|
From: Bart V. A. <bar...@gm...> - 2009-08-23 07:15:36
|
On Sun, Aug 23, 2009 at 12:20 AM, Nicholas Nethercote <n.n...@gm...> wrote: > > On Sat, Aug 22, 2009 at 5:49 PM, Bart Van > Assche<bar...@gm...> wrote: > > > > It would be great if someone familiar with the stabs format could have > > a look at the bug report https://bugs.kde.org/show_bug.cgi?id=162482. > > Reading the debug info of > > memcheck/tests/deep_templates causes a crash in Valgrind on certain > > PPC distro's. I'm not sure however whether this is an issue with the > > Valgrind stabs reader or that the stabs debug info of the test program > > is invalid. > > The problem with stabs is that there isn't a clear standard. There is > a document, from Red Hat I think, floating around which is a > quasi-standard, and I think Valgrind's reader is based on this. But > some compilers just do different things, eg. stabs on Darwin is close > to unusable. > > Given that stabs is increasingly rare, I think there are more useful > things that can be worked on. Especially if it has only so far > manifested as a single regtest failure. Every single regression test failure is annoying because after every regression test run it has to be analyzed again why the regression test failed. So I don't think that making regression tests pass is not useful. If the stabs format is increasingly rare, why is the deep_templates regression test compiled with the -gstabs flag ? From memcheck/tests/Makefile.am: deep_templates_SOURCES = deep_templates.cpp deep_templates_CXXFLAGS = $(AM_CFLAGS) -O -gstabs Bart. |