Zack Weinberg [Fri, 9 Jun 2017 13:32:43 +0000 (09:32 -0400)]
Polish the treatment of dl-tunable-list.h in Makeconfig.
Like basically all before-compile headers, dl-tunable-list.h should be
generated using a stamp file to minimize unnecessary rebuilding; it
wasn't being added to common-generated, so it wouldn't get cleaned up;
and it was in between the rules for libc-modules.h/.stmp and their own
addition to common-generated.
* Makeconfig (dl-tunable-list.h): Generate using a stamp file.
(common-generated): Add libc-modules.h and libc-modules.stmp in a
more appropriate location. Also add dl-tunable-list.h and
dl-tunable-list.stmp.
H.J. Lu [Fri, 9 Jun 2017 12:45:43 +0000 (05:45 -0700)]
x86-64: Optimize strrchr/wcsrchr with AVX2
Optimize strrchr/wcsrchr with AVX2 to check 32 bytes with vector
instructions. It is as fast as SSE2 version for small data sizes
and up to 1X faster for large data sizes on Haswell. Select AVX2
version on AVX2 machines where vzeroupper is preferred and AVX
unaligned load is fast.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
strrchr-sse2, strrchr-avx2, wcsrchr-sse2 and wcsrchr-avx2.
* sysdeps/x86_64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Add tests for __strrchr_avx2,
__strrchr_sse2, __wcsrchr_avx2 and __wcsrchr_sse2.
* sysdeps/x86_64/multiarch/strrchr-avx2.S: New file.
* sysdeps/x86_64/multiarch/strrchr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/strrchr.c: Likewise.
* sysdeps/x86_64/multiarch/wcsrchr-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/wcsrchr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/wcsrchr.c: Likewise.
H.J. Lu [Fri, 9 Jun 2017 12:44:30 +0000 (05:44 -0700)]
x86-64: Optimize memrchr with AVX2
Optimize memrchr with AVX2 to search 32 bytes with a single vector
compare instruction. It is as fast as SSE2 memrchr for small data
sizes and up to 1X faster for large data sizes on Haswell. Select
AVX2 memrchr on AVX2 machines where vzeroupper is preferred and AVX
unaligned load is fast.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
memrchr-sse2 and memrchr-avx2.
* sysdeps/x86_64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Add tests for __memrchr_avx2 and
__memrchr_sse2.
* sysdeps/x86_64/multiarch/memrchr-avx2.S: New file.
* sysdeps/x86_64/multiarch/memrchr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/memrchr.c: Likewise.
H.J. Lu [Fri, 9 Jun 2017 12:42:16 +0000 (05:42 -0700)]
x86-64: Optimize strchr/strchrnul/wcschr with AVX2
Optimize strchr/strchrnul/wcschr with AVX2 to search 32 bytes with vector
instructions. It is as fast as SSE2 versions for size <= 16 bytes and up
to 1X faster for or size > 16 bytes on Haswell. Select AVX2 version on
AVX2 machines where vzeroupper is preferred and AVX unaligned load is fast.
NB: It uses TZCNT instead of BSF since TZCNT produces the same result
as BSF for non-zero input. TZCNT is faster than BSF and is executed
as BSF if machine doesn't support TZCNT.
H.J. Lu [Fri, 9 Jun 2017 12:18:03 +0000 (05:18 -0700)]
x86-64: Optimize strlen/strnlen/wcslen/wcsnlen with AVX2
Optimize strlen/strnlen/wcslen/wcsnlen with AVX2 to check 32 bytes with
a single vector compare instruction. It is as fast as SSE2 versions for
size <= 16 bytes and up to 1X faster for or size > 16 bytes on Haswell.
Select AVX2 version on AVX2 machines where vzeroupper is preferred and
AVX unaligned load is fast.
NB: It uses TZCNT instead of BSF since TZCNT produces the same result
as BSF for non-zero input. TZCNT is faster than BSF and is executed
as BSF if machine doesn't support TZCNT.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
strlen-sse2, strnlen-sse2, strlen-avx2, strnlen-avx2,
wcslen-sse2, wcslen-avx2 and wcsnlen-avx2.
* sysdeps/x86_64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Add tests for __strlen_avx2,
__strlen_sse2, __strnlen_avx2, __strnlen_sse2, __wcslen_avx2,
__wcslen_sse2 and __wcsnlen_avx2.
* sysdeps/x86_64/multiarch/strlen-avx2.S: New file.
* sysdeps/x86_64/multiarch/strlen-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/strlen.c: Likewise.
* sysdeps/x86_64/multiarch/strnlen-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/strnlen-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/strnlen.c: Likewise.
* sysdeps/x86_64/multiarch/wcslen-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/wcslen-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/wcslen.c: Likewise.
* sysdeps/x86_64/multiarch/wcsnlen-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/wcsnlen.c (OPTIMIZE (avx2)): New.
(IFUNC_SELECTOR): Return OPTIMIZE (avx2) on AVX2 machines where
vzeroupper is preferred and AVX unaligned load is fast.
H.J. Lu [Fri, 9 Jun 2017 12:13:15 +0000 (05:13 -0700)]
x86-64: Optimize memchr/rawmemchr/wmemchr with SSE2/AVX2
SSE2 memchr is extended to support wmemchr. AVX2 memchr/rawmemchr/wmemchr
are added to search 32 bytes with a single vector compare instruction.
AVX2 memchr/rawmemchr/wmemchr are as fast as SSE2 memchr/rawmemchr/wmemchr
for small sizes and up to 1.5X faster for larger sizes on Haswell and
Skylake. Select AVX2 memchr/rawmemchr/wmemchr on AVX2 machines where
vzeroupper is preferred and AVX unaligned load is fast.
NB: It uses TZCNT instead of BSF since TZCNT produces the same result
as BSF for non-zero input. TZCNT is faster than BSF and is executed
as BSF if machine doesn't support TZCNT.
* sysdeps/x86_64/memchr.S (MEMCHR): New. Depending on if
USE_AS_WMEMCHR is defined.
(PCMPEQ): Likewise.
(memchr): Renamed to ...
(MEMCHR): This. Support wmemchr if USE_AS_WMEMCHR is defined.
Replace pcmpeqb with PCMPEQ.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
memchr-sse2, rawmemchr-sse2, memchr-avx2, rawmemchr-avx2,
wmemchr-sse4_1, wmemchr-avx2 and wmemchr-c.
* sysdeps/x86_64/multiarch/ifunc-avx2.h: New file.
* sysdeps/x86_64/multiarch/memchr-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/memchr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/memchr.c: Likewise.
* sysdeps/x86_64/multiarch/rawmemchr-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/rawmemchr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/rawmemchr.c: Likewise.
* sysdeps/x86_64/multiarch/wmemchr-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/wmemchr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/wmemchr.c: Likewise.
* sysdeps/x86_64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Test __memchr_avx2, __memchr_sse2,
__rawmemchr_avx2, __rawmemchr_sse2, __wmemchr_avx2 and
__wmemchr_sse2.
Florian Weimer [Fri, 9 Jun 2017 12:08:13 +0000 (14:08 +0200)]
support: Expose TEST_VERIFY_EXIT behavior to GCC optimizers
Previously, the implementation would conditionally exit based on the
status argument, which GCC did not know about. This leads to
false uninitialized variable warnings when data is accessed after a
TEST_VERIFY_EXIT failure (from code which would never execute).
Joseph Myers [Thu, 8 Jun 2017 22:34:58 +0000 (22:34 +0000)]
conformtest: Correct sys/wait.h expectations for XPG4.
This patch corrects the conformtest expectations for sys/wait.h for
XPG4. This does not change the test results for this header at the
makefile level since it fixes some spurious failures for this header
while introducing other failures previously hidden by incorrect
expectations. However, it shows up issues with stdlib.h for XPG4
defining W* names that are not permitted in this case (previously
wrongly allowed; the W* reservation is UX-shaded in XPG4.2 /
XSI-shaded in POSIX before 2008); that test is thus XFAILed.
Tested for x86_64.
* conform/data/sys/wait.h-data (WIFCONTINUED): Do not expect for
[XPG4].
(WEXITED): Likewise.
(WSTOPPED): Likewise.
(WCONTINUED): Likewise.
(WNOHANG): Likewise.
(WNOWAIT): Likewise.
(idtype_t): Likewise.
(P_ALL): Likewise.
(P_PID): Likewise.
(P_PGID): Likewise.
(id_t): Likewise.
(siginfo_t): Likewise.
(wait3): Likewise.
(waitid): Likewise.
(signal.h): Do not allow header for [XPG4].
(sys/resource.h): Likewise.
(si_*): Do not allow for [XPG4].
(W*): Likewise.
(P_*): Likewise.
(BUS_): Likewise.
(CLD_): Likewise.
(FPE_): Likewise.
(ILL_): Likewise.
(POLL_): Likewise.
(SEGV_): Likewise.
(SI_): Likewise.
(TRAP_): Likewise.
* conform/Makefile (test-xfail-XPG4/stdlib.h/conform): New
variable.
H.J. Lu [Thu, 8 Jun 2017 16:55:48 +0000 (09:55 -0700)]
Add more tests for memchr
This patch adds tests for len == 0 and tests for positions close to the
beginning, which are equivalent to positions close to the end for memchr.
* string/test-memrchr.c (test_main): Add tests for len == 0
and tests for positions close to the beginning, which are
equivalent to positions close to the end for memchr.
Add strfromf128 to stdlib when _Float128 support is enabled.
* stdio-common/printf-parsemb.c (__parse_one_specmb): Initialize
spec->info.is_binary128 to zero.
* stdio-common/printf.h (printf_info): Add new member is_binary128
to indicate that the number being converted to string is compatible
with the IEC 60559 binary128 format.
* stdio-common/printf_fp.c (__printf_fp_l): Add code to deal with
_Float128 numbers.
* stdio-common/printf_fphex.c: Include ieee754_float128.h and
ldbl-128/printf_fphex_macros.h
(__printf_fphex): Add code to deal with _Float128 numbers.
* stdio-common/printf_size.c (__printf_size): Likewise.
* stdio-common/vfprintf.c (process_arg): Initialize member
info.is_binary128 to zero.
* stdlib/fpioconst.h (FLT128_MAX_10_EXP_LOG): New macro.
* stdlib/stdlib.h: Include bits/floatn.h for _Float128 support.
(strfromf128): New declaration.
* stdlib/strfrom-skeleton.c (STRFROM): Set member info.is_binary128
to one.
* sysdeps/ieee754/float128/Makefile: Add strfromf128.
* sysdeps/ieee754/float128/Versions: Likewise.
* sysdeps/ieee754/float128/strfromf128.c: New file.
Refactor PRINT_FPHEX_LONG_DOUBLE into a reusable macro
This patch refactors the macro PRINT_FPHEX_LONG_DOUBLE from the file
sysdeps/ieee754/ldbl-128/printf_fphex.c into a function-like macro to
enable its use for both long double and _Float128, when they are
ABI-distinct.
* sysdeps/ieee754/ldbl-128/printf_fphex.c: Include
ldbl-128/printf_fphex_macros.h for the definition of PRINT_FPHEX.
(PRINT_FPHEX_LONG_DOUBLE): Define based on PRINT_FPHEX.
* sysdeps/ieee754/ldbl-128/printf_fphex_macros.h
(PRINT_FPHEX): New function-like macro that can be used for long
double, as well as for _Float128
Remove duplicated code from __printf_fp_l, __printf_fphex, and __printf_size
In __printf_fp_l, __printf_fphex, and __printf_size the blocks of code that are
used to read a double or long double argument, check for special values and
convert to multiprecision are similar. When adding float128 support to libc,
more code would be duplicated to deal with the extra type. This patch moves
the repetitive code to a macro which is now used by double and long double and
will be used for float128 when support is added, thus avoiding more
duplication.
Tested for powerpc64le and s390x.
* stdio-common/printf_fp.c (PRINTF_FP_FETCH): New macro.
(__printf_fp_l): Use the new macro to avoid duplicating code.
* stdio-common/printf_fphex.c (PRINTF_FPHEX_FETCH): New macro.
(__printf_fphex): Use the new macro to avoid duplicating code.
* stdio-common/printf_size.c (PRINTF_SIZE_FETCH): New macro.
(__printf_size): Use the new macro to avoid duplicating code.
Reuse the code for __mpn_extract_long_double to implement
__mpn_extract_float128.
* include/gmp.h: Include bits/floatn.h
(__mpn_extract_float128): Declare when __HAVE_DISTINCT_FLOAT128 is 1.
* stdlib/gmp-impl.h: Also check if alloca is not defined before
including stack-alloc.h. It could have been defined by other header
which not necessarily defines HAVE_ALLOCA.
* sysdeps/ieee754/float128/Makefile: New file.
* sysdeps/ieee754/float128/float1282mpn.c: New file.
* sysdeps/ieee754/float128/float128_private.h: Include gmp.h before
redefining __mpn_extract_long_double to __mpn_extract_float128, then
redefine __mpn_extract_long_double to __mpn_extract_float128.
* sysdeps/ieee754/ldbl-128/ldbl2mpn.c: Replace long double with
_Float128 to allow float128_private.h overrides.
Paul E. Murphy [Fri, 4 Nov 2016 11:57:27 +0000 (09:57 -0200)]
float128: Extend the power of ten tables
Update the power of ten tables used by the common implementation when long
double is not the most expressive real type.
* stdlib/fpioconst.h: Include bits/floatn.h.
(FPIOCONST_HAVE_EXTENDED_RANGE): New macro for testing how big the
power of ten table should be.
(FPIOCONST_POW10_ARRAY_SIZE): Use larger table if above is true.
* stdlib/fpioconst.c (__tens): Use FPIOCONST_HAVE_EXTENDED_RANGE
to include larger tables when _Float128 support is enabled.
(_fpioconst_pow10): Likewise.
Joseph Myers [Wed, 7 Jun 2017 17:59:50 +0000 (17:59 +0000)]
Fix include paths in include/bits/types/*.h.
Various include/bits/types/*.h files do
where the path specified is relative to the toplevel glibc source
directory.
That has the wrong number of ../ components to achieve the desired
effect; it actually searches relative to include/ for a file that does
not exist there, then goes on to search the #include <> paths
specified with -I, eventually finding the desired file via such a path
(e.g. sysdeps/nptl/) with the right number of directory components.
Before that it searches include/../.. because of the -Iinclude,
meaning that an appropriately named file outside the glibc source tree
can affect the build.
This patch changes all those files to do #include <path> without the
../../, as some such files already do.
Joseph Myers [Wed, 7 Jun 2017 17:14:28 +0000 (17:14 +0000)]
Fix tst-timezone race (bug 14096).
I observed a failure of timezone/tst-timezone with the same symptoms
as reported in bug 14096.
I was unable to reproduce the failure on testing again. However, from
inspection, the cause is as follows. tst-timezone uses time zones
compiled from the northamerica file, but has no dependency on any such
zones, so with a parallel build it is possible that they do not get
compiled until after tst-timezone is run. This patch adds a
dependency on the compiled America/New_York zone (the one used as a
makefile target to cause such zones to be compiled, rather than one
which is actually used in that test) to fix the race.
Tested for x86_64.
[BZ #14096]
* timezone/Makefile ($(objpfx)tst-timezone.out): Depend on
America/New_York.
Arjun Shankar [Mon, 29 May 2017 14:12:31 +0000 (16:12 +0200)]
Remove check for NULL buffer passed to `ptsname_r'
`ptsname_r' is declared in stdlib.h to only accept a `nonnull'
second argument and therefore GCC may choose to make optimizations
based on the assumption that this argument is NULL. This means
that potentially, GCC can optimize away the NULL check at some
point in the future. Since this is a programming interface, we
might as well remove the NULL check ourselves.
This also warrants a change to the `ptsname_r' manual page that
must be submitted to the corresponding mailing list.
In addition, remove the NULL buffer test in login/tst-ptsname.c.
Add support for routines in dl-procinfo.h to show string versions of
HWCAP entries when a program is invoked with the LD_SHOW_AUXV
environment variable set and also to aid in path resolution for
ldconfig.
The LD_HWCAP_MASK environment variable was ignored in static binaries,
which is inconsistent with the behaviour of dynamically linked
binaries. This seems to have been because of the inability of
ld_hwcap_mask being read early enough to influence anything but now
that it is in tunables, the mask is usable in static binaries as well.
This feature is important for aarch64, which relies on HWCAP_CPUID
being masked out to disable multiarch. A sanity test on x86_64 shows
that there are no failures. Likewise for aarch64.
* elf/dl-hwcaps.h [HAVE_TUNABLES]: Always read hwcap_mask.
* sysdeps/sparc/sparc32/dl-machine.h [HAVE_TUNABLES]:
Likewise.
* sysdeps/x86/cpu-features.c (init_cpu_features): Always set
up hwcap and hwcap_mask.
aarch64: Allow overriding HWCAP_CPUID feature check using HWCAP_MASK
Now that LD_HWCAP_MASK (or glibc.tune.hwcap_mask) is read early enough
to influence cpu feature check in aarch64, use it to influence
multiarch selection. Setting LD_HWCAP_MASK such that it clears
HWCAP_CPUID will now disable multiarch for the binary.
HWCAP_CPUID is also now set in HWCAP_IMPORTANT so that it is set by
default. With this patch, this feature is only usable with
dyanmically linked binaries because LD_HWCAP_MASK is not read for
static binaries. A future patch fixes that.
* sysdeps/unix/sysv/linux/aarch64/cpu-features.c
(init_cpu_features): Use glibc.tune.hwcap_mask.
* sysdeps/unix/sysv/linux/aarch64/dl-procinfo.h: New file.
The dl-procinfo.h for linux/s390 and linux/i386 don't have include
guards, which causes them to fail since addition of LD_HWCAP_MASK to
tunables. Add _DL_I386_PROCINFO_H guard to avoid redefining
_dl_procinfo on multiple includes and also allow the subsequent
include of another dl-procinfo.h to work.
Verified with a build test on i686.
* sysdeps/unix/sysv/linux/i386/dl-procinfo.h: Add include
guard.
* sysdeps/unix/sysv/linux/s390/dl-procinfo.h: Likewise.
Add LD_HWCAP_MASK to tunables in preparation of it being removed from
rtld.c. This allows us to read LD_HWCAP_MASK much earlier so that it
can influence IFUNC resolution in aarch64.
This patch does not actually do anything other than read the
LD_HWCAP_MASK variable and add the tunables way to set the
LD_HWCAP_MASK, i.e. via the glibc.tune.hwcap_mask tunable. In a
follow-up patch, the _dl_hwcap_mask will be replaced with
glibc.tune.hwcap_mask to complete the transition.
The TUNABLE_SET_VALUE and family of macros (and my later attempt to
add a TUNABLE_GET) never quite went together very well because the
overall interface was not clearly defined. This patch is an attempt
to do just that.
This patch consolidates the API to two simple sets of macros,
TUNABLE_GET* and TUNABLE_SET*. If TUNABLE_NAMESPACE is defined,
TUNABLE_GET takes just the tunable name, type and a (optionally NULL)
callback function to get the value of the tunable. The callback
function, if non-NULL, is called if the tunable was externally set
(i.e. via GLIBC_TUNABLES or any future mechanism). For example:
val = TUNABLE_GET (check, int32_t, check_callback)
returns the value of the glibc.malloc.check tunable (assuming
TUNABLE_NAMESPACE is set to malloc) as an int32_t into VAL after
calling check_callback.
Likewise, TUNABLE_SET can be used to set the value of the tunable,
although this is currently possible only in the dynamic linker before
it relocates itself. For example:
TUNABLE_SET (check, int32_t, 2)
will set glibc.malloc.check to 2. Of course, this is not possible
since we set (or read) glibc.malloc.check long after it is relocated.
To access or set a tunable outside of TUNABLE_NAMESPACE, use the
TUNABLE_GET_FULL and TUNABLE_SET_FULL macros, which have the following
prototype:
In future the tunable list may get split into mutable and immutable
tunables where mutable tunables can be modified by the library and
userspace after relocation as well and TUNABLE_SET will be more useful
than it currently is. However whenever we actually do that split, we
will have to ensure that the mutable tunables are protected with
locks.
* elf/Versions (__tunable_set_val): Rename to __tunable_get_val.
* elf/dl-tunables.c: Likewise.
(do_tunable_update_val): New function.
(__tunable_set_val): New function.
(__tunable_get_val): Call CB only if the tunable was externally
initialized.
(tunables_strtoul): Replace strval with initialized.
* elf/dl-tunables.h (strval): Replace with a bool initialized.
(TUNABLE_ENUM_NAME, TUNABLE_ENUM_NAME1): Adjust names to
prevent collision.
(__tunable_set_val): New function.
(TUNABLE_GET, TUNABLE_GET_FULL): New macros.
(TUNABLE_SET, TUNABLE_SET_FULL): Likewise.
(TUNABLE_SET_VAL): Remove.
(TUNABLE_SET_VAL_WITH_CALLBACK): Likewise.
* README.tunables: Document the new macros.
* malloc/arena.c (ptmalloc_init): Adjust.
Stefan Liebler [Tue, 6 Jun 2017 07:41:56 +0000 (09:41 +0200)]
Optimize generic spinlock code and use C11 like atomic macros.
This patch optimizes the generic spinlock code.
The type pthread_spinlock_t is a typedef to volatile int on all archs.
Passing a volatile pointer to the atomic macros which are not mapped to the
C11 atomic builtins can lead to extra stores and loads to stack if such
a macro creates a temporary variable by using "__typeof (*(mem)) tmp;".
Thus, those macros which are used by spinlock code - atomic_exchange_acquire,
atomic_load_relaxed, atomic_compare_exchange_weak - have to be adjusted.
According to the comment from Szabolcs Nagy, the type of a cast expression is
unqualified (see http://www.open-std.org/jtc1/sc22/wg14/www/docs/dr_423.htm):
__typeof ((__typeof (*(mem)) *(mem)) tmp;
Thus from spinlock perspective the variable tmp is of type int instead of
type volatile int. This patch adjusts those macros in include/atomic.h.
With this construct GCC >= 5 omits the extra stores and loads.
The atomic macros are replaced by the C11 like atomic macros and thus
the code is aligned to it. The pthread_spin_unlock implementation is now
using release memory order instead of sequentially consistent memory order.
The issue with passed volatile int pointers applies to the C11 like atomic
macros as well as the ones used before.
I've added a glibc_likely hint to the first atomic exchange in
pthread_spin_lock in order to return immediately to the caller if the lock is
free. Without the hint, there is an additional jump if the lock is free.
I've added the atomic_spin_nop macro within the loop of plain reads.
The plain reads are also realized by C11 like atomic_load_relaxed macro.
The new define ATOMIC_EXCHANGE_USES_CAS determines if the first try to acquire
the spinlock in pthread_spin_lock or pthread_spin_trylock is an exchange
or a CAS. This is defined in atomic-machine.h for all architectures.
The define SPIN_LOCK_READS_BETWEEN_CMPXCHG is now removed.
There is no technical reason for throwing in a CAS every now and then,
and so far we have no evidence that it can improve performance.
If that would be the case, we have to adjust other spin-waiting loops
elsewhere, too! Using a CAS loop without plain reads is not a good idea
on many targets and wasn't used by one. Thus there is now no option to
do so.
Architectures are now using the generic spinlock automatically if they
do not provide an own implementation. Thus the pthread_spin_lock.c files
in sysdeps folder are deleted.
H.J. Lu [Mon, 5 Jun 2017 23:20:00 +0000 (16:20 -0700)]
x86: Don't use dl_x86_cpu_features in cacheinfo.c
Since cpu_features is available, use it instead of dl_x86_cpu_features.
* sysdeps/x86/cacheinfo.c (intel_check_word): Accept cpu_features
and use it instead of dl_x86_cpu_features.
(handle_intel): Replace maxidx with cpu_features. Pass
cpu_features to intel_check_word.
(__cache_sysconf): Pass cpu_features to handle_intel.
(init_cacheinfo): Likewise. Use cpu_features instead of
dl_x86_cpu_features.
H.J. Lu [Mon, 5 Jun 2017 19:52:41 +0000 (12:52 -0700)]
x86-64: Optimize memcmp/wmemcmp with AVX2 and MOVBE
Optimize x86-64 memcmp/wmemcmp with AVX2. It uses vector compare as
much as possible. It is as fast as SSE4 memcmp for size <= 16 bytes
and up to 2X faster for size > 16 bytes on Haswell and Skylake. Select
AVX2 memcmp/wmemcmp on AVX2 machines where vzeroupper is preferred and
AVX unaligned load is fast.
NB: It uses TZCNT instead of BSF since TZCNT produces the same result
as BSF for non-zero input. TZCNT is faster than BSF and is executed
as BSF if machine doesn't support TZCNT.
Key features:
1. For size from 2 to 7 bytes, load as big endian with movbe and bswap
to avoid branches.
2. Use overlapping compare to avoid branch.
3. Use vector compare when size >= 4 bytes for memcmp or size >= 8
bytes for wmemcmp.
4. If size is 8 * VEC_SIZE or less, unroll the loop.
5. Compare 4 * VEC_SIZE at a time with the aligned first memory area.
6. Use 2 vector compares when size is 2 * VEC_SIZE or less.
7. Use 4 vector compares when size is 4 * VEC_SIZE or less.
8. Use 8 vector compares when size is 8 * VEC_SIZE or less.
* sysdeps/x86/cpu-features.h (index_cpu_MOVBE): New.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
memcmp-avx2 and wmemcmp-avx2.
* sysdeps/x86_64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Test __memcmp_avx2 and __wmemcmp_avx2.
* sysdeps/x86_64/multiarch/memcmp-avx2.S: New file.
* sysdeps/x86_64/multiarch/wmemcmp-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/memcmp.S: Use __memcmp_avx2 on AVX
2 machines if AVX unaligned load is fast and vzeroupper is
preferred.
* sysdeps/x86_64/multiarch/wmemcmp.S: Use __wmemcmp_avx2 on AVX
2 machines if AVX unaligned load is fast and vzeroupper is
preferred.
Since the ERMS versions of wmemset requires "rep stosl" instead of
"rep stosb", only the vector store stubs of SSE2/AVX2/AVX512 wmemset
are added. The SSE2 wmemset is about 3X faster and the AVX2 wmemset
is about 6X faster on Haswell.
H.J. Lu [Mon, 5 Jun 2017 14:58:11 +0000 (07:58 -0700)]
x86-64: Update strlen.S to support wcslen/wcsnlen
The difference between strlen and wcslen is byte vs int. We can
replace pminub and pcmpeqb with pminud and pcmpeqd to turn strlen
into wcslen.
* sysdeps/x86_64/strlen.S (PMINU): New.
(PCMPEQ): Likewise.
(SHIFT_RETURN): Likewise.
(FIND_ZERO): Replace pcmpeqb with PCMPEQ.
(strlen): Add SHIFT_RETURN before ret. Replace pcmpeqb and
pminub with PCMPEQ and PMINU.
* sysdeps/x86_64/wcsnlen.S: New file.
H.J. Lu [Mon, 5 Jun 2017 14:41:14 +0000 (07:41 -0700)]
x86_64: Remove redundant REX bytes from memrchr.S
By x86-64 specification, 32-bit destination registers are zero-extended
to 64 bits. There is no need to use 64-bit registers when only the lower
32 bits are non-zero. Also 2 instructions in:
mov %rdi, %rcx
and $15, %rcx
jz L(length_less16_offset0)
mov %rdi, %rcx <<< redundant
and $15, %rcx <<< redundant
are redundant.
* sysdeps/x86_64/memrchr.S (__memrchr): Use 32-bit registers for
the lower 32 bits. Remove redundant instructions.
H.J. Lu [Mon, 5 Jun 2017 14:21:44 +0000 (07:21 -0700)]
x86-64: Update LO_HI_LONG for p{readv,writev}{64}v2
The kernel interface for p{readv,writev}{64}v is
(unsigned long fd, {const }struct iovec *iov, unsigned long vlen,
unsigned long pos_l, unsigned long pos_h)
Except for targets which define __ARCH_WANT_COMPAT_SYS_PREADV64 and
__ARCH_WANT_COMPAT_SYS_PWRITEV64,
(unsigned long fd, {const }struct iovec *iov, unsigned long vlen,
off64_t pos)
is used for p{readv,writev}{64}v. X32 is the only such target. The
LO_HI_LONG macro is used to pass offset to the pos_l and pos_h pair.
Since pos_h is ignored when size of offset == sizeof of pos_l, x86-64
has
#define LO_HI_LONG(val) (val)
But the kernel interface for p{readv,writev}{64}v2 is
(unsigned long fd, {const }struct iovec *iov, unsigned long vlen,
unsigned long pos_l, unsigned long pos_h, int flags)
Except for targets which define __ARCH_WANT_COMPAT_SYS_PREADV64V2 and
__ARCH_WANT_COMPAT_SYS_PWRITEV64V2,
(unsigned long fd, {const }struct iovec *iov, unsigned long vlen,
off64_t pos, int flags)
is used for p{readv,writev}{64}v2. X32 is the only such target. Update
x86-64 LO_HI_LONG to pass 0 as the high part of the offset argument for
p{readv,writev}{64}v2 and define a different LO_HI_LONG for x32 to only
pass one argument for offset.
Tested on x32 and x86-64.
* sysdeps/unix/sysv/linux/x86_64/sysdep.h (LO_HI_LONG): Pass
0 as the high part of offset.
* sysdeps/unix/sysv/linux/x86_64/x32/sysdep.h (LO_HI_LONG): New.
Joseph Myers [Mon, 5 Jun 2017 10:19:03 +0000 (10:19 +0000)]
Define SIG_HOLD for XPG4 (bug 21538).
Various bits/signum.h headers define SIG_HOLD if __USE_UNIX98. That
should be __USE_XOPEN, as this macro is in XPG4. This patch fixes the
conditionals accordingly. Because of other header bugs, this does not
allow any XFAILs to be removed (however, the XPG4/signal.h/conform
XFAIL only depends on a few such straightforward header bugs, not on
the more complicated to fix ucontext_t issues, as ucontext_t isn't
included in signal.h in XPG4).
Joseph Myers [Mon, 5 Jun 2017 10:17:46 +0000 (10:17 +0000)]
Fix struct sigaltstack namespace (bug 21517).
glibc defines the stack_t type with the tag struct sigaltstack. This
is not permitted by POSIX; sigaltstack is only reserved with file
scope in the namespace of ordinary identifiers, not the tag namespace,
and in the case where stack_t is obtained from ucontext.h rather than
signal.h, it's not reserved with file scope at all.
This patch removes the tag accordingly and updates uses in glibc of
struct sigaltstack. This is similar to the removal of the "struct
siginfo" tag a few years ago: C++ name mangling changes are an
unavoidable consequence. A NEWS item is added to note the changed
mangling. There is inevitably some risk of breaking builds of
anything that relies on the struct sigaltstack name (though the first
few hits I looked at from codesearch.debian.net generally seemed to
involve code that could use the stack_t name conditionally, so
depending on how they determine the conditionals they may work with
glibc not defining the struct tag anyway).
Tested for x86_64 and x86, and with build-many-glibcs.py.
H.J. Lu [Sun, 4 Jun 2017 16:44:48 +0000 (09:44 -0700)]
benchtests: Add more tests for memrchr
bench-memchr.c is shared with bench-memrchr.c. This patch adds some
tests for positions close to the beginning for memrchr, which are
equivalent to positions close to the end for memchr.
* benchtests/bench-memchr.c (do_test): Print out both length
and position.
(test_main): Also test the position close to the beginning for
memrchr.
Zack Weinberg [Sun, 4 Jun 2017 15:14:43 +0000 (11:14 -0400)]
Include shlib-compat.h in many sunrpc/nis source files.
Every file that uses libc_hidden_nolink_sunrpc or
libnsl_hidden_nolink_def needs to include shlib-compat.h. Currently,
most of them are getting it via stdio.h, because libio.h refers to
SHLIB_COMPAT when _LIBC is defined, so it includes shlib-compat.h. My
experimental patch to not install libio.h breaks that chain; stdio.h
no longer pulls in libio.h even for internal users.
Accordingly, this patch adds #include <shlib-compat.h> to many files
in sunrpc/ and nis/. There are also a small number of really obvious
fixups to includes that caught my eye while proofreading the patch -
not including headers twice in a row, not worrying about portability
to Ultrix anymore, sort of thing.
* sunrpc/des_crypt.c, sunrpc/des_soft.c: No need to include
abi-versions.h as well as shlib-compat.h.
* sunrpc/get_myaddr.c: Remove obsolete comment.
* sunrpc/pmap_rmt.c: Remove obsolete comment and #undef.
* sunrpc/rpc_thread.c: Include libc-lock.h only once.
* resolv/res_libc.c: Include shlib-compat.h only once.
H.J. Lu [Sat, 3 Jun 2017 00:32:21 +0000 (17:32 -0700)]
x86: Update __x86_shared_non_temporal_threshold
__x86_shared_non_temporal_threshold was set to 6 times of per-core
shared cache size, based on the large memcpy micro benchmark in glibc
on a 8-core processor. For a processor with more than 8 cores, the
threshold is too low. Set __x86_shared_non_temporal_threshold to the
3/4 of the total shared cache size so that it is unchanged on 8-core
processors. On processors with less than 8 cores, the threshold is
lower.
* sysdeps/x86/cacheinfo.c (__x86_shared_non_temporal_threshold):
Set to the 3/4 of the total shared cache size.
Rical Jasan [Thu, 1 Jun 2017 09:07:01 +0000 (02:07 -0700)]
manual: Provide consistent errno documentation.
The @errno macro is extended to render the canonical error string in
every documented errno error. Redundant entries and "???" are
removed. Sixty-six errors now at least contain the error string as
the description, where no description (or "???") existed before.
* manual/errno.texi: Remove redundant error strings.
* manual/macros.texi (@errno): Render the error string in
every description.
Joseph Myers [Fri, 2 Jun 2017 19:47:03 +0000 (19:47 +0000)]
Fix sigevent namespace (bug 21543).
signal.h defines the sigevent structure and constants if defined
__USE_POSIX199309 || defined __USE_XOPEN_EXTENDED. The
__USE_XOPEN_EXTENDED condition is incorrect; this structure does not
appear in XSI standards before Unix98 (which implies
__USE_POSIX199309). This patch fixes the conditionals accordingly
(the existing conditional is correct for siginfo_t, just not for
sigevent).
Tested for x86_64. This does not allow any conform/ XFAILs to be
removed because of other signal.h namespace bugs.
[BZ #21543]
* signal/signal.h: Only include <bits/types/sigevent_t.h> and
<bits/sigevent-consts.h> if [__USE_POSIX199309], not if
[__USE_XOPEN_EXTENDED].
posix: Add missing build flags for p{write,read}v2
This patch adds the missing compiler flags for correct pthread cancellation
on some architectures for the p{read,write}v2 implementation (52bd9381692fd23).
Checked on x86_64-linux-gnu and powerpc64le-linux-gnu.
Florian Weimer [Fri, 2 Jun 2017 12:54:56 +0000 (14:54 +0200)]
getaddrinfo: Always allocate canonical name on the heap
A further simplification could eliminate the canon variable in
gaih_inet and replace it with canonbuf. However, canonbuf is
used as a flag in the nscd code, which makes this somewhat
non-straightforward.
Florian Weimer [Fri, 2 Jun 2017 09:59:28 +0000 (11:59 +0200)]
Add internal facility for dynamic array handling
This is intended as a type-safe alternative to obstacks and
hand-written realloc constructs. The implementation avoids
writing function pointers to the heap.
Include sys/param.h in stdlib/gmp-impl.h instead of redefining MAX/MIN
In stdlib/gmp-impl.h, the macros MAX and MIN are defined exactly the same
as in sys/param.h. This patch removes the redefinition and makes
gmp-impl.h include sys/param.h instead.
Tested for powerpc64le and s390x.
* stdlib/gmp-impl.h: Include sys/param.h instead of redefining the
macros MAX and MIN.
Joseph Myers [Thu, 1 Jun 2017 17:17:43 +0000 (17:17 +0000)]
conformtest: Correct signal.h expectations for XPG4 / XPG42.
Various of the signal.h conform/ failures for XPG4 and XPG42 are
actually the result of incorrect test expectations rather than header
bugs. This patch fixes the expectations to accord with those
standards (this does not however allow any XFAILs to be removed, as
some header bugs remain). Note for anyone comparing with the
standards: corrigendum U013/16 removes the mention of a sigmask
function in signal.h (C435 lists such a function in the definition of
signal.h, but without any actual specification for the function
itself), so sigmask is not included in the expectations.
Tested for x86_64.
* conform/data/signal.h-data (sa_sigaction): Do not expect for
[XPG4].
(SA_SIGINFO): Likewise.
(SA_ONSTACK): Likewise.
(SA_RESETHAND): Likewise.
(SA_RESTART): Likewise.
(SA_NOCLDWAIT): Likewise.
(SA_NODEFER): Likewise.
(SS_ONSTACK): Likewise.
(SS_DISABLE): Likewise.
(MINSIGSTKSZ): Likewise.
(SIGSTKSZ): Likewise.
(ucontext_t): Likewise.
(stack_t): Likewise.
(struct sigstack): Likewise.
(SI_USER): Do not expect for [XPG4 || XPG42].
(SI_QUEUE): Likewise.
(SI_TIMER): Likewise.
(SI_ASYNCIO): Likewise.
(SI_MESGQ): Likewise.
(bsd_signal): Do not expect for [XPG4].
(killpg): Likewise.
(sigaltstack): Likewise.
(sighold): Likewise.
(sigignore): Likewise.
(siginterrupt): Likewise.
(sigpause): Likewise.
(sigrelse): Likewise.
(sigset): Likewise.
(sigwait): Do not expect for [XPG4 || XPG42].
Zack Weinberg [Thu, 1 Jun 2017 14:56:10 +0000 (10:56 -0400)]
Add shim header for bits/syscall.h.
On Linux-based configurations, bits/syscall.h is a generated file.
To avoid build-ordering problems, the Linux sys/syscall.h only includes
bits/syscall.h if _LIBC is not defined. After the _ISOMAC-testsuite
changes, this means any test case that includes sys/syscall.h tries to
pull in bits/syscall.h. This would be fine, because it'll definitely
have been generated by the time we start compiling tests, except that
the generated <builddir>/misc/bits/syscall.h is not visible in the
include path, because nothing needed it till now. So we either get
the bits/syscall.h from the host system, or the build fails.
The fix is simple: add a shim header for bits/syscall.h. I put it in
sysdeps/unix/sysv/linux/include instead of the top-level include/
because bits/syscall.h doesn't exist at all on other configurations as
far as I can tell.
This is known to affect nptl/tst-cond2[45]. Thanks to John David
Anglin for noticing the problem.
[BZ #21514]
* sysdeps/unix/sysv/linux/include/bits/syscall.h: New shim header
pointing to the generated file in <builddir>/misc/bits/syscall.h.
Joseph Myers [Thu, 1 Jun 2017 14:07:40 +0000 (14:07 +0000)]
Fix more namespace issues in sys/ucontext.h (bug 21457).
Continuing the fixes for namespace issues in sys/ucontext.h, this
patch moves various symbols into the implementation namespace in the
absence of __USE_MISC. As with previous changes, it is nonexhaustive,
just covering more straightforward cases.
Structure fields are generally changed to have a prefix __ in the
absence of __USE_MISC, via a macro __ctx (used without a space before
the open parenthesis, since the result is a single identifier).
Various macros such as NGREG also have leading __ added. No changes
are made to structure tags (and thus to C++ name mangling), except
that in the (unused) file sysdeps/i386/sys/ucontext.h, structures
defined inside other structures as the type for a field have their
tags removed in the non-__USE_MISC case (those structure tags would
not in any case have been visible in C++, because in C++ the scope of
such a tag is limited to the containing structure). No changes are
made to the contents of bits/sigcontext.h, or to whether it is
included. Because of remaining namespace issues, this patch does not
yet fix the bug or allow any XFAILs to be removed.
Tested for x86_64 and x86, and with build-many-glibcs.py.
[BZ #21457]
* sysdeps/arm/sys/ucontext.h (NGREG): Rename to __NGREG and define
NGREG to __NGREG if [__USE_MISC].
(gregset_t): Define using __NGREG.
(__ctx): New macro.
(mcontext_t): Use __ctx in defining fields.
* sysdeps/i386/sys/ucontext.h (NGREG): Rename to __NGREG and
define NGREG to __NGREG if [__USE_MISC].
(gregset_t): Define using __NGREG.
(__ctx): New macro.
(__ctxt): Likewise.
(fpregset_t): Use __ctx and __ctxt in defining fields.
(mcontext_t): Likewise.
* sysdeps/m68k/sys/ucontext.h (NGREG): Rename to __NGREG and
define NGREG to __NGREG if [__USE_MISC].
(gregset_t): Define using __NGREG.
(__ctx): New macro.
(mcontext_t): Use __ctx in defining fields.
* sysdeps/mips/sys/ucontext.h (NGREG): Rename to __NGREG and
define NGREG to __NGREG if [__USE_MISC].
(gregset_t): Define using __NGREG.
(__ctx): New macro.
(fpregset_t): Use __ctx in defining fields.
(mcontext_t): Likewise.
* sysdeps/unix/sysv/linux/alpha/sys/ucontext.h (NGREG): Rename to
__NGREG and define NGREG to __NGREG if [__USE_MISC].
(gregset_t): Define using __NGREG.
(NFPREG): Rename to __NFPREG and define NFPREG to __NFPREG if
[__USE_MISC].
(fpregset_t): Define using __NFPREG.
* sysdeps/unix/sysv/linux/m68k/sys/ucontext.h (NGREG): Rename to
__NGREG and define NGREG to __NGREG if [__USE_MISC].
(gregset_t): Define using __NGREG.
(__ctx): New macro.
(fpregset_t): Use __ctx in defining fields.
(mcontext_t): Likewise.
* sysdeps/unix/sysv/linux/mips/sys/ucontext.h (NGREG): Rename to
__NGREG and define NGREG to __NGREG if [__USE_MISC].
(NFPREG): Rename to __NFPREG and define NFPREG to __NFPREG if
[__USE_MISC].
(gregset_t): Define using __NGREG.
(__ctx): New macro.
(fpregset_t): Use __ctx in defining fields.
(mcontext_t): Likewise.
* sysdeps/unix/sysv/linux/nios2/sys/ucontext.h (__ctx): New macro.
(mcontext_t): Use __ctx in defining fields.
* sysdeps/unix/sysv/linux/powerpc/sys/ucontext.h (__ctx): New
macro.
[__WORDSIZE == 32] (NGREG): Rename to __NGREG and define NGREG to
__NGREG if [__USE_MISC].
[__WORDSIZE == 32] (gregset_t): Define using __NGREG.
[__WORDSIZE == 32] (fpregset_t): Use __ctx in defining fields.
(mcontext_t): Likewise.
[__WORDSIZE != 32] (NGREG): Rename to __NGREG and define NGREG to
__NGREG if [__USE_MISC].
[__WORDSIZE != 32] (NFPREG): Rename to __NFPREG and define NFPREG
to __NFPREG if [__USE_MISC].
[__WORDSIZE != 32] (NVRREG): Rename to __NVRREG and define NVRREG
to __NVRREG if [__USE_MISC].
[__WORDSIZE != 32] (gregset_t): Define using __NGREG.
[__WORDSIZE != 32] (fpregset_t): Define using __NFPREG.
[__WORDSIZE != 32] (vscr_t): Use __ctx in defining fields.
[__WORDSIZE != 32] (vrregset_t): Likewise.
[__WORDSIZE != 32] (mcontext_t): Likewise.
* sysdeps/unix/sysv/linux/s390/sys/ucontext.h (__ctx): New macro.
(__psw_t): Use __ctx in defining fields.
(NGREG): Rename to __NGREG and define NGREG to __NGREG if
[__USE_MISC].
(gregset_t): Define using __NGREG.
(fpreg_t): Use __ctx in defining fields.
(fpregset_t): Likewise.
(mcontext_t): Likewise.
* sysdeps/unix/sysv/linux/sh/sys/ucontext.h (NGREG): Rename to
__NGREG and define NGREG to __NGREG if [__USE_MISC].
(gregset_t): Define using __NGREG.
(NFPREG): Rename to __NFPREG and define NFPREG to __NFPREG if
[__USE_MISC].
(fpregset_t): Define using __NFPREG.
(__ctx): New macro.
(mcontext_t): Use __ctx in defining fields.
* sysdeps/unix/sysv/linux/x86/sys/ucontext.h (__ctx): New macro.
[__x86_64__] (NGREG): Rename to __NGREG and define NGREG to
__NGREG if [__USE_MISC].
[__x86_64__] (gregset_t): Define using __NGREG.
[__x86_64__] (struct _libc_fpxreg): Use __ctx in defining fields.
[__x86_64__] (struct _libc_fpstate): Likewise.
[__x86_64__] (mcontext_t): Likewise.
[!__x86_64__] (NGREG): Rename to __NGREG and define NGREG to
__NGREG if [__USE_MISC].
[!__x86_64__] (gregset_t): Define using __NGREG.
[!__x86_64__] (struct _libc_fpreg): Use __ctx in defining fields.
[!__x86_64__] (struct _libc_fpstate): Likewise.
[!__x86_64__] (mcontext_t): Likewise.
Zack Weinberg [Wed, 31 May 2017 01:05:04 +0000 (21:05 -0400)]
Avoid tickling a linker bug from microblaze pt-vfork.S.
libpthread used to have its own vfork implementation that differed
from libc's only in having a pointless micro-optimization. There is
no longer any use to having a separate copy in libpthread, but the
historical ABI requires a compatibility shim. microblaze was trying
to be slightly too clever about how it did this, and tickled a linker
bug. The linker bug should get fixed eventually, but there's no
reason for us to keep tickling it in the meantime.
This doesn't reuse the generic pt-vfork.c because microblaze doesn't
have IFUNC support yet, and it doesn't reuse aarch64/pt-vfork.c
because that fails to generate a tailcall (with GCC 7.1.1).
* sysdeps/unix/sysv/linux/microblaze/pt-vfork.S: Don't include
alpha/pt-vfork.S. Provide own compat shim for vfork and __vfork.
* sysdeps/unix/sysv/linux/microblaze/vfork.S: Add __libc_vfork alias.
* sysdeps/unix/sysv/linux/microblaze/localplt.data:
libpthread.so no longer references __errno_location.
This patch adds support of preadv2 and pwritev2 which are similar to
preadv/pwritev but with an extra flag argument. As for preadv/pwritev
both interfaces are added a non-standard GNU API.
For default 'posix' implementation trying to emulate the Linux supported
flags is troublesome:
* We can not temporary change the file state of the O_DSYNC and O_SYNC
flags to emulate RWF_{D}SYNC (attempts to change the state of using
fcntl are silently ignored).
* IOCB_HIPRI requires the file opened in O_DIRECT and uses an internal
semantic not provided by any other flag (O_NONBLOCK for instance).
So default sysdeps/posix implementations fails with EOPNOTSUPP for any non
supported flag (which are none currently) calls generic preadv/pwritev.
Basically this implementation supports only preadv2 called as preadv (with
flags sets to 0).
The Linux one uses the preadv2/pwritev2 syscall if defined, otherwise it
call preadv/writev. Instead of using the previous __ASSUME_* to
unconditionally issue the syscall (and avoid building the fallback routine),
it call pread/write if the preadv2/pwritev2 syscalls fails. The idea
is just avoid adding another __ASSUME_* and checking each architecture
on every kernel bump and simplify code conditionals.
Checked on x86_64-linux-gnu and on i686-linux-gnu and a check with
run-built-tests=no on aarch64-linux-gnu, alpha-linux-gnu, arm-linux-gnueabihf,
ia64-linux-gnu, m68k-linux-gnu, microblaze-linux-gnu, mips{64,64n32}-linux-gnu,
nios2-linux-gnu, powerpc{64,64le}-linux-gnu, s390{x}-linux-gnu,
sparc{64,v9}-linux-gnu, tile{gx,pro}-linux-gnu, and sh4-linux-gnu (all using
gcc 6.3).
Delay initialization of CPU features struct in static binaries
Allow the CPU features structure set up to be overridden by tunables
by delaying it to until after tunables are initialized. The
initialization is already delayed in dynamically linked glibc, it is
only in static binaries that the initialization is set early to allow
it to influence IFUNC relocations that happen in libc-start. It is a
bit too early however and there is a good place between tunables
initialization and IFUNC relocations where this can be done.
Verified that this does not regress the testsuite.
* csu/libc-start.c [!ARCH_INIT_CPU_FEATURES]: Define
ARCH_INIT_CPU_FEATURES.
(LIBC_START_MAIN): Call it.
* sysdeps/unix/sysv/linux/aarch64/libc-start.c
(__libc_start_main): Remove.
(ARCH_INIT_CPU_FEATURES): New macro.
* sysdeps/x86/libc-start.c (__libc_start_main): Remove.
(ARCH_INIT_CPU_FEATURES): New macro.
Dennis Wölfing [Tue, 30 May 2017 21:26:19 +0000 (18:26 -0300)]
Add reallocarray function
The reallocarray function is an extension from OpenBSD. It is an
integer-overflow-safe replacement for realloc(p, X*Y) and
malloc(X*Y) (realloc(NULL, X*Y)). It can therefore help in preventing
certain security issues in code.
This is an updated version of a patch originally submitted by Rüdiger
Sonderfeld in May 2014 [1].
H.J. Lu [Tue, 30 May 2017 19:39:14 +0000 (12:39 -0700)]
x86_64: Remove redundant REX bytes from memchr.S
By x86-64 specification, 32-bit destination registers are zero-extended
to 64 bits. There is no need to use 64-bit registers when only the lower
32 bits are non-zero.
* sysdeps/x86_64/memchr.S (MEMCHR): Use 32-bit registers for
the lower 32 bits.
H.J. Lu [Thu, 25 May 2017 13:55:42 +0000 (06:55 -0700)]
Make __tunables_init hidden and avoid PLT
Since __tunables_init is internal to ld.so, we should mark it hidden
to avoid PLT. We should also avoid PLT when calling __tunable_set_val
within ld.so.
2017-05-25 Siddhesh Poyarekar <siddhesh@sourceware.org>
H.J. Lu <hongjiu.lu@intel.com>
* elf/dl-tunables.c (__tunable_set_val): Make a hidden alias.
* elf/dl-tunables.h (__tunables_init): Mark it hidden in rtld.
(__tunable_set_val): Likewise.
Paul E. Murphy [Mon, 8 Aug 2016 20:49:31 +0000 (15:49 -0500)]
float128: Add wrappers to override ldbl-128 as float128.
This change defines float128_private.h which contains
macros used to override long double naming conventions
when building a ldbl file.
* math/math.h [__HAVE_FLOAT128 && __GLIBC_USE (IEC_60559_TYPES_EXT)]
(SNANF128): New macro.
* math/e_sqrtf128.c: New file.
* math/s_fmaxmag_template.c: Include math_private.h in order to
make inline expansion of fabs128().
* math/s_fminmag_template.c: Likewise.
* sysdeps/ieee754/float128/e_acosf128.c: New file.
* sysdeps/ieee754/float128/e_acoshf128.c: New file.
* sysdeps/ieee754/float128/e_asinf128.c: New file.
* sysdeps/ieee754/float128/e_atan2f128.c: New file.
* sysdeps/ieee754/float128/e_atanhf128.c: New file.
* sysdeps/ieee754/float128/e_coshf128.c: New file.
* sysdeps/ieee754/float128/e_exp10f128.c: New file.
* sysdeps/ieee754/float128/e_expf128.c: New file.
* sysdeps/ieee754/float128/e_fmodf128.c: New file.
* sysdeps/ieee754/float128/e_gammaf128_r.c: New file.
* sysdeps/ieee754/float128/e_hypotf128.c: New file.
* sysdeps/ieee754/float128/e_ilogbf128.c: New file.
* sysdeps/ieee754/float128/e_j0f128.c: New file.
* sysdeps/ieee754/float128/e_j1f128.c: New file.
* sysdeps/ieee754/float128/e_jnf128.c: New file.
* sysdeps/ieee754/float128/e_lgammaf128_r.c: New file.
* sysdeps/ieee754/float128/e_log10f128.c: New file.
* sysdeps/ieee754/float128/e_log2f128.c: New file.
* sysdeps/ieee754/float128/e_logf128.c: New file.
* sysdeps/ieee754/float128/e_powf128.c: New file.
* sysdeps/ieee754/float128/e_rem_pio2f128.c: New file.
* sysdeps/ieee754/float128/e_remainderf128.c: New file.
* sysdeps/ieee754/float128/e_scalbf128.c: New file.
* sysdeps/ieee754/float128/e_sinhf128.c: New file.
* sysdeps/ieee754/float128/float128_private.h: New file.
* sysdeps/ieee754/float128/gamma_productf128.c: New file.
* sysdeps/ieee754/float128/ieee754_float128.h: New file.
* sysdeps/ieee754/float128/k_cosf128.c: New file.
* sysdeps/ieee754/float128/k_sincosf128.c: New file.
* sysdeps/ieee754/float128/k_sinf128.c: New file.
* sysdeps/ieee754/float128/k_tanf128.c: New file.
* sysdeps/ieee754/float128/lgamma_negf128.c: New file.
* sysdeps/ieee754/float128/lgamma_productf128.c: New file.
* sysdeps/ieee754/float128/s_asinhf128.c: New file.
* sysdeps/ieee754/float128/s_atanf128.c: New file.
* sysdeps/ieee754/float128/s_cbrtf128.c: New file.
* sysdeps/ieee754/float128/s_ceilf128.c: New file.
* sysdeps/ieee754/float128/s_copysignf128.c: New file.
* sysdeps/ieee754/float128/s_cosf128.c: New file.
* sysdeps/ieee754/float128/s_erff128.c: New file.
* sysdeps/ieee754/float128/s_expm1f128.c: New file.
* sysdeps/ieee754/float128/s_fabsf128.c: New file.
* sysdeps/ieee754/float128/s_finitef128.c: New file.
* sysdeps/ieee754/float128/s_floorf128.c: New file.
* sysdeps/ieee754/float128/s_fmaf128.c: New file.
* sysdeps/ieee754/float128/s_fpclassifyf128.c: New file.
* sysdeps/ieee754/float128/s_frexpf128.c: New file.
* sysdeps/ieee754/float128/s_fromfpf128.c: New file.
* sysdeps/ieee754/float128/s_fromfpxf128.c: New file.
* sysdeps/ieee754/float128/s_getpayloadf128.c: New file.
* sysdeps/ieee754/float128/s_isinff128.c: New file.
* sysdeps/ieee754/float128/s_isnanf128.c: New file.
* sysdeps/ieee754/float128/s_issignalingf128.c: New file.
* sysdeps/ieee754/float128/s_llrintf128.c: New file.
* sysdeps/ieee754/float128/s_llroundf128.c: New file.
* sysdeps/ieee754/float128/s_log1pf128.c: New file.
* sysdeps/ieee754/float128/s_logbf128.c: New file.
* sysdeps/ieee754/float128/s_lrintf128.c: New file.
* sysdeps/ieee754/float128/s_lroundf128.c: New file.
* sysdeps/ieee754/float128/s_modff128.c: New file.
* sysdeps/ieee754/float128/s_nearbyintf128.c: New file.
* sysdeps/ieee754/float128/s_nextafterf128.c: New file.
* sysdeps/ieee754/float128/s_nexttowardf128.c: New file.
* sysdeps/ieee754/float128/s_nextupf128.c: New file.
* sysdeps/ieee754/float128/s_remquof128.c: New file.
* sysdeps/ieee754/float128/s_rintf128.c: New file.
* sysdeps/ieee754/float128/s_roundevenf128.c: New file.
* sysdeps/ieee754/float128/s_roundf128.c: New file.
* sysdeps/ieee754/float128/s_scalblnf128.c: New file.
* sysdeps/ieee754/float128/s_scalbnf128.c: New file.
* sysdeps/ieee754/float128/s_setpayloadf128.c: New file.
* sysdeps/ieee754/float128/s_setpayloadsigf128.c: New file.
* sysdeps/ieee754/float128/s_signbitf128.c: New file.
* sysdeps/ieee754/float128/s_significandf128.c: New file.
* sysdeps/ieee754/float128/s_sincosf128.c: New file.
* sysdeps/ieee754/float128/s_sinf128.c: New file.
* sysdeps/ieee754/float128/s_tanf128.c: New file.
* sysdeps/ieee754/float128/s_tanhf128.c: New file.
* sysdeps/ieee754/float128/s_totalorderf128.c: New file.
* sysdeps/ieee754/float128/s_totalordermagf128.c: New file.
* sysdeps/ieee754/float128/s_truncf128.c: New file.
* sysdeps/ieee754/float128/s_ufromfpf128.c: New file.
* sysdeps/ieee754/float128/s_ufromfpxf128.c: New file.
* sysdeps/ieee754/float128/t_sincosf128.c: New file.
* sysdeps/ieee754/float128/x2y2m1f128.c: New file.
* sysdeps/ieee754/ldbl-128ibm/bits/iscanonical.h:
(__iscanonicalf128): Define as a macro.
Joseph Myers [Tue, 23 May 2017 16:47:29 +0000 (16:47 +0000)]
Fix sigstack namespace (bug 21511).
The sigstack function was removed in the 2001 edition of POSIX, and
the sigstack structure in the 2008 edition. signal.h wrongly includes
those declarations even for newer POSIX versions. This patch fixes
the conditions, and conform/ expectations, accordingly.
This patch makes the minimum change to when these declarations are
present, leaving them visible for __USE_MISC as they would previously
have been visible by default. Arguably these legacy declarations
should only be visible when an old standard is specifically requested,
but implementing that would require arranging for the various sigstack
implementations to be able to see the struct sigstack type despite it
not being in _GNU_SOURCE in that case.
Tested for x86_64.
[BZ #21511]
* signal/signal.h: Include <bits/types/struct_sigstack.h> only if
[(__USE_XOPEN_EXTENDED && !__USE_XOPEN2K8) || __USE_MISC].
(sigstack): Declare only if [(__USE_XOPEN_EXTENDED &&
!__USE_XOPEN2K) || __USE_MISC].
* conform/data/signal.h-data (struct sigstack): Expect type only
if [!XOPEN2K8 && !POSIX2008].
(sigstack): Expect function only if [XPG42 || UNIX98].
Joseph Myers [Tue, 23 May 2017 11:49:48 +0000 (11:49 +0000)]
Fix sys/ucontext.h namespace from signal.h etc. inclusion (bug 21457).
The various sys/ucontext.h headers include <signal.h> and all the
headers split out of <bits/sigstack.h>. (Except that the powerpc
version does not include <signal.h>.)
None of the standard versions defining ucontext.h require or permit
such inclusions; rather, they all say that the stack_t and sigset_t
types from signal.h are defined. This patch fixes the headers to
include just the bits/ headers for those types (and the existing
includes of bits/sigcontext.h). Since bits/types/sigset_t.h is now
being included instead of bits/types/__sigset_t.h, __sigset_t uses in
the headers are replaced by direct use of the public sigset_t type.
sysdeps/unix/sysv/linux/x86/bits/sigcontext.h was relying on the prior
inclusion of <signal.h> to define types such as __uint32_t, so gets a
bits/types.h include added to provide those types.
Although one could keep some or all of the includes under a __USE_MISC
conditional, that seems unnecessary to me, especially given the lack
of a <signal.h> include in the powerpc version meaning that portable
programs already cannot rely on such an include.
Tested for x86_64 and x86, and with build-many-glibcs.py. As with
other such fixes, more namespace issues remain so this does not permit
any XFAILs to be removed or bugs to be closed.
[BZ #21457]
* sysdeps/arm/sys/ucontext.h: Do not include <signal.h>,
<bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/generic/sys/ucontext.h: Do not include <signal.h>,
<bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/i386/sys/ucontext.h: Do not include <signal.h>,
<bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/m68k/sys/ucontext.h: Do not include <signal.h>,
<bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/mips/sys/ucontext.h: Do not include <signal.h>,
<bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/unix/sysv/linux/aarch64/sys/ucontext.h: Do not include
<signal.h>, <bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/unix/sysv/linux/alpha/sys/ucontext.h: Do not include
<signal.h>, <bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/unix/sysv/linux/arm/sys/ucontext.h: Do not include
<signal.h>, <bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/unix/sysv/linux/hppa/sys/ucontext.h: Do not include
<signal.h>, <bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/unix/sysv/linux/ia64/sys/ucontext.h: Do not include
<signal.h>, <bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h>.
* sysdeps/unix/sysv/linux/m68k/sys/ucontext.h: Do not include
<signal.h>, <bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/unix/sysv/linux/mips/sys/ucontext.h: Do not include
<signal.h>, <bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/unix/sysv/linux/nios2/sys/ucontext.h: Do not include
<signal.h>, <bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/unix/sysv/linux/powerpc/sys/ucontext.h: Do not include
<bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>.
* sysdeps/unix/sysv/linux/s390/sys/ucontext.h: Do not include
<signal.h>, <bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/unix/sysv/linux/sh/sys/ucontext.h: Do not include
<signal.h>, <bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/unix/sysv/linux/sparc/sys/ucontext.h: Do not include
<signal.h>, <bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/unix/sysv/linux/tile/sys/ucontext.h: Do not include
<signal.h>, <bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
* sysdeps/unix/sysv/linux/x86/bits/sigcontext.h: Include
<bits/types.h>.
* sysdeps/unix/sysv/linux/x86/sys/ucontext.h: Do not include
<signal.h>, <bits/sigstack.h>, <bits/types/struct_sigstack.h> or
<bits/ss_flags.h>. Include <bits/types/sigset_t.h> instead of
<bits/types/__sigset_t.h>.
(ucontext_t): Use sigset_t instead of __sigset_t.
Zack Weinberg [Tue, 23 May 2017 02:28:09 +0000 (22:28 -0400)]
Fix a bug in 'Remove __need macros from signal.h' (a992f506)
siginfo-arch.h is included in two different places, so the default
definitions of the macros that it might or might not define need to
be done conditionally afterward, not unconditionally beforehand.
* sysdeps/unix/sysv/linux/bits/siginfo-consts.h
(__SI_ASYNCIO_AFTER_SIGIO): Define default after including
bits/siginfo-arch.h, only if not already defined.
* sysdeps/unix/sysv/linux/bits/types/siginfo_t.h
(__SI_ALIGNMENT, __SI_BAND_TYPE, __SI_CLOCK_T)
(__SI_ERRNO_THEN_CODE, __SI_HAVE_SIGSYS, __SI_SEGFAULT_ADDL):
Likewise.
libio: Avoid dup already opened file descriptor [BZ#21393]
As described in BZ#21398 (close as dup of 21393) report current
freopen implementation fails when one tries to freopen STDIN_FILENO,
STDOUT_FILENO, or STDERR_FILENO. Although on bug report the
discussion leads to argue if a close followed by a freopen on the
standard file is a valid operation, the underlying issue is not
really the check for dup3 returned value, but rather calling it
if the returned file descriptor is equal as the input one.
So for a quality of implementation this patch avoid calling dup3
for the aforementioned case. It also adds a dup3 error case check
for the two possible failures, with one being Linux only: EINTR and
EBUSY. The EBUSY issue is better explained on this stackoverflow
thread [1], but in a short it is due the internal Linux
implementation which allows a race condition window for dup2 due
the logic dissociation of file descriptor allocation and actual
VFS 'install' operation. For both outliers failures all allocated
memory is freed and a NULL FILE* is returned.
With this patch the example on BZ#21398 is now actually possible
(I used as the testcase for the bug report). Checked on
x86_64-linux-gnu.
[BZ #21393]
* libio/freopen.c (freopen): Avoid dup already opened file descriptor
and add a check for dup3 failure.
* libio/freopen64.c (freopen64): Likewise.
* libio/tst-freopen.c (do_test): Rename to do_test_basic and use
libsupport.
(do_test_bz21398): New test.
* manual/stdio.texi (freopen): Add documentation of EBUSY failure.
Remove useless comment from sysdeps/sparc/sparc32/dl-machine.h
David Miller has not been shot yet AFAIK (yes, I googled for any news
that may seem relevant and I poked him on twitter some days ago) so
either nobody uses SPARC or the code is correct or nobody read the
instructions in the comment to shoot him. In all of those cases the
comment is clearly not useful, so getting rid of it.
Zack Weinberg [Fri, 2 Dec 2016 00:24:13 +0000 (19:24 -0500)]
Remove __need macros from signal.h.
The types affected are __sig_atomic_t, sig_atomic_t, __sigset_t,
sigset_t, sigval_t, sigevent_t, and siginfo_t. __sig_atomic_t is a
scalar, so it's now directly available from bits/types.h. The others
get bits/types/ headers.
Side effects include: There have been small changes to which
non-signal headers expose which subset of the signal-related types.
A couple of architectures' nested siginfo_t fields had to be renamed
to prevent undesired macro expansion. Internal code that wants to
manipulate signal masks must now include <sigsetops.h> (which is not
installed) and should be aware that __sigaddset, __sigandset,
__sigdelset, __sigemptyset, and __sigorset no longer return a value
(unlike the public API). Relatedly, the public signal.h no longer
declares any of those functions. The obsolete sigmask() macro no
longer has a system-specific definition -- in the cases where it
matters, it didn't work anyway.
New Linux architectures should create bits/siginfo-arch.h and/or
bits/siginfo-consts-arch.h to customize their siginfo_t, rather than
duplicating everything in bits/siginfo.h (which no longer exists).
Add new __SI_* macros if necessary. Ports to other operating systems
are strongly encouraged to generalize this scheme further.
* signal/Makefile, sysdeps/unix/sysv/linux/Makefile:
Update lists of installed headers.
* posix/bits/types.h: Define __sig_atomic_t here.
* signal/signal.h: Use the new bits headers; no need to handle
__need_sig_atomic_t nor __need_sigset_t. Don't use __sigmask
to define sigmask.
* include/signal.h: No need to handle __need_sig_atomic_t
nor __need_sigset_t. Don't define __sigemptyset.
* resolv/netdb.h, rt/mqueue.h: Use bits/types/sigevent_t.h.
* rt/aio.h: Use bits/types/sigevent_t.h and bits/sigevent-consts.h.
* socket/sys/socket.h: Don't include bits/sigset.h.
Zack Weinberg [Wed, 22 Mar 2017 13:55:25 +0000 (09:55 -0400)]
Remove __need_list_t and __need_res_state.
These __need macros are only used internally, by nptl/descr.h.
However, including all of resolv.h from descr.h causes build failures
due to resolv.h's dozens of pseudo-struct-field macros, some of which
collide with struct fields in NPTL internal data structures.
Similarly, including all of list.h from descr.h produces an include
cycle, atomic.h -> atomic-machine.h -> tls.h -> descr.h -> list.h ->
atomic.h, and then list.h tries to use atomic.h macros that haven't
been defined yet. So we do need mini-headers for these. In the
list.h case I called it include/list_t.h since it isn't going to be
installed.
* resolv/resolv.h: Remove __need_res_state logic.
Move definition of res_state and related constants to ...
* resolv/bits/types/res_state.h: ...this new file.
* resolv/Makefile: Install bits/types/res_state.h.
* include/bits/types/res_state.h: New wrapper.
* include/list.h: Remove __need_list_t logic.
Move definition of list_t to ...
* include/list_t.h: ...this new file.
* nptl/descr.h: Include list_t.h and bits/types/res_state.h
instead of list.h and resolv.h.
Zack Weinberg [Fri, 12 May 2017 00:36:15 +0000 (20:36 -0400)]
Remove sfi_* annotations from ARM assembly files.
This semi-mechanical patch removes all uses and definitions of the
sfi_breg, sfi_pld, and sfi_sp macros from various ARM-specific
assembly files. These were only used by NaCl.
Zack Weinberg [Fri, 12 May 2017 00:28:07 +0000 (20:28 -0400)]
Remove the bulk of the NaCl port.
The NaCl port has not been actively maintained since before the 2.25
release. The complementary GCC back-end was never contributed to GCC,
and we are given to understand that the current NaCl SDK has switched
to Clang and therefore cannot be used to build glibc anymore, so we
doubt that the port remains useful.
This commit simply removes the sysdeps/arm/nacl and sysdeps/nacl
directories and the abi-tags entry.
Remove the NaCl port.
* abi-tags: Remove .*-.*-nacl.* entry.
* sysdeps/arm/nacl: Remove directory and contents.
* sysdeps/nacl: Likewise.
Rical Jasan [Sat, 20 May 2017 05:44:25 +0000 (22:44 -0700)]
manual: Convert errno @comments to new @errno macro.
errno.texi documents error macros, their values, and error strings in
Texinfo @comments, some of which are also used for @standards. The
purpose of this commit is to separate the standards from the error
strings so that both the @standards conversion script picks up clean
@standards and the errno documentation framework is improved.
The error names, values, and messages are consolidated in a new custom
macro, @errno. It is not clear that scripts within the sources rely
on the special Texinfo @comment-based format to generate files used
throughout the library, so the definition of @errno in macros.texi now
provides a comment indicating the dependency. The dependent scripts
are updated to use @errno, which also simplifies them a bit. The
files those scripts generate were verified to be unchanged.
The @errno macro is not visibly rendered in any way at this time, but
it does use an @cindex command to add the error string to the Concept
Index, to facilitate searching on error messages.
* manual/errno.texi: Convert @comment-based errno
documentation to @errno.
* manual/macros.texi (@errno): New macro. Consolidate errors,
their values, and messages, adding the error string to the
Concept Index. Provide a warning in the comment about
external (to the manual) dependencies.
* sysdeps/gnu/errlist.awk: Use @errno instead of @comments.
* sysdeps/mach/hurd/errnos.awk: Likewise.
Joseph Myers [Fri, 19 May 2017 20:35:07 +0000 (20:35 +0000)]
Split up bits/sigstack.h.
bits/sigstack.h contains four things: the legacy struct sigstack type,
the preferred stack_t type, the SS_* enum values and macros for signal
stack sizes.
These vary in different ways between glibc configurations; in
particular, the stack sizes vary much more than any of the other
pieces. Furthermore, these pieces have different standard namespace
rules for when they should be visible (not currently visible in
conform/ results both because the relevant tests are XFAILed for
sys/ucontext.h namespace issues, and because some of the expectations
are incorrect in the same way as the headers, e.g. neither
expectations nor headers reflect that current POSIX no longer has
either the sigstack function or the sigstack structure).
To reduce duplication of identical definitions, and facilitate
namespace fixes without requiring the same feature test macro
conditions to be repeated in many versions of the same header, this
patch splits bits/sigstack.h up into four headers. It keeps the stack
size macros, while new bits/types/struct_sigstack.h,
bits/types/stack_t.h and bits/ss_flags.h are added for the other
pieces. bits/types/struct_sigstack.h is the same everywhere,
bits/types/stack_t.h has three variants different in the order of the
structure elements (generic = MIPS Linux, and other Linux), and
bits/ss_flags.h has generic and Linux variants.
This patch includes the new headers everywhere that included
<bits/sigstack.h>, so should cause no difference to what any public
header defines. Subsequent namespace fixes would then remove or
condition some of those includes.
There should be no conflicts with Zack's changes to signal.h types,
beyond the trivial conflict of both making additions to
signal/Makefile's headers list; the two patches affect disjoint sets
of types and other definitions.
Tested for x86_64 and x86, and with build-many-glibcs.py.
H.J. Lu [Fri, 19 May 2017 17:46:29 +0000 (10:46 -0700)]
x86: Optimize SSE2 memchr overflow calculation
SSE2 memchr computes "edx + ecx - 16" where ecx is less than 16. Use
"edx - (16 - ecx)", instead of satured math, to avoid possible addition
overflow. This replaces
add %ecx, %edx
sbb %eax, %eax
or %eax, %edx
sub $16, %edx
with
neg %ecx
add $16, %ecx
sub %ecx, %edx
It is the same for x86_64, except for rcx/rdx, instead of ecx/edx.
* sysdeps/i386/i686/multiarch/memchr-sse2.S (MEMCHR): Use
"edx + ecx - 16" to avoid possible addition overflow.
* sysdeps/x86_64/memchr.S (memchr): Likewise.
Fix makefile rules for vmsplice, splice, and open_by_handle_at
Commits ecade1c, afbbc18, and a5a34d2 added compiler flags for
vmsplice, splice, and open_by_handle_at respectively on default
misc/Makefile. However such symbols are build only for Linux and
the rules should on linux Makefile only.
Joseph Myers [Fri, 19 May 2017 11:30:26 +0000 (11:30 +0000)]
Also create and use ldbl-compat-choose.h.
This patch makes the glibc build generate an additional header
ldbl-compat-choose.h that defines LONG_DOUBLE_COMPAT_CHOOSE_* macros
for each libc and libm symbol, which select one or the other of their
arguments based on whether the symbol was added before a change to
long double != double.
The effect of this is that it is then possible to define a macro
maybe_long_double_symbol that automatically acts as either
long_double_symbol or weak_alias depending on when the symbol being
defined was added. This can be used when building long double
functions from type-generic templates. Thus, with this patch ldbl-opt
no longer needs special long double implementations of each new libm
function added using such a template, and the existing such
implementations are removed.
This is a step towards being able more generally to use common macros
to create all the aliases needed for a libm function, so reducing the
amount of special-case code needed in ldbl-opt and ldbl-64-128, and
facilitating subsequently adding *f32 / *f64 / *f128 / *f32x / *f64x
aliases to existing functions (where the set of aliases that a
function should have may depend on the architecture in various ways).
Tested with build-many-glibcs.py. Except for on
powerpc64le-linux-gnu, installed stripped shared libraries are
unchanged by the patch. powerpc64le-linux-gnu is the unique
configuration which used ldbl-opt from the start rather than adding a
new long double choice after originally only having had long double =
double. The effect of the patch there is that various cases that
previously used long_double_symbol unconditionally now use weak_alias
instead, so .os files contain e.g. a symbol cabsl instead of
cabsl@@GLIBC_2.17. The final dynamic symbols and versions in the
resulting shared libraries are unchanged (ABI tests pass), as is the
disassembly of the shared libraries, but the differences in the .os
files still result in different .gnu_hash contents in libm.so; the
differences are of no significance and logically using weak_alias is
what's most appropriate in those cases.
Joseph Myers [Fri, 19 May 2017 11:26:00 +0000 (11:26 +0000)]
Create and use first-versions.h with macros for function symbol versions.
This patch arranges for the glibc build to generate a header
first-versions.h that defines macros for the earliest symbol version
in which each public symbol (GLIBC_[0-9]* symbol version, name only
uses C identifier characters) is available.
This is used in sysdeps/ieee754/ldbl-opt/math-type-macros-double.h to
replace the manually defined LDOUBLE_*_libm_version macros for various
functions defined using type-generic templates, the purpose of which
is to use in LONG_DOUBLE_COMPAT tests "was this function originally
added before glibc supported long double != double on this platform?".
As discussed in
<https://sourceware.org/ml/libc-alpha/2016-12/msg00246.html>, I expect
this to be useful more generally in reducing the amount of
special-case code needed in ldbl-opt and ldbl-64-128.
Tested with build-many-glibcs.py that installed stripped shared
libraries are unchanged by this patch.
This patch consolidates the mq_timedsend Linux syscall generation
on sysdeps/unix/sysv/linux/mq_timedsend.c. It basically removes it
from architecture auto-generation list.
Checked on i686-linux-gnu, x86_64-linux-gnu, x86_64-linux-gnux32,
arch64-linux-gnu, arm-linux-gnueabihf, powerpc64le-linux-gnu,
sparc64-linux-gnu, and sparcv9-linux-gnu.
* rt/Makefile (CFLAGS-mq_timedsend.c): New flag.
* sysdeps/unix/sysv/linux/mq_timedsend.c: New file.
* sysdeps/unix/sysv/linux/syscalls.list (mq_timedsend): Remove from
auto-generation list.
This patch consolidates the mq_timedreceive Linux syscall generation
on sysdeps/unix/sysv/linux/mq_timedreceive.c. It basically removes it
from architecture auto-generation list.
Checked on i686-linux-gnu, x86_64-linux-gnu, x86_64-linux-gnux32,
arch64-linux-gnu, arm-linux-gnueabihf, powerpc64le-linux-gnu,
sparc64-linux-gnu, and sparcv9-linux-gnu.
* rt/Makefile (CFLAGS-mq_timedreceive.c): New flag.
* sysdeps/unix/sysv/linux/mq_timedreceive.c: New file.
* sysdeps/unix/sysv/linux/syscalls.list (mq_timedreceive): Remove
from auto-generation list.
linux: Consolidate Linux open_by_handle_at syscall
This patch consolidates the open_by_handle_at Linux syscall generation on
sysdeps/unix/sysv/linux/open_by_handle_at.c. It basically removes it from
architectures auto-generation list.
Checked on i686-linux-gnu, x86_64-linux-gnu, x86_64-linux-gnux32,
arch64-linux-gnu, arm-linux-gnueabihf, powerpc64le-linux-gnu,
sparc64-linux-gnu, and sparcv9-linux-gnu.
* sysdeps/unix/sysv/linux/Makefile (sysdep_routines): Add
open_by_handle_at.
(CFLAGS-open_by_handle_at.c): New flag.
* sysdeps/unix/sysv/linux/open_by_handle_at.c: New file.
* sysdeps/unix/sysv/linux/syscalls.list (open_by_handle_at): New
file.
* misc/Makefile (CFLAGS-open_by_handle_at.c): New rule.
This patch consolidates the splice Linux syscall generation on
sysdeps/unix/sysv/linux/splice.c. It basically removes it from
architectures auto-generation list.
Checked on i686-linux-gnu, x86_64-linux-gnu, x86_64-linux-gnux32,
arch64-linux-gnu, arm-linux-gnueabihf, powerpc64le-linux-gnu,
sparc64-linux-gnu, and sparcv9-linux-gnu.
* sysdeps/unix/sysv/linux/Makefile (sysdep_routines): Add splice.
(CFLAGS-splice.c): New flag.
* sysdeps/unix/sysv/linux/splice.c: New file.
* sysdeps/unix/sysv/linux/syscalls.list (splice): Remove from
auto-generation syscall list.
* misc/Makefile (CFLAGS-splice.c): New rule.
This patch consolidates the vmsplice Linux syscall generation on
sysdeps/unix/sysv/linux/vmsplice.c. It basically removes it from
architectures auto-generation list.
Checked on i686-linux-gnu, x86_64-linux-gnu, x86_64-linux-gnux32,
arch64-linux-gnu, arm-linux-gnueabihf, powerpc64le-linux-gnu,
sparc64-linux-gnu, and sparcv9-linux-gnu.
* sysdeps/unix/sysv/linux/Makefile (CFLAGS-vmsplice.c): New flag.
* sysdeps/unix/sysv/linux/syscalls.list (vmsplice): Remove from
auto-generation syscall list.
* sysdeps/unix/sysv/linux/vmsplice.c: New file.
* misc/Makefile (CFLAGS-vmsplice.c): New rule.