xhci: fix off-by-one-trb

2017-11-06 Thread Paul Irofti
The following diff fixes the overrun issues I have been seing with isoc transfers when using video(1). Doesn't mean video(1) works now :) OK? Index: xhci.c === RCS file: /cvs/src/sys/dev/usb/xhci.c,v retrieving revision 1.77 diff -u

Re: xhci: fix off-by-one-trb

2017-11-06 Thread Stefan Sperling
On Mon, Nov 06, 2017 at 08:53:59PM +0200, Paul Irofti wrote: > The following diff fixes the overrun issues I have been seing with isoc > transfers when using video(1). Doesn't mean video(1) works now :) > > OK? I doubt this is correct. I suspect your interrupt storm appears to be fixed only becau

Re: xhci: fix off-by-one-trb

2017-11-07 Thread Paul Irofti
> http://vcs.in-berlin.de/stsp_xhci/ci/e122929b48589657?sbs=0 Nice repo, would have been useful to know about it sooner :) > With your diff, the last TRB will not be processed because > xhci_xfer_tdsize() now returns zero for the second-to-last TRB. Okay, if that is true than xhci_device_generic

Re: xhci: fix off-by-one-trb

2017-11-07 Thread Martin Pieuchot
On 07/11/17(Tue) 17:16, Paul Irofti wrote: > > http://vcs.in-berlin.de/stsp_xhci/ci/e122929b48589657?sbs=0 > > Nice repo, would have been useful to know about it sooner :) > > > With your diff, the last TRB will not be processed because > > xhci_xfer_tdsize() now returns zero for the second-to-la