RE: [PATCH] usb: xhci: Renesas uPD720202 needs short TX quirk

[Yoshihiro Shimoda]

Hi,

Sorry, I missed this email...

> -----Original Message-----
> From: Felipe Balbi
> Sent: Friday, August 18, 2017 10:10 AM
> 
> Hi,
> 
> Kai-Heng Feng <kai.heng.feng@xxxxxxxxxxxxx> writes:
> >>> When plugging Logitech C920 webcam, warning messages filled up dmesg:
> >>> [77117.655018] xhci_hcd 0000:0c:00.0: WARN Successful completion on short TX: needs XHCI_TRUST_TX_LENGTH quirk?
> >>> [77117.659018] xhci_hcd 0000:0c:00.0: WARN Successful completion on short TX: needs XHCI_TRUST_TX_LENGTH quirk?
> >>
> >> have you confirmed this is needed for this controller?
> > I just found commit d2f48f05cd2a2 ("usb: xhci: ASMedia ASM1042A
> > chipset need shorts TX quirk") and did the same thing for this
> > controller.
> >
> >> Anybody from Renesas has confirmed it?
> > No, it's a user reported problem, please check the bug report in the link.
> >
> >> Do you have an errata document to refer to?
> > No. Probably need Renesas guy to provide it.
> 
> or confirm it. Yoshihiro, do you know if this is needed?

I know. This behavior is the same as R-Car H3. But, I didn't find any errata
documents. So, I'm asking HW guys why this environment causes it. However,
I don't get any feedback yet. I think we need any reason for applying such a quirk.

Best regards,
Yoshihiro Shimoda

> >>> [77122.622952] handle_tx_event: 541 callbacks suppressed
> >>>
> >>> No more warning messages with XHCI_TRUST_TX_LENGTH applied.
> >>>
> >>> BugLink: https://bugs.launchpad.net/bugs/1710548
> >>> Signed-off-by: Kai-Heng Feng <kai.heng.feng@xxxxxxxxxxxxx>
> >>> ---
> >>>  drivers/usb/host/xhci-pci.c | 4 +++-
> >>>  1 file changed, 3 insertions(+), 1 deletion(-)
> >>>
> >>> diff --git a/drivers/usb/host/xhci-pci.c b/drivers/usb/host/xhci-pci.c
> >>> index 8071c8fdd15e..8566b43e19ba 100644
> >>> --- a/drivers/usb/host/xhci-pci.c
> >>> +++ b/drivers/usb/host/xhci-pci.c
> >>> @@ -202,8 +202,10 @@ static void xhci_pci_quirks(struct device *dev, struct xhci_hcd *xhci)
> >>>               xhci->quirks |= XHCI_BROKEN_STREAMS;
> >>>       }
> >>>       if (pdev->vendor == PCI_VENDOR_ID_RENESAS &&
> >>> -                     pdev->device == 0x0015)
> >>> +                     pdev->device == 0x0015) {
> >>
> >> unnecessary
> >>
> >>>               xhci->quirks |= XHCI_RESET_ON_RESUME;
> >>> +             xhci->quirks |= XHCI_TRUST_TX_LENGTH;
> >>
> >>                 xhci->quirks |= XHCI_RESET_ON_RESUME |
> >>                                 XHCI_TRUST_TX_LENGTH;
> >>
> >>> +     }
> >>
> >> unnecessary
> >
> > Do you mean that this quirk just hide the warning, it doesn't fix the
> > root cause?
> 
> no, I meant that you need to add {} if you make a single statement out
> of the quirks.
> 
> --
> balbi