Re: strange threads problem using glibc-2.10.1

[Allin Cottrell <cottrell at wfu dot edu>]

On Sun, 17 May 2009, Carlos O'Donell wrote:

> On Wed, May 13, 2009 at 3:12 PM, Allin Cottrell <cottrell@wfu.edu> wrote:
> > On Wed, 13 May 2009, Allin Cottrell wrote:
> >
> >> Although gdb failed, valgrind produces something, for the segfault
> >> in ld-2.10.1.so when I run wine via make...
> >
> > Sorry to keep talking to myself, but I might as well put all my
> > info out there. ?The trouble I was having seems to be related to
> > the definition of __ASSUME_AT_RANDOM as it affects dl-osinfo.h.
> > I see this is activated for Linux kernels >= 2.6.29, and I used
> > --enable-kernel=2.6.29 when configuring glibc. ?I'm running kernel
> > 2.6.29.3.
> >
> > However, I rebuilt glibc-2.10.1 with __ASSUME_AT_RANDOM undef'd,
> > and the segfaults I mentioned have stopped. ?In case it's of any
> > interest I'm attaching .config from the kernel build.
>
> Excellent sleuthing, thanks for posting your final conclusion.

Thanks for the kind words!

> A couple of notes:
>
> 1) The error:
> ~~~
> warning: Cannot initialize thread debugging library: generic error
> ~~~
> Most likely indicates that your libthread_db.so is out of sync with
> your libpthread.so. The debugger must use the matched pair of
> libraries as built by glibc. If your new glibc is in a chroot, and
> your gdb is not in a chroot, then you must preload the appropriate
> libthread_db.so.

No, actually I'm not doing a chroot; all the relevant libraries
were as installed from my glibc 2.10.1 build.

> 2) You are using gcc 4.4.0, which is in regression fixing mode, and
> may not be stable enough for your uses. I suggest always using the
> most recent stable release branch which is 4.3.

OK, when I get the time, I'll try building with gcc 4.3.

Allin Cottrell