Re: segfault during profiling
scr schrieb:
>
> Hello all!
>
> I´m getting a segfault during profiling on Alpha (plain
> potato, first release) with certain sources.
>
> The call stack is as follows, it looks like it is as
> problem in the startup code:
>
> (gdb) where
> #0 0x120017884 in __errno_location ()
> [...]
A few additional notes to my original posting.
I´m getting this segfault on Alpha (ev56), but not on x86,
and only if I enable profiling (`-pg´).
The error occurs if I link statically against libpthread
or alternatively if I link against libpthread_p (in this
case both static and dynamic builds).
This happens only in a single project, which does not make
use of threads (but where usage of threads is intended).
All my attempts to create a small example project, which
shows this error, failed so far.
Because I can´t give such a test case, I thought why not
debugging into it myself. For this purpose I installed
the most recent stable (potato 2.2r2) libc6 packages on
the Alpha, but: The debug libs in /usr/lib/debug are binary
idential to their counterparts in /lib. Is this the result
of an oversight?
And could please someone point me to the docs where usage
of those debugging libs is explained.
Thanks,
Stefan Schroepfer
Reply to: