Mark Wielaard [Thu, 28 Aug 2014 14:59:04 +0000 (14:59 +0000)]
Bug 338615 suppress glibc 2.20 optimized strcmp implementation for ARMv7.
Add an add_hardwired_spec for strcmp in VG_(redir_initialise) for
ld-linux.so.3 and ld-linux-armhf.so.3 to use a simple strcmp
implementation in m_trampoline.S (compiled from the trivial .c code
to asm with gcc like the other implementations in that file).
338499 --sim-hints parsing broken due to wrong order in tokens (after introduction of no-nptl-pthread-stackcache)
Fix the token order in m_main.c
Somewhat retested by running the regression tests
(testing no-nptl-pthread-stackcache) and testing in an outer/inner setup
(testing enable-outer,no-inner-prefix).
It seems there is no regtest for the 2 other flags (lax-ioctls,fuse-compatible)
Julian Seward [Fri, 22 Aug 2014 19:07:12 +0000 (19:07 +0000)]
mc_LOADV_128_or_256_slow: change a constant from V_BITS8_DEFINED
to V_BITS64_DEFINED so as to be consistent with the rest of the
types in this function. Since both values are zero it gives no
functional change.
Mark Wielaard [Fri, 22 Aug 2014 10:14:28 +0000 (10:14 +0000)]
Tweak gdbserver_tests/hgtls.stdoutB.exp filter_gdb a little for older GDB.
Older GDB (7.2 on i386) don't print out which variable+offset an argument
pointer in a breakpoint function points to. The hgtls test already tests
whether the p pointer/test points to the expected tests array element.
So don't expect gdb to print it also and filter it out with filter_gdb.
Follow up to r14313: disable stack cache earlier
glibc is recycling memory for detached threads before a thread
termination => disable the stack cache earlier (i.e. once
a 'non main thread' is seen)
Mark Wielaard [Thu, 21 Aug 2014 10:04:04 +0000 (10:04 +0000)]
Check some known PATHs for mpicc in configure.
On some distributions (fedora) mpicc not installed on the default PATH.
Add a search path for finding mpicc by default if it is installed.
The user can still override the used mpicc compiler using --with-mpicc=.
Explicitly say we are checking for -mpreferred-stack-boundary=2.
Only 2 is ever used when it is supported. Some gcc versions/arches
(e.g. GCC 4.8 for x86_64) support -mpreferred-stack-boundary, but
only between 4 and 12. The message that -mpreferred-stack-boundary
wasn't supported was a little confusing. So explicitly say we checked
for -mpreferred-stack-boundary=2.
Florian Krohm [Wed, 20 Aug 2014 21:04:14 +0000 (21:04 +0000)]
Clean up confusion about VG_(args_the_exename) which was believed to
possibly be NULL in several places. Nowadays, VG_(ii_create_image) will
terminate the process if VG_(args_the_exename) is NULL.
Julian Seward [Wed, 20 Aug 2014 17:45:00 +0000 (17:45 +0000)]
Kind of a follow-up to r14237.
pre_mem_read_sockaddr: in the case where the caller doesn't
specify any address family (that is, the family is AF_UNSPEC)
don't perform any further checks on the supplied |sa| address
block, since doing so merely gives rise to false uninitialised
value errors.
Mark Wielaard [Wed, 20 Aug 2014 16:11:53 +0000 (16:11 +0000)]
configure should check for warning flags supported to disable them (#338205).
Configure would check whether gcc supported -Wno... flags. But gcc always
does. It is happy to just not warn about anything. So flip all configure
checks to test for the warning and only when gcc accepts the warning flag
use -Wno-...
Introduces two helper functions to make it easier to add new flag checks.
AC_GCC_WARNING_COND and AC_GCC_WARNING_SUBST_NO.
Add option a new sim-hint no-nptl-pthread-stackcache.
Activating this hint using --sim-hints=no-nptl-pthread-stackcache
means the glibc nptl stack cache will be disabled.
Disabling this stack/tls cache avoids helgrind false positive race conditions
errors when using __thread variables.
Note: disabling the stack cache is done by a kludge, dependent on
internal knowledge of glibc code, and using libpthread debug info.
So, this kludge might be broken with newer glibc version.
This has been tested on various platforms and various
glibc versions 2.11, 2.16 and 2.18
To check if the disabling works, you can do:
valgrind --tool=helgrind --sim-hints=no-nptl-pthread-stackcache -d -v ./helgrind/tests/tls_threads |& grep kludge
If you see the below 2 lines, then hopefully the stack cache has been disabled.
--12624-- deactivate nptl pthread stackcache via kludge: found symbol stack_cache_actsize at addr 0x3AF178
--12624:1:sched pthread stack cache size disabling done via kludge
The attached patch cleanups the clo processing
of clo which are (or should be) 'enum set'.
* pub_tool_options.h : add new macrox VG_USET_CLO and VG_USETX_CLO to
parse an 'enum set' command line option (with or without "all" keyword).
* use VG_USET_CLO for existing enum set clo options:
memcheck --errors-for-leak-kinds, --show-leak-kinds, --leak-check-heuristics
coregrind --vgdb-stop-at
* change --sim-hints and --kernel-variants to enum set
(this allows to detect user typos: currently, a typo in a sim-hint
or kernel variant is silently ignored. Now, an error will be given
to the user)
* The 2 new sets (--sim-hints and --kernel-variants) should not make
use of the 'all' keyword => VG_(parse_enum_set) has a new argument
to enable/disable the use of the "all" keyword.
* The macros defining an 'all enum' set definition was duplicating
all enum values (so addition of a new enum value could easily
give a bug). Removing these macros as they are unused
(to the exception of the leak-kind set).
For this set, the 'all macro' has been replaced by an 'all function',
coded using parse_enum_set parsing the "all" keyword.
* Unbreak build on macos (and on other platforms) that do not have
HAVE_DLINFO_RTLD_DI_TLS_MODID.
* add the copyright notice in auxprogs/getoff.c
* add more prereq to hgtls.vgtest
Some cleanup post QGetTlsAddr commit
* remove useless commented line in hgtls.vgtest
* avoid some #ifdef in arch specific code, hoping to discover more compile
time errors.
fix 338160: Implement QGetTlsAddr query so that GDB+V gdbsrv can print __thread variables.
To implement QGetTlsAddr, gdbsrv has to know how to get the glibc dtv
address and the module id from the link_map.
These 2 things are dependent on the internals of glibc.
The dependency is mostly isolated in a few lines of arch dependent
code or in an external utility that used a hack + -ldl lib to find
the offset of the modid in the link_map structure.
Tested on x86/amd64/ppc64/s390x. Somewhat tested on ppc32 and arm64.
Untested/a few #ifdef-ed lines not compiled on arm/mips32/mips64
and darwin.
For more background info about thread local storage handling, see
'ELF Handling For Thread-Local Storage' http://www.akkadia.org/drepper/tls.pdf
Changes:
* auxprogs/getoff.c new auxilliary program to get platform specific offsets
(currently only the offset for the module id in struct link_map).
* configure.ac : check for dlinfo(RTLD_DI_TLS_MODID) needed for getoff.c
* new gdbserver_tests/hgtls, testing various types of __thread variables
* various m_gdbserver files:
- implement decoding of the QGetTlsAddr query
- for each platform: platform specific code to get the dtv
- call to external program getoff-<platform> the first time an
__thread variable is printed.
Carl Love [Thu, 14 Aug 2014 16:54:48 +0000 (16:54 +0000)]
This commit is an update to Bugzilla 334836
There are two copies of the round test in none/tests/ppc32/round.c
and none/tests/ppc64/round.c. The two source files should be
identical. The LE functional test commit updated the round.c test for
ppc64 but was missing the ppc32 round updates. The round.c test was
updated to fix an issue where we were getting different outputs
depending on the compiler. The output is now consistent for the
compilers allowing the removal of the additional expect files for
ppc32 and ppc64.
Reduce memory needed for symbols, by having the tocptr and local_ep
(used for ppc64 platforms) #ifdef-ed and accessed by macros
that becomes NOP on non ppc64 platforms.
This decreases the debuginfo memory by about 2.5 Mb on a big 32 bit application.
Note : doing that, some questions were encountered in the way
tocptr and local_ep have (or do not have) to be copied/maintained
in storage.c canonicaliseSymtab
Florian Krohm [Tue, 12 Aug 2014 11:43:17 +0000 (11:43 +0000)]
Remove function VG_(sysnum_string_extra) as it was just a wrapper
around VG_(sysnum_string). Also remove associated macro
VG_SYSNUM_STRING_EXTRA.
The VG_SYSNUM_STRING macro returned a pointer to a variable which
is out of scope. Using that value may cause undefined behaviour.
Change VG_(sysnum_string) to return pointer to static buffer instead.
Fix call sites.
arm64 Improve fpsr gdbsrv handling.
let the compiler handle the ULong to UInt conversion rather than
play with addresses.
Tested manually GDB+vgdb that reading and setting fpsr works, using
code such as (provided by Julian, I cannot write a single line of
arm64 asm :)
void set_fpsr ( uint32_t val ) {
__asm__ __volatile__( "msr fpsr, %0" : : "r"(val) : "cc" );
}
Carl Love [Fri, 8 Aug 2014 22:29:10 +0000 (22:29 +0000)]
This commit is for Bugzilla 334834.
A performance regression was found due to the
-02 -m64 flags not being included when compiling
the VEX PPC64 code. This commit fixes the Makefile
to pass the needed flags when compiling VEX PPC64
code.
This patch implements reading the directory information for source
files in the dwarf3 reader.
Basically, the change consists in replacing in the DiInlLoc struct
const HChar* filename; /* caller source filename */
by
UInt fndn_ix; /* index in di->fndnpool of caller source
dirname/filename */
A similar change is done in DiVariable struct, as the
read_filename_Table code is shared between the inline info reader
and the varinfo reader.
Note however that outputting dirname in variable description
is not done. Unclear if that is desired or not.
It should be trivially doable however.
Replacing filename by fndn_ix implies a bunch of semi-mechanical
changes.
The code to read the directory names is in the new function
static
XArray* read_dirname_xa (struct _DebugInfo* di, const HChar *compdir,
Cursor *c,
Bool td3 )
Note that readdwarf.c and readdwarf3.c have significant duplicated
logic. Would be nice to integrate these 2 dwarf readers in one
single reader. This function is directly inspired from
an equivalent piece of code in readdwarf.c.
Modified memcheck/tests/varinfo5.vgtest to test the dirname appears
in the inlined functions.
Impact on memory is neglectable (a few Kb on a big executable).
Carl Love [Thu, 7 Aug 2014 23:49:27 +0000 (23:49 +0000)]
This commit is for Bugzilla 334836. The Bugzilla contains patch 3 of 3
to add PPC64 LE support. The other two patches can be found in Bugzillas
334384 and 334834. Note, there are no VEX changes in this patch.
PP64 Little Endian test case fixes.
This patch adds new LE and BE expect files where needed. In other
cases, the test was fixed to run correctly on LE and BE using based on
testing to see which platform is being used.
Where practical, the test cases have been changed so that the output
produced for BE and LE will be identical. The test cases that require
a major rewrite to make the output identical for BE and LE simply
had an additional expect file added.
Signed-off-by: Carl Love <carll@us.ibm.com>
git-svn-id: svn://svn.valgrind.org/valgrind/trunk@14240
Carl Love [Thu, 7 Aug 2014 23:35:54 +0000 (23:35 +0000)]
This commit is for Bugzilla 334834. The Bugzilla contains patch 2 of 3
to add PPC64 LE support. The other two patches can be found in Bugzillas
334384 and 334836.
POWER PC, add the functional Little Endian support, patch 2
The IBM POWER processor now supports both Big Endian and Little Endian.
The ABI for Little Endian also changes. Specifically, the function
descriptor is not used, the stack size changed, accessing the TOC
changed. Functions now have a local and a global entry point. Register
r2 contains the TOC for local calls and register r12 contains the TOC
for global calls. This patch makes the functional changes to the
Valgrind tool. The patch makes the changes needed for the
none/tests/ppc32 and none/tests/ppc64 Makefile.am. A number of the
ppc specific tests have Endian dependencies that are not fixed in
this patch. They are fixed in the next patch.
Per Julian's comments renamed coregrind/m_dispatch/dispatch-ppc64-linux.S
to coregrind/m_dispatch/dispatch-ppc64be-linux.S Created new file for LE
coregrind/m_dispatch/dispatch-ppc64le-linux.S. The same was done for
coregrind/m_syswrap/syscall-ppc-linux.S.
Signed-off-by: Carl Love <carll@us.ibm.com>
git-svn-id: svn://svn.valgrind.org/valgrind/trunk@14239
Carl Love [Thu, 7 Aug 2014 23:17:29 +0000 (23:17 +0000)]
This commit is for Bugzilla 334384. The Bugzilla contains patch 1 of 3
to add PPC64 LE support. The other two patches can be found in Bugzillas
334834 and 334836. The commit does not have a VEX commit associated with it.
POWER PC, add initial Little Endian support
The IBM POWER processor now supports both Big Endian and Little Endian.
This patch renames the #defines with the name ppc64 to ppc64be for the BE
specific code. This patch adds the Little Endian #define ppc64le to the
Additionally, a few functions are renamed to remove BE from the name if the
function is used by BE and LE. Functions that are BE specific have BE put
in the name.
The goals of this patch is to make sure #defines, function names and
variables consistently use PPC64/ppc64 if it refers to BE and LE,
PPC64BE/ppc64be if it is specific to BE, PPC64LE/ppc64le if it is LE
specific. The patch does not break the code for PPC64 Big Endian.
The test files memcheck/tests/atomic_incs.c, tests/power_insn_available.c
and tests/power_insn_available.c are also updated to the new #define
definition for PPC64 BE.
Signed-off-by: Carl Love <carll@us.ibm.com>
git-svn-id: svn://svn.valgrind.org/valgrind/trunk@14238
Julian Seward [Wed, 6 Aug 2014 19:52:12 +0000 (19:52 +0000)]
pre_mem_read_sockaddr: properly handle the NETLINK address family
rather than throwing to the default case. This stops Memcheck
reporting false positives for the NETLINK case.
fix 338024 inlined functions are not shown if DW_AT_ranges is used
Based on investigation and patch by Matthias Schwarzott.
(no small test found that reproduced the problem,
but the equivalent patch given in bug 338024 fixed the inlined stack
trace in a big shared lib).
Would be nice however to have a small test case ...
Bart Van Assche [Tue, 5 Aug 2014 11:35:48 +0000 (11:35 +0000)]
Linux system call wrappers: truncate ioctl request number to 32 bits
As explained in https://bugs.kde.org/show_bug.cgi?id=331829, when passing
an ioctl request number as an int to a function the request number will
be sign-extended to 64 bits on 64-bit systems. Avoid that this causes
Valgrind to fail to recognize an ioctl by truncating the request number
to 32 bits.
projects / thirdparty / valgrind.git / log
