stable.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	stable@vger.kernel.org, Alan Stern <stern@rowland.harvard.edu>
Subject: [ 24/34] USB: EHCI: unlink one async QH at a time
Date: Thu,  7 Feb 2013 16:57:20 -0800	[thread overview]
Message-ID: <20130208004631.272524117@linuxfoundation.org> (raw)
In-Reply-To: <20130208004627.375461662@linuxfoundation.org>

3.7-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Alan Stern <stern@rowland.harvard.edu>

commit 6e0c3339a6f19d748f16091d0a05adeb1e1f822b upstream.

This patch (as1648) fixes a regression affecting nVidia EHCI
controllers.  Evidently they don't like to have more than one async QH
unlinked at a time.  I can't imagine how they manage to mess it up,
but at least one of them does.

The patch changes the async unlink logic in two ways:

	Each time an IAA cycle is started, only the first QH on the
	async unlink list is handled (rather than all of them).

	Async QHs do not all get unlinked as soon as they have been
	empty for long enough.  Instead, only the last one (i.e., the
	one that has been on the schedule the longest) is unlinked,
	and then only if no other unlinks are in progress at the time.

This means that when multiple QHs are empty, they won't be unlinked as
quickly as before.  That's okay; it won't affect correct operation of
the driver or add an excessive load.  Multiple unlinks tend to be
relatively rare in any case.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Reported-and-tested-by: Piergiorgio Sartor <piergiorgio.sartor@nexgo.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

---
 drivers/usb/host/ehci-q.c |   50 +++++++++++++++++++++++++++-------------------
 1 file changed, 30 insertions(+), 20 deletions(-)

--- a/drivers/usb/host/ehci-q.c
+++ b/drivers/usb/host/ehci-q.c
@@ -1197,17 +1197,26 @@ static void start_iaa_cycle(struct ehci_
 	if (ehci->async_iaa || ehci->async_unlinking)
 		return;
 
-	/* Do all the waiting QHs at once */
-	ehci->async_iaa = ehci->async_unlink;
-	ehci->async_unlink = NULL;
-
 	/* If the controller isn't running, we don't have to wait for it */
 	if (unlikely(ehci->rh_state < EHCI_RH_RUNNING)) {
+
+		/* Do all the waiting QHs */
+		ehci->async_iaa = ehci->async_unlink;
+		ehci->async_unlink = NULL;
+
 		if (!nested)		/* Avoid recursion */
 			end_unlink_async(ehci);
 
 	/* Otherwise start a new IAA cycle */
 	} else if (likely(ehci->rh_state == EHCI_RH_RUNNING)) {
+		struct ehci_qh		*qh;
+
+		/* Do only the first waiting QH (nVidia bug?) */
+		qh = ehci->async_unlink;
+		ehci->async_iaa = qh;
+		ehci->async_unlink = qh->unlink_next;
+		qh->unlink_next = NULL;
+
 		/* Make sure the unlinks are all visible to the hardware */
 		wmb();
 
@@ -1255,34 +1264,35 @@ static void end_unlink_async(struct ehci
 	}
 }
 
+static void start_unlink_async(struct ehci_hcd *ehci, struct ehci_qh *qh);
+
 static void unlink_empty_async(struct ehci_hcd *ehci)
 {
-	struct ehci_qh		*qh, *next;
-	bool			stopped = (ehci->rh_state < EHCI_RH_RUNNING);
+	struct ehci_qh		*qh;
+	struct ehci_qh		*qh_to_unlink = NULL;
 	bool			check_unlinks_later = false;
+	int			count = 0;
 
-	/* Unlink all the async QHs that have been empty for a timer cycle */
-	next = ehci->async->qh_next.qh;
-	while (next) {
-		qh = next;
-		next = qh->qh_next.qh;
-
+	/* Find the last async QH which has been empty for a timer cycle */
+	for (qh = ehci->async->qh_next.qh; qh; qh = qh->qh_next.qh) {
 		if (list_empty(&qh->qtd_list) &&
 				qh->qh_state == QH_STATE_LINKED) {
-			if (!stopped && qh->unlink_cycle ==
-					ehci->async_unlink_cycle)
+			++count;
+			if (qh->unlink_cycle == ehci->async_unlink_cycle)
 				check_unlinks_later = true;
 			else
-				single_unlink_async(ehci, qh);
+				qh_to_unlink = qh;
 		}
 	}
 
-	/* Start a new IAA cycle if any QHs are waiting for it */
-	if (ehci->async_unlink)
-		start_iaa_cycle(ehci, false);
+	/* If nothing else is being unlinked, unlink the last empty QH */
+	if (!ehci->async_iaa && !ehci->async_unlink && qh_to_unlink) {
+		start_unlink_async(ehci, qh_to_unlink);
+		--count;
+	}
 
-	/* QHs that haven't been empty for long enough will be handled later */
-	if (check_unlinks_later) {
+	/* Other QHs will be handled later */
+	if (count > 0) {
 		ehci_enable_event(ehci, EHCI_HRTIMER_ASYNC_UNLINKS, true);
 		++ehci->async_unlink_cycle;
 	}



  parent reply	other threads:[~2013-02-08  0:57 UTC|newest]

Thread overview: 37+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-02-08  0:56 [ 00/34] 3.7.7-stable review Greg Kroah-Hartman
2013-02-08  0:56 ` [ 01/34] powerpc/mm: Fix hash computation function Greg Kroah-Hartman
2013-02-08  0:56 ` [ 02/34] digsig: Fix memory leakage in digsig_verify_rsa() Greg Kroah-Hartman
2013-02-08  0:56 ` [ 03/34] drm/radeon/evergreen+: wait for the MC to settle after MC blackout Greg Kroah-Hartman
2013-02-08  0:57 ` [ 04/34] drm/radeon: add WAIT_UNTIL to the non-VM safe regs list for cayman/TN Greg Kroah-Hartman
2013-02-08  0:57 ` [ 05/34] drm/radeon: add quirk for RV100 board Greg Kroah-Hartman
2013-02-08  0:57 ` [ 06/34] drm/radeon: fix MC blackout on evergreen+ Greg Kroah-Hartman
2013-02-08  0:57 ` [ 07/34] drm/radeon: fix backend map setup on 1 RB sumo boards Greg Kroah-Hartman
2013-02-08  0:57 ` [ 08/34] drm/radeon: protect against div by 0 in backend setup Greg Kroah-Hartman
2013-02-08  0:57 ` [ 09/34] drm/radeon: prevent crash in the ring space allocation Greg Kroah-Hartman
2013-02-08  0:57 ` [ 10/34] drm/radeon: Calling object_unrefer() when creating fb failure Greg Kroah-Hartman
2013-02-08  0:57 ` [ 11/34] x86-64: Replace left over sti/cli in ia32 audit exit code Greg Kroah-Hartman
2013-02-08  0:57 ` [ 12/34] sched/rt: Use root_domain of rt_rq not current processor Greg Kroah-Hartman
2013-02-08  0:57 ` [ 13/34] mtd: davinci_nand: fix modular build with CONFIG_OF=y Greg Kroah-Hartman
2013-02-08  0:57 ` [ 14/34] nilfs2: fix fix very long mount time issue Greg Kroah-Hartman
2013-02-08  0:57 ` [ 15/34] mm/hugetlb: set PTE as huge in hugetlb_change_protection and remove_migration_pte Greg Kroah-Hartman
2013-02-08  0:57 ` [ 16/34] drivers/rtc/rtc-isl1208.c: call rtc_update_irq() from the alarm irq handler Greg Kroah-Hartman
2013-02-08  0:57 ` [ 17/34] drivers/rtc/rtc-pl031.c: fix the missing operation on enable Greg Kroah-Hartman
2013-02-08  0:57 ` [ 18/34] USB: ftdi_sio: add Zolix FTDI PID Greg Kroah-Hartman
2013-02-08  0:57 ` [ 19/34] USB: ftdi_sio: add PID/VID entries for ELV WS 300 PC II Greg Kroah-Hartman
2013-02-08  0:57 ` [ 20/34] USB: option: add support for Telit LE920 Greg Kroah-Hartman
2013-02-08  0:57 ` [ 21/34] USB: option: add Changhong CH690 Greg Kroah-Hartman
2013-02-08  0:57 ` [ 22/34] USB: qcserial: add Telit Gobi QDL device Greg Kroah-Hartman
2013-02-08  0:57 ` [ 23/34] USB: EHCI: remove ASS/PSS polling timeout Greg Kroah-Hartman
2013-02-08  0:57 ` Greg Kroah-Hartman [this message]
2013-02-08  0:57 ` [ 25/34] USB: EHCI: fix timer bug affecting port resume Greg Kroah-Hartman
2013-02-08  0:57 ` [ 26/34] USB: EHCI: fix bug in scheduling periodic split transfers Greg Kroah-Hartman
2013-02-08  0:57 ` [ 27/34] usb: Using correct way to clear usb3.0 devices remote wakeup feature Greg Kroah-Hartman
2013-02-08  0:57 ` [ 28/34] USB: storage: Define a new macro for USB storage match rules Greg Kroah-Hartman
2013-02-08  0:57 ` [ 29/34] USB: storage: optimize to match the Huawei USB storage devices and support new switch command Greg Kroah-Hartman
2013-02-08  0:57 ` [ 30/34] drivers: xhci: fix incorrect bit test Greg Kroah-Hartman
2013-02-08  0:57 ` [ 31/34] xhci: Fix isoc TD encoding Greg Kroah-Hartman
2013-02-08  0:57 ` [ 32/34] xhci: Fix TD size for isochronous URBs Greg Kroah-Hartman
2013-02-08  0:57 ` [ 33/34] USB: XHCI: fix memory leak of URB-private data Greg Kroah-Hartman
2013-02-08  0:57 ` [ 34/34] usb: Prevent dead ports when xhci is not enabled Greg Kroah-Hartman
2013-02-08 20:29 ` [ 00/34] 3.7.7-stable review Shuah Khan
2013-02-09 12:06 ` Satoru Takeuchi

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20130208004631.272524117@linuxfoundation.org \
    --to=gregkh@linuxfoundation.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=stable@vger.kernel.org \
    --cc=stern@rowland.harvard.edu \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).