Re: [gentoo-user] emerge of glibc and gcc fails

[Yoav Luft]

On 10/28/07, Dan Farrell <dan@×××××××××.cx> wrote:
>
> On Sun, 28 Oct 2007 19:40:01 +0200
> "Yoav Luft" <yoav.luft@×××××.com> wrote:
>
> > On 10/28/07, Dan Farrell <dan@×××××××××.cx> wrote:
> > >
> > > On Sun, 28 Oct 2007 06:04:43 +0200
> > > "Yoav Luft" <yoav.luft@×××××.com> wrote:
> > >
> > > > hello,
> > > > during an emerge -uDN process, the compilation for gcc-4.2.1
> > > > failed, as well as for glibc, with same errors.
> > > > I have run revdep-rebuilt, and found out about a broken
> > > > libexpat.so.0thingy, which I solved following the instruction of
> > > > the gentoo irc bot.
> > > > after that, compiling gcc still fails.
> > > >
> > > > insn-emit.c: In function 'gen_leave':
> > > > insn-emit.c:1123: internal compiler error: Segmentation fault
> > > > . . .
> > > > I'm clueless about it, but do feel that not being able to updaate
> > > > gcc or glibc  is bad, and will probably mean that I will not be
> > > > able to update other things as well.
> > >
> > > Have you done a memtest-86, and waited for a few days, or at least a
> > > few error-free complete passes?
> > > --
> > > gentoo-user@g.o mailing list
> > >
> > >
> >
> > I have waited for a few days, and tried different versions of gcc and
> > glibc, all failed.
>
> did you remember to emerge --sync, maybe even emerge --metadata ?
> frankly, I am not sure what the last one does, but I believe it could
> update data concerning the ebuilds themselves, which could be a
> potential source of problems.
>
> > I did had an overheating problem a couple of months ago, right around
> > the time the problem started. So, it could be that the CPU is
> > semi-toasted? How do I run memtest-86?
>
> there are 2 ways I know of:
> 1) simply reboot on a gentoo boot cd and type 'memtest' or some such
>         at the boot: line.  (Refer to the on-disc documentation for help
>         concerning what exactly to type)
> 2) a)emerge memtest86+;
>    b) find memtest.bin and move it to /boot/memtest86plus
>    c)add a grub line that looks like this:
>         title=Memtest86Plus
>         root (hd0,1) # this refers to my boot partition.  you may have
>                      # to change this setting.
>         kernel /boot/memtest86plus/memtest.bin
>    d) reboot and use memtest from the hard drive.
>
> If, when you emerge memtest86+, you have /boot mounted, the emerge will
> plop the .bin file down right where it needs to be.  If not, the output
> from the emerge should giveyou the info you need.
>
> Of course, memtest.bin doesn't have to be on your boot partition, if
> grub can read the root partition.  In fact, you might not even have
> a /boot partition.  Just make sure the root line for memtest in grub is
> proper.
>
>
> > Is there anything else that's
> > worth checking?
>
> You could try a CPU stress test.  If the cpu isn't working properly,
> using it at 100% capacity for a little while ought to make that
> apparent.  But your results may vary.  A rudimentary query on google
> points me in the direction of:
>
> http://users.bigpond.net.au/cpuburn/
>
> However, i don't know if it's source and if not, there's always a valid
> security concern when running others' binaries.  something to keep in
> mind...
> --
> gentoo-user@g.o mailing list
>
>
I had run memtest, and returned no errors. In all occurrences of this
problem I had  emerge --sync before  I updated, as I was about the update
the whole system, and eventually emerge -UD world. I would try a cpu stress,
but that would have to wait for next week. Thanks fo rthe help so far.