summary refs log tree commit diff
Commit message (Collapse)AuthorAgeFilesLines
* x86: Set dl_platform and dl_hwcap from CPU features [BZ #21391]H.J. Lu2017-05-0312-138/+273
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | dl_platform and dl_hwcap are set from AT_PLATFORM and AT_HWCAP very early during startup. They are used by dynamic linker to determine platform and build an array of hardware capability names, which are added to search path when loading shared object. dl_platform and dl_hwcap are unused on x86-64. On i386, i386, i486, i586 and i686 platforms were supported and only SSE2 capability was used. On x86, usage of AT_PLATFORM and AT_HWCAP to determine platform and processor capabilities is obsolete since all information is available in dl_x86_cpu_features. This patch sets dl_platform and dl_hwcap from dl_x86_cpu_features in dynamic linker. On i386, the available plaforms are changed to i586 and i686 since i386 has been deprecated. On x86-64, the available plaforms are haswell, which is for Haswell class processors with BMI1, BMI2, LZCNT, MOVBE, POPCNT, AVX2 and FMA, and xeon_phi, which is for Xeon Phi class processors with AVX512F, AVX512CD, AVX512ER and AVX512PF. A capability, avx512_1, is also added to x86-64 for AVX512 ISAs: AVX512F, AVX512CD, AVX512BW, AVX512DQ and AVX512VL. [BZ #21391] * sysdeps/i386/dl-machine.h (dl_platform_init) [IS_IN (rtld)]: Only call init_cpu_features. [!IS_IN (rtld)]: Only set GLRO(dl_platform) to NULL if needed. * sysdeps/x86_64/dl-machine.h (dl_platform_init): Likewise. * sysdeps/i386/dl-procinfo.h: Removed. * sysdeps/unix/sysv/linux/i386/dl-procinfo.h: Don't include <sysdeps/i386/dl-procinfo.h> nor <ldsodefs.h>. Include <sysdeps/x86/dl-procinfo.h>. (_dl_procinfo): Replace _DL_HWCAP_COUNT with 32. * sysdeps/unix/sysv/linux/x86_64/dl-procinfo.h [!IS_IN (ldconfig)]: Include <sysdeps/x86/dl-procinfo.h> instead of <sysdeps/generic/dl-procinfo.h>. * sysdeps/x86/cpu-features.c: Include <dl-hwcap.h>. (init_cpu_features): Set dl_platform, dl_hwcap and dl_hwcap_mask. * sysdeps/x86/cpu-features.h (bit_cpu_LZCNT): New. (bit_cpu_MOVBE): Likewise. (bit_cpu_BMI1): Likewise. (bit_cpu_BMI2): Likewise. (index_cpu_BMI1): Likewise. (index_cpu_BMI2): Likewise. (index_cpu_LZCNT): Likewise. (index_cpu_MOVBE): Likewise. (index_cpu_POPCNT): Likewise. (reg_BMI1): Likewise. (reg_BMI2): Likewise. (reg_LZCNT): Likewise. (reg_MOVBE): Likewise. (reg_POPCNT): Likewise. * sysdeps/x86/dl-hwcap.h: New file. * sysdeps/x86/dl-procinfo.h: Likewise. * sysdeps/x86/dl-procinfo.c (_dl_x86_hwcap_flags): New. (_dl_x86_platforms): Likewise.
* Tweak realloc/MREMAP comment to be more accurate.DJ Delorie2017-05-032-3/+8
| | | | | | | | | MMap'd memory isn't shrunk without MREMAP, but IIRC this is intentional for performance reasons. Regardless, this patch tweaks the existing comment to be more accurate wrt the existing code. [BZ #21411] * malloc/malloc.c: Tweak realloc/MREMAP comment to be more accurate.
* Correct misplaced comments in struct ip_mreq_sourcePhil Blundell2017-05-032-2/+7
|
* Bug 20686: Add el_GR@euro support.Carlos O'Donell2017-05-032-0/+6
| | | | | | | | | Despite the fact that el_GR is ISO-8859-7:2003 which contains the euro symobl, it is not possible to know this apriori to selecting the el_GR locale. Therefore you don't know if el_GR can possibly have the 2003 ammendments which include the euro symbol. This is resolved by creating an el_GR@euro locale similar to all the other @euro locales for non-UTF8 charsets.
* Bug 20116: Clarify behaviour of PD->lock.Carlos O'Donell2017-05-032-3/+18
| | | | | Add comments to the concurrency notes to clarify the semaphore-like and mutex-like behaviours of PD->lock.
* Consolidate Linux epoll_wait syscallAdhemerval Zanella2017-05-035-5/+17
| | | | | | | | | | | | | | This patch consolidates the epoll_wait Linux syscall generation on sysdeps/unix/sysv/linux/epoll_wait.c. The implementation tries to use __NR_epoll_wait if defined, otherwise calls epoll_pwait. Checked on i686-linux-gnu, x86_64-linux-gnu, x86_64-linux-gnux32, arch64-linux-gnu, arm-linux-gnueabihf, and powerpc64le-linux-gnu. * sysdeps/unix/sysv/linux/epoll_wait.c: New file. * sysdeps/unix/sysv/linux/generic/epoll_wait.c: Remove file. * sysdeps/unix/sysv/linux/syscalls.list: Remove epoll_wait from auto-generation list.
* Consolidate Linux select implementationAdhemerval Zanella2017-05-037-11/+78
| | | | | | | | | | | | | | | | | | | | | | | | | | | This patch consolidates the select Linux syscall implementation on sysdeps/unix/sysv/linux/select.c. The changes are: 1. Remove select from auto-generation syscalls.list on the architecture that uses __NR_select. 2. Remove generic implementation add a default one that handle all current cases (with the expection of alpha) The new default implementation will either use __NR_select if available of fallback to __NR_pselect6 otherwise. 3. Add a alpha outlier implementation which requires old compatibility symbols. Checked on i686-linux-gnu, x86_64-linux-gnu, x86_64-linux-gnux32, arch64-linux-gnu, arm-linux-gnueabihf, and powerpc64le-linux-gnu. * sysdeps/unix/sysv/linux/alpha/Makefile (sysdep_routines): Add osf_select. * sysdeps/unix/sysv/linux/alpha/select.c: New file. * sysdeps/unix/sysv/linux/alpha/syscalls.list: Remove select and osf_select from auto-generation list. * sysdeps/unix/sysv/linux/syscalls.list: Likewise. * sysdeps/unix/sysv/linux/sparc/sparc64/syscalls.list: Likewise. * sysdeps/unix/sysv/linux/generic/select.c: Remove file. * sysdeps/unix/sysv/linux/select.c: New file.
* Consolidate Linux poll implementationAdhemerval Zanella2017-05-033-3/+13
| | | | | | | | | | | | | | | | This patch consolidates the poll Linux syscall implementation on sysdeps/unix/sysv/linux/poll.c. It basically removes poll from auto-generation list and add a default implementation that either call __NR_poll directly (if the kernel headers defines it) or ppoll adjusting the timeout argument (as the generic implementation). Checked on i686-linux-gnu, x86_64-linux-gnu, x86_64-linux-gnux32, aarch64-linux-gnu, arm-linux-gnueabihf, and powerpc64le-linux-gnu. * sysdeps/unix/sysv/linux/generic/poll.c: Remove file. * sysdeps/unix/sysv/linux/poll.c: New file. * sysdeps/unix/sysv/linux/syscalls.list: Remove poll from auto-generation list.
* Add HWCAP_ASIMDRDM from Linux 4.11 to AArch64 bits/hwcap.h.Joseph Myers2017-05-032-0/+6
| | | | | | | | | | This patch adds the HWCAP_ASIMDRDM macro from Linux 4.11 to the AArch64 bits/hwcap.h. Tested (compilation only) with build-many-glibcs.py. * sysdeps/unix/sysv/linux/aarch64/bits/hwcap.h (HWCAP_ASIMDRDM): New macro.
* sparc: handle R_SPARC_DISP64 and R_SPARC_REGISTER relocsVladimir Mezentsev2017-05-022-0/+12
| | | | | | | | | | | The Studio compiler generates relocation types which are not supported in glibc. Handle these relocs. Tested in sparc64-linux-gnu. No regressions. BZ #21179] * sysdeps/sparc/sparc64/dl-machine.h: Handle R_SPARC_DISP64 and R_SPARC_REGISTER relocations.
* posix: Fix internal p{read,write} plt usageAdhemerval Zanella2017-05-026-1/+17
| | | | | | | | | | | | | | | | | | | | | | | | This patch adds internal alias for __pread, __pread64, and __pwrite following the already in place one for __pwrite64. This is not used in any implementation but on microblaze on preadv/pwritev fallback (since it does not define __ASSUME_PREADV). In fact it was signaled by commit c35db50ff5 which update the expected localptl.data for the architecture based on resulted value. This patch updates the plt for microblaze now that p{read,write}{64} are correctly routed to use internal alias. Checked on x86_64-linux-gnu and a build for all supported architectures (no all variants although). * include/unistd.h (__pread): Add libc_hidden_proto. (__pread64): Likewise. (__pwrite): Likewise. * sysdeps/unix/sysv/linux/microblaze/localplt.data [libc.so] (__pread64): Remove. * sysdeps/unix/sysv/linux/pread.c (__pread64): Add libc_hidden_weak. * sysdeps/unix/sysv/linux/pread64.c (__pread64): Likewise. * sysdeps/unix/sysv/linux/pwrite.c (__pwrite): Likewise.
* powerpc: Update powerpc-fpu libm-test-ulpsTulio Magno Quites Machado Filho2017-05-022-16/+20
| | | | * sysdeps/powerpc/fpu/libm-test-ulps: Update.
* nptl: Using libsupport for tst-cancel4*Adhemerval Zanella2017-05-016-946/+250
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Checked on x86_64-linux-gnu. * nptl/tst-cancel4-common.c: Use libsupport. * nptl/tst-cancel4-common.h: Likewise. * nptl/tst-cancel4.c (tf_read): Likewise. (tf_readv): Likewise. (tf_write): Likewise. (tf_writev): Likewise. (tf_sleep): Likewise. (tf_usleep): Likewise. (tf_nanosleep): Likewise. (tf_select): Likewise. (tf_pselect): Likewise. (tf_poll): Likewise. (tf_ppoll): Likewise. (tf_wait): Likewise. (tf_waitpid): Likewise. (tf_waitid): Likewise. (tf_sigpause): Likewise. (tf_sigsuspend): Likewise. (tf_sigwait): Likewise. (tf_sigwaitinfo): Likewise. (tf_sigtimedwait): Likewise. (tf_pause): Likewise. (tf_accept): Likewise. (tf_send): Likewise. (tf_recv): Likewise. (tf_recvfrom): Likewise. (tf_recvmsg): Likewise. (tf_open): Likewise. (tf_close): Likewise. (tf_pread): Likewise. (tf_pwrite): Likewise. (tf_preadv): Likewise. (tf_pwritev): Likewise. (tf_fsync): Likewise. (tf_fdatasync): Likewise. (tf_msync): Likewise. (tf_sendto): Likewise. (tf_sendmsg): Likewise. (tf_creat): Likewise. (tf_connect): Likewise. (tf_tcdrain): Likewise. (tf_msgrcv): Likewise. (tf_msgsnd): Likewise. * nptl/tst-cancel4_1.c (tf_sendmmsg): Likewise. * nptl/tst-cancel4_2.c (tf_recvmmsg): Likewise.
* posix: Using libsupport for p{write,read}v testsAdhemerval Zanella2017-05-012-25/+30
| | | | | | | | Checked on x86_64-linux-gnu. * misc/tst-preadvwritev-common.c (do_prepare): Use libsupport expected arguments. (do_test): Adapt to use libsupport.
* Add TCP_FASTOPEN_CONNECT from Linux 4.11 to netinet/tcp.h.Joseph Myers2017-05-012-0/+3
| | | | | | | | | This patch adds the TCP_FASTOPEN_CONNECT macro from Linux 4.11 to sysdeps/gnu/netinet/tcp.h. Tested for x86-64. * sysdeps/gnu/netinet/tcp.h (TCP_FASTOPEN_CONNECT): New macro.
* Add PF_SMC, AF_SMC from Linux 4.11 to bits/socket.h.Joseph Myers2017-05-012-1/+7
| | | | | | | | | | | This patch adds the PF_SMC / AF_SMC macros from Linux 4.11 to sysdeps/unix/sysv/linux/bits/socket.h. Tested for x86_64. * sysdeps/unix/sysv/linux/bits/socket.h (PF_SMC): New macro. (PF_MAX): Set to 44. (AF_SMC): New macro.
* Default Linux kernel version in build-many-glibcs.py to 4.11.Joseph Myers2017-05-012-1/+6
| | | | | * scripts/build-many-glibcs.py (Context.checkout): Default Linux kernel version to 4.11.
* powerpc: Fix logbl on power7 [BZ# 21280]Tulio Magno Quites Machado Filho2017-04-282-6/+25
| | | | | | | | | | | | | | | | | 1. Fix the results for negative subnormals by ignoring the signal when normalizing the value. 2. Fix the output when the high part is a power of 2 and the low part is a nonzero number with opposite sign. This fix is based on commit 380bd0fd2418f8988217de950f8b8ff18af0cb2b. After applying this patch, logbl() tests pass cleanly on POWER >= 7. Tested on powerpc, powerpc64 and powerpc64le [BZ #21280] * sysdeps/powerpc/power7/fpu/s_logbl.c (__logbl): Ignore the signal of subnormals and adjust the exponent of power of 2 down when low part has opposite sign.
* powerpc64le: Create divergent sysdep directory for powerpc64le.Paul E. Murphy2017-04-2820-0/+43
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | float128 on powerpc64le requires the addition of the ieee754/float128 sysdep, whereas powerpc64 doesn't. This requires creating a bunch of submachine and cpu directories and Implies files which just point towards their powerpc64 equivalent. Tested on P7, P8, and generic powerpc64le targets with and without multiarch. * sysdeps/powerpc/powerpc64le/Implies: New file. * sysdeps/powerpc/powerpc64le/fpu/Implies: New file. * sysdeps/powerpc/powerpc64le/fpu/multiarch/Implies: New file. * sysdeps/powerpc/powerpc64le/multiarch/Implies: New file. * sysdeps/powerpc/powerpc64le/power7/Implies: New file. * sysdeps/powerpc/powerpc64le/power7/fpu/Implies: New file. * sysdeps/powerpc/powerpc64le/power7/fpu/multiarch/Implies: New file. * sysdeps/powerpc/powerpc64le/power7/multiarch/Implies: New file. * sysdeps/powerpc/powerpc64le/power8/Implies: New file. * sysdeps/powerpc/powerpc64le/power8/fpu/Implies: New file. * sysdeps/powerpc/powerpc64le/power8/fpu/multiarch/Implies: New file. * sysdeps/powerpc/powerpc64le/power8/multiarch/Implies: New file. * sysdeps/powerpc/powerpc64le/power9/Implies: New file. * sysdeps/powerpc/powerpc64le/power9/fpu/Implies: New file. * sysdeps/powerpc/powerpc64le/power9/fpu/multiarch/Implies: New file. * sysdeps/powerpc/powerpc64le/power9/multiarch/Implies: New file. * sysdeps/powerpc/preconfigure: New file. * sysdeps/unix/sysv/linux/powerpc/powerpc64le/Implies: New file. * sysdeps/unix/sysv/linux/powerpc/powerpc64le/fpu/Implies: New file.
* conformtest: Allow time.h inclusion from semaphore.h for XOPEN2K.Joseph Myers2017-04-283-2/+10
| | | | | | | | | | | | | | | | POSIX.1:2001 added sem_timedwait, but said nothing about defining struct timespec in semaphore.h. The 2008 edition added a corresponding permission to include <time.h> from <semaphore.h>. Since this is the normal POSIX approach in such cases, it seems appropriate to consider this a bug fix, and so this patch makes the conform/ tests allow that inclusion for POSIX.1:2001 as well. Tested for x86_64. * conform/data/semaphore.h-data [XOPEN2K] (time.h): Allow header inclusion. * conform/Makefile (test-xfail-XOPEN2K/semaphore.h/conform): Remove.
* conformtest: Fix XPG standard naming.Joseph Myers2017-04-2764-224/+291
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The conform/ tests test -D_XOPEN_SOURCE under the name "XPG3", and -D_XOPEN_SOURCE -D_XOPEN_SOURCE_EXTENDED under the name "XPG4". That naming is misleading. _XOPEN_SOURCE_EXTENDED actually means XPG4.2, including UX-shaded interfaces. _XOPEN_SOURCE actually means XPG4, or XPG4.2 without UX-shaded interfaces. (Prior XPG versions also used _XOPEN_SOURCE, but without any versioning of the values of the macro, so XPG4.2 without UX-shaded interfaces is the only sensible set of interfaces for glibc to expose given _XOPEN_SOURCE=1 without _XOPEN_SOURCE_EXTENDED defined.) This patch fixes the naming in the conform/ tests, so that what is now called XPG4 is changed to XPG42, and what is now called XPG3 is changed to XPG4. Tested for x86_64 (and verified the complete set of expectations is unchanged by the patch beyond the intended renaming). * conform/GlibcConform.pm (XPG4): Rename standard to XPG42. (XPG3): Rename standard to XPG4. * conform/Makefile: Likewise. * conform/list-header-symbols.pl: Likewise. * conform/data/aio.h-data: Likewise. * conform/data/arpa/inet.h-data: Likewise. * conform/data/complex.h-data: Likewise. * conform/data/ctype.h-data: Likewise. * conform/data/dlfcn.h-data: Likewise. * conform/data/fcntl.h-data: Likewise. * conform/data/fenv.h-data: Likewise. * conform/data/float.h-data: Likewise. * conform/data/fmtmsg.h-data: Likewise. * conform/data/ftw.h-data: Likewise. * conform/data/grp.h-data: Likewise. * conform/data/inttypes.h-data: Likewise. * conform/data/iso646.h-data: Likewise. * conform/data/langinfo.h-data: Likewise. * conform/data/libgen.h-data: Likewise. * conform/data/limits.h-data: Likewise. * conform/data/locale.h-data: Likewise. * conform/data/math.h-data: Likewise. * conform/data/mqueue.h-data: Likewise. * conform/data/ndbm.h-data: Likewise. * conform/data/net/if.h-data: Likewise. * conform/data/netdb.h-data: Likewise. * conform/data/netinet/in.h-data: Likewise. * conform/data/poll.h-data: Likewise. * conform/data/pthread.h-data: Likewise. * conform/data/pwd.h-data: Likewise. * conform/data/sched.h-data: Likewise. * conform/data/search.h-data: Likewise. * conform/data/semaphore.h-data: Likewise. * conform/data/signal.h-data: Likewise. * conform/data/spawn.h-data: Likewise. * conform/data/stdbool.h-data: Likewise. * conform/data/stdint.h-data: Likewise. * conform/data/stdio.h-data: Likewise. * conform/data/stdlib.h-data: Likewise. * conform/data/string.h-data: Likewise. * conform/data/strings.h-data: Likewise. * conform/data/stropts.h-data: Likewise. * conform/data/sys/mman.h-data: Likewise. * conform/data/sys/resource.h-data: Likewise. * conform/data/sys/select.h-data: Likewise. * conform/data/sys/socket.h-data: Likewise. * conform/data/sys/stat.h-data: Likewise. * conform/data/sys/statvfs.h-data: Likewise. * conform/data/sys/time.h-data: Likewise. * conform/data/sys/timeb.h-data: Likewise. * conform/data/sys/types.h-data: Likewise. * conform/data/sys/uio.h-data: Likewise. * conform/data/sys/un.h-data: Likewise. * conform/data/sys/wait.h-data: Likewise. * conform/data/syslog.h-data: Likewise. * conform/data/termios.h-data: Likewise. * conform/data/tgmath.h-data: Likewise. * conform/data/time.h-data: Likewise. * conform/data/ucontext.h-data: Likewise. * conform/data/unistd.h-data: Likewise. * conform/data/utmpx.h-data: Likewise. * conform/data/varargs.h-data: Likewise. * conform/data/wchar.h-data: Likewise. * conform/data/wctype.h-data: Likewise.
* Default build-many-glibcs.py to GCC 7 branch.Joseph Myers2017-04-262-1/+6
| | | | | | | | | | | | | Now there is a GCC 7 release branch, this patch makes build-many-glibcs.py default to using it instead of GCC 6. Relative to GCC 6, this fixes the MicroBlaze build but introduces ICEs building glibc testcases for SH (GCC PRs 78459, 78460; the latter is an out-of-memory issue so you may want to set memory limits when running build-many-glibcs.py). * scripts/build-many-glibcs.py (Context.checkout): Default gcc version to 7 branch.
* Add earlyclobber to sqrtt/sqrtf insns.Uros Bizjak2017-04-262-4/+10
| | | | | | | | | | | | When using software completions, we have to prevent assembler to match input and output operands of sqrtt/sqrtf insn. Add earlyclobber to output operand to avoid unwanted operand matching. 2017-04-14 Uros Bizjak <ubizjak@gmail.com> * sysdeps/alpha/fpu/math_private.h (__ieee754_sqrt): Add earlyclobber to output operand of sqrt insn. (__ieee754_sqrtf): Ditto.
* Fix sys/socket.h namespace issues from sys/uio.h inclusion (bug 21426).Joseph Myers2017-04-2511-21/+58
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | sys/socket.h includes sys/uio.h to get the definition of the iovec structure. POSIX allows sys/socket.h to make all sys/uio.h symbols visible. However, all of sys/uio.h is XSI-shaded, so for non-XSI POSIX this results in conformtest failures (for sys/socket.h and other headers that include it): Namespace violation: "UIO_MAXIOV" Namespace violation: "readv" Namespace violation: "writev" Now, there is some ambiguity in POSIX about what namespace reservations apply in this case - see http://austingroupbugs.net/view.php?id=1127 - but glibc convention would still avoid declaring readv and writev, for example, for feature test macros that don't include them (if only headers from the relevant standard are included), even if such declarations are permitted, so there is a bug here according to glibc conventions. This patch moves the struct iovec definition to a new bits/types/struct_iovec.h header and includes that from sys/socket.h instead of including the whole of sys/uio.h. This fixes the namespace issue; however, three files in glibc that were relying on the implicit inclusion needed to be updated to include sys/uio.h explicitly. So there is a question of whether sys/socket.h should continue to include sys/uio.h under some conditions, such as __USE_XOPEN or __USE_MISC or __USE_XOPEN || __USE_MISC, for greater compatibility with code that (wrongly) expects this optional inclusion to be present there. (I think the three affected files in glibc should still have explicit sys/uio.h inclusions added in any case, however.) Tested for x86_64. [BZ #21426] * misc/bits/types/struct_iovec.h: New file. * misc/Makefile (headers): Add bits/types/struct_iovec.h. * include/bits/types/struct_iovec.h: New file. * bits/uio.h (struct iovec): Replace by inclusion of <bits/types/struct_iovec.h>. * sysdeps/unix/sysv/linux/bits/uio.h (struct iovec): Likewise. * socket/sys/socket.h: Include <bits/types/struct_iovec.h> instead of <sys/uio.h>. * nptl/tst-cancel4.c: Include <sys/uio.h> * posix/test-errno.c: Likewise. * support/resolv_test.c: Likewise. * conform/Makefile (test-xfail-POSIX2008/arpa/inet.h/conform): Remove. (test-xfail-POSIX2008/netdb.h/conform): Likewise. (test-xfail-POSIX2008/netinet/in.h/conform): Likewise. (test-xfail-POSIX2008/sys/socket.h/conform): Likewise.
* conformtest: Allow *_t in sys/socket.h.Joseph Myers2017-04-252-0/+5
| | | | | | | | | | | | | | | The conformtest expectations in sys/socket.h are missing the standard POSIX allowance for all headers to define *_t names. For XSI standard versions that allowance comes in via the permission to include <sys/uio.h> (for which the expectations properly allow *_t), but for non-XSI POSIX nothing brings in that allowance and spurious namespace failures occur. This patch adds the required permission to the expectations to remove the spurious failures (by itself this does not allow any XFAILs to be removed). Tested for x86_64. * conform/data/sys/socket.h-data (*_t): Allow.
* posix: Remove ununsed posix_spawn internal assignmentAdhemerval Zanella2017-04-252-16/+19
| | | | | | | | | | The internal 'ret' variable in '__spawni_child' function is not used after assignment in most cases. Checked on x86_64-linux-gnu. * sysdeps/unix/sysv/linux/spawni.c (__spawni_child): Remove ununsed assignment.
* [BZ 21340] add support for POSIX_SPAWN_SETSIDAdhemerval Zanella2017-04-259-2/+131
| | | | | | | | | | | | | | | | | | | | | | | | This patch adds support for the POSIX_SPAWN_SETSID flag. It was recently accepted by the Austin Group: http://austingroupbugs.net/view.php?id=1044 Checked on x86_64 Daurnimator <quae@daurnimator.com> Adhemerval Zanella <adhemerval.zanella@linaro.org> [BZ #21340] * posix/Makefile (tests): Add tst-posix_spawn-setsid to list of tests. * posix/spawn.h: define POSIX_SPAWN_SETSID flag. * posix/spawnattr_setflags.c (ALL_FLAGS): Add POSIX_SPAWN_SETSID to valid flags. * posix/tst-posix_spawn-setsid.c: Add test for POSIX_SPAWN_SETSID. * sysdeps/mach/hurd/spawni.c (__spawni): Implementation of POSIX_SPAWN_SETSID. * sysdeps/posix/spawni.c (__spawni): Likewise. * sysdeps/unix/sysv/linux/spawni.c (__spawni_child): Likewise. * NEWS: Add note about POSIX_SPAWN_SETSID support.
* Macroize function declarations in math_private.hGabriel F. T. Gomes2017-04-243-170/+154
| | | | | | | | | | | | | | | | | This patch moves the declaration of many floating-point functions from math_private.h to math_private_calls.h and macroize the declaration to be dependent on floating-point type. For each of float, double, and long double, the new header file is included once. This reduces the amount of repetitive boilerplate that will be required when adding float128 versions of these functions. Tested for powerpc64le and s390x. * sysdeps/generic/math_private.h: Move the declaration of many functions to sysdeps/generic/math_private_calls.h. * sysdeps/generic/math_private_calls.h: New file with the declarations of the functions removed from math_private.h macroized by floating-point type.
* S390: Move utf8-utf32-z9.c to multiarch folder and use s390_libc_ifunc_expr ↵Stefan Liebler2017-04-213-103/+169
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | macro. The utf8-utf32-z9.c iconv module is using ifunc and thus the ifunc part should be in multiarch folder. Otherwise ifunc is used even if you configure with --disable-multi-arch. This patch moves the ifunc resolvers to the new file sysdeps/s390/multiarch/utf8-utf32-z9.c. The resolvers are now implemented with s390_libc_ifunc_expr macro instead of using gcc attribute ifunc directly. The ifunc versions are implemented in sysdeps/s390/utf8-utf32-z9.c. Each version is only implemented if needed or supported. Therefore there is a block at beginning of the file which selects the versions which should be defined depending on support for multiarch, vector-support and used minimum architecture level. This block defines HAVE_[FROM|TO]_[C|CU|VX] to 1 or 0. The code below is rearranged and surrounded by #if HAVE_[FROM|TO]_[C|CU|VX] == 1. There is no functional change. The cu instructions are z9 zarch instructions. As the major distros are already using the newer z196 as architecture level set, those instructions can be used as fallback version instead of the c-code. This behaviour is decided at compile time via HAVE_S390_MIN_Z196_ZARCH_ASM_SUPPORT. ChangeLog: * sysdeps/s390/multiarch/utf8-utf32-z9.c: New File. * sysdeps/s390/utf8-utf32-z9.c: Move ifunc resolvers to multiarch folder and define ifunc versions depending on HAVE_[FROM|TO]_[C|CU|VX]. (HAVE_FROM_C, HAVE_FROM_CU, HAVE_FROM_VX, HAVE_TO_C, HAVE_TO_VX, FROM_LOOP_DEFAULT, FROM_LOOP_C, FROM_LOOP_CU, FROM_LOOP_VX, TO_LOOP_DEFAULT, TO_LOOP_C, TO_LOOP_VX): New Define.
* S390: Move utf16-utf32-z9.c to multiarch folder and use s390_libc_ifunc_expr ↵Stefan Liebler2017-04-213-72/+123
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | macro. The utf16-utf32-z9.c iconv module is using ifunc and thus the ifunc part should be in multiarch folder. Otherwise ifunc is used even if you configure with --disable-multi-arch. This patch moves the ifunc resolvers to the new file sysdeps/s390/multiarch/utf16-utf32-z9.c. The resolvers are now implemented with s390_libc_ifunc_expr macro instead of using gcc attribute ifunc directly. The ifunc versions are implemented in sysdeps/s390/utf16-utf32-z9.c. Each version is only implemented if needed or supported. Therefore there is a block at beginning of the file which selects the versions which should be defined depending on support for multiarch, vector-support and used minimum architecture level. This block defines HAVE_[FROM|TO]_[C|VX] to 1 or 0. The code below is rearranged and surrounded by #if HAVE_[FROM|TO]_[C|VX] == 1. There is no functional change. ChangeLog: * sysdeps/s390/multiarch/utf16-utf32-z9.c: New File. * sysdeps/s390/utf16-utf32-z9.c: Move ifunc resolvers to multiarch folder and define ifunc versions depending on HAVE_[FROM|TO]_[C|VX]. (HAVE_FROM_C, HAVE_FROM_VX, HAVE_TO_C, HAVE_TO_VX, FROM_LOOP_DEFAULT, FROM_LOOP_C, FROM_LOOP_VX, TO_LOOP_DEFAULT, TO_LOOP_C, TO_LOOP_VX): New Define.
* S390: Move utf8-utf16-z9.c to multiarch folder and use s390_libc_ifunc_expr ↵Stefan Liebler2017-04-213-95/+159
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | macro. The utf8-utf16-z9.c iconv module is using ifunc and thus the ifunc part should be in multiarch folder. Otherwise ifunc is used even if you configure with --disable-multi-arch. This patch moves the ifunc resolvers to the new file sysdeps/s390/multiarch/utf8-utf16-z9.c. The resolvers are now implemented with s390_libc_ifunc_expr macro instead of using gcc attribute ifunc directly. The ifunc versions are implemented in sysdeps/s390/utf8-utf16-z9.c. Each version is only implemented if needed or supported. Therefore there is a block at beginning of the file which selects the versions which should be defined depending on support for multiarch, vector-support and used minimum architecture level. This block defines HAVE_[FROM|TO]_[C|CU|VX] to 1 or 0. The code below is rearranged and surrounded by #if HAVE_[FROM|TO]_[C|CU|VX] == 1. There is no functional change. The cu instructions are z9 zarch instructions. As the major distros are already using the newer z196 as architecture level set, those instructions can be used as fallback version instead of the c-code. This behaviour is decided at compile time via HAVE_S390_MIN_Z196_ZARCH_ASM_SUPPORT. ChangeLog: * sysdeps/s390/multiarch/utf8-utf16-z9.c: New File. * sysdeps/s390/utf8-utf16-z9.c: Move ifunc resolvers to multiarch folder and define ifunc versions depending on HAVE_[FROM|TO]_[C|CU|VX]. (HAVE_FROM_C, HAVE_FROM_CU, HAVE_FROM_VX, HAVE_TO_C, HAVE_TO_VX, FROM_LOOP_DEFAULT, FROM_LOOP_C, FROM_LOOP_CU, FROM_LOOP_VX, TO_LOOP_DEFAULT, TO_LOOP_C, TO_LOOP_VX): New Define.
* S390: Use new s390_libc_ifunc_expr macro in s390 8bit-generic.c.Stefan Liebler2017-04-213-29/+25
| | | | | | | | | | | | | | | | | | | This patch adds s390_libc_ifunc_expr macro which uses the __ifunc base macro in include/libc-symbols.h and lets the user define a generic expression to choose the correct ifunc variant. Furthermore as the base macro is used, the ifunc resolver functions are now also using inhibit_stack_protector. S390 needs its own version due to the hwcap argument of the ifunc resolver. This new macro is now used in iconv code in 8bit-generic.c instead of using gcc attribute ifunc directly. ChangeLog: * sysdeps/s390/multiarch/ifunc-resolve.h (s390_libc_ifunc_expr_init, s390_libc_ifunc_expr): New Define. * sysdeps/s390/multiarch/8bit-generic.c (__to_generic, __from_generic): Use s390_libc_ifunc_expr to define ifunc resolvers.
* manual: Document replacing malloc [BZ #20424]Florian Weimer2017-04-212-0/+72
|
* Do not use wildcard symbol names for public versions in Versions files.Joseph Myers2017-04-203-2/+29
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | As noted in <https://sourceware.org/ml/libc-alpha/2012-12/msg00240.html>, stdlib/Versions and wcsmbs/Versions list some functions as __strto*_internal and __wcsto*_internal rather than explicitly listing the symbols to be exported (so any new internal function matching one of those patterns would be wrongly added to version GLIBC_2.0), which seems like a bad idea. This patch changes those files to list the exported symbols explicitly. There are still entries in sysdeps/nacl/Versions for __nacl_irt_*, but as GLIBC_PRIVATE symbols that seems less significant. Tested with build-many-glibcs.py that installed stripped shared libraries are unchanged by the patch. * stdlib/Versions (__strtod_internal): List explicitly, not as wildcard. (__strtof_internal): Likewise. (__strtold_internal): Likewise. (__strtol_internal): Likewise. (__strtoll_internal): Likewise. (__strtoul_internal): Likewise. (__strtoull_internal): Likewise. (__strtoq_internal): Likewise. (__strtouq_internal): Likewise. * wcsmbs/Versions (__wcstod_internal): Likewise. (__wcstof_internal): Likewise. (__wcstold_internal): Likewise. (__wcstol_internal): Likewise. (__wcstoll_internal): Likewise. (__wcstoul_internal): Likewise. (__wcstoull_internal): Likewise.
* Fix missing timespec definition for sys/stat.h (BZ #21371)Adhemerval Zanella2017-04-202-1/+6
| | | | | | | | | | | | | | | | | | | As indicated by the bug report, the 'struct timespec' definition is not defined for '_XOPEN_SOURCE=700' and '_POSIX_C_SOURCE=200112L'. It is because current code only includes its definition if __USE_ATFILE is defined and the define is only set with: 1. _GNU_SOURCE and/or _ATFILE_SOURCE definition. 2. _POSIX_C_SOURCE >= 200809L However, the 'st_*' fields in 'struct stat' are defined if __USE_XOPEN2K8. This patch uses the same logic for 'struct timespec' inclusion. Tested on x86_64-linux-gnu. * io/sys/stat.h: Use __USE_XOPEN2K8 insteaf of __USE_ATFILE for struct timespec definition.
* nss_dns: Correct parentheses for the __glibc_unlikely argumentFlorian Weimer2017-04-192-1/+5
| | | | This fixes commit bee05c9d58a34ec5886faf3b56ecaa56355d94bf.
* rcmd/rexec: Fix typo in commentFlorian Weimer2017-04-193-2/+7
|
* A third round of inclusion fixes for _ISOMAC testsuite.Zack Weinberg2017-04-194-0/+9
| | | | | | * posix/tst-mmap-offset.c: Include stdint.h. * resolv/tst-ns_name_compress.c: Include string.h. * resolv/tst-resolv-edns.c: Include resolv.h.
* resolv: Replace __builtin_expect with __glibc_unlikely/__glibc_likelyFlorian Weimer2017-04-196-20/+41
|
* Bug 21399: Fix CP1254 comment for U+00ECChristopher Chittleborough2017-04-192-1/+6
|
* Create more sockets with SOCK_CLOEXEC [BZ #15722]Florian Weimer2017-04-199-10/+28
|
* Assume that accept4 is always available and worksFlorian Weimer2017-04-1916-138/+230
| | | | | | | | | | Simplify the Linux accept4 implementation based on the assumption that it is available in some way. __ASSUME_ACCEPT4_SOCKETCALL was previously unused, so remove it. For ia64, the accept4 system call (and socket call) were backported in kernel version 3.2.18. Reflect this in the installation instructions.
* conformtest: Enable tests when cross compiling.Joseph Myers2017-04-182-2/+5
| | | | | | | | | | | | | This patch enables the compilation part of the conformtest tests (the vast bulk of them) when cross compiling, so making it easy to run them across many configurations with build-many-glibcs.py. Tested with build-many-glibcs.py. * conform/Makefile (tests-special): Do not make addition of $(conformtest-header-tests) conditional on [$(cross-compiling) = no]. (generated): Do not make addition of $(conformtest-header-base) conditional on [$(cross-compiling) = no].
* Fix bits/socket.h IOC* namespace issues (bug 21267).Joseph Myers2017-04-182-0/+45
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | sysdeps/unix/sysv/linux/bits/socket.h includes asm/socket.h. That includes asm/sockios.h, which on MIPS includes asm/ioctl.h, resulting in namespace violations from IOC* macros. bits/socket.h already has code to handle asm/socket.h unconditionally defining macros that are only wanted for __USE_MISC. This patch extends it to handle the IOC* macros as well (always undefining them if not defined when bits/socket.h was included, as I don't think they are part of the intended API even for __USE_MISC). It's possible there should also be a kernel fix - it's not clear to me that IOC* belong in the uapi headers, and even if they do they might best be split out into another header to avoid getting defined by this particular path. But since glibc needs to deal with existing kernel headers, it also seems appropriate to extend the existing workaround to these macros. Tested (compilation only) with build-many-glibcs.py. [BZ #21267] * sysdeps/unix/sysv/linux/bits/socket.h (IOCSIZE_MASK): Undefine if defined by <asm/socket.h> and not previously defined. (IOCSIZE_SHIFT): Likewise. (IOC_IN): Likewise. (IOC_INOUT): Likewise. (IOC_OUT): Likewise.
* x86: Use AVX2 memcpy/memset on Skylake server [BZ #21396]H.J. Lu2017-04-1811-1/+45
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | On Skylake server, AVX512 load/store instructions in memcpy/memset may lead to lower CPU turbo frequency in certain situations. Use of AVX2 in memcpy/memset has been observed to have improved overall performance in many workloads due to the higher frequency. Since AVX512ER is unique to Xeon Phi, this patch sets Prefer_No_AVX512 if AVX512ER isn't available so that AVX2 versions of memcpy/memset are used on Skylake server. [BZ #21396] * sysdeps/x86/cpu-features.c (init_cpu_features): Set Prefer_No_AVX512 if AVX512ER isn't available. * sysdeps/x86/cpu-features.h (bit_arch_Prefer_No_AVX512): New. (index_arch_Prefer_No_AVX512): Likewise. * sysdeps/x86_64/multiarch/memcpy.S (__new_memcpy): Don't use AVX512 version if Prefer_No_AVX512 is set. * sysdeps/x86_64/multiarch/memcpy_chk.S (__memcpy_chk): Likewise. * sysdeps/x86_64/multiarch/memmove.S (__libc_memmove): Likewise. * sysdeps/x86_64/multiarch/memmove_chk.S (__memmove_chk): Likewise. * sysdeps/x86_64/multiarch/mempcpy.S (__mempcpy): Likewise. * sysdeps/x86_64/multiarch/mempcpy_chk.S (__mempcpy_chk): Likewise. * sysdeps/x86_64/multiarch/memset.S (memset): Likewise. * sysdeps/x86_64/multiarch/memset_chk.S (__memset_chk): Likewise.
* x86: Set Prefer_No_VZEROUPPER if AVX512ER is availableH.J. Lu2017-04-183-2/+42
| | | | | | | | | | | | | | | | | | | | | | | | | | | AVX512ER won't be implemented in any Xeon processors and will be in all Xeon Phi processors. Don't check CPU model number when setting Prefer_No_VZEROUPPER for Xeon Phi. Instead, set Prefer_No_VZEROUPPER if AVX512ER is available. It works with current and future Xeon Phi and non-Xeon Phi processors. * sysdeps/x86/cpu-features.c (init_cpu_features): Set Prefer_No_VZEROUPPER if AVX512ER is available. * sysdeps/x86/cpu-features.h (bit_cpu_AVX512PF): New. (bit_cpu_AVX512ER): Likewise. (bit_cpu_AVX512CD): Likewise. (bit_cpu_AVX512BW): Likewise. (bit_cpu_AVX512VL): Likewise. (index_cpu_AVX512PF): Likewise. (index_cpu_AVX512ER): Likewise. (index_cpu_AVX512CD): Likewise. (index_cpu_AVX512BW): Likewise. (index_cpu_AVX512VL): Likewise. (reg_AVX512PF): Likewise. (reg_AVX512ER): Likewise. (reg_AVX512CD): Likewise. (reg_AVX512BW): Likewise. (reg_AVX512VL): Likewise.
* Assume that O_CLOEXEC is always defined and worksFlorian Weimer2017-04-1827-565/+91
|
* Assume that dup3 is availableFlorian Weimer2017-04-189-82/+21
|
* Assume that pipe2 is always availableFlorian Weimer2017-04-186-96/+32
| | | | | The Debian patches for Hurd (which are already required to build glibc before this commit) contain an implementation of pipe2.
* malloc: Turn cfree into a compatibility symbolFlorian Weimer2017-04-187-19/+17
|
generated by cgit-pink 1.4.1 (git 2.36.1) at 2024-11-15 03:58:38 +0000