gmane.comp.lib.glibc.alpha

Remove __ASSUME_TRUNCATE64_SYSCALL

Joseph S. Myers — 2012-05-26T20:35:47

__ASSUME_TRUNCATE64_SYSCALL is used in:

sysdeps/unix/sysv/linux/ftruncate64.c
sysdeps/unix/sysv/linux/powerpc/powerpc32/ftruncate64.c
sysdeps/unix/sysv/linux/powerpc/powerpc32/truncate64.c
sysdeps/unix/sysv/linux/truncate64.c

None of these files is #included from anywhere else.
sysdeps/unix/sysv/linux/wordsize-64 overrides these files, as, in
ports, does MIPS, so they are only ever used by 32-bit Linux
architectures.

__ASSUME_TRUNCATE64_SYSCALL is defined (for all supported kernel
versions) for i386, sparc (32-bit), powerpc (32-bit) and sh (so the
only 32-bit libc architecture not defining it is s390-32).  In ports,
it is defined for all supported kernel versions for arm, hppa and
tile, and for 2.4.12 and later for m68k.  So the only 32-bit ports
architectures not defining it for all supported kernel versions, but
using files testing the macro, are am33 and m68k.

Examining 2.4.0-test1 sources shows that both s390 and m68k had the
syscall there, and examining 2.6.25 shows that when the mn10300/am33
port w

[PATCH] Avoid symbol versions in rtld syscalls

Richard Henderson — 2012-05-26T16:34:46

This is the root cause of the incorrect symbols being exported from
ld.so on alpha.  I had a look at ld about honoring the version-map
file above the symver symbols in the object file, and it isn't an
easy fix.  Whereas arranging for these symbols not to be defined in
the rtld-foo.os files is fairly trivial.

Sanity tested on x86_64.  Ok?


r~
* include/libc-symbols.h (nonrtld_default_symbol_version): New.
(nonrtld_symbol_version): New.
* sysdeps/unix/make-syscalls.sh: Use them.


diff --git a/include/libc-symbols.h b/include/libc-symbols.h
index 12b3639..5a01e89 100644
--- a/include/libc-symbols.h
+++ b/include/libc-symbols.h
< at >< at > -378,6 +378,17 < at >< at > for linking")
   strong_alias(real, name)
 #endif
 
+#ifdef IS_IN_rtld
+# define nonrtld_default_symbol_version(real, name, version) \
+  strong_alias(real, name)
+# define nonrtld_symbol_version(real, name, version)
+#else
+# define nonrtld_default_symbol_version(real, name, version) \
+  default_symbol_version(real, name, version)
+# define nonrtld_symbol_vers

[PATCH] Adjust abilist.awk for Alpha

Richard Henderson — 2012-05-26T16:26:29

Alpha uses the ST_OTHER field, which gets printed as field 7.  E.g.

000000000007d600 g    DF .text  0000000000000204  GLIBC_2.2   0x88 putwchar

with an additional complication of e.g.

00000000000fe7c0 g    D  .text  0000000000000060  GLIBC_2.0   0x80 __divlu

which is done so that callers of the division routines do *not* go through
a plt entry.  (Non-standard calling conventions.)

Sanity tested on x86_64.  Ok?


r~
* scripts/abilist.awk: Accept 8 fields.  Handle Alpha functions
marked to avoid plt entry.


diff --git a/scripts/abilist.awk b/scripts/abilist.awk
index 6a08839..f2e3857 100644
--- a/scripts/abilist.awk
+++ b/scripts/abilist.awk
< at >< at > -37,14 +37,14 < at >< at > $4 == "*UND*" { next }
 # Skip locals.
 $2 == "l" { next }
 
-$2 == "g" || $2 == "w" && NF == 7 {
+$2 == "g" || $2 == "w" && (NF == 7 || NF == 8) {
   weak = $2;
   type = $3;
   size = $5;
   sub(/^0*/, "", size);
   size = " 0x" size;
   version = $6;
-  symbol = $7;
+  symbol = (NF == 8 ? $8 : $7);
   gsub(/[()]/, "", version);
 
   if (versi

RFA: Port maintainers: Convert WORDSIZE[32|64]/ld to abi-variants

H.J. Lu — 2012-05-26T13:36:41

Hi, Glibc Port Maintainers,

Here is a patch to convert WORDSIZE[32|64]/ld in shlib-versions to
abi-variants in CPU/Makefile to support more than one ABI in
gnu/lib-names.h, gnu/stubs.h and bits/syscall.h:

1. For non-biarch architectures, there is nothing to do.
2. For biarch/triarch architectures where there are no soname differences,
you need to rename syscall-list-* variables in CPU/Makefile to abi-*.
If the default ABI for the build isn't the first one on abi-variants, you
should also define default-abi:

---
# The default ABI is 64.
default-abi := 64
---

3. For biarch/triarch architectures where there are soname differences,
in addition to renaming syscall-list-* variables in CPU/Makefile to
abi-* and defining default-abi, you also need to define

abi-XX-ld-soname := your ld.so soname.

where XX is the ABI variant, like

---
# We don't need any header files.
abi-includes :=

abi-variants := 32 64 x32

abi-32-options := -D__i386__ -U__x86_64__
abi-32-condition := !defined __x86_64__
abi-32-ld-soname :=

X32 is almost done

H.J. Lu — 2012-05-25T22:22:34

Hi,

Thanks for everyone, x32 is almost done.  The last remaining issue is

http://sourceware.org/bugzilla/show_bug.cgi?id=14112

I have a proposal on hjl/abi branch.  I updated x32 wiki page:

http://sourceware.org/glibc/wiki/x32

to describe how to bootstrap x32 GCC and GLIBC.  I verified the procedure
on openSUSE 12.1:

hjl< at >gnu-32:/tmp> cat /etc/os-release
NAME=openSUSE
VERSION = 12.1 (Asparagus)
VERSION_ID="12.1"
PRETTY_NAME="openSUSE 12.1 (Asparagus) (x86_64)"
ID=opensuse
hjl< at >gnu-32:/tmp> cat x.c
#include <stdio.h>

int
main (void)
{
  printf ("hello\n");
  return 0;
}
hjl< at >gnu-32:/tmp> /usr/gcc-x32-4.7/bin/x86_64-linux-gcc -mx32 x.c
hjl< at >gnu-32:/tmp> file a.out
a.out: ELF 32-bit LSB executable, x86-64, version 1 (SYSV),
dynamically linked (uses shared libs), for GNU/Linux 2.6.39, not
stripped
hjl< at >gnu-32:/tmp> ./a.out
hello
hjl< at >gnu-32:/tmp> uname -a
Linux gnu-32 3.4.0-1.9-desktop+ #3 SMP PREEMPT Fri May 25 05:30:28 PDT
2012 x86_64 x86_64 x86_64 GNU/Linux
hjl< at >gnu-32:/tmp>

[PATCH] Fix underflow and inexact signalling in soft-fp when packing.

David Miller — 2012-05-25T22:09:16

This fixes the long double math testsuite regressions that started
showing up on Sparc with Joseph's recent change to check underflow.

The corresponding changes needed in glibc-ports for powerpc/nofp,
mips, and alpha should be pretty straightforward.

In the main glibc tree, sparc is the only consumer of this code.

Ok to commit?

* soft-fp/op-common.h (_FP_PACK_CANONICAL): Only set underflow if
the underflow exception is enabled and the non-zero result is
tiny, or the non-zero result is tiny and there will be a loss of
accuracy.  Set inexact if overflow is detected after rounding, but
not if it is detected before.
* soft-fp/soft-fp.h (FP_CUR_EXCEPTIONS): Define.
(FP_TRAPPING_EXCEPTIONS): Provide default define.
* sysdeps/sparc/sparc32/soft-fp/sfp-machine.h (_FP_DECL_EX):
Initialize _fcw to zero.
(FP_TRAPPING_EXCEPTIONS): Define.
(FP_HANDLE_EXCEPTIONS): Add dummy use of _fcw.
* sysdeps/sparc/sparc64/soft-fp/sfp-machine.h (_FP_DECL_EX):
Initialize _fcw to zero.
(FP_TRAPPING_EXCEPTIONS): Define

Remove __ASSUME_NEW_GETRLIMIT_SYSCALL

Joseph S. Myers — 2012-05-25T21:29:10

__ASSUME_NEW_GETRLIMIT_SYSCALL is defined in kernel-features.h for x86
and powerpc, and in ports for arm, for m68k (>= 2.4.12) and tile.  It
is used in sysdeps/unix/sysv/linux/i386/getrlimit.c and
sysdeps/unix/sysv/linux/i386/setrlimit.c.  Those files are in turn
included for powerpc, s390-32, sh and (in ports) am33, arm and m68k.

This relates to system calls now called ugetrlimit and setrlimit.  As
of 2.4.0-test1, s390, sh and m68k all have those syscalls wired up
(this is about the syscall tables rather than what's in asm/unistd.h,
since we already require >= 2.6.19.1 kernel headers).  So although
those architectures do not define __ASSUME_NEW_GETRLIMIT_SYSCALL (at
all, or for 2.4.0 in the case of m68k) it's nevertheless safe to
assume the syscalls to be present for those architectures.  Finally,
am33 (mn10300) support was added in 2.6.25, and when added the support
for those syscalls was present (and kernel versions before support for
an architecture was officially added to Linux are not of concern to
gl

PATCH: Don't use header files in glibc configure

H.J. Lu — 2012-05-25T20:27:32

Hi,

Here is the patch adapted from the patch for the same problem in libgcc:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53183

When we configure glibc, we shouldn't use any headers.  Tested on
Linux/ia32,x32,x86-64. There are no changes in config.h nor config.make.
OK to install?

Thanks.


H.J.
---
* configure.in: Define the default includes to being none.
* configure: Regenerated.

diff --git a/configure b/configure
index 09a0637..7fa4941 100755
--- a/configure
+++ b/configure
< at >< at > -567,42 +567,7 < at >< at > PACKAGE_URL='http://www.gnu.org/software/glibc/'
 
 ac_unique_file="include/features.h"
 enable_option_checking=no
-# Factoring default headers for most tests.
-ac_includes_default="\
-#include <stdio.h>
-#ifdef HAVE_SYS_TYPES_H
-# include <sys/types.h>
-#endif
-#ifdef HAVE_SYS_STAT_H
-# include <sys/stat.h>
-#endif
-#ifdef STDC_HEADERS
-# include <stdlib.h>
-# include <stddef.h>
-#else
-# ifdef HAVE_STDLIB_H
-#  include <stdlib.h>
-# endif
-#endif
-#ifdef HAVE_STRING_H
-# if !defined STDC_HEADERS && defined

[PATCH] Suppress sign-conversion warning from FD_SET

Paul Pluzhnikov — 2012-05-25T16:27:01

Greetings,

Using glibc-2.15 to compile this file:

#include <sys/select.h>
int foo(int fd) {
  fd_set fds;
  FD_SET(fd, &fds);
  return FD_ISSET(fd, &fds);
}

with 'gcc -O2 -D_FORTIFY_SOURCE=2 -Werror=sign-conversion -c t.c', results in:

t.c: In function 'foo':
t.c:4: error: conversion to 'long unsigned int' from 'int' may change the sign of the result
t.c:5: error: conversion to 'long unsigned int' from 'int' may change the sign of the result

Attached trivial patch fixes that.

Thanks,

--
Paul Pluzhnikov


2012-05-25  Paul Pluzhnikov  <ppluzhnikov< at >google.com>

* misc/bits/select2.h (__FD_ELT): Add cast to suppress warning.



diff --git a/misc/bits/select2.h b/misc/bits/select2.h
index 9679925..4a20e11 100644
--- a/misc/bits/select2.h
+++ b/misc/bits/select2.h
< at >< at > -1,5 +1,5 < at >< at >
 /* Checking macros for select functions.
-   Copyright (C) 2011 Free Software Foundation, Inc.
+   Copyright (C) 2011, 2012 Free Software Foundation, Inc.
    This file is part of the GNU C Library.
 
    The GNU C Library is fre

[PATCH] Fix fma test for underflow

Marek Polacek — 2012-05-25T13:32:51

As discussed here
http://www.cygwin.com/ml/libc-alpha/2012-05/msg01782.html
and in BZ 14152, fma doesn't always set FE_UNDERFLOW.
So adjust the test.  Regtested on x86_64-linux.  Ok?

Or shall I ajdust the fmal test in the same manner?

2012-05-25  Marek Polacek  <polacek< at >redhat.com>

[BZ #14152]
* math/libm-test.inc (fma_test): Don't always expect underflow
exception.

--- libc/math/libm-test.inc.mp2012-05-25 14:34:08.639113677 +0200
+++ libc/math/libm-test.inc2012-05-25 14:44:29.347707291 +0200
< at >< at > -3970,7 +3970,8 < at >< at > fma_test (void)
   TEST_fff_f (fma, 0x1.4000004p-967, 0x1p-106, 0x0.000001p-1022, 0x0.0000010000003p-1022, UNDERFLOW_EXCEPTION);
   TEST_fff_f (fma, 0x1.4p-967, -0x1p-106, -0x0.000001p-1022, -0x0.0000010000002p-1022, UNDERFLOW_EXCEPTION);
   TEST_fff_f (fma, -0x1.19cab66d73e17p-959, 0x1.c7108a8c5ff51p-107, -0x0.80b0ad65d9b64p-1022, -0x0.80b0ad65d9d59p-1022, UNDERFLOW_EXCEPTION);
-  TEST_fff_f (fma, -0x1.d2eaed6e8e9d3p-979, -0x1.4e066c62ac9ddp-63, -0x0.9245e6b003454p-1022, -0x0.9245c09c5fb5

underflow exceptions

David Miller — 2012-05-25T00:11:32

Hey Joseph, just FYI your test to add underflow exceptions to
libm-test.inc is causing long double testsuite failures on 64-bit
sparc.

Specifically:

testing long double (without inline functions)
Failure: exp2 (-16400) == 0x1p-16400: Exception "Underflow" set
Failure: scalbn (min_value / 4, INT_MAX) == inf: Exception "Underflow" set
Failure: scalbln (min_value / 4, INT_MAX) == inf: Exception "Underflow" set
Failure: scalbln (min_value / 4, LONG_MAX) == inf: Exception "Underflow" set
Failure: scalbln (min_value / 4, 0x88000000) == inf: Exception "Underflow" set

Test suite completed:
  5478 test cases plus 4610 tests for exception flags executed.
  5 errors occurred.

I'll try to dig more deeply but I figured you might be able to tell
offhand whether this is some generic issue on 64-bit cpus or that
it might be prudent to relax the requirements.

PATCH: Convert WORDSIZE[32|64] to abi-variants

H.J. Lu — 2012-05-24T23:26:39

Hi,

I extended syscall-list support to cover all generated files on hjl/abi
branch.  If a target supports more than one ABI, it can define the
appropriate abi-*-*, like sysdeps/unix/sysv/linux/x86_64/Makefile has

# We don't need any header files.
abi-includes :=

abi-32-options := -D__i386__ -U__x86_64__
abi-32-condition := !defined __x86_64__
abi-32-ld-soname := ld-linux.so.2
abi-64-options := -U__i386__ -D__x86_64__ -U__ILP32__ -D__LP64__
abi-64-condition := defined __x86_64__ && defined __LP64__
abi-64-ld-soname := ld-linux-x86-64.so.2
abi-x32-options := -U__i386__ -D__x86_64__ -D__ILP32__ -U__LP64__
abi-x32-condition := defined __x86_64__ && defined __ILP32__
abi-x32-ld-soname := ld-linux-x32.so.2

abi-variants is a list of ABIs and
sysdeps/unix/sysv/linux/x86_64/64/Makefile has

# The default ABI is 64.
default-abi := 64

We have duplicated information of ld-soname in shlib-versions and
sysdeps/unix/sysv/linux/x86_64/Makefile.  I can use the ld soname
in CPU/Makefile for everything.  Any comments?

Th

Fix acosf underflow (bug 14153)

Joseph S. Myers — 2012-05-24T21:52:46

Bug 14153 is a spurious underflow exception from acosf (2e-17); the
cause is the acosf implementation calculating a value on the order of
the cube of its argument, which underflows.  There is a test for very
small arguments, maybe to avoid underflow, and if such an argument is
encountered it returns pi/2 (computed by adding high and low parts, I
suppose to get correct results for the current rounding mode and to
raise the inexact exception).  However, this tests for values no more
than 2**-57, as if the test was unchanged from a "double" version of
the function; it suffices to test for values no more than 2**-26,
which avoids the problem with internal underflows.  (For small x, the
correct value is essentially pi/2 - x.  Compute pi/2 rounded to 25
bits - one more bit than float - which is the nearest value that is
representable as float or half way between two representable values.
Compute the difference of pi/2 and that rounded values; if |x| is
smaller than that difference then the result of rounding pi/2

roland/systemtap branch

Roland McGrath — 2012-05-24T21:26:23

The roland/systemtap branch has been lingering for quite some time.
It never got merged, but there were never any coherent objections IIRC.
Related past threads I found:
http://sourceware.org/ml/libc-alpha/2011-03/msg00001.html 
http://sourceware.org/ml/libc-alpha/2011-02/msg00003.html 
http://sourceware.org/ml/libc-alpha/2011-01/msg00008.html 

I believe Fedora has been using this code since Fedora 15.
(I have Fedora 16/x86_64 handy and its libc.so.6 has the notes.)

I've just rebased the branch, fixed conflicts, updated copyright years
and log entries, and adapted the x86_64 assembly bits for x32.  (The
x86_64 pthread probes written in assembly use unadorned register
names, indicating 64-bit quantities for pointer values on x32, but I
don't think it really matters.)

I've tested the build on x86_64 but not really tested using the probes
with stap or gdb.  I just eyeballed the notes with 'readelf -n' and
they all look fine.

Are there any objections to (or endorsements for) merging the branch now?

The c

PATCH: Make chunk size a multiple of MALLOC_ALIGNMENT

H.J. Lu — 2012-05-24T19:30:00

Hi,

do_check_free_chunk has

static void do_check_free_chunk(mstate av, mchunkptr p)
{
  INTERNAL_SIZE_T sz = p->size & ~(PREV_INUSE|NON_MAIN_ARENA);
  mchunkptr next = chunk_at_offset(p, sz);

  do_check_chunk(av, p);

  /* Chunk must claim to be free ... */
  assert(!inuse(p));
  assert (!chunk_is_mmapped(p));

  /* Unless a special marker, must have OK fields */
  if ((unsigned long)(sz) >= MINSIZE)
  {
    assert((sz & MALLOC_ALIGN_MASK) == 0);

If a free chunk >= MINSIZE, it must be a multiple of MALLOC_ALIGNMENT.
However, when sysmalloc frees old top chunk with size >= MINSIZE, it
doesn't make sure that the size is a multiple of MALLOC_ALIGNMENT:

     /* Setup fencepost and free the old top chunk. */
      /* The fencepost takes at least MINSIZE bytes, because it might
         become the top chunk again later.  Note that a footer is set
         up, too, although the chunk is marked in use. */
      old_size -= MINSIZE;
      set_head(chunk_at_offset(old_top, old_size + 2*SIZE_SZ), 0|PREV_INUSE);

[Patch] Testcase for vfprintf bug

Jeff Law — 2012-05-24T18:20:24

I've recently had the need to review glibc's vfprintf code.  During the 
process I discovered that no testcase for bug #13446 was ever installed.

The original testcase in #13446 utilizes the printf hooks mechanism; the 
system where I needed the problem fixed & tested pre-dates printf hooks, 
so I mangled the existing stdio-common/bug23.c test to expose the memory 
allocation bug in #13446.

OK to install?

2012-05-24  Jeff Law  <law< at >redhat.com>

* stdio-common/Makefile (tests): Add bug25.
* stdio-common/bug25.c: New test.

diff --git a/stdio-common/Makefile b/stdio-common/Makefile
index 02684d2..a5b6f74 100644
--- a/stdio-common/Makefile
+++ b/stdio-common/Makefile
< at >< at > -56,7 +56,7 < at >< at > tests := tstscanf test_rdwr test-popen tstgetln test-fseek \
  tst-fwrite bug16 bug17 tst-swscanf tst-sprintf2 bug18 bug18a \
  bug19 bug19a tst-popen2 scanf13 scanf14 scanf15 bug20 bug21 bug22 \
  scanf16 scanf17 tst-setvbuf1 tst-grouping bug23 bug24 \
- bug-vfprintf-nargs tst-long-dbl-fphex tst-fphex-wide tst-sprintf3

Add tests for underflow exception bugs 10846, 14036

Joseph S. Myers — 2012-05-24T14:57:24

Now that libm-test.inc can test underflow exceptions, I propose this
patch to add tests for two bugs about missing underflow exceptions
(fixed by the changes to use -frounding-math).  Tested x86 and x86_64.

2012-05-24  Joseph Myers  <joseph< at >codesourcery.com>

[BZ #10846]
[BZ #14036]
* math/libm-test.inc (exp_test): Add test from bug 14036.
(pow_test): Add test from bug 10846.

diff --git a/math/libm-test.inc b/math/libm-test.inc
index 5424866..f545723 100644
--- a/math/libm-test.inc
+++ b/math/libm-test.inc
< at >< at > -3501,6 +3501,7 < at >< at > exp_test (void)
 
 #if !(defined TEST_LDOUBLE && LDBL_MAX_EXP > 1024)
   TEST_f_f (exp, 710, plus_infty, OVERFLOW_EXCEPTION);
+  TEST_f_f (exp, -1234, plus_zero, UNDERFLOW_EXCEPTION);
 #endif
   TEST_f_f (exp, 1e5, plus_infty, OVERFLOW_EXCEPTION);
   TEST_f_f (exp, max_value, plus_infty, OVERFLOW_EXCEPTION);
< at >< at > -6732,6 +6733,8 < at >< at > pow_test (void)
   TEST_ff_f (pow, 0x1.0000000000001p0L, -0x1.23456789abcdfp61L, 1.0118762747828234466621210689458255908670e-253L);
 #endif
 
+  TEST_f

PATCH: Add x32 abilist files

H.J. Lu — 2012-05-24T13:14:39

Hi,

This patch adds x32 abilist files. Tested on Linux/x32.  OK to
install?

Thanks.


H.J.
---
* sysdeps/unix/sysv/linux/x86_64/x32/nptl/ld.abilist: New file.
* sysdeps/unix/sysv/linux/x86_64/x32/nptl/libBrokenLocale.abilist:
Likewise.
* sysdeps/unix/sysv/linux/x86_64/x32/nptl/libanl.abilist:
Likewise.
* sysdeps/unix/sysv/linux/x86_64/x32/nptl/libc.abilist:
Likewise.
* sysdeps/unix/sysv/linux/x86_64/x32/nptl/libcrypt.abilist:
Likewise.
* sysdeps/unix/sysv/linux/x86_64/x32/nptl/libdl.abilist:
Likewise.
* sysdeps/unix/sysv/linux/x86_64/x32/nptl/libm.abilist:
Likewise.
* sysdeps/unix/sysv/linux/x86_64/x32/nptl/libnsl.abilist:
Likewise.
* sysdeps/unix/sysv/linux/x86_64/x32/nptl/libpthread.abilist:
Likewise.
* sysdeps/unix/sysv/linux/x86_64/x32/nptl/libresolv.abilist:
Likewise.
* sysdeps/unix/sysv/linux/x86_64/x32/nptl/librt.abilist:
Likewise.
* sysdeps/unix/sysv/linux/x86_64/x32/nptl/libthread_db.abilist:
Likewise.
* sysdeps/unix/sysv/linux/x86_64/x32/nptl/libutil.abilist:
Likewise.

d

PATCH: Add x32 C++ type data

H.J. Lu — 2012-05-24T13:10:33

Hi,

This patch adds x32 C++ type data.  Tested on Linux/x32.  OK to install?

Thanks.


H.J.
----
* scripts/data/c++-types-x32-linux-gnu.data: New file.
* sysdeps/x86_64/x32/shlib-versions: Add x32 ABI entry.

diff --git a/scripts/data/c++-types-x32-linux-gnu.data b/scripts/data/c++-types-x32-linux-gnu.data
new file mode 100644
index 0000000..348bf52
--- /dev/null
+++ b/scripts/data/c++-types-x32-linux-gnu.data
< at >< at > -0,0 +1,67 < at >< at >
+blkcnt64_t:x
+blkcnt_t:x
+blksize_t:x
+caddr_t:Pc
+clockid_t:i
+clock_t:x
+daddr_t:i
+dev_t:y
+fd_mask:l
+fsblkcnt64_t:y
+fsblkcnt_t:y
+fsfilcnt64_t:y
+fsfilcnt_t:y
+fsid_t:8__fsid_t
+gid_t:j
+id_t:j
+ino64_t:y
+ino_t:y
+int16_t:s
+int32_t:i
+int64_t:x
+int8_t:a
+intptr_t:i
+key_t:i
+loff_t:x
+mode_t:j
+nlink_t:y
+off64_t:x
+off_t:x
+pid_t:i
+pthread_attr_t:14pthread_attr_t
+pthread_barrier_t:17pthread_barrier_t
+pthread_barrierattr_t:21pthread_barrierattr_t
+pthread_cond_t:14pthread_cond_t
+pthread_condattr_t:18pthread_condattr_t
+pthread_key_t:j
+pthread_mutex_t:15pthread_mutex_

Don't include exceptions in libm-test-ulps test names

Joseph S. Myers — 2012-05-24T11:51:45

As noted in my underflow patch, the names of libm tests that appear in
libm-test-ulps include a description of the tests for exceptions.

This can cause undue churn in the libm-test-ulps files.  In
particular, if a test is made to permit an exception until a bug about
that spurious exception is fixed (with a comment referencing the bug
beside the test in question), then ulps for that test need updating,
and when the bug is fixed they need updating again, whether or not the
fix actually changed the value returned by the function.

I think this outweighs any value in actually having these
descriptions, and propose this patch which removes the exception
descriptions from the test names (and makes the corresponding
mechanical updates to libm-test-ulps files).  Tested x86 and x86_64.

2012-05-24  Joseph Myers  <joseph< at >codesourcery.com>

* math/gen-libm-test.pl (%beautify): Remove entries for exceptions
and other flags.
(special_function): Do not include flags in test name.
(parse_args): Likewise.
* sysdeps/i

[PATCH] Remove some sparc cob-webs leftover after Joseph's recent changes.

David Miller — 2012-05-24T04:52:02

Committed to master.

* sysdeps/unix/sysv/linux/sparc/sparc32/sigaction.c
(__libc_sigaction): Remove unused local variables.
---
 sysdeps/unix/sysv/linux/sparc/sparc32/sigaction.c |    4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/sysdeps/unix/sysv/linux/sparc/sparc32/sigaction.c b/sysdeps/unix/sysv/linux/sparc/sparc32/sigaction.c
index 7bdf5e0..9a535e8 100644
--- a/sysdeps/unix/sysv/linux/sparc/sparc32/sigaction.c
+++ b/sysdeps/unix/sysv/linux/sparc/sparc32/sigaction.c
< at >< at > -30,11 +30,9 < at >< at > static void __sigreturn_stub (void);
 int
 __libc_sigaction (int sig, const struct sigaction *act, struct sigaction *oact)
 {
-  struct old_kernel_sigaction k_sigact, k_osigact;
-  int ret;
-
   struct kernel_sigaction kact, koact;
   unsigned long stub = 0;
+  int ret;
 
   if (act)
     {