> > Wei, thanks for you feedback, this patch is just follow the function
> > 'quirk_msi_intx_disable_ati_bug' in quirks.c Is it ok if I revise code as:
> > static void quirk_msi_intx_disable_qca_bug(struct pci_dev *dev)
> > {
> > If (dev->revision < 0x18)
> > dev->dev_flags |= PCI_DEV_FLAGS_MSI_INTX_DISABLE_BUG;
> > }
>
> Per the comment in pci_ids.h, you should not add #defines there unless they
> are used in more than one place. So in this case, you would just use the hex
> constants directly in quirks.c, as we already do for the existing ATTANSIC
> quirks.
>
> As far as the loop with pci_get_device(), the revised patch is probably what
> you want, but it is not functionally equivalent to the original. Let's say you
> have two devices:
>
> 1) 8161 rev 0x18
> 2) 8171 rev 0x17
>
> The original patch will disable MSI for the 8161 because there is
> *another* device (the 8171) with rev < 0x18. I doubt that's what you want.
>
> I'd like to see a dev_info() saying that you're disabling MSI for this device so
> that when somebody complains "my device should be using MSI but isn't,"
> it's easy to figure out why not.
>
Bjorn, your example is very clear for me. I made a bug for the original patch :)
BTW. The flag of PCI_DEV_FLAGS_MSI_INTX_DISABLE_BUG isn't disable MSI on such devices,
But not enable PCI INT_DISABLE bit for PCI_COMMAND register when enable MSI function.
Some devices like AR8161 with revision lower than 0x18 will not issue MSI interrupt if INT_DISABLE bit is set.
Thanks
Xiong
--
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
