| Summary: | gcc downgrade gcc-3.3.2-r6 -> gcc-3.3.2.-r5 fails | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Daniel Seyffer <gentoo-bugs> |
| Component: | [OLD] Core system | Assignee: | Please assign to toolchain <gcc-porting> |
| Status: | RESOLVED WORKSFORME | ||
| Severity: | major | CC: | flash3001, vapier |
| Priority: | High | ||
| Version: | unspecified | ||
| Hardware: | x86 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: | emerge -u gcc output | ||
|
Description
Daniel Seyffer
2004-02-07 03:46:51 UTC
I don't seem to run into this problem on my second Gentoo box (a Pentium II), which is currently also downgrading from 3.3.2-r6 to r5 and so far compiles fine. Hmmm. :-/ that output doesnt have the error in it ... scroll further up to find it also, why would you downgrade ? :P Why? Don't ask me, ask portage! emerge -up gcc These are the packages that I would merge, in order: Calculating dependencies ...done! [ebuild UD] sys-devel/gcc-3.3.2-r5 [3.3.2-r6] ChangeLog says: 07 Feb 2004; Martin Schlemmer <azarah@gentoo.org> gcc-3.3.2-r6.ebuild: Mask again, as it have some issues, bug #40603. Will attached updated output at once. Created attachment 25135 [details]
emerge -u gcc output
Ups, now that's quite interesting:
"/usr/lib/gcc-lib/i686-pc-linux-gnu/3.3.2/libgcc.a: could not read symbols:
Malformed archive"
Check disk space - if possible, switch PORTAGE_TMPDIR to another partition. disk space is 3,5 GB of free space left. Any other pointers? Thanks. I get this as well with the same CFLAGS I have tried other cflags but it always fails at the same point. I have found a work around for this problem, I have locally masked r5 of gcc3.3.2 and r4 has compiled fine. I got a lot of errors compiling kde 3.2 with r5 mainly segfaults and downgraded to r4 then and all compiled fine. I notice on downgrading to r5 from r6 it does not scan at the begining of the compile for programs using the stack smashing code but going back to r4 does, may be a clue? I am leaving r5 masked for now until this has been resolved. Hmm, weird thing though, is that it tries to link with the libgcc.a in the live /. So this might be another issue rather of -r6? It only shows up here as there might be some changes to the build process might have changed. Colin, does -r5 merge from -r4 now? Well, if I try downgrading to -r4 I get exaclty the same error message as when trying to emerge -r5. Suppose my currently installed -r6 is quite broken, probably time to search for a binary gcc package... :-/ I have worked around this problem by emerging a binary gcc package (gcc-3.3.2-r2 from http://dev.gentoo.org/~avenj/bins/) and then emerging gcc-3.3.2-r5. My portage started yesterday trying to upgrade from 3.2.3-r3 to 3.3.2-r5 and
fails with yet a different problem:
No binaries with suspicious libgcc __guard@GCC dependencies detected
>>> Source unpacked.
* Configuring GCC...
Created "Makefile" in /var/tmp/portage/gcc-3.3.2-r5/work/build using "mt-frag"
cc1: invalid option `-O2-pipe'
*** The command 'gcc -o conftest -mcpu=athlon -O2-pipe conftest.c' failed.
You can see that the -O2-pipe flags need a space in between them. This smells
like a sed-script error to me. Anyone else seen this? I have it on two athlon
machines.
Should I open a new bug report for this?
this sounds like a broken flag-o-matic.eclass to me where filter-flags does some incidential concatenating of formerly separated CFLAGS You can close this old bug report - at least as far as I am concerned. (See #11) ;-) |
