On Mon, 27 May 2013, Oleksij Rempel wrote:
> > Hmmm. Maybe we can narrow this down. What happens if you apply only
> > parts of the commit?
> >
> > For example, on top of c97041a, try applying only the hunks that change
> > ehci-sched.c. If that works, try applying also only the last hunk
> > affecting ehci-q.c. If that works, try applying also the two previous
> > hunks.
> >
>
> after ehci-sched.c hank i have this problem again:
>
> +++ b/drivers/usb/host/ehci-sched.c
> @@ -792,7 +792,6 @@ static int qh_schedule(struct ehci_hcd *ehci, struct
> ehci_qh
> unsigned frame; /* 0..(qh->period - 1), or
> NO_FRAME */
> struct ehci_qh_hw *hw = qh->hw;
>
> - qh_refresh(ehci, qh);
> hw->hw_next = EHCI_LIST_END(ehci);
> frame = qh->start;
>
> @@ -844,8 +843,6 @@ static int qh_schedule(struct ehci_hcd *ehci, struct
> ehci_qh
> } else
> ehci_dbg (ehci, "reused qh %p schedule\n", qh);
>
> - /* stuff into the periodic schedule */
> - qh_link_periodic(ehci, qh);
> done:
> return status;
> }
> @@ -891,6 +888,12 @@ static int intr_submit (
> qh = qh_append_tds(ehci, urb, qtd_list, epnum, &urb->ep->hcpriv);
> BUG_ON (qh == NULL);
>
> + /* stuff into the periodic schedule */
> + if (qh->qh_state == QH_STATE_IDLE) {
> + qh_refresh(ehci, qh);
> + qh_link_periodic(ehci, qh);
> + }
> +
> /* ... update usbfs periodic stats */
> ehci_to_hcd(ehci)->self.bandwidth_int_reqs++;
Good grief, I can't believe I forgot to include this material in the
original commit. I really am getting careless...
Anyway, this should fix the problem.
Alan Stern
Index: usb-3.9/drivers/usb/host/ehci-sched.c
===================================================================
--- usb-3.9.orig/drivers/usb/host/ehci-sched.c
+++ usb-3.9/drivers/usb/host/ehci-sched.c
@@ -646,6 +646,10 @@ static void end_unlink_intr(struct ehci_
/* reschedule QH iff another request is queued */
if (!list_empty(&qh->qtd_list) && ehci->rh_state == EHCI_RH_RUNNING) {
rc = qh_schedule(ehci, qh);
+ if (rc == 0) {
+ qh_refresh(ehci, qh);
+ qh_link_periodic(ehci, qh);
+ }
/* An error here likely indicates handshake failure
* or no space left in the schedule. Neither fault
@@ -653,9 +657,10 @@ static void end_unlink_intr(struct ehci_
*
* FIXME kill the now-dysfunctional queued urbs
*/
- if (rc != 0)
+ else {
ehci_err(ehci, "can't reschedule qh %p, err %d\n",
qh, rc);
+ }
}
/* maybe turn off periodic schedule */
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
