3.3-rc4+: Reported regressions from 3.2

3.3-rc4+: Reported regressions from 3.2

[Rafael J. Wysocki]

Hi all,

We definitely aren't 100% in business yet with the tracking of regressions,
but since the Bugzilla is operational again, we can collect reports at least.

I'd like to use this opportunity to thank Maciej Rutecki and Florian Mickler
for their hard work on the regression tracking and to clarify some things that
seem to have caused some confusion to happen recently.  Namely, we use the
kernel Bugzilla (https://bugzilla.kernel.org) for the tracking of regressions
for the simple reason that our scripts generate the lists that are posted
(like the one below) out of Bugzilla entries created by us (and sometimes
by other people).  Those entries are used by us as a storage of information,
so we're going to add them even though some kernel developers don't seem to
like that.  We don't require developers to actually use those entries for
handling bug reports, we only need people to let us know when they should be
closed, because the bug has been fixed in the Linus' tree, or when there's
a patch available for the bug (in which case we mark it as "resolved", but
we don't close it just yet).  There is no reason for you to be openly hostile
to Maciej, who's been creating those entries recently, because he's just been
doing his job.  However, you may not want us to track kernel regressions at
all, in which case please let us know about that and we will find some other,
presumably equally interesting things to do. :-)

Yes, we might have been more verbose about what _exactly_ we've been using the
Bugzilla for and what our workflow is, but at the same time people might have
been a bit less harsh to someone who's been doing a service to the community,
completely voluntarily.

Thanks,
Rafael

---

This message contains a list of some regressions from 3.2,
for which there are no fixes in the mainline known to the tracking team.
If any of them have been fixed already, please let us know.

If you know of any other unresolved regressions from 3.2, please let us
know either and we'll add them to the list.  Also, please let us know
if any of the entries below are invalid.

Each entry from the list will be sent additionally in an automatic reply
to this message with CCs to the people involved in reporting and handling
the issue.


Listed regressions statistics:

  Date          Total  Pending  Unresolved
  ----------------------------------------
  2012-02-23       15       13          13


Unresolved regressions
----------------------

Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42798
Subject		: i915 regression with 3.3-rc3+git
Submitter	: Riccardo Magliocchetti <riccardo.magliocchetti@gmail.com>
Date		: 2012-02-17 20:48 (7 days old)
Message-ID	: <4F3EBD0D.7000906@gmail.com>
References	: http://marc.info/?l=linux-kernel&m=132951173002619&w=2


Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42776
Subject		: OF-related boot crash in 3.3.0-rc3-00188-g3ec1e88
Submitter	: Meelis Roos <mroos@linux.ee>
Date		: 2012-02-13 7:45 (11 days old)
Message-ID	: <alpine.SOC.1.00.1202130942030.1488@math.ut.ee>
References	: http://marc.info/?l=linux-kernel&m=132911916331615&w=2


Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42761
Subject		: Possible circular locking dependency (3.3-rc2)
Submitter	: Felipe Balbi <balbi@ti.com>
Date		: 2012-02-08 12:41 (16 days old)
Message-ID	: <20120208124147.GF16334@legolas.emea.dhcp.ti.com>
References	: http://marc.info/?l=linux-kernel&m=132870492311858&w=2


Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42746
Subject		: 3.3-rc2 snd_pcm lockdep backtrace
Submitter	: Josh Boyer <jwboyer@redhat.com>
Date		: 2012-02-06 14:56 (18 days old)
Message-ID	: <20120206145621.GA4771@zod.bos.redhat.com>
References	: http://marc.info/?l=linux-kernel&m=132854040916172&w=2


Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42735
Subject		: kobject (ffff88003ffbb4b8): tried to init an initialized object, something is seriously wrong.
Submitter	: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Date		: 2012-02-03 20:59 (21 days old)
Message-ID	: <20120203205953.GA2897@phenom.dumpdata.com>
References	: http://marc.info/?l=linux-kernel&m=132830315526901&w=2


Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42734
Subject		: dosemu graphics broken in v3.3-rc1
Submitter	: Hans de Bruin <jmdebruin@xmsnet.nl>
Date		: 2012-02-02 22:25 (22 days old)
First-Bad-Commit: http://git.kernel.org/linus/308e5bcbdb10452e8aba31aa21432fb67ee46d72
Message-ID	: <4F2B0D4D.3000304@xmsnet.nl>
References	: http://marc.info/?l=linux-kernel&m=132825899111982&w=2
Handled-By	: Jesse Barnes <jbarnes@virtuousgeek.org>


Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42733
Subject		: Regression 3.2 -> 3.3-rc1 10 sec hang at boot and resume, COMRESET failed
Submitter	: Norbert Preining <preining@logic.at>
Date		: 2012-02-02 5:12 (22 days old)
Message-ID	: <20120202051258.GA15550@gamma.logic.tuwien.ac.at>
References	: http://marc.info/?l=linux-kernel&m=132815978615112&w=2


Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42713
Subject		: Regression in skge that started around acb42a3 (so past v3.3-rc1)
Submitter	: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Date		: 2012-01-30 15:58 (25 days old)
Message-ID	: <20120130155816.GA1400@phenom.dumpdata.com>
References	: http://marc.info/?l=linux-kernel&m=132793944220262&w=2


Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42707
Subject		: Hang deconfiguring network interface (in shutdown) on 3.3-rc1
Submitter	: James Bottomley <James.Bottomley@hansenpartnership.com>
Date		: 2012-01-28 19:56 (27 days old)
Message-ID	: <1327780565.2924.24.camel@dabdike.int.hansenpartnership.com>
References	: http://marc.info/?l=linux-kernel&m=132778076214873&w=2


Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42686
Subject		: iwlagn is getting even worse with 3.3-rc1
Submitter	: Norbert Preining <preining@logic.at>
Date		: 2012-01-24 1:36 (31 days old)
Message-ID	: <20120124013625.GB18436@gamma.logic.tuwien.ac.at>
References	: http://marc.info/?l=linux-kernel&m=132736918325378&w=2


Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42683
Subject		: WARN... Device 'cpu1' does not have a release() function, it is broken and must be fixed. when doing 'xl vcpu-set <guest_id> 1'
Submitter	: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Date		: 2012-01-23 18:06 (32 days old)
Message-ID	: <20120123180601.GA24553@phenom.dumpdata.com>
References	: http://marc.info/?l=linux-kernel&m=132734233916154&w=2


Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42678
Subject		: [3.3-rc1] radeon stuck in kernel after lockup
Submitter	: Torsten Kaiser <just.for.lkml@googlemail.com>
Date		: 2012-01-21 19:03 (34 days old)
Message-ID	: <CAPVoSvSXMvRb=1itu9DjF+s=6zfAChvUxS-x=b8EV9kOZinNpA@mail.gmail.com>
References	: http://marc.info/?l=linux-kernel&m=132717279606670&w=2
Handled-By	: Jérôme Glisse <glisse@freedesktop.org>


Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42669
Subject		: 3.3-rc1: compiling problems  nvme , l2cap_sock , mc13892-regulator , and snd-pcsp don't work
Submitter	: werner <w.landgraf@ru.ru>
Date		: 2012-01-20 18:52 (35 days old)
Message-ID	: <web-682421181@zbackend1.aha.ru>
References	: http://marc.info/?l=linux-kernel&m=132708923719565&w=2


For details, please visit the bug entries and follow the links given in
references.

As you can see, there is a Bugzilla entry for each of the listed regressions.
There also is a Bugzilla entry used for tracking the regressions from 3.2,
unresolved as well as resolved, at:

http://bugzilla.kernel.org/show_bug.cgi?id=42644

Please let the tracking team know if there are any Bugzilla entries that
should be added to the list in there.

Thanks!

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

Re: 3.3-rc4+: Reported regressions from 3.2

[David Miller]

From: "Rafael J. Wysocki" <rjw@sisk.pl>
Date: Thu, 23 Feb 2012 23:51:20 +0100 (CET)

> Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42776
> Subject		: OF-related boot crash in 3.3.0-rc3-00188-g3ec1e88
> Submitter	: Meelis Roos <mroos@linux.ee>
> Date		: 2012-02-13 7:45 (11 days old)
> Message-ID	: <alpine.SOC.1.00.1202130942030.1488@math.ut.ee>
> References	: http://marc.info/?l=linux-kernel&m=132911916331615&w=2

Potential workaround/fix at:

http://marc.info/?l=linux-kernel&m=132993294018762&w=2

Re: 3.3-rc4+: Reported regressions from 3.2

[Rafael J. Wysocki]

On Friday, February 24, 2012, David Miller wrote:
> From: "Rafael J. Wysocki" <rjw@sisk.pl>
> Date: Thu, 23 Feb 2012 23:51:20 +0100 (CET)
> 
> > Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42776
> > Subject		: OF-related boot crash in 3.3.0-rc3-00188-g3ec1e88
> > Submitter	: Meelis Roos <mroos@linux.ee>
> > Date		: 2012-02-13 7:45 (11 days old)
> > Message-ID	: <alpine.SOC.1.00.1202130942030.1488@math.ut.ee>
> > References	: http://marc.info/?l=linux-kernel&m=132911916331615&w=2
> 
> Potential workaround/fix at:
> 
> http://marc.info/?l=linux-kernel&m=132993294018762&w=2

Thanks, updated.

Rafael

Re: [Bug #42707] Hang deconfiguring network interface (in shutdown) on 3.3-rc1

[James Bottomley]

On Thu, 2012-02-23 at 23:55 +0100, Rafael J. Wysocki wrote:
> This message has been generated automatically as a part of a summary report
> of recent regressions.
> 
> The following bug entry is on the current list of known regressions
> from 3.2.  Please verify if it still should be listed and let the tracking team
> know (either way).
> 
> 
> Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42707
> Subject		: Hang deconfiguring network interface (in shutdown) on 3.3-rc1
> Submitter	: James Bottomley <James.Bottomley@hansenpartnership.com>
> Date		: 2012-01-28 19:56 (27 days old)
> Message-ID	: <1327780565.2924.24.camel@dabdike.int.hansenpartnership.com>
> References	: http://marc.info/?l=linux-kernel&m=132778076214873&w=2

Still present in 3.3-rc4; I've bisected it back to this commit:

commit 92feeabf3f673767c6ee4cfc7fc224098446c1c1
Author: Matt Carlson <mcarlson@broadcom.com>
Date:   Thu Dec 8 14:40:14 2011 +0000

    tg3: Save stats across chip resets

and sure enough, just reverting this single commit on 3.3-rc4 fixes the
problem.

James

Re: [Bug #42707] Hang deconfiguring network interface (in shutdown) on 3.3-rc1

[Matt Carlson]

On Mon, Feb 27, 2012 at 05:44:34PM -0600, James Bottomley wrote:
> On Thu, 2012-02-23 at 23:55 +0100, Rafael J. Wysocki wrote:
> > This message has been generated automatically as a part of a summary report
> > of recent regressions.
> > 
> > The following bug entry is on the current list of known regressions
> > from 3.2.  Please verify if it still should be listed and let the tracking team
> > know (either way).
> > 
> > 
> > Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42707
> > Subject		: Hang deconfiguring network interface (in shutdown) on 3.3-rc1
> > Submitter	: James Bottomley <James.Bottomley@hansenpartnership.com>
> > Date		: 2012-01-28 19:56 (27 days old)
> > Message-ID	: <1327780565.2924.24.camel@dabdike.int.hansenpartnership.com>
> > References	: http://marc.info/?l=linux-kernel&m=132778076214873&w=2
> 
> Still present in 3.3-rc4; I've bisected it back to this commit:
> 
> commit 92feeabf3f673767c6ee4cfc7fc224098446c1c1
> Author: Matt Carlson <mcarlson@broadcom.com>
> Date:   Thu Dec 8 14:40:14 2011 +0000
> 
>     tg3: Save stats across chip resets
> 
> and sure enough, just reverting this single commit on 3.3-rc4 fixes the
> problem.
> 
> James

I don't see anything incorrect about the patch.  I'm guessing the patch
just changes the timing somehow.

tg3_reset_chip() does not take any driver-internal spinlocks.  That
happens at tg3_close().  I'm guessing the spinlock in the trace is coming
from synchronize_irq().

Re: [Bug #42707] Hang deconfiguring network interface (in shutdown) on 3.3-rc1

[David Miller]

From: "Matt Carlson" <mcarlson-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
Date: Mon, 27 Feb 2012 17:24:08 -0800

> On Mon, Feb 27, 2012 at 05:44:34PM -0600, James Bottomley wrote:
>> On Thu, 2012-02-23 at 23:55 +0100, Rafael J. Wysocki wrote:
>> > This message has been generated automatically as a part of a summary report
>> > of recent regressions.
>> > 
>> > The following bug entry is on the current list of known regressions
>> > from 3.2.  Please verify if it still should be listed and let the tracking team
>> > know (either way).
>> > 
>> > 
>> > Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42707
>> > Subject		: Hang deconfiguring network interface (in shutdown) on 3.3-rc1
>> > Submitter	: James Bottomley <James.Bottomley-JuX6DAaQMKPCXq6kfMZ53/egYHeGw8Jk@public.gmane.org>
>> > Date		: 2012-01-28 19:56 (27 days old)
>> > Message-ID	: <1327780565.2924.24.camel-sFMDBYUN5F8GjUHQrlYNx2Wm91YjaHnnhRte9Li2A+AAvxtiuMwx3w@public.gmane.org>
>> > References	: http://marc.info/?l=linux-kernel&m=132778076214873&w=2
>> 
>> Still present in 3.3-rc4; I've bisected it back to this commit:
>> 
>> commit 92feeabf3f673767c6ee4cfc7fc224098446c1c1
>> Author: Matt Carlson <mcarlson-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
>> Date:   Thu Dec 8 14:40:14 2011 +0000
>> 
>>     tg3: Save stats across chip resets
>> 
>> and sure enough, just reverting this single commit on 3.3-rc4 fixes the
>> problem.
>> 
>> James
> 
> I don't see anything incorrect about the patch.  I'm guessing the patch
> just changes the timing somehow.
> 
> tg3_reset_chip() does not take any driver-internal spinlocks.  That
> happens at tg3_close().  I'm guessing the spinlock in the trace is coming
> from synchronize_irq().

If you can't figure it out, we need to revert, so send me a revert patch
as soon as possible.

Re: [Bug #42707] Hang deconfiguring network interface (in shutdown) on 3.3-rc1

[Matt Carlson]

On Mon, Feb 27, 2012 at 05:44:34PM -0600, James Bottomley wrote:
> On Thu, 2012-02-23 at 23:55 +0100, Rafael J. Wysocki wrote:
> > This message has been generated automatically as a part of a summary report
> > of recent regressions.
> > 
> > The following bug entry is on the current list of known regressions
> > from 3.2.  Please verify if it still should be listed and let the tracking team
> > know (either way).
> > 
> > 
> > Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42707
> > Subject		: Hang deconfiguring network interface (in shutdown) on 3.3-rc1
> > Submitter	: James Bottomley <James.Bottomley-JuX6DAaQMKPCXq6kfMZ53/egYHeGw8Jk@public.gmane.org>
> > Date		: 2012-01-28 19:56 (27 days old)
> > Message-ID	: <1327780565.2924.24.camel-sFMDBYUN5F8GjUHQrlYNx2Wm91YjaHnnhRte9Li2A+AAvxtiuMwx3w@public.gmane.org>
> > References	: http://marc.info/?l=linux-kernel&m=132778076214873&w=2
> 
> Still present in 3.3-rc4; I've bisected it back to this commit:
> 
> commit 92feeabf3f673767c6ee4cfc7fc224098446c1c1
> Author: Matt Carlson <mcarlson-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
> Date:   Thu Dec 8 14:40:14 2011 +0000
> 
>     tg3: Save stats across chip resets
> 
> and sure enough, just reverting this single commit on 3.3-rc4 fixes the
> problem.
> 
> James

Are you dealing with a bcm5700 or bcm5701 device?  If so, can you try
the following patch?

Subject: [PATCH 1/1] tg3: Fix tg3_get_stats64 for 5700 / 5701 devs

tg3_get_stats64() takes tp->lock when dealing with non-serdes bcm5700
and bcm5701 devices.  However, functions that call tg3_halt() have
already acquired tp->lock.  When tg3_get_stats64() is called in
tg3_halt(), deadlock will occur.

This patch fixes the problem by separating the stat gathering code into
a new tg3_get_nstats() function.  tg3_get_stats64() is recoded to call
this function and take tp->lock.  The code that takes tp->lock in
tg3_calc_crc_errors() has been removed.  Function signatures have been
cleaned up too.

Signed-off-by: Matt Carlson <mcarlson-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
---
 drivers/net/ethernet/broadcom/tg3.c |   43 ++++++++++++++++++-----------------
 1 files changed, 22 insertions(+), 21 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/tg3.c b/drivers/net/ethernet/broadcom/tg3.c
index 97dcccd..76f33d5 100644
--- a/drivers/net/ethernet/broadcom/tg3.c
+++ b/drivers/net/ethernet/broadcom/tg3.c
@@ -8001,10 +8001,8 @@ static int tg3_chip_reset(struct tg3 *tp)
 	return 0;
 }
 
-static struct rtnl_link_stats64 *tg3_get_stats64(struct net_device *,
-						 struct rtnl_link_stats64 *);
-static struct tg3_ethtool_stats *tg3_get_estats(struct tg3 *,
-						struct tg3_ethtool_stats *);
+static void tg3_get_nstats(struct tg3 *, struct rtnl_link_stats64 *);
+static void tg3_get_estats(struct tg3 *, struct tg3_ethtool_stats *);
 
 /* tp->lock is held. */
 static int tg3_halt(struct tg3 *tp, int kind, int silent)
@@ -8025,7 +8023,7 @@ static int tg3_halt(struct tg3 *tp, int kind, int silent)
 
 	if (tp->hw_stats) {
 		/* Save the stats across chip resets... */
-		tg3_get_stats64(tp->dev, &tp->net_stats_prev),
+		tg3_get_nstats(tp, &tp->net_stats_prev);
 		tg3_get_estats(tp, &tp->estats_prev);
 
 		/* And make sure the next sample is new data */
@@ -10125,7 +10123,7 @@ static inline u64 get_stat64(tg3_stat64_t *val)
        return ((u64)val->high << 32) | ((u64)val->low);
 }
 
-static u64 calc_crc_errors(struct tg3 *tp)
+static u64 tg3_calc_crc_errors(struct tg3 *tp)
 {
 	struct tg3_hw_stats *hw_stats = tp->hw_stats;
 
@@ -10134,14 +10132,12 @@ static u64 calc_crc_errors(struct tg3 *tp)
 	     GET_ASIC_REV(tp->pci_chip_rev_id) == ASIC_REV_5701)) {
 		u32 val;
 
-		spin_lock_bh(&tp->lock);
 		if (!tg3_readphy(tp, MII_TG3_TEST1, &val)) {
 			tg3_writephy(tp, MII_TG3_TEST1,
 				     val | MII_TG3_TEST1_CRC_EN);
 			tg3_readphy(tp, MII_TG3_RXR_COUNTERS, &val);
 		} else
 			val = 0;
-		spin_unlock_bh(&tp->lock);
 
 		tp->phy_crc_errors += val;
 
@@ -10155,8 +10151,7 @@ static u64 calc_crc_errors(struct tg3 *tp)
 	estats->member =	old_estats->member + \
 				get_stat64(&hw_stats->member)
 
-static struct tg3_ethtool_stats *tg3_get_estats(struct tg3 *tp,
-					       struct tg3_ethtool_stats *estats)
+static void tg3_get_estats(struct tg3 *tp, struct tg3_ethtool_stats *estats)
 {
 	struct tg3_ethtool_stats *old_estats = &tp->estats_prev;
 	struct tg3_hw_stats *hw_stats = tp->hw_stats;
@@ -10238,20 +10233,13 @@ static struct tg3_ethtool_stats *tg3_get_estats(struct tg3 *tp,
 	ESTAT_ADD(nic_tx_threshold_hit);
 
 	ESTAT_ADD(mbuf_lwm_thresh_hit);
-
-	return estats;
 }
 
-static struct rtnl_link_stats64 *tg3_get_stats64(struct net_device *dev,
-						 struct rtnl_link_stats64 *stats)
+static void tg3_get_nstats(struct tg3 *tp, struct rtnl_link_stats64 *stats)
 {
-	struct tg3 *tp = netdev_priv(dev);
 	struct rtnl_link_stats64 *old_stats = &tp->net_stats_prev;
 	struct tg3_hw_stats *hw_stats = tp->hw_stats;
 
-	if (!hw_stats)
-		return old_stats;
-
 	stats->rx_packets = old_stats->rx_packets +
 		get_stat64(&hw_stats->rx_ucast_packets) +
 		get_stat64(&hw_stats->rx_mcast_packets) +
@@ -10294,15 +10282,13 @@ static struct rtnl_link_stats64 *tg3_get_stats64(struct net_device *dev,
 		get_stat64(&hw_stats->tx_carrier_sense_errors);
 
 	stats->rx_crc_errors = old_stats->rx_crc_errors +
-		calc_crc_errors(tp);
+		tg3_calc_crc_errors(tp);
 
 	stats->rx_missed_errors = old_stats->rx_missed_errors +
 		get_stat64(&hw_stats->rx_discards);
 
 	stats->rx_dropped = tp->rx_dropped;
 	stats->tx_dropped = tp->tx_dropped;
-
-	return stats;
 }
 
 static int tg3_get_regs_len(struct net_device *dev)
@@ -12213,6 +12199,21 @@ static const struct ethtool_ops tg3_ethtool_ops = {
 	.set_rxfh_indir		= tg3_set_rxfh_indir,
 };
 
+static struct rtnl_link_stats64 *tg3_get_stats64(struct net_device *dev,
+						struct rtnl_link_stats64 *stats)
+{
+	struct tg3 *tp = netdev_priv(dev);
+
+	if (!tp->hw_stats)
+		return &tp->net_stats_prev;
+
+	spin_lock_bh(&tp->lock);
+	tg3_get_nstats(tp, stats);
+	spin_unlock_bh(&tp->lock);
+
+	return stats;
+}
+
 static void tg3_set_rx_mode(struct net_device *dev)
 {
 	struct tg3 *tp = netdev_priv(dev);
-- 
1.7.3.4

Re: [Bug #42707] Hang deconfiguring network interface (in shutdown) on 3.3-rc1

[James Bottomley]

On Tue, 2012-02-28 at 15:32 -0800, Matt Carlson wrote:
> On Mon, Feb 27, 2012 at 05:44:34PM -0600, James Bottomley wrote:
> > On Thu, 2012-02-23 at 23:55 +0100, Rafael J. Wysocki wrote:
> > > This message has been generated automatically as a part of a summary report
> > > of recent regressions.
> > > 
> > > The following bug entry is on the current list of known regressions
> > > from 3.2.  Please verify if it still should be listed and let the tracking team
> > > know (either way).
> > > 
> > > 
> > > Bug-Entry	: http://bugzilla.kernel.org/show_bug.cgi?id=42707
> > > Subject		: Hang deconfiguring network interface (in shutdown) on 3.3-rc1
> > > Submitter	: James Bottomley <James.Bottomley-JuX6DAaQMKPCXq6kfMZ53/egYHeGw8Jk@public.gmane.org>
> > > Date		: 2012-01-28 19:56 (27 days old)
> > > Message-ID	: <1327780565.2924.24.camel-sFMDBYUN5F8GjUHQrlYNx2Wm91YjaHnnhRte9Li2A+AAvxtiuMwx3w@public.gmane.org>
> > > References	: http://marc.info/?l=linux-kernel&m=132778076214873&w=2
> > 
> > Still present in 3.3-rc4; I've bisected it back to this commit:
> > 
> > commit 92feeabf3f673767c6ee4cfc7fc224098446c1c1
> > Author: Matt Carlson <mcarlson-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
> > Date:   Thu Dec 8 14:40:14 2011 +0000
> > 
> >     tg3: Save stats across chip resets
> > 
> > and sure enough, just reverting this single commit on 3.3-rc4 fixes the
> > problem.
> > 
> > James
> 
> Are you dealing with a bcm5700 or bcm5701 device?

Yes, this is it:

20:02.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5701 Gigabit Ethernet (rev 15)
	Subsystem: Hewlett-Packard Company Core Lan 1000Base-T
	Flags: bus master, 66MHz, medium devsel, latency 128, IRQ 72
	Memory at ffffffff90000000 (64-bit, non-prefetchable) [size=64K]
	Capabilities: [40] PCI-X non-bridge device
	Capabilities: [48] Power Management version 2
	Capabilities: [50] Vital Product Data
	Capabilities: [58] MSI: Enable- Count=1/8 Maskable- 64bit+
	Kernel driver in use: tg3

>   If so, can you try
> the following patch?
> 
> Subject: [PATCH 1/1] tg3: Fix tg3_get_stats64 for 5700 / 5701 devs

Well, the patch needs some attention:

  CC [M]  drivers/net/ethernet/broadcom/tg3.o
drivers/net/ethernet/broadcom/tg3.c: In function 'tg3_get_estats':
drivers/net/ethernet/broadcom/tg3.c:9882: warning: 'return' with a value, in function returning void

It also didn't apply incredibly well (the fuzz factors and line offsets
are a lot higher than they should be):

patching file drivers/net/ethernet/broadcom/tg3.c
Hunk #1 succeeded at 7886 (offset -115 lines).
Hunk #2 succeeded at 7908 (offset -115 lines).
Hunk #3 succeeded at 9845 (offset -278 lines).
Hunk #4 succeeded at 9854 (offset -278 lines).
Hunk #5 succeeded at 9873 (offset -278 lines).
Hunk #6 succeeded at 9958 (offset -275 lines).
Hunk #7 succeeded at 10007 with fuzz 1 (offset -275 lines).
Hunk #8 succeeded at 10103 with fuzz 2 (offset -2096 lines).

but it seems to work.

Thanks,

James

Re: [Bug #42707] Hang deconfiguring network interface (in shutdown) on 3.3-rc1

[Michael Chan]

On Tue, 2012-02-28 at 18:58 -0600, James Bottomley wrote:
> Well, the patch needs some attention:
> 
>   CC [M]  drivers/net/ethernet/broadcom/tg3.o
> drivers/net/ethernet/broadcom/tg3.c: In function 'tg3_get_estats':
> drivers/net/ethernet/broadcom/tg3.c:9882: warning: 'return' with a
> value, in function returning void
> 
> It also didn't apply incredibly well (the fuzz factors and line
> offsets
> are a lot higher than they should be):
> 
> patching file drivers/net/ethernet/broadcom/tg3.c
> Hunk #1 succeeded at 7886 (offset -115 lines).
> Hunk #2 succeeded at 7908 (offset -115 lines).
> Hunk #3 succeeded at 9845 (offset -278 lines).
> Hunk #4 succeeded at 9854 (offset -278 lines).
> Hunk #5 succeeded at 9873 (offset -278 lines).
> Hunk #6 succeeded at 9958 (offset -275 lines).
> Hunk #7 succeeded at 10007 with fuzz 1 (offset -275 lines).
> Hunk #8 succeeded at 10103 with fuzz 2 (offset -2096 lines).
> 
> but it seems to work.

I think Matt did the patch for the net-next tree.  Here's the same patch
for the net tree which is the correct tree for this patch.  Thanks.


Subject: [PATCH net] tg3: Fix tg3_get_stats64 for 5700 / 5701 devs

From: Matt Carlson <mcarlson@broadcom.com>

tg3_get_stats64() takes tp->lock when dealing with non-serdes bcm5700
and bcm5701 devices.  However, functions that call tg3_halt() have
already acquired tp->lock.  When tg3_get_stats64() is called in
tg3_halt(), deadlock will occur.

This patch fixes the problem by separating the stat gathering code into
a new tg3_get_nstats() function.  tg3_get_stats64() is recoded to call
this function and take tp->lock.  The code that takes tp->lock in
tg3_calc_crc_errors() has been removed.  Function signatures have been
cleaned up too.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
---
 drivers/net/ethernet/broadcom/tg3.c |   45 ++++++++++++++++++-----------------
 1 files changed, 23 insertions(+), 22 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/tg3.c b/drivers/net/ethernet/broadcom/tg3.c
index a1f2e0f..423d023 100644
--- a/drivers/net/ethernet/broadcom/tg3.c
+++ b/drivers/net/ethernet/broadcom/tg3.c
@@ -7886,10 +7886,8 @@ static int tg3_chip_reset(struct tg3 *tp)
 	return 0;
 }
 
-static struct rtnl_link_stats64 *tg3_get_stats64(struct net_device *,
-						 struct rtnl_link_stats64 *);
-static struct tg3_ethtool_stats *tg3_get_estats(struct tg3 *,
-						struct tg3_ethtool_stats *);
+static void tg3_get_nstats(struct tg3 *, struct rtnl_link_stats64 *);
+static void tg3_get_estats(struct tg3 *, struct tg3_ethtool_stats *);
 
 /* tp->lock is held. */
 static int tg3_halt(struct tg3 *tp, int kind, int silent)
@@ -7910,7 +7908,7 @@ static int tg3_halt(struct tg3 *tp, int kind, int silent)
 
 	if (tp->hw_stats) {
 		/* Save the stats across chip resets... */
-		tg3_get_stats64(tp->dev, &tp->net_stats_prev),
+		tg3_get_nstats(tp, &tp->net_stats_prev),
 		tg3_get_estats(tp, &tp->estats_prev);
 
 		/* And make sure the next sample is new data */
@@ -9847,7 +9845,7 @@ static inline u64 get_stat64(tg3_stat64_t *val)
        return ((u64)val->high << 32) | ((u64)val->low);
 }
 
-static u64 calc_crc_errors(struct tg3 *tp)
+static u64 tg3_calc_crc_errors(struct tg3 *tp)
 {
 	struct tg3_hw_stats *hw_stats = tp->hw_stats;
 
@@ -9856,14 +9854,12 @@ static u64 calc_crc_errors(struct tg3 *tp)
 	     GET_ASIC_REV(tp->pci_chip_rev_id) == ASIC_REV_5701)) {
 		u32 val;
 
-		spin_lock_bh(&tp->lock);
 		if (!tg3_readphy(tp, MII_TG3_TEST1, &val)) {
 			tg3_writephy(tp, MII_TG3_TEST1,
 				     val | MII_TG3_TEST1_CRC_EN);
 			tg3_readphy(tp, MII_TG3_RXR_COUNTERS, &val);
 		} else
 			val = 0;
-		spin_unlock_bh(&tp->lock);
 
 		tp->phy_crc_errors += val;
 
@@ -9877,14 +9873,13 @@ static u64 calc_crc_errors(struct tg3 *tp)
 	estats->member =	old_estats->member + \
 				get_stat64(&hw_stats->member)
 
-static struct tg3_ethtool_stats *tg3_get_estats(struct tg3 *tp,
-					       struct tg3_ethtool_stats *estats)
+static void tg3_get_estats(struct tg3 *tp, struct tg3_ethtool_stats *estats)
 {
 	struct tg3_ethtool_stats *old_estats = &tp->estats_prev;
 	struct tg3_hw_stats *hw_stats = tp->hw_stats;
 
 	if (!hw_stats)
-		return old_estats;
+		return;
 
 	ESTAT_ADD(rx_octets);
 	ESTAT_ADD(rx_fragments);
@@ -9963,20 +9958,13 @@ static struct tg3_ethtool_stats *tg3_get_estats(struct tg3 *tp,
 	ESTAT_ADD(nic_tx_threshold_hit);
 
 	ESTAT_ADD(mbuf_lwm_thresh_hit);
-
-	return estats;
 }
 
-static struct rtnl_link_stats64 *tg3_get_stats64(struct net_device *dev,
-						 struct rtnl_link_stats64 *stats)
+static void tg3_get_nstats(struct tg3 *tp, struct rtnl_link_stats64 *stats)
 {
-	struct tg3 *tp = netdev_priv(dev);
 	struct rtnl_link_stats64 *old_stats = &tp->net_stats_prev;
 	struct tg3_hw_stats *hw_stats = tp->hw_stats;
 
-	if (!hw_stats)
-		return old_stats;
-
 	stats->rx_packets = old_stats->rx_packets +
 		get_stat64(&hw_stats->rx_ucast_packets) +
 		get_stat64(&hw_stats->rx_mcast_packets) +
@@ -10019,15 +10007,13 @@ static struct rtnl_link_stats64 *tg3_get_stats64(struct net_device *dev,
 		get_stat64(&hw_stats->tx_carrier_sense_errors);
 
 	stats->rx_crc_errors = old_stats->rx_crc_errors +
-		calc_crc_errors(tp);
+		tg3_calc_crc_errors(tp);
 
 	stats->rx_missed_errors = old_stats->rx_missed_errors +
 		get_stat64(&hw_stats->rx_discards);
 
 	stats->rx_dropped = tp->rx_dropped;
 	stats->tx_dropped = tp->tx_dropped;
-
-	return stats;
 }
 
 static inline u32 calc_crc(unsigned char *buf, int len)
@@ -15409,6 +15395,21 @@ static void __devinit tg3_init_coal(struct tg3 *tp)
 	}
 }
 
+static struct rtnl_link_stats64 *tg3_get_stats64(struct net_device *dev,
+						struct rtnl_link_stats64 *stats)
+{
+	struct tg3 *tp = netdev_priv(dev);
+
+	if (!tp->hw_stats)
+		return &tp->net_stats_prev;
+
+	spin_lock_bh(&tp->lock);
+	tg3_get_nstats(tp, stats);
+	spin_unlock_bh(&tp->lock);
+
+	return stats;
+}
+
 static const struct net_device_ops tg3_netdev_ops = {
 	.ndo_open		= tg3_open,
 	.ndo_stop		= tg3_close,
-- 
1.7.1

Re: [Bug #42707] Hang deconfiguring network interface (in shutdown) on 3.3-rc1

[David Miller]

From: "Michael Chan" <mchan-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
Date: Wed, 29 Feb 2012 01:33:37 -0800

> Subject: [PATCH net] tg3: Fix tg3_get_stats64 for 5700 / 5701 devs
> 
> From: Matt Carlson <mcarlson-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
> 
> tg3_get_stats64() takes tp->lock when dealing with non-serdes bcm5700
> and bcm5701 devices.  However, functions that call tg3_halt() have
> already acquired tp->lock.  When tg3_get_stats64() is called in
> tg3_halt(), deadlock will occur.
> 
> This patch fixes the problem by separating the stat gathering code into
> a new tg3_get_nstats() function.  tg3_get_stats64() is recoded to call
> this function and take tp->lock.  The code that takes tp->lock in
> tg3_calc_crc_errors() has been removed.  Function signatures have been
> cleaned up too.
> 
> Signed-off-by: Matt Carlson <mcarlson-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>
> Signed-off-by: Michael Chan <mchan-dY08KVG/lbpWk0Htik3J/w@public.gmane.org>

Applied, thanks everyone.