On Mon, Mar 10, 2014 at 01:09:04PM +0000, David Laight wrote: > From: Mathias Nyman > > On 03/09/2014 04:20 PM, Igor Gnatenko wrote: > > > After suspend another Renesas PCI-X USB 3.0 card doesn't work. > > > 03:00.0 USB controller [0c03]: Renesas Technology Corp. uPD720202 USB 3.0 Host Controller > > [1912:0015] (rev 02) (prog-if 30 [XHCI]) > ... > > > diff --git a/drivers/usb/host/xhci-pci.c b/drivers/usb/host/xhci-pci.c > > > index 04f986d..13d4add 100644 > > > --- a/drivers/usb/host/xhci-pci.c > > > +++ b/drivers/usb/host/xhci-pci.c > > > @@ -143,9 +143,7 @@ static void xhci_pci_quirks(struct device *dev, struct xhci_hcd *xhci) > > > xhci->quirks |= XHCI_TRUST_TX_LENGTH; > > > } > > > if (pdev->vendor == PCI_VENDOR_ID_RENESAS && > > > - pdev->device == 0x0015 && > > > - pdev->subsystem_vendor == PCI_VENDOR_ID_SAMSUNG && > > > - pdev->subsystem_device == 0xc0cd) > > > + pdev->device == 0x0015) > > > xhci->quirks |= XHCI_RESET_ON_RESUME; > > > if (pdev->vendor == PCI_VENDOR_ID_VIA) > > > xhci->quirks |= XHCI_RESET_ON_RESUME; > > > > > > > This will set the resume quirk for all devices with a Renesas uPD720202 > > host. > > > > Do they all need this quirk, or should just this card manufacturer be > > added to the quirk? (subsystem vendor/device) > > Given the number of systems that seem to need this quirk, it is almost > worth reversing the quick? No, we should avoid applying this quirk to all systems. > ISTR there was a system where it caused grief? The behavior of the quirk itself causes users grief. When the host is reset on resume from suspend, we tell the USB core the host lost power. The core will then disconnect all USB devices and re-enumerate them. If we apply this quirk broadly, we increase the time to bring up USB devices on resume (which is critical for mobile systems), and USB storage devices will be re-enumerated. Without this quirk, I can start watching a movie stored on a USB hard drive device, suspend the laptop, resume it, and the movie will continue playing. With this quirk, the storage device will re-enumerated as a new block device, and the movie will stop playing on resume. So, I would like to limit the impact of this quirk as much as possible. I wish I had limited it further to specific Etron and VIA versions; maybe we can dig that information out of the original bug reports and limit it to only impacted revisions. > Although it might be that some delays have been added that fix that. I think you're remembering a different quirk? 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
