aboutsummaryrefslogtreecommitdiffstats
path: root/src (follow)
Commit message (Collapse)AuthorAgeFilesLines
...
* estream: Fix modestring parsing.Justus Winter2016-10-181-1/+1
| | | | | | * src/estream.c (parse_mode): Fix parsing the 'sysopen' flag. Signed-off-by: Justus Winter <[email protected]>
* Add error code USER_ID_EXISTS, NAME_EXISTS, and DUP_NAME.Werner Koch2016-10-071-1/+3
| | | | Signed-off-by: Werner Koch <[email protected]>
* syscfg: Add support for {i686,x86_64}-apple-darwin.Werner Koch2016-10-073-0/+56
| | | | | | | | -- Thanks to Chris Ballinger <[email protected]> for the files. Signed-off-by: Werner Koch <[email protected]>
* Add error WINDOW_TOO_SMALL, WINDOW_TOO_LARGE, and MISSING_ENVVAR.Werner Koch2016-09-011-1/+5
| | | | Signed-off-by: Werner Koch <[email protected]>
* New error code GPG_ERR_ENGINE_TOO_OLDWerner Koch2016-08-161-1/+2
|
* Fix build without threadsYann E. MORIN2016-07-121-1/+3
| | | | | | | | | | | | | | | * src/gen-posix-lock-obj.c: properly guard inclusioin of pthread.h * tests/t-lock.c: likewise * tests/t-poll.c: likewise -- Although ./configure checks for thread support, gen-posix-lock-obj and two tests still include pthread.h unconditionally. Guard that inclusion using the same condition as all other uses of pthread-related code. Signed-off-by: "Yann E. MORIN" <[email protected]>
* Define EWOULDBLOCK in case it is not definedAndre Heinecke2016-07-051-0/+5
| | | | | | | * src/estream.c (EWOULDBLOCK): Define fallback. -- Older mingw versions (2.0) do not define EWOULDBLOCK in errno.h
* estream: Fix bug es_fclose_snatch if a seek has been used.Werner Koch2016-06-271-1/+1
| | | | | | * src/estream.c (func_mem_ioctl): Set LEN from DATA_LEN. Signed-off-by: Werner Koch <[email protected]>
* w32: Silence compiler warnings about redefined macros.Werner Koch2016-06-251-6/+19
| | | | | | * src/estream.c (S_IRGRP) [W32]: Protect against redefinition. Signed-off-by: Werner Koch <[email protected]>
* estream: Remove two compiler warning.Werner Koch2016-06-241-40/+41
| | | | | | | | | * src/estream.c (func_file_create): Remove dead assignment. (doreadline): Do not decrement SPACE_LEFT before breaking the loop. Add an extra block to limit the scope of that variable. -- Signed-off-by: Werner Koch <[email protected]>
* estream: Fix assertion failure due to es_flush.NIIBE Yutaka2016-06-151-3/+1
| | | | | | | | | | | | | | | | | | | | | | | | * src/estream.c (es_writen): Set writing flag even if no data was written. -- GnuPG-bug-id: 2371 Signed-off-by: Werner Koch <[email protected]> gniibe tracked the problem down to a fully valid change in GnuPG (gnupg commit 12af2630cf4d1a39179179925fac8f2cce7504ff). He wrote: This is the first instance for estream to do READ and WRITE (in the history of the code > 10 years :-). In the [gnupg] function agent_write_private_key, the pattern is: es_fopen es_fread es_fseek es_fwrite which should work well, but if results core dump by assertion failure in the function es_flush of libgpg-error.
* Adjust memory limit of es_fopenmem to the block size.Werner Koch2016-06-151-0/+8
| | | | | | | | | | * src/estream.c (func_mem_create): Round up memory limit. -- This is required so that giving a memory limit to es_fopenmem won't fail if it is below the block size. Signed-off-by: Werner Koch <[email protected]>
* Add GPG_ERR_SUBKEYS_EXP_OR_REV.Werner Koch2016-05-171-1/+2
|
* syscfg: Add a powerpc and a tilgegx architecture.Werner Koch2016-05-073-0/+50
| | | | | | | | | | * src/syscfg/lock-obj-pub.powerpc-unknown-linux-gnuspe.h: New. * src/syscfg/lock-obj-pub.tilegx-unknown-linux-gnu.h: New. * src/Makefile.am (lock_obj_pub): Add them. -- Debian-bug-id: 823630, 823631 Signed-off-by: Werner Koch <[email protected]>
* Release 1.22.libgpg-error-1.22Werner Koch2016-04-253-3/+3
| | | | | | * configure.ac: Set LT version to C18/A18/R0. Signed-off-by: Werner Koch <[email protected]>
* Fix for HPPA.NIIBE Yutaka2016-04-252-4/+11
| | | | | | | | | | | | | | * configure.ac (HAVE_GCC_ATTRIBUTE_ALIGNED): Revert. * src/gen-posix-lock-obj.c (USE_16BYTE_ALIGNMENT): Revert. * src/syscfg/lock-obj-pub.hppa-unknown-linux-gnu.h: Revert. -- Signed-off-by: NIIBE Yutaka <[email protected]> This fixes commit 5168b97. I considered wrongly that long double were 128-bit in HPPA. Currently, all that we can do for HPPA is assuming GCC and use its extension of "aligned" attribute.
* syscfg: Add powerpc-unknown-linux-gnuspeWerner Koch2016-04-221-0/+1
| | | | --
* w32: Add iconv functions.Werner Koch2016-04-214-399/+130
| | | | | | | | | | | | * src/w32-add.h (gpgrt_w32_iconv_t): New. (gpgrt_w32_iconv_open, gpgrt_w32_iconv_close, gpgrt_w32_iconv): New. ( GPGRT_ENABLE_W32_ICONV_MACROS): New * src/w32-iconv.c: Change license to LGPLv2.1+. Dispable mlang feature. Remove external DLL loading. Simplify iconv functions. Use cleaner context struct pattern. Use gpgrt namespace. * src/gpg-error.def.in: Add new functions. Signed-off-by: Werner Koch <[email protected]>
* Add file w32-iconv.cWerner Koch2016-04-201-0/+2086
| | | | --
* estream,w32: Temporary fix for gpgrt_poll.Werner Koch2016-04-051-0/+4
| | | | | | | | | | | * src/estream.c (_gpgrt_poll) [W32]: Do not use FD_ISSET. -- gpgrt_poll return an error on Windows anyway and thus it does not make sense to use a possible undefined macro here. Reported-by: Andre Heinecke Signed-off-by: Werner Koch <[email protected]>
* estream: Prepare for new mode flag "sysopen".Werner Koch2016-03-291-14/+47
| | | | | | * src/estream.c (parse_mode): Add arg "sysopen". Adjust all callers. Signed-off-by: Werner Koch <[email protected]>
* estream: Use simpler names for static functions.Werner Koch2016-03-291-44/+44
| | | | | | * src/estream.c: Replace all es_func_* to just func_*. Signed-off-by: Werner Koch <[email protected]>
* doc: Re-format comments in estream.cWerner Koch2016-03-291-150/+311
| | | | | -- Signed-off-by: Werner Koch <[email protected]>
* estream: Remove strange macro for better readability.Werner Koch2016-03-291-15/+11
| | | | | | | * src/estream.c (SET_UNLESS_NONZERO): Remove macro. (es_deinitialize): Replace that macro by direct code. Signed-off-by: Werner Koch <[email protected]>
* Add function gpgrt_annotate_leaked_object.Peter Wu2016-03-241-0/+30
| | | | | | | | | | | | | | | | | | | | | | * src/gpg-error.h.in: add gpgrt_annotate_leaked_object to support marking memory as non-leaked for Clang and GCC. -- This annotation can be used to mark objects as explicitly leaked such that it can be ignored in tools like LeakSanitizer. The GPGRT_HAVE_LEAK_SANITIZER macro is explicitly not undefined to support -fsanitize=leak, a user or configure script could then decide to add this macro when just -fsanitize=leak is given. Signed-off-by: Peter Wu <[email protected]> Additional changes by -wk: - But __GNUC__ guard around the entire GPGRT_HAVE_LEAK_SANITIZER detection. - Add NEWS entry. Signed-off-by: Werner Koch <[email protected]>
* syscfg: Add lock-obj-pub files for {armv5, armv6, x86_64}-musl targetsKylie McClain2016-03-144-0/+74
| | | | | | | | | | | | | | | * src/syscfg/lock-obj-pub.armv5-unknown-linux-musleabi.h: New. * src/syscfg/lock-obj-pub.armv6-unknown-linux-musleabihf.h New. * src/syscfg/lock-obj-pub.x86_64-pc-linux-musl.h: New. * src/Makefile.am (lock_obj_pub): Add files. -- This patch adds three new precompiled lock-obj-pub files: - armv5-unknown-linux-musleabi - armv6-unknown-linux-musleabihf - x86_64-pc-linux-musl ChangeLog lines and Makefile patch by -wk
* Fix detecting Solaris operating system.NIIBE Yutaka2016-03-031-1/+2
| | | | | | | | | | | | | | | | | | | | * src/gen-posix-lock-obj.c (USE_DOUBLE_FOR_ALIGNMENT): Check for the macro __sun. -- Signed-off-by: NIIBE Yutaka <[email protected]> This fixes commit 5168b97 and 34b0714. Thanks to Ibraheem Saleh and Thomas Klausner for testing. Thanks to Tom G. Christensen for suggesting the fix with the useful link: https://sourceforge.net/p/predef/wiki/Home/ Thanks to Nelson H. F. Beebe to inform the tool: http://www.math.utah.edu/~beebe/cc-defs GnuPG-bug-id: 2144
* Fix for Solaris.NIIBE Yutaka2016-03-011-1/+1
| | | | | | | | | | * src/gen-posix-lock-obj.c (USE_DOUBLE_FOR_ALIGNMENT): Check LP64. -- Signed-off-by: NIIBE Yutaka <[email protected]> This fixes commit 5168b97.
* Add support for Solaris, fixing HPPA.NIIBE Yutaka2016-02-262-10/+12
| | | | | | | | | | | | * configure.ac (HAVE_GCC_ATTRIBUTE_ALIGNED): Remove. (LIB_SCHED_YIELD): New. Check sched_yield in -lrt. * src/gen-posix-lock-obj.c (USE_16BYTE_ALIGNMENT): Remove. (USE_DOUBLE_FOR_ALIGNMENT, USE_LONG_DOUBLE_FOR_ALIGNMENT): New. * src/syscfg/lock-obj-pub.hppa-unknown-linux-gnu.h: Update. -- Signed-off-by: NIIBE Yutaka <[email protected]>
* w32: Confirm the 639-1 code for Venda is "ve"Daniel Kahn Gillmor2016-02-081-5/+1
| | | | | | | | | | | -- https://www.loc.gov/standards/iso639-2/php/code_changes.php says that ve is indeed the 2-character code for Venda: >> This alpha-2 ISO 639-1 code was approved in 1999 and included in >> ISO 639-1: 2002. It was mistakenly missing in earlier versions of >> the tables on the ISO639-2 web site. As of January 2003, it has >> been included.
* Convert http links to https where possible in the source.Daniel Kahn Gillmor2016-02-0826-27/+27
| | | | | | | | | | | | | | | | -- * use https for bug reporting * in comments and docs, use https to refer to: - www.gnu.org - creativecommons.org - translationproject.org - mail.gnome.org - www.perl.org - www.ctan.org - www.cl.cam.ac.uk - www.ntg.nl - cygwin.com - www.ethnologue.com
* Add syscfg alias for x86_64-pc-linux-gnuhardened1Werner Koch2016-01-271-0/+1
| | | | --
* Add GPG_ERR_DB_CORRUPTED.Werner Koch2016-01-191-2/+2
| | | | Signed-off-by: Werner Koch <[email protected]>
* Add new lock-obj-pub for NIOS2.Marek Vasut2016-01-132-0/+24
| | | | src/syscfg/lock-obj-pub.nios2-unknown-linux-gnu.h: New.
* Fix windows 8bit encoding conversionAndre Heinecke2015-12-071-5/+13
| | | | | | | | | | * src/w32-gettext.c (wchar_to_native): Convert to ConsoleOutputCP. -- Just using CP_ACP did not neccessarily convert to the correct codepage as codepages might differ between ConsoleOutput and GUI output (which it usually does). This fixes encoding issues on the Windows commandline.
* Fix typos found by codespell.Justus Winter2015-11-193-6/+6
| | | | | -- Signed-off-by: Justus Winter <[email protected]>
* Avoid 'malloc' corner case.Justus Winter2015-11-191-3/+3
| | | | | | | | | | | | | * src/init.c (_gpgrt_realloc): Avoid calling 'malloc(0)'. -- Previously, if '_gpgrt_realloc' was called with both A and N being zero, malloc is invoked with a size of zero. This happens e.g. when calling '_gpgrt_free' with a NULL pointer, which is supposed to be a no-op. Found using the Clang Static Analyzer. Signed-off-by: Justus Winter <[email protected]>
* Add error codes NO_NAME, NO_KEY, and SERVER_FAILURE.Werner Koch2015-10-211-1/+4
|
* estream: Avoid calling write(fd,NULL,n).Werner Koch2015-10-181-20/+32
| | | | | | | | | | | | | | | * src/estream.c (es_func_fd_write): Take care of a flush requests. (es_func_w32_write): Ditto. (es_func_fp_write): Ditto. -- The write handlers may be called with (BUFFER,SIZE) of (NULL,0) to propagate flush events to user supplied functions (es_fopencookie). However we need to take care to also do this with the internal handler. Might be a reason for system faults due to write or fwrite with a NULL buffer; on Linux this seems to be harmless. Signed-off-by: Werner Koch <[email protected]>
* estream: Keep track of EPIPE.Werner Koch2015-09-281-4/+25
| | | | | | | | | * src/estream.c (_gpgrt_stream_internal): Add indicators.hup. (init_stream_obj): Init it. (es_fill, es_flush, es_seek): Set that. (_gpgrt_poll): Set event. Signed-off-by: Werner Koch <[email protected]>
* Add GPG_ERR_FALSE and GPG_ERR_TRUE error codes.Werner Koch2015-09-281-1/+2
| | | | Signed-off-by: Werner Koch <[email protected]>
* estream: Add gpgrt_set_nonblock and gpgrt_poll.Werner Koch2015-09-257-4/+413
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * configure.ac (AC_CHECK_HEADERS): Add sys/select.h and sys/time.h. * src/estream.c: Include both header if available. (COOKIE_IOCTL_NONBLOCK): New. (struct estream_cookie_fd): Add field nonblock. (func_fd_create): Set nonblock from MODEFLAGS. (es_func_fd_ioctl): New. (parse_mode): Add modeflag "nonblock". (es_fill): Map EWOULDBLOCK to EAGAIN. Do not set error indicator for EAGAIN. (es_flush, es_seek, es_write_nbf): Map EWOULDBLOCK to EAGAIN. (do_fdopen): Call COOKIE_IOCTL_NONBLOCK. (_gpgrt_set_nonblock): New. (_gpgrt_get_nonblock): New. (_gpgrt_poll): New. * src/gpg-error.h.in (struct _gpgrt_poll_s): New. (gpgrt_poll_t, es_poll_t): New. (es_set_nonblock, es_get_nonblock, es_poll): New. * src/gpg-error.vers, src/gpg-error.def.in: Add gpgrt_set_nonblock, gpgrt_get_nonblock, and gpgrt_poll. * src/visibility.c (gpgrt_set_nonblock, gpgrt_get_nonblock): New. (gpgrt_poll): New. * tests/t-common.h (DIM): New. * tests/t-poll.c: New. * tests/Makefile.am (TESTS): Add t-poll. (t_poll_LDADD): New. -- The poll interface uses select(2) internally because that is more portable than poll(2). Signed-off-by: Werner Koch <[email protected]>
* estream: Replace indicator help functions to ease code reading.Werner Koch2015-09-241-33/+13
| | | | | | | * src/estream.c (es_set_indicators, es_get_indicator): Remove and change callers to set/get indicators directly. Signed-off-by: Werner Koch <[email protected]>
* Support i[456]86-pc{,-linux,-kfreebsd}-gnu.hDaniel Kahn Gillmor2015-09-175-27/+8
| | | | | | | | | | | | | | | | | | | | | | | | | * src/mkheader (canon_host_triplet): Add new entries. * src/syscfg/lock-obj-pub.i486-pc-gnu.h: Rename to ... * src/syscfg/lock-obj-pub.i686-pc-gnu.h: this. * src/syscfg/lock-obj-pub.i486-pc-kfreebsd-gnu.h: Rename to ... * src/syscfg/lock-obj-pub.i686-pc-kfreebsd-gnu.h: this. * src/syscfg/lock-obj-pub.i486-pc-linux-gnu.h: Remove. * src/Makefile.am (lock_obj_pub): Update. -- i486-pc, i586-pc, and i686-pc all use the same ABI given the GNU userland and a specific kernel. This changeset updates the arch-specific lock-obj header generation to treat the hardware aliases explicitly and should improve cross-building for anyone wanting to target any of these 9 architectures. It also removes src/syscfg/lock-obj-pub.i486-pc-linux-gnu.h, which should be handled by the alias to src/syscfg/lock-obj-pub.i686-pc-linux-gnu.h Debian-Bug-Id: 799177
* Add new version macros.Werner Koch2015-08-261-6/+9
| | | | | | | | | | | | | * src/gpg-error.h.in (GPGRT_VERSION): New. (GPGRT_VERSION_NUMBER): New. (GPG_ERROR_VERSION, GPG_ERROR_VERSION_NUMBER): Move to top of file. -- Eventually we will have a gpgrt.h file instead of gpg-error. To make things easier we already provide suitable named version macros. Miving them to the top makes it easier to find the version. Signed-off-by: Werner Koch <[email protected]>
* Add macro GPGRT_INLINE and avoid -Wundef warningsWerner Koch2015-08-261-27/+28
| | | | | | | | | | | | * src/gpg-error.h.in (GPG_ERR_INLINE): Use #if defined for possible undefined macros to avoid warnign with GCC's -Wundef option. (GPGRT_INLINE): New. -- I still consider using -Wundef for regular building a bad behaviour because undefined macros have a well defined value in a cpp conditional. That warning is useful for debugging build problems but should not be used as standard warning option. Anyway, here is a fix.
* w32: Make sure the setmode is called.Werner Koch2015-08-251-1/+7
| | | | | | * src/estream.c (HAVE_DOSISH_SYSTEM): Define if needed. Signed-off-by: Werner Koch <[email protected]>
* Add option --lib-version to the gpg-error tool.Werner Koch2015-07-271-4/+14
| | | | * src/gpg-error.c (main): Add new option.
* Add new public macros for GCC attributes.Werner Koch2015-07-242-22/+100
| | | | | | | | | | * src/gpg-error.h.in (GPGRT_GCC_VERSION): New. (GPGRT_ATTR_NORETURN, GPGRT_ATTR_PRINTF, GPGRT_ATTR_NR_PRINTF): New. (GPGRT_ATTR_FORMAT_ARG, GPGRT_ATTR_SENTINEL): New. (GPGRT_ATTR_USED, GPGRT_ATTR_UNUSED, GPGRT_ATTR_DEPRECATED): New. (GPGRT_ATTR_PURE, GPGRT_ATTR_MALLOC): New. (GPGRT_HAVE_MACRO_FUNCTION, GPGRT_HAVE_PRAGMA_GCC_PUSH): New. (_GPGRT_GCC_A_PRINTF): Replace GPGRT_ATTR_PRINTF.
* Allow building with --disable-threads.Werner Koch2015-06-154-20/+64
| | | | | | | | | | | | | | | | | | | | * src/posix-lock-obj.h (LOCK_ABI_NOT_AVAILABLE): New. (LOCK_ABI_VERSION): Define depending on USE_POSIX_THREADS. (_gpgrt_lock_t) [!USE_POSIX_THREADS]: Do not define the union. * src/gen-posix-lock-obj.c: Take care of USE_POSIX_THREADS. * src/posix-lock.c (_gpgrt_lock_init, _gpgrt_lock_lock) (_gpgrt_lock_trylock, _gpgrt_lock_unlock) (_gpgrt_lock_destroy): Return success for a no-threads version. * tests/t-lock.c: Disable tests if threads are not available. * src/mkheader.c (main): Add NO-THREADS to the printed comment. * configure.ac: Show NO-TRHEADS in the final summary. -- Warning: Using --disable-threads creates a different ABI which we can't encode in the the cpu-vendor-os triplet. The run time checks should detect this and abort the process. Signed-off-by: Werner Koch <[email protected]>