]> git.ipfire.org Git - thirdparty/valgrind.git/log
thirdparty/valgrind.git
20 years agoStuff about core-dump reinstate-ment in the manual and release notes.
Nicholas Nethercote [Fri, 18 Nov 2005 17:36:01 +0000 (17:36 +0000)] 
Stuff about core-dump reinstate-ment in the manual and release notes.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5193

20 years agoMake coredump files have names like vgcore.12345 rather than vgcore.pid12345
Nicholas Nethercote [Fri, 18 Nov 2005 17:12:26 +0000 (17:12 +0000)] 
Make coredump files have names like vgcore.12345 rather than vgcore.pid12345
to be consistent with all the other such cases (eg. --log-file names,
cachegrind.out.12345).

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5192

20 years agosuppression for Ubuntu with a stripped ld.so.
Nicholas Nethercote [Fri, 18 Nov 2005 17:09:36 +0000 (17:09 +0000)] 
suppression for Ubuntu with a stripped ld.so.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5191

20 years agoBack out r5180 (sorry Josef). On consideration it just increases the difficulty
Julian Seward [Fri, 18 Nov 2005 15:10:24 +0000 (15:10 +0000)] 
Back out r5180 (sorry Josef).  On consideration it just increases the difficulty
of testing the system properly.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5190

20 years agoUpdate documentation about --show-below-main.
Nicholas Nethercote [Fri, 18 Nov 2005 14:59:00 +0000 (14:59 +0000)] 
Update documentation about --show-below-main.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5189

20 years agoLast of the __libc_start_main fixes.
Tom Hughes [Fri, 18 Nov 2005 09:39:09 +0000 (09:39 +0000)] 
Last of the __libc_start_main fixes.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5188

20 years agoYet more __libc_start_main updates.
Tom Hughes [Fri, 18 Nov 2005 09:30:15 +0000 (09:30 +0000)] 
Yet more __libc_start_main updates.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5187

20 years agoUpdate XML results for __libc_start_main change.
Tom Hughes [Fri, 18 Nov 2005 09:26:14 +0000 (09:26 +0000)] 
Update XML results for __libc_start_main change.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5186

20 years agoFix argument count given to makecontext.
Tom Hughes [Fri, 18 Nov 2005 09:24:05 +0000 (09:24 +0000)] 
Fix argument count given to makecontext.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5185

20 years agoMore __libc_start_main updates.
Tom Hughes [Fri, 18 Nov 2005 09:23:28 +0000 (09:23 +0000)] 
More __libc_start_main updates.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5184

20 years agoUpdate some expected results in line with Julian's fix to stop
Tom Hughes [Fri, 18 Nov 2005 09:03:47 +0000 (09:03 +0000)] 
Update some expected results in line with Julian's fix to stop
tracing the stack at __libc_start_main.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5183

20 years agoSuSE 10 (ppc32) suppressions.
Julian Seward [Fri, 18 Nov 2005 00:15:18 +0000 (00:15 +0000)] 
SuSE 10 (ppc32) suppressions.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5182

20 years agocomment about the test
Nicholas Nethercote [Thu, 17 Nov 2005 21:33:42 +0000 (21:33 +0000)] 
comment about the test

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5181

20 years ago(ppc32) Only include hacky redirects for malloc-replacing tools.
Julian Seward [Thu, 17 Nov 2005 20:15:04 +0000 (20:15 +0000)] 
(ppc32) Only include hacky redirects for malloc-replacing tools.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5180

20 years agowibl
Nicholas Nethercote [Thu, 17 Nov 2005 19:47:37 +0000 (19:47 +0000)] 
wibl

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5179

20 years agogcc isn't so hot at reading English text
Julian Seward [Thu, 17 Nov 2005 19:44:27 +0000 (19:44 +0000)] 
gcc isn't so hot at reading English text

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5178

20 years agoI couldn't resist
Nicholas Nethercote [Thu, 17 Nov 2005 19:40:24 +0000 (19:40 +0000)] 
I couldn't resist

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5177

20 years agocopy log message as a comment
Nicholas Nethercote [Thu, 17 Nov 2005 19:38:09 +0000 (19:38 +0000)] 
copy log message as a comment

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5176

20 years agodocument
Nicholas Nethercote [Thu, 17 Nov 2005 19:32:59 +0000 (19:32 +0000)] 
document

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5175

20 years agoAllocate stack with mmap rather than malloc, so that on ppc32 we get an executable
Julian Seward [Thu, 17 Nov 2005 19:32:07 +0000 (19:32 +0000)] 
Allocate stack with mmap rather than malloc, so that on ppc32 we get an executable
stack.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5174

20 years agoChange the cannot-detect-cache-config message on PPC32 so we can filter
Nicholas Nethercote [Thu, 17 Nov 2005 19:28:18 +0000 (19:28 +0000)] 
Change the cannot-detect-cache-config message on PPC32 so we can filter
for it more precisely and thus tell if the auto-detection ever fails on
x86/AMD64.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5173

20 years agoEnable mq syscalls (untested).
Julian Seward [Thu, 17 Nov 2005 19:09:02 +0000 (19:09 +0000)] 
Enable mq syscalls (untested).

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5172

20 years agoFixes to make the .pdf build using the toolchain on SuSE10 (from Donna).
Julian Seward [Thu, 17 Nov 2005 18:04:58 +0000 (18:04 +0000)] 
Fixes to make the .pdf build using the toolchain on SuSE10 (from Donna).

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5171

20 years agoMake VG_(apply_StackTrace) stop if it looks like we've overshot main().
Julian Seward [Thu, 17 Nov 2005 15:12:34 +0000 (15:12 +0000)] 
Make VG_(apply_StackTrace) stop if it looks like we've overshot main().

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5170

20 years agosys_tgkill: hand the syscall to the kernel in the standard way, rather
Julian Seward [Thu, 17 Nov 2005 14:26:52 +0000 (14:26 +0000)] 
sys_tgkill: hand the syscall to the kernel in the standard way, rather
than doing it inline.  Doing it inline screws up on ppc32-linux if
we're sending an async signal to ourselves (the same thread) because
the kernel immediately hands the signal to async_sighandler() which
then dies at the assertion that this thread's state is VgTs_WaitSys.
From which I conclude this wrapper has always had a race against the
kernel which did not show up on x86 or amd64.  (and/or that I don't
understand this stuff too well)

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5169

20 years agoConnect properly to the tgkill wrapper.
Julian Seward [Thu, 17 Nov 2005 14:22:22 +0000 (14:22 +0000)] 
Connect properly to the tgkill wrapper.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5168

20 years agoThis won't work on ppc32 unless you ask for an alt stack with x permissions,
Julian Seward [Thu, 17 Nov 2005 13:04:46 +0000 (13:04 +0000)] 
This won't work on ppc32 unless you ask for an alt stack with x permissions,
since signal returning requires execution of code on the stack.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5167

20 years agoGet rid of bogus assertion.
Julian Seward [Thu, 17 Nov 2005 13:03:42 +0000 (13:03 +0000)] 
Get rid of bogus assertion.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5166

20 years agoUpdate README_DEVELOPERS to describe how to debug valgrind more
Tom Hughes [Thu, 17 Nov 2005 12:31:12 +0000 (12:31 +0000)] 
Update README_DEVELOPERS to describe how to debug valgrind more
easily. Based on patch from Yao Qi <qiyaoltc@cn.ibm.com>.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5165

20 years agoRemoved vki_ntohs definition and the infrastructure that supported
Tom Hughes [Thu, 17 Nov 2005 12:14:42 +0000 (12:14 +0000)] 
Removed vki_ntohs definition and the infrastructure that supported
it as it is no longer used.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5164

20 years agoUse VG_(ntohl) and VG_(ntohs) to decode IP addresses and ports. Note
Tom Hughes [Thu, 17 Nov 2005 12:02:58 +0000 (12:02 +0000)] 
Use VG_(ntohl) and VG_(ntohs) to decode IP addresses and ports. Note
that this also required reversing the order of the arguments to the
print call as the previous ordering assumed that the address was still
byte swapped.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5163

20 years agoProvide a full set of ntohl/htonl/ntohs/htons routines in the
Tom Hughes [Thu, 17 Nov 2005 12:01:56 +0000 (12:01 +0000)] 
Provide a full set of ntohl/htonl/ntohs/htons routines in the
valgrind C library.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5162

20 years agoKludgey endianness fix (for printing only) to stop fdleak_ipv4 failing on
Julian Seward [Thu, 17 Nov 2005 11:45:28 +0000 (11:45 +0000)] 
Kludgey endianness fix (for printing only) to stop fdleak_ipv4 failing on
ppc32.  I'm sure there's a Right Way to fix it, and this isn't it.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5161

20 years agoEnable sys_lstat on ppc32.
Julian Seward [Thu, 17 Nov 2005 11:01:55 +0000 (11:01 +0000)] 
Enable sys_lstat on ppc32.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5160

20 years agoFilter out I-have-no-idea-what-your-caches-are message, so these tests
Julian Seward [Thu, 17 Nov 2005 10:56:08 +0000 (10:56 +0000)] 
Filter out I-have-no-idea-what-your-caches-are message, so these tests
don't fail on ppc32.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5159

20 years agoA couple of these syscalls fail differently when running on an amd64
Julian Seward [Thu, 17 Nov 2005 03:08:16 +0000 (03:08 +0000)] 
A couple of these syscalls fail differently when running on an amd64
in 32-bit mode.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5158

20 years agoRecord XML output format changes.
Julian Seward [Thu, 17 Nov 2005 00:51:36 +0000 (00:51 +0000)] 
Record XML output format changes.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5157

20 years agorecord more email comments
Nicholas Nethercote [Wed, 16 Nov 2005 23:54:46 +0000 (23:54 +0000)] 
record more email comments

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5156

20 years agoImplemented checks for FPSCR and VSCR on leaving dispatcher
Cerion Armour-Brown [Wed, 16 Nov 2005 20:22:11 +0000 (20:22 +0000)] 
Implemented checks for FPSCR and VSCR on leaving dispatcher
 - required flags: FPSCR[RM] == 0, VSCR[NJ] == 1

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5155

20 years agoMinor tweaks.
Nicholas Nethercote [Wed, 16 Nov 2005 20:12:22 +0000 (20:12 +0000)] 
Minor tweaks.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5154

20 years agoFix up ppc32-linux handlers for sys_sigreturn and sys_rt_sigreturn. I really
Julian Seward [Wed, 16 Nov 2005 19:15:33 +0000 (19:15 +0000)] 
Fix up ppc32-linux handlers for sys_sigreturn and sys_rt_sigreturn.  I really
have no idea how these work (what's there is a conceptual transcription of
what we have for x86-linux) and so if those of you more Skilled In The Art
Of Signals than myself would care to peer at it, that would be appreciated.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5153

20 years agoEven more news.
Julian Seward [Wed, 16 Nov 2005 19:13:13 +0000 (19:13 +0000)] 
Even more news.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5152

20 years agoAdded tests for latest altivec fp insns
Cerion Armour-Brown [Wed, 16 Nov 2005 18:58:26 +0000 (18:58 +0000)] 
Added tests for latest altivec fp insns
Fixed fp nan argument values
Fixed setting of non-java bit in VSCR to be on for all cases.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5151

20 years agoEdit the release notes a bit and add our huge list of fixed bugs.
Julian Seward [Wed, 16 Nov 2005 18:39:25 +0000 (18:39 +0000)] 
Edit the release notes a bit and add our huge list of fixed bugs.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5150

20 years agoImplemented most of the remaining altivec fp ops:
Cerion Armour-Brown [Wed, 16 Nov 2005 18:02:58 +0000 (18:02 +0000)] 
Implemented most of the remaining altivec fp ops:
rounds (vrfi*), converts (vctu/sxs, vcfu/sx)

git-svn-id: svn://svn.valgrind.org/vex/trunk@1462

20 years agoAdded fp vector conversion/rounding irops to memcheck
Cerion Armour-Brown [Wed, 16 Nov 2005 17:21:49 +0000 (17:21 +0000)] 
Added fp vector conversion/rounding irops to memcheck

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5149

20 years agoYet more irops, for fp vector conversion/rounding.
Cerion Armour-Brown [Wed, 16 Nov 2005 17:21:10 +0000 (17:21 +0000)] 
Yet more irops, for fp vector conversion/rounding.

git-svn-id: svn://svn.valgrind.org/vex/trunk@1461

20 years agoRecord previous log message as a comment in the code.
Nicholas Nethercote [Wed, 16 Nov 2005 17:16:10 +0000 (17:16 +0000)] 
Record previous log message as a comment in the code.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5148

20 years agoMake the data segment (heap) executable. LinuxThreads on ppc32 creates
Julian Seward [Wed, 16 Nov 2005 17:12:38 +0000 (17:12 +0000)] 
Make the data segment (heap) executable.  LinuxThreads on ppc32 creates
trampolines in this area.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5147

20 years agoSync with all the bugzilla emails I have.
Julian Seward [Wed, 16 Nov 2005 17:11:52 +0000 (17:11 +0000)] 
Sync with all the bugzilla emails I have.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5146

20 years agoUpdate XML results.
Tom Hughes [Wed, 16 Nov 2005 16:54:02 +0000 (16:54 +0000)] 
Update XML results.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5145

20 years agoEnable sys_rt_sigsuspend. This is needed by LinuxThreads.
Julian Seward [Wed, 16 Nov 2005 12:51:34 +0000 (12:51 +0000)] 
Enable sys_rt_sigsuspend.  This is needed by LinuxThreads.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5144

20 years agoBug status updates.
Tom Hughes [Wed, 16 Nov 2005 10:23:15 +0000 (10:23 +0000)] 
Bug status updates.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5143

20 years agoSlightly reorder the preamble-printing order, and also print the CPU arch/subarch...
Julian Seward [Wed, 16 Nov 2005 03:51:02 +0000 (03:51 +0000)] 
Slightly reorder the preamble-printing order, and also print the CPU arch/subarch detected.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5142

20 years agoMove VG_(am_get_filename) to the tool accessible aspacemgr header file.
Tom Hughes [Wed, 16 Nov 2005 00:11:14 +0000 (00:11 +0000)] 
Move VG_(am_get_filename) to the tool accessible aspacemgr header file.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5141

20 years agoReinstate code to extent SegInfo ranges to cover all PT_LOAD segments
Tom Hughes [Wed, 16 Nov 2005 00:09:15 +0000 (00:09 +0000)] 
Reinstate code to extent SegInfo ranges to cover all PT_LOAD segments
when VG_(needs_data_syms) has been called by the tool.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5140

20 years agoUpdate bug status.
Tom Hughes [Wed, 16 Nov 2005 00:05:58 +0000 (00:05 +0000)] 
Update bug status.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5139

20 years agoFix stabs decoder to allow :: in a method name provided it is inside
Tom Hughes [Wed, 16 Nov 2005 00:04:58 +0000 (00:04 +0000)] 
Fix stabs decoder to allow :: in a method name provided it is inside
a template argument list. Fixes bug #113126.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5138

20 years agoComplete documentation trawl for 3.1.0.
Julian Seward [Tue, 15 Nov 2005 20:56:23 +0000 (20:56 +0000)] 
Complete documentation trawl for 3.1.0.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5137

20 years agowibble
Nicholas Nethercote [Tue, 15 Nov 2005 20:16:45 +0000 (20:16 +0000)] 
wibble

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5136

20 years agoUpdate manual for 3.1.0, sections <= manual-core.html.
Julian Seward [Tue, 15 Nov 2005 19:51:04 +0000 (19:51 +0000)] 
Update manual for 3.1.0, sections <= manual-core.html.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5135

20 years agoIncremented the xml output protocolversion to 2, following the change in the time...
Cerion Armour-Brown [Tue, 15 Nov 2005 19:07:53 +0000 (19:07 +0000)] 
Incremented the xml output protocolversion to 2, following the change in the time field format of r5042

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5134

20 years agoMore detail about pre-release testing.
Nicholas Nethercote [Tue, 15 Nov 2005 15:27:06 +0000 (15:27 +0000)] 
More detail about pre-release testing.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5133

20 years agoImplement SSE2 'clflush'.
Julian Seward [Tue, 15 Nov 2005 11:16:30 +0000 (11:16 +0000)] 
Implement SSE2 'clflush'.

git-svn-id: svn://svn.valgrind.org/vex/trunk@1460

20 years agodelete unused multiply primops
Julian Seward [Tue, 15 Nov 2005 10:21:19 +0000 (10:21 +0000)] 
delete unused multiply primops

git-svn-id: svn://svn.valgrind.org/vex/trunk@1459

20 years agoRemove horribly out-of-date comment.
Nicholas Nethercote [Tue, 15 Nov 2005 02:24:57 +0000 (02:24 +0000)] 
Remove horribly out-of-date comment.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5132

20 years agoHandle some ALSA ioctls that don't have size or direction hints.
Tom Hughes [Mon, 14 Nov 2005 17:03:54 +0000 (17:03 +0000)] 
Handle some ALSA ioctls that don't have size or direction hints.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5131

20 years agoIn XML mode, emit a closing </valgrindoutput> on various kinds of
Julian Seward [Mon, 14 Nov 2005 17:01:01 +0000 (17:01 +0000)] 
In XML mode, emit a closing </valgrindoutput> on various kinds of
failures, so as to help parsers stop parsing:

- after any kind of assertion failure or panic
- if suppression file is missing or has a syntax error

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5130

20 years agoMore man page updating, including:
Nicholas Nethercote [Mon, 14 Nov 2005 16:00:27 +0000 (16:00 +0000)] 
More man page updating, including:
- commented out stuff about addrcheck
- --log-file no longer puts "pid" in the filename
- other minor changes

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5129

20 years agoMake this test compile on my machine.
Nicholas Nethercote [Mon, 14 Nov 2005 15:35:20 +0000 (15:35 +0000)] 
Make this test compile on my machine.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5128

20 years agoEnable the prctl system call on amd64.
Tom Hughes [Mon, 14 Nov 2005 15:26:03 +0000 (15:26 +0000)] 
Enable the prctl system call on amd64.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5127

20 years agoSeparate the stack unwind logic for amd64 and x86, so that they can be
Julian Seward [Mon, 14 Nov 2005 15:18:25 +0000 (15:18 +0000)] 
Separate the stack unwind logic for amd64 and x86, so that they can be
differently performance-tuned.  amd64 needs to consult CFI first and
then if that fails (unlikely) follow the %rbp chain.  On x86, the CFI
is almost never helpful, but consulting it first wastes significant
time in allocation-intensive programs.  This commit pulls the two
archs apart and puts the CFI check second on x86.  This reduces start
time for ktuberling on x86 on memcheck from 78 seconds to 75.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5126

20 years agoOn amd64, when running allocation-intensive code in the presence of
Julian Seward [Mon, 14 Nov 2005 15:10:12 +0000 (15:10 +0000)] 
On amd64, when running allocation-intensive code in the presence of
many shared objects, finding the relevant CFI information for stack
unwinding becomes a significant performance overhead.  This change
slowly rearranges the SegInfo list to bring more popular entries to
the front during CFI lookup.  This reduces the startup time of
konqueror on memcheck on amd64 from 124 to 110 seconds.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5125

20 years agoUpdate manual page.
Tom Hughes [Mon, 14 Nov 2005 15:01:32 +0000 (15:01 +0000)] 
Update manual page.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5124

20 years agoDocument sign extension of si_code by the kernel.
Tom Hughes [Mon, 14 Nov 2005 14:37:02 +0000 (14:37 +0000)] 
Document sign extension of si_code by the kernel.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5123

20 years agoAdded tests for altivec load/stores
Cerion Armour-Brown [Mon, 14 Nov 2005 12:15:27 +0000 (12:15 +0000)] 
Added tests for altivec load/stores
Enabled altivec float tests

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5122

20 years agogcc4 picked up a typo.
Cerion Armour-Brown [Mon, 14 Nov 2005 03:32:23 +0000 (03:32 +0000)] 
gcc4 picked up a typo.

git-svn-id: svn://svn.valgrind.org/vex/trunk@1458

20 years agoCouple more irops for memcheck.
Cerion Armour-Brown [Mon, 14 Nov 2005 03:00:35 +0000 (03:00 +0000)] 
Couple more irops for memcheck.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5121

20 years agocomment out altivec tests for unsupported insns
Cerion Armour-Brown [Mon, 14 Nov 2005 02:43:09 +0000 (02:43 +0000)] 
comment out altivec tests for unsupported insns

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5120

20 years agoMore av insns: vmaddfp, vnmsubfp
Cerion Armour-Brown [Mon, 14 Nov 2005 02:37:44 +0000 (02:37 +0000)] 
More av insns: vmaddfp, vnmsubfp

Rough 'n ready IR used - results will be rounded along the way, not just at the end of the calculations, giving some error.

git-svn-id: svn://svn.valgrind.org/vex/trunk@1457

20 years agoFrontend
Cerion Armour-Brown [Mon, 14 Nov 2005 00:44:47 +0000 (00:44 +0000)] 
Frontend
--------
Added a bunch of altivec float insns:
vaddfp, vsubfp, vmaxfp, vminfp,
vrefp, vrsqrtefp
vcmpgefp, vcmpgtfp, vcmpbfp

Made use of fact that ppc backend for compare insns return
zero'd lanes if either of the corresponding args is a nan.
 - perhaps better to have an irop Iop_isNan32Fx4, but seems unecessary work until we get into running non-native code through vex.
 - better still, tighten down the spec for compare irops wrt nan

Backend
-------
Separated av float ops to own insn group - they're only ever type 32x4
Added av float unary insns
Added av float cmp insns - for irops that don't map directly to native insns, native behaviour wrt nan's is followed, requiring lane value==nan comparisons for each argument vector.

git-svn-id: svn://svn.valgrind.org/vex/trunk@1456

20 years agoNew irops: Iop_CmpGT32Fx4, Iop_CmpGE32Fx4
Cerion Armour-Brown [Mon, 14 Nov 2005 00:35:59 +0000 (00:35 +0000)] 
New irops: Iop_CmpGT32Fx4, Iop_CmpGE32Fx4

git-svn-id: svn://svn.valgrind.org/vex/trunk@1455

20 years agoMore profiling-induced speedups.
Julian Seward [Sun, 13 Nov 2005 20:30:24 +0000 (20:30 +0000)] 
More profiling-induced speedups.

git-svn-id: svn://svn.valgrind.org/vex/trunk@1454

20 years agoAdd some flag-specialisation cases that profiling showed the need for.
Julian Seward [Sun, 13 Nov 2005 19:51:04 +0000 (19:51 +0000)] 
Add some flag-specialisation cases that profiling showed the need for.

git-svn-id: svn://svn.valgrind.org/vex/trunk@1453

20 years agoHacks to enable self-hosting on amd64, so as to facilitate
Julian Seward [Sun, 13 Nov 2005 18:51:31 +0000 (18:51 +0000)] 
Hacks to enable self-hosting on amd64, so as to facilitate
cachegrinding it.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5119

20 years agowibble
Julian Seward [Sun, 13 Nov 2005 18:50:39 +0000 (18:50 +0000)] 
wibble

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5118

20 years agoInline cachesim_*_doref(). This gains about 5--10% in speed.
Nicholas Nethercote [Sun, 13 Nov 2005 17:57:32 +0000 (17:57 +0000)] 
Inline cachesim_*_doref().  This gains about 5--10% in speed.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5117

20 years agoMake the address space manager use fstat64 when it is available. There
Tom Hughes [Sun, 13 Nov 2005 16:52:56 +0000 (16:52 +0000)] 
Make the address space manager use fstat64 when it is available. There
are two reasons for this:

    - It can cope with manjor and minor device numbers outside the
      traditional 0-255 range.

    - It returns correct results for x86 binaries on amd64 systems
      where fstat returns uninitialised rubbish in the top 16 bits
      of the device number.

We also make the /proc/self/maps reading code encode device numbers in
the new style to cope with manjor and minor device numbers outside the
traditional 0-255 range.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5116

20 years agoaddition
Nicholas Nethercote [Sun, 13 Nov 2005 16:41:15 +0000 (16:41 +0000)] 
addition

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5115

20 years agoPreserve the sign of the bottom 16 bits of si_code when discarding
Tom Hughes [Sun, 13 Nov 2005 11:52:39 +0000 (11:52 +0000)] 
Preserve the sign of the bottom 16 bits of si_code when discarding
the top 16 bits - the pthread library uses negative code values in
the signals it sends between threads.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5114

20 years agoHacks needed for MontaVista Linux 3.1 (ppc32).
Julian Seward [Sun, 13 Nov 2005 02:42:23 +0000 (02:42 +0000)] 
Hacks needed for MontaVista Linux 3.1 (ppc32).

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5113

20 years agoEnable a couple more syscalls.
Julian Seward [Sun, 13 Nov 2005 02:41:58 +0000 (02:41 +0000)] 
Enable a couple more syscalls.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5112

20 years agoppc32 stack unwind: if the initial FP looks bad, don't use it.
Julian Seward [Sun, 13 Nov 2005 02:41:35 +0000 (02:41 +0000)] 
ppc32 stack unwind: if the initial FP looks bad, don't use it.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5111

20 years agoThe absolute bare minimum changes needed to make it work on an
Julian Seward [Sun, 13 Nov 2005 01:59:22 +0000 (01:59 +0000)] 
The absolute bare minimum changes needed to make it work on an
integer-only PPC processor (PPC440GX).

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5110

20 years agoUse revised PPC32 subarchitecture categories.
Julian Seward [Sun, 13 Nov 2005 00:53:33 +0000 (00:53 +0000)] 
Use revised PPC32 subarchitecture categories.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5109

20 years agoRevise the PPC32 subarchitecture kinds, so as to facilitated
Julian Seward [Sun, 13 Nov 2005 00:53:05 +0000 (00:53 +0000)] 
Revise the PPC32 subarchitecture kinds, so as to facilitated
supporting CPUs that have neither Altivec nor FPU.

git-svn-id: svn://svn.valgrind.org/vex/trunk@1452

20 years agoHook the ppc32 stuff up to the revised CPU detection machinery, and
Julian Seward [Sun, 13 Nov 2005 00:30:22 +0000 (00:30 +0000)] 
Hook the ppc32 stuff up to the revised CPU detection machinery, and
add a bunch of code to detect what the cpu can do at startup by
catching SIGILLs.  Shame PPC doesn't offer any sane mechanism for
finding out what instruction subsets the CPU is capable of (a la
x86/amd64 cpuid).

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5108

20 years agoUpdate bug status.
Tom Hughes [Sun, 13 Nov 2005 00:09:49 +0000 (00:09 +0000)] 
Update bug status.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5107

20 years agoEnable remaining *xattr system calls on amd64. Fixes bug #116200.
Tom Hughes [Sun, 13 Nov 2005 00:08:03 +0000 (00:08 +0000)] 
Enable remaining *xattr system calls on amd64. Fixes bug #116200.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5106

20 years agoMask out the top 16 bits of si_code at the start of the signal
Tom Hughes [Sun, 13 Nov 2005 00:01:20 +0000 (00:01 +0000)] 
Mask out the top 16 bits of si_code at the start of the signal
handlers because the kernel sometimes forgets to do so.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5105