On Mon, 2009-06-22 at 15:19 -0500, Robert Noland wrote: > On Mon, 2009-06-22 at 15:13 -0500, Robert Noland wrote: > > On Mon, 2009-06-22 at 20:08 +0000, John Baldwin wrote: > > > Author: jhb > > > Date: Mon Jun 22 20:08:06 2009 > > > New Revision: 194644 > > > URL: http://svn.freebsd.org/changeset/base/194644 > > > > > > Log: > > > Enable MSI in the MSI capability registers any time that the first > > > message > > > in an MSI group is enabled, not just if the address/data pair are not > > > initialized. > > > > This should fix the issues with broken interrupts with drm on Intel > > graphics chips using msi. That is the slow windows after VT Switch > > issue. It however is still currently broken on HEAD due to another > > issue, which will hopefully be resolved soon. > > It may actually work now on non-SMP, though I haven't tested that case.
Ok, I have confirmed that this alone fixes things on UP. SMP is still
broken.
robert.
> >
> > > Reported by: rnoland
> > > MFC after: 1 week
> > >
> > > Modified:
> > > head/sys/dev/pci/pci.c
> > >
> > > Modified: head/sys/dev/pci/pci.c
> > > ==============================================================================
> > > --- head/sys/dev/pci/pci.c Mon Jun 22 19:35:39 2009
> > > (r194643)
> > > +++ head/sys/dev/pci/pci.c Mon Jun 22 20:08:06 2009
> > > (r194644)
> > > @@ -2883,8 +2883,10 @@ pci_setup_intr(device_t dev, device_t ch
> > > goto bad;
> > > dinfo->cfg.msi.msi_addr = addr;
> > > dinfo->cfg.msi.msi_data = data;
> > > - pci_enable_msi(child, addr, data);
> > > }
> > > + if (dinfo->cfg.msi.msi_handlers == 0)
> > > + pci_enable_msi(child, dinfo->cfg.msi.msi_addr,
> > > + dinfo->cfg.msi.msi_data);
> > > dinfo->cfg.msi.msi_handlers++;
> > > } else {
> > > KASSERT(dinfo->cfg.msix.msix_alloc > 0,
--
Robert Noland <rnol...@freebsd.org>
FreeBSD
signature.asc
Description: This is a digitally signed message part
