https://bugs.kde.org/show_bug.cgi?id=432801
--- Comment #11 from Eyal <eyals...@gmail.com> --- I'm back to not being sure if this is an llvm issue. There are a few things at play here. One is sigaction(), which can foul up the the contents of xmm registers, especially xmm3, which is the only register that isn't specifically cleared out by the loop. The other issue is clang, which isn't specifically clearing out xmm3 but it sort of is doing thing that happen to clear it out. I'm still investigating. And finally, there is valgrind, which might just not know enough about an operation in order to determine that, yes, it's legal. I feel like I'm making progress. I'll update with more as I go along. If I do find that it is a memcheck problem in clang, is there someone that can help me write a software patch for valgrind? -- You are receiving this mail because: You are watching all bug changes.
