Summary: | gcc 4.1.1 fails while compiling menuconfig for kernel 2.6.17 | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Yan-Fa Li <yanfali> |
Component: | [OLD] Development | Assignee: | Gentoo Toolchain Maintainers <toolchain> |
Status: | RESOLVED INVALID | ||
Severity: | normal | CC: | rhill |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: | gcc output |
Description
Yan-Fa Li
2006-06-18 21:46:40 UTC
Created attachment 89506 [details]
gcc output
bwah. was this gentoo-sources? yes it was gentoo sources * sys-kernel/gentoo-sources Available versions: 2.4.31-r1 2.4.32-r4 ~2.4.32-r5 2.6.14-r7 2.6.15-r1 2.6.15-r7 2.6.15-r8 2.6.16 2.6.16-r1 2.6.16-r2 2.6.16-r3 2.6.16-r4 2.6.16-r5 2.6.16-r6 2.6.16-r7 2.6.16-r8 2.6.16-r9 2.6.16-r10 2.6.17 Installed: 2.6.16 2.6.16-r6 2.6.16-r7 2.6.16-r8 2.6.16-r9 2.6.16-r10 2.6.17 Homepage: http://dev.gentoo.org/~dsd/genpatches Description: Full sources including the gentoo patchset for the 2.6 kernel tree Actually. I'm having trouble with all var++ constructs all over the source so I'm rolling back to 4.1.0-r1 as we type. I will leave 4.1.1 installed for testing. I'm using the CFLAG=-Os, this could be the cause of the breakage. If 4.1.0-r1 also starts to exhibit problems I will build 4.1.1 without -Os and see what happens. i'm pretty sure something got screwed with your 4.1.1 installation though i can't imagine what. -Os shouldn't be messing with anything, as the toolchain eclass automatically strips it and replaces it with -O2 when build GCC. in addition to that, the kernel Makefile also throws away user defined CFLAGS when doing it's thing. you can see that in the first line of the preprocessor output you posted: // /usr/libexec/gcc/i686-pc-linux-gnu/4.1.1/cc1 -quiet -MD scripts/kconfig/.conf.o.d scripts/kconfig/conf.c -quiet -dumpbase conf.c -mtune=pentiumpro -auxbase-strip scripts/kconfig/conf.o -O2 -Wall -Wstrict-prototypes -fomit-frame-pointer -o - -frandom-seed=0 the only thing i can think of is some bizarre interaction due to using 4.1.1 on a stable keyworded system, though i've never run across this before. Thanks Ryan. The system is now downgraded to 4.1.0 and currently successfully building 2.6.17. I will try rebuilding gcc-4.1.1 again tomorrow and attempt to rebuild the kernel. If it doesn't exhibit the problem I'll put it down to a bizarre toolchain corruption issue and close the bug. downgrading to 4.1.0 and then rebuilding 4.1.1 with that allowed me to successfully build 2.6.17. Seems like something was corrupt in my gcc toolchain. |