Re: intel-agp and yenta-socket issues (was Re: 2.6.14-rc5-mm1

[Valdis.Kletnieks@vt.edu]

On Mon, 24 Oct 2005 22:32:23 PDT, Andrew Morton said:

> > intel-agp would hang during modprobe until I backed this one out.
> 
> A sysrq trace would be nice.

Did some further testing, this time passing a loglevel= so I'd see the output,
and sysrq-T showed that both modprobe issues were due to an even earlier
modprobe that went south.  Looking like a bad merge of a local patch caused
some locking to get corrupted (always a bad sign when atomic_dec_and_test
throws an "atomic counter underflow" message on a refcount).  It's now unclear
why backing out the agp-updates patch allowed modprobe to succeed - it
*shouldn't* have made a difference.  Oddness.

Attachment: pgpuHRRrjuLAc.pgp
Description: PGP signature