EHCI: improved logic for isochronous scheduling

This patch (as1608) reworks the logic used by ehci-hcd for scheduling
isochronous transfers.  Now the modular calculations are all based on
a window that starts at the last frame scanned for isochronous
completions.  No transfer descriptors for any earlier frames can
possibly remain on the schedule, so there can be no confusion from
schedule wrap-around.  This removes the need for a "slop" region of
arbitrary size.

There's no need to check for URBs that are longer than the schedule
length.  With the old code they could throw things off by wrapping
around and appearing to end in the near future rather than the distant
future.  Now such confusion isn't possible, and the existing test for
submissions that extend too far into the future will also catch those
that exceed the schedule length.  (But there still has to be an
initial test to handle the case where the schedule already extends as
far into the future as possible.)

Delays caused by IRQ latency won't confuse the algorithm unless they
are ridiculously long (over 250 ms); they will merely reduce how far
into the future new transfers can be scheduled.  A few people have
reported problems caused by delays of 50 ms or so.  Now instead of
failing completely, isochronous transfers will experience a brief
glitch and then continue normally.

(Whether this is truly a good thing is debatable.  A latency as large
as 50 ms generally indicates a bug is present, and complete failure of
audio or video transfers draws people's attention pretty vividly.
Making the transfers more robust also makes it easier for such bugs to
remain undetected.)

Finally, ehci->next_frame is renamed to ehci->last_iso_frame, because
that better describes what it is: the last frame to have been scanned
for isochronous completions.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drivers/usb/host/ehci-sched.c

@@ -1361,7 +1361,7 @@ sitd_slot_ok (
  * given EHCI_TUNE_FLS and the slop).  Or, write a smarter scheduler!
  */
 
-#define SCHEDULE_SLOP	80	/* microframes */
+#define SCHEDULING_DELAY	40	/* microframes */
 
 static int
 iso_stream_schedule (
@@ -1370,7 +1370,7 @@ iso_stream_schedule (
 	struct ehci_iso_stream	*stream
 )
 {
-	u32			now, next, start, period, span;
+	u32			now, base, next, start, period, span;
 	int			status;
 	unsigned		mod = ehci->periodic_size << 3;
 	struct ehci_iso_sched	*sched = urb->hcpriv;
@@ -1382,12 +1382,6 @@ iso_stream_schedule (
 		span <<= 3;
 	}
 
-	if (span > mod - SCHEDULE_SLOP) {
-		ehci_dbg (ehci, "iso request %p too long\n", urb);
-		status = -EFBIG;
-		goto fail;
-	}
-
 	now = ehci_read_frame_index(ehci) & (mod - 1);
 
 	/* Typical case: reuse current schedule, stream is still active.
@@ -1396,7 +1390,6 @@ iso_stream_schedule (
 	 * slot in the schedule, implicitly assuming URB_ISO_ASAP.
 	 */
 	if (likely (!list_empty (&stream->td_list))) {
-		u32	excess;
 
 		/* For high speed devices, allow scheduling within the
 		 * isochronous scheduling threshold.  For full speed devices
@@ -1408,36 +1401,41 @@ iso_stream_schedule (
 		else
 			next = now;
 
-		/* Fell behind (by up to twice the slop amount)?
-		 * We decide based on the time of the last currently-scheduled
-		 * slot, not the time of the next available slot.
+		/*
+		 * Use ehci->last_iso_frame as the base.  There can't be any
+		 * TDs scheduled for earlier than that.
 		 */
-		excess = (stream->next_uframe - period - next) & (mod - 1);
-		if (excess >= mod - 2 * SCHEDULE_SLOP)
-			start = next + excess - mod + period *
-					DIV_ROUND_UP(mod - excess, period);
-		else
-			start = next + excess + period;
-		if (start - now >= mod) {
-			ehci_dbg(ehci, "request %p would overflow (%d+%d >= %d)\n",
-					urb, start - now - period, period,
-					mod);
-			status = -EFBIG;
+		base = ehci->last_iso_frame << 3;
+		next = (next - base) & (mod - 1);
+		start = (stream->next_uframe - base) & (mod - 1);
+
+		/* Is the schedule already full? */
+		if (unlikely(start < period)) {
+			ehci_dbg(ehci, "iso sched full %p (%u-%u < %u mod %u)\n",
+					urb, stream->next_uframe, base,
+					period, mod);
+			status = -ENOSPC;
 			goto fail;
 		}
+
+		/* Behind the scheduling threshold?  Assume URB_ISO_ASAP. */
+		if (unlikely(start < next))
+			start += period * DIV_ROUND_UP(next - start, period);
+
+		start += base;
 	}
 
 	/* need to schedule; when's the next (u)frame we could start?
 	 * this is bigger than ehci->i_thresh allows; scheduling itself
-	 * isn't free, the slop should handle reasonably slow cpus.  it
+	 * isn't free, the delay should handle reasonably slow cpus.  it
 	 * can also help high bandwidth if the dma and irq loads don't
 	 * jump until after the queue is primed.
 	 */
 	else {
 		int done = 0;
-		start = SCHEDULE_SLOP + (now & ~0x07);
 
-		/* NOTE:  assumes URB_ISO_ASAP, to limit complexity/bugs */
+		base = now & ~0x07;
+		start = base + SCHEDULING_DELAY;
 
 		/* find a uframe slot with enough bandwidth.
 		 * Early uframes are more precious because full-speed
@@ -1464,19 +1462,16 @@ iso_stream_schedule (
 
 		/* no room in the schedule */
 		if (!done) {
-			ehci_dbg(ehci, "iso resched full %p (now %d max %d)\n",
-				urb, now, now + mod);
+			ehci_dbg(ehci, "iso sched full %p", urb);
 			status = -ENOSPC;
 			goto fail;
 		}
 	}
 
 	/* Tried to schedule too far into the future? */
-	if (unlikely(start - now + span - period
-				>= mod - 2 * SCHEDULE_SLOP)) {
-		ehci_dbg(ehci, "request %p would overflow (%d+%d >= %d)\n",
-				urb, start - now, span - period,
-				mod - 2 * SCHEDULE_SLOP);
+	if (unlikely(start - base + span - period >= mod)) {
+		ehci_dbg(ehci, "request %p would overflow (%u+%u >= %u)\n",
+				urb, start - base, span - period, mod);
 		status = -EFBIG;
 		goto fail;
 	}
@@ -1490,7 +1485,7 @@ iso_stream_schedule (
 
 	/* Make sure scan_isoc() sees these */
 	if (ehci->isoc_count == 0)
-		ehci->next_frame = now >> 3;
+		ehci->last_iso_frame = now >> 3;
 	return 0;
 
  fail:
@@ -2220,16 +2215,16 @@ static void scan_isoc(struct ehci_hcd *ehci)
 		now_frame = (uf >> 3) & fmask;
 		live = true;
 	} else  {
-		now_frame = (ehci->next_frame - 1) & fmask;
+		now_frame = (ehci->last_iso_frame - 1) & fmask;
 		live = false;
 	}
 	ehci->now_frame = now_frame;
 
-	frame = ehci->next_frame;
 	for (;;) {
 		union ehci_shadow	q, *q_p;
 		__hc32			type, *hw_p;
 
+		frame = ehci->last_iso_frame;
 restart:
 		/* scan each element in frame's queue for completions */
 		q_p = &ehci->pshadow [frame];
@@ -2334,7 +2329,6 @@ static void scan_isoc(struct ehci_hcd *ehci)
 		/* Stop when we have reached the current frame */
 		if (frame == now_frame)
 			break;
-		frame = (frame + 1) & fmask;
+		ehci->last_iso_frame = (frame + 1) & fmask;
 	}
-	ehci->next_frame = now_frame;
 }

drivers/usb/host/ehci.h

@@ -143,7 +143,7 @@ struct ehci_hcd {			/* one per controller */
 	struct ehci_qh		*intr_unlink_last;
 	unsigned		intr_unlink_cycle;
 	unsigned		now_frame;	/* frame from HC hardware */
-	unsigned		next_frame;	/* scan periodic, start here */
+	unsigned		last_iso_frame;	/* last frame scanned for iso */
 	unsigned		intr_count;	/* intr activity count */
 	unsigned		isoc_count;	/* isoc activity count */
 	unsigned		periodic_count;	/* periodic activity count */