summaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2012-02-19mmt: replace likely/unlikely with valgrind core variantsMarcin Slusarz1-35/+33
2012-02-05mmt: fix formatting of help messagesMarcin Slusarz1-4/+4
2011-10-22mmt: let last_used_region never be NULLMarcin Slusarz1-5/+6
Removes one comparison from the fast path and improves performance slightly.
2011-10-22mmt: force inlining of find_mmapMarcin Slusarz1-2/+3
Next commit triggers some heuristic in gcc which does not inline find_mmap into trace functions which in effect harms performance significantly instead of improving it. Force inlining it.
2011-09-17mmt: add runtime checksMarcin Slusarz3-2/+208
2011-09-17mmt: resolve as much as possible at instrumentation timeMarcin Slusarz3-158/+635
resolve read/write size and number of parameters at instrumentation time, instead of at runtime
2011-09-17mmt: use only as much stack as needed in fastpathsMarcin Slusarz1-18/+20
2011-09-17mmt: add some compiler hintsMarcin Slusarz1-14/+18
they give noticable performance boost
2011-09-17mmt: cache last 10 negative regionsMarcin Slusarz1-9/+222
2011-09-17mmt: cache last used regionMarcin Slusarz1-6/+25
2011-09-17mmt: use binary search to find matching mmap regionMarcin Slusarz3-143/+242
2011-09-17mmt: add new tests to mmaptestMarcin Slusarz4-50/+146
2011-09-17mmt: crash when there's not enough space for new mmap regionMarcin Slusarz2-0/+4
2011-09-17mmt: bump maximum number of mmaped regionsMarcin Slusarz1-1/+1
2011-09-17mmt: restore ability to print file namesMarcin Slusarz1-20/+17
guarded by MMT_PRINT_FILENAMES macro
2011-09-17mmt: fix nvidia post_mmap hookMarcin Slusarz3-4/+6
Fix bug introduced by "mmt: move nvidia ioctl tracing to separate file". (double registration of the same mmap region).
2011-09-17mmt: move tracing functions to separate fileMarcin Slusarz7-520/+535
2011-09-17mmt: move instrumentation functions to separate fileMarcin Slusarz5-565/+632
2011-09-17mmt: move nvidia ioctl tracing to separate fileMarcin Slusarz5-608/+729
2011-09-07core: add support for drm core and nouveau ioctlsMarcin Slusarz2-0/+1645
2011-09-07mmt: Add support for sending mmiotrace marks.Marcin Kościelnicki1-0/+36
2011-09-07mmt: Fix dumpmem for 0-length blocks.Marcin Kościelnicki1-0/+1
2011-09-07mmt: Also trace ioctls on /dev/nvidia0Marcin Kościelnicki1-2/+2
2011-09-07mmt: Add more object classes.Marcin Kościelnicki1-8/+95
2011-09-07mmt: add mmt testerMarcin Slusarz6-0/+196
2011-09-07mmt: rework to not touch memory 2 times in load tracingMarcin Slusarz1-22/+301
Reading pci memory can have side effects and value can change between reads, so read it only once. Reported by Marcin Kościelnicki.
2011-09-07mmt: importMarcin Slusarz5-0/+1362
2011-09-06Make some vgdb interface to callgrind_control internalweidendo3-6/+38
The vgdb "status" monitor command is still available, but used for pretty printing of status information now (acutally, just some place holder for real information up to now: just number of running threads). The internal interface used by callgrind_control to provide stack traces and event counts is using "status internal", and is not documented, as the format is not for human consumption. This also adds some documentation. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12014 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-09-06Allow overriding not-supported auto-detected cache configsweidendo6-283/+268
Patch by Philippe Waroquiers, slightly changed. This actually was a regression from 3.6.1, but the patch also improves on printed messages, and refactors common code between cachegrind and callgrind. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12013 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-09-06Change get_svn_revision to also work with older subversion releasesflorian1-1/+5
that do not allow specifying a revision with svn info. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12012 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-09-06Print the (pseudo-)soname when doing debug printing for Mach-Osewardj1-0/+6
debuginfo reading. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12011 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-09-05* whitespace changessewardj1-3/+71
* move MEMCPY(NONE, ZuintelZufastZumemcpy) to the correct ifdef; it got put in the wrong place by r12009 a couple of hours ago. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12010 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-09-05For intercepts in libc and the dynamic linker (ld.so or dyld), splitsewardj4-260/+579
the Linux and Darwin definitions so they are in completely separate ifdefs -- iow, remove any definitions that are common to both. This gives some duplication, but the upside is that it is now possible to edit the Darwin intercepts without fear of breaking the Linux ones. This will be important when it comes to supporting OSX 10.7. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12009 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-09-05Avoid hanging on older s390x systems.florian1-1/+6
See also bugzilla #268623 comment 3. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12008 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-09-05Add support for IBM Power ISA 2.06 -- stage 3 -- Test cases.sewardj10-4/+2872
Bug 279994 comment 2). (Maynard Johnson, maynardj@us.ibm.com) git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12007 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-09-05Add support for new primops added in vex r2199.sewardj1-0/+2
See Bug 279994 comment 1. (Maynard Johnson, maynardj@us.ibm.com) git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12006 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-09-02Add support for s390x model z114. See also VEX r2198florian1-0/+1
git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12005 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-08-30drd: Yet another regression test reliability improvement.bart1-0/+6
git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12004 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-08-24drd: Avoid reporting false positive race reports inside pthread_join().bart1-0/+6
Note: I haven't analyzed yet since which commit these false positives started to appear but it's most likely a commit applied after the 3.6 release. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12003 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-08-23fix the suppressions to also match libc-2.XX.y.sodirk1-10/+10
git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12002 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-08-23Make a copy of any environment string we are going to modify whentom1-4/+4
we are cleaning up the environment before an exec, otherwise we will seg fault if the string is read only. Fixes #270326. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12001 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-08-23Fix link error for 32-bit build on OSX 10.7 (Lion). This is thesewardj1-0/+98
failure referred to in #275168 comment 0. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@12000 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-08-22Remove the svn:executable property from a source filebart1-0/+0
git-svn-id: svn://svn.valgrind.org/valgrind/trunk@11999 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-08-20Make sure this gets built with -fomit-frame-pointer, even on x86-linux,sewardj1-10/+15
where it otherwise wouldn be. On x86-linux running Memcheck, gives a 6% instruction count reduction and a 10% reduction in memory traffic. (Duh!) git-svn-id: svn://svn.valgrind.org/valgrind/trunk@11998 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-08-20memcheck/tests/linux/timerfd-syscall: Avoid failure on heavily loaded systemsbart1-1/+1
git-svn-id: svn://svn.valgrind.org/valgrind/trunk@11997 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-08-19Add more info about cache sizes.florian1-0/+57
git-svn-id: svn://svn.valgrind.org/valgrind/trunk@11996 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-08-19Make callgrind report the detected cache configuration as well.tom1-0/+10
git-svn-id: svn://svn.valgrind.org/valgrind/trunk@11995 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-08-19Report the detected cache configuration with "-v -v" before wetom1-0/+10
validate it to make validation failures easier to diagnose. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@11994 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-08-18Add a new simulation hint, --sim-hints=fuse-compatible, which causessewardj8-3/+33
a bunch of file-related syscalls to be handled on the might-block syscall path rather than the fast syscall path. This fixes deadlocks when running some FUSE-specific filesystem codes. Fixes #278057. (Mike Shal, marfey@gmail.com) git-svn-id: svn://svn.valgrind.org/valgrind/trunk@11993 a5019735-40e9-0310-863c-91ae7b9d1cf9
2011-08-18x86-linux: don't add redirections for strchr/index in ld.so since theysewardj1-4/+6
are already hardwiredly-redirected at startup, and so these are redundant. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@11992 a5019735-40e9-0310-863c-91ae7b9d1cf9
generated by cgit v1.2.3 (git 2.39.1) at 2024-11-28 22:56:56 +0000