]> git.ipfire.org Git - thirdparty/valgrind.git/log
thirdparty/valgrind.git
20 years agoOops; forgot to update spec
Julian Seward [Tue, 19 Jul 2005 14:59:41 +0000 (14:59 +0000)] 
Oops; forgot to update spec

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4187

20 years agoNew command line option: --xml-user-comment=XMLTEXT, which allows
Julian Seward [Tue, 19 Jul 2005 14:18:56 +0000 (14:18 +0000)] 
New command line option: --xml-user-comment=XMLTEXT, which allows
copying of arbitrary bits of XML text to the XML output.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4186

20 years agoNew printf format specifier %t, which is like %s but escapes
Julian Seward [Tue, 19 Jul 2005 14:17:37 +0000 (14:17 +0000)] 
New printf format specifier %t, which is like %s but escapes
XML-unfriendly chars and thereby helps in the construction of valid
XML output.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4185

20 years agoFix the mmap2 pre-handler to avoid corrupting the arguments when
Tom Hughes [Tue, 19 Jul 2005 13:04:04 +0000 (13:04 +0000)] 
Fix the mmap2 pre-handler to avoid corrupting the arguments when
it fails with ENOMEM due to a lack of memory.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4184

20 years agoShow Valgrind's args too in the XML output.
Julian Seward [Tue, 19 Jul 2005 12:39:11 +0000 (12:39 +0000)] 
Show Valgrind's args too in the XML output.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4183

20 years agoAdd time stamps to XML START/FINISH indications.
Julian Seward [Tue, 19 Jul 2005 12:17:51 +0000 (12:17 +0000)] 
Add time stamps to XML START/FINISH indications.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4182

20 years agoMove the code for generating a human-readable time string into its own
Julian Seward [Tue, 19 Jul 2005 12:17:05 +0000 (12:17 +0000)] 
Move the code for generating a human-readable time string into its own
function.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4181

20 years agoNew command line option: --log-file-qualifier=VAR. When specified,
Julian Seward [Tue, 19 Jul 2005 11:25:02 +0000 (11:25 +0000)] 
New command line option: --log-file-qualifier=VAR.  When specified,
the contents of environment variable VAR (viz, $VAR) are incorporated
into logfile names.  This makes it easy to incorporate environmental
information into logfile names.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4180

20 years agoAdd some assertions.
Tom Hughes [Tue, 19 Jul 2005 07:37:51 +0000 (07:37 +0000)] 
Add some assertions.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4179

20 years agoEnable the rt_sigqueueinfo syscall on amd64.
Tom Hughes [Tue, 19 Jul 2005 07:21:06 +0000 (07:21 +0000)] 
Enable the rt_sigqueueinfo syscall on amd64.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4178

20 years agowibble
Nicholas Nethercote [Tue, 19 Jul 2005 03:38:09 +0000 (03:38 +0000)] 
wibble

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4177

20 years agoAdded a file containing Josef's explanation of Calltree's function
Nicholas Nethercote [Tue, 19 Jul 2005 03:30:31 +0000 (03:30 +0000)] 
Added a file containing Josef's explanation of Calltree's function
entry/exit tracking.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4176

20 years agoRemove unneeded #ifdef.
Nicholas Nethercote [Tue, 19 Jul 2005 01:22:26 +0000 (01:22 +0000)] 
Remove unneeded #ifdef.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4175

20 years agoAdd note about justified pedantry of overlap checking.
Nicholas Nethercote [Tue, 19 Jul 2005 00:48:55 +0000 (00:48 +0000)] 
Add note about justified pedantry of overlap checking.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4174

20 years agowibbles
Nicholas Nethercote [Tue, 19 Jul 2005 00:40:16 +0000 (00:40 +0000)] 
wibbles

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4173

20 years agoRemove valgrind.vs.
Nicholas Nethercote [Tue, 19 Jul 2005 00:36:23 +0000 (00:36 +0000)] 
Remove valgrind.vs.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4172

20 years agoallow to be build in a buildroot
Dirk Mueller [Mon, 18 Jul 2005 23:56:46 +0000 (23:56 +0000)] 
allow to be build in a buildroot

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4171

20 years agoremove the version script, doesn't work for executables
Dirk Mueller [Mon, 18 Jul 2005 23:52:10 +0000 (23:52 +0000)] 
remove the version script, doesn't work for executables

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4170

20 years agoFix crash when no environment is given to execve.
Tom Hughes [Mon, 18 Jul 2005 23:23:03 +0000 (23:23 +0000)] 
Fix crash when no environment is given to execve.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4169

20 years agoCheck members of the stack_t structure passed to sigaltstack
Tom Hughes [Mon, 18 Jul 2005 23:18:10 +0000 (23:18 +0000)] 
Check members of the stack_t structure passed to sigaltstack
individually to avoid problems with padding on 64 bit platforms.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4168

20 years agoCheck each member of the ifconf structure passed to SIOCGIFCONF
Tom Hughes [Mon, 18 Jul 2005 22:45:55 +0000 (22:45 +0000)] 
Check each member of the ifconf structure passed to SIOCGIFCONF
individually to avoid problems with padding bytes on 64 bit platforms.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4167

20 years agoCheck each field of the msghdr structure passed to sendmsg/recvmsg
Tom Hughes [Mon, 18 Jul 2005 22:41:33 +0000 (22:41 +0000)] 
Check each field of the msghdr structure passed to sendmsg/recvmsg
individually to avoid complaints due to uninitialised padding bytes
on 64 bit platforms.

Also fixed sendmsg to check things which should be initialised (the
msghdr structure and the iov array) properly instead of doing a write
check for everything.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4166

20 years agoUpdate svn:ignore lists.
Tom Hughes [Mon, 18 Jul 2005 17:53:07 +0000 (17:53 +0000)] 
Update svn:ignore lists.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4165

20 years agoIf the client program is a PIE executable, avoid mapping it at
Tom Hughes [Mon, 18 Jul 2005 15:52:30 +0000 (15:52 +0000)] 
If the client program is a PIE executable, avoid mapping it at
address zero. Instead do something like what the kernel does and
map it in the middle of the client address space.

Fix for bug #106283 based on patch from Sergey Vlasov.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4164

20 years agoadd comment
Nicholas Nethercote [Mon, 18 Jul 2005 14:10:12 +0000 (14:10 +0000)] 
add comment

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4163

20 years agoTake a copy of the environment given to execve before trying to mangle
Tom Hughes [Mon, 18 Jul 2005 14:02:11 +0000 (14:02 +0000)] 
Take a copy of the environment given to execve before trying to mangle
it as it might be in read only memory. Fixes bug #101881.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4162

20 years agoHandle the BLKGETSIZE64 ioctl. Fixes bug #104797.
Tom Hughes [Mon, 18 Jul 2005 13:26:55 +0000 (13:26 +0000)] 
Handle the BLKGETSIZE64 ioctl. Fixes bug #104797.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4161

20 years agoPreserve %esi across VG_(clone) as the caller may be relying on it
Tom Hughes [Mon, 18 Jul 2005 12:02:45 +0000 (12:02 +0000)] 
Preserve %esi across VG_(clone) as the caller may be relying on it
not being changed. Fix to bug #106713.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4160

20 years agoTrack API changes in vex r1272.
Julian Seward [Mon, 18 Jul 2005 11:41:15 +0000 (11:41 +0000)] 
Track API changes in vex r1272.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4159

20 years agoAllow a NULL argument to the times system call. Fixes bug #103509.
Tom Hughes [Mon, 18 Jul 2005 11:03:35 +0000 (11:03 +0000)] 
Allow a NULL argument to the times system call. Fixes bug #103509.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4158

20 years agoFixed debug code so that it compiles and fixed a printf to report the
Tom Hughes [Mon, 18 Jul 2005 09:41:57 +0000 (09:41 +0000)] 
Fixed debug code so that it compiles and fixed a printf to report the
start and end as (apparently) intended.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4157

20 years agoConsider blocks with both SF_MMAP and SF_STACK set as part of the root
Tom Hughes [Mon, 18 Jul 2005 09:40:29 +0000 (09:40 +0000)] 
Consider blocks with both SF_MMAP and SF_STACK set as part of the root
memory set as any additional stack space allocated by valgrind will
have both flags set. This fixes bug #106293.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4156

20 years agotweak
Nicholas Nethercote [Sun, 17 Jul 2005 23:12:33 +0000 (23:12 +0000)] 
tweak

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4155

20 years agoIntroduced get_pszB() to cover several common cases.
Nicholas Nethercote [Sun, 17 Jul 2005 18:12:00 +0000 (18:12 +0000)] 
Introduced get_pszB() to cover several common cases.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4154

20 years agotweaks
Nicholas Nethercote [Sun, 17 Jul 2005 18:00:57 +0000 (18:00 +0000)] 
tweaks

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4153

20 years agoIntroduce get_bszB() and get_bszB_as_is() to factor out some common
Nicholas Nethercote [Sun, 17 Jul 2005 17:55:42 +0000 (17:55 +0000)] 
Introduce get_bszB() and get_bszB_as_is() to factor out some common
combinations of function calls.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4152

20 years agoReplace is_inuse_bszB() with is_inuse_block(), which is higher-level
Nicholas Nethercote [Sun, 17 Jul 2005 17:20:30 +0000 (17:20 +0000)] 
Replace is_inuse_bszB() with is_inuse_block(), which is higher-level
and thus makes the code that uses it easier to understand.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4151

20 years agoRename 'other' as 'other_b' to make the fact it's a block more clear.
Nicholas Nethercote [Sun, 17 Jul 2005 17:12:24 +0000 (17:12 +0000)] 
Rename 'other' as 'other_b' to make the fact it's a block more clear.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4150

20 years agoMake VG_(percentify)() cope if the denominator is zero.
Nicholas Nethercote [Sun, 17 Jul 2005 16:16:41 +0000 (16:16 +0000)] 
Make VG_(percentify)() cope if the denominator is zero.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4149

20 years agoFix comments.
Nicholas Nethercote [Sun, 17 Jul 2005 16:12:59 +0000 (16:12 +0000)] 
Fix comments.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4148

20 years agoxml output for tool=none is fine too.
Cerion Armour-Brown [Wed, 13 Jul 2005 14:18:24 +0000 (14:18 +0000)] 
xml output for tool=none is fine too.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4147

20 years agoFixed bug #88678 -- debug info is now correctly gathered for files
Nicholas Nethercote [Tue, 12 Jul 2005 23:28:59 +0000 (23:28 +0000)] 
Fixed bug #88678 -- debug info is now correctly gathered for files
containing spaces.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4146

20 years agoRemove dead code.
Nicholas Nethercote [Sun, 10 Jul 2005 16:57:24 +0000 (16:57 +0000)] 
Remove dead code.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4145

20 years agoRemove very old info about the CVS repo at Sourceforge.
Nicholas Nethercote [Sun, 10 Jul 2005 15:10:29 +0000 (15:10 +0000)] 
Remove very old info about the CVS repo at Sourceforge.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4144

20 years agoGet rid of endianness assumptions in the PRRAn macro, which is
Julian Seward [Sun, 10 Jul 2005 00:55:26 +0000 (00:55 +0000)] 
Get rid of endianness assumptions in the PRRAn macro, which is
important for doing checks of scalar syscall args.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4143

20 years agoClean up endianness macros a bit. Now pub_tool_basics.h defines
Julian Seward [Sun, 10 Jul 2005 00:53:42 +0000 (00:53 +0000)] 
Clean up endianness macros a bit.  Now pub_tool_basics.h defines
VG_LITTLEENDIAN or VG_BIGENDIAN and that's what should be used.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4142

20 years agoFix up ppc32-linux stack unwinding.
Julian Seward [Sat, 9 Jul 2005 22:51:08 +0000 (22:51 +0000)] 
Fix up ppc32-linux stack unwinding.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4141

20 years agoAdd comments from PaulM.
Julian Seward [Sat, 9 Jul 2005 10:43:45 +0000 (10:43 +0000)] 
Add comments from PaulM.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4140

20 years agoAnother seemingly-small but critical change imported from 2.4.0-ppc:
Julian Seward [Sat, 9 Jul 2005 10:36:25 +0000 (10:36 +0000)] 
Another seemingly-small but critical change imported from 2.4.0-ppc:
don't read symbols from sections with w permissions.  For some reason
ppc-linux maps executables three times, not twice as on x86, and if
the w flag is ignored, we wind up reading debug info twice, getting
wrong redirections, and then it all goes to hell in a handbasket.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4139

20 years agoImport sigframe stuff from Paul's tree. With this and the other changes
Julian Seward [Fri, 8 Jul 2005 18:28:40 +0000 (18:28 +0000)] 
Import sigframe stuff from Paul's tree.  With this and the other changes
today, it's possible to run bash on ppc32-linux.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4138

20 years agoOops, this really should have been part of r4136.
Julian Seward [Fri, 8 Jul 2005 18:27:22 +0000 (18:27 +0000)] 
Oops, this really should have been part of r4136.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4137

20 years agoFix/reinstate a whole bunch of syscalls.
Julian Seward [Fri, 8 Jul 2005 18:26:37 +0000 (18:26 +0000)] 
Fix/reinstate a whole bunch of syscalls.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4136

20 years agoAssertion-failure-avoiding hacks from Paul's tree. I don't know why
Julian Seward [Fri, 8 Jul 2005 18:25:13 +0000 (18:25 +0000)] 
Assertion-failure-avoiding hacks from Paul's tree.  I don't know why
these are necessary, but they are.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4135

20 years agoA longer gdb-attach delay for ppc
Julian Seward [Fri, 8 Jul 2005 18:24:04 +0000 (18:24 +0000)] 
A longer gdb-attach delay for ppc

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4134

20 years agoCheck the obvious ..
Julian Seward [Fri, 8 Jul 2005 18:23:40 +0000 (18:23 +0000)] 
Check the obvious ..

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4133

20 years agoComment wibble.
Julian Seward [Fri, 8 Jul 2005 09:46:53 +0000 (09:46 +0000)] 
Comment wibble.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4132

20 years agoDon't bomb cachegrind at startup.
Julian Seward [Fri, 8 Jul 2005 09:45:43 +0000 (09:45 +0000)] 
Don't bomb cachegrind at startup.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4131

20 years agoAdd a simple random number generator to m_libcbase so we don't have
Nicholas Nethercote [Fri, 8 Jul 2005 04:08:59 +0000 (04:08 +0000)] 
Add a simple random number generator to m_libcbase so we don't have
to use the one from glibc.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4130

20 years agoCreate and use bigendian versions of MC_(helperc_{LOAD,STORE}V{2,4,8}).
Julian Seward [Fri, 8 Jul 2005 01:29:33 +0000 (01:29 +0000)] 
Create and use bigendian versions of MC_(helperc_{LOAD,STORE}V{2,4,8}).
This involved some serious nastyness from the Department of Cpp Abuse.

Memcheck still bombs on ppc32 for unknown reasons.

There are still endianness issues within these functions, I think.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4129

20 years agoEnable a couple more syscalls.
Julian Seward [Fri, 8 Jul 2005 00:18:25 +0000 (00:18 +0000)] 
Enable a couple more syscalls.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4128

20 years agoAdd a nested-functions test from Lee Kindness.
Julian Seward [Thu, 7 Jul 2005 14:29:21 +0000 (14:29 +0000)] 
Add a nested-functions test from Lee Kindness.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4127

20 years agoDon't allow vex to chase into any block for which we might want to create
Julian Seward [Thu, 7 Jul 2005 13:52:53 +0000 (13:52 +0000)] 
Don't allow vex to chase into any block for which we might want to create
a self-checking translation.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4126

20 years agoSelf-modifying-code check for amd64.
Julian Seward [Thu, 7 Jul 2005 13:20:31 +0000 (13:20 +0000)] 
Self-modifying-code check for amd64.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4125

20 years agoUpdate expected output.
Julian Seward [Thu, 7 Jul 2005 11:32:37 +0000 (11:32 +0000)] 
Update expected output.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4124

20 years agoModify this test so it no longer uses client requests, but instead
Julian Seward [Thu, 7 Jul 2005 11:24:14 +0000 (11:24 +0000)] 
Modify this test so it no longer uses client requests, but instead
relies on --smc-support=all to work correctly.  Hence it tests the
s-m-c support at least on x86.  Jump through various hoops to defeat
vex's basic-block-chasing optimisation, which has an annoying habit of
making this test work correctly even without --smc-support=all.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4123

20 years agoSupport for self modifying code on unfriendly platforms (x86, amd64)
Julian Seward [Thu, 7 Jul 2005 10:05:05 +0000 (10:05 +0000)] 
Support for self modifying code on unfriendly platforms (x86, amd64)
via the use of self-checking translations.  (Friendly platforms which
have icache-invalidation instructions we can observe, such as ppc32,
are already handled correctly.)  This should finally fix the
longstanding problem of V incorrectly handling calls of statically
nested functions (a gcc extension), and more generally make it a lot
easier to use V to debug dynamic code generation systems.

Since self-checking is a large performance overhead, there is some
control via a command line flag:

   --smc-support=none

      Don't make any translations self-checking.

   --smc-support=stack

      Add checking code for translations taken from segments which
      have the SF_GROWDOWN flag set -- stacks, basically.
      This is the default.  It should make gcc nested functions and
      GNU Ada work correctly with no intervention from the user.

   --smc-support=all

      Make all translations self-checking.  This is expensive and
      you want to do this if you're debugging a JIT compiler or
      some such.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4122

20 years agorm unused code
Julian Seward [Thu, 7 Jul 2005 08:22:51 +0000 (08:22 +0000)] 
rm unused code

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4121

20 years agoThis is x86 specific; re-home accordingly.
Julian Seward [Wed, 6 Jul 2005 20:22:16 +0000 (20:22 +0000)] 
This is x86 specific; re-home accordingly.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4120

20 years agosmc1 is an x86-specific test; re-home it accordingly. (How did this
Julian Seward [Wed, 6 Jul 2005 20:03:43 +0000 (20:03 +0000)] 
smc1 is an x86-specific test; re-home it accordingly.  (How did this
one evade detection for so long?)

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4119

20 years ago--time-stamp=yes is incompatible with --xml=yes, at least for the moment.
Julian Seward [Wed, 6 Jul 2005 19:46:48 +0000 (19:46 +0000)] 
--time-stamp=yes is incompatible with --xml=yes, at least for the moment.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4118

20 years agoScan the entire BB looking for "bogus literals"* before instrumenting
Julian Seward [Wed, 6 Jul 2005 19:42:23 +0000 (19:42 +0000)] 
Scan the entire BB looking for "bogus literals"* before instrumenting
any of it, so as to avoid any problems arising from switching from one
scheme to the other half-way through.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4117

20 years agoAdd a regression test for memcheck's ability do deal with partially
Julian Seward [Wed, 6 Jul 2005 19:01:53 +0000 (19:01 +0000)] 
Add a regression test for memcheck's ability do deal with partially
defined operands in equality comparisons.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4116

20 years agoExtensively re-analyse, re-check and revise the scheme for expensive
Julian Seward [Wed, 6 Jul 2005 18:48:59 +0000 (18:48 +0000)] 
Extensively re-analyse, re-check and revise the scheme for expensive
handling of integer EQ/NE, which can sometimes do better than the
naive scheme when the inputs are partially defined.  I never was
convinced it was correct before, but now I am.  Regtest to follow.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4115

20 years agoRename VG_(get_obj)() as VG_(find_seginfo) to be more consistent
Nicholas Nethercote [Wed, 6 Jul 2005 13:19:11 +0000 (13:19 +0000)] 
Rename VG_(get_obj)() as VG_(find_seginfo) to be more consistent
with the other function names.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4114

20 years agoThis test is nonsensical for ppc; it only applies to x86 and amd64.
Julian Seward [Wed, 6 Jul 2005 08:38:19 +0000 (08:38 +0000)] 
This test is nonsensical for ppc; it only applies to x86 and amd64.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4113

20 years agoDefine the system call numbers for pread64 and pwrite64.
Tom Hughes [Wed, 6 Jul 2005 08:17:07 +0000 (08:17 +0000)] 
Define the system call numbers for pread64 and pwrite64.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4112

20 years agoEnable some more amd64 syscalls.
Tom Hughes [Tue, 5 Jul 2005 23:34:44 +0000 (23:34 +0000)] 
Enable some more amd64 syscalls.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4111

20 years agoSort out the mess that is pread64/pwrite64 properly. All three platforms
Tom Hughes [Tue, 5 Jul 2005 23:25:17 +0000 (23:25 +0000)] 
Sort out the mess that is pread64/pwrite64 properly. All three platforms
that we currently support use the same handlers in the kernel without any
platform specific wrappers.

The final argument is a 64 bit argument however, which means that it
requires two registers on x86 and ppc32 and only one on amd64. The
reason it works in the kernel is that x86 and ppc32 calling conventions
inside the kernel work out correctly and the values get joined together.

For our purposes we make x86 and ppc32 use the generic veneer with
five arguments and amd64 use a platform specific one with four...

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4110

20 years agoset default fpu, altivec control words in dispatch-ppc.S (both zero)
Cerion Armour-Brown [Tue, 5 Jul 2005 18:24:22 +0000 (18:24 +0000)] 
set default fpu, altivec control words in dispatch-ppc.S (both zero)

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4109

20 years agoDisable PIE by default (sorry Tom), even on PIE-enabled platforms. It
Julian Seward [Tue, 5 Jul 2005 13:44:21 +0000 (13:44 +0000)] 
Disable PIE by default (sorry Tom), even on PIE-enabled platforms.  It
causes too much breakage.  PIE builds are still possible, but you have
to say --enable-pie to get them now.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4108

20 years agoFix quoting some more. Shell script sucks.
Julian Seward [Tue, 5 Jul 2005 00:59:08 +0000 (00:59 +0000)] 
Fix quoting some more.  Shell script sucks.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4107

20 years agoFix quoting a bit.
Julian Seward [Tue, 5 Jul 2005 00:00:40 +0000 (00:00 +0000)] 
Fix quoting a bit.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4106

20 years agoCheck args a bit.
Julian Seward [Mon, 4 Jul 2005 23:54:41 +0000 (23:54 +0000)] 
Check args a bit.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4105

20 years agoAdd a test script (recycled version of Tom's nightly/bin/nightly)
Julian Seward [Mon, 4 Jul 2005 23:44:10 +0000 (23:44 +0000)] 
Add a test script (recycled version of Tom's nightly/bin/nightly)
which is useful for doing automated test runs against the GNU
Scientific Library v 1.6 (gsl-1.6).  This has proven very helpful in
shaking out Vex simulation bugs.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4104

20 years agoTweak debug output.
Nicholas Nethercote [Sun, 3 Jul 2005 20:22:39 +0000 (20:22 +0000)] 
Tweak debug output.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4103

20 years agoComment changes only.
Nicholas Nethercote [Sun, 3 Jul 2005 20:22:07 +0000 (20:22 +0000)] 
Comment changes only.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4102

20 years agoRemove dead function.
Nicholas Nethercote [Sun, 3 Jul 2005 18:45:37 +0000 (18:45 +0000)] 
Remove dead function.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4101

20 years agoChanged m_hashtable.c to allow the size of the hash table to be specified
Nicholas Nethercote [Sun, 3 Jul 2005 17:53:11 +0000 (17:53 +0000)] 
Changed m_hashtable.c to allow the size of the hash table to be specified
when it is created.  Fortunately this didn't affect code outside this
module except for the calls to VG_(HT_construct)().

As a result, we save some memory because not all tables have to be as big
as the ones needed for malloc/free tracking.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4100

20 years agowhoops
Nicholas Nethercote [Sun, 3 Jul 2005 17:10:04 +0000 (17:10 +0000)] 
whoops

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4099

20 years agoRemove now-obsolete .supp file for glibc-2.1.
Nicholas Nethercote [Sun, 3 Jul 2005 16:08:58 +0000 (16:08 +0000)] 
Remove now-obsolete .supp file for glibc-2.1.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4098

20 years agoThese directories no longer exist.
Nicholas Nethercote [Sun, 3 Jul 2005 16:08:12 +0000 (16:08 +0000)] 
These directories no longer exist.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4097

20 years agogcc-4.0.1 rejects insn_{mmx,sse,sse2}.c despite the fact that
Julian Seward [Sun, 3 Jul 2005 10:41:02 +0000 (10:41 +0000)] 
gcc-4.0.1 rejects insn_{mmx,sse,sse2}.c despite the fact that
gcc-4.0.0 accepted it.  This commit adds flags to placate 4.0.1 (from
Dirk Mueller).

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4096

20 years agoglibc-2.1 is now so ancient that it's probably untrue we still support it.
Julian Seward [Sun, 3 Jul 2005 10:18:33 +0000 (10:18 +0000)] 
glibc-2.1 is now so ancient that it's probably untrue we still support it.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4095

20 years agoUpdate ignores.
Robert Walsh [Sun, 3 Jul 2005 06:07:16 +0000 (06:07 +0000)] 
Update ignores.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4094

20 years agoRemove trailing slash (some part of the autoswamp didn't like it)
Julian Seward [Sun, 3 Jul 2005 00:50:48 +0000 (00:50 +0000)] 
Remove trailing slash (some part of the autoswamp didn't like it)

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4093

20 years agoRemoved the remnants of the attempt at an ARM port, because it had
Nicholas Nethercote [Sat, 2 Jul 2005 23:13:59 +0000 (23:13 +0000)] 
Removed the remnants of the attempt at an ARM port, because it had
bit-rotted badly and was clogging up the code.

I put the useful remnants in docs/porting-to-ARM in case anyone ever
wants to try porting to ARM again.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4092

20 years agoMake the unique tags printed on XML errors actually unique.
Julian Seward [Sat, 2 Jul 2005 21:16:30 +0000 (21:16 +0000)] 
Make the unique tags printed on XML errors actually unique.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4091

20 years agoxml1: add another .stderr.exp file for slightly different systems, and
Nicholas Nethercote [Sat, 2 Jul 2005 21:14:19 +0000 (21:14 +0000)] 
xml1:  add another .stderr.exp file for slightly different systems, and
make the stderr.exp64 consistent with the improved filtering.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4090

20 years agoUpdate expected output. Hopefully this won't be so fragile now
Julian Seward [Sat, 2 Jul 2005 20:43:05 +0000 (20:43 +0000)] 
Update expected output.  Hopefully this won't be so fragile now
(famous last words).

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4089

20 years agoImprove filtering to get rid of suppressions, using magic incantations
Julian Seward [Sat, 2 Jul 2005 20:41:25 +0000 (20:41 +0000)] 
Improve filtering to get rid of suppressions, using magic incantations
from Tom H and Ashley Pittman.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@4088