> Hi all, > > replying to myself, here are a few other things I discovered: > > 1. The mca-bits (sp?) script still hangs the "make regtest" inside the rpm > process even after I set the CFLAGS and CXXFLAGS to "-O2 -g". > > 2. Outside the rpm build process with the default ./configure selected CFLAGS > and CXXFLAGS, then valgrind builds and completes the tests successfully. > > Can anyone shed any light on these problems?
I did a few patches in some test cases that rely on consistent behavior when reading writing to global variables by declaring the variable volatile, but ended up just setting CFLAGS and CXXFLAGS to "-O0 -g3", because, for example, it would still fail on tests using gdb to modify the program state, as variables that live on registers cannot be changed. But I wonder some may be side effects of the patch I made to enable it to compile, and I commented at http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47602 because the comment in the valgrind C file kind of tells me I should not manually save/restore %ebp because should not touch the stack... Patch I made is to revert to use valgrind 3.6.1 behavior: http://svn.mandriva.com/cgi-bin/viewvc.cgi/packages/cooker/valgrind/current/SOURCES/valgrind-3.7.0-pic-clobber.patch?view=markup > Regards, > > Shlomi Fish Paulo ------------------------------------------------------------------------------ Write once. Port to many. Get the SDK and tools to simplify cross-platform app development. Create new or port existing apps to sell to consumers worldwide. Explore the Intel AppUpSM program developer opportunity. appdeveloper.intel.com/join http://p.sf.net/sfu/intel-appdev _______________________________________________ Valgrind-users mailing list Valgrind-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/valgrind-users
