gcc-4.4.5 fails with bootstrap lean [SOLVED]

[madal]

Hola;

I'm generally decent @ tracking my compilation errors and finding solutions here or on the buglist, but this one has me truly stumped. With a recent update, gcc-4.4.5 fails, always at the same point, with a "bootstrap lean" error; even with all the steps I have taken to try to fix it, it borks at the same point. I have a 1 month old build on a new box. These are the suggestions I have found, and steps I have taken:
1) Re-emerged libstdc++-v3 multiple times.
2) Re-emerged glibc multiple times, up to and including the new version, glibc-2.12.2, that just came into stable.
3) Run memtest86+ through multiple cycles, sans errors. I have 1.8G of active swap.
4) Edited make.conf. Tried MAKEOPTS="-j1". Nuked -pipe in CFLAGS. Tried O1, O2 (my default), and O3 (just for fun) in CFLAGS.
5) Re-established the symlink from /lib --> /lib64, even though it was never gone.
6) Erased /var/tmp/portage and re-synced.
7) Tried the emerge both inside and outside of the X-window environment.
8. Numerous clean revdep-rebuilds.
9) IA32 emulation is in my kernel.

Here's what I get:

[Hu]

Please post the complete build log or at least the topmost error. Since you build with -j9, the useful error message scrolled away and is well above the point where you truncated the build log.

[madal]

Sorry about that. I am not familiar w/ using pastebin.........yet. The entire build.log file is 4.9MB. I re-emerge for you using MAKEOPTS="-j1". Please let me know if what I have posted here is still inadequate, and I will learn to pastebin. From my make.conf:

[djdunn]

internal compiler error: in dwarf2out_frame_debug_expr, at dwarf2out.c

and

{standard input}:8510: Error: open CFI at the end of file; missing .cfi_endproc directive


you can google it more but i did and you can try -fno-shrink-wrap optimization in your make.conf
_________________
“Music is a moral law. It gives a soul to the Universe, wings to the mind, flight to the imagination, a charm to sadness, gaiety and life to everything. It is the essence of order, and leads to all that is good and just and beautiful.”

― Plato

[madal]

Thanks djdunn, you got me on the right path. It seems that gcc-4.4 doesn't like Sandy Bridge too much. Found the answers here:

https://bugs.gentoo.org/show_bug.cgi?id=360927
and here:
https://bugs.gentoo.org/show_bug.cgi?id=357287

NEED to get rid of avx (sure, like that wasn't totally obvious):

[padoor]

i also had the same problem in a new desktop install while updating system sfter stage and portage install in the partition.
i thought because i had not mrged a kernel it does so.
i left it like that as the stage itself had same gcc and i don't know why it wanted to re install it
however it is working with the gcc came with the stage3.
_________________
reach out a little bit more to catch it (DON'T BELIEVE the advocate part under my user name)

[<3]

just had the same problem on a new install. That fix seems to work. Is there a way too get notified when this gets fixed?

[HeXiLeD]

I had the same problem in several computers with 4.5.3*.
Systems were 32bit and 64 bit and different archs
I tried a couple gcc versions and no luck.
None of the systems had -mno-avx.

I fixed the problem by turning off ccache en distcc.
_________________
Do you hear the sound of inevitability?
With age, comes great grumpiness and that, was 20 years ago...

CertFP: becbbd161d5a5c31de3c45171b77bf710911db29 / d985d21f89fe2977b593c4d381a1a86802e62990d9328d893db76d59f9935244