On Fri, 17 Dec 2010, Sarah Sharp wrote: > Alan, > > Can the USB core rely on hcd->irq being equal to pci_dev->irq in > suspend_common() for PCI host controllers that use legacy interrupts? I'm not aware of any PCI host controller drivers that change it (except for some reason, ohci-hcd.c sets hcd->irq to -1 in ohci_stop() -- this won't interfere with suspend). I also can't think of any reason why it would need to be changed. > > > Yes, that's possible. Here are the modifications: > > > > > > 1. Set hcd->irq = pdev->irq in xhci driver when using MSI; > > > 2. Synchronize irq in xhci_suspend() only when using MSI-X; > > > 3. Synchronize irq only if pci_dev->irq is equal to hcd->irq in > > > suspend_common(). > > Andiry, if Alan says we can't rely on that fact, you could change the > suspend_common() code to skip the irq sync if pci_dev->irq is equal to > -1. You would have to set pci_dev->irq to -1 when you enable MSI-X in > the xHCI driver. You want to be a little careful about setting ->irq to -1 and comparing the resulting values, because it is supposed to be an unsigned int. There have been repeated discussions regarding what irq value is supposed to mean "no irq", but I don't remember the outcomes. Why not just add an extra flag or two instead of relying on magic values? Alan Stern -- 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
