On Wed, 19 Nov 2014, David Malcolm wrote: > On Wed, 2014-11-19 at 10:09 -0700, Jeff Law wrote: > > On 11/19/14 04:47, Richard Biener wrote: > > > On Wed, Nov 19, 2014 at 11:46 AM, David Malcolm <dmalc...@redhat.com> > > > wrote: > > >> Valgrind complains about uninitialized data within sparseset_bit_p. > > >> Provide a suppression file to silence these warnings. > > >> > > >> Valgrind requires suppression files for C++ code to use the mangled > > >> names, so we do that here. > > > > > > There is --enable-valgrind-annotations to get the same effect by GCC > > > telling valgrind about this (and more). > > Right. See VALGRIND_DISCARD. Is that not covering this case? > > I simply didn't spot the option, and was running without it. > > I'll drop the new file, and document that people running the jit > testsuite under valgrind need to use that configure option.
IMHO, making --enable-valgrind-annotations the default when headers are found and when in gcc development is in DEV-PHASE = experimental (i.e. not for releases) would be even better. Anyone opposed? I thought it already was the default! The overhead is IIRC a few weird NOP instructions per VALGRIND_DISCARD (& Co.) annotation. brgds, H-P (PS. I care a little bit since I added them in the first place.)
