about summary refs log tree commit diff
Commit message (Collapse)AuthorAgeFilesLines
* x86-64: Improve branch predication in _dl_runtime_resolve_avx512_opt [BZ #21258]H.J. Lu2017-03-213-6/+15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | On Skylake server, _dl_runtime_resolve_avx512_opt is used to preserve the first 8 vector registers. The code layout is if only %xmm0 - %xmm7 registers are used preserve %xmm0 - %xmm7 registers if only %ymm0 - %ymm7 registers are used preserve %ymm0 - %ymm7 registers preserve %zmm0 - %zmm7 registers Branch predication always executes the fallthrough code path to preserve %zmm0 - %zmm7 registers speculatively, even though only %xmm0 - %xmm7 registers are used. This leads to lower CPU frequency on Skylake server. This patch changes the fallthrough code path to preserve %xmm0 - %xmm7 registers instead: if whole %zmm0 - %zmm7 registers are used preserve %zmm0 - %zmm7 registers if only %ymm0 - %ymm7 registers are used preserve %ymm0 - %ymm7 registers preserve %xmm0 - %xmm7 registers Tested on Skylake server. [BZ #21258] * sysdeps/x86_64/dl-trampoline.S (_dl_runtime_resolve_opt): Define only if _dl_runtime_resolve is defined to _dl_runtime_resolve_sse_vex. * sysdeps/x86_64/dl-trampoline.h (_dl_runtime_resolve_opt): Fallthrough to _dl_runtime_resolve_sse_vex.
* Regenerate INSTALL.Joseph Myers2017-03-212-0/+10
|
* Merge branch 'master' of ssh://sourceware.org/git/glibcThorsten Kukuk2017-03-212-1/+6
|\ | | | | | | | | Conflicts: ChangeLog
| * Fix failing test malloc/tst-interpose-nothread with GCC 7.Stefan Liebler2017-03-212-1/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The test malloc/tst-interpose-nothread fails on s390x if built with GCC 7 and glibc commit "Remove the str(n)dup inlines from string/bits/string2.h. Although inlining" (ae65d4f3c3995279ca458c460ebf8bab1885fa03) with output: error: free: 0x3fffdffa010: invalid allocation index: 0 (not less than 0) The destructor check_for_allocations in malloc/tst-interpose-aux.c is called twice. One time after the test-child-process has finished successfully and once after the test-parent-process finishes. During the latter invocation, allocation_index == 0. GCC 7 is now inlining the free function and calls unconditionally fail in get_header as header->allocation_index (type == size_t) is always >= allocation_index (= 0). Before the mentioned commit above, strdup was replaced by strlen, malloc and memcpy. The malloc call was also inlined and allocation_index was set to one. This patch moves the already existing compiler barrier before the invocation of free. ChangeLog: * malloc/tst-interpose-aux.c (check_for_allocations): Move compiler barrier before free.
* | Deprecate libnsl by default (only shared library will beThorsten Kukuk2017-03-2148-81/+300
|/ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | build for backward compatibility, no linking possible) and disable building of libnss_compat, libnss_nis and libnss_nisplus, except --enable-obsolete-nsl option is given to configure. * config.h.in: Add LINK_OBSOLETE_NSL. * config.make.in: Add build-obsolete-nsl. * configure.ac: Add obsolete-nsl option. * include/libc-symbols.h: Define libnsl_hidden_nolink_def. * include/rpcsvc/yp.h: Add missing functions as libnsl_hidden_proto. * include/rpcsvc/nislib.h: Likewise. * include/rpcsvc/ypclnt.h: Likewise. * manual/install.texi: Document --enable-obsolete-nsl. * nis/Makefile: Build only libnsl by default (add build-obsolete-nsl). * nis/nis_add.c: Replace libnsl_hidden_def with libnsl_hidden_nolink_def. * nis/nis_addmember.c: Likewise. * nis/nis_call.c: Likewise. * nis/nis_clone_obj.c: Likewise. * nis/nis_defaults.c: Likeise. * nis/nis_domain_of_r.c: Likewise. * nis/nis_error.c: Likewise. * nis/nis_file.c: Likewise. * nis/nis_free.c: Likewise. * nis_local_names.c: Likewise. * nis/nis_lookup.c: Likewise. * nis/nis_modify.c: Likewise. * nis/nis_print.c: Likewise. * nis/nis_remove.c: Likewise. * nis/nis_table.c: Likewise. * nis/nis_util.c: Likewise. * nis/nis_xdr.c: Likewise. * nis/yp_xdr.c: Likewise. * nis/ypclnt.c: Likewise. * nis/ypupdate_xdr.c: Likewise. * nis/nis_checkpoint.c: Add libnsl_hidden_nolink_def to all functions. * nis/nis_clone_dir.c: Likewise. * nis/nis_clone_res.c: Likewise. * nis/nis_creategroup.c: Likewise. * nis/nis_destroygroup.c: Likewise. * nis/nis_domain_of.c: Likewise. * nis/nis_getservlist.c: Likewise. * nis/nis_ismember.c: Likewise. * nis/nis_mkdir.c: Likewise. * nis/nis_ping.c: Likewise. * nis/nis_print_group_entry.c: Likewise. * nis/nis_removemember.c: Likewise. * nis/nis_rmdir.c: Likewise. * nis/nis_server.c: Likewise. * nis/nis_subr.c: Likewise. * nis/nis_verifygroup.c: Likewise. Signed-off-by: Thorsten Kukuk <kukuk@suse.com>
* conformtest: Add x32 XFAILs for mq_attr element types (bug 21279).Joseph Myers2017-03-203-5/+15
| | | | | | | | | | | | | | | | | | POSIX specifies long as the type of elements of struct mq_attr. For x32, they are __syscall_slong_t (i.e. long long). This patch XFAILs the corresponding tests for x32 in the conformtest expectations (the bug should not be closed without an actual fix). Tested with build-many-glibcs.py. [BZ #21279] * sysdeps/unix/sysv/linux/x86_64/x32/Makefile [$(subdir) = conform] (conformtest-xfail-conds): Update comment. * conform/data/mqueue.h-data (mq_attr.mq_flags): XFAIL for x86_64-x32-linux. (mq_attr.mq_maxmsg): Likewise. (mq_attr.mq_msgsize): Likewise. (mq_attr.mq_curmsgs): Likewise.
* conformtest: Add mips XFAIL for struct stat st_rdev type (bug 21278).Joseph Myers2017-03-203-2/+9
| | | | | | | | | | | | | | MIPS o32 struct stat has the wrong type of st_rdev. This patch XFAILs that test in the conformtest expectations for this case (the bug should not be closed without an actual fix, however). Tested with build-many-glibcs.py. [BZ #21278] * sysdeps/unix/sysv/linux/mips/mips32/Makefile [$(subdir) = conform] (conformtest-xfail-conds): Update comment. * conform/data/sys/stat.h-data (stat.st_rdev): XFAIL for mips-o32-linux.
* Fix alpha termios.h NL2, NL3 namespace (bug 21268).Joseph Myers2017-03-202-2/+11
| | | | | | | | | | | | | | | sysdeps/unix/sysv/linux/alpha/bits/termios.h defines NL2 and NL3 for __USE_MISC || __USE_XOPEN. These should only be defined for __USE_MISC as they are not part of any standard namespace. This patch conditions them accordingly, matching the powerpc version of the header (the only other one in glibc that defines these macros). Tested (compilation only) with build-many-glibcs.py. [BZ #21268] * sysdeps/unix/sysv/linux/alpha/bits/termios.h (NL2): Define only if [__USE_MISC] (NL3): Likewise.
* posix_spawn: fix stack setup on ia64 [BZ #21275]Mike Frysinger2017-03-202-5/+14
| | | | | | | | | The ia64-specific clone2 call expects the base of the stack mapping and the stack size as sep arguments, not an initial stack value as on other stack-grows-down architectures. Reuse the stack-grows-up macro so we pass in the right stack base. Reported-by: Matt Turner <mattst88@gentoo.org>
* linux ttyname and ttyname_r: do not return wrong resultsChristian Brauner2017-03-194-8/+71
| | | | | | | | | | | | | If a link (say /proc/self/fd/0) pointing to a device, say /dev/pts/2, in a parent mount namespace is passed to ttyname, and a /dev/pts/2 exists (in a different devpts) in the current namespace, then it returns /dev/pts/2. But /dev/pts/2 is NOT the current tty, it is a different file and device. Detect this case and return ENODEV. Userspace can choose to take this as a hint that the fd points to a tty device but to act on the fd rather than the link. Signed-off-by: Serge Hallyn <serge@hallyn.com> Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>
* conformtest: XFAIL tv_nsec tests for x32 (bug 16437).Joseph Myers2017-03-186-4/+22
| | | | | | | | | | | | | | | | | This patch XFAILs the conformtest tv_nsec tests for x32 so that the incorrect type does not potentially hide other failures. As this is not a fix for the bug, it should remain open in Bugzilla. Tested (compilation only) with build-many-glibcs.py. [BZ #16437] * sysdeps/unix/sysv/linux/x86_64/x32/Makefile [$(subdir) = conform] (conformtest-xfail-conds): New variable. * conform/data/signal.h-data (timespec.tv_nsec): XFAIL for x86_64-x32-linux. * conform/data/sys/select.h-data (timespec.tv_nsec): Likewise. * conform/data/sys/stat.h-data (timespec.tv_nsec): Likewise. * conform/data/time.h-data (timespec.tv_nsec): Likewise.
* Fix sparc64 bits/setjmp.h namespace (bug 21261).Joseph Myers2017-03-183-21/+30
| | | | | | | | | | | | | | | | | | sysdeps/unix/sysv/linux/sparc/bits/setjmp.h defines 64-bit __jmp_buf with a load of identifiers that are not part of any standard namespace, resulting in conform/ tests failing. This patch fixes this by moving those identifiers to the implementation namespace, so enabling the conform/ tests to pass for sparc64. Tested (compilation only) for sparc64 with build-many-glibcs.py. [BZ #21261] * sysdeps/unix/sysv/linux/sparc/bits/setjmp.h [__WORDSIZE == 64 && !_ASM] (__sparc64_jmp_buf): Use reserved names for all fields. * sysdeps/sparc/sparc64/jmpbuf-unwind.h (_JMPBUF_UNWINDS): Update for jmp_buf field renaming. (_JMPBUF_UNWINDS_ADJ): Likewise.
* conformtest: Handle conditional XFAILs with allow-header.Joseph Myers2017-03-182-1/+6
| | | | | | | | | | | This patch fixes the conformtest handling of headers listed in allow-header to process xfail[cond]- in the expectations for those headers. Tested with build-many-glibcs.py. * conform/conformtest.pl: Handle xfail[cond]- in header mentioned with allow-header.
* Further harden glibc malloc metadata against 1-byte overflows.DJ Delorie2017-03-172-0/+7
| | | | | | | | | Additional check for chunk_size == next->prev->chunk_size in unlink() 2017-03-17 Chris Evans <scarybeasts@gmail.com> * malloc/malloc.c (unlink): Add consistency check between size and next->prev->size, to further harden against 1-byte overflows.
* Use CPU_FEATURES_CPU_P to check if AVX is availableH.J. Lu2017-03-172-2/+6
| | | | | | | Don't use bit_cpu_AVX directly. * sysdeps/x86/cpu-features.c (init_cpu_features): Check AVX with CPU_FEATURES_CPU_P.
* Make alpha termios.h define IXANY unconditionally (bug 21259).Joseph Myers2017-03-172-1/+5
| | | | | | | | | | | | | | | | | sysdeps/unix/sysv/linux/alpha/bits/termios.h defines IXANY only if __USE_MISC. But it's in the base standard for POSIX.1:2008, and XSI-shaded in previous standards. This patch makes the header define it unconditionally, like other versions of this header do (it's always reserved by standards that don't require it, so defining unconditionally is OK by the standards). Tested (compilation only) for alpha with build-many-glibcs.py. Note that there are still termios.h conformtest failures after this patch because of other issues with the alpha version of this header. [BZ #21259] * sysdeps/unix/sysv/linux/alpha/bits/termios.h (IXANY): Define unconditionally, not just for [__USE_MISC].
* conformtest: Add mips XFAIL for struct stat st_dev type (bug 17786).Joseph Myers2017-03-173-1/+11
| | | | | | | | | | | | | | | | As noted in bug 17786, MIPS o32 struct stat has the wrong type of st_dev. This patch XFAILs that test in the conformtest expectations for this case (the test still fails after the patch because there's also a similar issue for st_rdev that needs reporting and XFAILing separately, and the bug should not be closed without an actual fix, not just XFAILing). Tested for mips with build-many-glibcs.py. [BZ #17786] * sysdeps/unix/sysv/linux/mips/mips32/Makefile: New file. * conform/data/sys/stat.h-data (stat.st_dev): XFAIL for mips-o32-linux.
* Add missing piece to last ChangeLog entry.Joseph Myers2017-03-171-0/+1
|
* conformtest: Add alpha XFAIL for struct netent n_net type (bug 21260).Joseph Myers2017-03-173-1/+11
| | | | | | | | | | | | | | As noted in bug 21260, sysdeps/unix/sysv/linux/alpha/bits/netdb.h defines struct netent with n_net of type unsigned long instead of the correct uint32_t. This patch XFAILs that test in the conformtest expectations for alpha. (This is not a fix for the bug, and it should not be closed without an actual fix.) Tested for alpha with build-many-glibcs.py. [BZ #21260] * sysdeps/unix/sysv/linux/alpha/Makefile [$(subdir) = conform] (conformtest-xfail-conds): New variable.
* conformtest: Skip execution tests when cross-compiling.Joseph Myers2017-03-173-2/+18
| | | | | | | | | | | | | | | | | | | | This patch makes conformtest skip execution tests when cross-compiling, as an interim step towards running most of these tests (presently disabled) in that case. It omits the (obvious) Makefile change to actually enable the conformtest tests when cross-compiling, as there are still enough failures seen with build-many-glibcs.py that I'd like to get the results cleaner before enabling these tests. Tested for x86_64, and with the tests actually enabled for cross-compilation with build-many-glibcs.py. * conform/conformtest.pl ($cross): New variable. (--cross): New command-line option. (runtest): Skip test execution when cross-compiling. * conform/Makefile (conformtest-cross): New variable. ($(conformtest-header-tests)): Pass $(conformtest-cross) to conformtest.pl.
* conformtest: Support system-specific XFAILs.Joseph Myers2017-03-173-2/+27
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | conformtest has an internal XFAIL mechanism to allow failures of individual expectations to be ignored, so that known hard-to-fix failures (e.g. those affecting ABIs or requiring kernel changes) do not cause the overall tests to FAIL and so hide other failures from the same (header, standard) pair. Various such bugs are system-specific, so this patch adds a mechanism to allow system-specific XFAILs. A system-independent XFAIL is achieved by putting "xfail-" at the start of the relevant expectation in the *-data files. A system-specific XFAIL instead uses "xfail[cond]-", where "cond" is a condition listed in conformtest-xfail-conds in a sysdeps makefile (so one for x32 might set conformtest-xfail-conds = x86_64-x32-linux, for example, and then an expectation for tv_nsec's type could use xfail[x86_64-x32-linux]-). The actual names are arbitrary, just needing to match between the makefiles and the expectations, and if necessary you can use "xfail[cond1|cond2]-" for a test that is expected to fail under multiple conditions. As with system-independent XFAILs, I think system-specific ones should have a bug filed in Bugzilla and a comment referencing that bug. Tested for x86_64, including with test expectations and makefiles changed to use the new facility. * conform/conformtest.pl ($xfail_str): New variable. (--xfail=): New command-line option. (top level): Handle expectations starting xfail[cond]-. * conform/Makefile (conformtest-xfail): New variable. ($(conformtest-header-tests)): Pass $(conformtest-xfail) to conformtest.pl.
* conformtest: Make more tests into compilation tests.Joseph Myers2017-03-162-8/+12
| | | | | | | | | | | | | | | | | | | | | | | | | | The conformtest header tests test some things through compilation tests and others through execution tests. This patch makes more of the tests into compilation tests, using _Static_assert (note: for float.h tests on floating-point values this is depending on a GNU extension, that those assertions are allowed in the absence of -pedantic although they aren't strictly integer constant expressions). The remaining execution tests are for values of things listed as "symbol" (in fact no such things have a value expectation listed) and for values of macros defined as string constants (three such values listed in total). This is intended as preparation for enabling the vast bulk of the tests to run for cross compilation. (Even the few remaining execution tests ought in principle to run for cross compilation when a test wrapper is defined, but that's more complicated. The existing execution tests for native builds in fact are linked and run with an existing installed libc that's required to exist to link against, rather than with the newly built libc; only the new headers are used.) Tested for x86_64. * conform/conformtest.pl: Use compilation instead of execution tests for testing values of constants and usability in #if.
* XFAIL catan and catanh tests on ibm128Tulio Magno Quites Machado Filho2017-03-164-1168/+1175
| | | | | | | | | | | | | | | Mark 4 catan and catanh tests as xfail-rounding:ibm128-libgcc. After this patch all catan and catanh tests pass on ibm128. Regenerated auto-libm-test-out using gmp 6.1.2, mpfr 3.1.5 and upstream mpc (9ef8030e50), Tested on powerpc, powerpc64 and powerpc64le. * math/auto-libm-test-in: Mark some catan and catanh as xfail-rounding:ibm128-libgcc. * math/auto-libm-test-out-catan: Regenerate. * math/auto-libm-test-out-catanh: Likewise.
* Remove C++ namespace handling from glibc headers.Joseph Myers2017-03-1625-411/+37
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | glibc headers include some code (not particularly consistent or systematic) to put various declarations in C++ namespaces std and __c99, if _GLIBCPP_USE_NAMESPACES is defined. As noted in <https://gcc.gnu.org/ml/libstdc++/2017-03/msg00025.html>, this macro was removed from libstdc++ in 2000. I don't expect compilation with such old versions of libstdc++ to work with current glibc headers anyway (whereas old *binaries* are expected to stay working with current glibc); this patch (which should be a no-op with any libstdc++ version postdating that removal) removes all this code from the glibc headers. The begin-end-check.pl test, whose comments say it is about checking these namespace macro calls, is also removed. The code in that test would have covered __BEGIN_DECLS / __END_DECLS as well, but if those weren't properly matched it would show up with the check-installed-headers-cxx tests, so I don't think there is an actual use for keeping begin-end-check.pl with the namespace code removed. Tested for x86_64 and x86 (testsuite, and that installed stripped shared libraries are unchanged by the patch). * misc/sys/cdefs.h (__BEGIN_NAMESPACE_STD): Remove macro. (__END_NAMESPACE_STD): Likewise. (__USING_NAMESPACE_STD): Likewise. (__BEGIN_NAMESPACE_C99): Likewise. (__END_NAMESPACE_C99): Likewise. (__USING_NAMESPACE_C99): Likewise. * math/math.h (_Mdouble_BEGIN_NAMESPACE): Do not define and undefine macro. (_Mdouble_END_NAMESPACE): Likewise. * ctype/ctype.h: Do not handle C++ namespaces. * libio/bits/stdio-ldbl.h: Likewise. * libio/stdio.h: Likewise. * locale/locale.h: Likewise. * math/bits/mathcalls.h: Likewise. * setjmp/setjmp.h: Likewise. * signal/signal.h: Likewise. * stdlib/bits/stdlib-float.h: Likewise. * stdlib/bits/stdlib-ldbl.h: Likewise. * stdlib/stdlib.h: Likewise. * string/string.h: Likewise. * sysdeps/x86/fpu/bits/mathinline.h: Likewise. * time/bits/types/clock_t.h: Likewise. * time/bits/types/struct_tm.h: Likewise. * time/bits/types/time_t.h: Likewise. * time/time.h: Likewise. * wcsmbs/bits/wchar-ldbl.h: Likewise. * wcsmbs/uchar.h: Likewise. * wcsmbs/wchar.h: Likewise. [_GLIBCPP_USE_NAMESPACES] (wint_t): Remove conditional definition. * wctype/wctype.h: Do not handle C++ namespaces. * scripts/begin-end-check.pl: Remove. * Makefile (installed-headers): Likewise. (tests-special): Do not add $(objpfx)begin-end-check.out. ($(objpfx)begin-end-check.out): Remove.
* Add ChangeLog entry for aarch64 ifunc support patch.Steve Ellcey2017-03-151-0/+12
| | | | Forgot to add ChangeLog entry.
* Add ifunc support for aarch64.Steve Ellcey2017-03-157-0/+205
| | | | | | | | | | | * sysdeps/aarch64/dl-machine.h: Include cpu-features.c. (DL_PLATFORM_INIT): New define. (dl_platform_init): New function. * sysdeps/aarch64/ldsodefs.h: Include cpu-features.h. * sysdeps/unix/sysv/linux/aarch64/cpu-features.c: New file. * sysdeps/unix/sysv/linux/aarch64/cpu-features.h: Likewise. * sysdeps/unix/sysv/linux/aarch64/dl-procinfo.c: Likewise. * sysdeps/unix/sysv/linux/aarch64/libc-start.c: Likewise.
* x86_64: fix static build of __mempcpy_chk for compilers defaulting to PIC/PIEMike Frysinger2017-03-152-1/+6
| | | | | | | | | | | | When glibc is compiled with gcc 6.2 that has been configured with to default to PIC/PIE, the static version of __mempcpy_chk is not built, as the test is done on PIC instead of SHARED. Fix the test to check for SHARED, like it is done for similar functions like __memcpy_chk. 2017-03-12 Mike Frysinger <vapier@gentoo.org> * sysdeps/x86_64/mempcpy_chk.S (__mempcpy_chk): Check for SHARED instead of PIC.
* Improve float range reduction accuracy near pi/2 (bug 21094).Joseph Myers2017-03-155-2/+595
| | | | | | | | | | | | | | | | | | | | | | | | | Bug 21094 reports 3ulp errors of cosf and tanf for certain arguments near pi/2 arising from the use of an insufficiently accurate range reduction. (To be clear, this is a quality-of-implementation issue relating to the apparent intent of those particular cosf and tanf implementations; 3ulp is within the general glibc accuracy goals, so not inherently a bug.) This patch fixes that error by making a wider range of cases use the existing more accurate range reduction for arguments close to pi/2. The wider range of values is still narrow enough for the "z -= pio2_2;" in the more accurate case to be exact, as the code expects. Tested for x86_64, x86 and mips64; no ulps updates needed (but at least on mips64, the larger ulps were seen if the tests were added without the substantive fix). [BZ #21094] * sysdeps/ieee754/flt-32/e_rem_pio2f.c (__ieee754_rem_pio2f): Use 24+24+24-bit pi for wider range of values around pi/2. * math/auto-libm-test-in: Add more tests of cos and tan. * math/auto-libm-test-out-cos: Regenerated. * math/auto-libm-test-out-tan: Likewise.
* hppa: Fix setting of __libc_stack_endJohn David Anglin2017-03-152-5/+9
| | | | | | | | | | | | | | The binutils package was recently changed to fix -z relro support on hppa. See ld/21000 for details: https://sourceware.org/bugzilla/show_bug.cgi?id=21000 This exposed a problem with the _dl_start_user function in the RTLD_START define. We need to set __libc_stack_end before it is made read only. For this, we need to define DL_STACK_END. The offset of 0x160 gives the same stack end as the code in _dl_start_user. A build log with the attached patch is here: https://buildd.debian.org/status/fetch.php?pkg=glibc&arch=hppa&ver=2.24-9&stamp=1487639205&raw=0
* Fix build with --enable-static-nss [BZ #21088]Alexey Neyman2017-03-152-1/+7
| | | | Signed-off-by: Alexey Neyman <stilor@att.net>
* Fix combreloc test with BSD grepAlexey Neyman2017-03-153-9/+14
| | | | | | | | | | | | | | The test for "-z combreloc" fails when cross-compiling on a machine that uses BSD grep (e.g. on macos). grep complains about empty subexpression and exits with non-zero status, which is interpreted by configure as "not found". As a result, support for "-z combreloc" (HAVE_Z_COMBRELOC) is not detected, leading to link failure on SPARC. While there, replace fgrep with 'grep -F', as fgrep is non-POSIX. * configure.ac: Avoid empty subexpression in grep. Signed-off-by: Alexey Neyman <stilor@att.net>
* Fix test-errno issuesAdhemerval Zanella2017-03-154-9/+53
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | This patch fixes multiple issues of test-errno.c (9a56f8718341): - Rename Linux test-errno.c to test-errno-linux.c to avoid build the same source for both tests. - Add a mlock check for 32 bits build running on 64 bits kernels. Althuough man pages states that mlock fails with EINVAL if final address overflows, kernels does not return it for aforementioned condition (it returns ENOMEM instead). Although it seems to be a kernel issue for compat syscall handling, I think it is worth to still check syscall return and document the behavior. - Initialize option lenght for setsockopt check. - Change open test from EINVAL to EISDIR. Checked on x86_64-linux-gnu and i686-linux-gnu (running on 64 bits kernel). * posix/test-errno.c (do_test): Initialize setsockopt optlen. * sysdeps/unix/sysv/linux/test-errno.c: Move to ... * sysdeps/unix/sysv/linux/test-errno-linux.c: ... here. (test_wrp_rv): Fix format. (test_wrp_rv2): New macro. (do_test): Handle mlock return on 64 bits kernels with 32 bits binaries.
* Fix test-math-vector-sincos.h aliasing.Joseph Myers2017-03-152-14/+23
| | | | | | | | | | | | | | | | | | | | | x86_64 libmvec tests have been failing to build lately with GCC mainline with -Wuninitialized errors, and Markus Trippelsdorf traced this to an aliasing issue <https://sourceware.org/ml/libc-alpha/2017-03/msg00169.html>. This patch fixes the aliasing issue, so that the vectors-of-pointers are initialized using a union instead of pointer casts. This also fixes the testsuite build failures with GCC mainline. Tested for x86_64 (full testsuite with GCC 6; testsuite build with GCC mainline with build-many-glibcs.py). * sysdeps/x86/fpu/test-math-vector-sincos.h (INIT_VEC_PTRS_LOOP): Use a union when storing pointers. (VECTOR_WRAPPER_fFF_2): Do not take address of integer vector and cast result when passing to INIT_VEC_PTRS_LOOP. (VECTOR_WRAPPER_fFF_3): Likewise. (VECTOR_WRAPPER_fFF_4): Likewise.
* Split helper classification macros from mathcalls.hGabriel F. T. Gomes2017-03-156-29/+59
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | The classification macros: finite, fpclassify, iseqsig, isinf, isnan, issignaling, and signbit are defined by ISO C11 and declared in mathcalls.h for each of the floating-point types: float, double, and long double. TS 18661-3 does not mention these macros for float128, however support for them must be present when _Float128 is present. This is true, even when the feature test macro __STDC_WANT_IEC_60559_TYPES_EXT__ is false. Other function declarations in mathcalls.h, on the other hand, depend on __STDC_WANT_IEC_60559_TYPES_EXT__. This patch splits the helper functions (__finite, __fpclassify, __iseqsig, __isinf, __isnan, __issignaling, and __signbit) from mathcalls.h, so that these helper functions can be declared for _Float128, even when __STDC_WANT_IEC_60559_TYPES_EXT__ is false. Tested for powerpc64le, s390x, and x86_64. * include/bits/mathcalls-helper-functions.h: New file. * math/Makefile (headers): Add bits/mathcalls-helper-functions.h. * math/bits/mathcalls.h (__finite, __fpclassify, __iseqsig) (__isinf, __isnan, __issignaling, __signbit): Move declarations to math/bits/mathcalls-helper-functions.h. * math/bits/mathcalls-helper-functions.h: New file. * math/math.h: Include bits/mathcalls-helper-functions.h for float, double, and long double.
* Use internal __feraiseexcept in __iseqsigGabriel F. T. Gomes2017-03-152-1/+6
| | | | | | | | | | Replace the use of feraiseexcept with __feraiseexcept in the helper function __iseqsig (math/s_iseqsig_template.c). Tested for powerpc64le, s390x, and x86_64. * math/s_iseqsig_template.c (__iseqsig): Use __feraiseexcept instead of feraiseexcept.
* support: Explain ignored failures of temporary file removal [BZ #21243]Florian Weimer2017-03-152-1/+10
|
* support: Add error checking to close system calls [BZ #21244]Florian Weimer2017-03-157-15/+63
|
* support_format_dns_packet: Fix CNAME and multiple RR handlingFlorian Weimer2017-03-154-7/+123
| | | | | | | | | | Before this change, the loop iterating over RRs in the answer section stopped at the first CNAME record, never printing them. The CNAME and PTR record contents was extracted from the wrong buffer (whole packet instead RDATA). This desynced the parsing after the first CNAME or PTR record. Also fix the AAAA record parsing by checking their sizes.
* Fix missing posix_fadvise64 mips64 static build (BZ #21232)Adhemerval Zanella2017-03-142-0/+8
| | | | | | | | | | | This patch fixes the missing posix_fadvise64 symbol for static build required for _FILE_OFFSET_BITS=64 on mips64 build. Checked on a mips64-linux-gnu build with run-built-tests=no. [BZ #21232] * sysdeps/unix/sysv/linux/mips/mips64/n64/posix_fadvise64.c: Add posix_fadvise64 weak_alias for static build.
* Remove _dl_platform_stringAndreas Schwab2017-03-147-36/+13
| | | | There are no non-trivial uses of _dl_platform_string.
* hurd: Make send/recv more posixishSamuel Thibault2017-03-133-2/+16
| | | | | | | | Thanks David Michael for the suggestion. * sysdeps/mach/hurd/send.c (__send): Convert hurdish error code into posix error code. * sysdeps/mach/hurd/recv.c (__recv): Likewise.
* build-many-glibcs: Remove no_isolate from SH configAdhemerval Zanella2017-03-132-19/+11
| | | | | | | | | | | | | Now with d40dbe7 SH build does not require more the no_isolate gcc options to correct build glibc (since SH build now does not generate a trap anymore). This patch removes the unrequired options from SH config. Checked with a build for sh3-linux-gnu, sh3eb-linux-gnu, sh4-linux-gnu, and sh4eb-linux-gnu. * scripts/build-many-glibcs.py (Context.add_all_configs): Remove no_isolate usage for SH.
* Remove the str(n)dup inlines from string/bits/string2.h. Although inliningWilco Dijkstra2017-03-1315-98/+41
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | calls with constant strings shows a small (~10%) performance gain, strdup is typically used in error reporting code, so not performance critical. Remove the now unused __need_malloc_and_calloc related defines from stdlib.h. Rename existing uses of str(n)dup to __str(n)dup so it no longer needs to be redirected to a builtin. Also building GLIBC with -Os now no longer shows localplt or linkname space failures (partial fix for BZ #15105 and BZ #19463). [BZ #15105] [BZ #19463] * elf/dl-cache.c (_dl_load_cache_lookup): Use __strdup. * inet/rcmd.c (rcmd_af): Likewise. * inet/rexec.c (rexec_af): Likewise. * intl/dcigettext.c (_LIBC): Likewise. * intl/finddomain.c (_nl_find_domain): Use strdup expansion. * locale/loadarchive.c (_nl_load_locale_from_archive): Use __strdup. * locale/setlocale.c (setlocale): Likewise. * posix/spawn_faction_addopen.c (posix_spawn_file_actions_addopen): Likewise. * stdlib/putenv.c (putenv): Use __strndup. * sunrpc/svc_simple.c (__registerrpc): Use __strdup. * sysdeps/posix/getaddrinfo.c (gaih_inet): Use __strdup/__strndup. * include/stdlib.h (__need_malloc_and_calloc): Remove uses. (__Need_M_And_C) Remove define/undef. * stdlib/stdlib.h (__need_malloc_and_calloc): Remove uses. (__malloc_and_calloc_defined): Remove define. * string/bits/string2.h (__strdup): Remove define. (strdup): Likewise. (__strndup): Likewise. (strndup): Likewise.
* Add more IPV6_* macros to sysdeps/unix/sysv/linux/bits/in.h.Joseph Myers2017-03-132-0/+26
| | | | | | | | | | | | | | | | | | | | | Linux 4.10 adds IPV6_RECVFRAGSIZE to include/uapi/linux/in6.h, which shows that several such IPV6_* macros are missing from glibc's sysdeps/unix/sysv/linux/bits/in.h (while older ones are present). I don't know whether any of these might be deliberately omitted, but this patch adds what appear to be the missing more recent macros to glibc. Tested for x86_64. * sysdeps/unix/sysv/linux/bits/in.h (IPV6_AUTOFLOWLABEL): New macro. (IPV6_ADDR_PREFERENCES): Likewise. (IPV6_MINHOPCOUNT): Likewise. (IPV6_ORIGDSTADDR): Likewise. (IPV6_RECVORIGDSTADDR): Likewise. (IPV6_TRANSPARENT): Likewise. (IPV6_UNICAST_IF): Likewise. (IPV6_RECVFRAGSIZE): Likewise.
* Add missing ChangeLog entries.Thorsten Kukuk2017-03-131-0/+9
|
* The rpcgen tests should not run if we don't build rpcgen.Thorsten Kukuk2017-03-131-0/+2
| | | | * sunrpc/Makefile: only run rpcgen tests if we build rpcgen.
* sh: Fix building with gcc5/6Alexey Neyman2017-03-122-1/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Build glibc for sh4-unknown-linux-gnu currently fails if one's using GCC5/6: in dl-conflict.c, the elf_machine_rela() function is called with NULL as its 3rd argument, sym. The implementation of that function in sysdeps/sh/dl-machine.h dereferences that pointer: const Elf32_Sym *const refsym = sym; ... if (map == &GL(dl_rtld_map)) value -= map->l_addr + refsym->st_value + reloc->r_addend; GCC discovers a null pointer dereference, and in accordance with -fdelete-null-pointer-checks (which is enabled in -O2) replaces this code with a trap - which, as SH does not implement a trap pattern in GCC, evaluates to an abort() call. This abort() call pulls many more objects from libc_nonshared.a, eventually resulting in link failure due to multiple definitions for a number of symbols. As far as I see, the conditional before this code is always false in rtld: _dl_resolve_conflicts() is called with main_map as the first argument, not GL(_dl_rtld_map), but since that call is in yet another compilation unit, GCC does not know about it. Patch that wraps this conditional into !defined RESOLVE_CONFLICT_FIND_MAP attached. * sysdeps/sh/dl-machine.h (elf_machine_rela): The condition in R_SH_DIR32 case is always false when inlined from dl-conflict.c. Ifdef out to prevent GCC from insertin an abort() call.
* Fix send consolidation typoMarko Myllynen2017-03-122-1/+7
| | | | | | | | Fix 60f9423b type for alpha kernel-features.h definition. * sysdeps/unix/sysv/linux/alpha/kernel-features.h (__ASSUME_RECV_SYSCALL): Replace duplicate by __ASSUME_SEND_SYSCALL.
* If sunrpc code is disabled, rpcsvc header files, rpcgen andThorsten Kukuk2017-03-112-6/+15
| | | | | | | | | librpcsvc.a should not be installed, too. * sunrpc/Makefile: don't build and install rpcsvc header files, rpcgen and librpcsvc.a by default. Signed-off-by: Thorsten Kukuk <kukuk@suse.com>
* Update auto-libm-test-out for catan / catanh.Stefan Liebler2017-03-104-912/+918
| | | | | | | | | | | | | | I've used gmp 6.1.2, mpfr 3.1.5 and upstream mpc with fix in mpc_atan (https://scm.gforge.inria.fr/anonscm/gitweb?p=mpc/mpc.git;a=commit;h=958aac9b15a659d6fb5edcb11778123f8a35b14f) to build gen-auto-libm-tests and regenerated catan / catanh out files. Regenerated ULPs for s390 from scratch. Now the catan / catanh tests are passing. ChangeLog: * math/auto-libm-test-out-catan: Regenerated. * math/auto-libm-test-out-catanh: Likewise. * sysdeps/s390/fpu/libm-test-ulps: Likewise.
generated by cgit-pink 1.4.1 (git 2.36.1) at 2024-08-05 15:13:06 +0000