From 4fb33bc53b4ce38e02f238165c9bed5e0f8dadba Mon Sep 17 00:00:00 2001 From: "timurrrr@chromium.org" Date: Thu, 19 Aug 2010 14:16:02 +0000 Subject: Move std::string ignores back to the common TSan ignore file to suppress a false report in unit BUG=http://code.google.com/p/data-race-test/issues/detail?id=40 TBR=glider Review URL: http://codereview.chromium.org/3169025 git-svn-id: svn://svn.chromium.org/chrome/trunk/src@56677 0039d316-1c4b-4281-b951-d872f2087c98 --- tools/valgrind/tsan/ignores.txt | 9 +++++++++ tools/valgrind/tsan/ignores_mac.txt | 15 --------------- 2 files changed, 9 insertions(+), 15 deletions(-) (limited to 'tools/valgrind') diff --git a/tools/valgrind/tsan/ignores.txt b/tools/valgrind/tsan/ignores.txt index 378d14c..1517faf 100644 --- a/tools/valgrind/tsan/ignores.txt +++ b/tools/valgrind/tsan/ignores.txt @@ -38,6 +38,15 @@ fun:vfprintf fun:fwrite fun:fflush +# False reports on std::string internals, see +# http://code.google.com/p/data-race-test/issues/detail?id=40 +fun:*_M_mutateE* +fun_r:*_M_set_length_and_sharable* +fun:*_M_is_leaked* +fun:*_M_is_shared* +fun:*_M_set_leaked* +fun:*_M_set_sharable* + # Don't instrument intercepts src:*ts_valgrind_intercepts.c diff --git a/tools/valgrind/tsan/ignores_mac.txt b/tools/valgrind/tsan/ignores_mac.txt index 4c3aa1b..20c7a13 100644 --- a/tools/valgrind/tsan/ignores_mac.txt +++ b/tools/valgrind/tsan/ignores_mac.txt @@ -10,21 +10,6 @@ fun:*base*subtle*Release_Store* fun:*base*subtle*NoBarrier_Store* fun:*base*subtle*NoBarrier_CompareAndSwap* -# 32-bit version of std::string -# There is something very strange inside these functions. -# Looks like they write zero to a global variable which -# value is already zero. I.e. the race is benign. -# Global var is _ZNSs4_Rep20_S_empty_rep_storageE -# The race can only be reported on Mac since empty_rep is ignored by name -# inside ThreadSanitizer. Unfortunately, we don't have global variable names -# when running on Mac. -fun:*_M_mutateE* -fun_r:*_M_set_length_and_sharable* -fun:*_M_is_leaked* -fun:*_M_is_shared* -fun:*_M_set_leaked* -fun:*_M_set_sharable* - # we ignore the Security libraries for now since # their instrumentation is very slow. # TODO(timurrrr): investigate whether we need to instrument them -- cgit v1.1