On Sat, 2014-05-03 at 10:29 +0800, Andrew Cooks wrote:
> Hi Alex
>
> On Fri, May 2, 2014 at 12:27 AM, Alex Williamson
> <alex.williamson@xxxxxxxxxx> wrote:
> > The existing quirk for these devices doesn't really solve the problem,
> > re-implement it using the DMA alias iterator. We'll come back later
> > and remove the existing quirk and dma_source interface.
> >
> > Signed-off-by: Alex Williamson <alex.williamson@xxxxxxxxxx>
> > ---
> > drivers/pci/quirks.c | 16 ++++++++++++++++
> > 1 file changed, 16 insertions(+)
> >
> > diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
> > index e729206..a458c6b 100644
> > --- a/drivers/pci/quirks.c
> > +++ b/drivers/pci/quirks.c
> > @@ -3333,6 +3333,22 @@ int pci_dev_specific_reset(struct pci_dev *dev, int probe)
> > return -ENOTTY;
> > }
> >
> > +static void quirk_dma_func0_alias(struct pci_dev *dev)
> > +{
> > + if (PCI_SLOT(dev->devfn) != 0)
> > + dev->dma_func_alias |= (1 << 0);
> > +}
> > +
> > +/*
> > + * https://bugzilla.redhat.com/show_bug.cgi?id=605888
> > + *
> > + * Some Ricoh devices use function 0 as the PCIe requester ID for DMA.
> > + */
> > +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_RICOH, 0xe822, quirk_dma_func0_alias);
> > +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_RICOH, 0xe230, quirk_dma_func0_alias);
> > +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_RICOH, 0xe832, quirk_dma_func0_alias);
> > +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_RICOH, 0xe832, quirk_dma_func0_alias);
>
> 0xe832 is listed twice.
oops, copy-paste error
> Previously only 0xe832 needed the dma alias on my thinkpad T410, which
> has all three devices.
>
> > +
> > static struct pci_dev *pci_func_0_dma_source(struct pci_dev *dev)
> > {
> > if (!PCI_FUNC(dev->devfn))
> >
>
> Unfortunately, this quirk doesn't work for me. I tried it without
> modification, as well as with each alias individually. I get:
>
> Set context mapping for 0d:00.3
> firewire_ohci 0000:0d:00.3: added OHCI v1.10 device as card 0, 4 IR +
> 4 IT contexts, quirks 0x11
> dmar: DRHD: handling fault status reg 2
> dmar: DMAR:[DMA Read] Request device [0d:00.0] fault addr fffff000
> DMAR:[fault reason 02] Present bit in context entry is clear
>
> I think I need to see
> Set context mapping for 0d:00.0
> before
> Set context mapping for 0d:00.3
It would actually be the reverse, we always set the device, then the
alias for the device.
> in the log, but it's not there. I'd love to look into this and
> understand it properly, but I don't have time for the next four weeks.
>
> The devices are attached as follows:
> BDF, device ID
> 0d:00.0, e822
> 0d:00.1, e230
> 0d:00.3, e832
>
> The kernel log is attached.
Hmm, there are only a few reasons why you'd never see 0d:00.3 followed
by 0d:00.0...
1) dma_func_alias bit 0 isn't getting set on 0d:00.3; we are building
with CONFIG_PCI_QUIRKS=y, right?
2) domain_context_mapping_one called from domain_context_mapping_cb
returns !0; there's only one possible non-zero return for a non-vm,
non-si domain
3) something is broken in the first loop of pci_for_each_dma_alias; I'm
not seeing anything obvious
Anyway, appreciate an debugging you're able to fit in, my only ricoh
device has only function 0. Thanks,
Alex
--
To unsubscribe from this list: send the line "unsubscribe linux-pci" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
