On Fri, May 21, 2010 at 10:58:19AM -0700, Dong Nguyen wrote:
> Hi Sarah,
>
> Thanks for your suggestion. My comments below:
>
> > What if you set hcd->driver->irq to NULL once you've set up MSI or
> > MSI-X, here:
> >
> > /*
> > * sucessfully setup msi/msi-x,
> > * need to disable the legacy INTx
> > */
> > if (!ret) {
> > if (hcd->irq)
> > free_irq(hcd->irq, hcd);
> > hcd->irq = -1;
> > hcd->driver->irq = NULL;
> > }
> >
> > Then you can check for that NULL pointer in your clean up function, and
> > only free the IRQ if you have setup MSI:
> >
> > else if (xhci_to_hcd(xhci)->driver->irq == NULL) {
> > free_irq(pdev->irq, xhci_to_hcd(xhci));
> > pdev->irq = -1;
> > }
> >
>
> The hcd->driver is read only memory and I don't think we can set to NULL
> once xhci initialization completed. That's why I introduced the
> msi/msi-x enable flag in first patch so I only call free_irq for
> msi/msi-x case and leave the legacy interrupt for core hcd to free irq.
> Adding one msi/msi-x flag probably easiest solution when clean up
> msi/msi-x.
Ok, why not check hcd->irq == -1 instead? Then you don't need the
flags.
Sarah Sharp
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
