Nicolas Turro <[email protected]> writes:
> On Tue, 2003-07-15 at 16:35, Philippe Troin wrote:
> > Nicolas Turro <[email protected]> writes:
> >
> > > Hi, i posted a bug about amd hanging at boot time a few week ago,
> > > does anybody has a fix for it ?
> > > The bug is described here :
> > >
> > > https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=90902
> > >
> > > it seems that several groups of users besides me encounter the same
> > > bug....
> > >
> > > If not, i'll have to post-install a 2.4.18 kernel on all my new
> > > RH9 boxes :-(
> >
> > Looks like YAFP (Yet Another Futex Problem).
> >
> > Have you tried running with LD_ASSUME_KERNEL=2.2.5 ?
>
> You are right, setting LD_ASSUME_KERNEL=2.2.5 (or
> LD_ASSUME_KERNEL=2.4.1) corrects the problem.
> Do you have more info on this ?bug' ?
As a rule of thumb, whenever a program deadlocks in a futex call,
setting this variable to a < 2.5.x value might prevent the deadlock.
Google around for the exact reason why.
AFAIU, it's a RH 9.0 bug.
Phil.