6 years agogta02: Add gsm power management device
Lars-Peter Clausen [Mon, 17 May 2010 18:25:20 +0000 (20:25 +0200)]
gta02: Add gsm power management device

6 years agogta02: Add gps power management device
Lars-Peter Clausen [Mon, 17 May 2010 18:22:47 +0000 (20:22 +0200)]
gta02: Add gps power management device

6 years agogta02: Add bt power management device
Lars-Peter Clausen [Mon, 17 May 2010 18:19:36 +0000 (20:19 +0200)]
gta02: Add bt power management device

6 years agogta02: Add notify handler to probe device children
Lars-Peter Clausen [Sun, 4 Apr 2010 18:37:28 +0000 (20:37 +0200)]
gta02: Add notify handler to probe device children

On the gta02 we often have a child parent relationship between different
devices. The child devices can only be probed after their parant has been.
Instead of adding a probe completed handler to each device we handle this in a
generic way with a bus notifier.

Signed-off-by: Lars-Peter Clausen <>
6 years agoARM: s3c2440: GTA02: Select SPARSE_IRQ
Lars-Peter Clausen [Wed, 29 Dec 2010 01:43:10 +0000 (02:43 +0100)]
ARM: s3c2440: GTA02: Select SPARSE_IRQ

With the conversion of the pcf50633 driver to genirq we need some extra irqs.

Signed-off-by: Lars-Peter Clausen <>
6 years agos3c2440: gta02: Configure pcf50633 gpios.
Lars-Peter Clausen [Sun, 4 Apr 2010 18:49:19 +0000 (20:49 +0200)]
s3c2440: gta02: Configure pcf50633 gpios.

6 years agogta02: Remove usage of pcf50633 gpio api
Lars-Peter Clausen [Mon, 17 May 2010 17:53:21 +0000 (19:53 +0200)]
gta02: Remove usage of pcf50633 gpio api

This is the only user of the pcf50633 gpio api. Since the custom interface is
going to be replaced with gpiolib all users need to be remove or replaced.
It is safe to be remove it in this case since it is used used to turn a gpio
off which is never going to be turned on.

Signed-off-by: Lars-Peter Clausen <>
6 years agogta02: Disable hardware ECC unless we get instructed to enable it
Holger Freyther [Wed, 19 Nov 2008 17:10:51 +0000 (17:10 +0000)]
gta02: Disable hardware ECC unless we get instructed to enable it

Early verions off uboot used for the gta02 flashed the nand with ecc information
incompatible to s3c2440 hardware ecc. Disable hardware error correction by
default, unless the bootloader explicitly enables it.

Signed-off-by: Lars-Peter Clausen <>
6 years agoARM: gta02: Add LED support
Lars-Peter Clausen [Thu, 24 Sep 2009 17:20:02 +0000 (19:20 +0200)]
ARM: gta02: Add LED support

This patch adds support for the LEDs found on the gta02 device.

Signed-off-by: Lars-Peter Clausen <>
6 years agoAdd glamo driver.
Lars-Peter Clausen [Sun, 4 Oct 2009 11:47:44 +0000 (13:47 +0200)]
Add glamo driver.

6 years agommc: test: add random fault injection in core.c
Per Forlin [Wed, 6 Apr 2011 19:07:10 +0000 (21:07 +0200)]
mmc: test: add random fault injection in core.c

This simple fault injection proved to be very useful to
test the error handling in the block.c rw_rq(). It may
still be useful to test if the host driver handle
pre_req() and post_req() correctly in case of errors.

Signed-off-by: Per Forlin <>
6 years agommc: add handling for two parallel block requests in issue_rw_rq
Per Forlin [Wed, 6 Apr 2011 19:07:09 +0000 (21:07 +0200)]
mmc: add handling for two parallel block requests in issue_rw_rq

Change mmc_blk_issue_rw_rq() to become asynchronous.
The execution flow looks like this:
The mmc-queue calls issue_rw_rq(), which sends the request
to the host and returns back to the mmc-queue. The mmc-queue calls
isuue_rw_rq() again with a new request. This new request is prepared,
in isuue_rw_rq(), then it waits for the active request to complete before
pushing it to the host. When to mmc-queue is empty it will call
isuue_rw_rq() with req=NULL to finish off the active request
without starting a new request.

Signed-off-by: Per Forlin <>
6 years agommc: add a second mmc queue request member
Per Forlin [Wed, 6 Apr 2011 19:07:08 +0000 (21:07 +0200)]
mmc: add a second mmc queue request member

Add an additional mmc queue request instance to make way for
two active block requests. One request may be active while the
other request is being prepared.

Signed-off-by: Per Forlin <>
6 years agommc: move error code in mmc_block_issue_rw_rq to a separate function.
Per Forlin [Wed, 6 Apr 2011 19:07:07 +0000 (21:07 +0200)]
mmc: move error code in mmc_block_issue_rw_rq to a separate function.

Break out code without functional changes. This simplifies the code and
makes way for handle two parallel request.

Signed-off-by: Per Forlin <>
6 years agommc: add a block request prepare function
Per Forlin [Wed, 6 Apr 2011 19:07:06 +0000 (21:07 +0200)]
mmc: add a block request prepare function

Break out code from mmc_blk_issue_rw_rq to create a
block request prepare function. This doesn't change
any functionallity. This helps when handling more
than one active block request.

Signed-off-by: Per Forlin <>
6 years agommc: add member in mmc queue struct to hold request data
Per Forlin [Wed, 6 Apr 2011 19:07:05 +0000 (21:07 +0200)]
mmc: add member in mmc queue struct to hold request data

The way the request data is organized in the mmc queue struct
it only allows processing of one request at the time.
This patch adds a new struct to hold mmc queue request data such as
sg list, request, blk request and bounce buffers, and updates any functions
depending on the mmc queue struct. This lies the ground for
using multiple active request for one mmc queue.

Signed-off-by: Per Forlin <>
6 years agommc: mmc_test: add test for none blocking transfers
Per Forlin [Wed, 6 Apr 2011 19:07:04 +0000 (21:07 +0200)]
mmc: mmc_test: add test for none blocking transfers

Add four tests for read and write performance per
different transfer size, 4k to 4M.
 * Read using blocking mmc request
 * Read using none blocking mmc request
 * Write using blocking mmc request
 * Write using none blocking mmc request

The host dirver must support pre_req() and post_req()
in order to run the none blocking test cases.

Signed-off-by: Per Forlin <>
6 years agommc: mmc_test: add debugfs file to list all tests
Per Forlin [Wed, 6 Apr 2011 19:07:03 +0000 (21:07 +0200)]
mmc: mmc_test: add debugfs file to list all tests

Add a debugfs file "testlist" to print all available tests

Signed-off-by: Per Forlin <>
6 years agommc: add none blocking mmc request function
Per Forlin [Wed, 6 Apr 2011 19:07:02 +0000 (21:07 +0200)]
mmc: add none blocking mmc request function

Previously there has only been one function mmc_wait_for_req
to start and wait for a request. This patch adds
 * mmc_start_req - starts a request wihtout waiting
 * mmc_wait_for_req_done - waits until request is done
 * mmc_pre_req - asks the host driver to prepare for the next job
 * mmc_post_req - asks the host driver to clean up after a completed job

The intention is to use pre_req() and post_req() to do cache maintenance
while a request is active. pre_req() can be called while a request is active
to minimize latency to start next job. post_req() can be used after the next
job is started to clean up the request. This will minimize the host driver
request end latency. post_req() is typically used before ending the block
request and handing over the buffer to the block layer.

Add a host-private member in mmc_data to be used by
pre_req to mark the data. The host driver will then
check this mark to see if the data is prepared or not.

Signed-off-by: Per Forlin <>
6 years agoAdd c fiq handler.
Lars-Peter Clausen [Mon, 5 Oct 2009 11:39:17 +0000 (13:39 +0200)]
Add c fiq handler.

6 years agoAdd s3c24xx_serial_console_set_silence
Lars-Peter Clausen [Mon, 5 Oct 2009 12:27:40 +0000 (14:27 +0200)]
Add s3c24xx_serial_console_set_silence

6 years agomtd: s3c2410_nand: Add config option to disable hw ecc at runtime
Holger Freyther [Wed, 19 Nov 2008 17:10:51 +0000 (17:10 +0000)]
mtd: s3c2410_nand: Add config option to disable hw ecc at runtime

This patch adds a flag to the s3c2410_nand platform data, which configures
whether hardware ecc is used for that chip.

Currently hardware ecc is used if it was compiled into the kernel. But if you
want to build a kernel which runs on multiple devices you might have a
configuration where you have devices which require hw ecc as well as devices
which want software ecc.

Signed-off-by: Lars-Peter Clausen <>
6 years agos3c24xx: Fix level irqs on external interrupts.
Lars-Peter Clausen [Sat, 8 May 2010 12:32:09 +0000 (14:32 +0200)]
s3c24xx: Fix level irqs on external interrupts.

Although the external interrupts support level and edge triggered irqs their
handler is currently always set to handle_edge_irq().
While being technically wrong for a level triggered irq to be handled by
handle_edge_irq() it will cause serious problems in combination with a oneshot
irq. handle_edge_irq() will unmask the irq immediately and as a result the irq
will be triggered again before the threaded irq handler had a chance to run and
clear the irq source.

Thus level triggered irqs should be handled by handle_level_irq.
According to the specs the irq controller will remember if an irq has been
triggered while it had been masked and will issue it when the irq gets
unmasked. Thus it is sufficient to use handle_level_irq() for edge triggered
irqs as well. Hence handle_level_irq() can always be used for external

Signed-off-by: Lars-Peter Clausen <>
6 years agoARM: s3c24xx: Set ARCH_NR_GPIOS according to the selected SoC types.
Lars-Peter Clausen [Fri, 17 Sep 2010 11:48:41 +0000 (13:48 +0200)]
ARM: s3c24xx: Set ARCH_NR_GPIOS according to the selected SoC types.

Currently the code in gpiolib.c tries to register GPIO BANKA-BANKM.
ARCH_NR_GPIOS on the other hand is set only to 256, which would be the
equivalent of BANKA-BANKH. Thus the registration of all other banks will fail.

This patch fixes this by setting S3C_GPIO_END according to the selected SoC
types. S3C_GPIO_END is set to the maximum of the number of GPIOs over all
selected SoC types. Thus it is ensured that memory is not wasted if support for
SoCs with higher GPIO numbers is not built-in.  When registering the bank it is
made sure that banks which are outside of that range are not even tried to be
registered.  Otherwise there would be problems with configs where
CONFIG_S3C24XX_GPIO_EXTRA is set to a non zero value.

Signed-off-by: Lars-Peter Clausen <>
6 years agoLinux
Greg Kroah-Hartman [Wed, 3 Aug 2011 19:43:28 +0000 (12:43 -0700)]

6 years agoalpha: fix several security issues
Dan Rosenberg [Wed, 15 Jun 2011 22:09:01 +0000 (15:09 -0700)]
alpha: fix several security issues

commit 21c5977a836e399fc710ff2c5367845ed5c2527f upstream.

Fix several security issues in Alpha-specific syscalls.  Untested, but
mostly trivial.

1. Signedness issue in osf_getdomainname allows copying out-of-bounds
kernel memory to userland.

2. Signedness issue in osf_sysinfo allows copying large amounts of
kernel memory to userland.

3. Typo (?) in osf_getsysinfo bounds minimum instead of maximum copy
size, allowing copying large amounts of kernel memory to userland.

4. Usage of user pointer in osf_wait4 while under KERNEL_DS allows
privilege escalation via writing return value of sys_wait4 to kernel

Signed-off-by: Dan Rosenberg <>
Cc: Richard Henderson <>
Cc: Ivan Kokshaysky <>
Cc: Matt Turner <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoproc: restrict access to /proc/PID/io
Vasiliy Kulikov [Fri, 24 Jun 2011 12:08:38 +0000 (16:08 +0400)]
proc: restrict access to /proc/PID/io

commit 1d1221f375c94ef961ba8574ac4f85c8870ddd51 upstream.

/proc/PID/io may be used for gathering private information.  E.g.  for
openssh and vsftpd daemons wchars/rchars may be used to learn the
precise password length.  Restrict it to processes being able to ptrace
the target process.

ptrace_may_access() is needed to prevent keeping open file descriptor of
"io" file, executing setuid binary and gathering io information of the
setuid'ed process.

Signed-off-by: Vasiliy Kulikov <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agooom: task->mm == NULL doesn't mean the memory was freed
Oleg Nesterov [Sat, 30 Jul 2011 14:35:02 +0000 (16:35 +0200)]
oom: task->mm == NULL doesn't mean the memory was freed

commit c027a474a68065391c8773f6e83ed5412657e369 upstream.

exit_mm() sets ->mm == NULL then it does mmput()->exit_mmap() which
frees the memory.

However select_bad_process() checks ->mm != NULL before TIF_MEMDIE,
so it continues to kill other tasks even if we have the oom-killed
task freeing its memory.

Change select_bad_process() to check ->mm after TIF_MEMDIE, but skip
the tasks which have already passed exit_notify() to ensure a zombie
with TIF_MEMDIE set can't block oom-killer. Alternatively we could
probably clear TIF_MEMDIE after exit_mmap().

Signed-off-by: Oleg Nesterov <>
Reviewed-by: KOSAKI Motohiro <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoFS-Cache: Fix __fscache_uncache_all_inode_pages()'s outer loop
Jan Beulich [Thu, 21 Jul 2011 14:02:43 +0000 (15:02 +0100)]
FS-Cache: Fix __fscache_uncache_all_inode_pages()'s outer loop

commit b307d4655a71749ac3f91c6dbe33d28cc026ceeb upstream.

The compiler, at least for ix86 and m68k, validly warns that the

next <= (loff_t)-1

is always true (and it's always true also for x86-64 and probably all
other arches - as long as pgoff_t isn't wider than loff_t).  The
intention appears to be to avoid wrapping of "next", so rather than
eliminating the pointless comparison, fix the loop to indeed get exited
when "next" would otherwise wrap.

On m68k the following warning is observed:

  fs/fscache/page.c: In function '__fscache_uncache_all_inode_pages':
  fs/fscache/page.c:979: warning: comparison is always false due to limited range of data type

Reported-by: Geert Uytterhoeven <>
Reported-by: Jan Beulich <>
Signed-off-by: Jan Beulich <>
Signed-off-by: David Howells <>
Cc: Suresh Jayaraman <>
Cc: Geert Uytterhoeven <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agogro: Only reset frag0 when skb can be pulled
Herbert Xu [Wed, 27 Jul 2011 13:16:28 +0000 (06:16 -0700)]
gro: Only reset frag0 when skb can be pulled

commit 17dd759c67f21e34f2156abcf415e1f60605a188 upstream.

Currently skb_gro_header_slow unconditionally resets frag0 and
frag0_len.  However, when we can't pull on the skb this leaves
the GRO fields in an inconsistent state.

This patch fixes this by only resetting those fields after the
pskb_may_pull test.

Signed-off-by: Herbert Xu <>
Signed-off-by: David S. Miller <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agobridge: send proper message_age in config BPDU
stephen hemminger [Fri, 22 Jul 2011 07:47:06 +0000 (07:47 +0000)]
bridge: send proper message_age in config BPDU

commit 0c03150e7ea8f7fcd03cfef29385e0010b22ee92 upstream.

A bridge topology with three systems:

      +------+  +------+
      | A(2) |--| B(1) |
      +------+  +------+
           \    /
          | C(3) |

What is supposed to happen:
 * bridge with the lowest ID is elected root (for example: B)
 * C detects that A->C is higher cost path and puts in blocking state

What happens. Bridge with lowest id (B) is elected correctly as
root and things start out fine initially. But then config BPDU
doesn't get transmitted from A -> C. Because of that
the link from A-C is transistioned to the forwarding state.

The root cause of this is that the configuration messages
is generated with bogus message age, and dropped before

In the standardmessage_age is supposed to be:
  the time since the generation of the Configuration BPDU by
  the Root that instigated the generation of this Configuration BPDU.

Reimplement this by recording the timestamp (age + jiffies) when
recording config information. The old code incorrectly used the time
elapsed on the ageing timer which was incorrect.

See also:

Signed-off-by: Stephen Hemminger <>
Signed-off-by: David S. Miller <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agocifs: lower default and max wsize to what 2.6.39 can handle
Jeff Layton [Mon, 1 Aug 2011 19:14:16 +0000 (15:14 -0400)]
cifs: lower default and max wsize to what 2.6.39 can handle

This patch is intended for 2.6.39-stable kernels only and is needed to
fix a regression introduced in 2.6.39. Prior to 2.6.39, when signing was
enabled on a socket the client only sent single-page writes. This
changed with commit ca83ce3, which made signed and unsigned connections
use the same codepaths for write calls.

This caused a regression when working with windows servers. Windows
machines will reject writes larger than the MaxBufferSize when signing
is active, but do not clear the CAP_LARGE_WRITE_X flag in the protocol
negotiation. The upshot is that when signing is active, windows servers
often reject large writes from the client in 2.6.39.

Because 3.0 adds support for larger wsize values, simply cherry picking
the upstream patches that fix the wsize negotiation isn't sufficient to
fix this issue. We also need to alter the maximum and default values to
something suitable for 2.6.39.

This patch also accounts for the change in field name from sec_mode to
secMode that went into 3.0.

Signed-off-by: Jeff Layton <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agocifs: fix wsize negotiation to respect max buffer size and active signing (try #4)
Jeff Layton [Wed, 22 Jun 2011 21:33:57 +0000 (17:33 -0400)]
cifs: fix wsize negotiation to respect max buffer size and active signing (try #4)

commit 1190f6a067bf27b2ee7e06ec0776a17fe0f6c4d8 upstream.

Hopefully last version. Base signing check on CAP_UNIX instead of
tcon->unix_ext, also clean up the comments a bit more.

According to Hongwei Sun's blog posting here:

CAP_LARGE_WRITEX is ignored when signing is active. Also, the maximum
size for a write without CAP_LARGE_WRITEX should be the maxBuf that
the server sent in the NEGOTIATE request.

Fix the wsize negotiation to take this into account. While we're at it,
alter the other wsize definitions to use sizeof(WRITE_REQ) to allow for
slightly larger amounts of data to potentially be written per request.

Signed-off-by: Jeff Layton <>
Signed-off-by: Steve French <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agovfs: fix race in rcu lookup of pruned dentry
Linus Torvalds [Tue, 26 Jul 2011 20:04:16 +0000 (13:04 -0700)]
vfs: fix race in rcu lookup of pruned dentry

Backport of commit 59430262401bec02d415179c43dbe5b8819c09ce
done by Hugh Dickins <>

Don't update *inode in __follow_mount_rcu() until we'd verified that
there is mountpoint there.  Kudos to Hugh Dickins for catching that
one in the first place and eventually figuring out the solution (and
catching a braino in the earlier version of patch).

Signed-off-by: Linus Torvalds <>
Cc: Hugh Dickins <>
Signed-off-by: Al Viro <>
6 years agoRevert "DiBxxxx: get rid of DMA buffer on stack"
Greg Kroah-Hartman [Mon, 1 Aug 2011 18:53:11 +0000 (11:53 -0700)]
Revert "DiBxxxx: get rid of DMA buffer on stack"

This reverts commit 673de951a6cbaa7a9c4518e03e78b88809828497 which was
commit 5a0deeed5741117ee8625d6305d0034e219f102c upstream.

Based on comments from users, this broke things.

Acked-by: Olivier Grenie <>
Cc: Florian Mickler <>
Cc: Patrick Boettcher <>
Cc: Mauro Carvalho Chehab <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agomm: vmscan: only read new_classzone_idx from pgdat when reclaiming successfully
Mel Gorman [Mon, 11 Jul 2011 09:11:23 +0000 (10:11 +0100)]
mm: vmscan: only read new_classzone_idx from pgdat when reclaiming successfully

commit 215ddd6664ced067afca7eebd2d1eb83f064ff5a upstream

During allocator-intensive workloads, kswapd will be woken frequently
causing free memory to oscillate between the high and min watermark.  This
is expected behaviour.  Unfortunately, if the highest zone is small, a
problem occurs.

When balance_pgdat() returns, it may be at a lower classzone_idx than it
started because the highest zone was unreclaimable.  Before checking if it
should go to sleep though, it checks pgdat->classzone_idx which when there
is no other activity will be MAX_NR_ZONES-1.  It interprets this as it has
been woken up while reclaiming, skips scheduling and reclaims again.  As
there is no useful reclaim work to do, it enters into a loop of shrinking
slab consuming loads of CPU until the highest zone becomes reclaimable for
a long period of time.

There are two problems here.  1) If the returned classzone or order is
lower, it'll continue reclaiming without scheduling.  2) if the highest
zone was marked unreclaimable but balance_pgdat() returns immediately at
DEF_PRIORITY, the new lower classzone is not communicated back to kswapd()
for sleeping.

This patch does two things that are related.  If the end_zone is
unreclaimable, this information is communicated back.  Second, if the
classzone or order was reduced due to failing to reclaim, new information
is not read from pgdat and instead an attempt is made to go to sleep.  Due
to this, it is also necessary that pgdat->classzone_idx be initialised
each time to pgdat->nr_zones - 1 to avoid re-reads being interpreted as

Signed-off-by: Mel Gorman <>
Reported-by: Pádraig Brady <>
Tested-by: Pádraig Brady <>
Tested-by: Andrew Lutomirski <>
Acked-by: Rik van Riel <>
Cc: Minchan Kim <>
Cc: KOSAKI Motohiro <>
Cc: Johannes Weiner <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agomm: vmscan: do not apply pressure to slab if we are not applying pressure to zone
Mel Gorman [Mon, 11 Jul 2011 09:11:21 +0000 (10:11 +0100)]
mm: vmscan: do not apply pressure to slab if we are not applying pressure to zone

commit d7868dae893c83c50c7824bc2bc75f93d114669f upstream

During allocator-intensive workloads, kswapd will be woken frequently
causing free memory to oscillate between the high and min watermark.  This
is expected behaviour.

When kswapd applies pressure to zones during node balancing, it checks if
the zone is above a high+balance_gap threshold.  If it is, it does not
apply pressure but it unconditionally shrinks slab on a global basis which
is excessive.  In the event kswapd is being kept awake due to a high small
unreclaimable zone, it skips zone shrinking but still calls shrink_slab().

Once pressure has been applied, the check for zone being unreclaimable is
being made before the check is made if all_unreclaimable should be set.
This miss of unreclaimable can cause has_under_min_watermark_zone to be
set due to an unreclaimable zone preventing kswapd backing off on

Signed-off-by: Mel Gorman <>
Reported-by: Pádraig Brady <>
Tested-by: Pádraig Brady <>
Tested-by: Andrew Lutomirski <>
Acked-by: Rik van Riel <>
Reviewed-by: Minchan Kim <>
Reviewed-by: KOSAKI Motohiro <>
Cc: Johannes Weiner <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agocifs: clean up wsize negotiation and allow for larger wsize
Jeff Layton [Thu, 19 May 2011 20:22:58 +0000 (16:22 -0400)]
cifs: clean up wsize negotiation and allow for larger wsize

commit f7910cbd9fa319ee4501074f1f3b5ce23c4b1518 upstream.

Now that we can handle larger wsizes in writepages, fix up the
negotiation of the wsize to allow for that. find_get_pages only seems to
give out a max of 256 pages at a time, so that gives us a reasonable
default of 1M for the wsize.

If the server however does not support large writes via POSIX
extensions, then we cap the wsize to (128k - PAGE_CACHE_SIZE). That
gives us a size that goes up to the max frame size specified in RFC1001.

Finally, if CAP_LARGE_WRITE_AND_X isn't set, then further cap it to the
largest size allowed by the protocol (USHRT_MAX).

Signed-off-by: Jeff Layton <>
Reviewed-and-Tested-by: Pavel Shilovsky <>
Signed-off-by: Steve French <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoxfs [stable only]: restart busy extent search after node removal
Eric Sandeen [Wed, 13 Jul 2011 15:48:46 +0000 (10:48 -0500)]
xfs [stable only]: restart busy extent search after node removal

A user on #xfs reported that a log replay was oopsing in
__rb_rotate_left() with a null pointer deref, and provided
an xfs_metadump image for reproduction and testing.

I traced this down to the fact that in xfs_alloc_busy_insert(),
we erased a node with rb_erase() when the new node overlapped,
but left the erased node specified as the parent node for the
new insertion.

So when we try to insert a new node with an erased node as
its parent, obviously things go very wrong.

97d3ac75e5e0ebf7ca38ae74cebd201c09b97ab2 xfs: exact busy extent tracking
actually fixed this, but as part of a much larger change.  Here's
the relevant code from that commit:

                * We also need to restart the busy extent search from the
                * tree root, because erasing the node can rearrange the
                * tree topology.
               rb_erase(&busyp->rb_node, &pag->pagb_tree);
               busyp->length = 0;
               return false;

We can do essentially the same thing to older codebases by restarting
the tree search after the erase.

This should apply to .35.y through .39.y, and was tested on .39
with the oopsing replay reproducer.

Signed-off-by: Eric Sandeen <>
Reviewed-by: Dave Chinner <>
Reviewed-by: Alex Elder <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agomm: compaction: abort compaction if too many pages are isolated and caller is asynchr...
Mel Gorman [Tue, 19 Jul 2011 09:15:51 +0000 (10:15 +0100)]
mm: compaction: abort compaction if too many pages are isolated and caller is asynchronous V2

commit: f9e35b3b41f47c4e17d8132edbcab305a6aaa4b0 upstream

Asynchronous compaction is used when promoting to huge pages.  This is all
very nice but if there are a number of processes in compacting memory, a
large number of pages can be isolated.  An "asynchronous" process can
stall for long periods of time as a result with a user reporting that
firefox can stall for 10s of seconds.  This patch aborts asynchronous
compaction if too many pages are isolated as it's better to fail a
hugepage promotion than stall a process.

[ return COMPACT_PARTIAL for abort]
Reported-and-tested-by: Ury Stankevich <>
Signed-off-by: Mel Gorman <>
Reviewed-by: Minchan Kim <>
Reviewed-by: Michal Hocko <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agomm: vmscan: do not use page_count without a page pin
Andrea Arcangeli [Tue, 19 Jul 2011 09:15:50 +0000 (10:15 +0100)]
mm: vmscan: do not use page_count without a page pin

commit: d179e84ba5da1d0024087d1759a2938817a00f3f upstream

It is unsafe to run page_count during the physical pfn scan because
compound_head could trip on a dangling pointer when reading
page->first_page if the compound page is being freed by another CPU.

[ split out patch]
Signed-off-by: Andrea Arcangeli <>
Signed-off-by: Mel Gorman <>
Reviewed-by: Michal Hocko <>
Reviewed-by: Minchan Kim <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agomm: compaction: ensure that the compaction free scanner does not move to the next...
Mel Gorman [Tue, 19 Jul 2011 09:15:49 +0000 (10:15 +0100)]
mm: compaction: ensure that the compaction free scanner does not move to the next zone

commit: 7454f4ba40b419eb999a3c61a99da662bf1a2bb8 upstream

Compaction works with two scanners, a migration and a free scanner.  When
the scanners crossover, migration within the zone is complete.  The
location of the scanner is recorded on each cycle to avoid excesive

When a zone is small and mostly reserved, it's very easy for the migration
scanner to be close to the end of the zone.  Then the following situation
can occurs

  o migration scanner isolates some pages near the end of the zone
  o free scanner starts at the end of the zone but finds that the
    migration scanner is already there
  o free scanner gets reinitialised for the next cycle as
    cc->migrate_pfn + pageblock_nr_pages
    moving the free scanner into the next zone
  o migration scanner moves into the next zone

When this happens, NR_ISOLATED accounting goes haywire because some of the
accounting happens against the wrong zone.  One zones counter remains
positive while the other goes negative even though the overall global
count is accurate.  This was reported on X86-32 with !SMP because !SMP
allows the negative counters to be visible.  The fact that it is the bug
should theoritically be possible there.

Signed-off-by: Mel Gorman <>
Reviewed-by: Minchan Kim <>
Reviewed-by: Michal Hocko <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoext4: fix oops in ext4_quota_off()
Amir Goldstein [Mon, 16 May 2011 13:59:13 +0000 (09:59 -0400)]
ext4: fix oops in ext4_quota_off()

commit 0b26859027ce0005ef89520af20351360e51ad76 upstream.

If quota is not enabled when ext4_quota_off() is called, we must not
dereference quota file inode since it is NULL.  Check properly for

This fixes a bug in commit 21f976975cbe (ext4: remove unnecessary
[cm]time update of quota file), which was merged for 2.6.39-rc3.

Reported-by: Amir Goldstein <>
Signed-off-by: Amir Goldstein <>
Signed-off-by: Jan Kara <>
Signed-off-by: "Theodore Ts'o" <>
Cc: Chris Dunlop <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agolibata: fix unexpectedly frozen port after ata_eh_reset()
Tejun Heo [Wed, 25 May 2011 11:19:39 +0000 (13:19 +0200)]
libata: fix unexpectedly frozen port after ata_eh_reset()

commit 8c56cacc724c7650b893d43068fa66044aa29a61 upstream.

To work around controllers which can't properly plug events while
reset, ata_eh_reset() clears error states and ATA_PFLAG_EH_PENDING
after reset but before RESET is marked done.  As reset is the final
recovery action and full verification of devices including onlineness
and classfication match is done afterwards, this shouldn't lead to
lost devices or missed hotplug events.

Unfortunately, it forgot to thaw the port when clearing EH_PENDING, so
if the condition happens after resetting an empty port, the port could
be left frozen and EH will end without thawing it, making the port
unresponsive to further hotplug events.

Thaw if the port is frozen after clearing EH_PENDING.  This problem is
reported by Bruce Stenning in the following thread.

stable: I think we should weather this patch a bit longer in -rcX
before sending it to -stable.  Please wait at least a month
after this patch makes upstream.  Thanks.

-v2: Fixed spelling in the comment per Dave Howorth.

Signed-off-by: Tejun Heo <>
Reported-by: Bruce Stenning <>
Cc: Dave Howorth <>
Signed-off-by: Jeff Garzik <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agovmscan: fix a livelock in kswapd
Shaohua Li [Tue, 19 Jul 2011 15:49:26 +0000 (08:49 -0700)]
vmscan: fix a livelock in kswapd

commit 4746efded84d7c5a9c8d64d4c6e814ff0cf9fb42 upstream.

I'm running a workload which triggers a lot of swap in a machine with 4
nodes.  After I kill the workload, I found a kswapd livelock.  Sometimes
kswapd3 or kswapd2 are keeping running and I can't access filesystem,
but most memory is free.

This looks like a regression since commit 08951e545918c159 ("mm: vmscan:
correct check for kswapd sleeping in sleeping_prematurely").

Node 2 and 3 have only ZONE_NORMAL, but balance_pgdat() will return 0
for classzone_idx.  The reason is end_zone in balance_pgdat() is 0 by
default, if all zones have watermark ok, end_zone will keep 0.

Later sleeping_prematurely() always returns true.  Because this is an
order 3 wakeup, and if classzone_idx is 0, both balanced_pages and
present_pages in pgdat_balanced() are 0.  We add a special case here.
If a zone has no page, we think it's balanced.  This fixes the livelock.

Signed-off-by: Shaohua Li <>
Acked-by: Mel Gorman <>
Cc: Minchan Kim <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agodavinci: DM365 EVM: fix video input mux bits
Jon Povey [Tue, 19 Jul 2011 03:30:11 +0000 (12:30 +0900)]
davinci: DM365 EVM: fix video input mux bits

commit 9daedd833a38edd90cf7baa1b1fcf61c3a0721e3 upstream.

Video input mux settings for tvp7002 and imager inputs were swapped.
Comment was correct.

Tested on EVM with tvp7002 input.

Signed-off-by: Jon Povey <>
Acked-by: Manjunath Hadli <>
Signed-off-by: Sekhar Nori <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agox86: Make Dell Latitude E5420 use reboot=pci
Daniel J Blueman [Fri, 13 May 2011 01:04:59 +0000 (09:04 +0800)]
x86: Make Dell Latitude E5420 use reboot=pci

commit b7798d28ec15d20fd34b70fa57eb13f0cf6d1ecd upstream.

Rebooting on the Dell E5420 often hangs with the keyboard or ACPI
methods, but is reliable via the PCI method.

[ hpa: this was deferred because we believed for a long time that the
  recent reshuffling of the boot priorities in commit
  660e34cebf0a11d54f2d5dd8838607452355f321 fixed this platform.
  Unfortunately that turned out to be incorrect. ]

Signed-off-by: Daniel J Blueman <>
Signed-off-by: H. Peter Anvin <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agommc: sdhci-esdhc-imx: SDHCI_CARD_PRESENT does not get cleared
Shawn Guo [Tue, 21 Jun 2011 14:41:49 +0000 (22:41 +0800)]
mmc: sdhci-esdhc-imx: SDHCI_CARD_PRESENT does not get cleared

commit 803862a6f7de4939e0a557214e5e4b37e36f87ff upstream.

The function esdhc_readl_le intends to clear bit SDHCI_CARD_PRESENT,
when the card detect gpio tells there is no card.  But it does not
clear the bit actually.  The patch gives a fix on that.

Signed-off-by: Shawn Guo <>
Acked-by: Wolfram Sang <>
Signed-off-by: Chris Ball <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agommc: Added quirks for Ricoh 1180:e823 lower base clock frequency
Manoj Iyer [Mon, 11 Jul 2011 21:28:35 +0000 (16:28 -0500)]
mmc: Added quirks for Ricoh 1180:e823 lower base clock frequency

commit 15bed0f2fa8e1d7db201692532c210a7823d2d21 upstream.

Ricoh 1180:e823 does not recognize certain types of SD/MMC cards,
as reported at  Lowering the SD
base clock frequency from 200Mhz to 50Mhz fixes this issue. This
solution was suggest by Koji Matsumuro, Ricoh Company, Ltd.

This change has no negative performance effect on standard SD
cards, though it's quite possible that there will be one on
UHS-1 cards.

Signed-off-by: Manoj Iyer <>
Tested-by: Daniel Manrique <>
Cc: Koji Matsumuro <>
Acked-by: Jesse Barnes <>
Signed-off-by: Chris Ball <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agohwmon: (max1111) Fix race condition causing NULL pointer exception
Pavel Herrmann [Sun, 17 Jul 2011 16:39:19 +0000 (18:39 +0200)]
hwmon: (max1111) Fix race condition causing NULL pointer exception

commit d3f684f2820a7f42acef68bea6622d9032127fb2 upstream.

spi_sync call uses its spi_message parameter to keep completion information,
using a drvdata structure is not thread-safe. Use a mutex to prevent
multiple access to shared driver data.

Signed-off-by: Pavel Herrmann <>
Acked-by: Russell King <>
Acked-by: Pavel Machek <>
Acked-by: Marek Vasut <>
Acked-by: Cyril Hrubis <>
Tested-by: Stanislav Brabec <>
Signed-off-by: Jean Delvare <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agohwmon: (asus_atk0110) Fix memory leak
Luca Tettamanti [Sun, 17 Jul 2011 16:39:18 +0000 (18:39 +0200)]
hwmon: (asus_atk0110) Fix memory leak

commit 0b8e77f12cb6bfe2e5a67f2cdc8c7af23abc4ccf upstream.

The object returned by atk_gitm is dynamically allocated and must be

Signed-off-by: Luca Tettamanti <>
Signed-off-by: Jean Delvare <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agohwmon: (it87) Fix label group removal
Jean Delvare [Sun, 17 Jul 2011 16:39:19 +0000 (18:39 +0200)]
hwmon: (it87) Fix label group removal

commit fa8b69758e65b406c8010936b541cd00deef804d upstream.

A copy-and-paste error caused it87_attributes_vid to be referenced
where it87_attributes_label should be. Thankfully the group is only
used for attribute removal, not attribute creation, so the effects of
this bug are limited, but let's fix it still.

Signed-off-by: Jean Delvare <>
Acked-by: Guenter Roeck <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoASoC: Fix shift in WM8958 accessory detection default implementation
Mark Brown [Fri, 15 Jul 2011 13:28:32 +0000 (22:28 +0900)]
ASoC: Fix shift in WM8958 accessory detection default implementation

commit b35e160a111aa9ae3fad6294e038be20d0da721b upstream.

Signed-off-by: Mark Brown <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoASoC: Correct WM8994 MICBIAS supply widget hookup
Mark Brown [Thu, 14 Jul 2011 09:21:37 +0000 (18:21 +0900)]
ASoC: Correct WM8994 MICBIAS supply widget hookup

commit b793eb60a01d5b5e4aaeb2fbc2b036dec0d9f84d upstream.

The WM8994 and WM8958 series of devices have two MICBIAS supplies rather
than one, the current widget actually manages the microphone detection
control register bit (which is managed separately by the relevant API).

Fix this, hooking the relevant supplies up to the MICBIAS1 and MICBIAS2

Signed-off-by: Mark Brown <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agosi4713-i2c: avoid potential buffer overflow on si4713
Mauro Carvalho Chehab [Sun, 17 Jul 2011 03:24:37 +0000 (00:24 -0300)]
si4713-i2c: avoid potential buffer overflow on si4713

commit dc6b845044ccb7e9e6f3b7e71bd179b3cf0223b6 upstream.

While compiling it with Fedora 15, I noticed this issue:

  inlined from ‘si4713_write_econtrol_string’ at drivers/media/radio/si4713-i2c.c:1065:24:
  arch/x86/include/asm/uaccess_32.h:211:26: error: call to ‘copy_from_user_overflow’ declared with attribute error: copy_from_user() buffer size is not provably correct

Signed-off-by: Mauro Carvalho Chehab <>
Acked-by: Sakari Ailus <>
Acked-by: Eduardo Valentin <>
Reviewed-by: Eugene Teo <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoRevert "V4L/DVB: cx23885: Enable Message Signaled Interrupts(MSI)"
Jarod Wilson [Wed, 29 Jun 2011 21:36:19 +0000 (18:36 -0300)]
Revert "V4L/DVB: cx23885: Enable Message Signaled Interrupts(MSI)"

commit d7515b8866b60c9526d2d7af37ebcd16c3c5ed97 upstream.

This reverts commit e38030f3ff02684eb9e25e983a03ad318a10a2ea.

MSI flat-out doesn't work right on cx2388x devices yet. There are now
multiple reports of cards that hard-lock systems when MSI is enabled,
including my own HVR-1250 when trying to use its built-in IR receiver.
Disable MSI and it works just fine. Similar for another user's HVR-1270.
Issues have also been reported with the HVR-1850 when MSI is enabled,
and the 1850 behavior sounds similar to an as-yet-undiagnosed issue I've
seen with an 1800.

CC: Steven Toth <>
CC: Kusanagi Kouichi <>
Signed-off-by: Jarod Wilson <>
Acked-by: Andy Walls <>
Signed-off-by: Mauro Carvalho Chehab <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoGFS2: make sure fallocate bytes is a multiple of blksize
Benjamin Marzinski [Tue, 26 Apr 2011 06:13:24 +0000 (01:13 -0500)]
GFS2: make sure fallocate bytes is a multiple of blksize

commit 6905d9e4dda6112f007e9090bca80507da158e63 upstream.

The GFS2 fallocate code chooses a target size to for allocating chunks of
space.  Whenever it can't find any resource groups with enough space free, it
halves its target. Since this target is in bytes, eventually it will no longer
be a multiple of blksize.  As long as there is more space available in the
resource group than the target, this isn't a problem, since gfs2 will use the
actual space available, which is always a multiple of blksize.  However,
when gfs couldn't fallocate a bigger chunk than the target, it was using the
non-blksize aligned number. This caused a BUG in later code that required
blksize aligned offsets.  GFS2 now ensures that bytes is always a multiple of

Signed-off-by: Benjamin Marzinski <>
Signed-off-by: Steven Whitehouse <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agodrm/radeon: fix oops in ttm reserve when pageflipping (v2)
Dave Airlie [Sun, 29 May 2011 07:48:32 +0000 (17:48 +1000)]
drm/radeon: fix oops in ttm reserve when pageflipping (v2)

commit 498c555f56a02ec1059bc150cde84411ba0ac010 upstream.

We need to take a reference to this object, pinning doesn't take a reference
so if userspace deletes the object it can disappear even if pinned.

v2: fix error paths to unreference properly also.

should fix:

Acked-By: Alex Deucher <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agodrm/radeon/kms: clean up multiple crtc handling for evergreen+ (v2)
Alex Deucher [Fri, 8 Jul 2011 15:44:56 +0000 (11:44 -0400)]
drm/radeon/kms: clean up multiple crtc handling for evergreen+ (v2)

commit b7eff394670366a42935bfbaef67a6f7185627d7 upstream.

evergreen+ asics have 2-6 crtcs.  Don't access crtc registers
for crtc regs that don't exist as they have very high latency
and may cause problems on some asics.  The previous code missed
a few cases and was not fine grained enough (missed the 4 crtc
case for example).


v2: fix typo noticed by Chris Bandy <>

Signed-off-by: Alex Deucher <>
Reviewed-by: Michel Dänzer <>
Tested-by: Simon Farnsworth <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoath5k: fix incorrect use of drvdata in PCI suspend/resume code
Pavel Roskin [Wed, 29 Jun 2011 19:39:43 +0000 (15:39 -0400)]
ath5k: fix incorrect use of drvdata in PCI suspend/resume code

commit 37000b305bff81bb1ee2f7f37b1319b670a08f76 upstream.

Signed-off-by: Pavel Roskin <>
Signed-off-by: John W. Linville <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoath5k: fix incorrect use of drvdata in sysfs code
Pavel Roskin [Wed, 29 Jun 2011 19:39:37 +0000 (15:39 -0400)]
ath5k: fix incorrect use of drvdata in sysfs code

commit 95acbd432b4c6498c5b4b2f92e0e05e3c032d4f8 upstream.

Signed-off-by: Pavel Roskin <>
Signed-off-by: John W. Linville <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoath9k: Fix tx throughput drops for AR9003 chips with AES encryption
Rajkumar Manoharan [Fri, 1 Jul 2011 13:07:33 +0000 (18:37 +0530)]
ath9k: Fix tx throughput drops for AR9003 chips with AES encryption

commit 4f6760b01bda625e9555e16d8e9ba8126a9c9498 upstream.

While sending aggregated frames in AES, the AR5416 chips
required additional padding b/w subframes. This workaround
is not needed for edma (AR9003 family) chips. With this patch
~4Mbps thoughput improvement was observed in clear environment.

Signed-off-by: Rajkumar Manoharan <>
Signed-off-by: John W. Linville <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agodrm/radeon/kms: add new NI pci ids
Alex Deucher [Fri, 15 Jul 2011 14:39:10 +0000 (14:39 +0000)]
drm/radeon/kms: add new NI pci ids

commit e3c1620434ac77b618ce74c024ace3559602ac99 upstream.

Signed-off-by: Alex Deucher <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoSUNRPC: Fix use of static variable in rpcb_getport_async
Ben Greear [Tue, 12 Jul 2011 17:27:55 +0000 (10:27 -0700)]
SUNRPC: Fix use of static variable in rpcb_getport_async

commit ec0dd267bf7d08cb30e321e45a75fd40edd7e528 upstream.

Because struct rpcbind_args *map was declared static, if two
threads entered this method at the same time, the values
assigned to map could be sent two two differen tasks.
This could cause all sorts of problems, include use-after-free
and double-free of memory.

Fix this by removing the static declaration so that the map
pointer is on the stack.

Signed-off-by: Ben Greear <>
Signed-off-by: Trond Myklebust <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoSUNRPC: Fix a race between work-queue and rpc_killall_tasks
Trond Myklebust [Wed, 6 Jul 2011 23:58:23 +0000 (19:58 -0400)]
SUNRPC: Fix a race between work-queue and rpc_killall_tasks

commit b55c59892e1f3b6c7d4b9ccffb4263e1486fb990 upstream.

Since rpc_killall_tasks may modify the rpc_task's tk_action field
without any locking, we need to be careful when dereferencing it.

Reported-by: Ben Greear <>
Tested-by: Ben Greear <>
Signed-off-by: Trond Myklebust <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agortlwifi: rtl8192cu: Add new USB ID for Netgear WNA1000M
Yoann DI-RUZZA [Fri, 1 Jul 2011 13:47:17 +0000 (08:47 -0500)]
rtlwifi: rtl8192cu: Add new USB ID for Netgear WNA1000M

commit 35cbcbc6f39da30c39bc0a1e679ec44506c4eb3d upstream.

Signed-off-by: Yoann DI-RUZZA <>
Signed-off-by: Larry Finger <>
Signed-off-by: John W. Linville <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agocarl9170: add NEC WL300NU-AG usbid
Christian Lamparter [Thu, 30 Jun 2011 19:06:17 +0000 (21:06 +0200)]
carl9170: add NEC WL300NU-AG usbid

commit 06a86ddbf557cb8a0f7ded54e872e9d456002d52 upstream.

Reported-by: Mark Davis
Signed-off-by: Christian Lamparter <>
Signed-off-by: John W. Linville <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agomac80211: fix TKIP replay vulnerability
Johannes Berg [Thu, 7 Jul 2011 16:24:54 +0000 (18:24 +0200)]
mac80211: fix TKIP replay vulnerability

commit 34459512ffa7236c849466e3bd604801389734e1 upstream.

Unlike CCMP, the presence or absence of the QoS
field doesn't change the encryption, only the
TID is used. When no QoS field is present, zero
is used as the TID value. This means that it is
possible for an attacker to take a QoS packet
with TID 0 and replay it as a non-QoS packet.

Unfortunately, mac80211 uses different IVs for
checking the validity of the packet's TKIP IV
when it checks TID 0 and when it checks non-QoS
packets. This means it is vulnerable to this
replay attack.

To fix this, use the same replay counter for
TID 0 and non-QoS packets by overriding the
rx->queue value to 0 if it is 16 (non-QoS).

This is a minimal fix for now. I caused this
issue in

commit 1411f9b531f0a910cd1c85a337737c1e6ffbae6a
Author: Johannes Berg <>
Date:   Thu Jul 10 10:11:02 2008 +0200

    mac80211: fix RX sequence number check

while fixing a sequence number issue (there,
a separate counter needs to be used).

Signed-off-by: Johannes Berg <>
Signed-off-by: John W. Linville <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoNFSv4.1: update nfs4_fattr_bitmap_maxsz
Andy Adamson [Mon, 11 Jul 2011 21:17:42 +0000 (17:17 -0400)]
NFSv4.1: update nfs4_fattr_bitmap_maxsz

commit e5012d1f3861d18c7f3814e757c1c3ab3741dbcd upstream.

Attribute IDs assigned in RFC 5661 now require three bitmaps.
Fixes hitting a BUG_ON in xdr_shrink_bufhead when getting ACLs.

Signed-off-by: Andy Adamson <>
Signed-off-by: Trond Myklebust <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agodrm/radeon/kms: fix backend map typo on juniper
Alex Deucher [Mon, 11 Jul 2011 19:45:32 +0000 (19:45 +0000)]
drm/radeon/kms: fix backend map typo on juniper

commit 9a4a0b9c9fa6ad9902f49801d7b770462b69c4d3 upstream.

Don't enable backends that don't exist.

Signed-off-by: Alex Deucher <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agofs/partitions/efi.c: corrupted GUID partition tables can cause kernel oops
Timo Warns [Thu, 26 May 2011 23:25:57 +0000 (16:25 -0700)]
fs/partitions/efi.c: corrupted GUID partition tables can cause kernel oops

commit 3eb8e74ec72736b9b9d728bad30484ec89c91dde upstream.

The kernel automatically evaluates partition tables of storage devices.
The code for evaluating GUID partitions (in fs/partitions/efi.c) contains
a bug that causes a kernel oops on certain corrupted GUID partition

This bug has security impacts, because it allows, for example, to
prepare a storage device that crashes a kernel subsystem upon connecting
the device (e.g., a "USB Stick of (Partial) Death").

crc = efi_crc32((const unsigned char *) (*gpt), le32_to_cpu((*gpt)->header_size));

computes a CRC32 checksum over gpt covering (*gpt)->header_size bytes.
There is no validation of (*gpt)->header_size before the efi_crc32 call.

A corrupted partition table may have large values for (*gpt)->header_size.
 In this case, the CRC32 computation access memory beyond the memory
allocated for gpt, which may cause a kernel heap overflow.

Validate value of GUID partition table header size.

[ fix layout and indenting]
Signed-off-by: Timo Warns <>
Cc: Matt Domsch <>
Cc: Eugene Teo <>
Cc: Dave Jones <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Cc: Moritz Muehlenhoff <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agobttv: fix s_tuner for radio
Hans Verkuil [Sun, 12 Jun 2011 10:02:43 +0000 (07:02 -0300)]
bttv: fix s_tuner for radio

commit a024c1a6b274e11596d124619e43c25560f64c01 upstream.

Fix typo: g_tuner should have been s_tuner.

Tested with a bttv card.

Signed-off-by: Hans Verkuil <>
Signed-off-by: Mauro Carvalho Chehab <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agopvrusb2: fix g/s_tuner support
Hans Verkuil [Sun, 12 Jun 2011 09:39:52 +0000 (06:39 -0300)]
pvrusb2: fix g/s_tuner support

commit 50e9efd60b213ce43ad6979bfc18e25eec2d8413 upstream.

The tuner-core subdev requires that the type field of v4l2_tuner is
filled in correctly. This is done in v4l2-ioctl.c, but pvrusb2 doesn't
use that yet, so we have to do it manually based on whether the current
input is radio or not.

Tested with my pvrusb2.

Signed-off-by: Hans Verkuil <>
Acked-by: Mike Isely <>
Signed-off-by: Mauro Carvalho Chehab <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agohwmon: (pmbus) Improve auto-detection of temperature status register
Guenter Roeck [Sun, 3 Jul 2011 20:08:03 +0000 (13:08 -0700)]
hwmon: (pmbus) Improve auto-detection of temperature status register

commit 22e6b2312d221376dcd98a8afb0f314b62d33a91 upstream.

It is possible that a PMBus device supports the READ_TEMPERATURE2 and/or
READ_TEMPERATURE3 registers but does not support READ_TEMPERATURE1.
Improve temperature status register detection to address this condition.

Reported-by: Jean Delvare <>
Signed-off-by: Guenter Roeck <>
Acked-by: Jean Delvare <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agotuner-core: fix tuner_resume: use t->mode instead of t->type
Hans Verkuil [Mon, 13 Jun 2011 12:21:56 +0000 (09:21 -0300)]
tuner-core: fix tuner_resume: use t->mode instead of t->type

commit 9bf0ef060ebae452c07cf5b0616247780740bb50 upstream.

set_mode is called with t->type, which is the tuner type. Instead, use
t->mode which is the actual tuner mode (i.e. radio vs tv).

Signed-off-by: Hans Verkuil <>
Signed-off-by: Mauro Carvalho Chehab <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agotuner-core: fix s_std and s_tuner
Hans Verkuil [Tue, 14 Jun 2011 06:56:09 +0000 (03:56 -0300)]
tuner-core: fix s_std and s_tuner

commit 4e4a31fb95d88518180517bae3098a23ebde9f9c upstream.

Both s_std and s_tuner are broken because set_mode_freq is called before the
new std (for s_std) and audmode (for s_tuner) are set.

This patch splits set_mode_freq in a set_mode and a set_freq and in s_std/s_tuner
first calls set_mode, and if that returns 0 (i.e. the mode is supported)
then they set t->std/t->audmode and call set_freq.

This fixes a bug where changing std or audmode would actually change it to
the previous value.

Discovered while testing analog TV standards for cx18 with a tda18271 tuner.

Signed-off-by: Hans Verkuil <>
Signed-off-by: Mauro Carvalho Chehab <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agov4l2-ioctl.c: prefill tuner type for g_frequency and g/s_tuner
Hans Verkuil [Sun, 12 Jun 2011 09:36:41 +0000 (06:36 -0300)]
v4l2-ioctl.c: prefill tuner type for g_frequency and g/s_tuner

commit 227690df75382e46a4f6ea1bbc5df855a674b47f upstream.

The subdevs are supposed to receive a valid tuner type for the g_frequency
and g/s_tuner subdev ops. Some drivers do this, others don't. So prefill
this in v4l2-ioctl.c based on whether the device node from which this is
called is a radio node or not.

The spec does not require applications to fill in the type, and if they
leave it at 0 then the 'check_mode' call in tuner-core.c will return
an error and the ioctl does nothing.

Signed-off-by: Hans Verkuil <>
Signed-off-by: Mauro Carvalho Chehab <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoRevert: "dell-laptop: Toggle the unsupported hardware killswitch"
Keng-Yu Lin [Mon, 27 Jun 2011 10:19:03 +0000 (11:19 +0100)]
Revert: "dell-laptop: Toggle the unsupported hardware killswitch"

commit be65dde82a4a402e9607c2f306f343bf0912623c upstream.

This reverts commit a3d77411e8b2ad661958c1fbee65beb476ec6d70,

as it causes a mess in the wireless rfkill status on some models.
It is probably a bad idea to toggle the rfkill for all dell models
without the respect to the claim that it is hardware-controlled.

Signed-off-by: Keng-Yu Lin <>
Signed-off-by: Matthew Garrett <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agofirewire: ohci: do not bind to Pinnacle cards, avert panic
Stefan Richter [Sat, 9 Jul 2011 22:23:03 +0000 (00:23 +0200)]
firewire: ohci: do not bind to Pinnacle cards, avert panic

commit 7f7e37115a8b6724f26d0637a04e1d35e3c59717 upstream.

When firewire-ohci is bound to a Pinnacle MovieBoard, eventually a
"Register access failure" is logged and an interrupt storm or a kernel
panic happens.

Until this is sorted out (if that is going to succeed at all), let's
just prevent firewire-ohci from touching these devices.

Signed-off-by: Stefan Richter <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoARM: pxa: fix PGSR register address calculation
Paul Parsons [Sun, 8 May 2011 01:54:33 +0000 (01:54 +0000)]
ARM: pxa: fix PGSR register address calculation

commit beb0c9b056b1c23d2029b46a425362e9ccbeba01 upstream.

The file mfp-pxa2xx.c defines a macro, PGSR(), which translates a gpio
bank number to a PGSR register address. The function pxa2xx_mfp_suspend()
erroneously passed in a gpio number instead of a gpio bank number.

Signed-off-by: Paul Parsons <>
Signed-off-by: Eric Miao <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoARM: pxa910: correct nand pmu setting
Lei Wen [Tue, 21 Jun 2011 09:54:18 +0000 (02:54 -0700)]
ARM: pxa910: correct nand pmu setting

commit d204b2c5b16df935fa9a546c528e168859fddcc0 upstream.

The original pair of <0x01db, 208000000> is invalid.
Correct to the valid value.

Signed-off-by: Lei Wen <>
Signed-off-by: Eric Miao <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoARM: pxa168: correct nand pmu setting
Lei Wen [Tue, 21 Jun 2011 12:37:47 +0000 (05:37 -0700)]
ARM: pxa168: correct nand pmu setting

commit 6662498e132dfa758925a160fd5ef80a083651c3 upstream.

The original pair of <0x01db, 208000000> is invalid. Correct it to
the valid value.

The 6th bit of the NFC APMU register indicates NFC works whether
at 156Mhz or 78Mhz. So 0x19b indicates NFC works at 156Mhz, and
0x1db indicates it works at 78Mhz.

Signed-off-by: Lei Wen <>
Signed-off-by: Eric Miao <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoARM: 6989/1: perf: do not start the PMU when no events are present
Will Deacon [Fri, 1 Jul 2011 13:38:12 +0000 (14:38 +0100)]
ARM: 6989/1: perf: do not start the PMU when no events are present

commit f4f38430c94c38187db73a2cf3892cc8b12a2713 upstream.

armpmu_enable can be called in situations where no events are present
(for example, from the event rotation tick after a profiled task has
exited). In this case, we currently start the PMU anyway which may
leave it active inevitably without any events being monitored.

This patch adds a simple check to the enabling code so that we avoid
starting the PMU when no events are present.

Reported-by: Ashwin Chaugle <>
Signed-off-by: Will Deacon <>
Signed-off-by: Russell King <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoASoC: ak4642: fixup snd_soc_update_bits mask for PW_MGMT2
Kuninori Morimoto [Thu, 7 Jul 2011 00:58:56 +0000 (17:58 -0700)]
ASoC: ak4642: fixup snd_soc_update_bits mask for PW_MGMT2

commit bd7fdbcaa2d06d446577fd3c9b81847b04469e01 upstream.

mask didn't cover update-data

Signed-off-by: Kuninori Morimoto <>
Acked-by: Liam Girdwood <>
Signed-off-by: Mark Brown <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoASoC: Ensure we delay long enough for WM8994 FLL to lock when starting
Mark Brown [Sat, 2 Jul 2011 00:24:46 +0000 (17:24 -0700)]
ASoC: Ensure we delay long enough for WM8994 FLL to lock when starting

commit 8e9ddf811ba021506d2316fcfe619faa0ab3f567 upstream.

This delay is very conservative.

Signed-off-by: Mark Brown <>
Acked-by: Liam Girdwood <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoASoC: Fix Blackfin I2S _pointer() implementation return in bounds values
Mark Brown [Mon, 13 Jun 2011 11:14:07 +0000 (12:14 +0100)]
ASoC: Fix Blackfin I2S _pointer() implementation return in bounds values

commit e999dc50404d401150a5429b6459473a691fd1a0 upstream.

The Blackfin DMA controller can report one frame beyond the end of the
buffer in the wraparound case but ALSA requires that the pointer always
be in the buffer. Do the wraparound to handle this. A similar bug is
likely to apply to the other Blackfin PCM drivers but the code is less
obvious to inspection and I don't have a user to test.

Reported-by: Kieran O'Leary <Kieran.O'>
Acked-by: Liam Girdwood <>
Signed-off-by: Mark Brown <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agohwmon: (pmbus) Use long variables for register to data conversions
Guenter Roeck [Mon, 11 Jul 2011 02:31:29 +0000 (19:31 -0700)]
hwmon: (pmbus) Use long variables for register to data conversions

commit f450c1504d3ff6aace3927c805bdea0ac0669efd upstream.

Using integer variable types for register to data conversions can cause
overflows especially for power calculations, which are in microwatt.
Use long variables instead.

Signed-off-by: Guenter Roeck <>
Acked-by: Jean Delvare <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agohwmon: (pmbus) Improve fan detection
Guenter Roeck [Thu, 30 Jun 2011 13:54:05 +0000 (06:54 -0700)]
hwmon: (pmbus) Improve fan detection

commit 81ae68142a0483b3791cba99eab268859858f508 upstream.

Some PMBus devices return no error when reading fan speed registers, but don't
really support fans. Strengthen fan detection by also checking if fan
configuration registers exist.

Signed-off-by: Guenter Roeck <>
Reviewed-by: Robert Coulson <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agohwmon: (pmbus) Drop check for PMBus revision register in probe function
Guenter Roeck [Sat, 25 Jun 2011 19:42:14 +0000 (12:42 -0700)]
hwmon: (pmbus) Drop check for PMBus revision register in probe function

commit 7f2c0662e5cfd05f12ca49109e8f787bf2d87b66 upstream.

Some PMBus devices do not support the PMBus revision register, so don't check
if it exists.

Signed-off-by: Guenter Roeck <>
Reviewed-by: Robert Coulson <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agohwmon: (pmbus) Auto-detect temp2 and temp3 registers/attributes
Guenter Roeck [Thu, 30 Jun 2011 13:57:41 +0000 (06:57 -0700)]
hwmon: (pmbus) Auto-detect temp2 and temp3 registers/attributes

commit 0e502ec889d33bfcb348e420d7e105bc61c45eb4 upstream.

Additional temperature attribute support is easy to detect, so do it.

Signed-off-by: Guenter Roeck <>
Reviewed-by: Robert Coulson <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agohwmon: (lm95241) Fix negative temperature results
Guenter Roeck [Thu, 30 Jun 2011 09:09:37 +0000 (02:09 -0700)]
hwmon: (lm95241) Fix negative temperature results

commit 0c2a40e2fe4f4af0410f57e84b95b817ec15aa70 upstream.

Negative temperatures were returned in degrees C instead of milli-Degrees C.
Also, negative temperatures were reported for remote temperature sensors even
if the chip was configured for positive-only results.

Fix by detecting temperature modes, and by treating negative temperatures
similar to positive temperatures, with appropriate sign extension.

Signed-off-by: Guenter Roeck <>
Acked-by: Jean Delvare <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agohwmon: (lm95241) Fix chip detection code
Guenter Roeck [Mon, 27 Jun 2011 18:22:46 +0000 (11:22 -0700)]
hwmon: (lm95241) Fix chip detection code

commit 27739e694a3c34b5e371575d74e500d60111c689 upstream.

The LM95241 driver accepts every chip ID equal to or larger than 0xA4 as its
own, and other chips such as LM95245 use chip IDs in the accepted ID range.
This results in false chip detection.

Fix problem by accepting only the known LM95241 chip ID.

Signed-off-by: Guenter Roeck <>
Acked-by: Jean Delvare <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agomm: vmscan: evaluate the watermarks against the correct classzone
Mel Gorman [Fri, 8 Jul 2011 22:39:39 +0000 (15:39 -0700)]
mm: vmscan: evaluate the watermarks against the correct classzone

commit da175d06b437093f93109ba9e5efbe44dfdf9409 upstream.

When deciding if kswapd is sleeping prematurely, the classzone is taken
into account but this is different to what balance_pgdat() and the
allocator are doing.  Specifically, the DMA zone will be checked based on
the classzone used when waking kswapd which could be for a GFP_KERNEL or
GFP_HIGHMEM request.  The lowmem reserve limit kicks in, the watermark is
not met and kswapd thinks it's sleeping prematurely keeping kswapd awake in

Signed-off-by: Mel Gorman <>
Reported-by: Pádraig Brady <>
Tested-by: Pádraig Brady <>
Tested-by: Andrew Lutomirski <>
Acked-by: Rik van Riel <>
Reviewed-by: Minchan Kim <>
Cc: KOSAKI Motohiro <>
Cc: Johannes Weiner <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agomm/nommu.c: fix remap_pfn_range()
Bob Liu [Fri, 8 Jul 2011 22:39:46 +0000 (15:39 -0700)]
mm/nommu.c: fix remap_pfn_range()

commit 8f3b1327aa454bc8283e96bca7669c3c88b83f79 upstream.

remap_pfn_range() means map physical address pfn<<PAGE_SHIFT to user addr.

For nommu arch it's implemented by vma->vm_start = pfn << PAGE_SHIFT which
is wrong acroding the original meaning of this function.  And some driver
developer using remap_pfn_range() with correct parameter will get
unexpected result because vm_start is changed.  It should be implementd
like addr = pfn << PAGE_SHIFT but which is meanless on nommu arch, this
patch just make it simply return.

Parameter name and setting of vma->vm_flags also be fixed.

Signed-off-by: Bob Liu <>
Cc: Geert Uytterhoeven <>
Cc: David Howells <>
Acked-by: Greg Ungerer <>
Cc: Mike Frysinger <>
Cc: Bob Liu <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agomm: vmscan: correct check for kswapd sleeping in sleeping_prematurely
Mel Gorman [Fri, 8 Jul 2011 22:39:36 +0000 (15:39 -0700)]
mm: vmscan: correct check for kswapd sleeping in sleeping_prematurely

commit 08951e545918c1594434d000d88a7793e2452a9b upstream.

During allocator-intensive workloads, kswapd will be woken frequently
causing free memory to oscillate between the high and min watermark.  This
is expected behaviour.  Unfortunately, if the highest zone is small, a
problem occurs.

This seems to happen most with recent sandybridge laptops but it's
probably a co-incidence as some of these laptops just happen to have a
small Normal zone.  The reproduction case is almost always during copying
large files that kswapd pegs at 100% CPU until the file is deleted or
cache is dropped.

The problem is mostly down to sleeping_prematurely() keeping kswapd awake
when the highest zone is small and unreclaimable and compounded by the
fact we shrink slabs even when not shrinking zones causing a lot of time
to be spent in shrinkers and a lot of memory to be reclaimed.

Patch 1 corrects sleeping_prematurely to check the zones matching
the classzone_idx instead of all zones.

Patch 2 avoids shrinking slab when we are not shrinking a zone.

Patch 3 notes that sleeping_prematurely is checking lower zones against
a high classzone which is not what allocators or balance_pgdat()
is doing leading to an artifical belief that kswapd should be
still awake.

Patch 4 notes that when balance_pgdat() gives up on a high zone that the
decision is not communicated to sleeping_prematurely()

This problem affects for certain and is expected to affect 2.6.39
and 3.0-rc4 as well.  If accepted, they need to go to -stable to be picked
up by distros and this series is against 3.0-rc4.  I've cc'd people that
reported similar problems recently to see if they still suffer from the
problem and if this fixes it.

This patch: correct the check for kswapd sleeping in sleeping_prematurely()

During allocator-intensive workloads, kswapd will be woken frequently
causing free memory to oscillate between the high and min watermark.  This
is expected behaviour.

A problem occurs if the highest zone is small.  balance_pgdat() only
considers unreclaimable zones when priority is DEF_PRIORITY but
sleeping_prematurely considers all zones.  It's possible for this sequence
to occur

  1. kswapd wakes up and enters balance_pgdat()
  2. At DEF_PRIORITY, marks highest zone unreclaimable
  3. At DEF_PRIORITY-1, ignores highest zone setting end_zone
  4. At DEF_PRIORITY-1, calls shrink_slab freeing memory from
        highest zone, clearing all_unreclaimable. Highest zone
        is still unbalanced
  5. kswapd returns and calls sleeping_prematurely
  6. sleeping_prematurely looks at *all* zones, not just the ones
     being considered by balance_pgdat. The highest small zone
     has all_unreclaimable cleared but the zone is not
     balanced. all_zones_ok is false so kswapd stays awake

This patch corrects the behaviour of sleeping_prematurely to check the
zones balance_pgdat() checked.

Signed-off-by: Mel Gorman <>
Reported-by: Pádraig Brady <>
Tested-by: Pádraig Brady <>
Tested-by: Andrew Lutomirski <>
Acked-by: Rik van Riel <>
Reviewed-by: Minchan Kim <>
Reviewed-by: KOSAKI Motohiro <>
Cc: Johannes Weiner <>
Signed-off-by: Andrew Morton <>
Signed-off-by: Linus Torvalds <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agodrm/radeon/kms: fix typo in IH_CNTL swap bitfield
Alex Deucher [Wed, 6 Jul 2011 19:52:27 +0000 (19:52 +0000)]
drm/radeon/kms: fix typo in IH_CNTL swap bitfield

commit fcb857abc4c0d512e99a08ba0415be9a6d65b4c0 upstream.

Only affects BE systems.

Signed-off-by: Alex Deucher <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agodrm/radeon/kms: fix typo in evergreen disp int status register
Alex Deucher [Wed, 6 Jul 2011 19:37:47 +0000 (19:37 +0000)]
drm/radeon/kms: fix typo in evergreen disp int status register

commit 37cba6c6f4687e694ac6e3adcf2c2b2974aa3aae upstream.

Signed-off-by: Alex Deucher <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agodrm/radeon/kms: free ib pool on module unloading
Jerome Glisse [Wed, 6 Jul 2011 18:30:09 +0000 (18:30 +0000)]
drm/radeon/kms: free ib pool on module unloading

commit ccd6895d401efad0c0e41d0e93fba4ef3085e268 upstream.

ib pool weren't free for various newer asic on module unload.
This doesn't cause much arm but still could be candidate for

Signed-off-by: Jerome Glisse <>
Reviewed-by: Alex Deucher <>
Signed-off-by: Dave Airlie <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoUSB: serial: add IDs for WinChipHead USB->RS232 adapter
Wolfgang Denk [Tue, 19 Jul 2011 09:25:38 +0000 (11:25 +0200)]
USB: serial: add IDs for WinChipHead USB->RS232 adapter

commit 026dfaf18973404a01f488d6aa556a8c466e06a4 upstream.

Add ID 4348:5523 for WinChipHead USB->RS 232 adapter with
Prolifec PL2303 chipset

Signed-off-by: Wolfgang Denk <>
Signed-off-by: Greg Kroah-Hartman <>
6 years agoLinux
Greg Kroah-Hartman [Sat, 9 Jul 2011 06:16:06 +0000 (23:16 -0700)]