Summary: | glibc-2.3.2-r9 fails to compile - emmintrin.h header error | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | scott.claridge |
Component: | [OLD] Core system | Assignee: | Please assign to toolchain <gcc-porting> |
Status: | RESOLVED FIXED | ||
Severity: | major | ||
Priority: | High | ||
Version: | unspecified | ||
Hardware: | x86 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
scott.claridge
2004-02-09 19:58:25 UTC
I realise there are lots of glibc 2.3.2-r9 bugs but this one looks a tad different.... I'm beginning to think that this is a gcc problem, not a glibc one per se. After failing to get any progress on glibc, I "injected" the -r9 stub and went on with my updates. (Probably not the best option but...) The next thing in the list was module0-init-tools 0.9.15_pre4. That failed with undefined "__builtin_va_start" errors.... At that stage I tried to go back to gcc 3.2.3-r3 which was my last working gcc. Of course now that doesn't compile... The same undefined "__builtin_va_start" error when linking genconstants... I may be in trouble... And have to ponder my next step. Have you tried without distcc yet? I've just built a binary glibc-2.3.2-r3 (on another machine) installed that and recompiled gcc-3.2.3-r3. Compiled no problems. (Which is much better than yesterday, when gcc barfed with all "lower level" app compiles. I followed that with a build of module-init-tools-0.9.15_pre4, which failed yesterday. This now works fine. This swings me back to the point that maybe is was the new glibc version which was the problem and not gcc. I might try the gcc update (3.3.2-r5) and see what havoc that creates.... Extra info: I remember that along the way I tweaked CFLAGS back to -O2, instead of the profile default of -O3. Well I got gcc-3.3.2-r5 to compile okay with all this so far. Stil had the problem at the end where the LD environment hasn't been reset to the new version. (A ldconfig;env-update;ldconfig fixes that.) The new 2.3 version of Python then seemed to compile okay. I guess the next step is to try glibc-2.3.2-r9 again.... What portage version did you have installed when you merged 3.3.2-r5 the second time? portage-2.0.50, which was upgraded just before the initial compiles. That problem is fixed with -r1 of portage (ldconfig issue) ... Well today I compiled up glibc-2.3.2-r9, and had absolutely no problems. Of course I may find downstream problems with other app and all, but this time it compiled fine. (I have upgraded to portage-2.0.50-r1.) I'm not able to quantify any real changes since my first attempt. Maybe the -O2 change from the default -O3, and taking out distcc (but does gcc and glibc make use of distcc anyway?) I hope its not a windoze-like issue - "lets all get out of the car and reboot". I'll advise of any further developments, should they occur. Hmm, distcc might do it. I've had a lot of weird failings lately with distcc enabled. seems to be fixed for reporter. |