about summary refs log tree commit diff
path: root/sysdeps
Commit message (Collapse)AuthorAgeFilesLines
...
* tile: add no-op fe*() routines for libc internal useChris Metcalf2014-12-261-0/+4
| | | | | | | | | These avoid having tile generate real calls to the no-op functions, which then causes linknamespace test failures. It might make sense to factor all of these out into a common header that can be shared by tile, microblaze, etc., but for now just fix the test failures.
* linux/clock_settime: remove unnecessary vDSO definitionsChris Metcalf2014-12-261-9/+0
| | | | | | | | | | | | These definitions were added back before __ASSUME_POSIX_CPU_TIMERS was removed. There used to be a vsyscall to clock_getres() in maybe_syscall_settime_cpu(), but that function was removed in commit 26889eac. The presence of the vsyscall definitions means that platforms that don't provide clock_getres as a vsyscall hit a symbol redefinition warning in this file, becoming fatal with -Werror. Removing the vsyscall definitions is the obvious fix. No change to generated code on x86_64.
* tilegx: fix sysdep.h to avoid a redefinition warningChris Metcalf2014-12-261-3/+3
| | | | | | | | The symbol for HAVE_CLOCK_GETTIME_VSYSCALL was being only conditionally defined under [SHARED]. However, it turns out this causes a preprocessor symbol redefinition warning when building clock_gettime.o. Move the symbol definition down to make it unconditional, like other platforms do.
* tilegx32: avoid a a -Werror warning from unwindingChris Metcalf2014-12-261-1/+1
| | | | | | | The _Unwind_GetCFA() routine returns a 64-bit value, which we interpret as a pointer. Add an intermediate cast to long so that in ILP32 mode we don't get a warning about casting a wrong-sized integer to a pointer.
* tilegx: enable wordsize-64 support for ieee745 dbl-64.Chris Metcalf2014-12-231-0/+1
| | | | | | I missed this during the initial port. Some testing shows that enabling this mode does, unsurprisingly, yield some nice speedups on the math functions in question.
* Make __ASSUME_UTIMES hppa-specific.Joseph Myers2014-12-226-44/+38
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This patch makes __ASSUME_UTIMES hppa-specific, removing mentions of the macro from architecture-independent code and code for other architectures. (All other architectures either have the utimes syscall in all relevant kernel versions, or use the asm-generic interface so only have utimensat and won't get the utimes syscall.) A similar approach is used to that used for futimesat for MicroBlaze: if the kernel is recent enough that the utimes syscall can be assumed to be present, use the implementation in terms of the utimes syscall, and otherwise use the linux/generic implementation in terms of utimensat. Tested x86_64 that the disassembly of installed shared libraries is unchanged by the patch. Not tested for hppa. * sysdeps/unix/sysv/linux/kernel-features.h (__ASSUME_UTIMES): Do not define. * sysdeps/unix/sysv/linux/utimes.c: Do not include <kernel-features.h>. (__utimes) [__NR_utimes]: Make code unconditional. (__utimes) [!__ASSUME_UTIMES]: Remove conditional code. * sysdeps/unix/sysv/linux/aarch64/kernel-features.h (__ASSUME_UTIMES): Do not undefine. * sysdeps/unix/sysv/linux/tile/kernel-features.h (__ASSUME_UTIMES): Likewise. * sysdeps/unix/sysv/linux/hppa/kernel-features.h (__ASSUME_UTIMES): Define for [__LINUX_KERNEL_VERSION >= 0x030e00] instead of undefining for [__LINUX_KERNEL_VERSION < 0x030e00]. * sysdeps/unix/sysv/linux/hppa/utimes.c: New file.
* Fix preprocessor indentation in sysdeps/mips/memcpy.S.Steve Ellcey2014-12-221-97/+97
| | | | Checked in as obvious.
* tilegx: remove implicit boolean conversion in strstr.Chris Metcalf2014-12-221-1/+1
| | | | | | | | [BZ #17746] The __builtin_expect() truncated a uint64_t to a 32-bit long in ILP32 mode, discarding the high 32 bits, and potentially missing the NUL terminator that we were searching for with SIMD operations. Explicitly compare to zero to fix the problem.
* 2014-12-22 Steve Ellcey <sellcey@imgtec.com>Steve Ellcey2014-12-221-0/+4
| | | | | * sysdeps/unix/mips/sysdep.h (__mips_isa_rev): Set to zero if not already defined.
* Split __kernel_standard* functions (fixes bug 17724).Joseph Myers2014-12-223-90/+154
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Bug 17724 reports references to fesetround being brought in by ldbl-128ibm rintl via references to __rintl from __kernel_standard_l. Because all three __kernel_standard* functions are in the same file, this gets brought in even though only the long double version __kernel_standard_l needs __rintl, and the C90 functions use only __kernel_standard. This patch fixes this by splitting the three versions into separate files; it's fine for long double functions to refer to fe* functions directly, unless they get called by C90 double functions. Tested for x86_64 (testsuite; the reordering of code means disassembly of shared libraries can't usefully be compared). Tested for powerpc that the relevant issue disappears from the linknamespace test output. [BZ #17724] * sysdeps/ieee754/k_standard.c: Don't include <float.h>. (__kernel_standard_f): Remove. Moved to k_standardf.c. (__kernel_standard_l): Remove. Moved to k_standardl.c with (char *) casts added. * sysdeps/ieee754/k_standardf.c: New file. * sysdeps/ieee754/k_standardl.c: Likewise. * math/Makefile (libm-support): Remove k_standard. (libm-calls): Add k_standard.
* Call libc_fetestexcept_aarch64.Wilco Dijkstra2014-12-221-7/+2
|
* Call libc_fesetround_aarch64.Wilco Dijkstra2014-12-221-20/+5
|
* Fix resolver bind, getsockname namespace (bug 17733).Joseph Myers2014-12-222-4/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | On Linux architectures using socketcall, the resolver ends up bringing in strong symbols for bind and getsockname, which are not in POSIX.1-1996. This causes linknamespace test failures: FAIL: conform/POSIX/pthread.h/linknamespace FAIL: conform/POSIX/sched.h/linknamespace FAIL: conform/POSIX/time.h/linknamespace These functions are defined as strong symbols with __bind and __getsockname as weak aliases. This patch switches this to the other way round by removing the NO_WEAK_ALIAS definitions and so letting the default case in socket.S act; I see no reason for the existing arrangements. Tested for x86 (testsuite, and that disassembly of installed shared libraries is unchanged by the patch). [BZ #17733] * sysdeps/unix/sysv/linux/bind.S (NO_WEAK_ALIAS): Do not define. (__bind): Do not define as weak alias. * sysdeps/unix/sysv/linux/getsockname.S (NO_WEAK_ALIAS): Do not define. (__getsockname): Do not define as weak alias.
* tile: separate ffsll from ffsChris Metcalf2014-12-212-11/+33
| | | | | This avoids a linknamespace failure when ffs is legal but ffsll is not.
* tile: remove localplt.data and use generic one again.Chris Metcalf2014-12-211-18/+0
| | | | With the __tls_get_addr removed, it works for tile.
* Add hidden __tls_get_addr/___tls_get_addr aliasH.J. Lu2014-12-215-15/+10
| | | | | | | | | | | | | | | | | | | | __tls_get_addr/___tls_get_addr is always defined in ld.so. There is no need to call them via PLT inside ld.so. This patch adds the hidden __tls_get_addr/___tls_get_addr aliases and calls them directly from _dl_tlsdesc_dynamic. There is no need to set up the EBX register in i386 _dl_tlsdesc_dynamic when calling the hidden ___tls_get_addr. * elf/dl-tls.c (__tls_get_addr): Provide the hidden definition if not defined. * sysdeps/i386/dl-tls.h (___tls_get_addr): Provide the hidden definition. * sysdeps/i386/dl-tlsdesc.S (_dl_tlsdesc_dynamic): Call the hidden ___tls_get_addr. * sysdeps/x86_64/dl-tlsdesc.S (_dl_tlsdesc_dynamic): Call the hidden __tls_get_addr. * sysdeps/generic/localplt.data (__tls_get_addr): Removed. * sysdeps/unix/sysv/linux/i386/localplt.data (___tls_get_addr): Likewise.
* m68k: remove @PLTPC from _dl_init callAndreas Schwab2014-12-211-1/+1
|
* Remove @PLT from "call _dl_init@PLT" in _dl_start_userH.J. Lu2014-12-212-2/+2
| | | | | | | | | | _dl_start_user in ld.so calls the local function _dl_init. There is no need to go through PLT. * sysdeps/i386/dl-machine.h (_dl_start_user): Remove @PLT from "call _dl_init@PLT". * sysdeps/x86_64/dl-machine.h (_dl_start_user): Likewise. from "call _dl_init@PLT".
* tile: provide localplt.data with __tls_get_addr optionalChris Metcalf2014-12-201-0/+18
|
* tilegx: fix strstr to build and link betterChris Metcalf2014-12-191-1/+2
| | | | | | | | The two_way_short_needle() routine included from str-two-way.h is not used, so mark it so to avoid compiler warnings. Calling strnlen() breaks linknamespace tests, so change it to __strnlen().
* Remove trailing white space.Steve Ellcey2014-12-191-1/+0
|
* 2014-12-19 Steve Ellcey <sellcey@imgtec.com>Steve Ellcey2014-12-196-0/+182
| | | | | | | | | | * sysdeps/mips/addmul_1.S (__mpn_addmul_1): Use mulu/muhu instead of multu on MIPSr6. * sysdeps/mips/mul_1.S (__mpn_mul_1): Ditto. * sysdeps/mips/submul_1.S (__mpn_submul_1): Ditto. * sysdeps/mips/mips64/addmul_1.S (__mpn_addmul_1): Ditto. * sysdeps/mips/mips64/mul_1.S (__mpn_mul_1): Ditto. * sysdeps/mips/mips64/submul_1.S (__mpn_submul_1): Ditto.
* 2014-12-19 Steve Ellcey <sellcey@imgtec.com>Steve Ellcey2014-12-192-4/+16
| | | | | | | | | | * sysdeps/mips/sys/asm.h (PTR_ADDU): Use addu on mips32r6/mips64r6. (PTR_ADDIU): Use addiu for mips32r6/mips64r6. (PTR_SUBU): Use subu for mips32r6/mips64r6. (PTR_SUBIU): Use subu for mips32r6/mips64r6 (subiu does not exist). * sysdeps/mips/machine-gmon.h (PTR_ADDU_STRING) Use addu for mips32r6/mips64r6. (PTR_SUBU_STRING) Use subu for mips32r6/mips64r6.
* Fix soft-fp build warning on sparc about strict aliasing.David S. Miller2014-12-191-6/+11
| | | | | * sysdeps/sparc/sparc32/soft-fp/q_neg.c (_Q_neg): Use a union to access the quad as both a long double and as a series of 4 words.
* Replace -Wno-error with -fno-builtin-lroundH.J. Lu2014-12-191-2/+2
|
* i386: Move futex functions from lowlevellock.h to lowlevellock-futex.h.Torvald Riegel2014-12-192-110/+157
|
* Use generic lowlevellock-futex.h in x86_64 lowlevellock.h.Torvald Riegel2014-12-192-120/+24
|
* sh: Remove custom lowlevellock, barrier, condvar, and rwlock implementations.Torvald Riegel2014-12-1919-5572/+0
|
* Compile s_llround.c with -Wno-error for x32 buildH.J. Lu2014-12-191-0/+6
| | | | | | Since x32 returns 32-bit long int and 64-bit long long int in the same 64-bit register, we make the 32b-bit lround an alias of the 64-bit llround. Add -Wno-error for x32 build to silence the compiler.
* Replace 1L with (mp_limb_t) 1H.J. Lu2014-12-191-1/+1
| | | | | | | | | | | | | | | X86-64 and x32 use sysdeps/i386/ldbl2mpn.c. res_ptr is a pointer to mp_limb_t, which is long for i386 and x86-64 and long long for x32. On x32, I got ../sysdeps/x86_64/../i386/ldbl2mpn.c: In function ‘__mpn_extract_long_double’: ../sysdeps/x86_64/../i386/ldbl2mpn.c:72:4: error: left shift count >= width of type [-Werror] res_ptr[N - 1] &= ~(1L << ((LDBL_MANT_DIG - 1) % BITS_PER_MP_LIMB)); ^ cc1: all warnings being treated as errors This patch replaces 1L with (mp_limb_t) 1. Verified on x32, i686 and x86-64 with GCC 4.8.3.
* NPTL: Move fork state variables to initializer files.Roland McGrath2014-12-171-8/+0
|
* NPTL: Remove gratuitous Linuxisms from gai_misc.h.Roland McGrath2014-12-171-6/+6
|
* Fix profil_counter namespace (bug 17725).Joseph Myers2014-12-1714-20/+29
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | On ARM, where profil_counter is not static, it is brought in by references to various standard functions, as noted in <https://sourceware.org/ml/libc-alpha/2014-11/msg00890.html>, although it is not a standard function itself. I don't know if this also causes test failures on SPARC, although I see no reason for it not to do so. This patch fixes this namespace issue. profil_counter is renamed to __profil_counter and made a weak alias on ARM and SPARC. Because of the uses in profil.c / sprofil.c it seems simplest to make the rename globally, including on the other architectures for which profil_counter was static and so the change is of no substance. The variant names profil_counter_* used in sprofil.c are also renamed to start with __ so that undesired function names do not get exported in static libc. As I noted in bug 17726, profil_counter should probably be a compat symbol on ARM and SPARC, so it wouldn't exist at all in static libc even as a weak alias. Since defining a compat symbol still requires an internal name as a target of an alias, this patch still seems reasonable as an intermediate step towards that goal: it wouldn't be possible for the function simply to be static profil_counter on ARM and SPARC with profil_counter also being the exported compat symbol name, so profil.c / sprofil.c would still need to be prepared to call the function under another name (here, __profil_counter). Tested for x86_64 (testsuite, and that stripped installed shared libraries are unchanged by the patch) and ARM (ABI and linknamespace tests - this patch reduces the number of linknamespace failures I see on ARM from 227 to 5, the residue being math.h failures for fe* functions and for j0l/j1n/jnl/y0l/y1l/ynl aliases). 2014-12-17 Joseph Myers <joseph@codesourcery.com> [BZ #17725] * sysdeps/generic/profil-counter.h (profil_counter): Rename to __profil_counter. * sysdeps/unix/sysv/linux/hppa/profil-counter.h (profil_counter): Likewise. * sysdeps/unix/sysv/linux/i386/profil-counter.h (profil_counter): Likewise. * sysdeps/unix/sysv/linux/ia64/profil-counter.h (profil_counter): Likewise. * sysdeps/unix/sysv/linux/s390/s390-32/profil-counter.h (profil_counter): Likewise. * sysdeps/unix/sysv/linux/s390/s390-64/profil-counter.h (profil_counter): Likewise. * sysdeps/unix/sysv/linux/sh/profil-counter.h (profil_counter): Likewise. * sysdeps/unix/sysv/linux/tile/profil-counter.h (profil_counter): Likewise. * sysdeps/unix/sysv/linux/x86_64/profil-counter.h (profil_counter): Likewise. * sysdeps/unix/sysv/linux/arm/profil-counter.h (profil_counter): Likewise. [!__profil_counter] (profil_counter): Define as weak alias of __profil_counter. * sysdeps/unix/sysv/linux/sparc/sparc32/profil-counter.h (profil_counter): Rename to __profil_counter. [!__profil_counter] (profil_counter): Define as weak alias of __profil_counter. * sysdeps/unix/sysv/linux/sparc/sparc64/profil-counter.h (profil_counter): Rename to __profil_counter. [!__profil_counter] (profil_counter): Define as weak alias of __profil_counter. * sysdeps/posix/profil.c: Update comment referring to profil_counter. (__profil): Use __profil_counter instead of profil_counter. * sysdeps/posix/sprofil.c (profil_counter): Rename to __profil_counter. Use __profil_counter_ushort and __profil_counter_uint in definitions. (__sprofil): Use __profil_counter_uint and __profil_counter_ushort instead of profil_counter_uint and profil_counter_ushort.
* Use PTR_MANGLE on libgcc unwinder function pointers.Roland McGrath2014-12-161-3/+12
|
* Revert "Use pragmas rather than makefiles for necessary options for unwind ↵Roland McGrath2014-12-164-6/+2
| | | | | | code." This reverts commit c324fcfe75c3976ae0b16943df00710e1d0d74f7.
* Use pragmas rather than makefiles for necessary options for unwind code.Roland McGrath2014-12-164-2/+6
|
* Fix x86_64 memrchr namespace (bug 17719).Joseph Myers2014-12-161-3/+3
| | | | | | | | | | | | | | | | | | | | | | | On x86_64, memrchr (not a standard function) is defined as a strong symbol, instead of a weak alias of __memrchr as on other architectures. This results in linknamespace test failures from the use of __memrchr from dirname. (Not a conformance issue because of the mem* reservation, but contrary to glibc conventions.) This patch makes x86_64 follow other architectures by defining memrchr as a weak alias. Tested for x86_64 (testsuite, and that disassembly of installed shared libraries is unchanged by the patch). [BZ #17719] * sysdeps/x86_64/memrchr.S (memrchr): Rename to __memrchr and define as weak alias of __memrchr. (__memrchr): Do not define as strong alias of memrchr. * conform/Makefile (test-xfail-XPG4/libgen.h/linknamespace): Remove variable. (test-xfail-UNIX98/libgen.h/linknamespace): Likewise. (test-xfail-XOPEN2K/libgen.h/linknamespace): Likewise. (test-xfail-XOPEN2K8/libgen.h/linknamespace): Likewise.
* Fix resolver if_* namespace (bug 17717).Joseph Myers2014-12-162-15/+26
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Resolver code, brought in by pthreads (at least), uses if_* interfaces that weren't in POSIX before 2001, resulting in linknamespace failures. This patch changes those interfaces to be weak aliases of __if_* and makes the resolver use __if_* directly. Tested for x86_64 (testsuite, and that disassembly of installed shared libraries is unchanged by this patch). [BZ #17717] * inet/if_index.c (if_nametoindex): Rename to __if_nametoindex and define as weak alias of __if_nametoindex. Use libc_hidden_weak. (if_indextoname): Rename to __if_indextoname and define as weak alias of __if_indextoname. Use libc_hidden_weak. (if_freenameindex): Rename to __if_freenameindex and define as weak alias of __if_freenameindex. (if_nameindex): Rename to __if_nameindex and define as weak alias of __if_nameindex. * sysdeps/mach/hurd/if_index.c (if_nametoindex): Rename to __if_nametoindex and define as weak alias of __if_nametoindex. Use libc_hidden_weak. (if_freenameindex): Rename to __if_freenameindex and define as weak alias of __if_freenameindex. (if_nameindex): Rename to __if_nameindex and define as weak alias of __if_nameindex. (if_indextoname): Rename to __if_indextoname and define as weak alias of __if_indextoname. Use libc_hidden_weak. * sysdeps/unix/sysv/linux/if_index.c (if_nametoindex): Rename to __if_nametoindex and define as weak alias of __if_nametoindex. Use libc_hidden_weak. (if_freenameindex): Rename to __if_freenameindex and define as weak alias of __if_freenameindex. Use libc_hidden_weak. (if_nameindex_netlink): Use __if_freenameindex instead of if_freenameindex. (if_nameindex): Rename to __if_nameindex and define as weak alias of __if_nameindex. Use libc_hidden_weak. (if_indextoname): Rename to __if_indextoname and define as weak alias of __if_indextoname. Use libc_hidden_weak. * include/net/if.h [!_ISOMAC] (__if_nametoindex): Declare and use libc_hidden_proto. [!_ISOMAC] (__if_freenameindex): Likewise. * resolv/res_init.c (__res_vinit): Use __if_nametoindex instead of if_nametoindex. * conform/Makefile (test-xfail-XPG4/grp.h/linknamespace): Remove variable. (test-xfail-XPG4/pwd.h/linknamespace): Likewise. (test-xfail-UNIX98/aio.h/linknamespace): Likewise. (test-xfail-UNIX98/grp.h/linknamespace): Likewise. (test-xfail-UNIX98/pthread.h/linknamespace): Likewise. (test-xfail-UNIX98/pwd.h/linknamespace): Likewise. (test-xfail-UNIX98/sched.h/linknamespace): Likewise. (test-xfail-UNIX98/time.h/linknamespace): Likewise.
* Add comments for the generic lowlevellock implementation.Torvald Riegel2014-12-151-9/+82
| | | | Patch by Bernard Ogden <bernie.ogden@linaro.org>.
* Remove custom pthread_once implementation on s390.Torvald Riegel2014-12-151-110/+0
|
* NPTL: Refactor named semaphore code to use shm-directory.hRoland McGrath2014-12-128-7/+61
|
* Refactor shm_{open,unlink} code to separate Linux-specific directory choice ↵Roland McGrath2014-12-118-282/+276
| | | | from POSIX-generic code.
* * Fix SH specific compiler warnings which are for integer-pointerKaz Kojima2014-12-124-5/+5
| | | | type conversions without cast.
* Move semaphore.h to sysdeps/pthread/.Joseph Myers2014-12-111-0/+78
| | | | | | | | | | | | | | | | | | | | | | | | | | | Carlos reported failures in conform/ tests in environments where the compiler used could only find headers in glibc's source and build trees, not any previously installed headers <https://sourceware.org/ml/libc-alpha/2014-09/msg00040.html>. This patch moves nptl/semaphore.h to sysdeps/pthread/semaphore.h so that it can be found by builds from all glibc subdirectories; it's not in any way NPTL-specific. (I left the Makefile setting to install this header in nptl/, but maybe it should move as well - it's just not clear to me what ifeq ($(subdir),...) conditional should be used to select the directory to associate the header with for installation purposes. The path in the toplevel Makefile used for begin-end-check also remains hardcoded; it's a known todo issue to rework that test to run in each subdirectory checking the headers installed from that subdirectory, rather than a separate hardcoded list.) Tested for x86_64 (testsuite, and that installed stripped shared libraries are unchanged by the patch). I did *not* test a configuration such as that in which Carlos saw failure. * nptl/semaphore.h: Move to ... * sysdeps/pthread/semaphore.h: ... here. * Makefile (installed-headers): Change nptl/semaphore.h to sysdeps/pthread/semaphore.h.
* tile: add inhibit_loop_to_libcall to string functionsChris Metcalf2014-12-113-3/+3
| | | | | | Without this, on gcc 4.8.2 the built glibc crashes when memcpy or memset are invoked, since they call themselves recursively. See commit 85c2e6110c9a01ec for the generic inhibit_loop_to_libcall.
* * sysdeps/mips/dl-trampoline.c: Modify switch expression to haveSteve Ellcey2014-12-111-1/+1
| | | | integer value instead of boolean.
* S/390: Get rid of assembler warning value truncated.Stefan Liebler2014-12-111-1/+1
|
* powerpc: Fix lgammal_r overflow warningsAdhemerval Zanella2014-12-111-1/+9
| | | | | | | ldbl-128ibm uses ldbl-128 e_lgammal_r implementation as is, however some constants definitions overflows for IBM long double range. This patch suppress the compiler warnings until the ldbl-128ibm implementation is fixed.
* Properly handle forced elision in pthread_mutex_trylock (bug 16657)Andreas Schwab2014-12-112-10/+0
|
* Fix MIPS waitid build.Joseph Myers2014-12-101-0/+7
| | | | | | | | | | | | | | | As previously discussed in <https://sourceware.org/ml/libc-alpha/2012-11/msg00798.html>, MIPS (o32) waitid has build warnings (now errors) because a function is declared inline but functions with five-argument syscalls cannot be inlined for MIPS o32. This patch disables the -Winline warnings for waitid.c using a MIPS-specific wrapper file. As it's whole-file disabling, there's no point in using push and pop, so just DIAG_IGNORE_NEEDS_COMMENT is used. * sysdeps/unix/sysv/linux/mips/mips32/waitid.c: New file.
generated by cgit-pink 1.4.1 (git 2.36.1) at 2024-11-18 16:45:07 +0000