Bugzilla Bug 179218

I tried to use gcc 4.2.0 in a chroot. Although it doesn't work too good.

First I used an existing chroot and updated from 4.1.1-r3 to 4.2. After
compiling this and that paludis (one of the compiled packages) didn't work
anymore. It only produced segmentation faults from then on.

As I was not able to recompile paludis without a working paludis I started from
scratch.

I unpacked a stage1, unmasked gcc 4.2.0 and ran bootstrap.sh. Now I got a
segfault while compiling glibc.

What's this bug about? Please don't add ~ppc64 to gcc 4.2.0 unless those
problems are sorted out.

Created an attachment (id=119787) [details]
log for glibc merge (incl. segfault)

ive been using gcc-4.2.0 on my quad without troubles, but that only extends so
far to ppc64 ... and we've shown in the past that ppc64 tends to be much more
unstable than ppc in the toolchain ...

ok. so let's find out the differences between your and mine config. could you
please send me your 'emerge --info' output?

when i said "ppc" i meant "ppc" :)

i havent tried ppc64, just ppc32

(In reply to comment #4)
> when i said "ppc" i meant "ppc" :)
> 
> i havent tried ppc64, just ppc32
> 

Sorry, I think I don't understand that "ive been using gcc-4.2.0 on my quad
without troubles, but that only extends so far to ppc64" then :-/

/me catches a dictionary..

I think he means he's using a 32bit userland corsair...

ahh.. stupid /me. I didn't thought about that possibility. thanks..

yeah, the ppc64 userland is just too unstable for me to use as a desktop, so i
have a 64bit kernel with a 32bit userland ... i need KDE/firefox actually
functioning :)

sorry about that

fwiw, firefox and thunderbird work quite nicely on ppc64 64ul now thanks to
some patches from the moz Gentoo team.  Sorry to derail the point of this
bug...

unless anyone has crazy ideas as to the source of the troubles, i think the way
to tackle this would be to take a stage3/gcc-4.1.2 system and install gcc-4.2.0
in parallel ...

then locate the simplest C-only package you can find and get it to bomb only
when compiled with gcc-4.2 ...

Check for out of memory issues, There is a major regression on 64bit systems as
far as gcc is concerned. More detail can be found here :

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30052

It does include a patch but needs to be updated in two locations to fix the ice
it introduces on fortran.

going to close this bug ...  gcc 4.2.2 seems to be dramatically improved ...
still under test .. but it's not dead yet. 4.2.0 has no chance of going stable
and we ain't fixing it.

as this bug was referenced by external parties I will document this here.

http://sourceware.org/ml/libc-alpha/2006-05/msg00017.html fixes this issue for
gcc 4.2.x

added ~ppc64 to gcc-4.2.2