Re: [PATCH] xhci: wait CNR when doing xhci resume

[Greg KH <gregkh@linuxfoundation.org>]

On Mon, Aug 12, 2019 at 03:24:52PM +0800, Rick Tseng wrote:
> From: Rick <rtseng@nvidia.com>
> 
> NVIDIA 3.1 xHCI card would lose power when moving power state into D3Cold.
> Thus we need to wait CNR bit to clear when xhci resmue as xhci init.
> 
> Signed-off-by: Rick <rtseng@nvidia.com>

We need a "full" name on the from and signed-off-by lines, please.

> ---
>  drivers/usb/host/xhci-pci.c | 20 ++++++++++++++++++++
>  1 file changed, 20 insertions(+)
> 
> diff --git a/drivers/usb/host/xhci-pci.c b/drivers/usb/host/xhci-pci.c
> index 1e0236e..857ad8a 100644
> --- a/drivers/usb/host/xhci-pci.c
> +++ b/drivers/usb/host/xhci-pci.c
> @@ -12,6 +12,7 @@
>  #include <linux/slab.h>
>  #include <linux/module.h>
>  #include <linux/acpi.h>
> +#include <linux/iopoll.h>
>  
>  #include "xhci.h"
>  #include "xhci-trace.h"
> @@ -455,6 +456,19 @@ static void xhci_pme_quirk(struct usb_hcd *hcd)
>  	readl(reg);
>  }
>  
> +static int xhci_poll_cnr(struct usb_hcd *hcd)
> +{
> +	struct xhci_hcd *xhci = hcd_to_xhci(hcd);
> +	void __iomem *reg = &xhci->op_regs->status;
> +	u32 result;
> +	int ret;
> +
> +	ret = readl_poll_timeout_atomic(reg, result,
> +					(result & STS_CNR) == 0,
> +					1, 100 * 1000);
> +	return ret;
> +}
> +
>  static int xhci_pci_suspend(struct usb_hcd *hcd, bool do_wakeup)
>  {
>  	struct xhci_hcd	*xhci = hcd_to_xhci(hcd);
> @@ -508,6 +522,12 @@ static int xhci_pci_resume(struct usb_hcd *hcd, bool hibernated)
>  	if (pdev->vendor == PCI_VENDOR_ID_INTEL)
>  		usb_enable_intel_xhci_ports(pdev);
>  
> +	if (pdev->vendor == PCI_VENDOR_ID_NVIDIA) {

So all devices of this vendor need that?  Are you _sure_?  Why not just
blacklist a single device?

thanks,

greg k-h