Re: ACPI interrupt problems? (Was: HEADS UP: Starting socketlocking merge)

From: John Baldwin <jhb_at_FreeBSD.org>
Date: Mon, 14 Jun 2004 13:16:37 -0400
On Monday 14 June 2004 01:14 am, Nate Lawson wrote:
> On Mon, 14 Jun 2004, Daniel O'Connor wrote:
> > On Mon, 14 Jun 2004 13:36, Nate Lawson wrote:
> > > > That's difficult for those of us who are affected by the ACPI
> > > > interrupt allocation problems. If ACPI isn't going to be fixed soon
> > > > (it's been 2 weeks already), then the last commits should be backed.
> > >
> > > It would help if you'd point out the email message with this problem
> > > and what commits caused it.  I'm not aware of any new interrupt
> > > problems and haven't seen email about it on acpi_at_.
> >
> > I don't know which commit but I believe John Baldwin is working on a fix.
> >
> > The problem occurs between 02:30 29/05/2004 UTC and 04:00 29/05/2004 UTC.
> > (ie code from the former works, code from the later doesn't)
> >
> > The problem exhibits itself as bfe0 watchdog timeouts (on my system
> > anyway)
>
> Have you isolated it further?  Try backing out rev 1.15 of
> acpi_pci_link.c, referenced by this email (click on link for a patch):
>
> http://docs.freebsd.org/cgi/getmsg.cgi?fetch=2053288+0+archive/2004/cvs-all
>/20040530.cvs-all
>
> I do believe John is working on this.  A temporary backout might be in
> order but I'll let him decide on that.  Either way, the right direction
> for this stuff is to keep moving forward and this commit was definitely a
> step that way.

It can be backed out for now I guess.  Basically, the code doesn't work at all 
with regards to routing virgin interrupts and this change just exposed that.

-- 
John Baldwin <jhb_at_FreeBSD.org>  <><  http://www.FreeBSD.org/~jhb/
"Power Users Use the Power to Serve"  =  http://www.FreeBSD.org
Received on Mon Jun 14 2004 - 15:16:39 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:57 UTC