7 years agodrm/radeon/kms: add some new pci ids
Alex Deucher [Thu, 1 Dec 2011 16:02:11 +0000]
drm/radeon/kms: add some new pci ids

commit 2ed4d9d648cbd4fb1c232a646dbdbdfdd373ca94 upstream.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoiwlwifi: allow pci_enable_msi fail
Wey-Yi Guy [Fri, 28 Oct 2011 00:19:39 +0000]
iwlwifi: allow pci_enable_msi fail

commit 8a39ef8ba0fa0410d71db8e981e887fe4fdeca88 upstream.

Continue the init process even fail to enable msi

out_iounmap is no longer used, remove it

Reported-by: werner <w.landgraf@ru.ru>
Tested-by: werner <w.landgraf@ru.ru>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agobridge: correct IPv6 checksum after pull
stephen hemminger [Tue, 15 Nov 2011 08:09:14 +0000]
bridge: correct IPv6 checksum after pull

commit fa2da8cdae1dd64f78fc915ca1d1a4a93c71e7cb upstream.

Bridge multicast snooping of ICMPv6 would incorrectly report a checksum problem
when used with Ethernet devices like sky2 that use CHECKSUM_COMPLETE.
When bytes are removed from skb, the computed checksum needs to be adjusted.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Tested-by: Martin Volf <martin.volf.42@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agohugetlb: release pages in the error path of hugetlb_cow()
Hillf Danton [Tue, 15 Nov 2011 22:36:12 +0000]
hugetlb: release pages in the error path of hugetlb_cow()

commit ea4039a34c4c206d015d34a49d0b00868e37db1d upstream.

If we fail to prepare an anon_vma, the {new, old}_page should be released,
or they will leak.

Signed-off-by: Hillf Danton <dhillf@gmail.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoSCSI: Silencing 'killing requests for dead queue'
Hannes Reinecke [Wed, 9 Nov 2011 07:39:24 +0000]
SCSI: Silencing 'killing requests for dead queue'

commit 745718132c3c7cac98a622b610e239dcd5217f71 upstream.

When we tear down a device we try to flush all outstanding
commands in scsi_free_queue(). However the check in
scsi_request_fn() is imperfect as it only signals that
we _might start_ aborting commands, not that we've actually
aborted some.
So move the printk inside the scsi_kill_request function,
this will also give us a hint about which commands are aborted.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Cc: Christoph Biedl <linux-kernel.bfrz@manchmal.in-ulm.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoxfs: use doalloc flag in xfs_qm_dqattach_one()
Mitsuo Hayasaka [Thu, 1 Dec 2011 23:27:44 +0000]
xfs: use doalloc flag in xfs_qm_dqattach_one()

commit db3e74b582915d66e10b0c73a62763418f54c340 upstream.

The doalloc arg in xfs_qm_dqattach_one() is a flag that indicates
whether a new area to handle quota information will be allocated
if needed. Originally, it was passed to xfs_qm_dqget(), but has
been removed by the following commit (probably by mistake):

commit 8e9b6e7fa4544ea8a0e030c8987b918509c8ff47
Author: Christoph Hellwig <hch@lst.de>
Date:   Sun Feb 8 21:51:42 2009 +0100

xfs: remove the unused XFS_QMOPT_DQLOCK flag

As the result, xfs_qm_dqget() called from xfs_qm_dqattach_one()
never allocates the new area even if it is needed.

This patch gives the doalloc arg to xfs_qm_dqget() in
xfs_qm_dqattach_one() to fix this problem.

Signed-off-by: Mitsuo Hayasaka <mitsuo.hayasaka.hu@hitachi.com>
Cc: Alex Elder <aelder@sgi.com>
Cc: Christoph Hellwig <hch@infradead.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoxfs: Fix possible memory corruption in xfs_readlink
Carlos Maiolino [Thu, 1 Dec 2011 23:27:43 +0000]
xfs: Fix possible memory corruption in xfs_readlink

commit b52a360b2aa1c59ba9970fb0f52bbb093fcc7a24 upstream.

Fixes a possible memory corruption when the link is larger than
MAXPATHLEN and XFS_DEBUG is not enabled. This also remove the
S_ISLNK assert, since the inode mode is checked previously in
xfs_readlink_by_handle() and via VFS.

Updated to address concerns raised by Ben Hutchings about the loose
attention paid to 32- vs 64-bit values, and the lack of handling a
potentially negative pathlen value:
 - Changed type of "pathlen" to be xfs_fsize_t, to match that of
   ip->i_d.di_size
 - Added checking for a negative pathlen to the too-long pathlen
   test, and generalized the message that gets reported in that case
   to reflect the change
As a result, if a negative pathlen were encountered, this function
would return EFSCORRUPTED (and would fail an assertion for a debug
build)--just as would a too-long pathlen.

Signed-off-by: Alex Elder <aelder@sgi.com>
Signed-off-by: Carlos Maiolino <cmaiolino@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Ben Myers <bpm@sgi.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoxfs: fix buffer flushing during unmount
Christoph Hellwig [Thu, 1 Dec 2011 23:27:42 +0000]
xfs: fix buffer flushing during unmount

commit 87c7bec7fc3377b3873eb3a0f4b603981ea16ebb upstream.

The code to flush buffers in the umount code is a bit iffy: we first
flush all delwri buffers out, but then might be able to queue up a
new one when logging the sb counts.  On a normal shutdown that one
would get flushed out when doing the synchronous superblock write in
xfs_unmountfs_writesb, but we skip that one if the filesystem has
been shut down.

Fix this by moving the delwri list flushing until just before unmounting
the log, and while we're at it also remove the superflous delwri list
and buffer lru flusing for the rt and log device that can never have
cached or delwri buffers.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reported-by: Amit Sahrawat <amit.sahrawat83@gmail.com>
Tested-by: Amit Sahrawat <amit.sahrawat83@gmail.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
Cc: Ben Myers <bpm@sgi.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoxfs: Return -EIO when xfs_vn_getattr() failed
Mitsuo Hayasaka [Thu, 1 Dec 2011 23:27:41 +0000]
xfs: Return -EIO when xfs_vn_getattr() failed

commit ed32201e65e15f3e6955cb84cbb544b08f81e5a5 upstream.

An attribute of inode can be fetched via xfs_vn_getattr() in XFS.
Currently it returns EIO, not negative value, when it failed.  As a
result, the system call returns not negative value even though an
error occured. The stat(2), ls and mv commands cannot handle this
error and do not work correctly.

This patch fixes this bug, and returns -EIO, not EIO when an error
is detected in xfs_vn_getattr().

Signed-off-by: Mitsuo Hayasaka <mitsuo.hayasaka.hu@hitachi.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alex Elder <aelder@sgi.com>
Cc: Ben Myers <bpm@sgi.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoxfs: avoid direct I/O write vs buffered I/O race
Christoph Hellwig [Thu, 1 Dec 2011 23:27:40 +0000]
xfs: avoid direct I/O write vs buffered I/O race

commit c58cb165bd44de8aaee9755a144136ae743be116 upstream.

Currently a buffered reader or writer can add pages to the pagecache
while we are waiting for the iolock in xfs_file_dio_aio_write.  Prevent
this by re-checking mapping->nrpages after we got the iolock, and if
nessecary upgrade the lock to exclusive mode.  To simplify this a bit
only take the ilock inside of xfs_file_aio_write_checks.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>
Cc: Ben Myers <bpm@sgi.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoxfs: don't serialise direct IO reads on page cache checks
Dave Chinner [Thu, 1 Dec 2011 23:27:39 +0000]
xfs: don't serialise direct IO reads on page cache checks

commit 0c38a2512df272b14ef4238b476a2e4f70da1479 upstream.

There is no need to grab the i_mutex of the IO lock in exclusive
mode if we don't need to invalidate the page cache. Taking these
locks on every direct IO effective serialises them as taking the IO
lock in exclusive mode has to wait for all shared holders to drop
the lock. That only happens when IO is complete, so effective it
prevents dispatch of concurrent direct IO reads to the same inode.

Fix this by taking the IO lock shared to check the page cache state,
and only then drop it and take the IO lock exclusively if there is
work to be done. Hence for the normal direct IO case, no exclusive
locking will occur.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Tested-by: Joern Engel <joern@logfs.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alex Elder <aelder@sgi.com>
Cc: Ben Myers <bpm@sgi.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agomac80211: fix race between the AGG SM and the Tx data path
Emmanuel Grumbach [Sun, 27 Nov 2011 13:29:44 +0000]
mac80211: fix race between the AGG SM and the Tx data path

commit 2a1e0fd175dcfd72096ba9291d31e3b1b5342e60 upstream.

When a packet is supposed to sent be as an a-MPDU, mac80211 sets
IEEE80211_TX_CTL_AMPDU to let the driver know. On the other
hand, mac80211 configures the driver for aggregration with the
ampdu_action callback.
There is race between these two mechanisms since the following
scenario can occur when the BA agreement is torn down:

Tx softIRQ   drv configuration
========== =================

check OPERATIONAL bit
Set the TX_CTL_AMPDU bit in the packet

clear OPERATIONAL bit
stop Tx AGG
Pass Tx packet to the driver.

In that case the driver would get a packet with TX_CTL_AMPDU set
although it has already been notified that the BA session has been
torn down.

To fix this, we need to synchronize all the Qdisc activity after we
cleared the OPERATIONAL bit. After that step, all the following
packets will be buffered until the driver reports it is ready to get
new packets for this RA / TID. This buffering allows not to run into
another race that would send packets with TX_CTL_AMPDU unset while
the driver hasn't been requested to tear down the BA session yet.

This race occurs in practice and iwlwifi complains with a WARN_ON
when it happens.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Reviewed-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agomac80211: don't stop a single aggregation session twice
Johannes Berg [Thu, 24 Nov 2011 19:06:14 +0000]
mac80211: don't stop a single aggregation session twice

commit 24f50a9d165745fd0701c6e089d35f58a229ea69 upstream.

Nikolay noticed (by code review) that mac80211 can
attempt to stop an aggregation session while it is
already being stopped. So to fix it, check whether
stop is already being done and bail out if so.

Also move setting the STOPPING state into the lock
so things are properly atomic.

Reported-by: Nikolay Martynov <mar.kolya@gmail.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agocfg80211: fix regulatory NULL dereference
Johannes Berg [Mon, 21 Nov 2011 09:44:00 +0000]
cfg80211: fix regulatory NULL dereference

commit de3584bd62d87b4c250129fbc46ca52c80330add upstream.

By the time userspace returns with a response to
the regulatory domain request, the wiphy causing
the request might have gone away. If this is so,
reject the update but mark the request as having
been processed anyway.

Cc: Luis R. Rodriguez <lrodriguez@qca.qualcomm.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agonl80211: fix MAC address validation
Eliad Peller [Thu, 24 Nov 2011 16:13:56 +0000]
nl80211: fix MAC address validation

commit e007b857e88097c96c45620bf3b04a4e309053d1 upstream.

MAC addresses have a fixed length. The current
policy allows passing < ETH_ALEN bytes, which
might result in reading beyond the buffer.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agort2x00: Fix efuse EEPROM reading on PPC32.
Gertjan van Wingerde [Wed, 16 Nov 2011 22:16:15 +0000]
rt2x00: Fix efuse EEPROM reading on PPC32.

commit 68fa64ef606bcee688fce46d07aa68f175070156 upstream.

Fix __le32 to __le16 conversion of the first word of an 8-word block
of EEPROM read via the efuse method.

Reported-and-tested-by: Ingvar Hagelund <ingvar@redpill-linpro.com>
Signed-off-by: Gertjan van Wingerde <gwingerde@gmail.com>
Acked-by: Helmut Schaa <helmut.schaa@googlemail.com>
Acked-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agop54spi: Fix workqueue deadlock
Michael Büsch [Wed, 16 Nov 2011 22:55:46 +0000]
p54spi: Fix workqueue deadlock

commit 2d1618170eb493d18f66f2ac03775409a6fb97c6 upstream.

priv->work must not be synced while priv->mutex is locked, because
the mutex is taken in the work handler.
Move cancel_work_sync down to after the device shutdown code.
This is safe, because the work handler checks fw_state and bails out
early in case of a race.

Signed-off-by: Michael Buesch <m@bues.ch>
Acked-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agop54spi: Add missing spin_lock_init
Michael Büsch [Wed, 16 Nov 2011 22:48:31 +0000]
p54spi: Add missing spin_lock_init

commit 32d3a3922d617a5a685a5e2d24b20d0e88f192a9 upstream.

The tx_lock is not initialized properly. Add spin_lock_init().

Signed-off-by: Michael Buesch <m@bues.ch>
Acked-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoclocksource: Avoid selecting mult values that might overflow when adjusted
John Stultz [Mon, 31 Oct 2011 21:06:35 +0000]
clocksource: Avoid selecting mult values that might overflow when adjusted

commit d65670a78cdbfae94f20a9e05ec705871d7cdf2b upstream.

For some frequencies, the clocks_calc_mult_shift() function will
unfortunately select mult values very close to 0xffffffff.  This
has the potential to overflow when NTP adjusts the clock, adding
to the mult value.

This patch adds a clocksource.maxadj value, which provides
an approximation of an 11% adjustment(NTP limits adjustments to
500ppm and the tick adjustment is limited to 10%), which could
be made to the clocksource.mult value. This is then used to both
check that the current mult value won't overflow/underflow, as
well as warning us if the timekeeping_adjust() code pushes over
that 11% boundary.

v2: Fix max_adjustment calculation, and improve WARN_ONCE
messages.

v3: Don't warn before maxadj has actually been set

CC: Yong Zhang <yong.zhang0@gmail.com>
CC: David Daney <ddaney.cavm@gmail.com>
CC: Thomas Gleixner <tglx@linutronix.de>
CC: Chen Jie <chenj@lemote.com>
CC: zhangfx <zhangfx@lemote.com>
Reported-by: Chen Jie <chenj@lemote.com>
Reported-by: zhangfx <zhangfx@lemote.com>
Tested-by: Yong Zhang <yong.zhang0@gmail.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agohrtimer: Fix extra wakeups from __remove_hrtimer()
Jeff Ohlstein [Fri, 18 Nov 2011 23:47:10 +0000]
hrtimer: Fix extra wakeups from __remove_hrtimer()

commit 27c9cd7e601632b3794e1c3344d37b86917ffb43 upstream.

__remove_hrtimer() attempts to reprogram the clockevent device when
the timer being removed is the next to expire. However,
__remove_hrtimer() reprograms the clockevent *before* removing the
timer from the timerqueue and thus when hrtimer_force_reprogram()
finds the next timer to expire it finds the timer we're trying to
remove.

This is especially noticeable when the system switches to NOHz mode
and the system tick is removed. The timer tick is removed from the
system but the clockevent is programmed to wakeup in another HZ
anyway.

Silence the extra wakeup by removing the timer from the timerqueue
before calling hrtimer_force_reprogram() so that we actually program
the clockevent for the next timer to expire.

This was broken by 998adc3 "hrtimers: Convert hrtimers to use
timerlist infrastructure".

Signed-off-by: Jeff Ohlstein <johlstei@codeaurora.org>
Link: http://lkml.kernel.org/r/1321660030-8520-1-git-send-email-johlstei@codeaurora.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agotimekeeping: add arch_offset hook to ktime_get functions
Hector Palacios [Mon, 14 Nov 2011 10:15:25 +0000]
timekeeping: add arch_offset hook to ktime_get functions

commit d004e024058a0eaca097513ce62cbcf978913e0a upstream.

ktime_get and ktime_get_ts were calling timekeeping_get_ns()
but later they were not calling arch_gettimeoffset() so architectures
using this mechanism returned 0 ns when calling these functions.

This happened for example when running Busybox's ping which calls
syscall(__NR_clock_gettime, CLOCK_MONOTONIC, ts) which eventually
calls ktime_get. As a result the returned ping travel time was zero.

Signed-off-by: Hector Palacios <hector.palacios@digi.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agocgroup_freezer: fix freezing groups with stopped tasks
Michal Hocko [Tue, 22 Nov 2011 15:44:47 +0000]
cgroup_freezer: fix freezing groups with stopped tasks

commit 884a45d964dd395eda945842afff5e16bcaedf56 upstream.

2d3cbf8b (cgroup_freezer: update_freezer_state() does incorrect state
transitions) removed is_task_frozen_enough and replaced it with a simple
frozen call. This, however, breaks freezing for a group with stopped tasks
because those cannot be frozen and so the group remains in CGROUP_FREEZING
state (update_if_frozen doesn't count stopped tasks) and never reaches
CGROUP_FROZEN.

Let's add is_task_frozen_enough back and use it at the original locations
(update_if_frozen and try_to_freeze_cgroup). Semantically we consider
stopped tasks as frozen enough so we should consider both cases when
testing frozen tasks.

Testcase:
mkdir /dev/freezer
mount -t cgroup -o freezer none /dev/freezer
mkdir /dev/freezer/foo
sleep 1h &
pid=$!
kill -STOP $pid
echo $pid > /dev/freezer/foo/tasks
echo FROZEN > /dev/freezer/foo/freezer.state
while true
do
cat /dev/freezer/foo/freezer.state
[ "`cat /dev/freezer/foo/freezer.state`" = "FROZEN" ] && break
sleep 1
done
echo OK

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Tomasz Buchert <tomasz.buchert@inria.fr>
Cc: Paul Menage <paul@paulmenage.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoregulator: twl: fix twl4030 support for smps regulators
Tero Kristo [Mon, 28 Nov 2011 14:53:19 +0000]
regulator: twl: fix twl4030 support for smps regulators

commit ba305e31e88ea5c2f598ff9fbc5424711a429e30 upstream.

SMPS regulator voltage control differs from the one of the LDO ones.
Current TWL code was using LDO regulator ops for controlling the SMPS
regulators, which fails. This was fixed fixed by adding separate
regulator type which uses correct logic and calculations for the
voltage levels.

Signed-off-by: Tero Kristo <t-kristo@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoregulator: aat2870: Fix the logic of checking if no id is matched in aat2870_get_regu...
Axel Lin [Mon, 28 Nov 2011 06:06:31 +0000]
regulator: aat2870: Fix the logic of checking if no id is matched in aat2870_get_regulator

commit d4d6373c1109b11c8118340be97ae31b8f94d66a upstream.

In current implementation, the pointer ri is not NULL if no id is matched.
Fix it by checking i == ARRAY_SIZE(aat2870_regulators) if no id is matched.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agogenirq: fix regression in irqfixup, irqpoll
Edward Donovan [Mon, 28 Nov 2011 04:07:34 +0000]
genirq: fix regression in irqfixup, irqpoll

commit 52553ddffad76ccf192d4dd9ce88d5818f57f62a upstream.

Commit fa27271bc8d2("genirq: Fixup poll handling") introduced a
regression that broke irqfixup/irqpoll for some hardware configurations.

Amidst reorganizing 'try_one_irq', that patch removed a test that
checked for 'action->handler' returning IRQ_HANDLED, before acting on
the interrupt.  Restoring this test back returns the functionality lost
since 2.6.39.  In the current set of tests, after 'action' is set, it
must precede '!action->next' to take effect.

With this and my previous patch to irq/spurious.c, c75d720fca8a, all
IRQ regressions that I have encountered are fixed.

Signed-off-by: Edward Donovan <edward.donovan@numble.net>
Reported-and-tested-by: Rogério Brito <rbrito@ime.usp.br>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agortlwifi: fix lps_lock deadlock
Stanislaw Gruszka [Mon, 28 Nov 2011 09:33:40 +0000]
rtlwifi: fix lps_lock deadlock

commit e55b32c110b025ce07b40227f620e99700bf8741 upstream.

rtl_lps_leave can be called from interrupt context, so we have to
disable interrupts when taking lps_lock.

Below is full lockdep info about deadlock:

[   93.815269] =================================
[   93.815390] [ INFO: inconsistent lock state ]
[   93.815472] 2.6.41.1-3.offch.fc15.x86_64.debug #1
[   93.815556] ---------------------------------
[   93.815635] inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage.
[   93.815743] swapper/0 [HC0[0]:SC1[1]:HE1:SE0] takes:
[   93.815832]  (&(&rtlpriv->locks.lps_lock)->rlock){+.?...}, at: [<ffffffffa025dad6>] rtl_lps_leave+0x26/0x103 [rtlwifi]
[   93.815947] {SOFTIRQ-ON-W} state was registered at:
[   93.815947]   [<ffffffff8108e10d>] __lock_acquire+0x369/0xd0c
[   93.815947]   [<ffffffff8108efb3>] lock_acquire+0xf3/0x13e
[   93.815947]   [<ffffffff814e981d>] _raw_spin_lock+0x45/0x79
[   93.815947]   [<ffffffffa025de34>] rtl_swlps_rf_awake+0x5a/0x76 [rtlwifi]
[   93.815947]   [<ffffffffa025aec0>] rtl_op_config+0x12a/0x32a [rtlwifi]
[   93.815947]   [<ffffffffa01d614b>] ieee80211_hw_config+0x124/0x129 [mac80211]
[   93.815947]   [<ffffffffa01e0af3>] ieee80211_dynamic_ps_disable_work+0x32/0x47 [mac80211]
[   93.815947]   [<ffffffff81075aa5>] process_one_work+0x205/0x3e7
[   93.815947]   [<ffffffff81076753>] worker_thread+0xda/0x15d
[   93.815947]   [<ffffffff8107a119>] kthread+0xa8/0xb0
[   93.815947]   [<ffffffff814f3184>] kernel_thread_helper+0x4/0x10
[   93.815947] irq event stamp: 547822
[   93.815947] hardirqs last  enabled at (547822): [<ffffffff814ea1a7>] _raw_spin_unlock_irqrestore+0x45/0x61
[   93.815947] hardirqs last disabled at (547821): [<ffffffff814e9987>] _raw_spin_lock_irqsave+0x22/0x8e
[   93.815947] softirqs last  enabled at (547790): [<ffffffff810623ed>] _local_bh_enable+0x13/0x15
[   93.815947] softirqs last disabled at (547791): [<ffffffff814f327c>] call_softirq+0x1c/0x30
[   93.815947]
[   93.815947] other info that might help us debug this:
[   93.815947]  Possible unsafe locking scenario:
[   93.815947]
[   93.815947]        CPU0
[   93.815947]        ----
[   93.815947]   lock(&(&rtlpriv->locks.lps_lock)->rlock);
[   93.815947]   <Interrupt>
[   93.815947]     lock(&(&rtlpriv->locks.lps_lock)->rlock);
[   93.815947]
[   93.815947]  *** DEADLOCK ***
[   93.815947]
[   93.815947] no locks held by swapper/0.
[   93.815947]
[   93.815947] stack backtrace:
[   93.815947] Pid: 0, comm: swapper Not tainted 2.6.41.1-3.offch.fc15.x86_64.debug #1
[   93.815947] Call Trace:
[   93.815947]  <IRQ>  [<ffffffff814dfd00>] print_usage_bug+0x1e7/0x1f8
[   93.815947]  [<ffffffff8101a849>] ? save_stack_trace+0x2c/0x49
[   93.815947]  [<ffffffff8108d55c>] ? print_irq_inversion_bug.part.18+0x1a0/0x1a0
[   93.815947]  [<ffffffff8108dc8a>] mark_lock+0x106/0x220
[   93.815947]  [<ffffffff8108e099>] __lock_acquire+0x2f5/0xd0c
[   93.815947]  [<ffffffff810152af>] ? native_sched_clock+0x34/0x36
[   93.830125]  [<ffffffff810152ba>] ? sched_clock+0x9/0xd
[   93.830125]  [<ffffffff81080181>] ? sched_clock_local+0x12/0x75
[   93.830125]  [<ffffffffa025dad6>] ? rtl_lps_leave+0x26/0x103 [rtlwifi]
[   93.830125]  [<ffffffff8108efb3>] lock_acquire+0xf3/0x13e
[   93.830125]  [<ffffffffa025dad6>] ? rtl_lps_leave+0x26/0x103 [rtlwifi]
[   93.830125]  [<ffffffff814e981d>] _raw_spin_lock+0x45/0x79
[   93.830125]  [<ffffffffa025dad6>] ? rtl_lps_leave+0x26/0x103 [rtlwifi]
[   93.830125]  [<ffffffff81422467>] ? skb_dequeue+0x62/0x6d
[   93.830125]  [<ffffffffa025dad6>] rtl_lps_leave+0x26/0x103 [rtlwifi]
[   93.830125]  [<ffffffffa025f677>] _rtl_pci_ips_leave_tasklet+0xe/0x10 [rtlwifi]
[   93.830125]  [<ffffffff8106281f>] tasklet_action+0x8d/0xee
[   93.830125]  [<ffffffff810629ce>] __do_softirq+0x112/0x25a
[   93.830125]  [<ffffffff814f327c>] call_softirq+0x1c/0x30
[   93.830125]  [<ffffffff81010bf6>] do_softirq+0x4b/0xa1
[   93.830125]  [<ffffffff81062d7d>] irq_exit+0x5d/0xcf
[   93.830125]  [<ffffffff814f3b7e>] do_IRQ+0x8e/0xa5
[   93.830125]  [<ffffffff814ea533>] common_interrupt+0x73/0x73
[   93.830125]  <EOI>  [<ffffffff8108b825>] ? trace_hardirqs_off+0xd/0xf
[   93.830125]  [<ffffffff812bb6d5>] ? intel_idle+0xe5/0x10c
[   93.830125]  [<ffffffff812bb6d1>] ? intel_idle+0xe1/0x10c
[   93.830125]  [<ffffffff813f8d5e>] cpuidle_idle_call+0x11c/0x1fe
[   93.830125]  [<ffffffff8100e2ef>] cpu_idle+0xab/0x101
[   93.830125]  [<ffffffff814c6373>] rest_init+0xd7/0xde
[   93.830125]  [<ffffffff814c629c>] ? csum_partial_copy_generic+0x16c/0x16c
[   93.830125]  [<ffffffff81d4bbb0>] start_kernel+0x3dd/0x3ea
[   93.830125]  [<ffffffff81d4b2c4>] x86_64_start_reservations+0xaf/0xb3
[   93.830125]  [<ffffffff81d4b140>] ? early_idt_handlers+0x140/0x140
[   93.830125]  [<ffffffff81d4b3ca>] x86_64_start_kernel+0x102/0x111

Resolves:
https://bugzilla.redhat.com/show_bug.cgi?id=755154

Reported-by: vjain02@students.poly.edu
Reported-and-tested-by: Oliver Paukstadt <pstadt@sourcentral.org>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoext4: fix racy use-after-free in ext4_end_io_dio()
Tejun Heo [Fri, 25 Nov 2011 00:22:24 +0000]
ext4: fix racy use-after-free in ext4_end_io_dio()

commit 4c81f045c0bd2cbb78cc6446a4cd98038fe11a2e upstream.

ext4_end_io_dio() queues io_end->work and then clears iocb->private;
however, io_end->work calls aio_complete() which frees the iocb
object.  If that slab object gets reallocated, then ext4_end_io_dio()
can end up clearing someone else's iocb->private, this use-after-free
can cause a leak of a struct ext4_io_end_t structure.

Detected and tested with slab poisoning.

[ Note: Can also reproduce using 12 fio's against 12 file systems with the
  following configuration file:

  [global]
  direct=1
  ioengine=libaio
  iodepth=1
  bs=4k
  ba=4k
  size=128m

  [create]
  filename=${TESTDIR}
  rw=write

  -- tytso ]

Google-Bug-Id: 5354697
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Reported-by: Kent Overstreet <koverstreet@google.com>
Tested-by: Kent Overstreet <koverstreet@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoSUNRPC: Ensure we return EAGAIN in xs_nospace if congestion is cleared
Trond Myklebust [Tue, 22 Nov 2011 12:44:28 +0000]
SUNRPC: Ensure we return EAGAIN in xs_nospace if congestion is cleared

commit 24ca9a847791fd53d9b217330b15f3c285827a18 upstream.

By returning '0' instead of 'EAGAIN' when the tests in xs_nospace() fail
to find evidence of socket congestion, we are making the RPC engine believe
that the message was incorrectly sent and so it disconnects the socket
instead of just retrying.

The bug appears to have been introduced by commit
5e3771ce2d6a69e10fcc870cdf226d121d868491 (SUNRPC: Ensure that xs_nospace
return values are propagated).

Reported-by: Andrew Cooper <andrew.cooper3@citrix.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Tested-by: Andrew Cooper <andrew.cooper3@citrix.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoASoC: Ensure WM8731 register cache is synced when resuming from disabled
Mark Brown [Mon, 21 Nov 2011 11:55:41 +0000]
ASoC: Ensure WM8731 register cache is synced when resuming from disabled

commit ed3e80c4c991a52f9fce3421536a78e331ae0949 upstream.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoASoC: wm8753: Skip noop reconfiguration of DAI mode
Timo Juhani Lindfors [Thu, 17 Nov 2011 00:52:50 +0000]
ASoC: wm8753: Skip noop reconfiguration of DAI mode

commit 2391a0e06789a3f1718dee30b282562f7ed28c87 upstream.

This patch makes it possible to set DAI mode to its currently applied
value even if codec is active. This is necessary to allow

aplay -t raw -r 44100 -f S16_LE -c 2 < /dev/urandom &
alsactl store -f backup.state
alsactl restore -f backup.state

to work without returning errors. This patch is based on a patch sent
by Klaus Kurzmann <mok@fluxnetz.de>.

Signed-off-by: Timo Juhani Lindfors <timo.lindfors@iki.fi>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoASoC: fsl_ssi: properly initialize the sysfs attribute object
Timur Tabi [Mon, 14 Nov 2011 22:35:26 +0000]
ASoC: fsl_ssi: properly initialize the sysfs attribute object

commit 0f768a7235d3dfb6f4833030a95a06419df089cb upstream.

Commit 6992f533 ("sysfs: Use one lockdep class per sysfs attribute")
requires 'struct attribute' objects to be initialized with sysfs_attr_init().

Signed-off-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoASoC: sta32x: preserve coefficient RAM
Johannes Stezenbach [Mon, 14 Nov 2011 16:23:16 +0000]
ASoC: sta32x: preserve coefficient RAM

commit 54dc6cabe684375b3cf549c7b0545613d694aba8 upstream.

The coefficient RAM must be saved in a shadow so it can
be restored when the codec is powered on using
regulator_bulk_enable().

Signed-off-by: Johannes Stezenbach <js@sig21.net>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoALSA: hda/realtek - Fix missing inits of item indices for auto-mic
Takashi Iwai [Wed, 23 Nov 2011 06:38:59 +0000]
ALSA: hda/realtek - Fix missing inits of item indices for auto-mic

commit 6759dc323826c2c806c998cd93945c5476688dd2 upstream.

When the imux entries are rebuilt in alc_rebuild_imux_for_auto_mic(),
the initialization of index field is missing.  It may work without it
casually when the original imux was created by the auto-parser, but
it's definitely broken in the case of static configs where no imux was
parsed beforehand.  Because of this, the auto-mic switching doesn't
work properly on some model options.

This patch adds the missing initialization of index field.

Reported-by: Dmitry Nezhevenko <dion@inhex.net>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoALSA: lx6464es - fix device communication via command bus
Tim Blechmann [Tue, 22 Nov 2011 10:15:45 +0000]
ALSA: lx6464es - fix device communication via command bus

commit a29878553a9a7b4c06f93c7e383527cf014d4ceb upstream.

commit 6175ddf06b6172046a329e3abfd9c901a43efd2e optimized the mem*io
functions that have been used to send commands to the device. these
optimizations somehow corrupted the communication with the lx6464es,
that resulted the device to be unusable with kernels after 2.6.33.

this patch emulates the memcpy_*_io functions via a loop to avoid these
problems.

Signed-off-by: Tim Blechmann <tim@klingt.org>
LKML-Reference: <4ECB5257.4040600@ladisch.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoALSA: hda - Fix jack-detection control of VT1708
Takashi Iwai [Thu, 24 Nov 2011 15:33:09 +0000]
ALSA: hda - Fix jack-detection control of VT1708

commit 187d333edc0a8e1bb507900ce89853ffe3bd2c84 upstream.

VT1708 has no support for unsolicited events per jack-plug, the driver
implements the workq for polling the jack-detection.  The mixer element
"Jack Detect" was supposed to control this behavior on/off, but this
doesn't work properly as is now.  The workq is always started and the
HP automute is always enabled.

This patch fixes the jack-detect control behavior by triggering / stopping
the work appropriately at the state change.  Also the work checks the
internal state to continue scheduling or not.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoARM: 7161/1: errata: no automatic store buffer drain
Will Deacon [Mon, 14 Nov 2011 16:24:58 +0000]
ARM: 7161/1: errata: no automatic store buffer drain

commit 11ed0ba1754841316d4095478944300acf19acc3 upstream.

This patch implements a workaround for PL310 erratum 769419. On
revisions of the PL310 prior to r3p2, the Store Buffer does not
automatically drain. This can cause normal, non-cacheable writes to be
retained when the memory system is idle, leading to suboptimal I/O
performance for drivers using coherent DMA.

This patch adds an optional wmb() call to the cpu_idle loop. On systems
with an outer cache, this causes an explicit flush of the store buffer.

Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Tested-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoARM: OMAP: PM: only register TWL with voltage layer when device is present
Kevin Hilman [Wed, 23 Nov 2011 22:43:01 +0000]
ARM: OMAP: PM: only register TWL with voltage layer when device is present

commit 46232a3622c6e33605906ee6690dfef372925f53 upstream.

Current code registers voltage layer details for TWL PMIC even when a TWL
has not been registered.  Fix this to only register the TWL with voltage
layer when the TWL PMIC is initialized by board-level code.

Signed-off-by: Kevin Hilman <khilman@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoARM: OMAP2: select ARM_AMBA if OMAP3_EMU is defined
Ming Lei [Wed, 23 Nov 2011 22:44:50 +0000]
ARM: OMAP2: select ARM_AMBA if OMAP3_EMU is defined

commit a8a6565c7615cab3608d75af95b5c8a3522cd7c4 upstream.

This patch selects ARM_AMBA if OMAP3_EMU is defined because
OC_ETM depends on ARM_AMBA, so fix the link failure[1].

[1],
arch/arm/kernel/built-in.o: In function `etm_remove':
/home/tom/git/omap/linux-2.6-omap/arch/arm/kernel/etm.c:609: undefined
reference to `amba_release_regions'
arch/arm/kernel/built-in.o: In function `etb_remove':
/home/tom/git/omap/linux-2.6-omap/arch/arm/kernel/etm.c:409: undefined
reference to `amba_release_regions'
arch/arm/kernel/built-in.o: In function `etm_init':
/home/tom/git/omap/linux-2.6-omap/arch/arm/kernel/etm.c:640: undefined
reference to `amba_driver_register'
/home/tom/git/omap/linux-2.6-omap/arch/arm/kernel/etm.c:646: undefined
reference to `amba_driver_register'
/home/tom/git/omap/linux-2.6-omap/arch/arm/kernel/etm.c:648: undefined
reference to `amba_driver_unregister'
arch/arm/kernel/built-in.o: In function `etm_probe':
/home/tom/git/omap/linux-2.6-omap/arch/arm/kernel/etm.c:545: undefined
reference to `amba_request_regions'
/home/tom/git/omap/linux-2.6-omap/arch/arm/kernel/etm.c:595: undefined
reference to `amba_release_regions'
arch/arm/kernel/built-in.o: In function `etb_probe':
/home/tom/git/omap/linux-2.6-omap/arch/arm/kernel/etm.c:347: undefined
reference to `amba_request_regions'
/home/tom/git/omap/linux-2.6-omap/arch/arm/kernel/etm.c:392: undefined
reference to `amba_release_regions'
arch/arm/mach-omap2/built-in.o: In function `emu_init':
/home/tom/git/omap/linux-2.6-omap/arch/arm/mach-omap2/emu.c:62:
undefined reference to `amba_device_register'
/home/tom/git/omap/linux-2.6-omap/arch/arm/mach-omap2/emu.c:63:
undefined reference to `amba_device_register'
make: *** [.tmp_vmlinux1] Error 1
making modules

Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoARM: OMAP: hwmod: Fix the addr space, irq, dma count APIs
sricharan [Wed, 23 Nov 2011 22:35:07 +0000]
ARM: OMAP: hwmod: Fix the addr space, irq, dma count APIs

commit cc1b0765da6078b906772b79ff211b88cc0ae958 upstream.

The address spaces, irqs and dma reqs count APIs return the
number of corresponding entries in a hwmod including a additional
null value or a -1 terminator in the structure introduced recently.
More information here:

212738a4: omap_hwmod: use a terminator record with omap_hwmod_mpu_irqs
  arrays

78183f3f: omap_hwmod: use a null structure record to terminate
  omap_hwmod_addr_space arrays

bc614958: omap_hwmod: use a terminator record with omap_hwmod_dma_info
  arrays

The issue with irqs and dma info was originally reported by Benoit Cousson.

The devices which have multiple hwmods and use device_build_ss are
broken with this, as their resources are populated with a extra null
value, subsequently the probe fails. So fix the API not to include
the array terminator in the count.

Reported-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Santosh Shilimkar <santosh.shilimar@ti.com>
Signed-off-by: sricharan <r.sricharan@ti.com>
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoARM: OMAP: smartreflex: fix IRQ handling bug
Felipe Balbi [Wed, 23 Nov 2011 22:43:37 +0000]
ARM: OMAP: smartreflex: fix IRQ handling bug

commit 5a4f1844c2ba21f804d7729306d9b16eaeb724a8 upstream.

Fix a bug which has been on this driver since
it was added by the original commit 984aa6db
which would never clear IRQSTATUS bits.

Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoarm: mx28: fix bit operation in clock setting
Wolfram Sang [Sat, 10 Sep 2011 10:26:07 +0000]
arm: mx28: fix bit operation in clock setting

commit c2735391fbc68feae10d6d14e60956c8106e725f upstream.

reg | (1 << clk->enable_shift) always evaluates to true. Switch it
to & which makes much more sense. Same fix as 13be9f00 (ARM i.MX28: fix
bit operation) at a different location.

Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Shawn Guo <shawn.guo@freescale.com>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoARM: pxa: fix inconsistent CONFIG_USB_PXA27X
Haojian Zhuang [Wed, 9 Nov 2011 23:13:07 +0000]
ARM: pxa: fix inconsistent CONFIG_USB_PXA27X

commit c0a39151a4055332897cba615623d3de2f3896df upstream.

Since CONFIG_USB_GADGET_PXA27X and other macros are renamed to
CONFIG_USB_PXA27X. Update them in arch/arm/mach-pxa and arch/arm/configs
to keep consistent.

Signed-off-by: Haojian Zhuang <haojian.zhuang@marvell.com>
Acked-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Eric Miao <eric.y.miao@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoviafb: correct sync polarity for OLPC DCON
Daniel Drake [Mon, 21 Nov 2011 15:05:56 +0000]
viafb: correct sync polarity for OLPC DCON

commit a32839696a8eef813a1aff604fbad9a32dff6c95 upstream.

While the OLPC display appears to be able to handle either positive
or negative sync, the Display Controller only recognises positive sync.

This brings viafb (for XO-1.5) in line with lxfb (for XO-1) and
fixes a recent regression where the XO-1.5 DCON could no longer be
frozen. Thanks to Florian Tobias Schandinat for helping identify
the fix.

Test case: from a vt,
echo 1 > /sys/devices/platform/dcon/freeze
should cause the current screen contents to freeze, rather than garbage being
displayed.

Signed-off-by: Daniel Drake <dsd@laptop.org>
Signed-off-by: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agodrm/radeon/kms: fix up gpio i2c mask bits for r4xx
Alex Deucher [Mon, 14 Nov 2011 19:32:01 +0000]
drm/radeon/kms: fix up gpio i2c mask bits for r4xx

commit 6c47e5c23aa2a7c54ad7ac13af4bd56cd9e703bf upstream.

Fixes i2c test failures when i2c_algo_bit.bit_test=1.

The hw doesn't actually require a mask, so just set it
to the default mask bits for r1xx-r4xx radeon ddc.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoPCI hotplug: shpchp: don't blindly claim non-AMD 0x7450 device IDs
Bjorn Helgaas [Tue, 23 Aug 2011 16:16:43 +0000]
PCI hotplug: shpchp: don't blindly claim non-AMD 0x7450 device IDs

commit 4cac2eb158c6da0c761689345c6cc5df788a6292 upstream.

Previously we claimed device ID 0x7450, regardless of the vendor, which is
clearly wrong.  Now we'll claim that device ID only for AMD.

I suspect this was just a typo in the original code, but it's possible this
change will break shpchp on non-7450 AMD bridges.  If so, we'll have to fix
them as we find them.

Reference: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=638863
Reported-by: Ralf Jung <ralfjung-e@gmx.de>
Cc: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agodrm/i915: Fix inconsistent backlight level during disabled
Takashi Iwai [Wed, 16 Nov 2011 09:58:03 +0000]
drm/i915: Fix inconsistent backlight level during disabled

commit 04b38670cf46c096705f24e92a8747d1ab89e53c upstream.

When the brightness property is inquired while the backlight is disabled,
the driver returns a wrong value (zero) because it probes the value after
the backlight was turned off.  This caused a black screen even after the
backlight is enabled again.  It should return the internal backlight_level
instead, so that it won't be influenced by the backlight-enable state.

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=41926
BugLink: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/872652

Tested-by: Kamal Mostafa <kamal@canonical.com>
Cc: Alex Davis <alex14641@yahoo.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agodrm/i915: Turn on another required clock gating bit on gen6.
Eric Anholt [Tue, 8 Nov 2011 00:07:05 +0000]
drm/i915: Turn on another required clock gating bit on gen6.

commit 9ca1d10d748e56964de95e3ed80211b192f56cf4 upstream.

Unlike the previous one, I don't have known testcases it fixes.  I'd
rather not go through the same debug cycle on whatever testcases those
might be.

Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agodrm/i915: Turn on a required 3D clock gating bit on Sandybridge.
Eric Anholt [Tue, 8 Nov 2011 00:07:04 +0000]
drm/i915: Turn on a required 3D clock gating bit on Sandybridge.

commit 406478dc911e16677fbd9c84d1d50cdffbc031ab upstream.

Fixes rendering failures in Unigine Tropics and Sanctuary and the mesa
"fire" demo.

Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agodrm/i915: Ivybridge still has fences!
Daniel Vetter [Sun, 9 Oct 2011 19:52:01 +0000]
drm/i915: Ivybridge still has fences!

commit 775d17b6ca4357048f36c22151335addfe15db4b upstream.

So don't forget to restore them on resume and dump them into
the error state.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agodrm/radeon/kms: fix up gpio i2c mask bits for r4xx for real
Alex Deucher [Mon, 21 Nov 2011 17:10:14 +0000]
drm/radeon/kms: fix up gpio i2c mask bits for r4xx for real

commit d724502a9d7a46f4a56a1663b1f50d2dc9d1ef40 upstream.

Fixes i2c test failures when i2c_algo_bit.bit_test=1.

The hw doesn't actually require a mask, so just set it
to the default mask bits for r1xx-r4xx radeon ddc.

I missed this part the first time through.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agodrm: integer overflow in drm_mode_dirtyfb_ioctl()
Xi Wang [Wed, 23 Nov 2011 06:12:01 +0000]
drm: integer overflow in drm_mode_dirtyfb_ioctl()

commit a5cd335165e31db9dbab636fd29895d41da55dd2 upstream.

There is a potential integer overflow in drm_mode_dirtyfb_ioctl()
if userspace passes in a large num_clips.  The call to kmalloc would
allocate a small buffer, and the call to fb->funcs->dirty may result
in a memory corruption.

Reported-by: Haogang Chen <haogangchen@gmail.com>
Signed-off-by: Xi Wang <xi.wang@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agocrypto: mv_cesa - fix hashing of chunks > 1920 bytes
Phil Sutter [Wed, 16 Nov 2011 17:28:01 +0000]
crypto: mv_cesa - fix hashing of chunks > 1920 bytes

commit 274252862f386b7868f35bf5ceaa5391a8ccfdf3 upstream.

This was broken by commit 7759995c75ae0cbd4c861582908449f6b6208e7a (yes,
myself). The basic problem here is since the digest state is only saved
after the last chunk, the state array is only valid when handling the
first chunk of the next buffer. Broken since linux-3.0.

Signed-off-by: Phil Sutter <phil.sutter@viprinet.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoeCryptfs: Extend array bounds for all filename chars
Tyler Hicks [Wed, 23 Nov 2011 17:31:24 +0000]
eCryptfs: Extend array bounds for all filename chars

commit 0f751e641a71157aa584c2a2e22fda52b52b8a56 upstream.

From mhalcrow's original commit message:

    Characters with ASCII values greater than the size of
    filename_rev_map[] are valid filename characters.
    ecryptfs_decode_from_filename() will access kernel memory beyond
    that array, and ecryptfs_parse_tag_70_packet() will then decrypt
    those characters. The attacker, using the FNEK of the crafted file,
    can then re-encrypt the characters to reveal the kernel memory past
    the end of the filename_rev_map[] array. I expect low security
    impact since this array is statically allocated in the text area,
    and the amount of memory past the array that is accessible is
    limited by the largest possible ASCII filename character.

This patch solves the issue reported by mhalcrow but with an
implementation suggested by Linus to simply extend the length of
filename_rev_map[] to 256. Characters greater than 0x7A are mapped to
0x00, which is how invalid characters less than 0x7A were previously
being handled.

Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Reported-by: Michael Halcrow <mhalcrow@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoi2c-algo-bit: Generate correct i2c address sequence for 10-bit target
Jeffrey (Sheng-Hui) Chu [Wed, 23 Nov 2011 10:33:07 +0000]
i2c-algo-bit: Generate correct i2c address sequence for 10-bit target

commit cc6bcf7d2ec2234e7b41770185e4dc826390185e upstream.

The wrong bits were put on the wire, fix that.

This fixes kernel bug #42562.

Signed-off-by: Sheng-Hui J. Chu <jeffchu@broadcom.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoeCryptfs: Flush file in vma close
Tyler Hicks [Mon, 21 Nov 2011 23:31:29 +0000]
eCryptfs: Flush file in vma close

commit 32001d6fe9ac6b0423e674a3093aa56740849f3b upstream.

Dirty pages weren't being written back when an mmap'ed eCryptfs file was
closed before the mapping was unmapped. Since f_ops->flush() is not
called by the munmap() path, the lower file was simply being released.
This patch flushes the eCryptfs file in the vm_ops->close() path.

https://launchpad.net/bugs/870326

Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoeCryptfs: Prevent file create race condition
Tyler Hicks [Mon, 21 Nov 2011 23:31:02 +0000]
eCryptfs: Prevent file create race condition

commit b59db43ad4434519feb338eacb01d77eb50825c5 upstream.

The file creation path prematurely called d_instantiate() and
unlock_new_inode() before the eCryptfs inode info was fully
allocated and initialized and before the eCryptfs metadata was written
to the lower file.

This could result in race conditions in subsequent file and inode
operations leading to unexpected error conditions or a null pointer
dereference while attempting to use the unallocated memory.

https://launchpad.net/bugs/813146

Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoLinux 3.1.4
Greg Kroah-Hartman [Mon, 28 Nov 2011 22:48:14 +0000]
Linux 3.1.4

7 years agoRevert "USB: EHCI: fix HUB TT scheduling issue with iso transfer"
Greg Kroah-Hartman [Mon, 28 Nov 2011 22:40:10 +0000]
Revert "USB: EHCI: fix HUB TT scheduling issue with iso transfer"

This reverts commit f0cc710a6dec5b808a6f13f1f8853c094fce5f12.

Cc: Matthieu Castet <matthieu.castet@parrot.com>
Cc: Thomas Poussevin <thomas.poussevin@parrot.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoLinux 3.1.3
Greg Kroah-Hartman [Sat, 26 Nov 2011 17:08:57 +0000]
Linux 3.1.3

7 years agovmscan: fix shrinker callback bug in fs/super.c
Mikulas Patocka [Tue, 1 Nov 2011 00:08:57 +0000]
vmscan: fix shrinker callback bug in fs/super.c

commit 09f363c7363eb10cfb4b82094bd7064e5608258b upstream.

The callback must not return -1 when nr_to_scan is zero. Fix the bug in
fs/super.c and add this requirement to the callback specification.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: Dave Chinner <david@fromorbit.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agobatman-adv: unify hash_entry field position in tt_local/global_entry
Antonio Quartulli [Fri, 21 Oct 2011 22:55:39 +0000]
batman-adv: unify hash_entry field position in tt_local/global_entry

commit 93840ac40bb0d0f177ef8af74e64671be67e8c37 upstream.

Function tt_response_fill_table() actually uses a tt_local_entry pointer to
iterate either over the local or the global table entries (it depends on the
what hash table is passed as argument). To iterate over such entries the
hlist_for_each_entry_rcu() macro has to access their "hash_entry" field which
MUST be at the same position in both the tt_global/local_entry structures.

Reported-by: Simon Wunderlich <siwu@hrz.tu-chemnitz.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agobatman-adv: add sanity check when removing global tts
Simon Wunderlich [Wed, 19 Oct 2011 08:28:26 +0000]
batman-adv: add sanity check when removing global tts

commit 6e8014947d6469df1566e9e253805557c5c0e4e0 upstream.

After removing the batman-adv module, the hash may be already gone
when tt_global_del_orig() tries to clean the hash. This patch adds
a sanity check to avoid this.

Signed-off-by: Simon Wunderlich <siwu@hrz.tu-chemnitz.de>
Tested-by: Alexey Fisher <bug-track@fisher-privat.net>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agobatman-adv: remove references for global tt entries
Simon Wunderlich [Wed, 19 Oct 2011 09:02:25 +0000]
batman-adv: remove references for global tt entries

commit 531027fcddbcf81c9937dd04f08a7e8f11fd47d2 upstream.

struct tt_global_entry holds a reference to an orig_node which must be
decremented before deallocating the structure.

Signed-off-by: Simon Wunderlich <siwu@hrz.tu-chemnitz.de>
Tested-by: Alexey Fisher <bug-track@fisher-privat.net>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agobatman-adv: correctly set the data field in the TT_REPONSE packet
Antonio Quartulli [Mon, 17 Oct 2011 12:25:13 +0000]
batman-adv: correctly set the data field in the TT_REPONSE packet

commit 9d8523931f7f5eb8900077f0da0fbe6b8ad0010b upstream.

In the TT_RESPONSE packet, the number of carried entries is not correctly set.
This leads to a wrong interpretation of the packet payload on the receiver side
causing random entries to be added to the global translation table. Therefore
the latter gets always corrupted, triggering a table recovery all the time.

Signed-off-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agobatman-adv: fix tt_local_reset_flags() function
Antonio Quartulli [Sun, 16 Oct 2011 16:53:37 +0000]
batman-adv: fix tt_local_reset_flags() function

commit 31901264511cf20c5ed33b8649a3ca9ce28df60b upstream.

Currently the counter of tt_local_entry structures (tt_local_num) is incremented
each time the tt_local_reset_flags() is invoked causing the node to send wrong
TT_REPONSE packets containing a copy of non-initialised memory thus corrupting
other nodes global translation table and making higher level communication
impossible.

Reported-by: Junkeun Song <jun361@gmail.com>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
Acked-by: Junkeun Song <jun361@gmail.com>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agodrm/i915: always set FDI composite sync bit
Jesse Barnes [Mon, 10 Oct 2011 21:28:52 +0000]
drm/i915: always set FDI composite sync bit

commit c4f9c4c2b3f1831e932e04db992cf6fe92c2a95a upstream.

It's needed for 3 pipe support as well as just regular functionality
(e.g. DisplayPort).

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Tested-by: Adam Jackson <ajax@redhat.com>
Tested-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Robert Hooker <robert.hooker@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agodrm/i915: fix IVB cursor support
Jesse Barnes [Wed, 12 Oct 2011 18:10:21 +0000]
drm/i915: fix IVB cursor support

commit 65a21cd65316145f9302594be8e69074369e1050 upstream.

The cursor regs have moved around, add the offsets and new macros for
getting at them.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Tested-By: Eugeni Dodonov <eugeni.dodonov@intel.com>
Reviewed-By: Eugeni Dodonov <eugeni.dodonov@intel.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Robert Hooker <robert.hooker@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoUSB: quirks: adding more quirky webcams to avoid squeaky audio
sordna [Fri, 28 Oct 2011 04:06:26 +0000]
USB: quirks: adding more quirky webcams to avoid squeaky audio

commit 0d145d7d4a241c321c832a810bb6edad18e2217b upstream.

The following patch contains additional affected webcam models, on top of the
patches commited to linux-next 2394d67e446bf616a0885167d5f0d397bdacfdfc
and 5b253d88cc6c65a23cefc457a5a4ef139913c5fc

Signed-off-by: sordna <sordna@gmail.com>
Cc: Oliver Neukum <oliver@neukum.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoUSB: add quirk for Logitech C600 web cam
Josh Boyer [Wed, 26 Oct 2011 17:53:17 +0000]
USB: add quirk for Logitech C600 web cam

commit 60c71ca972a2dd3fd9d0165b405361c8ad48349b upstream.

We've had another report of the "chipmunk" sound on a Logitech C600 webcam.
This patch resolves the issue.

Signed-off-by: Josh Boyer <jwboyer@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoUSB: EHCI: fix HUB TT scheduling issue with iso transfer
Thomas Poussevin [Thu, 27 Oct 2011 16:46:48 +0000]
USB: EHCI: fix HUB TT scheduling issue with iso transfer

commit 811c926c538f7e8d3c08b630dd5844efd7e000f6 upstream.

The current TT scheduling doesn't allow to play and then record on a
full-speed device connected to a high speed hub.

The IN iso stream can only start on the first uframe (0-2 for a 165 us)
because of CSPLIT transactions.
For the OUT iso stream there no such restriction. uframe 0-5 are possible.

The idea of this patch is that the first uframe are precious (for IN TT iso
stream) and we should allocate the last uframes first if possible.

For that we reverse the order of uframe allocation (last uframe first).

Here an example :

hid interrupt stream
----------------------------------------------------------------------
uframe                |  0  |  1  |  2  |  3  |  4  |  5  |  6  |  7  |
----------------------------------------------------------------------
max_tt_usecs          | 125 | 125 | 125 | 125 | 125 | 125 | 30  |  0  |
----------------------------------------------------------------------
used usecs on a frame | 13  |  0  |  0  |  0  |  0  |  0  |  0  |  0  |
----------------------------------------------------------------------

iso OUT stream
----------------------------------------------------------------------
uframe                |  0  |  1  |  2  |  3  |  4  |  5  |  6  |  7  |
----------------------------------------------------------------------
max_tt_usecs          | 125 | 125 | 125 | 125 | 125 | 125 | 30  |  0  |
----------------------------------------------------------------------
used usecs on a frame | 13  | 125 |  39 |  0  |  0  |  0  |  0  |  0  |
----------------------------------------------------------------------

There no place for iso IN stream  (uframe 0-2 are used) and we got "cannot
submit datapipe for urb 0, error -28: not enough bandwidth" error.

With the patch this become.

iso OUT stream
----------------------------------------------------------------------
uframe                |  0  |  1  |  2  |  3  |  4  |  5  |  6  |  7  |
----------------------------------------------------------------------
max_tt_usecs          | 125 | 125 | 125 | 125 | 125 | 125 | 30  |  0  |
----------------------------------------------------------------------
used usecs on a frame |  13 |  0  |  0  |  0  | 125 |  39 |  0  |  0  |
----------------------------------------------------------------------

iso IN stream
----------------------------------------------------------------------
uframe                |  0  |  1  |  2  |  3  |  4  |  5  |  6  |  7  |
----------------------------------------------------------------------
max_tt_usecs          | 125 | 125 | 125 | 125 | 125 | 125 | 30  |  0  |
----------------------------------------------------------------------
used usecs on a frame |  13 |  0  | 125 | 40  | 125 |  39 |  0  |  0  |
----------------------------------------------------------------------

Signed-off-by: Matthieu Castet <matthieu.castet@parrot.com>
Signed-off-by: Thomas Poussevin <thomas.poussevin@parrot.com>
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agousb-storage: Accept 8020i-protocol commands longer than 12 bytes
Alan Stern [Tue, 25 Oct 2011 14:50:58 +0000]
usb-storage: Accept 8020i-protocol commands longer than 12 bytes

commit 2f640bf4c94324aeaa1b6385c10aab8c5ad1e1cf upstream.

The 8020i protocol (also 8070i and QIC-157) uses 12-byte commands;
shorter commands must be padded.  Simon Detheridge reports that his
3-TB USB disk drive claims to use the 8020i protocol (which is
normally meant for ATAPI devices like CD drives), and because of its
large size, the disk drive requires the use of 16-byte commands.
However the usb_stor_pad12_command() routine in usb-storage always
sets the command length to 12, making the drive impossible to use.

Since the SFF-8020i specification allows for 16-byte commands in
future extensions, we may as well accept them.  This patch (as1490)
changes usb_stor_pad12_command() to leave commands larger than 12
bytes alone rather than truncating them.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Tested-by: Simon Detheridge <simon@widgit.com>
CC: Matthew Dharm <mdharm-usb@one-eyed-alien.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoUSB: Fix Corruption issue in USB ftdi driver ftdi_sio.c
Andrew Worsley [Fri, 18 Nov 2011 12:13:33 +0000]
USB: Fix Corruption issue in USB ftdi driver ftdi_sio.c

commit b1ffb4c851f185e9051ba837c16d9b84ef688d26 upstream.

Fix for ftdi_set_termios() glitching output

ftdi_set_termios() is constantly setting the baud rate, data bits and parity
unnecessarily on every call, . When called while characters are being
transmitted can cause the FTDI chip to corrupt the serial port bit stream
output by stalling the output half a bit during the output of a character.
Simple fix by skipping this setting if the baud rate/data bits/parity are
unchanged.

Signed-off-by: Andrew Worsley <amworsley@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoUSB: ark3116 initialisation fix
Bart Hartgers [Wed, 26 Oct 2011 11:29:42 +0000]
USB: ark3116 initialisation fix

commit 583182ba5f02c8c9be82ea550f2051eaec15b975 upstream.

This patch for the usb serial ark3116 driver fixes an initialisation
ordering bug that gets triggered on hotplug when using at least recent
debian/ubuntu userspace. Without it, ark3116 serial cables don't work.

Signed-off-by: Bart Hartgers <bart.hartgers@gmail.com>
Tested-by: law_ence.dev@ntlworld.com
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoUSB: workaround for bug in old version of GCC
Alan Stern [Thu, 27 Oct 2011 15:20:21 +0000]
USB: workaround for bug in old version of GCC

commit 97ff22ee3b4cb3a334f7385e269773141aed702f upstream.

This patch (as1491) works around a bug in GCC-3.4.6, which is still
supposed to be supported.  The number of microseconds in the udelay()
call in quirk_usb_disable_ehci() is fixed at 100, but the compiler
doesn't understand this and generates a link-time error.  So we
replace the otherwise unused variable "delta" with a simple constant
100.  This same pattern is already used in other delay loops in that
source file.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Reported-by: Konrad Rzepecki <krzepecki@dentonet.pl>
Tested-by: Konrad Rzepecki <krzepecki@dentonet.pl>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoUSB: cdc-acm: Fix disconnect() vs close() race
Havard Skinnemoen [Wed, 9 Nov 2011 21:47:38 +0000]
USB: cdc-acm: Fix disconnect() vs close() race

commit 5dc2470c602da8851907ec18942cd876c3b4ecc1 upstream.

There's a race between the USB disconnect handler and the TTY close
handler which may cause the acm object to be freed while it's still
being used. This may lead to things like

http://article.gmane.org/gmane.linux.usb.general/54250

and

https://lkml.org/lkml/2011/5/29/64

This is the simplest fix I could come up with. Holding on to open_mutex
while closing the TTY device prevents acm_disconnect() from freeing the
acm object between acm->port.count drops to 0 and the TTY side of the
cleanups are finalized.

Signed-off-by: Havard Skinnemoen <hskinnemoen@google.com>
Cc: Oliver Neukum <oliver@neukum.name>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoUSB: pch_udc: Support new device LAPIS Semiconductor ML7831 IOH
Tomoya MORINAGA [Fri, 28 Oct 2011 00:37:34 +0000]
USB: pch_udc: Support new device LAPIS Semiconductor ML7831 IOH

commit 731ad81e2dd97e3f222361f7b3ff4b35639e46af upstream.

ML7831 is companion chip for Intel Atom E6xx series.

Signed-off-by: Tomoya MORINAGA <tomoya-linux@dsn.lapis-semi.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoUSB: serial: pl2303: rm duplicate id
wangyanqing [Thu, 10 Nov 2011 06:04:08 +0000]
USB: serial: pl2303: rm duplicate id

commit 0c16595539b612fe948559433dda08ff96a8bdc7 upstream.

I get report from customer that his usb-serial
converter doesn't work well,it sometimes work,
but sometimes it doesn't.

The usb-serial converter's id:
vendor_id product_id
0x4348    0x5523

Then I search the usb-serial codes, and there are
two drivers announce support this device, pl2303
and ch341, commit 026dfaf1 cause it. Through many
times to test, ch341 works well with this device,
and pl2303 doesn't work quite often(it just work quite little).

ch341 works well with this device, so we doesn't
need pl2303 to support.I try to revert 026dfaf1 first,
but it failed. So I prepare this patch by hand to revert it.

Signed-off-by: Wang YanQing <Udknight@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoUSB: option: add PID of Huawei E173s 3G modem
Ferenc Wagner [Thu, 17 Nov 2011 15:44:58 +0000]
USB: option: add PID of Huawei E173s 3G modem

commit 4aa3648c719265bac9c2742c9ebb043e6dbdd790 upstream.

Signed-off-by: Ferenc Wagner <wferi@niif.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoUSB: option: release new PID for ZTE 3G modem
zheng.zhijian@zte.com.cn [Thu, 17 Nov 2011 11:23:25 +0000]
USB: option: release new PID for ZTE 3G modem

commit 46b5a277ed90317a4d17e936c16037e76011b219 upstream.

This patch adds new PIDs for ZTE 3G modem, after we confirm it and tested.
Thanks for Dan's work at kernel option devier.

Signed-off-by: Alvin.Zheng <zheng.zhijian@zte.com.cn>
Signed-off-by: wsalvin <wsalvin@yahoo.com.cn>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoUSB: XHCI: resume root hubs when the controller resumes
Alan Stern [Thu, 3 Nov 2011 15:37:10 +0000]
USB: XHCI: resume root hubs when the controller resumes

commit f69e3120df82391a0ee8118e0a156239a06b2afb upstream.

This patch (as1494) fixes a problem in xhci-hcd's resume routine.
When the controller is runtime-resumed, this can only mean that one of
the two root hubs has made a wakeup request and therefore needs to be
resumed as well.  Rather than try to determine which root hub requires
attention (which might be difficult in the case where a new
non-SuperSpeed device has been plugged in), the patch simply resumes
both root hubs.

Without this change, there is a race: The controller might be put back
to sleep before it can activate its IRQ line, and the wakeup condition
might never get handled.

The patch also simplifies the logic in xhci_resume a little, combining
some repeated flag settings into a single pair of statements.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agousb, xhci: fix lockdep warning on endpoint timeout
Don Zickus [Fri, 21 Oct 2011 03:52:14 +0000]
usb, xhci: fix lockdep warning on endpoint timeout

commit f43d623164022dcbf6750ef220b7a1133a1183eb upstream.

While debugging a usb3 problem, I stumbled upon this lockdep warning.

Oct 18 21:41:17 dhcp47-74 kernel: =================================
Oct 18 21:41:17 dhcp47-74 kernel: [ INFO: inconsistent lock state ]
Oct 18 21:41:17 dhcp47-74 kernel: 3.1.0-rc4nmi+ #456
Oct 18 21:41:17 dhcp47-74 kernel: ---------------------------------
Oct 18 21:41:17 dhcp47-74 kernel: inconsistent {IN-HARDIRQ-W} -> {HARDIRQ-ON-W} usage.
Oct 18 21:41:17 dhcp47-74 kernel: swapper/0 [HC0[0]:SC1[1]:HE1:SE0] takes:
Oct 18 21:41:17 dhcp47-74 kernel: (&(&xhci->lock)->rlock){?.-...}, at: [<ffffffffa0228990>] xhci_stop_endpoint_command_watchdog+0x30/0x340 [xhci_hcd]
Oct 18 21:41:17 dhcp47-74 kernel: {IN-HARDIRQ-W} state was registered at:
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff8109a941>] __lock_acquire+0x781/0x1660
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff8109bed7>] lock_acquire+0x97/0x170
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff81501b46>] _raw_spin_lock+0x46/0x80
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffffa02299fa>] xhci_irq+0x3a/0x1960 [xhci_hcd]
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffffa022b351>] xhci_msi_irq+0x31/0x40 [xhci_hcd]
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff810d2305>] handle_irq_event_percpu+0x85/0x320
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff810d25e8>] handle_irq_event+0x48/0x70
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff810d537d>] handle_edge_irq+0x6d/0x130
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff810048c9>] handle_irq+0x49/0xa0
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff8150d56d>] do_IRQ+0x5d/0xe0
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff815029b0>] ret_from_intr+0x0/0x13
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff81388aca>] usb_set_device_state+0x8a/0x180
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff8138f038>] usb_add_hcd+0x2b8/0x730
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffffa022ed7e>] xhci_pci_probe+0x9e/0xd4 [xhci_hcd]
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff8127915f>] local_pci_probe+0x5f/0xd0
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff8127a569>] pci_device_probe+0x119/0x120
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff81334473>] driver_probe_device+0xa3/0x2c0
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff8133473b>] __driver_attach+0xab/0xb0
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff8133373c>] bus_for_each_dev+0x6c/0xa0
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff813341fe>] driver_attach+0x1e/0x20
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff81333b88>] bus_add_driver+0x1f8/0x2b0
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff81334df6>] driver_register+0x76/0x140
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff8127a7c6>] __pci_register_driver+0x66/0xe0
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffffa013c04a>] snd_timer_find+0x4a/0x70 [snd_timer]
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffffa013c00e>] snd_timer_find+0xe/0x70 [snd_timer]
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff810001d3>] do_one_initcall+0x43/0x180
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff810a9ed2>] sys_init_module+0x92/0x1f0
Oct 18 21:41:17 dhcp47-74 kernel:  [<ffffffff8150ab6b>] system_call_fastpath+0x16/0x1b
Oct 18 21:41:17 dhcp47-74 kernel: irq event stamp: 631984
Oct 18 21:41:17 dhcp47-74 kernel: hardirqs last  enabled at (631984): [<ffffffff81502720>] _raw_spin_unlock_irq+0x30/0x50
Oct 18 21:41:17 dhcp47-74 kernel: hardirqs last disabled at (631983): [<ffffffff81501c49>] _raw_spin_lock_irq+0x19/0x90
Oct 18 21:41:17 dhcp47-74 kernel: softirqs last  enabled at (631980): [<ffffffff8105ff63>] _local_bh_enable+0x13/0x20
Oct 18 21:41:17 dhcp47-74 kernel: softirqs last disabled at (631981): [<ffffffff8150ce6c>] call_softirq+0x1c/0x30
Oct 18 21:41:17 dhcp47-74 kernel:
Oct 18 21:41:17 dhcp47-74 kernel: other info that might help us debug this:
Oct 18 21:41:17 dhcp47-74 kernel: Possible unsafe locking scenario:
Oct 18 21:41:17 dhcp47-74 kernel:
Oct 18 21:41:17 dhcp47-74 kernel:       CPU0
Oct 18 21:41:17 dhcp47-74 kernel:       ----
Oct 18 21:41:17 dhcp47-74 kernel:  lock(&(&xhci->lock)->rlock);
Oct 18 21:41:17 dhcp47-74 kernel:  <Interrupt>
Oct 18 21:41:17 dhcp47-74 kernel:    lock(&(&xhci->lock)->rlock);
Oct 18 21:41:17 dhcp47-74 kernel:
Oct 18 21:41:17 dhcp47-74 kernel: *** DEADLOCK ***
Oct 18 21:41:17 dhcp47-74 kernel:
Oct 18 21:41:17 dhcp47-74 kernel: 1 lock held by swapper/0:
Oct 18 21:41:17 dhcp47-74 kernel: #0:  (&ep->stop_cmd_timer){+.-...}, at: [<ffffffff8106abf2>] run_timer_softirq+0x162/0x570
Oct 18 21:41:17 dhcp47-74 kernel:
Oct 18 21:41:17 dhcp47-74 kernel: stack backtrace:
Oct 18 21:41:17 dhcp47-74 kernel: Pid: 0, comm: swapper Tainted: G        W   3.1.0-rc4nmi+ #456
Oct 18 21:41:17 dhcp47-74 kernel: Call Trace:
Oct 18 21:41:17 dhcp47-74 kernel: <IRQ>  [<ffffffff81098ed7>] print_usage_bug+0x227/0x270
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff810999c6>] mark_lock+0x346/0x410
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff8109a7de>] __lock_acquire+0x61e/0x1660
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff81099893>] ? mark_lock+0x213/0x410
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff8109bed7>] lock_acquire+0x97/0x170
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffffa0228990>] ? xhci_stop_endpoint_command_watchdog+0x30/0x340 [xhci_hcd]
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff81501b46>] _raw_spin_lock+0x46/0x80
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffffa0228990>] ? xhci_stop_endpoint_command_watchdog+0x30/0x340 [xhci_hcd]
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffffa0228990>] xhci_stop_endpoint_command_watchdog+0x30/0x340 [xhci_hcd]
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff8106abf2>] ? run_timer_softirq+0x162/0x570
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff8106ac9d>] run_timer_softirq+0x20d/0x570
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff8106abf2>] ? run_timer_softirq+0x162/0x570
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffffa0228960>] ? xhci_queue_isoc_tx_prepare+0x8e0/0x8e0 [xhci_hcd]
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff810604d2>] __do_softirq+0xf2/0x3f0
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff81020edd>] ? lapic_next_event+0x1d/0x30
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff81090d4e>] ? clockevents_program_event+0x5e/0x90
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff8150ce6c>] call_softirq+0x1c/0x30
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff8100484d>] do_softirq+0x8d/0xc0
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff8105ff35>] irq_exit+0xe5/0x100
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff8150d65e>] smp_apic_timer_interrupt+0x6e/0x99
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff8150b6f0>] apic_timer_interrupt+0x70/0x80
Oct 18 21:41:17 dhcp47-74 kernel: <EOI>  [<ffffffff81095d8d>] ? trace_hardirqs_off+0xd/0x10
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff812ddb76>] ? acpi_idle_enter_bm+0x227/0x25b
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff812ddb71>] ? acpi_idle_enter_bm+0x222/0x25b
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff813eda63>] cpuidle_idle_call+0x103/0x290
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff81002155>] cpu_idle+0xe5/0x160
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff814e7f50>] rest_init+0xe0/0xf0
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff814e7e70>] ? csum_partial_copy_generic+0x170/0x170
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff81df8e23>] start_kernel+0x3fc/0x407
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff81df8321>] x86_64_start_reservations+0x131/0x135
Oct 18 21:41:17 dhcp47-74 kernel: [<ffffffff81df8412>] x86_64_start_kernel+0xed/0xf4
Oct 18 21:41:17 dhcp47-74 kernel: xhci_hcd 0000:00:14.0: xHCI host not responding to stop endpoint command.
Oct 18 21:41:17 dhcp47-74 kernel: xhci_hcd 0000:00:14.0: Assuming host is dying, halting host.
Oct 18 21:41:17 dhcp47-74 kernel: xhci_hcd 0000:00:14.0: HC died; cleaning up
Oct 18 21:41:17 dhcp47-74 kernel: usb 3-4: device descriptor read/8, error -110
Oct 18 21:41:17 dhcp47-74 kernel: usb 3-4: device descriptor read/8, error -22
Oct 18 21:41:17 dhcp47-74 kernel: hub 3-0:1.0: cannot disable port 4 (err = -19)

Basically what is happening is in xhci_stop_endpoint_command_watchdog()
the xhci->lock is grabbed with just spin_lock.  What lockdep deduces is
that if an interrupt occurred while in this function it would deadlock
with xhci_irq because that function also grabs the xhci->lock.

Fixing it is trivial by using spin_lock_irqsave instead.

This should be queued to stable kernels as far back as 2.6.33.

Signed-off-by: Don Zickus <dzickus@redhat.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agousb, xhci: Clear warm reset change event during init
Don Zickus [Thu, 3 Nov 2011 13:07:18 +0000]
usb, xhci: Clear warm reset change event during init

commit 79c3dd8150fd5236d95766a9e662e3e932b462c9 upstream.

I noticed on my Panther Point system that I wasn't getting hotplug events
for my usb3.0 disk on a usb3 port.  I tracked it down to the fact that the
system had the warm reset change bit still set.  This seemed to block future
events from being received, including a hotplug event.

Clearing this bit during initialization allowed the hotplug event to be
received and the disk to be recognized correctly.

This patch should be backported to kernels as old as 2.6.39.

Signed-off-by: Don Zickus <dzickus@redhat.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoxhci: Set slot and ep0 flags for address command.
Sarah Sharp [Thu, 3 Nov 2011 20:06:08 +0000]
xhci: Set slot and ep0 flags for address command.

commit d31c285b3a71cf9056e6a060de41f37780b0af86 upstream.

Matt's AsMedia xHCI host controller was responding with a Context Error
to an address device command after a configured device reset.  Some
sequence of events leads both the slot and endpoint zero add flags
cleared to zero, which the AsMedia host doesn't like:

[  223.701839] xhci_hcd 0000:03:00.0: Slot ID 1 Input Context:
[  223.701841] xhci_hcd 0000:03:00.0: @ffff880137b25000 (virt) @ffffc000 (dma) 0x000000 - drop flags
[  223.701843] xhci_hcd 0000:03:00.0: @ffff880137b25004 (virt) @ffffc004 (dma) 0x000000 - add flags
[  223.701846] xhci_hcd 0000:03:00.0: @ffff880137b25008 (virt) @ffffc008 (dma) 0x000000 - rsvd2[0]
[  223.701848] xhci_hcd 0000:03:00.0: @ffff880137b2500c (virt) @ffffc00c (dma) 0x000000 - rsvd2[1]
[  223.701850] xhci_hcd 0000:03:00.0: @ffff880137b25010 (virt) @ffffc010 (dma) 0x000000 - rsvd2[2]
[  223.701852] xhci_hcd 0000:03:00.0: @ffff880137b25014 (virt) @ffffc014 (dma) 0x000000 - rsvd2[3]
[  223.701854] xhci_hcd 0000:03:00.0: @ffff880137b25018 (virt) @ffffc018 (dma) 0x000000 - rsvd2[4]
[  223.701857] xhci_hcd 0000:03:00.0: @ffff880137b2501c (virt) @ffffc01c (dma) 0x000000 - rsvd2[5]
[  223.701858] xhci_hcd 0000:03:00.0: Slot Context:
[  223.701860] xhci_hcd 0000:03:00.0: @ffff880137b25020 (virt) @ffffc020 (dma) 0x8400000 - dev_info
[  223.701862] xhci_hcd 0000:03:00.0: @ffff880137b25024 (virt) @ffffc024 (dma) 0x010000 - dev_info2
[  223.701864] xhci_hcd 0000:03:00.0: @ffff880137b25028 (virt) @ffffc028 (dma) 0x000000 - tt_info
[  223.701866] xhci_hcd 0000:03:00.0: @ffff880137b2502c (virt) @ffffc02c (dma) 0x000000 - dev_state
[  223.701869] xhci_hcd 0000:03:00.0: @ffff880137b25030 (virt) @ffffc030 (dma) 0x000000 - rsvd[0]
[  223.701871] xhci_hcd 0000:03:00.0: @ffff880137b25034 (virt) @ffffc034 (dma) 0x000000 - rsvd[1]
[  223.701873] xhci_hcd 0000:03:00.0: @ffff880137b25038 (virt) @ffffc038 (dma) 0x000000 - rsvd[2]
[  223.701875] xhci_hcd 0000:03:00.0: @ffff880137b2503c (virt) @ffffc03c (dma) 0x000000 - rsvd[3]
[  223.701877] xhci_hcd 0000:03:00.0: Endpoint 00 Context:
[  223.701879] xhci_hcd 0000:03:00.0: @ffff880137b25040 (virt) @ffffc040 (dma) 0x000000 - ep_info
[  223.701881] xhci_hcd 0000:03:00.0: @ffff880137b25044 (virt) @ffffc044 (dma) 0x2000026 - ep_info2
[  223.701883] xhci_hcd 0000:03:00.0: @ffff880137b25048 (virt) @ffffc048 (dma) 0xffffe8e0 - deq
[  223.701885] xhci_hcd 0000:03:00.0: @ffff880137b25050 (virt) @ffffc050 (dma) 0x000000 - tx_info
[  223.701887] xhci_hcd 0000:03:00.0: @ffff880137b25054 (virt) @ffffc054 (dma) 0x000000 - rsvd[0]
[  223.701889] xhci_hcd 0000:03:00.0: @ffff880137b25058 (virt) @ffffc058 (dma) 0x000000 - rsvd[1]
[  223.701892] xhci_hcd 0000:03:00.0: @ffff880137b2505c (virt) @ffffc05c (dma) 0x000000 - rsvd[2]
...
[  223.701927] xhci_hcd 0000:03:00.0: // Ding dong!
[  223.701992] xhci_hcd 0000:03:00.0: Setup ERROR: address device command for slot 1.

The xHCI spec says that both flags must be set to one for the Address
Device command.  When the device is first enumerated,
xhci_setup_addressable_virt_dev() does set those flags.  However, when
the device is addressed after it has been reset in the configured state,
xhci_setup_addressable_virt_dev() is not called, and
xhci_copy_ep0_dequeue_into_input_ctx() is called instead.  That function
relies on the flags being set up by previous commands, which apparently
isn't a good assumption.

Move the setting of the flags into the common parent function.

This should be queued for stable kernels as old as 2.6.35, since that
was the first introduction of xhci_copy_ep0_dequeue_into_input_ctx.

Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Matt <mdm@iinet.net.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agodrivers/base/node.c: fix compilation error with older versions of gcc
Claudio Scordino [Thu, 17 Nov 2011 10:08:32 +0000]
drivers/base/node.c: fix compilation error with older versions of gcc

commit 91a13c281d7d4648c0b32dede11a0144c4e7984c upstream.

Patch to fix the error message "directives may not be used inside a macro
argument" which appears when the kernel is compiled for the cris architecture.

Signed-off-by: Claudio Scordino <claudio@evidence.eu.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agopcie-gadget-spear: Add "platform:" prefix for platform modalias
Axel Lin [Mon, 31 Oct 2011 02:20:28 +0000]
pcie-gadget-spear: Add "platform:" prefix for platform modalias

commit 161f14191dc166c4e3f37f68af1bc199c6868b7d upstream.

Since 43cc71eed1250755986da4c0f9898f9a635cb3bf (platform: prefix MODALIAS
with "platform:"), the platform modalias is prefixed with "platform:".

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Acked-by: Pratyush Anand <pratyush.anand@st.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoALSA: hda - fix ELD memory leak
Wu Fengguang [Tue, 22 Nov 2011 08:58:35 +0000]
ALSA: hda - fix ELD memory leak

Backported from commit b95d68b8179764e29558b75cec35ef4a6a98925b.

memset(eld) clears eld->proc_entry which will leak the struct
snd_info_entry when unloading module.

Fix it by
- memset only the fields before eld->eld_buffer
- set eld->eld_valid to true _after_ all eld fields have been filled

Cc: Pierre-louis Bossart <pierre-louis.bossart@intel.com>
Acked-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agonfs: when attempting to open a directory, fall back on normal lookup (try #5)
Jeff Layton [Fri, 4 Nov 2011 17:31:21 +0000]
nfs: when attempting to open a directory, fall back on normal lookup (try #5)

commit 1788ea6e3b2a58cf4fb00206e362d9caff8d86a7 upstream.

commit d953126 changed how nfs_atomic_lookup handles an -EISDIR return
from an OPEN call. Prior to that patch, that caused the client to fall
back to doing a normal lookup. When that patch went in, the code began
returning that error to userspace. The d_revalidate codepath however
never had the corresponding change, so it was still possible to end up
with a NULL ctx->state pointer after that.

That patch caused a regression. When we attempt to open a directory that
does not have a cached dentry, that open now errors out with EISDIR. If
you attempt the same open with a cached dentry, it will succeed.

Fix this by reverting the change in nfs_atomic_lookup and allowing
attempts to open directories to fall back to a normal lookup

Also, add a NFSv4-specific f_ops->open routine that just returns
-ENOTDIR. This should never be called if things are working properly,
but if it ever is, then the dprintk may help in debugging.

To facilitate this, a new file_operations field is also added to the
nfs_rpc_ops struct.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoTTY: ldisc, wait for ldisc infinitely in hangup
Jiri Slaby [Wed, 16 Nov 2011 15:27:09 +0000]
TTY: ldisc, wait for ldisc infinitely in hangup

commit 0c73c08ec73dbe080b9ec56696ee21d32754d918 upstream.

For /dev/console case, we do not kill all ldisc users. It's due to
redirected_tty_write test in __tty_hangup. In that case there still
might be a process waiting e.g. in n_tty_read for input.

We wait for such processes to disappear. The problem is that we use a
timeout. After this timeout, we continue closing the ldisc and start
freeing tty resources. It obviously leads to crashes when the other
process is woken.

So to fix this, we wait infinitely before reiniting the ldisc. (The
tiocsetd remains untouched -- times out after 5s.)

This is nicely reproducible with this run from shell:
  exec 0<>/dev/console 1<>/dev/console 2<>/dev/console
and stopping a getty like:
  systemctl stop serial-getty@ttyS0.service

The crash proper may be produced only under load or with constified
timing the same as for 92f6fa09b.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Dave Young <hidave.darkstar@gmail.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Ben Hutchings <ben@decadent.org.uk>
Cc: Dmitriy Matrosov <sgf.dma@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoTTY: ldisc, move wait idle to caller
Jiri Slaby [Wed, 16 Nov 2011 15:27:08 +0000]
TTY: ldisc, move wait idle to caller

commit 300420722e0734a4254f3b634e0f82664495d210 upstream.

It is the only place where reinit is called from. And we really need
to wait for the old ldisc to go once. Actually this is the place where
the waiting originally was (before removed and re-added later).

This will make the fix in the following patch easier to implement.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Dave Young <hidave.darkstar@gmail.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Ben Hutchings <ben@decadent.org.uk>
Cc: Dmitriy Matrosov <sgf.dma@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoTTY: ldisc, allow waiting for ldisc arbitrarily long
Jiri Slaby [Wed, 16 Nov 2011 15:27:07 +0000]
TTY: ldisc, allow waiting for ldisc arbitrarily long

commit df92d0561de364de53c42abc5d43e04ab6f326a5 upstream.

To fix a nasty bug in ldisc hup vs. reinit we need to wait infinitely
long for ldisc to be gone. So here we add a parameter to
tty_ldisc_wait_idle to allow that.

This is only a preparation for the real fix which is done in the
following patches.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Dave Young <hidave.darkstar@gmail.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Ben Hutchings <ben@decadent.org.uk>
Cc: Dmitriy Matrosov <sgf.dma@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agotty: hvc_dcc: Fix duplicate character inputs
Stephen Boyd [Wed, 26 Oct 2011 02:19:43 +0000]
tty: hvc_dcc: Fix duplicate character inputs

commit c2a3e84f950e7ddba1f3914b005861d46ae60359 upstream.

Reading from the DCC grabs a character from the buffer and
clears the status bit. Since this is a context-changing
operation, instructions following the character read that rely on
the status bit being accurate need to be synchronized with an
ISB.

In this case, the status bit check needs to execute after the
character read otherwise we run the risk of reading the character
and checking the status bit before the read can clear the status
bit in the first place. When this happens, the user will see the
same character they typed twice, instead of once.

Add an ISB after the read and the write, so that the status check
is synchronized with the read/write operations.

Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agopch_uart: Support new device LAPIS Semiconductor ML7831 IOH
Tomoya MORINAGA [Fri, 28 Oct 2011 00:38:49 +0000]
pch_uart: Support new device LAPIS Semiconductor ML7831 IOH

commit 8249f743f732ccbc3056428945ab1d9bd36d46bf upstream.

ML7831 is companion chip for Intel Atom E6xx series.

Signed-off-by: Tomoya MORINAGA <tomoya-linux@dsn.lapis-semi.com>
Acked-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agopch_uart: Fix DMA resource leak issue
Tomoya MORINAGA [Fri, 11 Nov 2011 01:55:27 +0000]
pch_uart: Fix DMA resource leak issue

commit 90f04c2926cfb5bf74533b0a7766bc896f6a0c0e upstream.

Changing UART mode PIO->DMA->PIO->DMA like below, pch_uart driver can't get
DMA channel resource.

setserial /dev/ttyPCH0 ^low_latency
setserial /dev/ttyPCH0 low_latency

CAUSE:
Changing mode using setserial command, ".startup" function which gets DMA
channel is called before ".verify_port" function which sets
dma-flag(use_dma/use_dma_flag) as 1.

PIO->DMA
  .startup: Since dma-flag is 0, DMA channel is not requested.
  .verify_port: dma-flag is set as 1.
  .shutdown: N/A

DMA->PIO
  .startup: Since dma-flag is 1, DMA channel is requested.
  .verify_port: dma-flag is set as 0.
  .shutdown: Since dma-flag is 0, DMA channel is not released.

This means DMA channel resource leak occurs.
Next time, this driver can't get DMA channel resource forever.

MODIFICATION:
  Currently, when release DMA channel resource, this driver checks dma-flag.
  However, this specification occurs the above issue.
  This driver must check whether dma_request_channel is executed or not.
  The values are saved in private data variable "chan_tx/chan_tx".
  These variables mean if the value is NULL, DMA channel is not requested,
  if not NULL, DMA channel is requested.

This patch fixes the issue.

Signed-off-by: Tomoya MORINAGA <tomoya.rohm@gmail.com>
Acked-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agopch_uart: Fix hw-flow control issue
Tomoya MORINAGA [Thu, 27 Oct 2011 06:45:18 +0000]
pch_uart: Fix hw-flow control issue

commit a1d7cfe29f13cf45f8094929864b9c66bf0cd91b upstream.

Using hardware flow control,
currently, register of the control-bit(AFE) is not set.
This patch fixes the issue.

Signed-off-by: Tomoya MORINAGA <tomoya-linux@dsn.lapis-semi.com>
Acked-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agopch_phub: Fix MAC address writing issue for LAPIS ML7831
Tomoya MORINAGA [Fri, 11 Nov 2011 01:12:18 +0000]
pch_phub: Fix MAC address writing issue for LAPIS ML7831

commit 2a9887919457c6e1bd482e8448223be59d19010a upstream.

ISSUE:
Using ML7831, MAC address writing doesn't work well.

CAUSE:
ML7831 and EG20T have the same register map for MAC address access.
However, this driver processes the writing the same as ML7223.
This is not true.
This driver must process the writing the same as EG20T.
This patch fixes the issue.

Signed-off-by: Tomoya MORINAGA <tomoya.rohm@gmail.com>
Cc: Masayuki Ohtak <masa-korg@dsn.okisemi.com>
Cc: Alexander Stein <alexander.stein@systec-electronic.com>
Cc: Denis Turischev <denis@compulab.co.il>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agopch_phub: Support new device LAPIS Semiconductor ML7831 IOH
Tomoya MORINAGA [Fri, 28 Oct 2011 00:33:13 +0000]
pch_phub: Support new device LAPIS Semiconductor ML7831 IOH

commit 584ad00ce4bfe594e4c4a89944b3c635187a1ca1 upstream.

ML7831 is companion chip for Intel Atom E6xx series.

Signed-off-by: Tomoya MORINAGA <tomoya-linux@dsn.lapis-semi.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoPM / driver core: disable device's runtime PM during shutdown
Peter Chen [Tue, 15 Nov 2011 20:52:29 +0000]
PM / driver core: disable device's runtime PM during shutdown

commit af8db1508f2c9f3b6e633e2d2d906c6557c617f9 upstream.

There may be an issue when the user issue "reboot/shutdown" command, then
the device has shut down its hardware, after that, this runtime-pm featured
device's driver will probably be scheduled to do its suspend routine,
and at its suspend routine, it may access hardware, but the device has
already shutdown physically, then the system hang may be occurred.

I ran out this issue using an auto-suspend supported USB devices, like
3G modem, keyboard. The usb runtime suspend routine may be scheduled
after the usb controller has been shut down, and the usb runtime suspend
routine will try to suspend its roothub(controller), it will access
register, then the system hang occurs as the controller is shutdown.

Signed-off-by: Peter Chen <peter.chen@freescale.com>
Acked-by: Ming Lei <tom.leiming@gmail.com>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agoip6_tunnel: copy parms.name after register_netdevice
Josh Boyer [Thu, 10 Nov 2011 15:10:23 +0000]
ip6_tunnel: copy parms.name after register_netdevice

commit 731abb9cb27aef6013ce60808a04e04a545f3f4e upstream.

Commit 1c5cae815d removed an explicit call to dev_alloc_name in ip6_tnl_create
because register_netdevice will now create a valid name.  This works for the
net_device itself.

However the tunnel keeps a copy of the name in the parms structure for the
ip6_tnl associated with the tunnel.  parms.name is set by copying the net_device
name in ip6_tnl_dev_init_gen.  That function is called from ip6_tnl_dev_init in
ip6_tnl_create, but it is done before register_netdevice is called so the name
is set to a bogus value in the parms.name structure.

This shows up if you do a simple tunnel add, followed by a tunnel show:

[root@localhost ~]# ip -6 tunnel add remote fec0::100 local fec0::200
[root@localhost ~]# ip -6 tunnel show
ip6tnl0: ipv6/ipv6 remote :: local :: encaplimit 0 hoplimit 0 tclass 0x00 flowlabel 0x00000 (flowinfo 0x00000000)
ip6tnl%d: ipv6/ipv6 remote fec0::100 local fec0::200 encaplimit 4 hoplimit 64 tclass 0x00 flowlabel 0x00000 (flowinfo 0x00000000)
[root@localhost ~]#

Fix this by moving the strcpy out of ip6_tnl_dev_init_gen, and calling it after
register_netdevice has successfully returned.

Signed-off-by: Josh Boyer <jwboyer@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agocfg80211: fix bug on regulatory core exit on access to last_request
Luis R. Rodriguez [Tue, 8 Nov 2011 22:28:06 +0000]
cfg80211: fix bug on regulatory core exit on access to last_request

commit 58ebacc66bd11be2327edcefc79de94bd6f5bb4a upstream.

Commit 4d9d88d1 by Scott James Remnant <keybuk@google.com> added
the .uevent() callback for the regulatory device used during
the platform device registration. The change was done to account
for queuing up udev change requests through udevadm triggers.
The change also meant that upon regulatory core exit we will now
send a uevent() but the uevent() callback, reg_device_uevent(),
also accessed last_request. Right before commiting device suicide
we free'd last_request but never set it to NULL so
platform_device_unregister() would lead to bogus kernel paging
request. Fix this and also simply supress uevents right before
we commit suicide as they are pointless.

This fix is required for kernels >= v2.6.39

$ git describe --contains 4d9d88d1
v2.6.39-rc1~468^2~25^2^2~21

The impact of not having this present is that a bogus paging
access may occur (only read) upon cfg80211 unload time. You
may also get this BUG complaint below. Although Johannes
could not reproduce the issue this fix is theoretically correct.

mac80211_hwsim: unregister radios
mac80211_hwsim: closing netlink
BUG: unable to handle kernel paging request at ffff88001a06b5ab
IP: [<ffffffffa030df9a>] reg_device_uevent+0x1a/0x50 [cfg80211]
PGD 1836063 PUD 183a063 PMD 1ffcb067 PTE 1a06b160
Oops: 0000 [#1] PREEMPT SMP DEBUG_PAGEALLOC
CPU 0
Modules linked in: cfg80211(-) [last unloaded: mac80211]

Pid: 2279, comm: rmmod Tainted: G        W   3.1.0-wl+ #663 Bochs Bochs
RIP: 0010:[<ffffffffa030df9a>]  [<ffffffffa030df9a>] reg_device_uevent+0x1a/0x50 [cfg80211]
RSP: 0000:ffff88001c5f9d58  EFLAGS: 00010286
RAX: 0000000000000000 RBX: ffff88001d2eda88 RCX: ffff88001c7468fc
RDX: ffff88001a06b5a0 RSI: ffff88001c7467b0 RDI: ffff88001c7467b0
RBP: ffff88001c5f9d58 R08: 000000000000ffff R09: 000000000000ffff
R10: 0000000000000000 R11: 0000000000000001 R12: ffff88001c7467b0
R13: ffff88001d2eda78 R14: ffffffff8164a840 R15: 0000000000000001
FS:  00007f8a91d8a6e0(0000) GS:ffff88001fc00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: ffff88001a06b5ab CR3: 000000001c62e000 CR4: 00000000000006f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process rmmod (pid: 2279, threadinfo ffff88001c5f8000, task ffff88000023c780)
Stack:
 ffff88001c5f9d98 ffffffff812ff7e5 ffffffff8176ab3d ffff88001c7468c2
 000000000000ffff ffff88001d2eda88 ffff88001c7467b0 ffff880000114820
 ffff88001c5f9e38 ffffffff81241dc7 ffff88001c5f9db8 ffffffff81040189
Call Trace:
 [<ffffffff812ff7e5>] dev_uevent+0xc5/0x170
 [<ffffffff81241dc7>] kobject_uevent_env+0x1f7/0x490
 [<ffffffff81040189>] ? sub_preempt_count+0x29/0x60
 [<ffffffff814cab1a>] ? _raw_spin_unlock_irqrestore+0x4a/0x90
 [<ffffffff81305307>] ? devres_release_all+0x27/0x60
 [<ffffffff8124206b>] kobject_uevent+0xb/0x10
 [<ffffffff812fee27>] device_del+0x157/0x1b0
 [<ffffffff8130377d>] platform_device_del+0x1d/0x90
 [<ffffffff81303b76>] platform_device_unregister+0x16/0x30
 [<ffffffffa030fffd>] regulatory_exit+0x5d/0x180 [cfg80211]
 [<ffffffffa032bec3>] cfg80211_exit+0x2b/0x45 [cfg80211]
 [<ffffffff8109a84c>] sys_delete_module+0x16c/0x220
 [<ffffffff8108a23e>] ? trace_hardirqs_on_caller+0x7e/0x120
 [<ffffffff814cba02>] system_call_fastpath+0x16/0x1b
Code: <all your base are belong to me>
RIP  [<ffffffffa030df9a>] reg_device_uevent+0x1a/0x50 [cfg80211]
 RSP <ffff88001c5f9d58>
CR2: ffff88001a06b5ab
---[ end trace 147c5099a411e8c0 ]---

Reported-by: Johannes Berg <johannes@sipsolutions.net>
Cc: Scott James Remnant <keybuk@google.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

7 years agonl80211: fix HT capability attribute validation
Johannes Berg [Thu, 3 Nov 2011 08:27:01 +0000]
nl80211: fix HT capability attribute validation

commit 6c7394197af90f6a332180e33f5d025d3037d883 upstream.

Since the NL80211_ATTR_HT_CAPABILITY attribute is
used as a struct, it needs a minimum, not maximum
length. Enforce that properly. Not doing so could
potentially lead to reading after the buffer.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>