- Feb 01, 2020
-
-
Siddhesh Poyarekar authored
We're ready to tag!
-
Siddhesh Poyarekar authored
-
Siddhesh Poyarekar authored
-
Siddhesh Poyarekar authored
Updated versions of gcc, binutils, awk, bison and perl.
-
Siddhesh Poyarekar authored
-
Siddhesh Poyarekar authored
It was discovered that RISC-V 64-bit still needs kernel headers >= 5.0 to be installed to build correctly: https://gitlab.com/kubu93/toolchains-builder/-/jobs/422726962 Fix up the NEWS item to make that clear.
-
Siddhesh Poyarekar authored
-
- Jan 31, 2020
-
-
Szabolcs Nagy authored
This internal change ideally should not affect the public API or ABI, but there is a widespread use of seccomp sandboxes, even on 32-bit targets, that don't handle new Linux syscall usage well, so it's worth mentioning in the NEWS. Co-authored-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
-
- Jan 30, 2020
-
-
Florian Weimer authored
Commit 1c3f9acf ("nptl: Add struct_mutex.h") replaced a zero constant with the identifier PTHREAD_MUTEX_DEFAULT in the macro PTHREAD_MUTEX_INITIALIZER. However, that constant is not available in ISO C11 mode: In file included from /usr/include/bits/thread-shared-types.h:74, from /usr/include/bits/pthreadtypes.h:23, from /usr/include/pthread.h:26, from bug25271.c:1: bug25271.c:3:21: error: ‘PTHREAD_MUTEX_DEFAULT’ undeclared here (not in a function) 3 | pthread_mutex_t m = PTHREAD_MUTEX_INITIALIZER; | ^~~~~~~~~~~~~~~~~~~~~~~~~ This commit change the constant to the equivalent PTHREAD_MUTEX_TIMED_NP, which is in the POSIX extension namespace and thus always available.
-
- Jan 24, 2020
-
-
Joseph Myers authored
In testing glibc for Arm and MIPS, I see: FAIL: misc/tst-sigcontext-get_pc If this test - backtracing through a call to raise - is valid, then raise needs to be built with -fasynchronous-unwind-tables (as the test itself is) to have the required unwind information for that backtracing to work. Adding that option, which this patch does, causes the test for pass for Arm. For MIPS, the test still does not pass (the backtrace has an address that is 2 bytes after the "address in signal handler", for unknown reasons), although the patch allows a longer backtrace to be produced.
-
Joseph Myers authored
The locale/tst-locale-locpath test unsets LANG, then runs a test with test_wrapper_env and expects LANG to remain unset for that test. This does not work for cross-testing with cross-test-ssh.sh when sshd (on the system specified as an argument to cross-test-ssh.sh) is configured to have a default LANG setting. The general design used in cross testing, after commit 8540f6d2 ("Don't require test wrappers to preserve environment variables, use more consistent environment.", 6 June 2014), is that environment settings required by tests should be passed explicitly to $(test-wrapper-env). This patch changes tst-locale-locpath.sh to pass an explicit LANG= rather than expecting "unset LANG" to be in effect for the program run under test_wrapper_env. Note that this does slightly change the environment in which the test is run natively (empty LANG instead of unset LANG) but that difference does not appear relevant to what it is trying to test. Tested for Arm that this fixes the failure seen for that test in cross-testing.
-
Joseph Myers authored
As noted in <https://sourceware.org/ml/libc-alpha/2019-06/msg00824.html>, elf/tst-rtld-preload fails when cross-testing because it attempts to run the test wrapper with itself. Unfortunately, that thread never resulted in a complete and correct patch for that test. This patch addresses the issues with that test more thoroughly. The test is changed not to use the wrapper twice, including updating the message it prints about the command it runs to be more complete and accurate after the change; the Makefile is changed not to pass the redundant '$(test-wrapper)' argument. Tested for Arm that this fixes the failure seen for that test in cross-testing.
-
- Jan 23, 2020
-
-
Joseph Myers authored
The tests elf/tst-ifunc-fault-bindnow and elf/tst-ifunc-fault-lazy fail in cross-testing because they run the dynamic linker directly without using the test wrapper. This patch fixes them to use the test wrapper instead. Tested that this fixes the failure of those two tests for powerpc soft-float.
-
Siddhesh Poyarekar authored
The ChangeLog automation scripts were incorporated in gnulib as vcs-to-changelog for a while now since other projects expressed the desire to use and extend this script. In the interest of avoiding duplication of code, drop the glibc version of gitlog-to-changelog and use the gnulib one directly. The only file that remains is vcstocl_quirks.py, which specifies properties and quirks of the glibc project source code. This patch also drops the shebang at the start of vcstocl_quirks.py since the file is not intended to be directly executable.
-
Dragan Mladjenovic authored
The change was introduced in: commit 33bc9efd Author: Dragan Mladjenovic <dmladjenovic@wavecomp.com> Date: Fri Aug 23 16:38:04 2019 +0000 mips: Force RWX stack for hard-float builds that can run on pre-4.8 kernels and probably requires a small explanation. Co-authored-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
-
- Jan 21, 2020
-
-
Andreas Schwab authored
When unwinding through a signal frame the backtrace function on PowerPC didn't check array bounds when storing the frame address. Fixes commit d400dcac ("PowerPC: fix backtrace to handle signal trampolines").
-
- Jan 20, 2020
-
-
Florian Weimer authored
Filip Ochnik spotted that one of the error jumps in gethosts fails to call __resolv_context_put to release the resolver context. Fixes commit 352f4ff9 ("resolv: Introduce struct resolv_context [BZ #21668]") and commit 964263bb ("getaddrinfo: Release resolver context on error in gethosts [BZ #21885]"). Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
-
Joseph Myers authored
benchtests/timing-type is built with the newly built libc, so should be run with it like actual tests and benchmarks.
-
Stefan Liebler authored
The commit 446997ff introduced this new usage of resplen. If build with gcc 9 -march>=z13 on s390x, the following warning occurs: res_send.c: In function ‘__res_context_send’: res_send.c:539:6: error: ‘resplen’ may be used uninitialized in this function [-Werror=maybe-uninitialized] 539 | if (resplen > HFIXEDSZ) | ^ Therefore this patch adds a further DIAG_IGNORE_NEEDS_COMMENT in the same way as it was previously done for usages of resplen or n. See commit d1bc2cbb.
-
- Jan 19, 2020
-
-
Siddhesh Poyarekar authored
Update translations after adding msgattrib to the update-translations makefile target.
-
Siddhesh Poyarekar authored
The translation project coordinator Benno Schulenberg suggested that we could save space in our tarball by trimming the generated po files by using msgattrib and dropping all untranslated, fuzzy and obsolete messages. This patch updates the update-translations target to do that. Testing indicates that the current po files reduce by over 65K lines due to this trimming.
-
Siddhesh Poyarekar authored
Update translations from the translation project.
-
Siddhesh Poyarekar authored
The latest translations in the translationproject URL need to be merged in using msgmerge for the po files to be correctly updated, otherwise we may end up getting odd results, such as the previous translations update. This patch adds another step to the update-translations Makefile target which does a msgmerge of the downloaded po file with libc.pot and then uses that as the final result.
-
- Jan 18, 2020
-
-
Matheus Castanho authored
The build has been failing on powerpc64le-linux-gnu with GCC 10 due to a maybe-uninitialized error: ../sysdeps/ieee754/dbl-64/mpa.c:875:6: error: ‘w.e’ may be used uninitialized in this function [-Werror=maybe-uninitialized] 875 | EY -= EX; | ^~ The warning is thrown because when __inv is called by __dvd *y is not initialized and if t == 0 before calling __dbl_mp, EY will stay uninitialized, as the function does not touch it in this case. However, since t will be set to 1/t before calling __dbl_mp, t == 0 will never happen, so we can instruct the compiler to ignore this case, which suppresses the warning. Tested on powerpc64le. Reviewed-by: Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
-
- Jan 17, 2020
-
-
Andreas Schwab authored
-
Florian Weimer authored
pvalloc is guarantueed to round up the allocation size to the page size, so applications can assume that the memory region is larger than the passed-in argument. The alloc_size attribute cannot express that. The test case is based on a suggestion from Jakub Jelinek. This fixes commit 9bf8e29c ("malloc: make malloc fail with requests larger than PTRDIFF_MAX (BZ#23741)"). Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
-
Lucas A. M. Magalhaes authored
This test was failing in some powerpc systems as it was not checking for ENOSPC return. As said on the Linux man-pages and can be observed by the implementation at mm/mprotect.c in the Linux Kernel source. The syscall pkey_alloc can return EINVAL or ENOSPC. ENOSPC will indicate either that all keys are in use or that the kernel does not support pkeys. Reviewed-by: Gabriel F. T. Gomes <gabriel@inconstante.net.br>
-
Tulio Magno Quites Machado Filho authored
GCC 10.0 enabled -fno-common by default and this started to point that __cache_line_size had been implemented in 2 different places: loader and libc. In order to avoid this duplication, the libc variable has been removed and the loader variable is moved to rtld_global_ro. File sysdeps/unix/sysv/linux/powerpc/dl-auxv.h has been added in order to reuse code for both static and dynamic linking scenarios. Reviewed-by: Carlos O'Donell <carlos@redhat.com>
-
Tulio Magno Quites Machado Filho authored
Initialize dl_auxv, dl_hwcap and dl_hwcap2 in rtld_global_ro for DSOs that have been statically dlopen'ed. Reviewed-by: Carlos O'Donell <carlos@redhat.com>
-
Siddhesh Poyarekar authored
The update-translation target seems to have downloaded outdated translations for these languages. Revert them and try to figure out if this is a problem with the target or an error in translations.
-
Siddhesh Poyarekar authored
The parser cannot identify the __nonnull off the bat and confuses __typeof in the change db25266c to be a function declaration.
-
- Jan 16, 2020
-
-
Florian Weimer authored
Without CET, a jump into a newly loaded object through an overwritten link map often does not crash, it just executes some random code. CET detects this in some cases because the function pointer does not point to the start of a function in the replacement shared object, so there is no ENDBR instruction. The new test uses a small shared object and the existing dangling link map to trigger the bug. Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
-
H.J. Lu authored
We should clear GL(dl_initfirst) when freeing its link_map memory. Tested on Fedora 31/x86-64 with CET. Reviewed-by: Florian Weimer <fweimer@redhat.com>
-
Siddhesh Poyarekar authored
Update translations from translationproject.org for 2.30.9000.
-
- Jan 13, 2020
-
-
Samuel Thibault authored
The C11 threads implementation will need it.
-
Samuel Thibault authored
-
Joseph Myers authored
This patch updates build-many-glibcs.py for the move of GCC to git, teaching it to do the initial checkout from git, to replace an SVN checkout with a git one if --replace-sources is used, and to get the commit identifier from a git checkout after updating it.
-
Stefan Liebler authored
This new test was introduced with recent commit 591236f1. If run on 32bit, it fails while renaming tst-ldconfig-ld-mod.so as there is no /usr/lib64 directory. This patch is constructing the file name with help of support_libdir_prefix. Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
-
Samuel Thibault authored
C11 threads should not expose them.
-
Samuel Thibault authored
-