luck/tmp_suning_uos_patched
1
0
Fork 1
Code Issues Pull Requests Packages Projects Releases Wiki Activity

USB: ehci completes high speed ISO URBs sooner

Browse Source 
This has some bugfixes for the EHCI driver's ISO transfer scanning
logic.  It was leaving ITDs and SITDs on the schedule too long, for
a few different reasons, which caused trouble.

  (a)	Look at all microframes for high speed transfers, not just
	the ones we expect to have finished.  This way transfers
	ending mid-frame will complete without needing another IRQ.
	This also minimizes bogus scheduling underruns (e.g. EL2NSYNC).

  (b)	When we encounter an ISO transfer (either speed, but this
	hits mostly at full speed) that's not yet been completed,
	immediately stop scanning; we've caught up to the hardware,
	no matter what other indications might say.

  (c)	Always clean up ITDs (for high speed transfers) when the HC
	is no longer running.

I'm not sure whether the last one has been observed before, but both
the others have been reported with "real world" audio and video code.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This commit is contained in:
David Brownell 2008-01-07 00:47:42 -08:00 committed by Greg Kroah-Hartman
parent aa16ca307e
commit 79592b722e
1 changed files with 27 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

41
drivers/usb/host/ehci-sched.c
View File

@ -2123,17 +2123,9 @@ scan_periodic (struct ehci_hcd *ehci)
for (;;) {
union ehci_shadow q, *q_p;
__hc32 type, *hw_p;
unsigned uframes;
unsigned incomplete = false;
/* don't scan past the live uframe */
frame = now_uframe >> 3;
if (frame == (clock >> 3))
uframes = now_uframe & 0x07;
else {
/* safe to scan the whole frame at once */
now_uframe |= 0x07;
uframes = 8;
}
restart:
/* scan each element in frame's queue for completions */
@ -2171,12 +2163,15 @@ scan_periodic (struct ehci_hcd *ehci)
q = q.fstn->fstn_next;
break;
case Q_TYPE_ITD:
/* skip itds for later in the frame */
/* If this ITD is still active, leave it for
* later processing ... check the next entry.
*/
rmb ();
for (uf = live ? uframes : 8; uf < 8; uf++) {
for (uf = 0; uf < 8 && live; uf++) {
if (0 == (q.itd->hw_transaction [uf]
& ITD_ACTIVE(ehci)))
continue;
incomplete = true;
q_p = &q.itd->itd_next;
hw_p = &q.itd->hw_next;
type = Q_NEXT_TYPE(ehci,
@ -2184,10 +2179,12 @@ scan_periodic (struct ehci_hcd *ehci)
q = *q_p;
break;
}
if (uf != 8)
if (uf < 8 && live)
break;
/* this one's ready ... HC won't cache the
/* Take finished ITDs out of the schedule
* and process them: recycle, maybe report
* URB completion. HC won't cache the
* pointer for much longer, if at all.
*/
*q_p = q.itd->itd_next;
@ -2198,8 +2195,12 @@ scan_periodic (struct ehci_hcd *ehci)
q = *q_p;
break;
case Q_TYPE_SITD:
/* If this SITD is still active, leave it for
* later processing ... check the next entry.
*/
if ((q.sitd->hw_results & SITD_ACTIVE(ehci))
&& live) {
incomplete = true;
q_p = &q.sitd->sitd_next;
hw_p = &q.sitd->hw_next;
type = Q_NEXT_TYPE(ehci,
@ -2207,6 +2208,11 @@ scan_periodic (struct ehci_hcd *ehci)
q = *q_p;
break;
}
/* Take finished SITDs out of the schedule
* and process them: recycle, maybe report
* URB completion.
*/
*q_p = q.sitd->sitd_next;
*hw_p = q.sitd->hw_next;
type = Q_NEXT_TYPE(ehci, q.sitd->hw_next);
@ -2232,7 +2238,14 @@ scan_periodic (struct ehci_hcd *ehci)
}
}
/* stop when we catch up to the HC */
/* If we can tell we caught up to the hardware, stop now.
* We can't advance our scan without collecting the ISO
* transfers that are still pending in this frame.
*/
if (incomplete && HC_IS_RUNNING(ehci_to_hcd(ehci)->state)) {
ehci->next_uframe = now_uframe;
break;
}
// FIXME: this assumes we won't get lapped when
// latencies climb; that should be rare, but...