Commit Graph

27816 Commits

Author SHA1 Message Date
Joseph Myers
d44a052c49 Fix powerpc32 __get_clockfreq for non-power4 (bug 17263).
In my powerpc32 testing I've observed misc/test-gettimebasefreq
failing.

This is a glibc build (soft-float, though that's not relevant here)
without any --with-cpu and without any special configuration of the
default CPU for GCC either.  In particular, it's one not using
sysdeps/powerpc/powerpc32/power4/hp-timing.h (although in fact the
processor I'm using for testing is POWER4-based), so hp_timing_t is
32-bit not 64-bit.  But the VDSO call being used by
INTERNAL_VSYSCALL_NO_SYSCALL_FALLBACK is generating a 64-bit result
(high part in r3, low part in r4).  The code extracting that result,
however, expects a result of the type hp_timing_t as passed to
INTERNAL_VSYSCALL_NO_SYSCALL_FALLBACK, meaning that only r3 (= 0) is
used and the value in r4 is ignored.  This patch fixes this by always
using uint64_t as the type in INTERNAL_VSYSCALL_NO_SYSCALL_FALLBACK -
reflecting the actual ABI (unconditional in the kernel) of that VDSO
call.  This is the minimal change for this issue - no check for
overflow, no change of the type of the timebase_freq variable or the
return type of __get_clockfreq to something other than hp_timing_t
(such a change would simply move the implicit conversions to the over
callers of that function), no change to hp_timing_t itself.

Tested for powerpc32 soft float.

	[BZ #17263]
	* sysdeps/unix/sysv/linux/powerpc/get_clockfreq.c: Include
	<stdint.h>.
	(__get_clockfreq): Use uint64_t instead of hp_timing_t in
	INTERNAL_VSYSCALL_NO_SYSCALL_FALLBACK call.
2014-08-13 16:06:18 +00:00
Siddhesh Poyarekar
e44c8b0eb7 Fix comment error that Jakub pointed out but I forgot to fix 2014-08-13 18:46:41 +05:30
Siddhesh Poyarekar
508ce3acd9 Disable x87 inline functions for x86_64 and SSE [BZ #17262]
Since:

commit 409e00bd69
Author: H.J. Lu <hjl.tools@gmail.com>
Date:   Wed Jan 29 07:51:41 2014 -0800

    Disable x87 inline functions for SSE2 math

    When i386 and x86-64 mathinline.h was merged into a single mathinline.h,
    "gcc -m32" enables x87 inline functions on x86-64 even when -mfpmath=sse
    and SSE2 is enabled.  It is a regression on x86-64.  We should check
    __SSE2_MATH__ instead of __x86_64__ when disabling x87 inline functions.

gcc-3.2 is unable to correctly compile x86_64 routines for llrint
since it gets redefined.  This is because gcc 3.2 does not set
__SSE2_MATH__ for x86_64, thus exposing the duplicate definition.

The correct fix ought to be to check for both __SSE2_MATH__ and
__x86_64__ and enable those bits only when neither are defined.

Tested fix with the reproducer for
409e00bd69 as well as with gcc-3.2.
2014-08-13 14:05:16 +05:30
H.J. Lu
b4acef1ffe Replace cpuid asm statement with __cpuid_count
The compiler doesn't know that the cpuid asm statement in intel_check_word
will trash RBX.  We are lucky that it doesn't cause any problems since
RBX is also used by compiler for other purposes so that RBX is saved and
restored.  This patch replaces it with __cpuid_count.

	[BZ #17259]
	* sysdeps/x86_64/cacheinfo.c (intel_check_word): Replace cpuid
	asm statement with __cpuid_count.
2014-08-12 17:02:51 -07:00
Mike Frysinger
61b1ef6e7c detect broken linker handling of __ehdr_start
Older versions of ld on ia64 support __ehdr_start, but generate relocs
when they shouldn't.  This causes the ld.so to not run because it tries
to resolve the __ehdr_start symbol (but it's not exported).
2014-08-12 19:54:02 -04:00
Joseph Myers
898c62f488 Fix powerpc-nofpu __fe_enabled_env and __fe_nonieee_env (bug 17261).
On powerpc, floating-point environment macros are defined as pointers
to constants in the library that contain the bit-patterns of the
desired environment, instead of being magic constants cast to pointer
type.

For soft-float, the bit-patterns used for fenv_t are not laid out the
same as for hard-float.  (e500 has a third layout used; that's not an
ABI issue because these values are only meaningful within a single
process, all of whose glibc libraries must come from the same build of
glibc.)  While the __fe_dfl_env value for soft-float was appropriate
for the soft-float fenv_t representation, the other two constants had
the same bit-patterns as for hard-float.  Those bit patterns had the
effect of having exceptions already raised, causing
math/test-fenv-return to fail; this patch fixes the patterns used.
(__fe_nonieee_env also had exceptions unmasked, though they should be
masked to match hard-float semantics.  Since there is no separate
non-IEEE mode for soft-float, it's most appropriate for
__fe_nonieee_env to be the same as __fe_dfl_env; this patch makes it
an alias.)

Tested for powerpc-nofpu.

	[BZ #17261]
	* sysdeps/powerpc/nofpu/fenv_const.c (__fe_enabled_env): Change
	value to 0.
	(__fe_nonieee_env): Define as an alias for __fe_dfl_env.
2014-08-12 20:31:54 +00:00
Roland McGrath
9a8a57209a NEWS: Mention x86-64 ld.so use of Intel MPX instructions. 2014-08-12 09:49:58 -07:00
Will Newton
7fda3e6f16 Add BZ #16892 to NEWS 2014-08-12 13:01:13 +01:00
Bernard Ogden
4e75751cbb Check value of futex before updating in __lll_timedlock
2014-08-12  Bernard Ogden  <bernie.ogden@linaro.org>

	[BZ #16892]
	* sysdeps/nptl/lowlevellock.h (__lll_timedlock): Use
	atomic_compare_and_exchange_bool_acq rather than atomic_exchange_acq.
2014-08-12 12:57:27 +01:00
Sean Anderson
bb2ce41656 malloc: fix comment typo 2014-08-12 05:24:29 -04:00
Allan McRae
7950a4ee59 Regenerate libc.po 2014-08-09 17:14:35 +10:00
Mike Frysinger
f469c49f92 tst-gettext2: make setup more robust
This test should be more robust about setting up its lang dirs.
I had two completely different systems (ia64 & x86_64) get wedged
in a way where the test just kept FAILing on me due to some of the
files missing.  This probably wasn't a big deal until the recent
commit which made checking of the locale dirs more robust (for
security reasons).
2014-08-08 22:55:48 -04:00
Wilco Dijkstra
656b84c2ef This patch adds new function libc_feholdsetround_noex_aarch64_ctx, enabling
further optimization. libc_feholdsetround_aarch64_ctx now only needs to
read the FPCR in the typical case, avoiding a redundant FPSR read.
Performance results show a good improvement (5-10% on sin()) on cores with
expensive FPCR/FPSR instructions.
2014-08-07 16:29:55 +00:00
Wilco
538e9e454d Fix performance issue in misaligned strcpy. 2014-08-07 14:30:31 +00:00
Adhemerval Zanella
d4795e4a43 PowerPC: Fix termios definitions
This patch fixes the incorrect guard by __USE_MISC of struct winsize and
struct termio in powerpc termios header.  Current states leads to build
failures if the program defines _XOPEN_SOURCE, but not _DEFAULT_SOURCE
or either _BSD_SOURCE or _SVID_SOURCE.  Without any definition,
__USE_MISC will not be defined and neither the struct definitions.

This patch copies the default Linux ioctl-types.h by adjusting only the
character control field (c_cc) size in struct termio.
2014-08-06 14:31:54 -04:00
Richard Henderson
27bb6dc0db alpha: Fix exception raising from soft-fp
Use the SSI_IEEE_RAISE_EXCEPTION function as from feraiseexcept,
instead of __ieee_get+set_fp_status.  Always raise the FP exceptions
from float-to-integer conversion.
2014-08-05 13:36:36 -07:00
Richard Henderson
d3f364d3c7 alpha: Remove linux lowlevellock.h
For real this time.
2014-08-05 09:06:36 -07:00
Richard Henderson
64accb62b1 alpha: Remove linux lowlevellock.h 2014-08-05 08:31:47 -07:00
Bernard Ogden
e5292691ea MIPS - Remove mips lowlevellock.h.
* sysdeps/unix/sysv/linux/mips/lowlevellock.h: Remove file.
2014-08-04 22:12:03 +00:00
Will Newton
5b4bd2d4f0 AArch64: Remove lowlevellock.h
Remove lowlevellock.h in favour of the generic implementation. The
generic implementation was tested natively and introduces no
regressions.

ChangeLog:

2014-08-04  Will Newton  <will.newton@linaro.org>

	* sysdeps/unix/sysv/linux/aarch64/lowlevellock.h: Remove
	file.
2014-08-04 16:11:51 +01:00
Mike Frysinger
fcadaeef47 add ChangeLog for previous commit 2014-08-04 10:31:01 -04:00
Mike Frysinger
b76ae12144 ia64: define nocancel entry points in PSEUDO
The previous set of not-cancel.h headers (prior to the commit
2fbdf5339a) did not require the
arch to define nocancel entry points, so ia64 never did.
However, after the various files were merged, it became a hard
requirement for arches which mean ia64 failed to build.

Here we add dedicated entry points.  It'd be nice to merge
with the existing stubs like other arches do, but the ia64
asm does not lend itself to interleaving of functions.  If
someone has a suggestion on merging these, that'd be great,
but at least now we build & pass tests again.
2014-08-04 10:05:51 -04:00
Roland McGrath
43a94c647d IA64: Consolidate nptl/ subdirectories under linux/... 2014-08-04 10:00:52 -04:00
Roland McGrath
8f3c1fa42d IA64: Move NPTL public headers to sysdeps/ia64/nptl/. 2014-08-04 09:58:14 -04:00
Roland McGrath
ca08746f9f IA64: Define TLS_DEFINE_INIT_TP 2014-08-04 09:53:49 -04:00
Roland McGrath
9c925ff964 IA64: Consolidate NPTL/non versions of vfork 2014-08-04 09:50:41 -04:00
Roland McGrath
e641054cd8 IA64: Consolidate NPTL/non versions of clone 2014-08-04 09:06:51 -04:00
Andreas Schwab
dacdc86717 Fix missing <math_private.h> in ldbl-96 fma 2014-08-04 10:20:20 +02:00
Mike Frysinger
41f725ff29 add ChangeLog for previous commit 2014-08-03 11:12:35 -04:00
Mike Frysinger
a1d8c6215d tst-setcontext: fix style 2014-08-03 08:55:20 -04:00
Roland McGrath
277cfb315b Remove sysdeps/unix/sysv/linux/arm/lowlevellock.h 2014-08-01 18:59:48 -07:00
Richard Henderson
4896f04920 Force eval for fma implementations 2014-08-01 12:13:50 -10:00
Roland McGrath
6f64e7d75f ARM: Move PTR_MANGLE et al out of Linux-specific file. 2014-08-01 11:28:49 -07:00
Siddhesh Poyarekar
a476ac4b45 Fix -Wundef warnings in regex_internal.h 2014-08-01 14:30:43 +05:30
Siddhesh Poyarekar
78dd658a02 Check if DEBUG is defined in regex_internal.c
The DEBUG macro is checked for its value in one place and if it is
defined in another.  Make this consistent across the two cases and use
the same style that we did in mktime.c, which is to check if the macro
is defined and it is set.
2014-08-01 14:24:41 +05:30
Siddhesh Poyarekar
3f3dd810e0 Fix -Wundef warnings in fnmatch.c
Fixes -Wundef warnings for MBSTATE_T and HAVE_STRING_H.  The
HAVE_STRING_H bit is a merge from the gnulib fnmatch.c and the
MBSTATE_T fix has been posted to gnulib for inclusion.
2014-08-01 14:23:38 +05:30
Stefan Liebler
95ee7fb13b NEWS: Explain the s390 jmp_buf / ucontext_t ABI change reversal. 2014-08-01 09:49:31 +02:00
Roland McGrath
f61a113fe0 Add __safe_fatal and use it in __pthread_unwind forwarder fallback. 2014-07-31 14:00:33 -07:00
Roland McGrath
9fe7e787ad Use __builtin_trap for ABORT_INSTRUCTION. 2014-07-31 13:07:19 -07:00
Roland McGrath
23fe486beb Split nptl-signals.h out from pthreadP.h 2014-07-31 11:37:48 -07:00
Stefan Liebler
f349489e7e [PATCH] S390: Fix remaining ONE_DIRECTION warning messages
This patch fixes the remaining ONE_DIRECTION warnings for s390 specific conversions.
It defines ONE_DIRECTION to 0 like the patch from Steve Ellcey:
https://www.sourceware.org/ml/libc-alpha/2014-05/msg00039.html

    Changelog:
    * sysdeps/s390/s390-64/utf16-utf32-z9.c
    (ONE_DIRECTION): Define.
    * sysdeps/s390/s390-64/utf8-utf16-z9.c
    (ONE_DIRECTION): Define.
    * sysdeps/s390/s390-64/utf8-utf32-z9.c
    (ONE_DIRECTION): Define.
2014-07-31 20:10:47 +02:00
Stefan Liebler
2f438e20ab S/390: Revert the jmp_buf/ucontext_t ABI change. 2014-07-31 20:04:54 +02:00
Andreas Schwab
6c9578a24b Fix -Wundef warning for HAVE_IFUNC 2014-07-31 17:05:19 +02:00
Ling Ma
05f3633da4 Improve 64bit memcpy performance for Haswell CPU with AVX instruction
In this patch we take advantage of HSW memory bandwidth, manage to
reduce miss branch prediction by avoiding using branch instructions and
force destination to be aligned with avx instruction.

The CPU2006 403.gcc benchmark indicates this patch improves performance
from 2% to 10%.
2014-07-30 08:02:35 -07:00
Adhemerval Zanella
a53fbd8e6c PowerPC: Fix gprof entry point for LE
This patch fixes the ELFv2 gprof entry point since the ABI
does not define function descriptors.  It fixes BZ#17213.
2014-07-30 09:01:25 -03:00
Jeff Layton
0961f7e1e3 fcntl-linux.h: add new definitions and manual updates for open file description locks
Open file description locks have been merged into the Linux kernel for
v3.15.  Add the appropriate command-value definitions and an update to
the manual that describes their usage.
2014-07-29 23:27:21 -04:00
Allan McRae
cc5fb300fc Update Spanish translations 2014-07-23 17:06:01 +10:00
Allan McRae
9bbc09c85d Add fixed bug to NEWS
Fixed in commit 7ee03f00.
2014-07-23 16:55:16 +10:00
Allan McRae
8125aedc0e Add missing changelog entry for commit ab7ac0f2 2014-07-23 16:52:13 +10:00
Marcus Shawcroft
33ef2f0c76 Revert "aarch64: Add hp-timing.h"
This reverts commit 4052993954.

Conflicts:
	sysdeps/aarch64/hp-timing.h
2014-07-22 12:09:44 +01:00