Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1290550 > unrolled thread

[PATCH v2 01/10] usb: host: ehci-sched: refactor scan_isoc function

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  [PATCH v2 01/10] usb: host: ehci-sched: refactor scan_isoc function "Geyslan G. Bem" <geyslan@gmail.com> - 2015-12-13 05:10 +0100
    Re: [PATCH v2 01/10] usb: host: ehci-sched: refactor scan_isoc  function Alan Stern <stern@rowland.harvard.edu> - 2015-12-13 16:30 +0100
      Re: [PATCH v2 01/10] usb: host: ehci-sched: refactor scan_isoc function "Geyslan G. Bem" <geyslan@gmail.com> - 2015-12-13 17:40 +0100

#1290550 — [PATCH v2 01/10] usb: host: ehci-sched: refactor scan_isoc function

This patch removes an infinite 'for' loop and makes use of the already
existing 'restart' tag instead, reducing one leading tab.

It also puts the easier evaluation (live variable) to be the first in
two conditionals.

  if (live && frame == now_frame) { ...

  if (live && ((frame == now_frame) || ...

The comments and code were corrected conforming file coding style.

Tested by compilation only.
Caught by checkpatch:
 WARNING: Too many leading tabs - consider code refactoring

Signed-off-by: Geyslan G. Bem <geyslan@gmail.com>
---
  v2:
    Fixes file coding indentation.
    Puts 'live' variable as the first evaluation in two ifs.
---
 drivers/usb/host/ehci-sched.c | 201 +++++++++++++++++++++---------------------
 1 file changed, 100 insertions(+), 101 deletions(-)

diff --git a/drivers/usb/host/ehci-sched.c b/drivers/usb/host/ehci-sched.c
index f9a3327..eb14143 100644
--- a/drivers/usb/host/ehci-sched.c
+++ b/drivers/usb/host/ehci-sched.c
@@ -2379,9 +2379,11 @@ static int sitd_submit (struct ehci_hcd *ehci, struct urb *urb,
 
 static void scan_isoc(struct ehci_hcd *ehci)
 {
-	unsigned	uf, now_frame, frame;
-	unsigned	fmask = ehci->periodic_size - 1;
-	bool		modified, live;
+	unsigned		uf, now_frame, frame;
+	unsigned		fmask = ehci->periodic_size - 1;
+	bool			modified, live;
+	union ehci_shadow	q, *q_p;
+	__hc32			type, *hw_p;
 
 	/*
 	 * When running, scan from last scan point up to "now"
@@ -2399,119 +2401,116 @@ static void scan_isoc(struct ehci_hcd *ehci)
 	ehci->now_frame = now_frame;
 
 	frame = ehci->last_iso_frame;
-	for (;;) {
-		union ehci_shadow	q, *q_p;
-		__hc32			type, *hw_p;
 
 restart:
-		/* scan each element in frame's queue for completions */
-		q_p = &ehci->pshadow [frame];
-		hw_p = &ehci->periodic [frame];
-		q.ptr = q_p->ptr;
-		type = Q_NEXT_TYPE(ehci, *hw_p);
-		modified = false;
-
-		while (q.ptr != NULL) {
-			switch (hc32_to_cpu(ehci, type)) {
-			case Q_TYPE_ITD:
-				/* If this ITD is still active, leave it for
-				 * later processing ... check the next entry.
-				 * No need to check for activity unless the
-				 * frame is current.
-				 */
-				if (frame == now_frame && live) {
-					rmb();
-					for (uf = 0; uf < 8; uf++) {
-						if (q.itd->hw_transaction[uf] &
-							    ITD_ACTIVE(ehci))
-							break;
-					}
-					if (uf < 8) {
-						q_p = &q.itd->itd_next;
-						hw_p = &q.itd->hw_next;
-						type = Q_NEXT_TYPE(ehci,
-							q.itd->hw_next);
-						q = *q_p;
+	/* Scan each element in frame's queue for completions */
+	q_p = &ehci->pshadow[frame];
+	hw_p = &ehci->periodic[frame];
+	q.ptr = q_p->ptr;
+	type = Q_NEXT_TYPE(ehci, *hw_p);
+	modified = false;
+
+	while (q.ptr != NULL) {
+		switch (hc32_to_cpu(ehci, type)) {
+		case Q_TYPE_ITD:
+			/*
+			 * If this ITD is still active, leave it for
+			 * later processing ... check the next entry.
+			 * No need to check for activity unless the
+			 * frame is current.
+			 */
+			if (live && frame == now_frame) {
+				rmb();
+				for (uf = 0; uf < 8; uf++) {
+					if (q.itd->hw_transaction[uf] &
+							ITD_ACTIVE(ehci))
 						break;
-					}
 				}
-
-				/* 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;
-				if (!ehci->use_dummy_qh ||
-				    q.itd->hw_next != EHCI_LIST_END(ehci))
-					*hw_p = q.itd->hw_next;
-				else
-					*hw_p = cpu_to_hc32(ehci,
-							ehci->dummy->qh_dma);
-				type = Q_NEXT_TYPE(ehci, q.itd->hw_next);
-				wmb();
-				modified = itd_complete (ehci, q.itd);
-				q = *q_p;
-				break;
-			case Q_TYPE_SITD:
-				/* If this SITD is still active, leave it for
-				 * later processing ... check the next entry.
-				 * No need to check for activity unless the
-				 * frame is current.
-				 */
-				if (((frame == now_frame) ||
-				     (((frame + 1) & fmask) == now_frame))
-				    && live
-				    && (q.sitd->hw_results &
-					SITD_ACTIVE(ehci))) {
-
-					q_p = &q.sitd->sitd_next;
-					hw_p = &q.sitd->hw_next;
+				if (uf < 8) {
+					q_p = &q.itd->itd_next;
+					hw_p = &q.itd->hw_next;
 					type = Q_NEXT_TYPE(ehci,
-							q.sitd->hw_next);
+							q.itd->hw_next);
 					q = *q_p;
 					break;
 				}
+			}
+
+			/*
+			 * 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;
+			if (!ehci->use_dummy_qh ||
+					q.itd->hw_next != EHCI_LIST_END(ehci))
+				*hw_p = q.itd->hw_next;
+			else
+				*hw_p = cpu_to_hc32(ehci, ehci->dummy->qh_dma);
+			type = Q_NEXT_TYPE(ehci, q.itd->hw_next);
+			wmb();
+			modified = itd_complete(ehci, q.itd);
+			q = *q_p;
+			break;
+		case Q_TYPE_SITD:
+			/*
+			 * If this SITD is still active, leave it for
+			 * later processing ... check the next entry.
+			 * No need to check for activity unless the
+			 * frame is current.
+			 */
+			if (live && ((frame == now_frame) ||
+					(((frame + 1) & fmask) == now_frame))
+				&& (q.sitd->hw_results & SITD_ACTIVE(ehci))) {
 
-				/* Take finished SITDs out of the schedule
-				 * and process them:  recycle, maybe report
-				 * URB completion.
-				 */
-				*q_p = q.sitd->sitd_next;
-				if (!ehci->use_dummy_qh ||
-				    q.sitd->hw_next != EHCI_LIST_END(ehci))
-					*hw_p = q.sitd->hw_next;
-				else
-					*hw_p = cpu_to_hc32(ehci,
-							ehci->dummy->qh_dma);
+				q_p = &q.sitd->sitd_next;
+				hw_p = &q.sitd->hw_next;
 				type = Q_NEXT_TYPE(ehci, q.sitd->hw_next);
-				wmb();
-				modified = sitd_complete (ehci, q.sitd);
 				q = *q_p;
 				break;
-			default:
-				ehci_dbg(ehci, "corrupt type %d frame %d shadow %p\n",
-					type, frame, q.ptr);
-				// BUG ();
-				/* FALL THROUGH */
-			case Q_TYPE_QH:
-			case Q_TYPE_FSTN:
-				/* End of the iTDs and siTDs */
-				q.ptr = NULL;
-				break;
 			}
 
-			/* assume completion callbacks modify the queue */
-			if (unlikely(modified && ehci->isoc_count > 0))
-				goto restart;
-		}
-
-		/* Stop when we have reached the current frame */
-		if (frame == now_frame)
+			/*
+			 * Take finished SITDs out of the schedule
+			 * and process them:  recycle, maybe report
+			 * URB completion.
+			 */
+			*q_p = q.sitd->sitd_next;
+			if (!ehci->use_dummy_qh ||
+					q.sitd->hw_next != EHCI_LIST_END(ehci))
+				*hw_p = q.sitd->hw_next;
+			else
+				*hw_p = cpu_to_hc32(ehci, ehci->dummy->qh_dma);
+			type = Q_NEXT_TYPE(ehci, q.sitd->hw_next);
+			wmb();
+			modified = sitd_complete(ehci, q.sitd);
+			q = *q_p;
 			break;
+		default:
+			ehci_dbg(ehci, "corrupt type %d frame %d shadow %p\n",
+					type, frame, q.ptr);
+			/* BUG(); */
+			/* FALL THROUGH */
+		case Q_TYPE_QH:
+		case Q_TYPE_FSTN:
+			/* End of the iTDs and siTDs */
+			q.ptr = NULL;
+			break;
+		}
 
-		/* The last frame may still have active siTDs */
-		ehci->last_iso_frame = frame;
-		frame = (frame + 1) & fmask;
+		/* Assume completion callbacks modify the queue */
+		if (unlikely(modified && ehci->isoc_count > 0))
+			goto restart;
 	}
+
+	/* Stop when we have reached the current frame */
+	if (frame == now_frame)
+		return;
+
+	/* The last frame may still have active siTDs */
+	ehci->last_iso_frame = frame;
+	frame = (frame + 1) & fmask;
+
+	goto restart;
 }
-- 
2.6.3

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [next] | [standalone]

#1290631 — Re: [PATCH v2 01/10] usb: host: ehci-sched: refactor scan_isoc function

On Sun, 13 Dec 2015, Geyslan G. Bem wrote:

> This patch removes an infinite 'for' loop and makes use of the already
> existing 'restart' tag instead, reducing one leading tab.
> 
> It also puts the easier evaluation (live variable) to be the first in
> two conditionals.
> 
>   if (live && frame == now_frame) { ...
> 
>   if (live && ((frame == now_frame) || ...

You should never do this.  The first test in a conditional should be 
the one that is most likely to cause the expression to fail.  In this 
case, live is almost always true whereas frame == now_frame is true 
only some percentage of the time.  Therefore the frame == now_frame 
test should come first.

Alan Stern

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1290642

2015-12-13 12:29 GMT-03:00 Alan Stern <stern@rowland.harvard.edu>:
> On Sun, 13 Dec 2015, Geyslan G. Bem wrote:
>
>> This patch removes an infinite 'for' loop and makes use of the already
>> existing 'restart' tag instead, reducing one leading tab.
>>
>> It also puts the easier evaluation (live variable) to be the first in
>> two conditionals.
>>
>>   if (live && frame == now_frame) { ...
>>
>>   if (live && ((frame == now_frame) || ...
>
> You should never do this.  The first test in a conditional should be
> the one that is most likely to cause the expression to fail.  In this
> case, live is almost always true whereas frame == now_frame is true
> only some percentage of the time.  Therefore the frame == now_frame
> test should come first.
I see.
Doing v3 and undoing the reordering.

>
> Alan Stern
>



-- 
Regards,

Geyslan G. Bem
hackingbits.com
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web