As long as I test on Win64 PGO with VC9 x64, it is unnecessary to hack for bug 640494. It should be on MSVC x86 only. Win64 build uses VC9SP1 on tinderbox.
7 years ago
Are you aware there's bug 653322, already?
Damn copy/paste. I meant bug 653056.
Note that this is most likely a compiler problem, so even x86 VC9 might be working. I'm trying to get a local build off x86 VC8sp1, but that it fails to link xul.lib in the first pass...
(In reply to comment #4) > Note that this is most likely a compiler problem, so even x86 VC9 might be > working. I'm trying to get a local build off x86 VC8sp1, but that it fails to > link xul.lib in the first pass... Although I don't test on VC9SP1 x86, should I test using VC9SP1 x86? PGO compiler sometimes generates incorrect code if using MMX, SSE and FP (such as bug 413019).
Also, as long as I know, we have a plan to switch to VC10 if Bug 515492 is fixed. (VC10 has no build-able CRT code.) So it may be unnecessary to test VC9SP1 x86.