cvs commit: src/sys/dev/bge if_bge.c
jhb at freebsd.org
Wed Dec 27 22:10:36 PST 2006
On Saturday 16 December 2006 09:19, John Baldwin wrote:
> On Friday 15 December 2006 20:42, John Polstra wrote:
> > On 15-Dec-2006 David O'Brien wrote:
> > > On Fri, Dec 15, 2006 at 01:26:55PM -0800, Peter Grehan wrote:
> > >> >I have a Tyan K8W (s2885) which I experienced this on:
> > >>
> > >> That's using the the AMD 8131 right ?
> > >
> > > Yep.
> > >
> > >> Known not to support MSI.
> > >
> > > I assumed the code would check and not attempt MSI if not supported.
> > That would be the job of the PCI infrastructure, not of the bge
> > driver. Quite a few motherboards/chipsets indicate MSI support
> > in their PCI configuration registers but do not in fact have
> > working MSI support. Scott and I recently added quirk support in
> > sys/dev/pci/pci.c to deal with that. Feel free to add a new entry
> > there.
> But I have a separate patch for the 8131 as it is treated differently.
> Please try http://www.FreeBSD.org/~jhb/patches/msi_pcib.patch
Can someone with a system with an AMD 8131 HT-PCIX bridge please test this
patch? I know Matt (cc'd) has one and I think David has one. The patch
should disable MSI for any devices behind the 8131 bridge, but not
necessarily all devices in the system.
More information about the cvs-all