OSDN Git Service

uclinux-h8/uClibc.git
17 years agoSupport SecurePLTs for PowerPC. You need a toolchain that supports
Joakim Tjernlund [Fri, 5 Jan 2007 10:20:37 +0000 (10:20 -0000)]
Support SecurePLTs for PowerPC. You need a toolchain that supports
config option --enable-secureplt. The assembler must also
supports R_PPC_REL16* relocations. gcc 4.1.1 and binutils 2.17
is known to do this.

17 years agoAtle Nissestad writes: The attached patch fixes compilation of the current svn on...
Mike Frysinger [Fri, 5 Jan 2007 09:09:22 +0000 (09:09 -0000)]
Atle Nissestad writes: The attached patch fixes compilation of the current svn on the nios2 platform, and updates the crt1/n/i.S files to get CTOR/DTOR-support to work.

17 years ago- s/Attemt/Attempt/
Bernhard Reutner-Fischer [Wed, 3 Jan 2007 22:53:41 +0000 (22:53 -0000)]
- s/Attemt/Attempt/

17 years ago- r16732 forgot to checkin the sync of the __UCLIBC_SUSV3_LEGACY_MACROS__ from string.h
Bernhard Reutner-Fischer [Sat, 23 Dec 2006 14:12:40 +0000 (14:12 -0000)]
- r16732 forgot to checkin the sync of the __UCLIBC_SUSV3_LEGACY_MACROS__ from string.h

17 years ago- provide a hidden_def for fputc_unlocked that is aliased to __fputc_unlocked.
Bernhard Reutner-Fischer [Fri, 22 Dec 2006 14:03:06 +0000 (14:03 -0000)]
- provide a hidden_def for fputc_unlocked that is aliased to __fputc_unlocked.
  Fixes undefined references to __fputc_unlocked when calling fputc().

17 years agomake sure UCLIBC_LDSO can be found via path
Mike Frysinger [Wed, 20 Dec 2006 16:17:33 +0000 (16:17 -0000)]
make sure UCLIBC_LDSO can be found via path

17 years agoPaul Brook writes:
Mike Frysinger [Wed, 20 Dec 2006 04:30:25 +0000 (04:30 -0000)]
Paul Brook writes:
When building an arm-uclinux toolchain withthreading disabled I get:

libc/stdio/getchar.c:26: error: 'getchar' aliased to undefined
symbol 'getchar_unlocked'

It looks like it's missing libc_hidden_def(getchar_unlocked). Patch below
fixes this.

17 years agorename local "brk" var so it doesnt namespace collide with the brk() function
Mike Frysinger [Sun, 17 Dec 2006 20:56:01 +0000 (20:56 -0000)]
rename local "brk" var so it doesnt namespace collide with the brk() function

17 years agofor now, dont allow mmu users to select file format
Mike Frysinger [Sun, 17 Dec 2006 20:49:10 +0000 (20:49 -0000)]
for now, dont allow mmu users to select file format

17 years agoactually scrub it, dont just comment it out
Mike Frysinger [Sat, 16 Dec 2006 19:18:08 +0000 (19:18 -0000)]
actually scrub it, dont just comment it out

17 years agostart pulling in top level Rules.mak so we get access to import CFLAGS/LDFLAGS/CPPFLAGS
Mike Frysinger [Sat, 16 Dec 2006 19:13:21 +0000 (19:13 -0000)]
start pulling in top level Rules.mak so we get access to import CFLAGS/LDFLAGS/CPPFLAGS

17 years agoscrub prototype so we dont conflict with the pthread header
Mike Frysinger [Sat, 16 Dec 2006 19:12:45 +0000 (19:12 -0000)]
scrub prototype so we dont conflict with the pthread header

17 years agoclean up
Mike Frysinger [Sat, 16 Dec 2006 19:12:19 +0000 (19:12 -0000)]
clean up

17 years agosmall cleanups
Mike Frysinger [Sat, 16 Dec 2006 19:08:54 +0000 (19:08 -0000)]
small cleanups

17 years agoPatch from Bernhard Fischer:
Eric Andersen [Fri, 15 Dec 2006 00:12:23 +0000 (00:12 -0000)]
Patch from Bernhard Fischer:
fix compilation if PUTC_MACRO and/or GETC_MACRO is turned off

17 years agofix building when __NR_sigalstack does not exist
Mike Frysinger [Thu, 14 Dec 2006 00:37:57 +0000 (00:37 -0000)]
fix building when __NR_sigalstack does not exist

17 years agoif __NR_lseek is not defined, then we fall back to calling the 64bit lseek
Mike Frysinger [Thu, 14 Dec 2006 00:33:55 +0000 (00:33 -0000)]
if __NR_lseek is not defined, then we fall back to calling the 64bit lseek

17 years agoforce LFS support since blackfin no longer supports non-lfs
Mike Frysinger [Wed, 13 Dec 2006 23:01:56 +0000 (23:01 -0000)]
force LFS support since blackfin no longer supports non-lfs

17 years agobe more specific as to what arguments to __uClibc_main() people can skim on during...
Mike Frysinger [Wed, 13 Dec 2006 04:26:28 +0000 (04:26 -0000)]
be more specific as to what arguments to __uClibc_main() people can skim on during initial port

17 years agoFix/disable locking when pthreads is disabled
Eric Andersen [Tue, 12 Dec 2006 21:43:59 +0000 (21:43 -0000)]
Fix/disable locking when pthreads is disabled

17 years ago- fix compilation of __LDSO_CACHE_SUPPORT__
Bernhard Reutner-Fischer [Tue, 12 Dec 2006 12:15:04 +0000 (12:15 -0000)]
- fix compilation of __LDSO_CACHE_SUPPORT__
  s/dprintf(2/fprintf(stderr/g

17 years agosort out a few issues that show up on mips
Eric Andersen [Tue, 12 Dec 2006 06:02:05 +0000 (06:02 -0000)]
sort out a few issues that show up on mips

17 years agosince we dont symlink the headers, theres nothing to filter
Mike Frysinger [Tue, 12 Dec 2006 03:56:42 +0000 (03:56 -0000)]
since we dont symlink the headers, theres nothing to filter

17 years agowe just want the kernel headers, not the whole kernel source ... so people may need...
Mike Frysinger [Tue, 12 Dec 2006 03:30:02 +0000 (03:30 -0000)]
we just want the kernel headers, not the whole kernel source ... so people may need to update their paths slightly

17 years agomostly revert the locking changes for 'malloc', as the implementation
Eric Andersen [Mon, 11 Dec 2006 05:28:24 +0000 (05:28 -0000)]
mostly revert the locking changes for 'malloc', as the implementation
does not easily lend itself to becoming complete pthread cancelation
safe without first investing in some deep and serious thought...
The other malloc implementations are pthread cancelation safe, and
this one is mostly used for uClinux, where the lack is at least less
likely to be a common problem.

17 years agofix building when pthread support is disabled
Mike Frysinger [Mon, 11 Dec 2006 04:24:55 +0000 (04:24 -0000)]
fix building when pthread support is disabled

17 years agodisable LDSO when FLAT format is selected
Mike Frysinger [Mon, 11 Dec 2006 04:19:50 +0000 (04:19 -0000)]
disable LDSO when FLAT format is selected

17 years agodont try and compile if the MMU is disabled
Mike Frysinger [Mon, 11 Dec 2006 04:09:06 +0000 (04:09 -0000)]
dont try and compile if the MMU is disabled

17 years agoneed endian.h for __LONG_LONG_PAIR() macro
Mike Frysinger [Mon, 11 Dec 2006 04:06:55 +0000 (04:06 -0000)]
need endian.h for __LONG_LONG_PAIR() macro

17 years agocleanup style
Mike Frysinger [Mon, 11 Dec 2006 04:06:36 +0000 (04:06 -0000)]
cleanup style

17 years agokill off the need for fix_includes.sh by utilizing KERNEL_SOURCE directly
Mike Frysinger [Mon, 11 Dec 2006 00:52:12 +0000 (00:52 -0000)]
kill off the need for fix_includes.sh by utilizing KERNEL_SOURCE directly

17 years agoset some default vars to prevent env pollution
Mike Frysinger [Mon, 11 Dec 2006 00:32:40 +0000 (00:32 -0000)]
set some default vars to prevent env pollution

17 years agotie PICFLAG:=-mfdpic to format rather than architecture
Mike Frysinger [Mon, 11 Dec 2006 00:27:48 +0000 (00:27 -0000)]
tie PICFLAG:=-mfdpic to format rather than architecture

17 years agocreate new menu so people can select what format they wish to build uClibc
Mike Frysinger [Mon, 11 Dec 2006 00:25:23 +0000 (00:25 -0000)]
create new menu so people can select what format they wish to build uClibc

17 years agothese weaks are also needed in libc.so for the time being
Eric Andersen [Sat, 9 Dec 2006 08:50:41 +0000 (08:50 -0000)]
these weaks are also needed in libc.so for the time being

17 years agooops, use __pthread_mutex_unlock() not pthread_mutex_unlock()
Eric Andersen [Sat, 9 Dec 2006 08:35:37 +0000 (08:35 -0000)]
oops, use __pthread_mutex_unlock() not pthread_mutex_unlock()

17 years agoAdd missing sys/types.h header
Eric Andersen [Sat, 9 Dec 2006 04:57:11 +0000 (04:57 -0000)]
Add missing sys/types.h header

17 years agosilence a lot of warnings and fix a few genuine bugs while I'm at it.
Eric Andersen [Sat, 9 Dec 2006 01:32:13 +0000 (01:32 -0000)]
silence a lot of warnings and fix a few genuine bugs while I'm at it.

17 years agoRichard Sandiford writes: add support for init/fini arrays in shared flat libraries
Mike Frysinger [Fri, 8 Dec 2006 22:53:40 +0000 (22:53 -0000)]
Richard Sandiford writes: add support for init/fini arrays in shared flat libraries

17 years agoseems this is the newfangled way to tell folks this stuff
Eric Andersen [Fri, 8 Dec 2006 21:52:02 +0000 (21:52 -0000)]
seems this is the newfangled way to tell folks this stuff
is really really and we mean it this time, for uClibc only

17 years agoonly define msync cancellation point if the syscall exists
Mike Frysinger [Fri, 8 Dec 2006 21:24:15 +0000 (21:24 -0000)]
only define msync cancellation point if the syscall exists

17 years agoTake Mike Frysinger's comments into account -- make certain that user
Eric Andersen [Fri, 8 Dec 2006 21:05:37 +0000 (21:05 -0000)]
Take Mike Frysinger's comments into account -- make certain that user
applications stop using _syscall#() and use syscall() instead.  Cleanup
internal handling of syscall includes to use the correct header file.

17 years agoSorry psm but contrary to commit 12927, kernel provided syscalls
Eric Andersen [Fri, 8 Dec 2006 19:28:27 +0000 (19:28 -0000)]
Sorry psm but contrary to commit 12927, kernel provided syscalls
are NOT suitable for user-space.  The libc ones are the ones that
application code must use.  If a problem is found with the libc
ones, then the libc syscall macros should be fixed.

17 years agoadd some warnings to archs that need a bit of attention
Eric Andersen [Fri, 8 Dec 2006 19:21:11 +0000 (19:21 -0000)]
add some warnings to archs that need a bit of attention

17 years agoif __NR_swapon does not exist, dont try and create a syscall for it
Mike Frysinger [Fri, 8 Dec 2006 18:53:12 +0000 (18:53 -0000)]
if __NR_swapon does not exist, dont try and create a syscall for it

17 years agoif __NR_swapoff does not exist, dont try and create a syscall for it
Mike Frysinger [Fri, 8 Dec 2006 18:52:40 +0000 (18:52 -0000)]
if __NR_swapoff does not exist, dont try and create a syscall for it

17 years agoif __NR_msync does not exist, dont try and create a syscall for it
Mike Frysinger [Fri, 8 Dec 2006 18:47:59 +0000 (18:47 -0000)]
if __NR_msync does not exist, dont try and create a syscall for it

17 years agoonly define mremap() if the syscall exists
Mike Frysinger [Fri, 8 Dec 2006 16:55:54 +0000 (16:55 -0000)]
only define mremap() if the syscall exists

17 years agoif __NR_mprotect does not exist, dont try and create a syscall for it
Mike Frysinger [Fri, 8 Dec 2006 03:39:59 +0000 (03:39 -0000)]
if __NR_mprotect does not exist, dont try and create a syscall for it

17 years agothis stuff is for uClibc internal use only
Eric Andersen [Fri, 8 Dec 2006 03:39:07 +0000 (03:39 -0000)]
this stuff is for uClibc internal use only

17 years agofinish the pthread locking cleanups. This should get things compiling again
Eric Andersen [Fri, 8 Dec 2006 03:19:36 +0000 (03:19 -0000)]
finish the pthread locking cleanups.  This should get things compiling again
though I still need to silence the warnings about _pthread_cleanup_push_defer

17 years ago__resolv_lock is supposed to be shared, not static
Eric Andersen [Fri, 8 Dec 2006 00:31:47 +0000 (00:31 -0000)]
__resolv_lock is supposed to be shared, not static

17 years agoFill in a few more of the missing pieces...
Eric Andersen [Fri, 8 Dec 2006 00:19:36 +0000 (00:19 -0000)]
Fill in a few more of the missing pieces...

17 years agoMajor cleanup of internal mutex locking. Be more consistant in how we do
Eric Andersen [Thu, 7 Dec 2006 23:24:19 +0000 (23:24 -0000)]
Major cleanup of internal mutex locking.  Be more consistant in how we do
things, and avoid potential deadlocks caused when a thread holding a uClibc
internal lock get canceled and terminates without releasing the lock.  This
change also provides a single place, bits/uClibc_mutex.h, for thread libraries
to modify to change all instances of internal locking.

17 years agoMajor cleanup of internal mutex locking. Be more consistant in how we do
Eric Andersen [Thu, 7 Dec 2006 23:24:02 +0000 (23:24 -0000)]
Major cleanup of internal mutex locking.  Be more consistant in how we do
things, and avoid potential deadlocks caused when a thread holding a uClibc
internal lock get canceled and terminates without releasing the lock.  This
change also provides a single place, bits/uClibc_mutex.h, for thread libraries
to modify to change all instances of internal locking.

17 years agostdio update from mjn3
Eric Andersen [Wed, 6 Dec 2006 22:41:21 +0000 (22:41 -0000)]
stdio update from mjn3

17 years agoupdate from mjn3
Eric Andersen [Wed, 6 Dec 2006 21:46:56 +0000 (21:46 -0000)]
update from mjn3

17 years agoprepare to later remove bits/kernel_types.h from bits/types.h
Eric Andersen [Wed, 6 Dec 2006 20:46:21 +0000 (20:46 -0000)]
prepare to later remove bits/kernel_types.h from bits/types.h

17 years agobits/kernel_stat.h is for internal uClibc use only, fix a few other
Eric Andersen [Wed, 6 Dec 2006 20:22:42 +0000 (20:22 -0000)]
bits/kernel_stat.h is for internal uClibc use only, fix a few other
minor include file issues

17 years agopulling in kernel config.h wont fly
Mike Frysinger [Wed, 6 Dec 2006 01:55:59 +0000 (01:55 -0000)]
pulling in kernel config.h wont fly

17 years agonote test subdir problems with multilib
Mike Frysinger [Wed, 6 Dec 2006 01:53:40 +0000 (01:53 -0000)]
note test subdir problems with multilib

17 years agomake ABI into a list
Mike Frysinger [Sun, 3 Dec 2006 20:13:25 +0000 (20:13 -0000)]
make ABI into a list

17 years agolist ABI first and tweak ISA defaults with suggestions from Jim Gifford
Mike Frysinger [Sun, 3 Dec 2006 20:10:28 +0000 (20:10 -0000)]
list ABI first and tweak ISA defaults with suggestions from Jim Gifford

17 years agorevert previous change
Mike Frysinger [Sun, 3 Dec 2006 19:52:48 +0000 (19:52 -0000)]
revert previous change

17 years agoPatch from Jim Gifford (of Cross Linux From Scratch) fixing the Mips
Rob Landley [Sun, 3 Dec 2006 18:22:06 +0000 (18:22 -0000)]
Patch from Jim Gifford (of Cross Linux From Scratch) fixing the Mips
failures from Mike's nightly builds.

17 years agosecond SUSV3 legacy check should be for macros as pointed out by psm
Mike Frysinger [Sun, 3 Dec 2006 02:50:45 +0000 (02:50 -0000)]
second SUSV3 legacy check should be for macros as pointed out by psm

17 years agofix building when susv3 legacy stuff is disabled
Mike Frysinger [Sat, 2 Dec 2006 19:31:40 +0000 (19:31 -0000)]
fix building when susv3 legacy stuff is disabled

17 years agoPTHREAD_THREADS_MAX is supposed to be in pthread-specific dir, but for now it lives...
Mike Frysinger [Sat, 2 Dec 2006 19:19:38 +0000 (19:19 -0000)]
PTHREAD_THREADS_MAX is supposed to be in pthread-specific dir, but for now it lives in bits/uClibc_local_lim.h

17 years ago(no commit message)
"Steven J. Hill" [Sat, 2 Dec 2006 02:01:57 +0000 (02:01 -0000)]

17 years agoRemove definition of TIMER_MAX which breaks builds of GCC-4.2 and in glibc code it...
"Steven J. Hill" [Sat, 2 Dec 2006 01:42:11 +0000 (01:42 -0000)]
Remove definition of TIMER_MAX which breaks builds of GCC-4.2 and in glibc code it is not defined.

17 years agoremove redundant headers local to libpthread, causing header
Eric Andersen [Wed, 29 Nov 2006 21:29:28 +0000 (21:29 -0000)]
remove redundant headers local to libpthread, causing header
file content skew between the versions

17 years ago- introduce UCLIBC_SUSV3_LEGACY_MACROS to be able to do away with all LEGACY
Bernhard Reutner-Fischer [Wed, 29 Nov 2006 21:12:13 +0000 (21:12 -0000)]
- introduce UCLIBC_SUSV3_LEGACY_MACROS to be able to do away with all LEGACY
  stuff, even the fallback defines.

17 years ago- honor __UCLIBC_HAS_LFS__
Bernhard Reutner-Fischer [Wed, 29 Nov 2006 21:10:04 +0000 (21:10 -0000)]
- honor __UCLIBC_HAS_LFS__

17 years ago- Cleanout the only occurance of the SUSv3 LEGACY function getwd() that was
Bernhard Reutner-Fischer [Wed, 29 Nov 2006 21:07:00 +0000 (21:07 -0000)]
- Cleanout the only occurance of the SUSv3 LEGACY function getwd() that was
  furthermore ifdef'ed out anyway.

17 years ago- gcvt() is marked LEGACY in XSI and mandatory in BSD.
Bernhard Reutner-Fischer [Wed, 29 Nov 2006 21:04:35 +0000 (21:04 -0000)]
- gcvt() is marked LEGACY in XSI and mandatory in BSD.
  Also make compilation conditional on float support.

17 years agofix from cmetcalf to make sure we clear the auxvt table before using it
Mike Frysinger [Mon, 27 Nov 2006 15:25:32 +0000 (15:25 -0000)]
fix from cmetcalf to make sure we clear the auxvt table before using it

17 years agoindent
Mike Frysinger [Mon, 27 Nov 2006 15:11:28 +0000 (15:11 -0000)]
indent

17 years ago- fix typo
Bernhard Reutner-Fischer [Wed, 22 Nov 2006 17:09:42 +0000 (17:09 -0000)]
- fix typo

17 years agoCorrect -march cflag value for thumb capable ARM processors: Thanks Paul Brook
Khem Raj [Wed, 22 Nov 2006 02:20:09 +0000 (02:20 -0000)]
Correct -march cflag value for thumb capable ARM processors: Thanks Paul Brook

17 years agoUse constant pool instead of and adr
Khem Raj [Tue, 21 Nov 2006 02:50:42 +0000 (02:50 -0000)]
Use constant pool instead of and adr
instruction in unoptimized builds and thumb: thanks Paul Brook

17 years agoAdd CHECK_ONLY option to allow for running the tests on the target instead of attempt...
"Steven J. Hill" [Sat, 18 Nov 2006 15:04:54 +0000 (15:04 -0000)]
Add CHECK_ONLY option to allow for running the tests on the target instead of attempting to link or compile them. Perhaps I don't fully understand the test structure, but for TLS and NPTL the tests cannot be compiled on the target without a full source tree.

17 years agofix from Paul Brook for ISO C violation in terms of handling of modifying automatic...
Mike Frysinger [Fri, 17 Nov 2006 18:18:46 +0000 (18:18 -0000)]
fix from Paul Brook for ISO C violation in terms of handling of modifying automatic local variables between setjmp/longjmp

17 years agoBernd Schmidt writes: [blackfin updates] add support for FDPIC and include L1 functions
Mike Frysinger [Fri, 17 Nov 2006 06:51:34 +0000 (06:51 -0000)]
Bernd Schmidt writes: [blackfin updates] add support for FDPIC and include L1 functions

17 years agoBernd Schmidt writes:
Mike Frysinger [Fri, 17 Nov 2006 06:07:30 +0000 (06:07 -0000)]
Bernd Schmidt writes:
This reintroduces a mechanism identical to the DL_BOOT macro present in
older versions of uClibc.  On Blackfin and FRV, we want to pass more
than one argument to _dl_start.  We also want to do something special
before returning, so delete the warning when the START macro is defined.

17 years agoget rid of the notion of .E files as .i is the proper usage
Mike Frysinger [Fri, 17 Nov 2006 05:55:24 +0000 (05:55 -0000)]
get rid of the notion of .E files as .i is the proper usage

17 years agorevert PTHREAD_THREADS_MAX back to former size since the large majority of people...
Mike Frysinger [Fri, 17 Nov 2006 05:50:31 +0000 (05:50 -0000)]
revert PTHREAD_THREADS_MAX back to former size since the large majority of people out there will not need more than 1k threads (and people who do can change this themselves)

17 years agofix comment typos
Mike Frysinger [Thu, 16 Nov 2006 23:29:14 +0000 (23:29 -0000)]
fix comment typos

17 years agoadd a test for making sure vfork/signals are handled properly across different setjmp...
Mike Frysinger [Thu, 16 Nov 2006 23:09:36 +0000 (23:09 -0000)]
add a test for making sure vfork/signals are handled properly across different setjmp/longjmps

17 years agoexercise the signal related jump code
Mike Frysinger [Thu, 16 Nov 2006 21:50:57 +0000 (21:50 -0000)]
exercise the signal related jump code

17 years agomake default output silence and have the exit status indicate error
Mike Frysinger [Thu, 16 Nov 2006 21:47:57 +0000 (21:47 -0000)]
make default output silence and have the exit status indicate error

17 years agorevert rev 16527 changes to ldso/ldso/dl-hash.c
Eric Andersen [Thu, 16 Nov 2006 20:52:28 +0000 (20:52 -0000)]
revert rev 16527 changes to ldso/ldso/dl-hash.c

17 years agoPaul Brook writes:
Eric Andersen [Thu, 16 Nov 2006 17:09:49 +0000 (17:09 -0000)]
Paul Brook writes:

The patch below fixes a bug in the new ARM _ld_linux_resolve implementation.
I'm don't know if/how the current implementation was tested, but it's
completely broken.

This patch makes the prologue and epilogue agree on how big the stack frame
is, and also makes sure EABI doubleword stack alignment is preserved.

Tested on arm-linux-gnueabi.

17 years agoThis change reimplements the ARM _dl_linux_resolve entry point - this is
Eric Andersen [Fri, 10 Nov 2006 20:41:49 +0000 (20:41 -0000)]
This change reimplements the ARM _dl_linux_resolve entry point - this is
called to resolve DLL PLT entries.  The assembler is changed to be thumb
compatible and slightly faster, the C function, _dl_linux_resolver (note
the extra r) is changed to take a byte address in place of an 8 byte
count (faster in caller and callee, and slightly easier to understand).

17 years agoarm thumb:
Eric Andersen [Fri, 10 Nov 2006 20:37:45 +0000 (20:37 -0000)]
arm thumb:
Put the call_via_rx code into each executable - call_via_ip cannot
possibly work if called through the PLT!  ldso requires this code
too as it is not linked with the crt stuff and thumb ldso does
make calls via a register.

The patch puts the code into crti.S so that it is linked into
every normally built application (if thumb or interworking is
selected).  This is only 30 extra bytes and it works - the previous
code did not because nothing both implemented and exported the
APIs (they were in libgcc, but not in the version script).

crti.S and crtn.S is also brought up to date with GCC 3.4.4 - this
is essential for thumb support because the .init and .fini sections
must use arm or thumb code to match the compilation of the libraries.

Note that code which pushes stuff into .init or .fini must be
compiled with or without -mthumb to match the uclibc compilation -
and gcc itself (which does do this) must therefore be compiled to
match.

17 years agoprevent cancellation functions from being called more than once
Mike Frysinger [Thu, 9 Nov 2006 08:14:00 +0000 (08:14 -0000)]
prevent cancellation functions from being called more than once

17 years agobackport from upstream:
Mike Frysinger [Thu, 9 Nov 2006 08:11:33 +0000 (08:11 -0000)]
backport from upstream:
2001-04-10  Ulrich Drepper  <drepper@redhat.com>

    * join.c (pthread_exit): Move code to new function __pthread_do_exit
    which takes an extra parameter with the current frame pointer.
    Call new function with CURRENT_STACK_FRAME.
    (__pthread_do_exit): New function.  Call __pthread_perform_cleanup
    with the new parameter.
    (pthread_join): Call __pthread_do_exit instead of pthread_exit.
    * cancel.c (__pthread_perform_cleanup): Takes extra parameter.  Use
    this parameter as the initial value the cleanup handler records are
    compared against.  No active cleanup handler record must have an
    address lower than the previous one and the initial record must be
    above (below on PA) the frame address passed in.
    (pthread_setcancelstate): Call __pthread_do_exit instead of
    pthread_exit.
    (pthread_setcanceltype): Likewise.
    (pthread_testcancel): Likewise.
    (_pthread_cleanup_pop_restore): Likewise.
    * condvar.c (pthread_cond_wait): Likewise.
    (pthread_cond_timedwait_relative): Likewise.
    * manager.c (pthread_start_thread): Likewise.
    * oldsemaphore.c (__old_sem_wait): Likewise.
    * pthread.c (pthread_handle_sigcancel): Likewise.
    * semaphore.c (__new_sem_wait): Likewise.
    (sem_timedwait): Likewise.
    * ptlongjmp.c (pthread_cleanup_upto): Also use current stack frame
    to limit the cleanup handlers which get run.
    * internals.h: Add prototype for __pthread_do_exit.  Adjust prototype
    for __pthread_perform_cleanup.

17 years agobackport from upstream:
Mike Frysinger [Thu, 9 Nov 2006 08:07:43 +0000 (08:07 -0000)]
backport from upstream:
2001-04-11  Ulrich Drepper  <drepper@redhat.com>

    * cancel.c (_pthread_cleanup_push): Catch invalid __prev buffer
    and remove it.
    (_pthread_cleanup_push_defer): Likewise.

17 years agoI failed to check the patch adding support for the new (stupid) US 2007
Eric Andersen [Tue, 7 Nov 2006 20:50:21 +0000 (20:50 -0000)]
I failed to check the patch adding support for the new (stupid) US 2007
daylight savings time rules when xlocale was enabled.  Fix compile error.

17 years agomips64 patch from Atsushi Nemoto:
Eric Andersen [Mon, 6 Nov 2006 17:49:24 +0000 (17:49 -0000)]
mips64 patch from Atsushi Nemoto:
We should not use old stype ipc_perm for 64-bit mips kernel.

I suppose other 64-bit platforms shoule define __IPC_64 also and we
can define it unconditionally, but not sure for other archs ...

17 years agomips64 patch from Atsushi Nemoto:
Eric Andersen [Mon, 6 Nov 2006 17:46:44 +0000 (17:46 -0000)]
mips64 patch from Atsushi Nemoto:
We should use "PTR_LA" instead of "la" for N64 ABI.