No I think that's correct. Both the MSI and MSI-X init functions will leave all MSIs masked after init. The IRQ enable/disable code will then end up unmasking the IRQs as needed. Maybe your device's IRQ isn't getting initialized properly? Jesse On Tuesday, January 20, 2009 7:04 am Venugopal Busireddy wrote: > I think a similar masking problem exists in msix_capability_init() > too... The current code leaves the per vector mask bits in the default > reset state (masked!). I am working around the problem by clearing the > mask bits towards the end of that function. I am not sure if the driver > is supposed to clear these bits and I have a buggy driver! > > Venu > > > -----Original Message----- > > From: linux-pci-owner@xxxxxxxxxxxxxxx [mailto:linux-pci- > > owner@xxxxxxxxxxxxxxx] On Behalf Of Matthew Wilcox > > Sent: Monday, January 19, 2009 06:40 PM > > To: linux-pci@xxxxxxxxxxxxxxx > > Cc: Matthew Wilcox; Jike Song; Jesse Barnes; stable@xxxxxxxxxx > > Subject: [PATCH] Fix MSI masking bug > > > > Commit 5993760f7fc75b77e4701f1e56dc84c0d6cf18d5 replaced: > > > > - pci_write_config_dword(dev, > > - msi_mask_bits_reg(pos, > > is_64bit_address(control)), > > > - maskbits); > > + pci_write_config_dword(dev, entry->msi_attrib.is_64, > > maskbits); > > > > which, instead of writing to an offset from the MSI capability, writes > > to the PCI vendor and device IDs. Fortunately these are specified to > > be read-only, and we won't even do the write if the capability is > > 64-bit > > > (due to the write being misaligned), but it would be nice to actually > > do > > > the masking. > > > > Signed-off-by: Matthew Wilcox <willy@xxxxxxxxxxxxxxx> > > Cc: Jike Song <albcamus@xxxxxxxxx> > > Cc: Jesse Barnes <jbarnes@xxxxxxxxxxxxxxxx> > > Cc: stable@xxxxxxxxxx > > --- > > drivers/pci/msi.c | 4 +++- > > 1 files changed, 3 insertions(+), 1 deletions(-) > > > > diff --git a/drivers/pci/msi.c b/drivers/pci/msi.c > > index b4a90ba..429c596 100644 > > --- a/drivers/pci/msi.c > > +++ b/drivers/pci/msi.c > > @@ -412,7 +412,9 @@ static int msi_capability_init(struct pci_dev > > *dev) > > > temp = (1 << multi_msi_capable(control)); > > temp = ((temp - 1) & ~temp); > > maskbits |= temp; > > - pci_write_config_dword(dev, entry->msi_attrib.is_64, > > maskbits); > > + pci_write_config_dword(dev, > > + msi_mask_bits_reg(pos, entry->msi_attrib.is_64), > > + maskbits); > > entry->msi_attrib.maskbits_mask = temp; > > } > > list_add_tail(&entry->list, &dev->msi_list); > > -- > > 1.5.6.5 > > > > -- > > 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 -- Jesse Barnes, Intel Open Source Technology Center -- 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
