Re: ext2fs.static: thread-cancel.c:55: hurd_thread_cancel: Assertion '! __spin_lock_locked (&ss->critical_section_lock)' failed

[Thomas Schwinge]

Hello!

On Sun, Aug 06, 2006 at 01:57:34PM -0700, Roland McGrath wrote:
> > Samuel Thibault, le Sat 05 Aug 2006 02:45:34 +0200, a écrit :
> > > Samuel Thibault, le Sat 05 Aug 2006 02:27:35 +0200, a écrit :
> > > >   assert (! __spin_lock_locked (&ss->critical_section_lock));
> > > >   __spin_lock (&ss->critical_section_lock);

I've now also hit that one (for the first time, as far as I can tell).

> > > > Why should one check that a spinlock is free before taking it?? Here the
> > > > purpose is to cancel _another_ thread, so it _can_ happen that the lock
> > > > be taken any time...
> > > > 
> > > > Shouldn't at least these assertion just be dropped?
> > > 
> > > Maybe not the ones from other files, which are not trivial, but at least
> > > this one (thread-cancel), which was reported to actually happen, and
> > > which to my mind is really arguable.
> > 
> > I got it again this afternoon...
> 
> Please show the backtraces of relevant threads.

Given that I got it on the root file system server and we (so far) have
no measures of reproducing it reliably, that'd be a bit difficult.


Regards,
 Thomas

signature.asc
Description: Digital signature