Re: SIGSEGV on recursive pthread_join

[Richard Guenther]

On Wed, 7 Jan 2004, Jeroen Dekkers wrote:

> On Tue, Jan 06, 2004 at 04:38:02PM +0100, Richard Guenther wrote:
> > On Mon, 5 Jan 2004, Jeroen Dekkers wrote:
> >
> > > On Mon, Jan 05, 2004 at 07:24:14PM +0100, Richard Guenther wrote:
> > > > On Mon, 5 Jan 2004, Jeroen Dekkers wrote:
> > > >
> > > > > On Mon, Jan 05, 2004 at 05:17:47PM +0100, Richard Guenther wrote:
> > > > > > I'm experiencing a weird behavior with the NPTL threads, as that my
> > > > > > application (sf.net/projects/glame) segfaults upon invocation of a
> > > > > > pthread_join while the main process is currently joining the thread 
> > > > > > that
> > > > > > tries to do the join. I.e.
> > > > >
> > > > > You can't join a thread more than once.
> > > >
> > > > I'm not doing this. I'm joining the joining thread ;) I.e. each thread 
> > > > is
> > > > exactly joined one time (see the numbering above).
> > >
> > > Oh, sorry, I misunderstood you.
> >
> > It was indeed the case I tried to join a thread twice - the code was just
> > somehow obfuscated.  So my problem is fixed.  But still, old libc did not
> > segfault in this case - can't we detect this somehow and return ESRCH
> > instead?
>
> Looking at the code i think it should return EINVAL, I have no idea
> why it doesn't. But POSIX says that joining a thread twice results in
> undefined behaviour, so it's not really wrong to crash, just not very
> friendly.

Indeed.  And somehow it "works" when using a libc with debugging symbols
in...

Richard.

--
Richard Guenther <richard dot guenther at uni-tuebingen dot de>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/