| Summary: | media-libs/xine-lib-1.1.2_pre20060328-r5 fails on x86 w/ -fforce-addr | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Martux <root> |
| Component: | New packages | Assignee: | Gentoo Developers for the x86 Architecture <x86> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | bakirov, flameeyes, jakub, Stefano |
| Priority: | High | ||
| Version: | 2006.0 | ||
| Hardware: | x86 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
|
Description
Martux
2006-05-06 01:03:12 UTC
Are you kidding, you ricer? Stop using that stupid CXXFLAGS and try again.. Register starvation goes to x86 arch team. Ok, this is a CFLAGS issue. I reproduced the bug with CFLAGS="-march=pentium-m -O2 -pipe -fforce-addr -fomit-frame-pointer -funroll-loops -falign-functions=4 -maccumulate-outgoing-args but xine-lib compiled with: -march=pentium3 -O2 -pipe -fomit-frame-pointer -falign-functions=4 -march=pentium3 -O2 -pipe -fomit-frame-pointer -funroll-loops -march=pentium3 -O2 -pipe -fomit-frame-pointer -maccumulate-outgoing-args I can't be bothered to spend anymore time trying to track down what's causing it. The user needs to fix his CFLAGS and this should be resolved as invalid in my opinion... Full ack David, he should fix his CFLAGS and CXXFLAGS instead of ricing. Stephan, please keep comments that don't actually *help* the bug, off of the bug. The problem is most likely due to -funroll-loops or -maccumulate-outgoing-args. Pick a sane set of global CFLAGS and you shouldn't have a problem. (In reply to comment #5) > Stephan, please keep comments that don't actually *help* the bug, off of the > bug. > > The problem is most likely due to -funroll-loops or -maccumulate-outgoing-args. > Pick a sane set of global CFLAGS and you shouldn't have a problem. > Hello! Truly sorry for this mess :( I didn't even knew, i was ricing. Fixed my CFLAGS and now it works! Thank you! What about filter_flags fforce-addr for this ebuild? Oh, it fails here with CFLAGS="-O2 -march=pentium3 -pipe -fomit-frame-pointer -fforce-addr" (didn't check that one before because I didn't think -fforce-addr would be the problem). So yeah, we could filter it on x86... Confirmed here, -fforce-addr breaks compile. Reopening. So, x86 arch team, this is up to you, feel free to fix it as you find pleasure :P *** Bug 132210 has been marked as a duplicate of this bug. *** Adding filter line. Seems that the filter function don't filter (CXX||CPP)FLAGS. (In reply to comment #9) > Confirmed here, -fforce-addr breaks compile. Reopening. > omitting the flag -fomit-frame-pointer caused this bug for me. interestingly the gentoo installer did this... -b |