https://bugs.kde.org/show_bug.cgi?id=386945
--- Comment #8 from Julian Seward <jsew...@acm.org> --- (In reply to Markus Trippelsdorf from comment #7) > Thanks. > The gcc crash is gone, but I still get lots of invalid read warnings. > > However the amount of errors is much lower now: > > From (valgrind trunk): > ERROR SUMMARY: 89817 errors from 298 contexts > To (your patch and --expensive-definedness-checks=yes): > ERROR SUMMARY: 6495 errors from 85 contexts I re-checked my fixes. I cannot figure out why V still reports undefined value errors now. Can you point me at the bit of the gcc source that generates this code? Maybe that has some further explanation of how this should work. > I still get lots of invalid read warnings. I get those too (see below). These are misaligned (non-8-byte-aligned) ldbrx instructions (I checked) as part of this inlined strcmp. Is the strcmp-inlining expected to produce misaligned loads, or not? -- You are receiving this mail because: You are watching all bug changes.
