David S. Miller [Wed, 30 Apr 2014 19:57:51 +0000 (12:57 -0700)]
Fix v9/64-bit strcmp when string ends in multiple zero bytes.
[BZ #16885]
* sysdeps/sparc/sparc64/strcmp.S: Fix end comparison handling when
multiple zero bytes exist at the end of a string.
Reported by Aurelien Jarno <aurelien@aurel32.net>
* string/test-strcmp.c (check): Add explicit test for situations where
there are multiple zero bytes after the first.
Carlos O'Donell [Thu, 30 Jan 2014 20:03:04 +0000 (15:03 -0500)]
[PATCH] Make sunrpc code usable again
New configure option --enable-obsolete-rpc makes the deprecated RPC
headers and functions available at compile time as they were before
version 2.14. This option will be removed at some time in the future
after the TI-RPC library becomes fully sufficient for the needs of
existing applications.
* stdio-common/Makefile (tst-sprintf-ENV): Set environment
for testcase.
* stdio-common/tst-sprintf.c: Include <locale.h>
(main): Test sprintf's handling of incomplete multibyte
characters.
Andreas Schwab [Fri, 22 Jun 2012 18:10:31 +0000 (11:10 -0700)]
Fix invalid memory access in do_lookup_x.
[BZ #13579] Do not free l_initfini and allow it to be reused
on subsequent dl_open calls for the same library. This fixes
the invalid memory access in do_lookup_x when the previously
free'd l_initfini was accessed through l_searchlist when a
library had been opened for the second time.
Maxim Kuvyrkov [Mon, 23 Dec 2013 20:44:50 +0000 (09:44 +1300)]
Fix race in free() of fastbin chunk: BZ #15073
Perform sanity check only if we have_lock. Due to lockless nature of fastbins
we need to be careful derefencing pointers to fastbin entries (chunksize(old)
in this case) in multithreaded environments.
The fix is to add have_lock to the if-condition checks. The rest of the patch
only makes code more readable.
* malloc/malloc.c (_int_free): Perform sanity check only if we
have_lock.
Jeff Law [Wed, 7 Nov 2012 18:58:37 +0000 (11:58 -0700)]
2012-11-07 Andreas Jaeger <aj@suse.de>
[BZ #14809]
* sysdeps/unix/sysv/linux/sys/sysctl.h (_UAPI_LINUX_KERNEL_H)
(_UAPI_LINUX_TYPES_H): Starting with Linux 3.7, the include header
guards are changed. Only define if not yet defined, #undef back
after including linux/sysctl.h if defined here.
(cherry picked from commit 01f34a3bd8c087ca4be0bd24857e454c8d29f20b)
Ulrich Drepper [Thu, 26 Jan 2012 14:45:54 +0000 (09:45 -0500)]
Really fix AVX tests
There is no problem with strcmp, it doesn't use the YMM registers.
The math routines might since gcc perhaps generates such code.
Introduce bit_YMM_USBALE and use it in the math routines.
David S. Miller [Mon, 27 Feb 2012 22:51:45 +0000 (14:51 -0800)]
Fix ldbl-128 rintl/nearbyintl just like flt-32 variants.
* sysdeps/ieee754/ldbl-128/s_nearbyintl.c (__nearbyintl): Do not
manipulate bits before adding and subtracting TWO112[sx].
* sysdeps/ieee754/ldbl-128/s_rintl.c (__rintl): Likewise.
Carlos O'Donell [Sun, 22 Apr 2012 16:06:09 +0000 (12:06 -0400)]
Redefine __i686 for use with older compilers.
This is a conservative fix for BZ #411 which is already fixed
on trunk. In this fix we don't touch any of the thunks or
do any other work other than to redefine the compiler define.
This is the minimum change required to build on a broader range
of distributions.
Jakub Jelinek [Fri, 30 Dec 2011 22:13:56 +0000 (17:13 -0500)]
Fix up regcomp/regexec
The problem is that parse_bracket_symbol is miscompiled, and it turns
out it is because of an incorrect attribute on re_string_fetch_byte_case.
Unlike re_string_peek_byte_case, this one is really not pure, it modifies memory
(increments pstr->cur_idx), and with the pure attribute GCC assumed it doesn't
and it cached the presumed value of regexp->cur_idx in a variable across the
for (;; ++i)
{
if (i >= BRACKET_NAME_BUF_SIZE)
return REG_EBRACK;
if (token->type == OP_OPEN_CHAR_CLASS)
ch = re_string_fetch_byte_case (regexp);
else
ch = re_string_fetch_byte (regexp);
if (re_string_eoi(regexp))
return REG_EBRACK;
if (ch == delim && re_string_peek_byte (regexp, 0) == ']')
break;
elem->opr.name[i] = ch;
}
Liubov Dmitrieva [Fri, 23 Dec 2011 13:50:39 +0000 (08:50 -0500)]
Fix wrong copying processing for last bytes in x86-32 wcscpy
Wrong copy algorithm for last bytes, not thread safety.
In some particular cases it uses the destination
memory beyond the string end for
16-byte load, puts changes into that part that is relevant
to destination string and writes whole 16-byte chunk into memory.
I have a test case where the memory beyond the string end contains
malloc/free data, that appear corrupted in case free() updates
it in between the 16-byte read and 16-byte write.