inst); } @@ -2296,8 +2297,7 @@ static bool allocate_usb4_bandwidth_for_stream(struct dc_stream_state *stream, i link->dpia_bw_alloc_config.remote_sink_req_bw[sink_index] = bw; } - /* get dp overhead for dp tunneling */ - link->dpia_bw_alloc_config.dp_overhead = link_dp_dpia_get_dp_overhead_in_dp_tunneling(link); + link->dpia_bw_alloc_config.dp_overhead = link_dpia_get_dp_overhead(link); req_bw += link->dpia_bw_alloc_config.dp_overhead; link_dp_dpia_allocate_usb4_bandwidth_for_stream(link, req_bw); diff --git a/drivers/gpu/drm/amd/display/dc/link/link_factory.c b/drivers/gpu/drm/amd/display/dc/link/link_factory.c index 1a04f4b74585..c5f4e803be84 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_factory.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_factory.c @@ -100,7 +100,7 @@ static void construct_link_service_validation(struct link_service *link_srv) { link_srv->validate_mode_timing = link_validate_mode_timing; link_srv->dp_link_bandwidth_kbps = dp_link_bandwidth_kbps; - link_srv->validate_dpia_bandwidth = link_validate_dpia_bandwidth; + link_srv->validate_dp_tunnel_bandwidth = link_validate_dp_tunnel_bandwidth; link_srv->dp_required_hblank_size_bytes = dp_required_hblank_size_bytes; } diff --git a/drivers/gpu/drm/amd/display/dc/link/link_validation.c b/drivers/gpu/drm/amd/display/dc/link/link_validation.c index 29606fda029d..aecaf37eee35 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_validation.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_validation.c @@ -86,6 +86,10 @@ static bool dp_active_dongle_validate_timing( if (!dongle_caps->is_dp_hdmi_ycbcr420_pass_through) return false; break; + case PIXEL_ENCODING_UNDEFINED: + /* These color depths are currently not supported */ + ASSERT(false); + break; default: /* Invalid Pixel Encoding*/ return false; @@ -104,6 +108,10 @@ static bool dp_active_dongle_validate_timing( if (dongle_caps->dp_hdmi_max_bpc < 12) return false; break; + case COLOR_DEPTH_UNDEFINED: + /* These color depths are currently not supported */ + ASSERT(false); + break; case COLOR_DEPTH_141414: case COLOR_DEPTH_161616: default: @@ -255,6 +263,14 @@ uint32_t dp_link_bandwidth_kbps( return link_rate_per_lane_kbps * link_settings->lane_count / 10000 * total_data_bw_efficiency_x10000; } +static uint32_t dp_get_timing_bandwidth_kbps( + const struct dc_crtc_timing *timing, + const struct dc_link *link) +{ + return dc_bandwidth_in_kbps_from_timing(timing, + dc_link_get_highest_encoding_format(link)); +} + static bool dp_validate_mode_timing( struct dc_link *link, const struct dc_crtc_timing *timing) @@ -351,63 +367,81 @@ enum dc_status link_validate_mode_timing( return DC_OK; } +static const struct dc_tunnel_settings *get_dp_tunnel_settings(const struct dc_state *context, + const struct dc_stream_state *stream) +{ + int i; + const struct dc_tunnel_settings *dp_tunnel_settings = NULL; + + for (i = 0; i < MAX_PIPES; i++) { + if (context->res_ctx.pipe_ctx[i].stream && (context->res_ctx.pipe_ctx[i].stream == stream)) { + dp_tunnel_settings = &context->res_ctx.pipe_ctx[i].link_config.dp_tunnel_settings; + break; + } + } + + return dp_tunnel_settings; +} + /* - * This function calculates the bandwidth required for the stream timing - * and aggregates the stream bandwidth for the respective dpia link - * - * @stream: pointer to the dc_stream_state struct instance - * @num_streams: number of streams to be validated + * Calculates the DP tunneling bandwidth required for the stream timing + * and aggregates the stream bandwidth for the respective DP tunneling link * - * return: true if validation is succeeded + * return: dc_status */ -bool link_validate_dpia_bandwidth(const struct dc_stream_state *stream, const unsigned int num_streams) +enum dc_status link_validate_dp_tunnel_bandwidth(const struct dc *dc, const struct dc_state *new_ctx) { - int bw_needed[MAX_DPIA_NUM] = {0}; - struct dc_link *dpia_link[MAX_DPIA_NUM] = {0}; - int num_dpias = 0; + struct dc_validation_dpia_set dpia_link_sets[MAX_DPIA_NUM] = { 0 }; + uint8_t link_count = 0; + enum dc_status result = DC_OK; - for (unsigned int i = 0; i < num_streams; ++i) { - if (stream[i].signal == SIGNAL_TYPE_DISPLAY_PORT) { - /* new dpia sst stream, check whether it exceeds max dpia */ - if (num_dpias >= MAX_DPIA_NUM) - return false; + // Iterate through streams in the new context + for (uint8_t i = 0; (i < MAX_PIPES && i < new_ctx->stream_count); i++) { + const struct dc_stream_state *stream = new_ctx->streams[i]; + const struct dc_link *link; + const struct dc_tunnel_settings *dp_tunnel_settings; + uint32_t timing_bw; - dpia_link[num_dpias] = stream[i].link; - bw_needed[num_dpias] = dc_bandwidth_in_kbps_from_timing(&stream[i].timing, - dc_link_get_highest_encoding_format(dpia_link[num_dpias])); - num_dpias++; - } else if (stream[i].signal == SIGNAL_TYPE_DISPLAY_PORT_MST) { - uint8_t j = 0; - /* check whether its a known dpia link */ - for (; j < num_dpias; ++j) { - if (dpia_link[j] == stream[i].link) - break; - } + if (stream == NULL) + continue; + + link = stream->link; + + if (!(link && (stream->signal == SIGNAL_TYPE_DISPLAY_PORT + || stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST) + && link->hpd_status)) + continue; + + dp_tunnel_settings = get_dp_tunnel_settings(new_ctx, stream); - if (j == num_dpias) { - /* new dpia mst stream, check whether it exceeds max dpia */ - if (num_dpias >= MAX_DPIA_NUM) - return false; - else { - dpia_link[j] = stream[i].link; - num_dpias++; - } + if ((dp_tunnel_settings == NULL) || (dp_tunnel_settings->should_use_dp_bw_allocation == false)) + continue; + + timing_bw = dp_get_timing_bandwidth_kbps(&stream->timing, link); + + // Find an existing entry for this 'link' in 'dpia_link_sets' + for (uint8_t j = 0; j < MAX_DPIA_NUM; j++) { + bool is_new_slot = false; + + if (dpia_link_sets[j].link == NULL) { + is_new_slot = true; + link_count++; + dpia_link_sets[j].required_bw = 0; + dpia_link_sets[j].link = link; } - bw_needed[j] += dc_bandwidth_in_kbps_from_timing(&stream[i].timing, - dc_link_get_highest_encoding_format(dpia_link[j])); + if (is_new_slot || (dpia_link_sets[j].link == link)) { + dpia_link_sets[j].tunnel_settings = dp_tunnel_settings; + dpia_link_sets[j].required_bw += timing_bw; + break; + } } } - /* Include dp overheads */ - for (uint8_t i = 0; i < num_dpias; ++i) { - int dp_overhead = 0; - - dp_overhead = link_dp_dpia_get_dp_overhead_in_dp_tunneling(dpia_link[i]); - bw_needed[i] += dp_overhead; - } + if (link_count && link_dpia_validate_dp_tunnel_bandwidth(dpia_link_sets, link_count) == false) + result = DC_FAIL_DP_TUNNEL_BW_VALIDATE; - return dpia_validate_usb4_bw(dpia_link, bw_needed, num_dpias); + return result; } struct dp_audio_layout_config { diff --git a/drivers/gpu/drm/amd/display/dc/link/link_validation.h b/drivers/gpu/drm/amd/display/dc/link/link_validation.h index bf398c49c3e8..9553c81053fe 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_validation.h +++ b/drivers/gpu/drm/amd/display/dc/link/link_validation.h @@ -30,9 +30,9 @@ enum dc_status link_validate_mode_timing( const struct dc_stream_state *stream, struct dc_link *link, const struct dc_crtc_timing *timing); -bool link_validate_dpia_bandwidth( - const struct dc_stream_state *stream, - const unsigned int num_streams); +enum dc_status link_validate_dp_tunnel_bandwidth( + const struct dc *dc, + const struct dc_state *new_ctx); uint32_t dp_link_bandwidth_kbps( const struct dc_link *link, const struct dc_link_settings *link_settings); diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c index a5127c2d47ef..0f965380a9b4 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c @@ -385,9 +385,15 @@ bool dp_is_128b_132b_signal(struct pipe_ctx *pipe_ctx) bool dp_is_lttpr_present(struct dc_link *link) { /* Some sink devices report invalid LTTPR revision, so don't validate against that cap */ - return (dp_parse_lttpr_repeater_count(link->dpcd_caps.lttpr_caps.phy_repeater_cnt) != 0 && + uint32_t lttpr_count = dp_parse_lttpr_repeater_count(link->dpcd_caps.lttpr_caps.phy_repeater_cnt); + bool is_lttpr_present = (lttpr_count > 0 && link->dpcd_caps.lttpr_caps.max_lane_count > 0 && link->dpcd_caps.lttpr_caps.max_lane_count <= 4); + + if (lttpr_count > 0 && !is_lttpr_present) + DC_LOG_ERROR("LTTPR count is nonzero but invalid lane count reported. Assuming no LTTPR present.\n"); + + return is_lttpr_present; } /* in DP compliance test, DPR-120 may have @@ -1551,6 +1557,8 @@ enum dc_status dp_retrieve_lttpr_cap(struct dc_link *link) uint8_t lttpr_dpcd_data[10] = {0}; enum dc_status status; bool is_lttpr_present; + uint32_t lttpr_count; + uint32_t closest_lttpr_offset; /* Logic to determine LTTPR support*/ bool vbios_lttpr_interop = link->dc->caps.vbios_lttpr_aware; @@ -1602,20 +1610,22 @@ enum dc_status dp_retrieve_lttpr_cap(struct dc_link *link) lttpr_dpcd_data[DP_LTTPR_ALPM_CAPABILITIES - DP_LT_TUNABLE_PHY_REPEATER_FIELD_DATA_STRUCTURE_REV]; + lttpr_count = dp_parse_lttpr_repeater_count(link->dpcd_caps.lttpr_caps.phy_repeater_cnt); + /* If this chip cap is set, at least one retimer must exist in the chain * Override count to 1 if we receive a known bad count (0 or an invalid value) */ if (((link->chip_caps & AMD_EXT_DISPLAY_PATH_CAPS__EXT_CHIP_MASK) == AMD_EXT_DISPLAY_PATH_CAPS__DP_FIXED_VS_EN) && - (dp_parse_lttpr_repeater_count(link->dpcd_caps.lttpr_caps.phy_repeater_cnt) == 0)) { + lttpr_count == 0) { /* If you see this message consistently, either the host platform has FIXED_VS flag * incorrectly configured or the sink device is returning an invalid count. */ DC_LOG_ERROR("lttpr_caps phy_repeater_cnt is 0x%x, forcing it to 0x80.", link->dpcd_caps.lttpr_caps.phy_repeater_cnt); link->dpcd_caps.lttpr_caps.phy_repeater_cnt = 0x80; + lttpr_count = 1; DC_LOG_DC("lttpr_caps forced phy_repeater_cnt = %d\n", link->dpcd_caps.lttpr_caps.phy_repeater_cnt); } - /* Attempt to train in LTTPR transparent mode if repeater count exceeds 8. */ is_lttpr_present = dp_is_lttpr_present(link); DC_LOG_DC("is_lttpr_present = %d\n", is_lttpr_present); @@ -1623,11 +1633,25 @@ enum dc_status dp_retrieve_lttpr_cap(struct dc_link *link) if (is_lttpr_present) { CONN_DATA_DETECT(link, lttpr_dpcd_data, sizeof(lttpr_dpcd_data), "LTTPR Caps: "); - core_link_read_dpcd(link, DP_LTTPR_IEEE_OUI, link->dpcd_caps.lttpr_caps.lttpr_ieee_oui, sizeof(link->dpcd_caps.lttpr_caps.lttpr_ieee_oui)); - CONN_DATA_DETECT(link, link->dpcd_caps.lttpr_caps.lttpr_ieee_oui, sizeof(link->dpcd_caps.lttpr_caps.lttpr_ieee_oui), "LTTPR IEEE OUI: "); + // Identify closest LTTPR to determine if workarounds required for known embedded LTTPR + closest_lttpr_offset = dp_get_closest_lttpr_offset(lttpr_count); - core_link_read_dpcd(link, DP_LTTPR_DEVICE_ID, link->dpcd_caps.lttpr_caps.lttpr_device_id, sizeof(link->dpcd_caps.lttpr_caps.lttpr_device_id)); - CONN_DATA_DETECT(link, link->dpcd_caps.lttpr_caps.lttpr_device_id, sizeof(link->dpcd_caps.lttpr_caps.lttpr_device_id), "LTTPR Device ID: "); + core_link_read_dpcd(link, (DP_LTTPR_IEEE_OUI + closest_lttpr_offset), + link->dpcd_caps.lttpr_caps.lttpr_ieee_oui, sizeof(link->dpcd_caps.lttpr_caps.lttpr_ieee_oui)); + core_link_read_dpcd(link, (DP_LTTPR_DEVICE_ID + closest_lttpr_offset), + link->dpcd_caps.lttpr_caps.lttpr_device_id, sizeof(link->dpcd_caps.lttpr_caps.lttpr_device_id)); + + if (lttpr_count > 1) { + CONN_DATA_DETECT(link, link->dpcd_caps.lttpr_caps.lttpr_ieee_oui, sizeof(link->dpcd_caps.lttpr_caps.lttpr_ieee_oui), + "Closest LTTPR To Host's IEEE OUI: "); + CONN_DATA_DETECT(link, link->dpcd_caps.lttpr_caps.lttpr_device_id, sizeof(link->dpcd_caps.lttpr_caps.lttpr_device_id), + "Closest LTTPR To Host's LTTPR Device ID: "); + } else { + CONN_DATA_DETECT(link, link->dpcd_caps.lttpr_caps.lttpr_ieee_oui, sizeof(link->dpcd_caps.lttpr_caps.lttpr_ieee_oui), + "LTTPR IEEE OUI: "); + CONN_DATA_DETECT(link, link->dpcd_caps.lttpr_caps.lttpr_device_id, sizeof(link->dpcd_caps.lttpr_caps.lttpr_device_id), + "LTTPR Device ID: "); + } } return status; diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia.c index 22bfdced64ab..9b2f1a7da1d1 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia.c @@ -75,12 +75,15 @@ enum dc_status dpcd_get_tunneling_device_data(struct dc_link *link) if (link->dpcd_caps.usb4_dp_tun_info.dp_tun_cap.bits.dpia_bw_alloc) { status = core_link_read_dpcd(link, USB4_DRIVER_BW_CAPABILITY, - dpcd_dp_tun_data, 1); + dpcd_dp_tun_data, 2); if (status != DC_OK) goto err; - link->dpcd_caps.usb4_dp_tun_info.driver_bw_cap.raw = dpcd_dp_tun_data[0]; + link->dpcd_caps.usb4_dp_tun_info.driver_bw_cap.raw = + dpcd_dp_tun_data[USB4_DRIVER_BW_CAPABILITY - USB4_DRIVER_BW_CAPABILITY]; + link->dpcd_caps.usb4_dp_tun_info.dpia_tunnel_info.raw = + dpcd_dp_tun_data[DP_IN_ADAPTER_TUNNEL_INFO - USB4_DRIVER_BW_CAPABILITY]; } DC_LOG_DEBUG("%s: Link[%d] DP tunneling support (RouterId=%d AdapterId=%d) " @@ -155,8 +158,14 @@ void link_decide_dp_tunnel_settings(struct dc_stream_state *stream, link->dpcd_caps.usb4_dp_tun_info.dp_tun_cap.bits.dp_tunneling; if (link->dpcd_caps.usb4_dp_tun_info.dp_tun_cap.bits.dpia_bw_alloc - && link->dpcd_caps.usb4_dp_tun_info.driver_bw_cap.bits.driver_bw_alloc_support) + && link->dpcd_caps.usb4_dp_tun_info.driver_bw_cap.bits.driver_bw_alloc_support) { dp_tunnel_setting->should_use_dp_bw_allocation = true; + dp_tunnel_setting->cm_id = link->dpcd_caps.usb4_dp_tun_info.usb4_driver_id & 0x0F; + dp_tunnel_setting->group_id = link->dpcd_caps.usb4_dp_tun_info.dpia_tunnel_info.bits.group_id; + dp_tunnel_setting->estimated_bw = link->dpia_bw_alloc_config.estimated_bw; + dp_tunnel_setting->allocated_bw = link->dpia_bw_alloc_config.allocated_bw; + dp_tunnel_setting->bw_granularity = link->dpia_bw_alloc_config.bw_granularity; + } } } diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c index 3af7564a84f1..819bf2d8ba53 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c @@ -35,6 +35,8 @@ #define Kbps_TO_Gbps (1000 * 1000) +#define MST_TIME_SLOT_COUNT 64 + // ------------------------------------------------------------------ // PRIVATE FUNCTIONS // ------------------------------------------------------------------ @@ -160,78 +162,6 @@ static void retrieve_usb4_dp_bw_allocation_info(struct dc_link *link) link->dpia_bw_alloc_config.nrd_max_lane_count); } -static uint8_t get_lowest_dpia_index(struct dc_link *link) -{ - const struct dc *dc_struct = link->dc; - uint8_t idx = 0xFF; - int i; - - for (i = 0; i < MAX_LINKS; ++i) { - - if (!dc_struct->links[i] || - dc_struct->links[i]->ep_type != DISPLAY_ENDPOINT_USB4_DPIA) - continue; - - if (idx > dc_struct->links[i]->link_index) { - idx = dc_struct->links[i]->link_index; - break; - } - } - - return idx; -} - -/* - * Get the maximum dp tunnel banwidth of host router - * - * @dc: pointer to the dc struct instance - * @hr_index: host router index - * - * return: host router maximum dp tunnel bandwidth - */ -static int get_host_router_total_dp_tunnel_bw(const struct dc *dc, uint8_t hr_index) -{ - uint8_t lowest_dpia_index = get_lowest_dpia_index(dc->links[0]); - uint8_t hr_index_temp = 0; - struct dc_link *link_dpia_primary, *link_dpia_secondary; - int total_bw = 0; - - for (uint8_t i = 0; i < MAX_LINKS - 1; ++i) { - - if (!dc->links[i] || dc->links[i]->ep_type != DISPLAY_ENDPOINT_USB4_DPIA) - continue; - - hr_index_temp = (dc->links[i]->link_index - lowest_dpia_index) / 2; - - if (hr_index_temp == hr_index) { - link_dpia_primary = dc->links[i]; - link_dpia_secondary = dc->links[i + 1]; - - /** - * If BW allocation enabled on both DPIAs, then - * HR BW = Estimated(dpia_primary) + Allocated(dpia_secondary) - * otherwise HR BW = Estimated(bw alloc enabled dpia) - */ - if ((link_dpia_primary->hpd_status && - link_dpia_primary->dpia_bw_alloc_config.bw_alloc_enabled) && - (link_dpia_secondary->hpd_status && - link_dpia_secondary->dpia_bw_alloc_config.bw_alloc_enabled)) { - total_bw += link_dpia_primary->dpia_bw_alloc_config.estimated_bw + - link_dpia_secondary->dpia_bw_alloc_config.allocated_bw; - } else if (link_dpia_primary->hpd_status && - link_dpia_primary->dpia_bw_alloc_config.bw_alloc_enabled) { - total_bw = link_dpia_primary->dpia_bw_alloc_config.estimated_bw; - } else if (link_dpia_secondary->hpd_status && - link_dpia_secondary->dpia_bw_alloc_config.bw_alloc_enabled) { - total_bw += link_dpia_secondary->dpia_bw_alloc_config.estimated_bw; - } - break; - } - } - - return total_bw; -} - /* * Cleanup function for when the dpia is unplugged to reset struct * and perform any required clean up @@ -251,32 +181,40 @@ static void dpia_bw_alloc_unplug(struct dc_link *link) static void link_dpia_send_bw_alloc_request(struct dc_link *link, int req_bw) { - uint8_t requested_bw; - uint32_t temp; + uint8_t request_reg_val; + uint32_t temp, request_bw; - /* Error check whether request bw greater than allocated */ - if (req_bw > link->dpia_bw_alloc_config.estimated_bw) { - DC_LOG_ERROR("%s: Request BW greater than estimated BW for link(%d)\n", - __func__, link->link_index); - req_bw = link->dpia_bw_alloc_config.estimated_bw; + if (link->dpia_bw_alloc_config.bw_granularity == 0) { + DC_LOG_ERROR("%s: Link[%d]: bw_granularity is zero!", __func__, link->link_index); + return; } temp = req_bw * link->dpia_bw_alloc_config.bw_granularity; - requested_bw = temp / Kbps_TO_Gbps; + request_reg_val = temp / Kbps_TO_Gbps; /* Always make sure to add more to account for floating points */ if (temp % Kbps_TO_Gbps) - ++requested_bw; + ++request_reg_val; - /* Error check whether requested and allocated are equal */ - req_bw = requested_bw * (Kbps_TO_Gbps / link->dpia_bw_alloc_config.bw_granularity); - if (req_bw && (req_bw == link->dpia_bw_alloc_config.allocated_bw)) { - DC_LOG_ERROR("%s: Request BW equals to allocated BW for link(%d)\n", - __func__, link->link_index); + request_bw = request_reg_val * (Kbps_TO_Gbps / link->dpia_bw_alloc_config.bw_granularity); + + if (request_bw > link->dpia_bw_alloc_config.estimated_bw) { + DC_LOG_ERROR("%s: Link[%d]: Request BW (%d --> %d) > Estimated BW (%d)... Set to Estimated BW!", + __func__, link->link_index, + req_bw, request_bw, link->dpia_bw_alloc_config.estimated_bw); + req_bw = link->dpia_bw_alloc_config.estimated_bw; + + temp = req_bw * link->dpia_bw_alloc_config.bw_granularity; + request_reg_val = temp / Kbps_TO_Gbps; + if (temp % Kbps_TO_Gbps) + ++request_reg_val; } + link->dpia_bw_alloc_config.allocated_bw = request_bw; + DC_LOG_DC("%s: Link[%d]: Request BW: %d", __func__, link->link_index, request_bw); + core_link_write_dpcd(link, REQUESTED_BW, - &requested_bw, + &request_reg_val, sizeof(uint8_t)); } @@ -304,14 +242,16 @@ bool link_dpia_enable_usb4_dp_bw_alloc_mode(struct dc_link *link) link->dpia_bw_alloc_config.bw_alloc_enabled = true; ret = true; - /* - * During DP tunnel creation, CM preallocates BW and reduces estimated BW of other - * DPIA. CM release preallocation only when allocation is complete. Do zero alloc - * to make the CM to release preallocation and update estimated BW correctly for - * all DPIAs per host router - */ - // TODO: Zero allocation can be removed once the MSFT CM fix has been released - link_dp_dpia_allocate_usb4_bandwidth_for_stream(link, 0); + if (link->dc->debug.dpia_debug.bits.enable_usb4_bw_zero_alloc_patch) { + /* + * During DP tunnel creation, the CM preallocates BW + * and reduces the estimated BW of other DPIAs. + * The CM releases the preallocation only when the allocation is complete. + * Perform a zero allocation to make the CM release the preallocation + * and correctly update the estimated BW for all DPIAs per host router. + */ + link_dp_dpia_allocate_usb4_bandwidth_for_stream(link, 0); + } } else DC_LOG_DEBUG("%s: link[%d] failed to enable DPTX BW allocation mode", __func__, link->link_index); } @@ -329,19 +269,17 @@ bool link_dpia_enable_usb4_dp_bw_alloc_mode(struct dc_link *link) */ void link_dp_dpia_handle_bw_alloc_status(struct dc_link *link, uint8_t status) { + link->dpia_bw_alloc_config.estimated_bw = get_estimated_bw(link); + if (status & DP_TUNNELING_BW_REQUEST_SUCCEEDED) { DC_LOG_DEBUG("%s: BW Allocation request succeeded on link(%d)", __func__, link->link_index); } else if (status & DP_TUNNELING_BW_REQUEST_FAILED) { - link->dpia_bw_alloc_config.estimated_bw = get_estimated_bw(link); - DC_LOG_DEBUG("%s: BW Allocation request failed on link(%d) allocated/estimated BW=%d", __func__, link->link_index, link->dpia_bw_alloc_config.estimated_bw); link_dpia_send_bw_alloc_request(link, link->dpia_bw_alloc_config.estimated_bw); } else if (status & DP_TUNNELING_ESTIMATED_BW_CHANGED) { - link->dpia_bw_alloc_config.estimated_bw = get_estimated_bw(link); - DC_LOG_DEBUG("%s: Estimated BW changed on link(%d) new estimated BW=%d", __func__, link->link_index, link->dpia_bw_alloc_config.estimated_bw); } @@ -374,9 +312,13 @@ void dpia_handle_usb4_bandwidth_allocation_for_link(struct dc_link *link, int pe void link_dp_dpia_allocate_usb4_bandwidth_for_stream(struct dc_link *link, int req_bw) { - DC_LOG_DEBUG("%s: ENTER: link(%d), hpd_status(%d), current allocated_bw(%d), req_bw(%d)\n", + link->dpia_bw_alloc_config.estimated_bw = get_estimated_bw(link); + + DC_LOG_DEBUG("%s: ENTER: link[%d] hpd(%d) Allocated_BW: %d Estimated_BW: %d Req_BW: %d", __func__, link->link_index, link->hpd_status, - link->dpia_bw_alloc_config.allocated_bw, req_bw); + link->dpia_bw_alloc_config.allocated_bw, + link->dpia_bw_alloc_config.estimated_bw, + req_bw); if (link_dp_is_bw_alloc_available(link)) link_dpia_send_bw_alloc_request(link, req_bw); @@ -384,73 +326,116 @@ void link_dp_dpia_allocate_usb4_bandwidth_for_stream(struct dc_link *link, int r DC_LOG_DEBUG("%s: BW Allocation mode not available", __func__); } -bool dpia_validate_usb4_bw(struct dc_link **link, int *bw_needed_per_dpia, const unsigned int num_dpias) +uint32_t link_dpia_get_dp_overhead(const struct dc_link *link) { - bool ret = true; - int bw_needed_per_hr[MAX_HR_NUM] = { 0, 0 }, host_router_total_dp_bw = 0; - uint8_t lowest_dpia_index, i, hr_index; + uint32_t link_dp_overhead = 0; + + if ((link->type == dc_connection_mst_branch) && + !link->dpcd_caps.channel_coding_cap.bits.DP_128b_132b_SUPPORTED) { + /* For 8b/10b encoding: MTP is 64 time slots long, slot 0 is used for MTPH + * MST overhead is 1/64 of link bandwidth (excluding any overhead) + */ + const struct dc_link_settings *link_cap = dc_link_get_link_cap(link); + + if (link_cap) { + uint32_t link_bw_in_kbps = (uint32_t)link_cap->link_rate * + (uint32_t)link_cap->lane_count * + LINK_RATE_REF_FREQ_IN_KHZ * 8; + link_dp_overhead = (link_bw_in_kbps / MST_TIME_SLOT_COUNT) + + ((link_bw_in_kbps % MST_TIME_SLOT_COUNT) ? 1 : 0); + } + } - if (!num_dpias || num_dpias > MAX_DPIA_NUM) - return ret; + return link_dp_overhead; +} - lowest_dpia_index = get_lowest_dpia_index(link[0]); +/* + * Aggregates the DPIA bandwidth usage for the respective USB4 Router. + * And then validate if the required bandwidth is within the router's capacity. + * + * @dc_validation_dpia_set: pointer to the dc_validation_dpia_set + * @count: number of DPIA validation sets + * + * return: true if validation is succeeded + */ +bool link_dpia_validate_dp_tunnel_bandwidth(const struct dc_validation_dpia_set *dpia_link_sets, uint8_t count) +{ + uint32_t granularity_Gbps; + const struct dc_link *link; + uint32_t link_bw_granularity; + uint32_t link_required_bw; + struct usb4_router_validation_set router_sets[MAX_HOST_ROUTERS_NUM] = { 0 }; + uint8_t i; + bool is_success = true; + uint8_t router_count = 0; - /* get total Host Router BW with granularity for the given modes */ - for (i = 0; i < num_dpias; ++i) { - int granularity_Gbps = 0; - int bw_granularity = 0; + if ((dpia_link_sets == NULL) || (count == 0)) + return is_success; - if (!link[i]->dpia_bw_alloc_config.bw_alloc_enabled) - continue; + // Iterate through each DP tunneling link (DPIA). + // Aggregate its bandwidth requirements onto the respective USB4 router. + for (i = 0; i < count; i++) { + link = dpia_link_sets[i].link; + link_required_bw = dpia_link_sets[i].required_bw; + const struct dc_tunnel_settings *dp_tunnel_settings = dpia_link_sets[i].tunnel_settings; - if (link[i]->link_index < lowest_dpia_index) - continue; + if ((link == NULL) || (dp_tunnel_settings == NULL) || dp_tunnel_settings->bw_granularity == 0) + break; - granularity_Gbps = (Kbps_TO_Gbps / link[i]->dpia_bw_alloc_config.bw_granularity); - bw_granularity = (bw_needed_per_dpia[i] / granularity_Gbps) * granularity_Gbps + - ((bw_needed_per_dpia[i] % granularity_Gbps) ? granularity_Gbps : 0); + if (link->type == dc_connection_mst_branch) + link_required_bw += link_dpia_get_dp_overhead(link); - hr_index = (link[i]->link_index - lowest_dpia_index) / 2; - bw_needed_per_hr[hr_index] += bw_granularity; - } + granularity_Gbps = (Kbps_TO_Gbps / dp_tunnel_settings->bw_granularity); + link_bw_granularity = (link_required_bw / granularity_Gbps) * granularity_Gbps + + ((link_required_bw % granularity_Gbps) ? granularity_Gbps : 0); - /* validate against each Host Router max BW */ - for (hr_index = 0; hr_index < MAX_HR_NUM; ++hr_index) { - if (bw_needed_per_hr[hr_index]) { - host_router_total_dp_bw = get_host_router_total_dp_tunnel_bw(link[0]->dc, hr_index); - if (bw_needed_per_hr[hr_index] > host_router_total_dp_bw) { - ret = false; + // Find or add the USB4 router associated with the current DPIA link + for (uint8_t j = 0; j < MAX_HOST_ROUTERS_NUM; j++) { + if (router_sets[j].is_valid == false) { + router_sets[j].is_valid = true; + router_sets[j].cm_id = dp_tunnel_settings->cm_id; + router_count++; + } + + if (router_sets[j].cm_id == dp_tunnel_settings->cm_id) { + uint32_t remaining_bw = + dp_tunnel_settings->estimated_bw - dp_tunnel_settings->allocated_bw; + + router_sets[j].allocated_bw += dp_tunnel_settings->allocated_bw; + + if (remaining_bw > router_sets[j].remaining_bw) + router_sets[j].remaining_bw = remaining_bw; + + // Get the max estimated BW within the same CM_ID + if (dp_tunnel_settings->estimated_bw > router_sets[j].estimated_bw) + router_sets[j].estimated_bw = dp_tunnel_settings->estimated_bw; + + router_sets[j].required_bw += link_bw_granularity; + router_sets[j].dpia_count++; break; } } } - return ret; -} + // Validate bandwidth for each unique router found. + for (i = 0; i < router_count; i++) { + uint32_t total_bw = 0; -int link_dp_dpia_get_dp_overhead_in_dp_tunneling(struct dc_link *link) -{ - int dp_overhead = 0, link_mst_overhead = 0; + if (router_sets[i].is_valid == false) + break; - if (!link_dp_is_bw_alloc_available(link)) - return dp_overhead; + // Determine the total available bandwidth for the current router based on aggregated data + if ((router_sets[i].dpia_count == 1) || (router_sets[i].allocated_bw == 0)) + total_bw = router_sets[i].estimated_bw; + else + total_bw = router_sets[i].allocated_bw + router_sets[i].remaining_bw; - /* if its mst link, add MTPH overhead */ - if ((link->type == dc_connection_mst_branch) && - !link->dpcd_caps.channel_coding_cap.bits.DP_128b_132b_SUPPORTED) { - /* For 8b/10b encoding: MTP is 64 time slots long, slot 0 is used for MTPH - * MST overhead is 1/64 of link bandwidth (excluding any overhead) - */ - const struct dc_link_settings *link_cap = - dc_link_get_link_cap(link); - uint32_t link_bw_in_kbps = (uint32_t)link_cap->link_rate * - (uint32_t)link_cap->lane_count * - LINK_RATE_REF_FREQ_IN_KHZ * 8; - link_mst_overhead = (link_bw_in_kbps / 64) + ((link_bw_in_kbps % 64) ? 1 : 0); + if (router_sets[i].required_bw > total_bw) { + is_success = false; + break; + } } - /* add all the overheads */ - dp_overhead = link_mst_overhead; - - return dp_overhead; + return is_success; } + diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.h b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.h index 801965b5f9a4..41efcb3e44e2 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.h +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.h @@ -28,10 +28,6 @@ #include "link.h" -/* Number of Host Routers per motherboard is 2 */ -#define MAX_HR_NUM 2 -/* Number of DPIA per host router is 2 */ -#define MAX_DPIA_NUM (MAX_HR_NUM * 2) /* * Host Router BW type @@ -42,6 +38,16 @@ enum bw_type { HOST_ROUTER_BW_INVALID, }; +struct usb4_router_validation_set { + bool is_valid; + uint8_t cm_id; + uint8_t dpia_count; + uint32_t required_bw; + uint32_t allocated_bw; + uint32_t estimated_bw; + uint32_t remaining_bw; +}; + /* * Enable USB4 DP BW allocation mode * @@ -74,25 +80,13 @@ void link_dp_dpia_allocate_usb4_bandwidth_for_stream(struct dc_link *link, int r void dpia_handle_usb4_bandwidth_allocation_for_link(struct dc_link *link, int peak_bw); /* - * Handle the validation of total BW here and confirm that the bw used by each - * DPIA doesn't exceed available BW for each host router (HR) - * - * @link[]: array of link pointer to all possible DPIA links - * @bw_needed[]: bw needed for each DPIA link based on timing - * @num_dpias: Number of DPIAs for the above 2 arrays. Should always be <= MAX_DPIA_NUM - * - * return: TRUE if bw used by DPIAs doesn't exceed available BW else return FALSE - */ -bool dpia_validate_usb4_bw(struct dc_link **link, int *bw_needed, const unsigned int num_dpias); - -/* * Obtain all the DP overheads in dp tunneling for the dpia link * * @link: pointer to the dc_link struct instance * * return: DP overheads in DP tunneling */ -int link_dp_dpia_get_dp_overhead_in_dp_tunneling(struct dc_link *link); +uint32_t link_dpia_get_dp_overhead(const struct dc_link *link); /* * Handle DP BW allocation status register @@ -104,4 +98,15 @@ int link_dp_dpia_get_dp_overhead_in_dp_tunneling(struct dc_link *link); */ void link_dp_dpia_handle_bw_alloc_status(struct dc_link *link, uint8_t status); +/* + * Aggregates the DPIA bandwidth usage for the respective USB4 Router. + * + * @dc_validation_dpia_set: pointer to the dc_validation_dpia_set + * @count: number of DPIA validation sets + * + * return: true if validation is succeeded + */ +bool link_dpia_validate_dp_tunnel_bandwidth(const struct dc_validation_dpia_set *dpia_link_sets, uint8_t count); + #endif /* DC_INC_LINK_DP_DPIA_BW_H_ */ + diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c index da74c2b5854f..035795042a01 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c @@ -173,6 +173,15 @@ bool edp_set_backlight_level_nits(struct dc_link *link, target_luminance = (struct target_luminance_value *)&backlight_millinits; + //make sure we disable AMD ABC first. + core_link_read_dpcd(link, DP_SOURCE_BACKLIGHT_CONTROL, + &backlight_enable, sizeof(uint8_t)); + if (backlight_enable) { + backlight_enable = 0; + core_link_write_dpcd(link, DP_SOURCE_BACKLIGHT_CONTROL, + &backlight_enable, 1); + } + core_link_read_dpcd(link, DP_EDP_BACKLIGHT_MODE_SET_REGISTER, &backlight_enable, sizeof(uint8_t)); @@ -193,10 +202,22 @@ bool edp_set_backlight_level_nits(struct dc_link *link, *(uint16_t *)&dpcd_backlight_set.backlight_transition_time_ms = (uint16_t)transition_time_in_ms; uint8_t backlight_control = isHDR ? 1 : 0; + uint8_t backlight_enable = 0; + // OLEDs have no PWM, they can only use AUX if (link->dpcd_sink_ext_caps.bits.oled == 1) backlight_control = 1; + //make sure we disable VESA ABC first. + core_link_read_dpcd(link, DP_EDP_BACKLIGHT_MODE_SET_REGISTER, + &backlight_enable, sizeof(uint8_t)); + + if (backlight_enable & DP_EDP_PANEL_LUMINANCE_CONTROL_ENABLE) { + backlight_enable &= ~DP_EDP_PANEL_LUMINANCE_CONTROL_ENABLE; + core_link_write_dpcd(link, DP_EDP_BACKLIGHT_MODE_SET_REGISTER, + &backlight_enable, sizeof(backlight_enable)); + } + if (core_link_write_dpcd(link, DP_SOURCE_BACKLIGHT_LEVEL, (uint8_t *)(&dpcd_backlight_set), sizeof(dpcd_backlight_set)) != DC_OK) diff --git a/drivers/gpu/drm/amd/display/dc/mpc/Makefile b/drivers/gpu/drm/amd/display/dc/mpc/Makefile index 1e2e66508192..5402c3529f5e 100644 --- a/drivers/gpu/drm/amd/display/dc/mpc/Makefile +++ b/drivers/gpu/drm/amd/display/dc/mpc/Makefile @@ -68,5 +68,5 @@ MPC_DCN401 = dcn401_mpc.o AMD_DAL_MPC_DCN401 = $(addprefix $(AMDDALPATH)/dc/mpc/dcn401/,$(MPC_DCN401)) AMD_DISPLAY_FILES += $(AMD_DAL_MPC_DCN401) -endif +endif diff --git a/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.c b/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.c index b4cea2b8cb2a..6f0e017a8ae2 100644 --- a/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.c +++ b/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.c @@ -30,7 +30,6 @@ #include "basics/conversion.h" #include "dcn10/dcn10_cm_common.h" #include "dc.h" -#include "dcn401/dcn401_mpc.h" #define REG(reg)\ mpc30->mpc_regs->reg @@ -879,7 +878,7 @@ void mpc32_set3dlut_ram10( } -static void mpc32_set_3dlut_mode( +void mpc32_set_3dlut_mode( struct mpc *mpc, enum dc_lut_mode mode, bool is_color_channel_12bits, @@ -1022,8 +1021,6 @@ static const struct mpc_funcs dcn32_mpc_funcs = { .power_on_mpc_mem_pwr = mpc3_power_on_ogam_lut, .get_mpc_out_mux = mpc1_get_mpc_out_mux, .set_bg_color = mpc1_set_bg_color, - .set_movable_cm_location = mpc401_set_movable_cm_location, - .populate_lut = mpc401_populate_lut, }; diff --git a/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.h b/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.h index 9622518826c9..8c9b20bcca85 100644 --- a/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.h +++ b/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.h @@ -391,4 +391,12 @@ void mpc32_select_3dlut_ram( enum dc_lut_mode mode, bool is_color_channel_12bits, uint32_t mpcc_id); + +void mpc32_set_3dlut_mode( + struct mpc *mpc, + enum dc_lut_mode mode, + bool is_color_channel_12bits, + bool is_lut_size17x17x17, + uint32_t mpcc_id); + #endif //__DC_MPCC_DCN32_H__ diff --git a/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.c b/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.c index 98cf0cbd59ba..f3fb3fe13757 100644 --- a/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.c +++ b/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.c @@ -294,7 +294,7 @@ void mpc401_program_3dlut_size(struct mpc *mpc, bool is_17x17x17, int mpcc_id) REG_UPDATE(MPCC_MCM_3DLUT_MODE[mpcc_id], MPCC_MCM_3DLUT_SIZE, is_17x17x17 ? 0 : 1); } -static void program_gamut_remap( +void mpc_program_gamut_remap( struct mpc *mpc, unsigned int mpcc_id, const uint16_t *regval, @@ -426,7 +426,7 @@ void mpc401_set_gamut_remap( if (adjust->gamut_adjust_type != GRAPHICS_GAMUT_ADJUST_TYPE_SW) { /* Bypass / Disable if type is bypass or hw */ - program_gamut_remap(mpc, mpcc_id, NULL, + mpc_program_gamut_remap(mpc, mpcc_id, NULL, adjust->mpcc_gamut_remap_block_id, MPCC_GAMUT_REMAP_MODE_SELECT_0); } else { struct fixed31_32 arr_matrix[12]; @@ -460,12 +460,12 @@ void mpc401_set_gamut_remap( else mode_select = MPCC_GAMUT_REMAP_MODE_SELECT_2; - program_gamut_remap(mpc, mpcc_id, arr_reg_val, + mpc_program_gamut_remap(mpc, mpcc_id, arr_reg_val, adjust->mpcc_gamut_remap_block_id, mode_select); } } -static void read_gamut_remap(struct mpc *mpc, +void mpc_read_gamut_remap(struct mpc *mpc, int mpcc_id, uint16_t *regval, enum mpcc_gamut_remap_id gamut_remap_block_id, @@ -561,9 +561,9 @@ void mpc401_get_gamut_remap(struct mpc *mpc, struct mpc_grph_gamut_adjustment *adjust) { uint16_t arr_reg_val[12] = {0}; - uint32_t mode_select; + uint32_t mode_select = MPCC_GAMUT_REMAP_MODE_SELECT_0; - read_gamut_remap(mpc, mpcc_id, arr_reg_val, adjust->mpcc_gamut_remap_block_id, &mode_select); + mpc_read_gamut_remap(mpc, mpcc_id, arr_reg_val, adjust->mpcc_gamut_remap_block_id, &mode_select); if (mode_select == MPCC_GAMUT_REMAP_MODE_SELECT_0) { adjust->gamut_adjust_type = GRAPHICS_GAMUT_ADJUST_TYPE_BYPASS; diff --git a/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.h b/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.h index 8e35ebc603a9..eb0c68d0b0c7 100644 --- a/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.h +++ b/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.h @@ -241,6 +241,19 @@ void mpc401_update_3dlut_fast_load_select( int mpcc_id, int hubp_idx); +void mpc_program_gamut_remap( + struct mpc *mpc, + unsigned int mpcc_id, + const uint16_t *regval, + enum mpcc_gamut_remap_id gamut_remap_block_id, + enum mpcc_gamut_remap_mode_select mode_select); + +void mpc_read_gamut_remap(struct mpc *mpc, + int mpcc_id, + uint16_t *regval, + enum mpcc_gamut_remap_id gamut_remap_block_id, + uint32_t *mode_select); + void mpc401_update_3dlut_fast_load_select( struct mpc *mpc, int mpcc_id, diff --git a/drivers/gpu/drm/amd/display/dc/os_types.h b/drivers/gpu/drm/amd/display/dc/os_types.h index f2ba76c1e0c0..782316348941 100644 --- a/drivers/gpu/drm/amd/display/dc/os_types.h +++ b/drivers/gpu/drm/amd/display/dc/os_types.h @@ -31,6 +31,7 @@ #include <linux/kgdb.h> #include <linux/delay.h> #include <linux/mm.h> +#include <linux/vmalloc.h> #include <asm/byteorder.h> diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c index 84f73fdb0f95..3a51be63f020 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c @@ -839,7 +839,7 @@ static enum dc_status build_mapped_resource( static enum dc_status dce100_validate_bandwidth( struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { int i; bool at_least_one_pipe = false; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c index f3d5baac11bf..cccde5a6f3cd 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c @@ -963,7 +963,7 @@ static enum dc_status build_mapped_resource( static enum dc_status dce110_validate_bandwidth( struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool result = false; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c index 4225cae68c10..164ba796f64c 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c @@ -886,7 +886,7 @@ static enum dc_status build_mapped_resource( enum dc_status dce112_validate_bandwidth( struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool result = false; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.h index 6221d749246d..3efc4c55d2d2 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.h @@ -45,7 +45,7 @@ enum dc_status dce112_validate_with_context( enum dc_status dce112_validate_bandwidth( struct dc *dc, struct dc_state *context, - bool fast_validate); + enum dc_validate_mode validate_mode); enum dc_status dce112_add_stream_to_ctx( struct dc *dc, diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c index d9ffdded5ce1..58b59d52dc9d 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c @@ -866,7 +866,7 @@ static void dce60_resource_destruct(struct dce110_resource_pool *pool) static enum dc_status dce60_validate_bandwidth( struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { int i; bool at_least_one_pipe = false; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c index bd5811f97531..3e8b0ac11d90 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c @@ -872,7 +872,7 @@ static void dce80_resource_destruct(struct dce110_resource_pool *pool) static enum dc_status dce80_validate_bandwidth( struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { int i; bool at_least_one_pipe = false; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn10/dcn10_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn10/dcn10_resource.c index be4ade0853e9..652c05c35494 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn10/dcn10_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn10/dcn10_resource.c @@ -1129,12 +1129,12 @@ static void dcn10_destroy_resource_pool(struct resource_pool **pool) static enum dc_status dcn10_validate_bandwidth( struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool voltage_supported; DC_FP_START(); - voltage_supported = dcn_validate_bandwidth(dc, context, fast_validate); + voltage_supported = dcn_validate_bandwidth(dc, context, validate_mode); DC_FP_END(); return voltage_supported ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c index 3405be07f5e3..067a93420a23 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c @@ -2007,7 +2007,7 @@ bool dcn20_fast_validate_bw( int *pipe_cnt_out, int *pipe_split_from, int *vlevel_out, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool out = false; int split[MAX_PIPES] = { 0 }; @@ -2021,7 +2021,7 @@ bool dcn20_fast_validate_bw( dcn20_merge_pipes_for_validate(dc, context); DC_FP_START(); - pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, fast_validate); + pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, validate_mode); DC_FP_END(); *pipe_cnt_out = pipe_cnt; @@ -2125,7 +2125,7 @@ validate_out: } enum dc_status dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool voltage_supported; display_e2e_pipe_params_st *pipes; @@ -2135,7 +2135,7 @@ enum dc_status dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, return DC_FAIL_BANDWIDTH_VALIDATE; DC_FP_START(); - voltage_supported = dcn20_validate_bandwidth_fp(dc, context, fast_validate, pipes); + voltage_supported = dcn20_validate_bandwidth_fp(dc, context, validate_mode, pipes); DC_FP_END(); kfree(pipes); diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.h index c0e062c7407d..e997d35a8b86 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.h @@ -119,7 +119,7 @@ void dcn20_set_mcif_arb_params( struct dc_state *context, display_e2e_pipe_params_st *pipes, int pipe_cnt); -enum dc_status dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, bool fast_validate); +enum dc_status dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, enum dc_validate_mode validate_mode); void dcn20_merge_pipes_for_validate( struct dc *dc, struct dc_state *context); @@ -158,7 +158,7 @@ bool dcn20_fast_validate_bw( int *pipe_cnt_out, int *pipe_split_from, int *vlevel_out, - bool fast_validate); + enum dc_validate_mode validate_mode); enum dc_status dcn20_build_mapped_resource(const struct dc *dc, struct dc_state *context, struct dc_stream_state *stream); enum dc_status dcn20_add_stream_to_ctx(struct dc *dc, struct dc_state *new_ctx, struct dc_stream_state *dc_stream); diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.c index 9ab01b65b177..238d7f8beb7c 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.c @@ -769,7 +769,7 @@ bool dcn21_fast_validate_bw(struct dc *dc, int *pipe_cnt_out, int *pipe_split_from, int *vlevel_out, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool out = false; int split[MAX_PIPES] = { 0 }; @@ -783,7 +783,7 @@ bool dcn21_fast_validate_bw(struct dc *dc, dcn20_merge_pipes_for_validate(dc, context); DC_FP_START(); - pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, fast_validate); + pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, validate_mode); DC_FP_END(); *pipe_cnt_out = pipe_cnt; @@ -924,7 +924,7 @@ validate_out: * dcn20_validate_bandwidth in dcn20_resource.c. */ static enum dc_status dcn21_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool voltage_supported; display_e2e_pipe_params_st *pipes; @@ -934,7 +934,7 @@ static enum dc_status dcn21_validate_bandwidth(struct dc *dc, struct dc_state *c return DC_FAIL_BANDWIDTH_VALIDATE; DC_FP_START(); - voltage_supported = dcn21_validate_bandwidth_fp(dc, context, fast_validate, pipes); + voltage_supported = dcn21_validate_bandwidth_fp(dc, context, validate_mode, pipes); DC_FP_END(); kfree(pipes); diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.h index f7ecc002c2f7..a017fd9854d1 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.h @@ -51,6 +51,6 @@ bool dcn21_fast_validate_bw( int *pipe_cnt_out, int *pipe_split_from, int *vlevel_out, - bool fast_validate); + enum dc_validate_mode validate_mode); #endif /* _DCN21_RESOURCE_H_ */ diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c index f631ae34e320..4d4635e01eb6 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c @@ -1319,13 +1319,13 @@ static struct clock_source *dcn30_clock_source_create( int dcn30_populate_dml_pipes_from_context( struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate) + enum dc_validate_mode validate_mode) { int i, pipe_cnt; struct resource_context *res_ctx = &context->res_ctx; DC_FP_START(); - dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); + dcn20_populate_dml_pipes_from_context(dc, context, pipes, validate_mode); DC_FP_END(); for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) { @@ -1627,7 +1627,7 @@ noinline bool dcn30_internal_validate_bw( display_e2e_pipe_params_st *pipes, int *pipe_cnt_out, int *vlevel_out, - bool fast_validate, + enum dc_validate_mode validate_mode, bool allow_self_refresh_only) { bool out = false; @@ -1646,7 +1646,7 @@ noinline bool dcn30_internal_validate_bw( context->bw_ctx.dml.vba.VoltageLevel = 0; context->bw_ctx.dml.vba.DRAMClockChangeSupport[0][0] = dm_dram_clock_change_vactive; dc->res_pool->funcs->update_soc_for_wm_a(dc, context); - pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, fast_validate); + pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, validate_mode); if (!pipe_cnt) { out = true; @@ -1655,7 +1655,7 @@ noinline bool dcn30_internal_validate_bw( dml_log_pipe_params(&context->bw_ctx.dml, pipes, pipe_cnt); - if (!fast_validate || !allow_self_refresh_only) { + if (validate_mode == DC_VALIDATE_MODE_AND_PROGRAMMING || !allow_self_refresh_only) { /* * DML favors voltage over p-state, but we're more interested in * supporting p-state over voltage. We can't support p-state in @@ -1669,7 +1669,7 @@ noinline bool dcn30_internal_validate_bw( vlevel = dcn20_validate_apply_pipe_split_flags(dc, context, vlevel, split, merge); } if (allow_self_refresh_only && - (fast_validate || vlevel == context->bw_ctx.dml.soc.num_states || + (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING || vlevel == context->bw_ctx.dml.soc.num_states || vba->DRAMClockChangeSupport[vlevel][vba->maxMpcComb] == dm_dram_clock_change_unsupported)) { /* * If mode is unsupported or there's still no p-state support @@ -1678,7 +1678,7 @@ noinline bool dcn30_internal_validate_bw( * We don't actually support prefetch mode 2, so require that we * at least support prefetch mode 1. */ - context->bw_ctx.dml.validate_max_state = fast_validate; + context->bw_ctx.dml.validate_max_state = (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING); context->bw_ctx.dml.soc.allow_dram_self_refresh_or_dram_clock_change_in_vblank = dm_allow_self_refresh; @@ -1865,7 +1865,7 @@ noinline bool dcn30_internal_validate_bw( } if (repopulate_pipes) - pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, fast_validate); + pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, validate_mode); context->bw_ctx.dml.vba.VoltageLevel = vlevel; *vlevel_out = vlevel; *pipe_cnt_out = pipe_cnt; @@ -2037,7 +2037,7 @@ void dcn30_calculate_wm_and_dlg( enum dc_status dcn30_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool out = false; @@ -2055,7 +2055,7 @@ enum dc_status dcn30_validate_bandwidth(struct dc *dc, goto validate_fail; DC_FP_START(); - out = dcn30_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, fast_validate, true); + out = dcn30_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, validate_mode, true); DC_FP_END(); if (pipe_cnt == 0) @@ -2066,7 +2066,7 @@ enum dc_status dcn30_validate_bandwidth(struct dc *dc, BW_VAL_TRACE_END_VOLTAGE_LEVEL(); - if (fast_validate) { + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) { BW_VAL_TRACE_SKIP(fast); goto validate_out; } diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.h index 689d9bdace81..2c967fe55712 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.h @@ -57,14 +57,14 @@ unsigned int dcn30_calc_max_scaled_time( unsigned int urgent_watermark); enum dc_status dcn30_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate); + enum dc_validate_mode validate_mode); bool dcn30_internal_validate_bw( struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, int *pipe_cnt_out, int *vlevel_out, - bool fast_validate, + enum dc_validate_mode validate_mode, bool allow_self_refresh_only); void dcn30_calculate_wm_and_dlg( struct dc *dc, struct dc_state *context, @@ -78,7 +78,7 @@ void dcn30_populate_dml_writeback_from_context( int dcn30_populate_dml_pipes_from_context( struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate); + enum dc_validate_mode validate_mode); bool dcn30_acquire_post_bldn_3dlut( struct resource_context *res_ctx, diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c index 7e0af5297dc4..88afa59d17b0 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c @@ -1616,14 +1616,14 @@ static bool is_dual_plane(enum surface_pixel_format format) int dcn31x_populate_dml_pipes_from_context(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate) + enum dc_validate_mode validate_mode) { uint32_t pipe_cnt; int i; dc_assert_fp_enabled(); - pipe_cnt = dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); + pipe_cnt = dcn20_populate_dml_pipes_from_context(dc, context, pipes, validate_mode); for (i = 0; i < pipe_cnt; i++) { pipes[i].pipe.src.gpuvm = 1; @@ -1641,7 +1641,7 @@ int dcn31x_populate_dml_pipes_from_context(struct dc *dc, int dcn31_populate_dml_pipes_from_context( struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate) + enum dc_validate_mode validate_mode) { int i, pipe_cnt; struct resource_context *res_ctx = &context->res_ctx; @@ -1649,7 +1649,7 @@ int dcn31_populate_dml_pipes_from_context( bool upscaled = false; DC_FP_START(); - dcn31x_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); + dcn31x_populate_dml_pipes_from_context(dc, context, pipes, validate_mode); DC_FP_END(); for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) { @@ -1760,7 +1760,7 @@ dcn31_set_mcif_arb_params(struct dc *dc, enum dc_status dcn31_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool out = false; @@ -1778,19 +1778,19 @@ enum dc_status dcn31_validate_bandwidth(struct dc *dc, goto validate_fail; DC_FP_START(); - out = dcn30_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, fast_validate, true); + out = dcn30_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, validate_mode, true); DC_FP_END(); - // Disable fast_validate to set min dcfclk in calculate_wm_and_dlg + // Disable DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX to set min dcfclk in calculate_wm_and_dlg if (pipe_cnt == 0) - fast_validate = false; + validate_mode = DC_VALIDATE_MODE_AND_PROGRAMMING; if (!out) goto validate_fail; BW_VAL_TRACE_END_VOLTAGE_LEVEL(); - if (fast_validate) { + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) { BW_VAL_TRACE_SKIP(fast); goto validate_out; } @@ -1850,7 +1850,9 @@ static struct resource_funcs dcn31_res_pool_funcs = { .patch_unknown_plane_state = dcn20_patch_unknown_plane_state, .get_panel_config_defaults = dcn31_get_panel_config_defaults, .get_det_buffer_size = dcn31_get_det_buffer_size, - .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe + .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe, + .update_dc_state_for_encoder_switch = dcn31_update_dc_state_for_encoder_switch, + .build_pipe_pix_clk_params = dcn20_build_pipe_pix_clk_params }; static struct clock_source *dcn30_clock_source_create( @@ -1954,6 +1956,9 @@ static bool dcn31_resource_construct( dc->caps.color.mpc.ogam_rom_caps.hlg = 0; dc->caps.color.mpc.ocsc = 1; + dc->caps.num_of_host_routers = 2; + dc->caps.num_of_dpias_per_host_router = 2; + /* Use pipe context based otg sync logic */ dc->config.use_pipe_ctx_sync_logic = true; dc->config.disable_hbr_audio_dp2 = true; @@ -2228,3 +2233,35 @@ struct resource_pool *dcn31_create_resource_pool( kfree(pool); return NULL; } + +enum dc_status dcn31_update_dc_state_for_encoder_switch(struct dc_link *link, + struct dc_link_settings *link_setting, + uint8_t pipe_count, + struct pipe_ctx *pipes, + struct audio_output *audio_output) +{ + struct dc_state *state = link->dc->current_state; + int i; + +#if defined(CONFIG_DRM_AMD_DC_FP) + for (i = 0; i < state->stream_count; i++) + if (state->streams[i] && state->streams[i]->link && state->streams[i]->link == link) + link->dc->hwss.calculate_pix_rate_divider((struct dc *)link->dc, state, state->streams[i]); + + for (i = 0; i < pipe_count; i++) { + link->dc->res_pool->funcs->build_pipe_pix_clk_params(&pipes[i]); + + // Setup audio + if (pipes[i].stream_res.audio != NULL) + build_audio_output(state, &pipes[i], &audio_output[i]); + } +#else + /* This DCN requires rate divider updates and audio reprogramming to allow DP1<-->DP2 link rate switching, + * but the above will not compile on architectures without an FPU. + */ + DC_LOG_WARNING("%s: DP1<-->DP2 link retraining will not work on this DCN on non-FPU platforms", __func__); + ASSERT(0); +#endif + + return DC_OK; +} diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.h index dd82815d7efe..c32c85ef0ba4 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.h @@ -39,7 +39,7 @@ struct dcn31_resource_pool { enum dc_status dcn31_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate); + enum dc_validate_mode validate_mode); void dcn31_calculate_wm_and_dlg( struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, @@ -48,7 +48,7 @@ void dcn31_calculate_wm_and_dlg( int dcn31_populate_dml_pipes_from_context( struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate); + enum dc_validate_mode validate_mode); void dcn31_populate_dml_writeback_from_context(struct dc *dc, struct resource_context *res_ctx, @@ -66,6 +66,12 @@ struct resource_pool *dcn31_create_resource_pool( unsigned int dcn31_get_det_buffer_size( const struct dc_state *context); +enum dc_status dcn31_update_dc_state_for_encoder_switch(struct dc_link *link, + struct dc_link_settings *link_setting, + uint8_t pipe_count, + struct pipe_ctx *pipes, + struct audio_output *audio_output); + /*temp: B0 specific before switch to dcn313 headers*/ #ifndef regPHYPLLF_PIXCLK_RESYNC_CNTL #define regPHYPLLF_PIXCLK_RESYNC_CNTL 0x007e diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c index d96bc6cb73ad..dedf7bce6ece 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c @@ -1667,12 +1667,12 @@ static struct clock_source *dcn31_clock_source_create( static int dcn314_populate_dml_pipes_from_context( struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate) + enum dc_validate_mode validate_mode) { int pipe_cnt; DC_FP_START(); - pipe_cnt = dcn314_populate_dml_pipes_from_context_fpu(dc, context, pipes, fast_validate); + pipe_cnt = dcn314_populate_dml_pipes_from_context_fpu(dc, context, pipes, validate_mode); DC_FP_END(); return pipe_cnt; @@ -1696,7 +1696,7 @@ static void dcn314_get_panel_config_defaults(struct dc_panel_config *panel_confi enum dc_status dcn314_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool out = false; @@ -1715,19 +1715,19 @@ enum dc_status dcn314_validate_bandwidth(struct dc *dc, DC_FP_START(); // do not support self refresh only - out = dcn30_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, fast_validate, false); + out = dcn30_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, validate_mode, false); DC_FP_END(); - // Disable fast_validate to set min dcfclk in calculate_wm_and_dlg + // Disable DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX to set min dcfclk in calculate_wm_and_dlg if (pipe_cnt == 0) - fast_validate = false; + validate_mode = DC_VALIDATE_MODE_AND_PROGRAMMING; if (!out) goto validate_fail; BW_VAL_TRACE_END_VOLTAGE_LEVEL(); - if (fast_validate) { + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) { BW_VAL_TRACE_SKIP(fast); goto validate_out; } @@ -1779,7 +1779,9 @@ static struct resource_funcs dcn314_res_pool_funcs = { .get_panel_config_defaults = dcn314_get_panel_config_defaults, .get_preferred_eng_id_dpia = dcn314_get_preferred_eng_id_dpia, .get_det_buffer_size = dcn31_get_det_buffer_size, - .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe + .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe, + .update_dc_state_for_encoder_switch = dcn31_update_dc_state_for_encoder_switch, + .build_pipe_pix_clk_params = dcn20_build_pipe_pix_clk_params }; static struct clock_source *dcn30_clock_source_create( @@ -1885,6 +1887,9 @@ static bool dcn314_resource_construct( dc->caps.max_disp_clock_khz_at_vmin = 650000; + dc->caps.num_of_host_routers = 2; + dc->caps.num_of_dpias_per_host_router = 2; + /* Use pipe context based otg sync logic */ dc->config.use_pipe_ctx_sync_logic = true; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.h index f8ba531d6342..ac9bb7f097d5 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.h @@ -41,7 +41,7 @@ struct dcn314_resource_pool { enum dc_status dcn314_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate); + enum dc_validate_mode validate_mode); struct resource_pool *dcn314_create_resource_pool( const struct dc_init_data *init_data, diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c index 6c2bb3f63be1..d110be626bc2 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c @@ -1664,7 +1664,7 @@ static bool allow_pixel_rate_crb(struct dc *dc, struct dc_state *context) static int dcn315_populate_dml_pipes_from_context( struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate) + enum dc_validate_mode validate_mode) { int i, pipe_cnt, crb_idx, crb_pipes; struct resource_context *res_ctx = &context->res_ctx; @@ -1674,7 +1674,7 @@ static int dcn315_populate_dml_pipes_from_context( bool pixel_rate_crb = allow_pixel_rate_crb(dc, context); DC_FP_START(); - dcn31x_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); + dcn31x_populate_dml_pipes_from_context(dc, context, pipes, validate_mode); DC_FP_END(); for (i = 0, pipe_cnt = 0, crb_pipes = 0; i < dc->res_pool->pipe_count; i++) { @@ -1844,7 +1844,9 @@ static struct resource_funcs dcn315_res_pool_funcs = { .get_panel_config_defaults = dcn315_get_panel_config_defaults, .get_power_profile = dcn315_get_power_profile, .get_det_buffer_size = dcn31_get_det_buffer_size, - .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe + .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe, + .update_dc_state_for_encoder_switch = dcn31_update_dc_state_for_encoder_switch, + .build_pipe_pix_clk_params = dcn20_build_pipe_pix_clk_params }; static bool dcn315_resource_construct( diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c index 568094827212..939811858ff7 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c @@ -1610,7 +1610,7 @@ static bool is_dual_plane(enum surface_pixel_format format) static int dcn316_populate_dml_pipes_from_context( struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate) + enum dc_validate_mode validate_mode) { int i, pipe_cnt; struct resource_context *res_ctx = &context->res_ctx; @@ -1618,7 +1618,7 @@ static int dcn316_populate_dml_pipes_from_context( const int max_usable_det = context->bw_ctx.dml.ip.config_return_buffer_size_in_kbytes - DCN3_16_MIN_COMPBUF_SIZE_KB; DC_FP_START(); - dcn31x_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); + dcn31x_populate_dml_pipes_from_context(dc, context, pipes, validate_mode); DC_FP_END(); for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) { @@ -1720,7 +1720,9 @@ static struct resource_funcs dcn316_res_pool_funcs = { .patch_unknown_plane_state = dcn20_patch_unknown_plane_state, .get_panel_config_defaults = dcn316_get_panel_config_defaults, .get_det_buffer_size = dcn31_get_det_buffer_size, - .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe + .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe, + .update_dc_state_for_encoder_switch = dcn31_update_dc_state_for_encoder_switch, + .build_pipe_pix_clk_params = dcn20_build_pipe_pix_clk_params }; static bool dcn316_resource_construct( diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c index bb0dae0be5b8..9ffa10189eee 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c @@ -1742,7 +1742,7 @@ void dcn32_add_phantom_pipes(struct dc *dc, struct dc_state *context, } } -static bool dml1_validate(struct dc *dc, struct dc_state *context, bool fast_validate) +static bool dml1_validate(struct dc *dc, struct dc_state *context, enum dc_validate_mode validate_mode) { bool out = false; @@ -1767,7 +1767,7 @@ static bool dml1_validate(struct dc *dc, struct dc_state *context, bool fast_val goto validate_fail; DC_FP_START(); - out = dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, fast_validate); + out = dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, validate_mode); DC_FP_END(); if (pipe_cnt == 0) @@ -1778,7 +1778,7 @@ static bool dml1_validate(struct dc *dc, struct dc_state *context, bool fast_val BW_VAL_TRACE_END_VOLTAGE_LEVEL(); - if (fast_validate) { + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) { BW_VAL_TRACE_SKIP(fast); goto validate_out; } @@ -1809,7 +1809,7 @@ validate_out: enum dc_status dcn32_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { unsigned int i; enum dc_status status; @@ -1827,11 +1827,11 @@ enum dc_status dcn32_validate_bandwidth(struct dc *dc, if (dc->debug.using_dml2) status = dml2_validate(dc, context, context->power_source == DC_POWER_SOURCE_DC ? context->bw_ctx.dml2_dc_power_source : context->bw_ctx.dml2, - fast_validate) ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; + validate_mode) ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; else - status = dml1_validate(dc, context, fast_validate) ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; + status = dml1_validate(dc, context, validate_mode) ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; - if (!fast_validate && status == DC_OK && dc_state_is_subvp_in_use(context)) { + if (validate_mode == DC_VALIDATE_MODE_AND_PROGRAMMING && status == DC_OK && dc_state_is_subvp_in_use(context)) { /* check new stream configuration still supports cursor if subvp used */ for (i = 0; i < context->stream_count; i++) { stream = context->streams[i]; @@ -1846,14 +1846,14 @@ enum dc_status dcn32_validate_bandwidth(struct dc *dc, }; } - if (!fast_validate && status == DC_FAIL_HW_CURSOR_SUPPORT) { + if (validate_mode == DC_VALIDATE_MODE_AND_PROGRAMMING && status == DC_FAIL_HW_CURSOR_SUPPORT) { /* attempt to validate again with subvp disabled due to cursor */ if (dc->debug.using_dml2) status = dml2_validate(dc, context, context->power_source == DC_POWER_SOURCE_DC ? context->bw_ctx.dml2_dc_power_source : context->bw_ctx.dml2, - fast_validate) ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; + validate_mode) ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; else - status = dml1_validate(dc, context, fast_validate) ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; + status = dml1_validate(dc, context, validate_mode) ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; } return status; @@ -1862,7 +1862,7 @@ enum dc_status dcn32_validate_bandwidth(struct dc *dc, int dcn32_populate_dml_pipes_from_context( struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate) + enum dc_validate_mode validate_mode) { int i, pipe_cnt; struct resource_context *res_ctx = &context->res_ctx; @@ -1878,7 +1878,7 @@ int dcn32_populate_dml_pipes_from_context( int num_subvp_none = 0; int odm_slice_count; - dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); + dcn20_populate_dml_pipes_from_context(dc, context, pipes, validate_mode); /* For single display subvp, look for subvp main so if we have phantom * pipe, we can set odm policy to match main pipe @@ -1960,7 +1960,7 @@ int dcn32_populate_dml_pipes_from_context( /* Only populate DML input with subvp info for full updates. * This is just a workaround -- needs a proper fix. */ - if (!fast_validate) { + if (validate_mode == DC_VALIDATE_MODE_AND_PROGRAMMING) { switch (dc_state_get_pipe_subvp_type(context, pipe)) { case SUBVP_MAIN: pipes[pipe_cnt].pipe.src.use_mall_for_pstate_change = dm_use_mall_pstate_change_sub_viewport; @@ -2061,21 +2061,15 @@ void dcn32_calculate_wm_and_dlg(struct dc *dc, struct dc_state *context, static void dcn32_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params) { - struct dml2_configuration_options *dml2_opt = &dc->dml2_tmp; - - memcpy(dml2_opt, &dc->dml2_options, sizeof(dc->dml2_options)); - DC_FP_START(); dcn32_update_bw_bounding_box_fpu(dc, bw_params); - dml2_opt->use_clock_dc_limits = false; if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2) - dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2); + dml2_reinit(dc, &dc->dml2_options, &dc->current_state->bw_ctx.dml2); - dml2_opt->use_clock_dc_limits = true; if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2_dc_power_source) - dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2_dc_power_source); + dml2_reinit(dc, &dc->dml2_dc_power_options, &dc->current_state->bw_ctx.dml2_dc_power_source); DC_FP_END(); } @@ -2257,7 +2251,7 @@ static bool dcn32_resource_construct( dc->caps.color.dpp.gamma_corr = 1; dc->caps.color.dpp.dgam_rom_for_yuv = 0; - dc->caps.color.dpp.hw_3d_lut = 1; + dc->caps.color.dpp.hw_3d_lut = 0; dc->caps.color.dpp.ogam_ram = 0; // no OGAM in DPP since DCN1 // no OGAM ROM on DCN2 and later ASICs dc->caps.color.dpp.ogam_rom_caps.srgb = 0; @@ -2276,6 +2270,7 @@ static bool dcn32_resource_construct( dc->caps.color.mpc.ogam_rom_caps.pq = 0; dc->caps.color.mpc.ogam_rom_caps.hlg = 0; dc->caps.color.mpc.ocsc = 1; + dc->caps.color.mpc.preblend = true; /* Use pipe context based otg sync logic */ dc->config.use_pipe_ctx_sync_logic = true; @@ -2519,7 +2514,6 @@ static bool dcn32_resource_construct( } dc->dml2_options.dcn_pipe_count = pool->base.pipe_count; - dc->dml2_options.use_native_pstate_optimization = false; dc->dml2_options.use_native_soc_bb_construction = true; dc->dml2_options.minimize_dispclk_using_odm = true; @@ -2551,6 +2545,10 @@ static bool dcn32_resource_construct( if (ASICREV_IS_GC_11_0_3(dc->ctx->asic_id.hw_internal_rev) && (dc->config.sdpif_request_limit_words_per_umc == 0)) dc->config.sdpif_request_limit_words_per_umc = 16; + /* init DC limited DML2 options */ + memcpy(&dc->dml2_dc_power_options, &dc->dml2_options, sizeof(struct dml2_configuration_options)); + dc->dml2_dc_power_options.use_clock_dc_limits = true; + return true; create_fail: diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.h index d60ed77eda80..82f966cf4ed2 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.h @@ -100,12 +100,12 @@ void dcn32_add_phantom_pipes(struct dc *dc, enum dc_status dcn32_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate); + enum dc_validate_mode validate_mode); int dcn32_populate_dml_pipes_from_context( struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, - bool fast_validate); + enum dc_validate_mode validate_mode); void dcn32_calculate_wm_and_dlg( struct dc *dc, struct dc_state *context, diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c index 7db1f7a5613f..c53266e16c58 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c @@ -1580,21 +1580,15 @@ static struct dc_cap_funcs cap_funcs = { static void dcn321_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params) { - struct dml2_configuration_options *dml2_opt = &dc->dml2_tmp; - - memcpy(dml2_opt, &dc->dml2_options, sizeof(dc->dml2_options)); - DC_FP_START(); dcn321_update_bw_bounding_box_fpu(dc, bw_params); - dml2_opt->use_clock_dc_limits = false; if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2) - dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2); + dml2_reinit(dc, &dc->dml2_options, &dc->current_state->bw_ctx.dml2); - dml2_opt->use_clock_dc_limits = true; if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2_dc_power_source) - dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2_dc_power_source); + dml2_reinit(dc, &dc->dml2_dc_power_options, &dc->current_state->bw_ctx.dml2_dc_power_source); DC_FP_END(); } @@ -1761,8 +1755,8 @@ static bool dcn321_resource_construct( dc->caps.color.dpp.gamma_corr = 1; dc->caps.color.dpp.dgam_rom_for_yuv = 0; - dc->caps.color.dpp.hw_3d_lut = 1; - dc->caps.color.dpp.ogam_ram = 1; + dc->caps.color.dpp.hw_3d_lut = 0; + dc->caps.color.dpp.ogam_ram = 0; // no OGAM ROM on DCN2 and later ASICs dc->caps.color.dpp.ogam_rom_caps.srgb = 0; dc->caps.color.dpp.ogam_rom_caps.bt2020 = 0; @@ -1780,6 +1774,7 @@ static bool dcn321_resource_construct( dc->caps.color.mpc.ogam_rom_caps.pq = 0; dc->caps.color.mpc.ogam_rom_caps.hlg = 0; dc->caps.color.mpc.ocsc = 1; + dc->caps.color.mpc.preblend = true; /* Use pipe context based otg sync logic */ dc->config.use_pipe_ctx_sync_logic = true; @@ -2018,7 +2013,6 @@ static bool dcn321_resource_construct( } dc->dml2_options.dcn_pipe_count = pool->base.pipe_count; - dc->dml2_options.use_native_pstate_optimization = false; dc->dml2_options.use_native_soc_bb_construction = true; dc->dml2_options.minimize_dispclk_using_odm = true; @@ -2046,6 +2040,10 @@ static bool dcn321_resource_construct( dc->dml2_options.max_segments_per_hubp = 18; dc->dml2_options.det_segment_size = DCN3_2_DET_SEG_SIZE; + /* init DC limited DML2 options */ + memcpy(&dc->dml2_dc_power_options, &dc->dml2_options, sizeof(struct dml2_configuration_options)); + dc->dml2_dc_power_options.use_clock_dc_limits = true; + return true; create_fail: diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c index 72c6cf047db0..e327dee9be21 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c @@ -1734,15 +1734,15 @@ static void dcn35_get_panel_config_defaults(struct dc_panel_config *panel_config static enum dc_status dcn35_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool out = false; out = dml2_validate(dc, context, context->power_source == DC_POWER_SOURCE_DC ? context->bw_ctx.dml2_dc_power_source : context->bw_ctx.dml2, - fast_validate); + validate_mode); - if (fast_validate) + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) return out ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; DC_FP_START(); @@ -1786,7 +1786,9 @@ static struct resource_funcs dcn35_res_pool_funcs = { .get_panel_config_defaults = dcn35_get_panel_config_defaults, .get_preferred_eng_id_dpia = dcn35_get_preferred_eng_id_dpia, .get_det_buffer_size = dcn31_get_det_buffer_size, - .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe + .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe, + .update_dc_state_for_encoder_switch = dcn31_update_dc_state_for_encoder_switch, + .build_pipe_pix_clk_params = dcn20_build_pipe_pix_clk_params }; static bool dcn35_resource_construct( @@ -1874,7 +1876,7 @@ static bool dcn35_resource_construct( dc->caps.color.dpp.gamma_corr = 1; dc->caps.color.dpp.dgam_rom_for_yuv = 0; - dc->caps.color.dpp.hw_3d_lut = 1; + dc->caps.color.dpp.hw_3d_lut = 0; dc->caps.color.dpp.ogam_ram = 0; // no OGAM in DPP since DCN1 // no OGAM ROM on DCN301 dc->caps.color.dpp.ogam_rom_caps.srgb = 0; @@ -1893,6 +1895,10 @@ static bool dcn35_resource_construct( dc->caps.color.mpc.ogam_rom_caps.pq = 0; dc->caps.color.mpc.ogam_rom_caps.hlg = 0; dc->caps.color.mpc.ocsc = 1; + dc->caps.color.mpc.preblend = true; + + dc->caps.num_of_host_routers = 2; + dc->caps.num_of_dpias_per_host_router = 2; /* max_disp_clock_khz_at_vmin is slightly lower than the STA value in order * to provide some margin. @@ -2153,7 +2159,6 @@ static bool dcn35_resource_construct( dc->dcn_ip->max_num_dpp = pool->base.pipe_count; dc->dml2_options.dcn_pipe_count = pool->base.pipe_count; - dc->dml2_options.use_native_pstate_optimization = true; dc->dml2_options.use_native_soc_bb_construction = true; dc->dml2_options.minimize_dispclk_using_odm = false; if (dc->config.EnableMinDispClkODM) diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c index 989a270f7dea..2f2976afc229 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c @@ -1714,15 +1714,15 @@ static void dcn35_get_panel_config_defaults(struct dc_panel_config *panel_config static enum dc_status dcn351_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool out = false; out = dml2_validate(dc, context, context->power_source == DC_POWER_SOURCE_DC ? context->bw_ctx.dml2_dc_power_source : context->bw_ctx.dml2, - fast_validate); + validate_mode); - if (fast_validate) + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) return out ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; DC_FP_START(); @@ -1758,7 +1758,9 @@ static struct resource_funcs dcn351_res_pool_funcs = { .get_panel_config_defaults = dcn35_get_panel_config_defaults, .get_preferred_eng_id_dpia = dcn351_get_preferred_eng_id_dpia, .get_det_buffer_size = dcn31_get_det_buffer_size, - .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe + .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe, + .update_dc_state_for_encoder_switch = dcn31_update_dc_state_for_encoder_switch, + .build_pipe_pix_clk_params = dcn20_build_pipe_pix_clk_params }; static bool dcn351_resource_construct( @@ -1846,7 +1848,7 @@ static bool dcn351_resource_construct( dc->caps.color.dpp.gamma_corr = 1; dc->caps.color.dpp.dgam_rom_for_yuv = 0; - dc->caps.color.dpp.hw_3d_lut = 1; + dc->caps.color.dpp.hw_3d_lut = 0; dc->caps.color.dpp.ogam_ram = 0; // no OGAM in DPP since DCN1 // no OGAM ROM on DCN301 dc->caps.color.dpp.ogam_rom_caps.srgb = 0; @@ -1865,6 +1867,10 @@ static bool dcn351_resource_construct( dc->caps.color.mpc.ogam_rom_caps.pq = 0; dc->caps.color.mpc.ogam_rom_caps.hlg = 0; dc->caps.color.mpc.ocsc = 1; + dc->caps.color.mpc.preblend = true; + + dc->caps.num_of_host_routers = 2; + dc->caps.num_of_dpias_per_host_router = 2; /* max_disp_clock_khz_at_vmin is slightly lower than the STA value in order * to provide some margin. @@ -2125,7 +2131,6 @@ static bool dcn351_resource_construct( dc->dcn_ip->max_num_dpp = pool->base.pipe_count; dc->dml2_options.dcn_pipe_count = pool->base.pipe_count; - dc->dml2_options.use_native_pstate_optimization = true; dc->dml2_options.use_native_soc_bb_construction = true; dc->dml2_options.minimize_dispclk_using_odm = false; if (dc->config.EnableMinDispClkODM) diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c index 48e1f234185f..5b7848496a70 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c @@ -1715,15 +1715,15 @@ static void dcn35_get_panel_config_defaults(struct dc_panel_config *panel_config static enum dc_status dcn35_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { bool out = false; out = dml2_validate(dc, context, context->power_source == DC_POWER_SOURCE_DC ? context->bw_ctx.dml2_dc_power_source : context->bw_ctx.dml2, - fast_validate); + validate_mode); - if (fast_validate) + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) return out ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; DC_FP_START(); @@ -1759,7 +1759,9 @@ static struct resource_funcs dcn36_res_pool_funcs = { .patch_unknown_plane_state = dcn20_patch_unknown_plane_state, .get_panel_config_defaults = dcn35_get_panel_config_defaults, .get_preferred_eng_id_dpia = dcn36_get_preferred_eng_id_dpia, - .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe + .get_vstartup_for_pipe = dcn10_get_vstartup_for_pipe, + .update_dc_state_for_encoder_switch = dcn31_update_dc_state_for_encoder_switch, + .build_pipe_pix_clk_params = dcn20_build_pipe_pix_clk_params }; static bool dcn36_resource_construct( @@ -1847,7 +1849,7 @@ static bool dcn36_resource_construct( dc->caps.color.dpp.gamma_corr = 1; dc->caps.color.dpp.dgam_rom_for_yuv = 0; - dc->caps.color.dpp.hw_3d_lut = 1; + dc->caps.color.dpp.hw_3d_lut = 0; dc->caps.color.dpp.ogam_ram = 0; // no OGAM in DPP since DCN1 // no OGAM ROM on DCN301 dc->caps.color.dpp.ogam_rom_caps.srgb = 0; @@ -1866,6 +1868,10 @@ static bool dcn36_resource_construct( dc->caps.color.mpc.ogam_rom_caps.pq = 0; dc->caps.color.mpc.ogam_rom_caps.hlg = 0; dc->caps.color.mpc.ocsc = 1; + dc->caps.color.mpc.preblend = true; + + dc->caps.num_of_host_routers = 2; + dc->caps.num_of_dpias_per_host_router = 2; /* max_disp_clock_khz_at_vmin is slightly lower than the STA value in order * to provide some margin. @@ -2126,7 +2132,6 @@ static bool dcn36_resource_construct( dc->dcn_ip->max_num_dpp = pool->base.pipe_count; dc->dml2_options.dcn_pipe_count = pool->base.pipe_count; - dc->dml2_options.use_native_pstate_optimization = true; dc->dml2_options.use_native_soc_bb_construction = true; dc->dml2_options.minimize_dispclk_using_odm = false; if (dc->config.EnableMinDispClkODM) diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c index f420c4dafa03..a9d989f20405 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c @@ -70,7 +70,6 @@ #include "dml/dcn30/display_mode_vba_30.h" #include "vm_helper.h" #include "dcn20/dcn20_vmid.h" -#include "dml/dcn401/dcn401_fpu.h" #include "dc_state_priv.h" @@ -1608,10 +1607,6 @@ static struct dc_cap_funcs cap_funcs = { static void dcn401_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params) { - struct dml2_configuration_options *dml2_opt = &dc->dml2_tmp; - - memcpy(dml2_opt, &dc->dml2_options, sizeof(dc->dml2_options)); - /* re-calculate the available MALL size if required */ if (bw_params->num_channels > 0) { dc->caps.max_cab_allocation_bytes = dcn401_calc_num_avail_chans_for_mall( @@ -1622,15 +1617,11 @@ static void dcn401_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *b DC_FP_START(); - dcn401_update_bw_bounding_box_fpu(dc, bw_params); - - dml2_opt->use_clock_dc_limits = false; if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2) - dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2); + dml2_reinit(dc, &dc->dml2_options, &dc->current_state->bw_ctx.dml2); - dml2_opt->use_clock_dc_limits = true; if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2_dc_power_source) - dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2_dc_power_source); + dml2_reinit(dc, &dc->dml2_dc_power_options, &dc->current_state->bw_ctx.dml2_dc_power_source); DC_FP_END(); } @@ -1644,7 +1635,7 @@ enum dc_status dcn401_patch_unknown_plane_state(struct dc_plane_state *plane_sta enum dc_status dcn401_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate) + enum dc_validate_mode validate_mode) { unsigned int i; enum dc_status status = DC_OK; @@ -1662,9 +1653,9 @@ enum dc_status dcn401_validate_bandwidth(struct dc *dc, if (dc->debug.using_dml2) status = dml2_validate(dc, context, context->power_source == DC_POWER_SOURCE_DC ? context->bw_ctx.dml2_dc_power_source : context->bw_ctx.dml2, - fast_validate) ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; + validate_mode) ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; - if (!fast_validate && status == DC_OK && dc_state_is_subvp_in_use(context)) { + if (validate_mode == DC_VALIDATE_MODE_AND_PROGRAMMING && status == DC_OK && dc_state_is_subvp_in_use(context)) { /* check new stream configuration still supports cursor if subvp used */ for (i = 0; i < context->stream_count; i++) { stream = context->streams[i]; @@ -1679,12 +1670,12 @@ enum dc_status dcn401_validate_bandwidth(struct dc *dc, }; } - if (!fast_validate && status == DC_FAIL_HW_CURSOR_SUPPORT) { + if (validate_mode == DC_VALIDATE_MODE_AND_PROGRAMMING && status == DC_FAIL_HW_CURSOR_SUPPORT) { /* attempt to validate again with subvp disabled due to cursor */ if (dc->debug.using_dml2) status = dml2_validate(dc, context, context->power_source == DC_POWER_SOURCE_DC ? context->bw_ctx.dml2_dc_power_source : context->bw_ctx.dml2, - fast_validate) ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; + validate_mode) ? DC_OK : DC_FAIL_BANDWIDTH_VALIDATE; } return status; @@ -1957,8 +1948,30 @@ static bool dcn401_resource_construct( dc->caps.color.mpc.ogam_rom_caps.pq = 0; dc->caps.color.mpc.ogam_rom_caps.hlg = 0; dc->caps.color.mpc.ocsc = 1; + dc->caps.color.mpc.preblend = true; dc->config.use_spl = true; dc->config.prefer_easf = true; + + dc->config.dcn_sharpness_range.sdr_rgb_min = 0; + dc->config.dcn_sharpness_range.sdr_rgb_max = 1750; + dc->config.dcn_sharpness_range.sdr_rgb_mid = 750; + dc->config.dcn_sharpness_range.sdr_yuv_min = 0; + dc->config.dcn_sharpness_range.sdr_yuv_max = 3500; + dc->config.dcn_sharpness_range.sdr_yuv_mid = 1500; + dc->config.dcn_sharpness_range.hdr_rgb_min = 0; + dc->config.dcn_sharpness_range.hdr_rgb_max = 2750; + dc->config.dcn_sharpness_range.hdr_rgb_mid = 1500; + + dc->config.dcn_override_sharpness_range.sdr_rgb_min = 0; + dc->config.dcn_override_sharpness_range.sdr_rgb_max = 3250; + dc->config.dcn_override_sharpness_range.sdr_rgb_mid = 1250; + dc->config.dcn_override_sharpness_range.sdr_yuv_min = 0; + dc->config.dcn_override_sharpness_range.sdr_yuv_max = 3500; + dc->config.dcn_override_sharpness_range.sdr_yuv_mid = 1500; + dc->config.dcn_override_sharpness_range.hdr_rgb_min = 0; + dc->config.dcn_override_sharpness_range.hdr_rgb_max = 2750; + dc->config.dcn_override_sharpness_range.hdr_rgb_mid = 1500; + dc->config.dc_mode_clk_limit_support = true; dc->config.enable_windowed_mpo_odm = true; dc->config.set_pipe_unlock_order = true; /* Need to ensure DET gets freed before allocating */ @@ -2195,7 +2208,6 @@ static bool dcn401_resource_construct( dc->config.sdpif_request_limit_words_per_umc = 16; dc->dml2_options.dcn_pipe_count = pool->base.pipe_count; - dc->dml2_options.use_native_pstate_optimization = false; dc->dml2_options.use_native_soc_bb_construction = true; dc->dml2_options.minimize_dispclk_using_odm = true; dc->dml2_options.map_dc_pipes_with_callbacks = true; @@ -2228,6 +2240,10 @@ static bool dcn401_resource_construct( /* SPL */ dc->caps.scl_caps.sharpener_support = true; + /* init DC limited DML2 options */ + memcpy(&dc->dml2_dc_power_options, &dc->dml2_options, sizeof(struct dml2_configuration_options)); + dc->dml2_dc_power_options.use_clock_dc_limits = true; + return true; create_fail: diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h index dc52a30991af..2ae6831c31ef 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h @@ -24,7 +24,7 @@ enum dc_status dcn401_patch_unknown_plane_state(struct dc_plane_state *plane_sta enum dc_status dcn401_validate_bandwidth(struct dc *dc, struct dc_state *context, - bool fast_validate); + enum dc_validate_mode validate_mode); void dcn401_prepare_mcache_programming(struct dc *dc, struct dc_state *context); diff --git a/drivers/gpu/drm/amd/display/dc/sspl/dc_spl.c b/drivers/gpu/drm/amd/display/dc/sspl/dc_spl.c index e0008c5f08ad..55b929ca7982 100644 --- a/drivers/gpu/drm/amd/display/dc/sspl/dc_spl.c +++ b/drivers/gpu/drm/amd/display/dc/sspl/dc_spl.c @@ -196,7 +196,12 @@ static struct spl_rect calculate_mpc_slice_in_timing_active( int epimo = mpc_slice_count - plane_clip_rec->width % mpc_slice_count - 1; struct spl_rect mpc_rec; - if (use_recout_width_aligned) { + if (spl_in->basic_in.custom_width != 0) { + mpc_rec.width = spl_in->basic_in.custom_width; + mpc_rec.x = spl_in->basic_in.custom_x; + mpc_rec.height = plane_clip_rec->height; + mpc_rec.y = plane_clip_rec->y; + } else if (use_recout_width_aligned) { mpc_rec.width = recout_width_align; if ((mpc_rec.width * (mpc_slice_idx + 1)) > plane_clip_rec->width) { mpc_rec.width = plane_clip_rec->width % recout_width_align; @@ -219,7 +224,7 @@ static struct spl_rect calculate_mpc_slice_in_timing_active( /* extra pixels in the division remainder need to go to pipes after * the extra pixel index minus one(epimo) defined here as: */ - if (mpc_slice_idx > epimo) { + if (mpc_slice_idx > epimo && spl_in->basic_in.custom_width == 0) { mpc_rec.x += mpc_slice_idx - epimo - 1; mpc_rec.width += 1; } @@ -252,10 +257,10 @@ static struct spl_rect calculate_odm_slice_in_timing_active(struct spl_in *spl_i odm_rec.x = odm_slice_width * odm_slice_idx; odm_rec.width = is_last_odm_slice ? - /* last slice width is the reminder of h_active */ - h_active - odm_slice_width * (odm_slice_count - 1) : - /* odm slice width is the floor of h_active / count */ - odm_slice_width; + /* last slice width is the reminder of h_active */ + h_active - odm_slice_width * (odm_slice_count - 1) : + /* odm slice width is the floor of h_active / count */ + odm_slice_width; odm_rec.y = 0; odm_rec.height = v_active; @@ -884,7 +889,9 @@ static bool spl_get_isharp_en(struct spl_in *spl_in, /* Calculate number of tap with adaptive scaling off */ static void spl_get_taps_non_adaptive_scaler( - struct spl_scratch *spl_scratch, const struct spl_taps *in_taps, bool always_scale) + struct spl_scratch *spl_scratch, + const struct spl_taps *in_taps, + bool is_subsampled) { bool check_max_downscale = false; @@ -945,14 +952,15 @@ static void spl_get_taps_non_adaptive_scaler( SPL_ASSERT(check_max_downscale); - if (IDENTITY_RATIO(spl_scratch->scl_data.ratios.horz) && !always_scale) + if (IDENTITY_RATIO(spl_scratch->scl_data.ratios.horz)) spl_scratch->scl_data.taps.h_taps = 1; - if (IDENTITY_RATIO(spl_scratch->scl_data.ratios.vert) && !always_scale) + if (IDENTITY_RATIO(spl_scratch->scl_data.ratios.vert)) spl_scratch->scl_data.taps.v_taps = 1; - if (IDENTITY_RATIO(spl_scratch->scl_data.ratios.horz_c) && !always_scale) + if (IDENTITY_RATIO(spl_scratch->scl_data.ratios.horz_c) && !is_subsampled) spl_scratch->scl_data.taps.h_taps_c = 1; - if (IDENTITY_RATIO(spl_scratch->scl_data.ratios.vert_c) && !always_scale) + if (IDENTITY_RATIO(spl_scratch->scl_data.ratios.vert_c) && !is_subsampled) spl_scratch->scl_data.taps.v_taps_c = 1; + } /* Calculate optimal number of taps */ @@ -965,15 +973,13 @@ static bool spl_get_optimal_number_of_taps( unsigned int max_taps_y, max_taps_c; unsigned int min_taps_y, min_taps_c; enum lb_memory_config lb_config; - bool skip_easf = false; - bool always_scale = spl_in->basic_out.always_scale; + bool skip_easf = false; bool is_subsampled = spl_is_subsampled_format(spl_in->basic_in.format); - if (spl_scratch->scl_data.viewport.width > spl_scratch->scl_data.h_active && max_downscale_src_width != 0 && spl_scratch->scl_data.viewport.width > max_downscale_src_width) { - spl_get_taps_non_adaptive_scaler(spl_scratch, in_taps, always_scale); + spl_get_taps_non_adaptive_scaler(spl_scratch, in_taps, is_subsampled); *enable_easf_v = false; *enable_easf_h = false; *enable_isharp = false; @@ -982,7 +988,7 @@ static bool spl_get_optimal_number_of_taps( /* Disable adaptive scaler and sharpener when integer scaling is enabled */ if (spl_in->scaling_quality.integer_scaling) { - spl_get_taps_non_adaptive_scaler(spl_scratch, in_taps, always_scale); + spl_get_taps_non_adaptive_scaler(spl_scratch, in_taps, is_subsampled); *enable_easf_v = false; *enable_easf_h = false; *enable_isharp = false; @@ -997,8 +1003,9 @@ static bool spl_get_optimal_number_of_taps( * From programming guide: taps = min{ ceil(2*H_RATIO,1), 8} for downscaling * taps = 4 for upscaling */ - if (skip_easf) - spl_get_taps_non_adaptive_scaler(spl_scratch, in_taps, always_scale); + if (skip_easf) { + spl_get_taps_non_adaptive_scaler(spl_scratch, in_taps, is_subsampled); + } else { if (spl_is_video_format(spl_in->basic_in.format)) { spl_scratch->scl_data.taps.h_taps = 6; @@ -1124,7 +1131,6 @@ static bool spl_get_optimal_number_of_taps( (IDENTITY_RATIO(spl_scratch->scl_data.ratios.vert))) { spl_scratch->scl_data.taps.h_taps = 1; spl_scratch->scl_data.taps.v_taps = 1; - if (IDENTITY_RATIO(spl_scratch->scl_data.ratios.horz_c) && !is_subsampled) spl_scratch->scl_data.taps.h_taps_c = 1; @@ -1149,6 +1155,7 @@ static bool spl_get_optimal_number_of_taps( if ((!*enable_easf_v) && !is_subsampled && (IDENTITY_RATIO(spl_scratch->scl_data.ratios.vert_c))) spl_scratch->scl_data.taps.v_taps_c = 1; + } } return true; diff --git a/drivers/gpu/drm/amd/display/dc/sspl/dc_spl_types.h b/drivers/gpu/drm/amd/display/dc/sspl/dc_spl_types.h index 36a284305a70..23d254dea18f 100644 --- a/drivers/gpu/drm/amd/display/dc/sspl/dc_spl_types.h +++ b/drivers/gpu/drm/amd/display/dc/sspl/dc_spl_types.h @@ -460,6 +460,8 @@ struct basic_in { enum spl_color_space color_space; // Color Space unsigned int max_luminance; // Max Luminance TODO: Is determined in dc_hw_sequencer.c is_sdr bool film_grain_applied; // Film Grain Applied // TODO: To check from where to get this? + int custom_width; // Width for non-standard segmentation - used when != 0 + int custom_x; // Start x for non-standard segmentation - used when custom_width != 0 }; // Basic output information diff --git a/drivers/gpu/drm/amd/display/dmub/dmub_srv.h b/drivers/gpu/drm/amd/display/dmub/dmub_srv.h index 3f3fa1b6a69e..0bafb6710761 100644 --- a/drivers/gpu/drm/amd/display/dmub/dmub_srv.h +++ b/drivers/gpu/drm/amd/display/dmub/dmub_srv.h @@ -129,7 +129,9 @@ enum dmub_window_id { DMUB_WINDOW_5_TRACEBUFF, DMUB_WINDOW_6_FW_STATE, DMUB_WINDOW_7_SCRATCH_MEM, + DMUB_WINDOW_IB_MEM, DMUB_WINDOW_SHARED_STATE, + DMUB_WINDOW_LSDMA_BUFFER, DMUB_WINDOW_TOTAL, }; @@ -355,6 +357,7 @@ struct dmub_diagnostic_data { uint8_t is_traceport_en : 1; uint8_t is_cw0_enabled : 1; uint8_t is_cw6_enabled : 1; + uint8_t is_pwait : 1; }; struct dmub_srv_inbox { @@ -539,6 +542,7 @@ struct dmub_srv { uint32_t fw_version; bool is_virtual; struct dmub_fb scratch_mem_fb; + struct dmub_fb ib_mem_gart; volatile struct dmub_shared_state_feature_block *shared_state; volatile const struct dmub_fw_state *fw_state; @@ -576,6 +580,7 @@ struct dmub_srv { enum dmub_srv_power_state_type power_state; struct dmub_diagnostic_data debug; + struct dmub_fb lsdma_rb_fb; }; /** @@ -602,14 +607,6 @@ struct dmub_notification { }; }; -/* enum dmub_ips_mode - IPS mode identifier */ -enum dmub_ips_mode { - DMUB_IPS_MODE_IPS1_MAX = 0, - DMUB_IPS_MODE_IPS2, - DMUB_IPS_MODE_IPS1_RCG, - DMUB_IPS_MODE_IPS1_ONO2_ON -}; - /** * DMUB firmware version helper macro - useful for checking if the version * of a firmware to know if feature or functionality is supported or present. diff --git a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h index b66bd10cdc9b..5cf5dd5831fc 100644 --- a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h @@ -104,6 +104,14 @@ */ #define DMUB_MAX_FPO_STREAMS 4 +/* Define to ensure that the "common" members always appear in the same + * order in different structs for back compat purposes + */ +#define COMMON_STREAM_STATIC_SUB_STATE \ + struct dmub_fams2_cmd_legacy_stream_static_state legacy; \ + struct dmub_fams2_cmd_subvp_stream_static_state subvp; \ + struct dmub_fams2_cmd_drr_stream_static_state drr; + /* Maximum number of streams on any ASIC. */ #define DMUB_MAX_STREAMS 6 @@ -291,6 +299,31 @@ union dmub_addr { } u; /*<< Low/high bit access */ uint64_t quad_part; /*<< 64 bit address */ }; + +/* Flattened structure containing SOC BB parameters stored in the VBIOS + * It is not practical to store the entire bounding box in VBIOS since the bounding box struct can gain new parameters. + * This also prevents alighment issues when new parameters are added to the SoC BB. + * The following parameters should be added since these values can't be obtained elsewhere: + * -dml2_soc_power_management_parameters + * -dml2_soc_vmin_clock_limits + */ +struct dmub_soc_bb_params { + uint32_t dram_clk_change_blackout_ns; + uint32_t dram_clk_change_read_only_ns; + uint32_t dram_clk_change_write_only_ns; + uint32_t fclk_change_blackout_ns; + uint32_t g7_ppt_blackout_ns; + uint32_t stutter_enter_plus_exit_latency_ns; + uint32_t stutter_exit_latency_ns; + uint32_t z8_stutter_enter_plus_exit_latency_ns; + uint32_t z8_stutter_exit_latency_ns; + uint32_t z8_min_idle_time_ns; + uint32_t type_b_dram_clk_change_blackout_ns; + uint32_t type_b_ppt_blackout_ns; + uint32_t vmin_limit_dispclk_khz; + uint32_t vmin_limit_dcfclk_khz; + uint32_t g7_temperature_read_blackout_ns; +}; #pragma pack(pop) /** @@ -762,6 +795,17 @@ enum dmub_ips_rcg_disable_type { #define DMUB_IPS1_COMMIT_MASK 0x00000004 #define DMUB_IPS2_COMMIT_MASK 0x00000008 +enum dmub_ips_comand_type { + /** + * Start/stop IPS residency measurements for a given IPS mode + */ + DMUB_CMD__IPS_RESIDENCY_CNTL = 0, + /** + * Query IPS residency information for a given IPS mode + */ + DMUB_CMD__IPS_QUERY_RESIDENCY_INFO = 1, +}; + /** * union dmub_fw_boot_options - Boot option definitions for SCRATCH14 */ @@ -856,7 +900,9 @@ union dmub_shared_state_ips_driver_signals { uint32_t allow_ips0_rcg : 1; /**< 1 is IPS0 RCG is allowed */ uint32_t allow_ips1_rcg : 1; /**< 1 is IPS1 RCG is allowed */ uint32_t allow_ips1z8 : 1; /**< 1 is IPS1 Z8 Retention is allowed */ - uint32_t reserved_bits : 24; /**< Reversed bits */ + uint32_t allow_dynamic_ips1 : 1; /**< 1 if IPS1 is allowed in dynamic use cases such as VPB */ + uint32_t allow_dynamic_ips1_z8: 1; /**< 1 if IPS1 z8 ret is allowed in dynamic use cases such as VPB */ + uint32_t reserved_bits : 22; /**< Reversed bits */ } bits; uint32_t all; }; @@ -1508,6 +1554,16 @@ enum dmub_cmd_type { */ DMUB_CMD__FUSED_IO = 89, + /** + * Command type used for all LSDMA commands. + */ + DMUB_CMD__LSDMA = 90, + + /** + * Command type use for all IPS commands. + */ + DMUB_CMD__IPS = 91, + DMUB_CMD__VBIOS = 128, }; @@ -1918,6 +1974,121 @@ struct dmub_rb_cmd_fams2_flip { struct dmub_fams2_flip_info flip_info; }; +struct dmub_cmd_lsdma_data { + union { + struct lsdma_init_data { + union dmub_addr gpu_addr_base; + uint32_t ring_size; + } init_data; + struct lsdma_tiled_copy_data { + uint32_t src_addr_lo; + uint32_t src_addr_hi; + uint32_t dst_addr_lo; + uint32_t dst_addr_hi; + + uint32_t src_x : 16; + uint32_t src_y : 16; + + uint32_t src_width : 16; + uint32_t src_height : 16; + + uint32_t dst_x : 16; + uint32_t dst_y : 16; + + uint32_t dst_width : 16; + uint32_t dst_height : 16; + + uint32_t rect_x : 16; + uint32_t rect_y : 16; + + uint32_t src_swizzle_mode : 5; + uint32_t src_mip_max : 5; + uint32_t src_mip_id : 5; + uint32_t dst_mip_max : 5; + uint32_t dst_swizzle_mode : 5; + uint32_t dst_mip_id : 5; + uint32_t tmz : 1; + uint32_t dcc : 1; + + uint32_t data_format : 6; + uint32_t padding1 : 4; + uint32_t dst_element_size : 3; + uint32_t num_type : 3; + uint32_t src_element_size : 3; + uint32_t write_compress : 2; + uint32_t cache_policy_dst : 2; + uint32_t cache_policy_src : 2; + uint32_t read_compress : 2; + uint32_t src_dim : 2; + uint32_t dst_dim : 2; + uint32_t max_uncom : 1; + + uint32_t max_com : 2; + uint32_t padding : 30; + } tiled_copy_data; + struct lsdma_linear_copy_data { + uint32_t count : 30; + uint32_t cache_policy_dst : 2; + + uint32_t tmz : 1; + uint32_t cache_policy_src : 2; + uint32_t padding : 29; + + uint32_t src_lo; + uint32_t src_hi; + uint32_t dst_lo; + uint32_t dst_hi; + } linear_copy_data; + struct lsdma_reg_write_data { + uint32_t reg_addr; + uint32_t reg_data; + } reg_write_data; + struct lsdma_pio_copy_data { + union { + struct { + uint32_t byte_count : 26; + uint32_t src_loc : 1; + uint32_t dst_loc : 1; + uint32_t src_addr_inc : 1; + uint32_t dst_addr_inc : 1; + uint32_t overlap_disable : 1; + uint32_t constant_fill : 1; + } fields; + uint32_t raw; + } packet; + uint32_t src_lo; + uint32_t src_hi; + uint32_t dst_lo; + uint32_t dst_hi; + } pio_copy_data; + struct lsdma_pio_constfill_data { + union { + struct { + uint32_t byte_count : 26; + uint32_t src_loc : 1; + uint32_t dst_loc : 1; + uint32_t src_addr_inc : 1; + uint32_t dst_addr_inc : 1; + uint32_t overlap_disable : 1; + uint32_t constant_fill : 1; + } fields; + uint32_t raw; + } packet; + uint32_t dst_lo; + uint32_t dst_hi; + uint32_t data; + } pio_constfill_data; + + uint32_t all[14]; + } u; + +}; + +struct dmub_rb_cmd_lsdma { + struct dmub_cmd_header header; + struct dmub_cmd_lsdma_data lsdma_data; +}; + struct dmub_optc_state_v2 { uint32_t v_total_min; uint32_t v_total_max; @@ -1949,6 +2120,28 @@ enum fams2_stream_type { FAMS2_STREAM_TYPE_SUBVP = 4, }; +struct dmub_rect16 { + /** + * Dirty rect x offset. + */ + uint16_t x; + + /** + * Dirty rect y offset. + */ + uint16_t y; + + /** + * Dirty rect width. + */ + uint16_t width; + + /** + * Dirty rect height. + */ + uint16_t height; +}; + /* static stream state */ struct dmub_fams2_legacy_stream_static_state { uint8_t vactive_det_fill_delay_otg_vlines; @@ -2021,11 +2214,13 @@ union dmub_fams2_stream_static_sub_state { }; //v0 union dmub_fams2_cmd_stream_static_sub_state { - struct dmub_fams2_cmd_legacy_stream_static_state legacy; - struct dmub_fams2_cmd_subvp_stream_static_state subvp; - struct dmub_fams2_cmd_drr_stream_static_state drr; + COMMON_STREAM_STATIC_SUB_STATE }; //v1 +union dmub_fams2_stream_static_sub_state_v2 { + COMMON_STREAM_STATIC_SUB_STATE +}; //v2 + struct dmub_fams2_stream_static_state { enum fams2_stream_type type; uint32_t otg_vline_time_ns; @@ -2091,7 +2286,7 @@ struct dmub_fams2_cmd_stream_static_base_state { struct dmub_fams2_stream_static_state_v1 { struct dmub_fams2_cmd_stream_static_base_state base; - union dmub_fams2_cmd_stream_static_sub_state sub_state; + union dmub_fams2_stream_static_sub_state_v2 sub_state; }; //v1 /** @@ -2139,6 +2334,11 @@ union dmub_cmd_fams2_config { } stream_v1; //v1 }; +struct dmub_fams2_config_v2 { + struct dmub_cmd_fams2_global_config global; + struct dmub_fams2_stream_static_state_v1 stream_v1[DMUB_MAX_STREAMS]; //v1 +}; + /** * DMUB rb command definition for FAMS2 (merged SubVP, FPO, Legacy) */ @@ -2148,6 +2348,22 @@ struct dmub_rb_cmd_fams2 { }; /** + * Indirect buffer descriptor + */ +struct dmub_ib_data { + union dmub_addr src; // location of indirect buffer in memory + uint16_t size; // indirect buffer size in bytes +}; + +/** + * DMUB rb command definition for commands passed over indirect buffer + */ +struct dmub_rb_cmd_ib { + struct dmub_cmd_header header; + struct dmub_ib_data ib_data; +}; + +/** * enum dmub_cmd_idle_opt_type - Idle optimization command type. */ enum dmub_cmd_idle_opt_type { @@ -2170,6 +2386,11 @@ enum dmub_cmd_idle_opt_type { * DCN hardware notify power state. */ DMUB_CMD__IDLE_OPT_SET_DC_POWER_STATE = 3, + + /** + * DCN notify to release HW. + */ + DMUB_CMD__IDLE_OPT_RELEASE_HW = 4, }; /** @@ -2315,7 +2536,8 @@ struct dmub_dig_transmitter_control_data_v1_7 { uint8_t connobj_id; /**< Connector Object Id defined in ObjectId.h */ uint8_t HPO_instance; /**< HPO instance (0: inst0, 1: inst1) */ uint8_t reserved1; /**< For future use */ - uint8_t reserved2[3]; /**< For future use */ + uint8_t skip_phy_ssc_reduction; + uint8_t reserved2[2]; /**< For future use */ uint32_t reserved3[11]; /**< For future use */ }; @@ -2933,6 +3155,7 @@ enum dmub_cmd_fams_type { DMUB_CMD__FAMS2_CONFIG = 4, DMUB_CMD__FAMS2_DRR_UPDATE = 5, DMUB_CMD__FAMS2_FLIP = 6, + DMUB_CMD__FAMS2_IB_CONFIG = 7, }; /** @@ -4416,6 +4639,37 @@ enum dmub_cmd_abm_type { DMUB_CMD__ABM_GET_HISTOGRAM_DATA = 11, }; +/** + * LSDMA command sub-types. + */ +enum dmub_cmd_lsdma_type { + /** + * Initialize parameters for LSDMA. + * Ring buffer is mapped to the ring buffer + */ + DMUB_CMD__LSDMA_INIT_CONFIG = 0, + /** + * LSDMA copies data from source to destination linearly + */ + DMUB_CMD__LSDMA_LINEAR_COPY = 1, + /** + * Send the tiled-to-tiled copy command + */ + DMUB_CMD__LSDMA_TILED_TO_TILED_COPY = 2, + /** + * Send the poll reg write command + */ + DMUB_CMD__LSDMA_POLL_REG_WRITE = 3, + /** + * Send the pio copy command + */ + DMUB_CMD__LSDMA_PIO_COPY = 4, + /** + * Send the pio constfill command + */ + DMUB_CMD__LSDMA_PIO_CONSTFILL = 5, +}; + struct abm_ace_curve { /** * @offsets: ACE curve offsets. @@ -5621,6 +5875,59 @@ struct dmub_rb_cmd_assr_enable { }; /** + * Current definition of "ips_mode" from driver + */ +enum ips_residency_mode { + IPS_RESIDENCY__IPS1_MAX, + IPS_RESIDENCY__IPS2, + IPS_RESIDENCY__IPS1_RCG, + IPS_RESIDENCY__IPS1_ONO2_ON, +}; + +#define NUM_IPS_HISTOGRAM_BUCKETS 16 + +/** + * IPS residency statistics to be sent to driver - subset of struct dmub_ips_residency_stats + */ +struct dmub_ips_residency_info { + uint32_t residency_millipercent; + uint32_t entry_counter; + uint32_t histogram[NUM_IPS_HISTOGRAM_BUCKETS]; + uint64_t total_time_us; + uint64_t total_inactive_time_us; +}; + +/** + * Data passed from driver to FW in a DMUB_CMD__IPS_RESIDENCY_CNTL command. + */ +struct dmub_cmd_ips_residency_cntl_data { + uint8_t panel_inst; + uint8_t start_measurement; + uint8_t padding[2]; // align to 4-byte boundary +}; + +struct dmub_rb_cmd_ips_residency_cntl { + struct dmub_cmd_header header; + struct dmub_cmd_ips_residency_cntl_data cntl_data; +}; + +/** + * Data passed from FW to driver in a DMUB_CMD__IPS_QUERY_RESIDENCY_INFO command. + */ +struct dmub_cmd_ips_query_residency_info_data { + union dmub_addr dest; + uint32_t size; + uint32_t ips_mode; + uint8_t panel_inst; + uint8_t padding[3]; // align to 4-byte boundary +}; + +struct dmub_rb_cmd_ips_query_residency_info { + struct dmub_cmd_header header; + struct dmub_cmd_ips_query_residency_info_data info_data; +}; + +/** * union dmub_rb_cmd - DMUB inbox command. */ union dmub_rb_cmd { @@ -5926,13 +6233,25 @@ union dmub_rb_cmd { * Definition of a DMUB_CMD__PSP_ASSR_ENABLE command. */ struct dmub_rb_cmd_assr_enable assr_enable; + struct dmub_rb_cmd_fams2 fams2_config; + struct dmub_rb_cmd_ib ib_fams2_config; + struct dmub_rb_cmd_fams2_drr_update fams2_drr_update; struct dmub_rb_cmd_fams2_flip fams2_flip; struct dmub_rb_cmd_fused_io fused_io; + + /** + * Definition of a DMUB_CMD__LSDMA command. + */ + struct dmub_rb_cmd_lsdma lsdma; + + struct dmub_rb_cmd_ips_residency_cntl ips_residency_cntl; + + struct dmub_rb_cmd_ips_query_residency_info ips_query_residency_info; }; /** diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c index a308bd604677..3f38db752b84 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c @@ -416,7 +416,7 @@ uint32_t dmub_dcn31_get_current_time(struct dmub_srv *dmub) void dmub_dcn31_get_diagnostic_data(struct dmub_srv *dmub) { - uint32_t is_dmub_enabled, is_soft_reset, is_sec_reset; + uint32_t is_dmub_enabled, is_soft_reset, is_sec_reset, is_pwait; uint32_t is_traceport_enabled, is_cw0_enabled, is_cw6_enabled; struct dmub_timeout_info timeout = {0}; @@ -466,6 +466,9 @@ void dmub_dcn31_get_diagnostic_data(struct dmub_srv *dmub) REG_GET(DMCUB_CNTL, DMCUB_ENABLE, &is_dmub_enabled); dmub->debug.is_dmcub_enabled = is_dmub_enabled; + REG_GET(DMCUB_CNTL, DMCUB_PWAIT_MODE_STATUS, &is_pwait); + dmub->debug.is_pwait = is_pwait; + REG_GET(DMCUB_CNTL2, DMCUB_SOFT_RESET, &is_soft_reset); dmub->debug.is_dmcub_soft_reset = is_soft_reset; diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c index 72a0f078cd1a..2228d62adc7e 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c @@ -92,19 +92,15 @@ void dmub_dcn35_reset(struct dmub_srv *dmub) uint32_t in_reset, is_enabled, scratch, i, pwait_mode; REG_GET(DMCUB_CNTL2, DMCUB_SOFT_RESET, &in_reset); + REG_GET(DMCUB_CNTL, DMCUB_ENABLE, &is_enabled); - if (in_reset == 0) { + if (in_reset == 0 && is_enabled != 0) { cmd.bits.status = 1; cmd.bits.command_code = DMUB_GPINT__STOP_FW; cmd.bits.param = 0; dmub->hw_funcs.set_gpint(dmub, cmd); - /** - * Timeout covers both the ACK and the wait - * for remaining work to finish. - */ - for (i = 0; i < timeout; ++i) { if (dmub->hw_funcs.is_gpint_acked(dmub, cmd)) break; @@ -130,11 +126,9 @@ void dmub_dcn35_reset(struct dmub_srv *dmub) /* Force reset in case we timed out, DMCUB is likely hung. */ } - REG_GET(DMCUB_CNTL, DMCUB_ENABLE, &is_enabled); - if (is_enabled) { REG_UPDATE(DMCUB_CNTL2, DMCUB_SOFT_RESET, 1); - REG_UPDATE(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET, 1); + udelay(1); REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); } @@ -160,11 +154,7 @@ void dmub_dcn35_reset_release(struct dmub_srv *dmub) LONO_SOCCLK_GATE_DISABLE, 1, LONO_DMCUBCLK_GATE_DISABLE, 1); - REG_UPDATE(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET, 1); - udelay(1); REG_UPDATE_2(DMCUB_CNTL, DMCUB_ENABLE, 1, DMCUB_TRACEPORT_EN, 1); - REG_UPDATE(DMCUB_CNTL2, DMCUB_SOFT_RESET, 1); - udelay(1); REG_UPDATE(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET, 0); REG_UPDATE(DMCUB_CNTL2, DMCUB_SOFT_RESET, 0); } @@ -464,7 +454,7 @@ uint32_t dmub_dcn35_get_current_time(struct dmub_srv *dmub) void dmub_dcn35_get_diagnostic_data(struct dmub_srv *dmub) { - uint32_t is_dmub_enabled, is_soft_reset; + uint32_t is_dmub_enabled, is_soft_reset, is_pwait; uint32_t is_traceport_enabled, is_cw6_enabled; struct dmub_timeout_info timeout = {0}; @@ -515,6 +505,9 @@ void dmub_dcn35_get_diagnostic_data(struct dmub_srv *dmub) REG_GET(DMCUB_CNTL, DMCUB_ENABLE, &is_dmub_enabled); dmub->debug.is_dmcub_enabled = is_dmub_enabled; + REG_GET(DMCUB_CNTL, DMCUB_PWAIT_MODE_STATUS, &is_pwait); + dmub->debug.is_pwait = is_pwait; + REG_GET(DMCUB_CNTL2, DMCUB_SOFT_RESET, &is_soft_reset); dmub->debug.is_dmcub_soft_reset = is_soft_reset; diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c index 2575dbc448f7..b31adbd0d685 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c @@ -413,7 +413,7 @@ uint32_t dmub_dcn401_get_current_time(struct dmub_srv *dmub) void dmub_dcn401_get_diagnostic_data(struct dmub_srv *dmub) { - uint32_t is_dmub_enabled, is_soft_reset, is_sec_reset; + uint32_t is_dmub_enabled, is_soft_reset, is_sec_reset, is_pwait; uint32_t is_traceport_enabled, is_cw0_enabled, is_cw6_enabled; struct dmub_timeout_info timeout = {0}; @@ -464,6 +464,9 @@ void dmub_dcn401_get_diagnostic_data(struct dmub_srv *dmub) REG_GET(DMCUB_CNTL, DMCUB_ENABLE, &is_dmub_enabled); dmub->debug.is_dmcub_enabled = is_dmub_enabled; + REG_GET(DMCUB_CNTL, DMCUB_PWAIT_MODE_STATUS, &is_pwait); + dmub->debug.is_pwait = is_pwait; + REG_GET(DMCUB_CNTL2, DMCUB_SOFT_RESET, &is_soft_reset); dmub->debug.is_dmcub_soft_reset = is_soft_reset; diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c index acca7943a8c8..b17a19400c06 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c @@ -65,6 +65,12 @@ /* Default scratch mem size. */ #define DMUB_SCRATCH_MEM_SIZE (1024) +/* Default indirect buffer size. */ +#define DMUB_IB_MEM_SIZE (1280) + +/* Default LSDMA ring buffer size. */ +#define DMUB_LSDMA_RB_SIZE (64 * 1024) + /* Number of windows in use. */ #define DMUB_NUM_WINDOWS (DMUB_WINDOW_TOTAL) /* Base addresses. */ @@ -559,7 +565,9 @@ enum dmub_status window_sizes[DMUB_WINDOW_5_TRACEBUFF] = trace_buffer_size; window_sizes[DMUB_WINDOW_6_FW_STATE] = fw_state_size; window_sizes[DMUB_WINDOW_7_SCRATCH_MEM] = DMUB_SCRATCH_MEM_SIZE; + window_sizes[DMUB_WINDOW_IB_MEM] = DMUB_IB_MEM_SIZE; window_sizes[DMUB_WINDOW_SHARED_STATE] = max(DMUB_FW_HEADER_SHARED_STATE_SIZE, shared_state_size); + window_sizes[DMUB_WINDOW_LSDMA_BUFFER] = DMUB_LSDMA_RB_SIZE; out->fb_size = dmub_srv_calc_regions_for_memory_type(params, out, window_sizes, DMUB_WINDOW_MEMORY_TYPE_FB); @@ -645,6 +653,7 @@ enum dmub_status dmub_srv_hw_init(struct dmub_srv *dmub, struct dmub_fb *tracebuff_fb = params->fb[DMUB_WINDOW_5_TRACEBUFF]; struct dmub_fb *fw_state_fb = params->fb[DMUB_WINDOW_6_FW_STATE]; struct dmub_fb *scratch_mem_fb = params->fb[DMUB_WINDOW_7_SCRATCH_MEM]; + struct dmub_fb *ib_mem_gart = params->fb[DMUB_WINDOW_IB_MEM]; struct dmub_fb *shared_state_fb = params->fb[DMUB_WINDOW_SHARED_STATE]; struct dmub_rb_init_params rb_params, outbox0_rb_params; @@ -655,7 +664,7 @@ enum dmub_status dmub_srv_hw_init(struct dmub_srv *dmub, return DMUB_STATUS_INVALID; if (!inst_fb || !stack_fb || !data_fb || !bios_fb || !mail_fb || - !tracebuff_fb || !fw_state_fb || !scratch_mem_fb) { + !tracebuff_fb || !fw_state_fb || !scratch_mem_fb || !ib_mem_gart) { ASSERT(0); return DMUB_STATUS_INVALID; } @@ -741,6 +750,8 @@ enum dmub_status dmub_srv_hw_init(struct dmub_srv *dmub, dmub->scratch_mem_fb = *scratch_mem_fb; + dmub->ib_mem_gart = *ib_mem_gart; + if (dmub->hw_funcs.setup_windows) dmub->hw_funcs.setup_windows(dmub, &cw2, &cw3, &cw4, &cw5, &cw6, ®ion6); diff --git a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c index 3ba9b62ba70b..71efd2770c99 100644 --- a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c +++ b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c @@ -147,7 +147,7 @@ unsigned int mod_freesync_calc_v_total_from_refresh( ((unsigned int)(div64_u64((1000000000ULL * 1000000), refresh_in_uhz))); - if (MICRO_HZ_TO_HZ(refresh_in_uhz) <= stream->timing.min_refresh_in_uhz) { + if (refresh_in_uhz <= stream->timing.min_refresh_in_uhz) { /* When the target refresh rate is the minimum panel refresh rate, * round down the vtotal value to avoid stretching vblank over * panel's vtotal boundary. @@ -155,6 +155,14 @@ unsigned int mod_freesync_calc_v_total_from_refresh( v_total = div64_u64(div64_u64(((unsigned long long)( frame_duration_in_ns) * (stream->timing.pix_clk_100hz / 10)), stream->timing.h_total), 1000000); + } else if (refresh_in_uhz >= stream->timing.max_refresh_in_uhz) { + /* When the target refresh rate is the maximum panel refresh rate + * round up the vtotal value to prevent off-by-one error causing + * v_total_min to be below the panel's lower bound + */ + v_total = div64_u64(div64_u64(((unsigned long long)( + frame_duration_in_ns) * (stream->timing.pix_clk_100hz / 10)), + stream->timing.h_total) + (1000000 - 1), 1000000); } else { v_total = div64_u64(div64_u64(((unsigned long long)( frame_duration_in_ns) * (stream->timing.pix_clk_100hz / 10)), diff --git a/drivers/gpu/drm/amd/include/amd_shared.h b/drivers/gpu/drm/amd/include/amd_shared.h index 11374a2cbab8..bfb446736ca8 100644 --- a/drivers/gpu/drm/amd/include/amd_shared.h +++ b/drivers/gpu/drm/amd/include/amd_shared.h @@ -396,6 +396,7 @@ enum amd_dpm_forced_level; * (such as allocating any required memory) * @suspend: handles IP specific hw/sw changes for suspend * @resume: handles IP specific hw/sw changes for resume + * @complete: handles IP specific changes after resume * @is_idle: returns current IP block idle status * @wait_for_idle: poll for idle * @check_soft_reset: check soft reset the IP block @@ -427,6 +428,7 @@ struct amd_ip_funcs { int (*prepare_suspend)(struct amdgpu_ip_block *ip_block); int (*suspend)(struct amdgpu_ip_block *ip_block); int (*resume)(struct amdgpu_ip_block *ip_block); + void (*complete)(struct amdgpu_ip_block *ip_block); bool (*is_idle)(struct amdgpu_ip_block *ip_block); int (*wait_for_idle)(struct amdgpu_ip_block *ip_block); bool (*check_soft_reset)(struct amdgpu_ip_block *ip_block); diff --git a/drivers/gpu/drm/amd/include/kgd_pp_interface.h b/drivers/gpu/drm/amd/include/kgd_pp_interface.h index f4d914dc731f..e2b1ea7467b0 100644 --- a/drivers/gpu/drm/amd/include/kgd_pp_interface.h +++ b/drivers/gpu/drm/amd/include/kgd_pp_interface.h @@ -108,6 +108,8 @@ enum pp_clock_type { PP_VCLK1, PP_DCLK, PP_DCLK1, + PP_ISPICLK, + PP_ISPXCLK, OD_SCLK, OD_MCLK, OD_VDDC_CURVE, diff --git a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c index 5c1cbdc122d2..71d986dd7a6e 100644 --- a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c +++ b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c @@ -98,6 +98,7 @@ int amdgpu_dpm_set_powergating_by_smu(struct amdgpu_device *adev, case AMD_IP_BLOCK_TYPE_GMC: case AMD_IP_BLOCK_TYPE_ACP: case AMD_IP_BLOCK_TYPE_VPE: + case AMD_IP_BLOCK_TYPE_ISP: if (pp_funcs && pp_funcs->set_powergating_by_smu) ret = (pp_funcs->set_powergating_by_smu( (adev)->powerplay.pp_handle, block_type, gate, 0)); @@ -852,22 +853,16 @@ int amdgpu_dpm_set_soft_freq_range(struct amdgpu_device *adev, uint32_t max) { struct smu_context *smu = adev->powerplay.pp_handle; - int ret = 0; - - if (type != PP_SCLK) - return -EINVAL; if (!is_support_sw_smu(adev)) return -EOPNOTSUPP; - mutex_lock(&adev->pm.mutex); - ret = smu_set_soft_freq_range(smu, - SMU_SCLK, + guard(mutex)(&adev->pm.mutex); + + return smu_set_soft_freq_range(smu, + type, min, max); - mutex_unlock(&adev->pm.mutex); - - return ret; } int amdgpu_dpm_write_watermarks_table(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/pm/legacy-dpm/kv_dpm.c b/drivers/gpu/drm/amd/pm/legacy-dpm/kv_dpm.c index 34e71727b27d..307ebf7e3226 100644 --- a/drivers/gpu/drm/amd/pm/legacy-dpm/kv_dpm.c +++ b/drivers/gpu/drm/amd/pm/legacy-dpm/kv_dpm.c @@ -1242,7 +1242,7 @@ static void kv_dpm_enable_bapm(void *handle, bool enable) if (pi->bapm_enable) { ret = amdgpu_kv_smc_bapm_enable(adev, enable); if (ret) - DRM_ERROR("amdgpu_kv_smc_bapm_enable failed\n"); + drm_err(adev_to_drm(adev), "amdgpu_kv_smc_bapm_enable failed\n"); } } @@ -1266,40 +1266,40 @@ static int kv_dpm_enable(struct amdgpu_device *adev) ret = kv_process_firmware_header(adev); if (ret) { - DRM_ERROR("kv_process_firmware_header failed\n"); + drm_err(adev_to_drm(adev), "kv_process_firmware_header failed\n"); return ret; } kv_init_fps_limits(adev); kv_init_graphics_levels(adev); ret = kv_program_bootup_state(adev); if (ret) { - DRM_ERROR("kv_program_bootup_state failed\n"); + drm_err(adev_to_drm(adev), "kv_program_bootup_state failed\n"); return ret; } kv_calculate_dfs_bypass_settings(adev); ret = kv_upload_dpm_settings(adev); if (ret) { - DRM_ERROR("kv_upload_dpm_settings failed\n"); + drm_err(adev_to_drm(adev), "kv_upload_dpm_settings failed\n"); return ret; } ret = kv_populate_uvd_table(adev); if (ret) { - DRM_ERROR("kv_populate_uvd_table failed\n"); + drm_err(adev_to_drm(adev), "kv_populate_uvd_table failed\n"); return ret; } ret = kv_populate_vce_table(adev); if (ret) { - DRM_ERROR("kv_populate_vce_table failed\n"); + drm_err(adev_to_drm(adev), "kv_populate_vce_table failed\n"); return ret; } ret = kv_populate_samu_table(adev); if (ret) { - DRM_ERROR("kv_populate_samu_table failed\n"); + drm_err(adev_to_drm(adev), "kv_populate_samu_table failed\n"); return ret; } ret = kv_populate_acp_table(adev); if (ret) { - DRM_ERROR("kv_populate_acp_table failed\n"); + drm_err(adev_to_drm(adev), "kv_populate_acp_table failed\n"); return ret; } kv_program_vc(adev); @@ -1310,39 +1310,39 @@ static int kv_dpm_enable(struct amdgpu_device *adev) if (pi->enable_auto_thermal_throttling) { ret = kv_enable_auto_thermal_throttling(adev); if (ret) { - DRM_ERROR("kv_enable_auto_thermal_throttling failed\n"); + drm_err(adev_to_drm(adev), "kv_enable_auto_thermal_throttling failed\n"); return ret; } } ret = kv_enable_dpm_voltage_scaling(adev); if (ret) { - DRM_ERROR("kv_enable_dpm_voltage_scaling failed\n"); + drm_err(adev_to_drm(adev), "kv_enable_dpm_voltage_scaling failed\n"); return ret; } ret = kv_set_dpm_interval(adev); if (ret) { - DRM_ERROR("kv_set_dpm_interval failed\n"); + drm_err(adev_to_drm(adev), "kv_set_dpm_interval failed\n"); return ret; } ret = kv_set_dpm_boot_state(adev); if (ret) { - DRM_ERROR("kv_set_dpm_boot_state failed\n"); + drm_err(adev_to_drm(adev), "kv_set_dpm_boot_state failed\n"); return ret; } ret = kv_enable_ulv(adev, true); if (ret) { - DRM_ERROR("kv_enable_ulv failed\n"); + drm_err(adev_to_drm(adev), "kv_enable_ulv failed\n"); return ret; } kv_start_dpm(adev); ret = kv_enable_didt(adev, true); if (ret) { - DRM_ERROR("kv_enable_didt failed\n"); + drm_err(adev_to_drm(adev), "kv_enable_didt failed\n"); return ret; } ret = kv_enable_smc_cac(adev, true); if (ret) { - DRM_ERROR("kv_enable_smc_cac failed\n"); + drm_err(adev_to_drm(adev), "kv_enable_smc_cac failed\n"); return ret; } @@ -1350,7 +1350,7 @@ static int kv_dpm_enable(struct amdgpu_device *adev) ret = amdgpu_kv_smc_bapm_enable(adev, false); if (ret) { - DRM_ERROR("amdgpu_kv_smc_bapm_enable failed\n"); + drm_err(adev_to_drm(adev), "amdgpu_kv_smc_bapm_enable failed\n"); return ret; } @@ -1358,7 +1358,7 @@ static int kv_dpm_enable(struct amdgpu_device *adev) kv_is_internal_thermal_sensor(adev->pm.int_thermal_type)) { ret = kv_set_thermal_temperature_range(adev, KV_TEMP_RANGE_MIN, KV_TEMP_RANGE_MAX); if (ret) { - DRM_ERROR("kv_set_thermal_temperature_range failed\n"); + drm_err(adev_to_drm(adev), "kv_set_thermal_temperature_range failed\n"); return ret; } amdgpu_irq_get(adev, &adev->pm.dpm.thermal.irq, @@ -1382,7 +1382,7 @@ static void kv_dpm_disable(struct amdgpu_device *adev) err = amdgpu_kv_smc_bapm_enable(adev, false); if (err) - DRM_ERROR("amdgpu_kv_smc_bapm_enable failed\n"); + drm_err(adev_to_drm(adev), "amdgpu_kv_smc_bapm_enable failed\n"); if (adev->asic_type == CHIP_MULLINS) kv_enable_nb_dpm(adev, false); @@ -1920,7 +1920,7 @@ static int kv_dpm_set_power_state(void *handle) if (pi->bapm_enable) { ret = amdgpu_kv_smc_bapm_enable(adev, adev->pm.ac_power); if (ret) { - DRM_ERROR("amdgpu_kv_smc_bapm_enable failed\n"); + drm_err(adev_to_drm(adev), "amdgpu_kv_smc_bapm_enable failed\n"); return ret; } } @@ -1931,7 +1931,7 @@ static int kv_dpm_set_power_state(void *handle) kv_update_dfs_bypass_settings(adev, new_ps); ret = kv_calculate_ds_divider(adev); if (ret) { - DRM_ERROR("kv_calculate_ds_divider failed\n"); + drm_err(adev_to_drm(adev), "kv_calculate_ds_divider failed\n"); return ret; } kv_calculate_nbps_level_settings(adev); @@ -1947,7 +1947,7 @@ static int kv_dpm_set_power_state(void *handle) ret = kv_update_vce_dpm(adev, new_ps, old_ps); if (ret) { - DRM_ERROR("kv_update_vce_dpm failed\n"); + drm_err(adev_to_drm(adev), "kv_update_vce_dpm failed\n"); return ret; } kv_update_sclk_t(adev); @@ -1960,7 +1960,7 @@ static int kv_dpm_set_power_state(void *handle) kv_update_dfs_bypass_settings(adev, new_ps); ret = kv_calculate_ds_divider(adev); if (ret) { - DRM_ERROR("kv_calculate_ds_divider failed\n"); + drm_err(adev_to_drm(adev), "kv_calculate_ds_divider failed\n"); return ret; } kv_calculate_nbps_level_settings(adev); @@ -1972,7 +1972,7 @@ static int kv_dpm_set_power_state(void *handle) kv_set_enabled_levels(adev); ret = kv_update_vce_dpm(adev, new_ps, old_ps); if (ret) { - DRM_ERROR("kv_update_vce_dpm failed\n"); + drm_err(adev_to_drm(adev), "kv_update_vce_dpm failed\n"); return ret; } kv_update_acp_boot_level(adev); @@ -2521,7 +2521,7 @@ static int kv_set_thermal_temperature_range(struct amdgpu_device *adev, if (high_temp > max_temp) high_temp = max_temp; if (high_temp < low_temp) { - DRM_ERROR("invalid thermal range: %d - %d\n", low_temp, high_temp); + drm_err(adev_to_drm(adev), "invalid thermal range: %d - %d\n", low_temp, high_temp); return -EINVAL; } @@ -2563,7 +2563,7 @@ static int kv_parse_sys_info_table(struct amdgpu_device *adev) data_offset); if (crev != 8) { - DRM_ERROR("Unsupported IGP table: %d %d\n", frev, crev); + drm_err(adev_to_drm(adev), "Unsupported IGP table: %d %d\n", frev, crev); return -EINVAL; } pi->sys_info.bootup_sclk = le32_to_cpu(igp_info->info_8.ulBootUpEngineClock); @@ -2579,7 +2579,7 @@ static int kv_parse_sys_info_table(struct amdgpu_device *adev) else pi->sys_info.htc_hyst_lmt = igp_info->info_8.ucHtcHystLmt; if (pi->sys_info.htc_tmp_lmt <= pi->sys_info.htc_hyst_lmt) { - DRM_ERROR("The htcTmpLmt should be larger than htcHystLmt.\n"); + drm_err(adev_to_drm(adev), "The htcTmpLmt should be larger than htcHystLmt.\n"); } if (le32_to_cpu(igp_info->info_8.ulSystemConfig) & (1 << 3)) @@ -2886,16 +2886,18 @@ kv_dpm_print_power_state(void *handle, void *request_ps) struct kv_ps *ps = kv_get_ps(rps); struct amdgpu_device *adev = (struct amdgpu_device *)handle; - amdgpu_dpm_print_class_info(rps->class, rps->class2); - amdgpu_dpm_print_cap_info(rps->caps); - printk("\tuvd vclk: %d dclk: %d\n", rps->vclk, rps->dclk); + amdgpu_dpm_dbg_print_class_info(adev, rps->class, rps->class2); + amdgpu_dpm_dbg_print_cap_info(adev, rps->caps); + drm_dbg(adev_to_drm(adev), "vclk: %d, dclk: %d\n", + rps->vclk, rps->dclk); for (i = 0; i < ps->num_levels; i++) { struct kv_pl *pl = &ps->levels[i]; - printk("\t\tpower level %d sclk: %u vddc: %u\n", - i, pl->sclk, - kv_convert_8bit_index_to_voltage(adev, pl->vddc_index)); + drm_dbg(adev_to_drm(adev), + "power level %d sclk: %u vddc: %u\n", + i, pl->sclk, + kv_convert_8bit_index_to_voltage(adev, pl->vddc_index)); } - amdgpu_dpm_print_ps_status(adev, rps); + amdgpu_dpm_dbg_print_ps_status(adev, rps); } static void kv_dpm_fini(struct amdgpu_device *adev) @@ -3013,13 +3015,13 @@ static int kv_dpm_sw_init(struct amdgpu_ip_block *ip_block) adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps; if (amdgpu_dpm == 1) amdgpu_pm_print_power_states(adev); - DRM_INFO("amdgpu: dpm initialized\n"); + drm_info(adev_to_drm(adev), "dpm initialized\n"); return 0; dpm_failed: kv_dpm_fini(adev); - DRM_ERROR("amdgpu: dpm initialization failed\n"); + drm_err(adev_to_drm(adev), "dpm initialization failed: %d\n", ret); return ret; } diff --git a/drivers/gpu/drm/amd/pm/legacy-dpm/legacy_dpm.c b/drivers/gpu/drm/amd/pm/legacy-dpm/legacy_dpm.c index c7518b13e787..ea3ace882a10 100644 --- a/drivers/gpu/drm/amd/pm/legacy-dpm/legacy_dpm.c +++ b/drivers/gpu/drm/amd/pm/legacy-dpm/legacy_dpm.c @@ -47,7 +47,7 @@ #define amdgpu_dpm_check_state_equal(adev, cps, rps, equal) \ ((adev)->powerplay.pp_funcs->check_state_equal((adev)->powerplay.pp_handle, (cps), (rps), (equal))) -void amdgpu_dpm_print_class_info(u32 class, u32 class2) +void amdgpu_dpm_dbg_print_class_info(struct amdgpu_device *adev, u32 class, u32 class2) { const char *s; @@ -66,71 +66,45 @@ void amdgpu_dpm_print_class_info(u32 class, u32 class2) s = "performance"; break; } - printk("\tui class: %s\n", s); - printk("\tinternal class:"); + drm_dbg(adev_to_drm(adev), "\tui class: %s\n", s); if (((class & ~ATOM_PPLIB_CLASSIFICATION_UI_MASK) == 0) && (class2 == 0)) - pr_cont(" none"); - else { - if (class & ATOM_PPLIB_CLASSIFICATION_BOOT) - pr_cont(" boot"); - if (class & ATOM_PPLIB_CLASSIFICATION_THERMAL) - pr_cont(" thermal"); - if (class & ATOM_PPLIB_CLASSIFICATION_LIMITEDPOWERSOURCE) - pr_cont(" limited_pwr"); - if (class & ATOM_PPLIB_CLASSIFICATION_REST) - pr_cont(" rest"); - if (class & ATOM_PPLIB_CLASSIFICATION_FORCED) - pr_cont(" forced"); - if (class & ATOM_PPLIB_CLASSIFICATION_3DPERFORMANCE) - pr_cont(" 3d_perf"); - if (class & ATOM_PPLIB_CLASSIFICATION_OVERDRIVETEMPLATE) - pr_cont(" ovrdrv"); - if (class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE) - pr_cont(" uvd"); - if (class & ATOM_PPLIB_CLASSIFICATION_3DLOW) - pr_cont(" 3d_low"); - if (class & ATOM_PPLIB_CLASSIFICATION_ACPI) - pr_cont(" acpi"); - if (class & ATOM_PPLIB_CLASSIFICATION_HD2STATE) - pr_cont(" uvd_hd2"); - if (class & ATOM_PPLIB_CLASSIFICATION_HDSTATE) - pr_cont(" uvd_hd"); - if (class & ATOM_PPLIB_CLASSIFICATION_SDSTATE) - pr_cont(" uvd_sd"); - if (class2 & ATOM_PPLIB_CLASSIFICATION2_LIMITEDPOWERSOURCE_2) - pr_cont(" limited_pwr2"); - if (class2 & ATOM_PPLIB_CLASSIFICATION2_ULV) - pr_cont(" ulv"); - if (class2 & ATOM_PPLIB_CLASSIFICATION2_MVC) - pr_cont(" uvd_mvc"); - } - pr_cont("\n"); + drm_dbg(adev_to_drm(adev), "\tinternal class: none\n"); + else + drm_dbg(adev_to_drm(adev), "\tinternal class: %s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s\n", + (class & ATOM_PPLIB_CLASSIFICATION_BOOT) ? " boot" : "", + (class & ATOM_PPLIB_CLASSIFICATION_THERMAL) ? " thermal" : "", + (class & ATOM_PPLIB_CLASSIFICATION_LIMITEDPOWERSOURCE) ? " limited_pwr" : "", + (class & ATOM_PPLIB_CLASSIFICATION_REST) ? " rest" : "", + (class & ATOM_PPLIB_CLASSIFICATION_FORCED) ? " forced" : "", + (class & ATOM_PPLIB_CLASSIFICATION_3DPERFORMANCE) ? " 3d_perf" : "", + (class & ATOM_PPLIB_CLASSIFICATION_OVERDRIVETEMPLATE) ? " ovrdrv" : "", + (class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE) ? " uvd" : "", + (class & ATOM_PPLIB_CLASSIFICATION_3DLOW) ? " 3d_low" : "", + (class & ATOM_PPLIB_CLASSIFICATION_ACPI) ? " acpi" : "", + (class & ATOM_PPLIB_CLASSIFICATION_HD2STATE) ? " uvd_hd2" : "", + (class & ATOM_PPLIB_CLASSIFICATION_HDSTATE) ? " uvd_hd" : "", + (class & ATOM_PPLIB_CLASSIFICATION_SDSTATE) ? " uvd_sd" : "", + (class2 & ATOM_PPLIB_CLASSIFICATION2_LIMITEDPOWERSOURCE_2) ? " limited_pwr2" : "", + (class2 & ATOM_PPLIB_CLASSIFICATION2_ULV) ? " ulv" : "", + (class2 & ATOM_PPLIB_CLASSIFICATION2_MVC) ? " uvd_mvc" : ""); } -void amdgpu_dpm_print_cap_info(u32 caps) +void amdgpu_dpm_dbg_print_cap_info(struct amdgpu_device *adev, u32 caps) { - printk("\tcaps:"); - if (caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY) - pr_cont(" single_disp"); - if (caps & ATOM_PPLIB_SUPPORTS_VIDEO_PLAYBACK) - pr_cont(" video"); - if (caps & ATOM_PPLIB_DISALLOW_ON_DC) - pr_cont(" no_dc"); - pr_cont("\n"); + drm_dbg(adev_to_drm(adev), "\tcaps: %s%s%s\n", + (caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY) ? " single_disp" : "", + (caps & ATOM_PPLIB_SUPPORTS_VIDEO_PLAYBACK) ? " video" : "", + (caps & ATOM_PPLIB_DISALLOW_ON_DC) ? " no_dc" : ""); } -void amdgpu_dpm_print_ps_status(struct amdgpu_device *adev, +void amdgpu_dpm_dbg_print_ps_status(struct amdgpu_device *adev, struct amdgpu_ps *rps) { - printk("\tstatus:"); - if (rps == adev->pm.dpm.current_ps) - pr_cont(" c"); - if (rps == adev->pm.dpm.requested_ps) - pr_cont(" r"); - if (rps == adev->pm.dpm.boot_ps) - pr_cont(" b"); - pr_cont("\n"); + drm_dbg(adev_to_drm(adev), "\tstatus:%s%s%s\n", + rps == adev->pm.dpm.current_ps ? " c" : "", + rps == adev->pm.dpm.requested_ps ? " r" : "", + rps == adev->pm.dpm.boot_ps ? " b" : ""); } void amdgpu_pm_print_power_states(struct amdgpu_device *adev) @@ -699,64 +673,64 @@ void amdgpu_add_thermal_controller(struct amdgpu_device *adev) adev->pm.fan_max_rpm = controller->ucFanMaxRPM; } if (controller->ucType == ATOM_PP_THERMALCONTROLLER_RV6xx) { - DRM_INFO("Internal thermal controller %s fan control\n", + drm_info(adev_to_drm(adev), "Internal thermal controller %s fan control\n", (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); adev->pm.int_thermal_type = THERMAL_TYPE_RV6XX; } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_RV770) { - DRM_INFO("Internal thermal controller %s fan control\n", + drm_info(adev_to_drm(adev), "Internal thermal controller %s fan control\n", (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); adev->pm.int_thermal_type = THERMAL_TYPE_RV770; } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_EVERGREEN) { - DRM_INFO("Internal thermal controller %s fan control\n", + drm_info(adev_to_drm(adev), "Internal thermal controller %s fan control\n", (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); adev->pm.int_thermal_type = THERMAL_TYPE_EVERGREEN; } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_SUMO) { - DRM_INFO("Internal thermal controller %s fan control\n", + drm_info(adev_to_drm(adev), "Internal thermal controller %s fan control\n", (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); adev->pm.int_thermal_type = THERMAL_TYPE_SUMO; } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_NISLANDS) { - DRM_INFO("Internal thermal controller %s fan control\n", + drm_info(adev_to_drm(adev), "Internal thermal controller %s fan control\n", (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); adev->pm.int_thermal_type = THERMAL_TYPE_NI; } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_SISLANDS) { - DRM_INFO("Internal thermal controller %s fan control\n", + drm_info(adev_to_drm(adev), "Internal thermal controller %s fan control\n", (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); adev->pm.int_thermal_type = THERMAL_TYPE_SI; } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_CISLANDS) { - DRM_INFO("Internal thermal controller %s fan control\n", + drm_info(adev_to_drm(adev), "Internal thermal controller %s fan control\n", (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); adev->pm.int_thermal_type = THERMAL_TYPE_CI; } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_KAVERI) { - DRM_INFO("Internal thermal controller %s fan control\n", + drm_info(adev_to_drm(adev), "Internal thermal controller %s fan control\n", (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); adev->pm.int_thermal_type = THERMAL_TYPE_KV; } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_EXTERNAL_GPIO) { - DRM_INFO("External GPIO thermal controller %s fan control\n", + drm_info(adev_to_drm(adev), "External GPIO thermal controller %s fan control\n", (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); adev->pm.int_thermal_type = THERMAL_TYPE_EXTERNAL_GPIO; } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_ADT7473_WITH_INTERNAL) { - DRM_INFO("ADT7473 with internal thermal controller %s fan control\n", + drm_info(adev_to_drm(adev), "ADT7473 with internal thermal controller %s fan control\n", (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); adev->pm.int_thermal_type = THERMAL_TYPE_ADT7473_WITH_INTERNAL; } else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_EMC2103_WITH_INTERNAL) { - DRM_INFO("EMC2103 with internal thermal controller %s fan control\n", + drm_info(adev_to_drm(adev), "EMC2103 with internal thermal controller %s fan control\n", (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with"); adev->pm.int_thermal_type = THERMAL_TYPE_EMC2103_WITH_INTERNAL; } else if (controller->ucType < ARRAY_SIZE(pp_lib_thermal_controller_names)) { - DRM_INFO("Possible %s thermal controller at 0x%02x %s fan control\n", + drm_info(adev_to_drm(adev), "Possible %s thermal controller at 0x%02x %s fan control\n", pp_lib_thermal_controller_names[controller->ucType], controller->ucI2cAddress >> 1, (controller->ucFanParameters & @@ -772,7 +746,7 @@ void amdgpu_add_thermal_controller(struct amdgpu_device *adev) i2c_new_client_device(&adev->pm.i2c_bus->adapter, &info); } } else { - DRM_INFO("Unknown thermal controller type %d at 0x%02x %s fan control\n", + drm_info(adev_to_drm(adev), "Unknown thermal controller type %d at 0x%02x %s fan control\n", controller->ucType, controller->ucI2cAddress >> 1, (controller->ucFanParameters & @@ -943,9 +917,9 @@ static int amdgpu_dpm_change_power_state_locked(struct amdgpu_device *adev) return -EINVAL; if (amdgpu_dpm == 1 && pp_funcs->print_power_state) { - printk("switching from power state:\n"); + drm_dbg(adev_to_drm(adev), "switching from power state\n"); amdgpu_dpm_print_power_state(adev, adev->pm.dpm.current_ps); - printk("switching to power state:\n"); + drm_dbg(adev_to_drm(adev), "switching to power state\n"); amdgpu_dpm_print_power_state(adev, adev->pm.dpm.requested_ps); } diff --git a/drivers/gpu/drm/amd/pm/legacy-dpm/legacy_dpm.h b/drivers/gpu/drm/amd/pm/legacy-dpm/legacy_dpm.h index 93bd3973330c..7120eef30509 100644 --- a/drivers/gpu/drm/amd/pm/legacy-dpm/legacy_dpm.h +++ b/drivers/gpu/drm/amd/pm/legacy-dpm/legacy_dpm.h @@ -23,10 +23,9 @@ #ifndef __LEGACY_DPM_H__ #define __LEGACY_DPM_H__ -void amdgpu_dpm_print_class_info(u32 class, u32 class2); -void amdgpu_dpm_print_cap_info(u32 caps); -void amdgpu_dpm_print_ps_status(struct amdgpu_device *adev, - struct amdgpu_ps *rps); +void amdgpu_dpm_dbg_print_class_info(struct amdgpu_device *adev, u32 class, u32 class2); +void amdgpu_dpm_dbg_print_cap_info(struct amdgpu_device *adev, u32 caps); +void amdgpu_dpm_dbg_print_ps_status(struct amdgpu_device *adev, struct amdgpu_ps *rps); int amdgpu_get_platform_caps(struct amdgpu_device *adev); int amdgpu_parse_extended_power_table(struct amdgpu_device *adev); void amdgpu_free_extended_power_table(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.c b/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.c index 4c0e976004ba..52e732be59e3 100644 --- a/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.c +++ b/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.c @@ -7951,15 +7951,15 @@ static void si_dpm_print_power_state(void *handle, struct rv7xx_pl *pl; int i; - amdgpu_dpm_print_class_info(rps->class, rps->class2); - amdgpu_dpm_print_cap_info(rps->caps); - DRM_INFO("\tuvd vclk: %d dclk: %d\n", rps->vclk, rps->dclk); + amdgpu_dpm_dbg_print_class_info(adev, rps->class, rps->class2); + amdgpu_dpm_dbg_print_cap_info(adev, rps->caps); + drm_dbg(adev_to_drm(adev), "\tuvd vclk: %d dclk: %d\n", rps->vclk, rps->dclk); for (i = 0; i < ps->performance_level_count; i++) { pl = &ps->performance_levels[i]; - DRM_INFO("\t\tpower level %d sclk: %u mclk: %u vddc: %u vddci: %u pcie gen: %u\n", + drm_dbg(adev_to_drm(adev), "\t\tpower level %d sclk: %u mclk: %u vddc: %u vddci: %u pcie gen: %u\n", i, pl->sclk, pl->mclk, pl->vddc, pl->vddci, pl->pcie_gen + 1); } - amdgpu_dpm_print_ps_status(adev, rps); + amdgpu_dpm_dbg_print_ps_status(adev, rps); } static int si_dpm_early_init(struct amdgpu_ip_block *ip_block) diff --git a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c index d79a1d94661a..756afe78a6e5 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c +++ b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c @@ -76,6 +76,7 @@ static void smu_power_profile_mode_get(struct smu_context *smu, enum PP_SMC_POWER_PROFILE profile_mode); static void smu_power_profile_mode_put(struct smu_context *smu, enum PP_SMC_POWER_PROFILE profile_mode); +static enum smu_clk_type smu_convert_to_smuclk(enum pp_clock_type type); static int smu_sys_get_pp_feature_mask(void *handle, char *buf) @@ -134,12 +135,17 @@ int smu_get_status_gfxoff(struct smu_context *smu, uint32_t *value) } int smu_set_soft_freq_range(struct smu_context *smu, - enum smu_clk_type clk_type, + enum pp_clock_type type, uint32_t min, uint32_t max) { + enum smu_clk_type clk_type; int ret = 0; + clk_type = smu_convert_to_smuclk(type); + if (clk_type == SMU_CLK_COUNT) + return -EINVAL; + if (smu->ppt_funcs->set_soft_freq_limited_range) ret = smu->ppt_funcs->set_soft_freq_limited_range(smu, clk_type, @@ -307,6 +313,26 @@ static int smu_dpm_set_vpe_enable(struct smu_context *smu, return ret; } +static int smu_dpm_set_isp_enable(struct smu_context *smu, + bool enable) +{ + struct smu_power_context *smu_power = &smu->smu_power; + struct smu_power_gate *power_gate = &smu_power->power_gate; + int ret; + + if (!smu->ppt_funcs->dpm_set_isp_enable) + return 0; + + if (atomic_read(&power_gate->isp_gated) ^ enable) + return 0; + + ret = smu->ppt_funcs->dpm_set_isp_enable(smu, enable); + if (!ret) + atomic_set(&power_gate->isp_gated, !enable); + + return ret; +} + static int smu_dpm_set_umsch_mm_enable(struct smu_context *smu, bool enable) { @@ -408,6 +434,12 @@ static int smu_dpm_set_power_gate(void *handle, dev_err(smu->adev->dev, "Failed to power %s VPE!\n", gate ? "gate" : "ungate"); break; + case AMD_IP_BLOCK_TYPE_ISP: + ret = smu_dpm_set_isp_enable(smu, !gate); + if (ret) + dev_err(smu->adev->dev, "Failed to power %s ISP!\n", + gate ? "gate" : "ungate"); + break; default: dev_err(smu->adev->dev, "Unsupported block type!\n"); return -EINVAL; @@ -1004,6 +1036,21 @@ static int smu_fini_fb_allocations(struct smu_context *smu) return 0; } +static void smu_update_gpu_addresses(struct smu_context *smu) +{ + struct smu_table_context *smu_table = &smu->smu_table; + struct smu_table *pm_status_table = smu_table->tables + SMU_TABLE_PMSTATUSLOG; + struct smu_table *driver_table = &(smu_table->driver_table); + struct smu_table *dummy_read_1_table = &smu_table->dummy_read_1_table; + + if (pm_status_table->bo) + pm_status_table->mc_address = amdgpu_bo_fb_aper_addr(pm_status_table->bo); + if (driver_table->bo) + driver_table->mc_address = amdgpu_bo_fb_aper_addr(driver_table->bo); + if (dummy_read_1_table->bo) + dummy_read_1_table->mc_address = amdgpu_bo_fb_aper_addr(dummy_read_1_table->bo); +} + /** * smu_alloc_memory_pool - allocate memory pool in the system memory * @@ -1285,6 +1332,7 @@ static int smu_sw_init(struct amdgpu_ip_block *ip_block) atomic_set(&smu->smu_power.power_gate.vcn_gated[i], 1); atomic_set(&smu->smu_power.power_gate.jpeg_gated, 1); atomic_set(&smu->smu_power.power_gate.vpe_gated, 1); + atomic_set(&smu->smu_power.power_gate.isp_gated, 1); atomic_set(&smu->smu_power.power_gate.umsch_mm_gated, 1); smu_init_power_profile(smu); @@ -1672,37 +1720,6 @@ static int smu_smc_hw_setup(struct smu_context *smu) } } - ret = smu_system_features_control(smu, true); - if (ret) { - dev_err(adev->dev, "Failed to enable requested dpm features!\n"); - return ret; - } - - smu_init_xgmi_plpd_mode(smu); - - ret = smu_feature_get_enabled_mask(smu, &features_supported); - if (ret) { - dev_err(adev->dev, "Failed to retrieve supported dpm features!\n"); - return ret; - } - bitmap_copy(feature->supported, - (unsigned long *)&features_supported, - feature->feature_num); - - if (!smu_is_dpm_running(smu)) - dev_info(adev->dev, "dpm has been disabled\n"); - - /* - * Set initialized values (get from vbios) to dpm tables context such as - * gfxclk, memclk, dcefclk, and etc. And enable the DPM feature for each - * type of clks. - */ - ret = smu_set_default_dpm_table(smu); - if (ret) { - dev_err(adev->dev, "Failed to setup default dpm clock tables!\n"); - return ret; - } - if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN5) pcie_gen = 4; else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN4) @@ -1738,6 +1755,37 @@ static int smu_smc_hw_setup(struct smu_context *smu) return ret; } + ret = smu_system_features_control(smu, true); + if (ret) { + dev_err(adev->dev, "Failed to enable requested dpm features!\n"); + return ret; + } + + smu_init_xgmi_plpd_mode(smu); + + ret = smu_feature_get_enabled_mask(smu, &features_supported); + if (ret) { + dev_err(adev->dev, "Failed to retrieve supported dpm features!\n"); + return ret; + } + bitmap_copy(feature->supported, + (unsigned long *)&features_supported, + feature->feature_num); + + if (!smu_is_dpm_running(smu)) + dev_info(adev->dev, "dpm has been disabled\n"); + + /* + * Set initialized values (get from vbios) to dpm tables context such as + * gfxclk, memclk, dcefclk, and etc. And enable the DPM feature for each + * type of clks. + */ + ret = smu_set_default_dpm_table(smu); + if (ret) { + dev_err(adev->dev, "Failed to setup default dpm clock tables!\n"); + return ret; + } + ret = smu_get_thermal_temperature_range(smu); if (ret) { dev_err(adev->dev, "Failed to get thermal temperature ranges!\n"); @@ -1780,6 +1828,9 @@ static int smu_start_smc_engine(struct smu_context *smu) struct amdgpu_device *adev = smu->adev; int ret = 0; + if (amdgpu_virt_xgmi_migrate_enabled(adev)) + smu_update_gpu_addresses(smu); + smu->smc_fw_state = SMU_FW_INIT; if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) { @@ -2935,6 +2986,12 @@ static enum smu_clk_type smu_convert_to_smuclk(enum pp_clock_type type) clk_type = SMU_DCLK; break; case PP_DCLK1: clk_type = SMU_DCLK1; break; + case PP_ISPICLK: + clk_type = SMU_ISPICLK; + break; + case PP_ISPXCLK: + clk_type = SMU_ISPXCLK; + break; case OD_SCLK: clk_type = SMU_OD_SCLK; break; case OD_MCLK: diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h b/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h index 9aacc7bc1c69..b52e194397e2 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h @@ -402,6 +402,7 @@ struct smu_power_gate { atomic_t vcn_gated[AMDGPU_MAX_VCN_INSTANCES]; atomic_t jpeg_gated; atomic_t vpe_gated; + atomic_t isp_gated; atomic_t umsch_mm_gated; }; @@ -1436,6 +1437,12 @@ struct pptable_funcs { int (*dpm_set_vpe_enable)(struct smu_context *smu, bool enable); /** + * @dpm_set_isp_enable: Enable/disable ISP engine dynamic power + * management. + */ + int (*dpm_set_isp_enable)(struct smu_context *smu, bool enable); + + /** * @dpm_set_umsch_mm_enable: Enable/disable UMSCH engine dynamic power * management. */ @@ -1635,7 +1642,7 @@ int smu_write_watermarks_table(struct smu_context *smu); int smu_get_dpm_freq_range(struct smu_context *smu, enum smu_clk_type clk_type, uint32_t *min, uint32_t *max); -int smu_set_soft_freq_range(struct smu_context *smu, enum smu_clk_type clk_type, +int smu_set_soft_freq_range(struct smu_context *smu, enum pp_clock_type clk_type, uint32_t min, uint32_t max); int smu_set_gfx_power_up_by_imu(struct smu_context *smu); diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu14_driver_if_v14_0_0.h b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu14_driver_if_v14_0_0.h index 1bc30db22f9c..cd44f4254134 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu14_driver_if_v14_0_0.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu14_driver_if_v14_0_0.h @@ -106,6 +106,7 @@ typedef struct { #define NUM_FCLK_DPM_LEVELS 8 #define NUM_MEM_PSTATE_LEVELS 4 +#define ISP_ALL_TILES_MASK 0x7FF typedef struct { uint32_t UClk; diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_12_pmfw.h b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_12_pmfw.h index d7505cfc433a..0a2ca544f4e3 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_12_pmfw.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_12_pmfw.h @@ -86,8 +86,10 @@ typedef enum { /*36*/ FEATURE_PIT = 36, /*37*/ FEATURE_DVO = 37, /*38*/ FEATURE_XVMINORPSM_CLKSTOP_DS = 38, +/*39*/ FEATURE_GLOBAL_DPM = 39, +/*40*/ FEATURE_NODE_POWER_MANAGER = 40, -/*39*/ NUM_FEATURES = 39 +/*41*/ NUM_FEATURES = 41 } FEATURE_LIST_e; //enum for MPIO PCIe gen speed msgs @@ -133,7 +135,7 @@ typedef enum { GFX_DVM_MARGIN_COUNT } GFX_DVM_MARGIN_e; -#define SMU_METRICS_TABLE_VERSION 0x12 +#define SMU_METRICS_TABLE_VERSION 0x13 typedef struct __attribute__((packed, aligned(4))) { uint64_t AccumulationCounter; @@ -275,6 +277,16 @@ typedef struct { //PSNs uint64_t PublicSerialNumber_AID[4]; uint64_t PublicSerialNumber_XCD[8]; + + //XGMI + uint32_t MaxXgmiWidth; + uint32_t MaxXgmiBitrate; + + // Telemetry + uint32_t InputTelemetryVoltageInmV; + + // General info + uint32_t pldmVersion[2]; } StaticMetricsTable_t; #pragma pack(pop) diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_types.h b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_types.h index eefdaa0b5df6..d7a9e41820fa 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_types.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_types.h @@ -305,6 +305,8 @@ enum smu_clk_type { SMU_MCLK, SMU_PCIE, SMU_LCLK, + SMU_ISPICLK, + SMU_ISPXCLK, SMU_OD_CCLK, SMU_OD_SCLK, SMU_OD_MCLK, diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c index 7fad5dfb39c4..aac202d0c30e 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c @@ -2444,7 +2444,8 @@ static int navi10_update_pcie_parameters(struct smu_context *smu, struct smu_11_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context; PPTable_t *pptable = smu->smu_table.driver_pptable; uint32_t smu_pcie_arg; - int ret, i; + int ret = 0; + int i; /* lclk dpm table setup */ for (i = 0; i < MAX_PCIE_CONF; i++) { @@ -2453,25 +2454,27 @@ static int navi10_update_pcie_parameters(struct smu_context *smu, } for (i = 0; i < NUM_LINK_LEVELS; i++) { - smu_pcie_arg = (i << 16) | - ((pptable->PcieGenSpeed[i] <= pcie_gen_cap) ? (pptable->PcieGenSpeed[i] << 8) : - (pcie_gen_cap << 8)) | ((pptable->PcieLaneCount[i] <= pcie_width_cap) ? - pptable->PcieLaneCount[i] : pcie_width_cap); - ret = smu_cmn_send_smc_msg_with_param(smu, - SMU_MSG_OverridePcieParameters, - smu_pcie_arg, - NULL); - - if (ret) - return ret; - - if (pptable->PcieGenSpeed[i] > pcie_gen_cap) - dpm_context->dpm_tables.pcie_table.pcie_gen[i] = pcie_gen_cap; - if (pptable->PcieLaneCount[i] > pcie_width_cap) - dpm_context->dpm_tables.pcie_table.pcie_lane[i] = pcie_width_cap; + if (pptable->PcieGenSpeed[i] > pcie_gen_cap || + pptable->PcieLaneCount[i] > pcie_width_cap) { + dpm_context->dpm_tables.pcie_table.pcie_gen[i] = + pptable->PcieGenSpeed[i] > pcie_gen_cap ? + pcie_gen_cap : pptable->PcieGenSpeed[i]; + dpm_context->dpm_tables.pcie_table.pcie_lane[i] = + pptable->PcieLaneCount[i] > pcie_width_cap ? + pcie_width_cap : pptable->PcieLaneCount[i]; + smu_pcie_arg = i << 16; + smu_pcie_arg |= pcie_gen_cap << 8; + smu_pcie_arg |= pcie_width_cap; + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_OverridePcieParameters, + smu_pcie_arg, + NULL); + if (ret) + break; + } } - return 0; + return ret; } static inline void navi10_dump_od_table(struct smu_context *smu, diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c index 115e3fa456bc..d57591509aed 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c @@ -2145,7 +2145,8 @@ static int sienna_cichlid_update_pcie_parameters(struct smu_context *smu, uint8_t min_gen_speed, max_gen_speed; uint8_t min_lane_width, max_lane_width; uint32_t smu_pcie_arg; - int ret, i; + int ret = 0; + int i; GET_PPTABLE_MEMBER(PcieGenSpeed, &table_member1); GET_PPTABLE_MEMBER(PcieLaneCount, &table_member2); @@ -2170,19 +2171,22 @@ static int sienna_cichlid_update_pcie_parameters(struct smu_context *smu, pcie_table->pcie_lane[1] = max_lane_width; for (i = 0; i < NUM_LINK_LEVELS; i++) { - smu_pcie_arg = (i << 16 | + if (!(smu->adev->pm.pp_feature & PP_PCIE_DPM_MASK) || + table_member1[i] > pcie_gen_cap || table_member2[i] > pcie_width_cap) { + smu_pcie_arg = (i << 16 | pcie_table->pcie_gen[i] << 8 | pcie_table->pcie_lane[i]); - ret = smu_cmn_send_smc_msg_with_param(smu, - SMU_MSG_OverridePcieParameters, - smu_pcie_arg, - NULL); - if (ret) - return ret; + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_OverridePcieParameters, + smu_pcie_arg, + NULL); + if (ret) + break; + } } - return 0; + return ret; } static int sienna_cichlid_get_dpm_ultimate_freq(struct smu_context *smu, diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c index a55ea76d7399..2c9869feba61 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c @@ -666,7 +666,6 @@ static int vangogh_print_clk_levels(struct smu_context *smu, { DpmClocks_t *clk_table = smu->smu_table.clocks_table; SmuMetrics_t metrics; - struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); int i, idx, size = 0, ret = 0; uint32_t cur_value = 0, value = 0, count = 0; bool cur_value_match_level = false; @@ -682,31 +681,25 @@ static int vangogh_print_clk_levels(struct smu_context *smu, switch (clk_type) { case SMU_OD_SCLK: - if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_MANUAL) { - size += sysfs_emit_at(buf, size, "%s:\n", "OD_SCLK"); - size += sysfs_emit_at(buf, size, "0: %10uMhz\n", - (smu->gfx_actual_hard_min_freq > 0) ? smu->gfx_actual_hard_min_freq : smu->gfx_default_hard_min_freq); - size += sysfs_emit_at(buf, size, "1: %10uMhz\n", - (smu->gfx_actual_soft_max_freq > 0) ? smu->gfx_actual_soft_max_freq : smu->gfx_default_soft_max_freq); - } + size += sysfs_emit_at(buf, size, "%s:\n", "OD_SCLK"); + size += sysfs_emit_at(buf, size, "0: %10uMhz\n", + (smu->gfx_actual_hard_min_freq > 0) ? smu->gfx_actual_hard_min_freq : smu->gfx_default_hard_min_freq); + size += sysfs_emit_at(buf, size, "1: %10uMhz\n", + (smu->gfx_actual_soft_max_freq > 0) ? smu->gfx_actual_soft_max_freq : smu->gfx_default_soft_max_freq); break; case SMU_OD_CCLK: - if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_MANUAL) { - size += sysfs_emit_at(buf, size, "CCLK_RANGE in Core%d:\n", smu->cpu_core_id_select); - size += sysfs_emit_at(buf, size, "0: %10uMhz\n", - (smu->cpu_actual_soft_min_freq > 0) ? smu->cpu_actual_soft_min_freq : smu->cpu_default_soft_min_freq); - size += sysfs_emit_at(buf, size, "1: %10uMhz\n", - (smu->cpu_actual_soft_max_freq > 0) ? smu->cpu_actual_soft_max_freq : smu->cpu_default_soft_max_freq); - } + size += sysfs_emit_at(buf, size, "CCLK_RANGE in Core%d:\n", smu->cpu_core_id_select); + size += sysfs_emit_at(buf, size, "0: %10uMhz\n", + (smu->cpu_actual_soft_min_freq > 0) ? smu->cpu_actual_soft_min_freq : smu->cpu_default_soft_min_freq); + size += sysfs_emit_at(buf, size, "1: %10uMhz\n", + (smu->cpu_actual_soft_max_freq > 0) ? smu->cpu_actual_soft_max_freq : smu->cpu_default_soft_max_freq); break; case SMU_OD_RANGE: - if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_MANUAL) { - size += sysfs_emit_at(buf, size, "%s:\n", "OD_RANGE"); - size += sysfs_emit_at(buf, size, "SCLK: %7uMhz %10uMhz\n", - smu->gfx_default_hard_min_freq, smu->gfx_default_soft_max_freq); - size += sysfs_emit_at(buf, size, "CCLK: %7uMhz %10uMhz\n", - smu->cpu_default_soft_min_freq, smu->cpu_default_soft_max_freq); - } + size += sysfs_emit_at(buf, size, "%s:\n", "OD_RANGE"); + size += sysfs_emit_at(buf, size, "SCLK: %7uMhz %10uMhz\n", + smu->gfx_default_hard_min_freq, smu->gfx_default_soft_max_freq); + size += sysfs_emit_at(buf, size, "CCLK: %7uMhz %10uMhz\n", + smu->cpu_default_soft_min_freq, smu->cpu_default_soft_max_freq); break; case SMU_SOCCLK: /* the level 3 ~ 6 of socclk use the same frequency for vangogh */ diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c index 9481f897432d..e97b0cf19197 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c @@ -497,7 +497,6 @@ static int renoir_print_clk_levels(struct smu_context *smu, int i, idx, size = 0, ret = 0; uint32_t cur_value = 0, value = 0, count = 0, min = 0, max = 0; SmuMetrics_t metrics; - struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); bool cur_value_match_level = false; memset(&metrics, 0, sizeof(metrics)); @@ -510,28 +509,24 @@ static int renoir_print_clk_levels(struct smu_context *smu, switch (clk_type) { case SMU_OD_RANGE: - if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_MANUAL) { - ret = smu_cmn_send_smc_msg_with_param(smu, - SMU_MSG_GetMinGfxclkFrequency, - 0, &min); - if (ret) - return ret; - ret = smu_cmn_send_smc_msg_with_param(smu, - SMU_MSG_GetMaxGfxclkFrequency, - 0, &max); - if (ret) - return ret; - size += sysfs_emit_at(buf, size, "OD_RANGE\nSCLK: %10uMhz %10uMhz\n", min, max); - } + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_GetMinGfxclkFrequency, + 0, &min); + if (ret) + return ret; + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_GetMaxGfxclkFrequency, + 0, &max); + if (ret) + return ret; + size += sysfs_emit_at(buf, size, "OD_RANGE\nSCLK: %10uMhz %10uMhz\n", min, max); break; case SMU_OD_SCLK: - if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_MANUAL) { - min = (smu->gfx_actual_hard_min_freq > 0) ? smu->gfx_actual_hard_min_freq : smu->gfx_default_hard_min_freq; - max = (smu->gfx_actual_soft_max_freq > 0) ? smu->gfx_actual_soft_max_freq : smu->gfx_default_soft_max_freq; - size += sysfs_emit_at(buf, size, "OD_SCLK\n"); - size += sysfs_emit_at(buf, size, "0:%10uMhz\n", min); - size += sysfs_emit_at(buf, size, "1:%10uMhz\n", max); - } + min = (smu->gfx_actual_hard_min_freq > 0) ? smu->gfx_actual_hard_min_freq : smu->gfx_default_hard_min_freq; + max = (smu->gfx_actual_soft_max_freq > 0) ? smu->gfx_actual_soft_max_freq : smu->gfx_default_soft_max_freq; + size += sysfs_emit_at(buf, size, "OD_SCLK\n"); + size += sysfs_emit_at(buf, size, "0:%10uMhz\n", min); + size += sysfs_emit_at(buf, size, "1:%10uMhz\n", max); break; case SMU_GFXCLK: case SMU_SCLK: diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c index a7167668d189..1a1f2a6b2e52 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c @@ -58,6 +58,7 @@ MODULE_FIRMWARE("amdgpu/aldebaran_smc.bin"); MODULE_FIRMWARE("amdgpu/smu_13_0_0.bin"); +MODULE_FIRMWARE("amdgpu/smu_13_0_0_kicker.bin"); MODULE_FIRMWARE("amdgpu/smu_13_0_7.bin"); MODULE_FIRMWARE("amdgpu/smu_13_0_10.bin"); @@ -92,7 +93,7 @@ const int pmfw_decoded_link_width[7] = {0, 1, 2, 4, 8, 12, 16}; int smu_v13_0_init_microcode(struct smu_context *smu) { struct amdgpu_device *adev = smu->adev; - char ucode_prefix[15]; + char ucode_prefix[30]; int err = 0; const struct smc_firmware_header_v1_0 *hdr; const struct common_firmware_header *header; @@ -103,8 +104,13 @@ int smu_v13_0_init_microcode(struct smu_context *smu) return 0; amdgpu_ucode_ip_version_decode(adev, MP1_HWIP, ucode_prefix, sizeof(ucode_prefix)); - err = amdgpu_ucode_request(adev, &adev->pm.fw, AMDGPU_UCODE_REQUIRED, - "amdgpu/%s.bin", ucode_prefix); + + if (amdgpu_is_kicker_fw(adev)) + err = amdgpu_ucode_request(adev, &adev->pm.fw, AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_kicker.bin", ucode_prefix); + else + err = amdgpu_ucode_request(adev, &adev->pm.fw, AMDGPU_UCODE_REQUIRED, + "amdgpu/%s.bin", ucode_prefix); if (err) goto out; @@ -2380,7 +2386,8 @@ int smu_v13_0_update_pcie_parameters(struct smu_context *smu, &dpm_context->dpm_tables.pcie_table; int num_of_levels = pcie_table->num_of_link_levels; uint32_t smu_pcie_arg; - int ret, i; + int ret = 0; + int i; if (!num_of_levels) return 0; @@ -2396,30 +2403,38 @@ int smu_v13_0_update_pcie_parameters(struct smu_context *smu, for (i = 0; i < num_of_levels; i++) { pcie_table->pcie_gen[i] = pcie_gen_cap; pcie_table->pcie_lane[i] = pcie_width_cap; + smu_pcie_arg = i << 16; + smu_pcie_arg |= pcie_table->pcie_gen[i] << 8; + smu_pcie_arg |= pcie_table->pcie_lane[i]; + + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_OverridePcieParameters, + smu_pcie_arg, + NULL); + if (ret) + break; } } else { for (i = 0; i < num_of_levels; i++) { - if (pcie_table->pcie_gen[i] > pcie_gen_cap) + if (pcie_table->pcie_gen[i] > pcie_gen_cap || + pcie_table->pcie_lane[i] > pcie_width_cap) { pcie_table->pcie_gen[i] = pcie_gen_cap; - if (pcie_table->pcie_lane[i] > pcie_width_cap) pcie_table->pcie_lane[i] = pcie_width_cap; - } - } - - for (i = 0; i < num_of_levels; i++) { - smu_pcie_arg = i << 16; - smu_pcie_arg |= pcie_table->pcie_gen[i] << 8; - smu_pcie_arg |= pcie_table->pcie_lane[i]; + smu_pcie_arg = i << 16; + smu_pcie_arg |= pcie_table->pcie_gen[i] << 8; + smu_pcie_arg |= pcie_table->pcie_lane[i]; - ret = smu_cmn_send_smc_msg_with_param(smu, - SMU_MSG_OverridePcieParameters, - smu_pcie_arg, - NULL); - if (ret) - return ret; + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_OverridePcieParameters, + smu_pcie_arg, + NULL); + if (ret) + break; + } + } } - return 0; + return ret; } int smu_v13_0_disable_pmfw_state(struct smu_context *smu) diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c index 5a9711e8cf68..e084ed99ec0e 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c @@ -572,8 +572,6 @@ static int smu_v13_0_0_set_default_dpm_table(struct smu_context *smu) PPTable_t *pptable = table_context->driver_pptable; SkuTable_t *skutable = &pptable->SkuTable; struct smu_13_0_dpm_table *dpm_table; - struct smu_13_0_pcie_table *pcie_table; - uint32_t link_level; int ret = 0; /* socclk dpm table setup */ @@ -689,24 +687,6 @@ static int smu_v13_0_0_set_default_dpm_table(struct smu_context *smu) dpm_table->max = dpm_table->dpm_levels[0].value; } - /* lclk dpm table setup */ - pcie_table = &dpm_context->dpm_tables.pcie_table; - pcie_table->num_of_link_levels = 0; - for (link_level = 0; link_level < NUM_LINK_LEVELS; link_level++) { - if (!skutable->PcieGenSpeed[link_level] && - !skutable->PcieLaneCount[link_level] && - !skutable->LclkFreq[link_level]) - continue; - - pcie_table->pcie_gen[pcie_table->num_of_link_levels] = - skutable->PcieGenSpeed[link_level]; - pcie_table->pcie_lane[pcie_table->num_of_link_levels] = - skutable->PcieLaneCount[link_level]; - pcie_table->clk_freq[pcie_table->num_of_link_levels] = - skutable->LclkFreq[link_level]; - pcie_table->num_of_link_levels++; - } - /* dcefclk dpm table setup */ dpm_table = &dpm_context->dpm_tables.dcef_table; if (smu_cmn_feature_is_enabled(smu, SMU_FEATURE_DPM_DCN_BIT)) { @@ -3150,6 +3130,90 @@ static int smu_v13_0_0_set_power_limit(struct smu_context *smu, return 0; } +static int smu_v13_0_0_update_pcie_parameters(struct smu_context *smu, + uint8_t pcie_gen_cap, + uint8_t pcie_width_cap) +{ + struct smu_13_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context; + struct smu_13_0_pcie_table *pcie_table = + &dpm_context->dpm_tables.pcie_table; + int num_of_levels; + uint32_t smu_pcie_arg; + uint32_t link_level; + struct smu_table_context *table_context = &smu->smu_table; + PPTable_t *pptable = table_context->driver_pptable; + SkuTable_t *skutable = &pptable->SkuTable; + int ret = 0; + int i; + + pcie_table->num_of_link_levels = 0; + + for (link_level = 0; link_level < NUM_LINK_LEVELS; link_level++) { + if (!skutable->PcieGenSpeed[link_level] && + !skutable->PcieLaneCount[link_level] && + !skutable->LclkFreq[link_level]) + continue; + + pcie_table->pcie_gen[pcie_table->num_of_link_levels] = + skutable->PcieGenSpeed[link_level]; + pcie_table->pcie_lane[pcie_table->num_of_link_levels] = + skutable->PcieLaneCount[link_level]; + pcie_table->clk_freq[pcie_table->num_of_link_levels] = + skutable->LclkFreq[link_level]; + pcie_table->num_of_link_levels++; + } + + num_of_levels = pcie_table->num_of_link_levels; + if (!num_of_levels) + return 0; + + if (!(smu->adev->pm.pp_feature & PP_PCIE_DPM_MASK)) { + if (pcie_table->pcie_gen[num_of_levels - 1] < pcie_gen_cap) + pcie_gen_cap = pcie_table->pcie_gen[num_of_levels - 1]; + + if (pcie_table->pcie_lane[num_of_levels - 1] < pcie_width_cap) + pcie_width_cap = pcie_table->pcie_lane[num_of_levels - 1]; + + /* Force all levels to use the same settings */ + for (i = 0; i < num_of_levels; i++) { + pcie_table->pcie_gen[i] = pcie_gen_cap; + pcie_table->pcie_lane[i] = pcie_width_cap; + smu_pcie_arg = i << 16; + smu_pcie_arg |= pcie_table->pcie_gen[i] << 8; + smu_pcie_arg |= pcie_table->pcie_lane[i]; + + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_OverridePcieParameters, + smu_pcie_arg, + NULL); + if (ret) + break; + } + } else { + for (i = 0; i < num_of_levels; i++) { + if (pcie_table->pcie_gen[i] > pcie_gen_cap || + pcie_table->pcie_lane[i] > pcie_width_cap) { + pcie_table->pcie_gen[i] = pcie_table->pcie_gen[i] > pcie_gen_cap ? + pcie_gen_cap : pcie_table->pcie_gen[i]; + pcie_table->pcie_lane[i] = pcie_table->pcie_lane[i] > pcie_width_cap ? + pcie_width_cap : pcie_table->pcie_lane[i]; + smu_pcie_arg = i << 16; + smu_pcie_arg |= pcie_table->pcie_gen[i] << 8; + smu_pcie_arg |= pcie_table->pcie_lane[i]; + + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_OverridePcieParameters, + smu_pcie_arg, + NULL); + if (ret) + break; + } + } + } + + return ret; +} + static const struct pptable_funcs smu_v13_0_0_ppt_funcs = { .get_allowed_feature_mask = smu_v13_0_0_get_allowed_feature_mask, .set_default_dpm_table = smu_v13_0_0_set_default_dpm_table, @@ -3179,7 +3243,7 @@ static const struct pptable_funcs smu_v13_0_0_ppt_funcs = { .feature_is_enabled = smu_cmn_feature_is_enabled, .print_clk_levels = smu_v13_0_0_print_clk_levels, .force_clk_levels = smu_v13_0_0_force_clk_levels, - .update_pcie_parameters = smu_v13_0_update_pcie_parameters, + .update_pcie_parameters = smu_v13_0_0_update_pcie_parameters, .get_thermal_temperature_range = smu_v13_0_0_get_thermal_temperature_range, .register_irq_handler = smu_v13_0_register_irq_handler, .enable_thermal_alert = smu_v13_0_enable_thermal_alert, diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c index e0d356f93ab0..b3adeb6e43a8 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c @@ -187,8 +187,34 @@ int smu_v13_0_12_get_max_metrics_size(void) return max(sizeof(StaticMetricsTable_t), sizeof(MetricsTable_t)); } +static void smu_v13_0_12_init_xgmi_data(struct smu_context *smu, + StaticMetricsTable_t *static_metrics) +{ + struct smu_table_context *smu_table = &smu->smu_table; + uint16_t max_speed; + uint8_t max_width; + int ret; + + if (smu_table->tables[SMU_TABLE_SMU_METRICS].version >= 0x13) { + max_width = (uint8_t)static_metrics->MaxXgmiWidth; + max_speed = (uint16_t)static_metrics->MaxXgmiBitrate; + ret = 0; + } else { + MetricsTable_t *metrics = (MetricsTable_t *)smu_table->metrics_table; + + ret = smu_v13_0_6_get_metrics_table(smu, NULL, true); + if (!ret) { + max_width = (uint8_t)metrics->XgmiWidth; + max_speed = (uint16_t)metrics->XgmiBitrate; + } + } + if (!ret) + amgpu_xgmi_set_max_speed_width(smu->adev, max_speed, max_width); +} + int smu_v13_0_12_setup_driver_pptable(struct smu_context *smu) { + struct smu_13_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context; struct smu_table_context *smu_table = &smu->smu_table; StaticMetricsTable_t *static_metrics = (StaticMetricsTable_t *)smu_table->metrics_table; struct PPTable_t *pptable = @@ -237,6 +263,18 @@ int smu_v13_0_12_setup_driver_pptable(struct smu_context *smu) if (ret) return ret; + if (smu_v13_0_6_cap_supported(smu, SMU_CAP(BOARD_VOLTAGE))) { + if (!static_metrics->InputTelemetryVoltageInmV) { + dev_warn(smu->adev->dev, "Invalid board voltage %d\n", + static_metrics->InputTelemetryVoltageInmV); + } + dpm_context->board_volt = static_metrics->InputTelemetryVoltageInmV; + } + if (smu_v13_0_6_cap_supported(smu, SMU_CAP(PLDM_VERSION)) && + static_metrics->pldmVersion[0] != 0xFFFFFFFF) + smu->adev->firmware.pldm_version = + static_metrics->pldmVersion[0]; + smu_v13_0_12_init_xgmi_data(smu, static_metrics); pptable->Init = true; } diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c index f00ef7f3f355..68624afe7d83 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c @@ -345,6 +345,11 @@ static void smu_v13_0_12_init_caps(struct smu_context *smu) if (fw_ver >= 0x00562500) smu_v13_0_6_cap_set(smu, SMU_CAP(HST_LIMIT_METRICS)); + + if (fw_ver >= 0x04560100) { + smu_v13_0_6_cap_set(smu, SMU_CAP(BOARD_VOLTAGE)); + smu_v13_0_6_cap_set(smu, SMU_CAP(PLDM_VERSION)); + } } static void smu_v13_0_6_init_caps(struct smu_context *smu) @@ -685,8 +690,8 @@ static int smu_v13_0_6_get_allowed_feature_mask(struct smu_context *smu, return 0; } -static int smu_v13_0_6_get_metrics_table(struct smu_context *smu, - void *metrics_table, bool bypass_cache) +int smu_v13_0_6_get_metrics_table(struct smu_context *smu, void *metrics_table, + bool bypass_cache) { struct smu_table_context *smu_table = &smu->smu_table; uint32_t table_size = smu_table->tables[SMU_TABLE_SMU_METRICS].size; @@ -800,6 +805,8 @@ static int smu_v13_0_6_setup_driver_pptable(struct smu_context *smu) int version = smu_v13_0_6_get_metrics_version(smu); int ret, i, retry = 100; uint32_t table_version; + uint16_t max_speed; + uint8_t max_width; if (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 12) && smu_v13_0_6_cap_supported(smu, SMU_CAP(STATIC_METRICS))) @@ -835,6 +842,9 @@ static int smu_v13_0_6_setup_driver_pptable(struct smu_context *smu) SMUQ10_ROUND(GET_METRIC_FIELD(MaxGfxclkFrequency, version)); pptable->MinGfxclkFrequency = SMUQ10_ROUND(GET_METRIC_FIELD(MinGfxclkFrequency, version)); + max_width = (uint8_t)GET_METRIC_FIELD(XgmiWidth, version); + max_speed = (uint16_t)GET_METRIC_FIELD(XgmiBitrate, version); + amgpu_xgmi_set_max_speed_width(smu->adev, max_speed, max_width); for (i = 0; i < 4; ++i) { pptable->FclkFrequencyTable[i] = @@ -1377,8 +1387,9 @@ static int smu_v13_0_6_print_clk_levels(struct smu_context *smu, return ret; } - min_clk = pstate_table->gfxclk_pstate.curr.min; - max_clk = pstate_table->gfxclk_pstate.curr.max; + single_dpm_table = &(dpm_context->dpm_tables.gfx_table); + min_clk = single_dpm_table->min; + max_clk = single_dpm_table->max; if (now < SMU_13_0_6_DSCLK_THRESHOLD) { size += sysfs_emit_at(buf, size, "S: %uMhz *\n", diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.h b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.h index d38d6d76b1e7..67b30674fd31 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.h +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.h @@ -74,6 +74,8 @@ enum smu_v13_0_6_caps { extern void smu_v13_0_6_set_ppt_funcs(struct smu_context *smu); bool smu_v13_0_6_cap_supported(struct smu_context *smu, enum smu_v13_0_6_caps cap); int smu_v13_0_6_get_static_metrics_table(struct smu_context *smu); +int smu_v13_0_6_get_metrics_table(struct smu_context *smu, void *metrics_table, + bool bypass_cache); bool smu_v13_0_12_is_dpm_running(struct smu_context *smu); int smu_v13_0_12_get_max_metrics_size(void); diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c index c8f4f6fb4083..c96fa5e49ed6 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c @@ -579,8 +579,6 @@ static int smu_v13_0_7_set_default_dpm_table(struct smu_context *smu) PPTable_t *driver_ppt = smu->smu_table.driver_pptable; SkuTable_t *skutable = &driver_ppt->SkuTable; struct smu_13_0_dpm_table *dpm_table; - struct smu_13_0_pcie_table *pcie_table; - uint32_t link_level; int ret = 0; /* socclk dpm table setup */ @@ -687,24 +685,6 @@ static int smu_v13_0_7_set_default_dpm_table(struct smu_context *smu) dpm_table->max = dpm_table->dpm_levels[0].value; } - /* lclk dpm table setup */ - pcie_table = &dpm_context->dpm_tables.pcie_table; - pcie_table->num_of_link_levels = 0; - for (link_level = 0; link_level < NUM_LINK_LEVELS; link_level++) { - if (!skutable->PcieGenSpeed[link_level] && - !skutable->PcieLaneCount[link_level] && - !skutable->LclkFreq[link_level]) - continue; - - pcie_table->pcie_gen[pcie_table->num_of_link_levels] = - skutable->PcieGenSpeed[link_level]; - pcie_table->pcie_lane[pcie_table->num_of_link_levels] = - skutable->PcieLaneCount[link_level]; - pcie_table->clk_freq[pcie_table->num_of_link_levels] = - skutable->LclkFreq[link_level]; - pcie_table->num_of_link_levels++; - } - /* dcefclk dpm table setup */ dpm_table = &dpm_context->dpm_tables.dcef_table; if (smu_cmn_feature_is_enabled(smu, SMU_FEATURE_DPM_DCN_BIT)) { @@ -2739,6 +2719,89 @@ static int smu_v13_0_7_set_power_limit(struct smu_context *smu, return 0; } +static int smu_v13_0_7_update_pcie_parameters(struct smu_context *smu, + uint8_t pcie_gen_cap, + uint8_t pcie_width_cap) +{ + struct smu_13_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context; + struct smu_13_0_pcie_table *pcie_table = + &dpm_context->dpm_tables.pcie_table; + int num_of_levels; + int link_level; + uint32_t smu_pcie_arg; + struct smu_table_context *table_context = &smu->smu_table; + PPTable_t *pptable = table_context->driver_pptable; + SkuTable_t *skutable = &pptable->SkuTable; + int ret = 0; + int i; + + pcie_table->num_of_link_levels = 0; + for (link_level = 0; link_level < NUM_LINK_LEVELS; link_level++) { + if (!skutable->PcieGenSpeed[link_level] && + !skutable->PcieLaneCount[link_level] && + !skutable->LclkFreq[link_level]) + continue; + + pcie_table->pcie_gen[pcie_table->num_of_link_levels] = + skutable->PcieGenSpeed[link_level]; + pcie_table->pcie_lane[pcie_table->num_of_link_levels] = + skutable->PcieLaneCount[link_level]; + pcie_table->clk_freq[pcie_table->num_of_link_levels] = + skutable->LclkFreq[link_level]; + pcie_table->num_of_link_levels++; + } + + num_of_levels = pcie_table->num_of_link_levels; + if (!num_of_levels) + return 0; + + if (!(smu->adev->pm.pp_feature & PP_PCIE_DPM_MASK)) { + if (pcie_table->pcie_gen[num_of_levels - 1] < pcie_gen_cap) + pcie_gen_cap = pcie_table->pcie_gen[num_of_levels - 1]; + + if (pcie_table->pcie_lane[num_of_levels - 1] < pcie_width_cap) + pcie_width_cap = pcie_table->pcie_lane[num_of_levels - 1]; + + /* Force all levels to use the same settings */ + for (i = 0; i < num_of_levels; i++) { + pcie_table->pcie_gen[i] = pcie_gen_cap; + pcie_table->pcie_lane[i] = pcie_width_cap; + smu_pcie_arg = i << 16; + smu_pcie_arg |= pcie_table->pcie_gen[i] << 8; + smu_pcie_arg |= pcie_table->pcie_lane[i]; + + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_OverridePcieParameters, + smu_pcie_arg, + NULL); + if (ret) + break; + } + } else { + for (i = 0; i < num_of_levels; i++) { + if (pcie_table->pcie_gen[i] > pcie_gen_cap || + pcie_table->pcie_lane[i] > pcie_width_cap) { + pcie_table->pcie_gen[i] = pcie_table->pcie_gen[i] > pcie_gen_cap ? + pcie_gen_cap : pcie_table->pcie_gen[i]; + pcie_table->pcie_lane[i] = pcie_table->pcie_lane[i] > pcie_width_cap ? + pcie_width_cap : pcie_table->pcie_lane[i]; + smu_pcie_arg = i << 16; + smu_pcie_arg |= pcie_table->pcie_gen[i] << 8; + smu_pcie_arg |= pcie_table->pcie_lane[i]; + + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_OverridePcieParameters, + smu_pcie_arg, + NULL); + if (ret) + break; + } + } + } + + return ret; +} + static const struct pptable_funcs smu_v13_0_7_ppt_funcs = { .get_allowed_feature_mask = smu_v13_0_7_get_allowed_feature_mask, .set_default_dpm_table = smu_v13_0_7_set_default_dpm_table, @@ -2768,7 +2831,7 @@ static const struct pptable_funcs smu_v13_0_7_ppt_funcs = { .feature_is_enabled = smu_cmn_feature_is_enabled, .print_clk_levels = smu_v13_0_7_print_clk_levels, .force_clk_levels = smu_v13_0_7_force_clk_levels, - .update_pcie_parameters = smu_v13_0_update_pcie_parameters, + .update_pcie_parameters = smu_v13_0_7_update_pcie_parameters, .get_thermal_temperature_range = smu_v13_0_7_get_thermal_temperature_range, .register_irq_handler = smu_v13_0_register_irq_handler, .enable_thermal_alert = smu_v13_0_enable_thermal_alert, diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c index 84f9b007b59f..fe00c84b1cc6 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c @@ -1207,11 +1207,13 @@ static int smu_v14_0_0_print_clk_levels(struct smu_context *smu, static int smu_v14_0_0_set_soft_freq_limited_range(struct smu_context *smu, enum smu_clk_type clk_type, - uint32_t min, - uint32_t max) + u32 min, + u32 max, + bool __always_unused automatic) { - enum smu_message_type msg_set_min, msg_set_max; - int ret = 0; + enum smu_message_type msg_set_min = SMU_MSG_MAX_COUNT; + enum smu_message_type msg_set_max = SMU_MSG_MAX_COUNT; + int ret = -EINVAL; if (!smu_v14_0_0_clk_dpm_is_enabled(smu, clk_type)) return -EINVAL; @@ -1240,16 +1242,23 @@ static int smu_v14_0_0_set_soft_freq_limited_range(struct smu_context *smu, msg_set_min = SMU_MSG_SetHardMinVcn1; msg_set_max = SMU_MSG_SetSoftMaxVcn1; break; + case SMU_ISPICLK: + msg_set_min = SMU_MSG_SetHardMinIspiclkByFreq; + break; + case SMU_ISPXCLK: + msg_set_min = SMU_MSG_SetHardMinIspxclkByFreq; + break; default: return -EINVAL; } - ret = smu_cmn_send_smc_msg_with_param(smu, msg_set_min, min, NULL); - if (ret) - return ret; + if (min && msg_set_min != SMU_MSG_MAX_COUNT) + ret = smu_cmn_send_smc_msg_with_param(smu, msg_set_min, min, NULL); + + if (max && msg_set_max != SMU_MSG_MAX_COUNT) + ret = smu_cmn_send_smc_msg_with_param(smu, msg_set_max, max, NULL); - return smu_cmn_send_smc_msg_with_param(smu, msg_set_max, - max, NULL); + return ret; } static int smu_v14_0_0_force_clk_levels(struct smu_context *smu, @@ -1278,7 +1287,7 @@ static int smu_v14_0_0_force_clk_levels(struct smu_context *smu, if (ret) break; - ret = smu_v14_0_0_set_soft_freq_limited_range(smu, clk_type, min_freq, max_freq); + ret = smu_v14_0_0_set_soft_freq_limited_range(smu, clk_type, min_freq, max_freq, false); break; default: ret = -EINVAL; @@ -1426,7 +1435,8 @@ static int smu_v14_0_common_set_performance_level(struct smu_context *smu, ret = smu_v14_0_0_set_soft_freq_limited_range(smu, SMU_SCLK, sclk_min, - sclk_max); + sclk_max, + false); if (ret) return ret; @@ -1438,7 +1448,8 @@ static int smu_v14_0_common_set_performance_level(struct smu_context *smu, ret = smu_v14_0_0_set_soft_freq_limited_range(smu, SMU_FCLK, fclk_min, - fclk_max); + fclk_max, + false); if (ret) return ret; } @@ -1447,7 +1458,8 @@ static int smu_v14_0_common_set_performance_level(struct smu_context *smu, ret = smu_v14_0_0_set_soft_freq_limited_range(smu, SMU_SOCCLK, socclk_min, - socclk_max); + socclk_max, + false); if (ret) return ret; } @@ -1456,7 +1468,8 @@ static int smu_v14_0_common_set_performance_level(struct smu_context *smu, ret = smu_v14_0_0_set_soft_freq_limited_range(smu, SMU_VCLK, vclk_min, - vclk_max); + vclk_max, + false); if (ret) return ret; } @@ -1465,7 +1478,8 @@ static int smu_v14_0_common_set_performance_level(struct smu_context *smu, ret = smu_v14_0_0_set_soft_freq_limited_range(smu, SMU_VCLK1, vclk1_min, - vclk1_max); + vclk1_max, + false); if (ret) return ret; } @@ -1474,7 +1488,8 @@ static int smu_v14_0_common_set_performance_level(struct smu_context *smu, ret = smu_v14_0_0_set_soft_freq_limited_range(smu, SMU_DCLK, dclk_min, - dclk_max); + dclk_max, + false); if (ret) return ret; } @@ -1483,7 +1498,8 @@ static int smu_v14_0_common_set_performance_level(struct smu_context *smu, ret = smu_v14_0_0_set_soft_freq_limited_range(smu, SMU_DCLK1, dclk1_min, - dclk1_max); + dclk1_max, + false); if (ret) return ret; } @@ -1533,6 +1549,14 @@ static int smu_v14_0_0_set_vpe_enable(struct smu_context *smu, 0, NULL); } +static int smu_v14_0_0_set_isp_enable(struct smu_context *smu, + bool enable) +{ + return smu_cmn_send_smc_msg_with_param(smu, enable ? + SMU_MSG_PowerUpIspByTile : SMU_MSG_PowerDownIspByTile, + ISP_ALL_TILES_MASK, NULL); +} + static int smu_v14_0_0_set_umsch_mm_enable(struct smu_context *smu, bool enable) { @@ -1662,6 +1686,7 @@ static const struct pptable_funcs smu_v14_0_0_ppt_funcs = { .gfx_off_control = smu_v14_0_gfx_off_control, .mode2_reset = smu_v14_0_0_mode2_reset, .get_dpm_ultimate_freq = smu_v14_0_common_get_dpm_ultimate_freq, + .set_soft_freq_limited_range = smu_v14_0_0_set_soft_freq_limited_range, .od_edit_dpm_table = smu_v14_0_od_edit_dpm_table, .print_clk_levels = smu_v14_0_0_print_clk_levels, .force_clk_levels = smu_v14_0_0_force_clk_levels, @@ -1669,6 +1694,7 @@ static const struct pptable_funcs smu_v14_0_0_ppt_funcs = { .set_fine_grain_gfx_freq_parameters = smu_v14_0_common_set_fine_grain_gfx_freq_parameters, .set_gfx_power_up_by_imu = smu_v14_0_set_gfx_power_up_by_imu, .dpm_set_vpe_enable = smu_v14_0_0_set_vpe_enable, + .dpm_set_isp_enable = smu_v14_0_0_set_isp_enable, .dpm_set_umsch_mm_enable = smu_v14_0_0_set_umsch_mm_enable, .get_dpm_clock_table = smu_v14_0_common_get_dpm_table, .set_mall_enable = smu_v14_0_common_set_mall_enable, diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c index 82c2db972491..3aea32baea3d 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c @@ -502,8 +502,6 @@ static int smu_v14_0_2_set_default_dpm_table(struct smu_context *smu) PPTable_t *pptable = table_context->driver_pptable; SkuTable_t *skutable = &pptable->SkuTable; struct smu_14_0_dpm_table *dpm_table; - struct smu_14_0_pcie_table *pcie_table; - uint32_t link_level; int ret = 0; /* socclk dpm table setup */ @@ -619,27 +617,6 @@ static int smu_v14_0_2_set_default_dpm_table(struct smu_context *smu) dpm_table->max = dpm_table->dpm_levels[0].value; } - /* lclk dpm table setup */ - pcie_table = &dpm_context->dpm_tables.pcie_table; - pcie_table->num_of_link_levels = 0; - for (link_level = 0; link_level < NUM_LINK_LEVELS; link_level++) { - if (!skutable->PcieGenSpeed[link_level] && - !skutable->PcieLaneCount[link_level] && - !skutable->LclkFreq[link_level]) - continue; - - pcie_table->pcie_gen[pcie_table->num_of_link_levels] = - skutable->PcieGenSpeed[link_level]; - pcie_table->pcie_lane[pcie_table->num_of_link_levels] = - skutable->PcieLaneCount[link_level]; - pcie_table->clk_freq[pcie_table->num_of_link_levels] = - skutable->LclkFreq[link_level]; - pcie_table->num_of_link_levels++; - - if (link_level == 0) - link_level++; - } - /* dcefclk dpm table setup */ dpm_table = &dpm_context->dpm_tables.dcef_table; if (smu_cmn_feature_is_enabled(smu, SMU_FEATURE_DPM_DCN_BIT)) { @@ -1487,10 +1464,31 @@ static int smu_v14_0_2_update_pcie_parameters(struct smu_context *smu, struct smu_14_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context; struct smu_14_0_pcie_table *pcie_table = &dpm_context->dpm_tables.pcie_table; - int num_of_levels = pcie_table->num_of_link_levels; + int num_of_levels; uint32_t smu_pcie_arg; - int ret, i; + uint32_t link_level; + struct smu_table_context *table_context = &smu->smu_table; + PPTable_t *pptable = table_context->driver_pptable; + SkuTable_t *skutable = &pptable->SkuTable; + int ret = 0; + int i; + pcie_table->num_of_link_levels = 0; + for (link_level = 0; link_level < NUM_LINK_LEVELS; link_level++) { + if (!skutable->PcieGenSpeed[link_level] && + !skutable->PcieLaneCount[link_level] && + !skutable->LclkFreq[link_level]) + continue; + + pcie_table->pcie_gen[pcie_table->num_of_link_levels] = + skutable->PcieGenSpeed[link_level]; + pcie_table->pcie_lane[pcie_table->num_of_link_levels] = + skutable->PcieLaneCount[link_level]; + pcie_table->clk_freq[pcie_table->num_of_link_levels] = + skutable->LclkFreq[link_level]; + pcie_table->num_of_link_levels++; + } + num_of_levels = pcie_table->num_of_link_levels; if (!num_of_levels) return 0; @@ -1505,30 +1503,40 @@ static int smu_v14_0_2_update_pcie_parameters(struct smu_context *smu, for (i = 0; i < num_of_levels; i++) { pcie_table->pcie_gen[i] = pcie_gen_cap; pcie_table->pcie_lane[i] = pcie_width_cap; + smu_pcie_arg = i << 16; + smu_pcie_arg |= pcie_table->pcie_gen[i] << 8; + smu_pcie_arg |= pcie_table->pcie_lane[i]; + + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_OverridePcieParameters, + smu_pcie_arg, + NULL); + if (ret) + break; } } else { for (i = 0; i < num_of_levels; i++) { - if (pcie_table->pcie_gen[i] > pcie_gen_cap) - pcie_table->pcie_gen[i] = pcie_gen_cap; - if (pcie_table->pcie_lane[i] > pcie_width_cap) - pcie_table->pcie_lane[i] = pcie_width_cap; - } - } - - for (i = 0; i < num_of_levels; i++) { - smu_pcie_arg = i << 16; - smu_pcie_arg |= pcie_table->pcie_gen[i] << 8; - smu_pcie_arg |= pcie_table->pcie_lane[i]; + if (pcie_table->pcie_gen[i] > pcie_gen_cap || + pcie_table->pcie_lane[i] > pcie_width_cap) { + pcie_table->pcie_gen[i] = pcie_table->pcie_gen[i] > pcie_gen_cap ? + pcie_gen_cap : pcie_table->pcie_gen[i]; + pcie_table->pcie_lane[i] = pcie_table->pcie_lane[i] > pcie_width_cap ? + pcie_width_cap : pcie_table->pcie_lane[i]; + smu_pcie_arg = i << 16; + smu_pcie_arg |= pcie_table->pcie_gen[i] << 8; + smu_pcie_arg |= pcie_table->pcie_lane[i]; - ret = smu_cmn_send_smc_msg_with_param(smu, + ret = smu_cmn_send_smc_msg_with_param(smu, SMU_MSG_OverridePcieParameters, smu_pcie_arg, NULL); - if (ret) - return ret; + if (ret) + break; + } + } } - return 0; + return ret; } static const struct smu_temperature_range smu14_thermal_policy[] = { diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c index 7eaf58fd7f9a..59f9abd0f7b8 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c @@ -86,6 +86,7 @@ static void smu_cmn_read_arg(struct smu_context *smu, #define SMU_RESP_BUSY_OTHER 0xFC #define SMU_RESP_DEBUG_END 0xFB +#define SMU_RESP_UNEXP (~0U) /** * __smu_cmn_poll_stat -- poll for a status from the SMU * @smu: a pointer to SMU context @@ -171,6 +172,15 @@ static void __smu_cmn_reg_print_error(struct smu_context *smu, dev_err_ratelimited(adev->dev, "SMU: I'm debugging!"); break; + case SMU_RESP_UNEXP: + if (amdgpu_device_bus_status_check(smu->adev)) { + /* print error immediately if device is off the bus */ + dev_err(adev->dev, + "SMU: response:0x%08X for index:%d param:0x%08X message:%s?", + reg_c2pmsg_90, msg_index, param, message); + break; + } + fallthrough; default: dev_err_ratelimited(adev->dev, "SMU: response:0x%08X for index:%d param:0x%08X message:%s?", diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h index 7473672abd2a..a608cdbdada4 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h +++ b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h @@ -40,28 +40,29 @@ #define SMU_IH_INTERRUPT_CONTEXT_ID_FAN_ABNORMAL 0x8 #define SMU_IH_INTERRUPT_CONTEXT_ID_FAN_RECOVERY 0x9 -#define smu_cmn_init_soft_gpu_metrics(ptr, frev, crev) \ - do { \ - typecheck(struct gpu_metrics_v##frev##_##crev, \ - typeof(*(ptr))); \ - struct metrics_table_header *header = \ - (struct metrics_table_header *)(ptr); \ - memset(header, 0xFF, sizeof(*(ptr))); \ - header->format_revision = frev; \ - header->content_revision = crev; \ - header->structure_size = sizeof(*(ptr)); \ +#define smu_cmn_init_soft_gpu_metrics(ptr, frev, crev) \ + do { \ + typecheck(struct gpu_metrics_v##frev##_##crev *, (ptr)); \ + struct gpu_metrics_v##frev##_##crev *tmp = (ptr); \ + struct metrics_table_header *header = \ + (struct metrics_table_header *)tmp; \ + memset(header, 0xFF, sizeof(*tmp)); \ + header->format_revision = frev; \ + header->content_revision = crev; \ + header->structure_size = sizeof(*tmp); \ } while (0) -#define smu_cmn_init_partition_metrics(ptr, frev, crev) \ - do { \ - typecheck(struct amdgpu_partition_metrics_v##frev##_##crev, \ - typeof(*(ptr))); \ - struct metrics_table_header *header = \ - (struct metrics_table_header *)(ptr); \ - memset(header, 0xFF, sizeof(*(ptr))); \ - header->format_revision = frev; \ - header->content_revision = crev; \ - header->structure_size = sizeof(*(ptr)); \ +#define smu_cmn_init_partition_metrics(ptr, fr, cr) \ + do { \ + typecheck(struct amdgpu_partition_metrics_v##fr##_##cr *, \ + (ptr)); \ + struct amdgpu_partition_metrics_v##fr##_##cr *tmp = (ptr); \ + struct metrics_table_header *header = \ + (struct metrics_table_header *)tmp; \ + memset(header, 0xFF, sizeof(*tmp)); \ + header->format_revision = fr; \ + header->content_revision = cr; \ + header->structure_size = sizeof(*tmp); \ } while (0) extern const int link_speed[]; |