linux-edison.git
12 years agodrm/gma500: Add fb gtt offset to fb base
Patrik Jakobsson [Tue, 14 May 2013 12:37:10 +0000 (14:37 +0200)]
drm/gma500: Add fb gtt offset to fb base

Old code assumed framebuffer starts at base of stolen memory. Since the
addition of hardware cursors, this might not be true anymore so add the
gtt offset to the calculation.

Reported-by: Holger Schurig <holgerschurig@gmail.com>
Tested-by: Holger Schurig <holgerschurig@gmail.com>
Signed-off-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
12 years agonetfilter: log: netns NULL ptr bug when calling from conntrack
Hans Schillstrom [Wed, 15 May 2013 01:23:45 +0000 (01:23 +0000)]
netfilter: log: netns NULL ptr bug when calling from conntrack

Since (69b34fb netfilter: xt_LOG: add net namespace support
for xt_LOG), we hit this:

[ 4224.708977] BUG: unable to handle kernel NULL pointer dereference at 0000000000000388
[ 4224.709074] IP: [<ffffffff8147f699>] ipt_log_packet+0x29/0x270

when callling log functions from conntrack both in and out
are NULL i.e. the net pointer is invalid.

Adding struct net *net in call to nf_logfn() will secure that
there always is a vaild net ptr.

Reported as netfilter's bugzilla bug 818:
https://bugzilla.netfilter.org/show_bug.cgi?id=818

Reported-by: Ronald <ronald645@gmail.com>
Signed-off-by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
12 years agopinctrl/lantiq: Free mapping configs for both pin and groups
Laurent Pinchart [Tue, 14 May 2013 15:00:38 +0000 (17:00 +0200)]
pinctrl/lantiq: Free mapping configs for both pin and groups

When creating mappings from DT both pin config and group config mappings
are allocated. Free them both when destroying the mappings.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Acked-by: John Crispin <blogic@openwrt.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
12 years agos390/ftrace: fix mcount adjustment
Heiko Carstens [Mon, 13 May 2013 12:48:52 +0000 (14:48 +0200)]
s390/ftrace: fix mcount adjustment

Tony Jones reported that the ftrace self tests on s390 do not work:

<6>Testing dynamic ftrace ops #1: (0 0 0 0 0) FAILED!
<6>Testing tracer irqsoff:
<3>failed to start irqsoff tracer
<4>.. no entries found ..FAILED!
<6>Testing tracer wakeup:
<3>failed to start wakeup tracer
<4>.. no entries found ..FAILED!
<6>Testing tracer function_graph:
<4>Failed to init function_graph tracer, init returned -19
<4>FAILED!

This happens because we forgot to adjust the instruction pointer that gets
passed to the ftrace trace function by MCOUNT_INSN_SIZE.

In addition change MCOUNT_INSN_SIZE to the correct value on 31 bit.
It only worked so far because the to be patched instruction was identical.

Reported-by: Tony Jones <tonyj@suse.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
12 years agos390: fix gmap_ipte_notifier vs. software dirty pages
Christian Borntraeger [Wed, 8 May 2013 13:25:38 +0000 (15:25 +0200)]
s390: fix gmap_ipte_notifier vs. software dirty pages

On heavy paging load some guest cpus started to loop in gmap_ipte_notify.
This was visible as stalled cpus inside the guest. The gmap_ipte_notifier
tries to map a user page and then made sure that the pte is valid and
writable. Turns out that with the software change bit tracking the pte
can become read-only (and only software writable) if the page is clean.
Since we loop in this code, the page would stay clean and, therefore,
be never writable again.
Let us just use fixup_user_fault, that guarantees to call handle_mm_fault.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
12 years agoARM: at91: rm9200 fix time support
Jean-Christophe PLAGNIOL-VILLARD [Wed, 15 May 2013 10:12:33 +0000 (12:12 +0200)]
ARM: at91: rm9200 fix time support

since commit 838a2ae80a6ab52139fb1bf0a93ea8c5eff94488
Author: Shawn Guo <shawn.guo@linaro.org>
Date:   Sat Jan 12 11:50:05 2013 +0000

ARM: use clockevents_config_and_register() where possible

The timer is wrongly configured and result in a nice crash
so revert it on rm9200 timer.

Signed-off-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
Cc: stable <stable@vger.kernel.org> # 3.8+
Cc: Shawn Guo <shawn.guo@linaro.org>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
12 years agotarget: close target_put_sess_cmd() vs. core_tmr_abort_task() race
Joern Engel [Mon, 13 May 2013 20:30:06 +0000 (16:30 -0400)]
target: close target_put_sess_cmd() vs. core_tmr_abort_task() race

It is possible for one thread to to take se_sess->sess_cmd_lock in
core_tmr_abort_task() before taking a reference count on
se_cmd->cmd_kref, while another thread in target_put_sess_cmd() drops
se_cmd->cmd_kref before taking se_sess->sess_cmd_lock.

This introduces kref_put_spinlock_irqsave() and uses it in
target_put_sess_cmd() to close the race window.

Signed-off-by: Joern Engel <joern@logfs.org>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
12 years agotarget: removed unused transport_state flag
Joern Engel [Mon, 13 May 2013 20:30:05 +0000 (16:30 -0400)]
target: removed unused transport_state flag

Signed-off-by: Joern Engel <joern@logfs.org>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
12 years agotarget/iblock: Fix WCE=1 + DPOFUA=1 backend WRITE regression
Nicholas Bellinger [Wed, 15 May 2013 06:41:04 +0000 (23:41 -0700)]
target/iblock: Fix WCE=1 + DPOFUA=1 backend WRITE regression

This patch fixes a regression bug introduced in v3.9-rc1 where if the
underlying struct block_device for a IBLOCK backend is configured with
WCE=1 + DPOFUA=1 settings, the rw = WRITE assignment no longer occurs
in iblock_execute_rw(), and rw = 0 is passed to iblock_submit_bios()
in effect causing a READ bio operation to occur.

The offending commit is:

commit d0c8b259f8970d39354c1966853363345d401330
Author: Nicholas Bellinger <nab@linux-iscsi.org>
Date:   Tue Jan 29 22:10:06 2013 -0800

    target/iblock: Use backend REQ_FLUSH hint for WriteCacheEnabled status

Note the WCE=1 + DPOFUA=0, WCE=0 + DPOFUA=1, and WCE=0 + DPOFUA=0 cases
are not affected by this regression bug.

Reported-by: Chris Boot <bootc@bootc.net>
Tested-by: Chris Boot <bootc@bootc.net>
Reported-by: Hannes Reinecke <hare@suse.de>
Cc: <stable@vger.kernel.org>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
12 years agoMerge branch 'bcache-for-upstream' of git://evilpiepirate.org/~kent/linux-bcache...
Jens Axboe [Wed, 15 May 2013 08:36:25 +0000 (10:36 +0200)]
Merge branch 'bcache-for-upstream' of git://evilpiepirate.org/~kent/linux-bcache into for-linus

Kent writes:

Jens - couple more bcache patches. Bug fixes and a doc update.

12 years agomtip32xx: Correctly handle bio->bi_idx != 0 conditions
Sam Bradshaw [Wed, 15 May 2013 08:09:05 +0000 (10:09 +0200)]
mtip32xx: Correctly handle bio->bi_idx != 0 conditions

Stacking drivers may append bvecs to existing bio's, resulting
in non-zero bi_idx conditions.  This patch counts the loops of
bio_for_each_segment() rather than inheriting the bi_idx value
to pass as a segment count to the hardware submission routine.

Signed-off-by: Sam Bradshaw <sbradshaw@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
12 years agomtip32xx: Fix NULL pointer dereference during module unload
Sam Bradshaw [Wed, 15 May 2013 08:04:34 +0000 (10:04 +0200)]
mtip32xx: Fix NULL pointer dereference during module unload

An open file-handle to one or more of the driver exported debugfs
nodes causes raciness in recursive removal during module unload;
sometimes a stale parent dentry is dereferenced when more than 1
pci device is present.

Signed-off-by: Sam Bradshaw <sbradshaw@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
12 years agobcache: Fix error handling in init code
Kent Overstreet [Wed, 15 May 2013 07:11:26 +0000 (00:11 -0700)]
bcache: Fix error handling in init code

This code appears to have rotted... fix various bugs and do some
refactoring.

Signed-off-by: Kent Overstreet <koverstreet@google.com>
12 years agobcache: clarify free/available/unused space
Gabriel [Wed, 24 Apr 2013 17:51:02 +0000 (19:51 +0200)]
bcache: clarify free/available/unused space

Don't describe bcache_available_percent as free space but as
non-writeback space.  Describe priority_stats in more detail
and point to that for total bcache occupation.

Signed-off-by: Gabriel de Perthuis <g2p.code+bcache@gmail.com>
Signed-off-by: Kent Overstreet <koverstreet@google.com>
12 years agobcache: drop "select CLOSURES"
Paul Bolle [Mon, 13 May 2013 08:35:21 +0000 (10:35 +0200)]
bcache: drop "select CLOSURES"

The Kconfig entry for BCACHE selects CLOSURES. But there's no Kconfig
symbol CLOSURES. That symbol was used in development versions of bcache,
but was removed when the closures code was no longer provided as a
kernel library. It can safely be dropped.

Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
12 years agobcache: Fix incompatible pointer type warning
Emil Goode [Thu, 9 May 2013 20:39:26 +0000 (22:39 +0200)]
bcache: Fix incompatible pointer type warning

The function pointer release in struct block_device_operations
should point to functions declared as void.

Sparse warnings:

drivers/md/bcache/super.c:656:27: warning:
incorrect type in initializer (different base types)
drivers/md/bcache/super.c:656:27:
expected void ( *release )( ... )
drivers/md/bcache/super.c:656:27:
got int ( static [toplevel] *<noident> )( ... )

drivers/md/bcache/super.c:656:2: warning:
initialization from incompatible pointer type [enabled by default]

drivers/md/bcache/super.c:656:2: warning:
(near initialization for ‘bcache_ops.release’) [enabled by default]

Signed-off-by: Emil Goode <emilgoode@gmail.com>
Signed-off-by: Kent Overstreet <koverstreet@google.com>
12 years agoARM: sunxi: select ARCH_REQUIRE_GPIOLIB
Maxime Ripard [Sun, 12 May 2013 13:06:51 +0000 (15:06 +0200)]
ARM: sunxi: select ARCH_REQUIRE_GPIOLIB

CONFIG_ARCH_SUNXI currently doesn't enable a gpiolib, which causes build
problems when building a kernel with only the sunxi platform enabled.
Select ARCH_REQUIRE_GPIOLIB to solve this.

Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Tested-by: Emilio López <emilio@elopez.com.ar>
12 years agoARC: [TB10x] Remove redundant abilis,simple-pinctrl mechanism
Christian Ruppert [Fri, 10 May 2013 14:05:50 +0000 (16:05 +0200)]
ARC: [TB10x] Remove redundant abilis,simple-pinctrl mechanism

The TB10x platform port includes a custom mechanism using to set up
default pin controller configurations using abilis,simple-default
pin configurations of nodes compatible with abilis,simple-pinctrl. This
mechanism is redundant with the Linux standard "default" pin
configuration, see commit ab78029ecc347debbd737f06688d788bd9d60c1d
"drivers/pinctrl: grab default handles from device core".
This patch removes the TB10x custom mechanism in favour of the Linux
standard.

Signed-off-by: Christian Ruppert <christian.ruppert@abilis.com>
Reviewed-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
12 years agonet/macb: fix ISR clear-on-write behavior only for some SoC
Nicolas Ferre [Tue, 14 May 2013 03:00:16 +0000 (03:00 +0000)]
net/macb: fix ISR clear-on-write behavior only for some SoC

Commit 749a2b6 (net/macb: clear tx/rx completion flags in ISR)
introduces clear-on-write on ISR register. This behavior is not always
implemented when using Cadence MACB/GEM and is breaking other platforms.
We are using the Design Configuration Register 1 information and a capability
property to actually activate this clear-on-write behavior on ISR.

Reported-by: Hein Tibosch <hein_tibosch@yahoo.es>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Tested-by: Hein Tibosch <hein_tibosch@yahoo.es>
Acked-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet/802/mrp: fix lockdep splat
Eric Dumazet [Mon, 13 May 2013 02:24:11 +0000 (02:24 +0000)]
net/802/mrp: fix lockdep splat

commit fb745e9a037895 ("net/802/mrp: fix possible race condition when
calling mrp_pdu_queue()") introduced a lockdep splat.

[   19.735147] =================================
[   19.735235] [ INFO: inconsistent lock state ]
[   19.735324] 3.9.2-build-0063 #4 Not tainted
[   19.735412] ---------------------------------
[   19.735500] inconsistent {IN-SOFTIRQ-W} -> {SOFTIRQ-ON-W} usage.
[   19.735592] rmmod/1840 [HC0[0]:SC0[0]:HE1:SE1] takes:
[   19.735682]  (&(&app->lock)->rlock#2){+.?...}, at: [<f862bb5b>]
mrp_uninit_applicant+0x69/0xba [mrp]

app->lock is normally taken under softirq context, so disable BH to
avoid the splat.

Reported-by: Denys Fedoryshchenko <denys@visp.net.lb>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: David Ward <david.ward@ll.mit.edu>
Cc: Cong Wang <amwang@redhat.com>
Tested-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotime: Revert ALWAYS_USE_PERSISTENT_CLOCK compile time optimizaitons
John Stultz [Wed, 24 Apr 2013 18:32:56 +0000 (11:32 -0700)]
time: Revert ALWAYS_USE_PERSISTENT_CLOCK compile time optimizaitons

Kay Sievers noted that the ALWAYS_USE_PERSISTENT_CLOCK config,
which enables some minor compile time optimization to avoid
uncessary code in mostly the suspend/resume path could cause
problems for userland.

In particular, the dependency for RTC_HCTOSYS on
!ALWAYS_USE_PERSISTENT_CLOCK, which avoids setting the time
twice and simplifies suspend/resume, has the side effect
of causing the /sys/class/rtc/rtcN/hctosys flag to always be
zero, and this flag is commonly used by udev to setup the
/dev/rtc symlink to /dev/rtcN, which can cause pain for
older applications.

While the udev rules could use some work to be less fragile,
breaking userland should strongly be avoided. Additionally
the compile time optimizations are fairly minor, and the code
being optimized is likely to be reworked in the future, so
lets revert this change.

Reported-by: Kay Sievers <kay@vrfy.org>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Cc: stable <stable@vger.kernel.org> #3.9
Cc: Feng Tang <feng.tang@intel.com>
Cc: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Link: http://lkml.kernel.org/r/1366828376-18124-1-git-send-email-john.stultz@linaro.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
12 years agoworkqueue: Make schedule_work() available again to non GPL modules
Marc Dionne [Mon, 6 May 2013 21:44:55 +0000 (17:44 -0400)]
workqueue: Make schedule_work() available again to non GPL modules

Commit 8425e3d5bdbe ("workqueue: inline trivial wrappers") changed
schedule_work() and schedule_delayed_work() to inline wrappers,
but these rely on some symbols that are EXPORT_SYMBOL_GPL, while
the original functions were EXPORT_SYMBOL.  This has the effect of
changing the licensing requirement for these functions and making
them unavailable to non GPL modules.

Make them available again by removing the restriction on the
required symbols.

Signed-off-by: Marc Dionne <marc.dionne@your-file-system.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
12 years agoworkqueue: correct handling of the pool spin_lock
Joonsoo Kim [Tue, 30 Apr 2013 15:07:00 +0000 (00:07 +0900)]
workqueue: correct handling of the pool spin_lock

When we fail to mutex_trylock(), we release the pool spin_lock and do
mutex_lock(). After that, we should regrab the pool spin_lock, but,
regrabbing is missed in current code. So correct it.

Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
12 years agomv643xx_eth: fix NAPI weight being > 64
Andrew Lunn [Mon, 13 May 2013 23:54:20 +0000 (23:54 +0000)]
mv643xx_eth: fix NAPI weight being > 64

3.10-rc1 issues the following warning:

netif_napi_add() called with weight 128 on device eth%d

This patch reduce the weight to 64, using NAPI_POLL_WEIGHT.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotcp: fix tcp_md5_hash_skb_data()
Eric Dumazet [Mon, 13 May 2013 21:25:52 +0000 (21:25 +0000)]
tcp: fix tcp_md5_hash_skb_data()

TCP md5 communications fail [1] for some devices, because sg/crypto code
assume page offsets are below PAGE_SIZE.

This was discovered using mlx4 driver [2], but I suspect loopback
might trigger the same bug now we use order-3 pages in tcp_sendmsg()

[1] Failure is giving following messages.

huh, entered softirq 3 NET_RX ffffffff806ad230 preempt_count 00000100,
exited with 00000101?

[2] mlx4 driver uses order-2 pages to allocate RX frags

Reported-by: Matt Schnall <mischnal@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Bernhard Beck <bbeck@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Avoid double insertion of vlan tags.
Sarveshwar Bandi [Mon, 13 May 2013 20:28:20 +0000 (20:28 +0000)]
be2net: Avoid double insertion of vlan tags.

Fix to avoid double insertion of  vlan tags into the packet while
handling an asic workaroud (issue introduced by net next Commit bc0c340)

Signed-off-by: Sarveshwar Bandi <sarveshwar.bandi@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosfc: Reduce RX scatter buffer size, and reduce alignment if appropriate
Ben Hutchings [Mon, 13 May 2013 12:01:22 +0000 (12:01 +0000)]
sfc: Reduce RX scatter buffer size, and reduce alignment if appropriate

efx_start_datapath() asserts that we can fit 2 RX scatter buffers plus
a software structure, each appropriately aligned, into a single page.
Where L1_CACHE_BYTES == 256 and PAGE_SIZE == 4096, which is the case
on s390, this assertion fails.

The current scatter buffer size is also not a multiple of 64 or 128,
which are more common cache line sizes.  If we can make both the start
and end of a scatter buffer cache-aligned, this will reduce the need
for read-modify-write operations on inter- processor links.

Fix the alignment by reducing EFX_RX_USR_BUF_SIZE to 2048 - 256 ==
1792.  (We could use 2048 - L1_CACHE_BYTES, but EFX_RX_USR_BUF_SIZE
also affects user-level networking where a larger amount of
housekeeping data may be needed.  Although this version of the driver
does not support user-level networking, I prefer to keep scattering
behaviour consistent with the out-of-tree version.)

This still doesn't fix the s390 build because like most architectures
it has NET_IP_ALIGN == 2.  When NET_IP_ALIGN != 0 we cannot achieve
cache line alignment at either the start or end of a scatter buffer,
so there is actually no point in padding the buffers to a multiple of
the cache line size.  All we need is 4-byte alignment of the network
header, so do that.

Adjust the assertions accordingly.

Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reported-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosfc: Delete EFX_PAGE_IP_ALIGN, equivalent to NET_IP_ALIGN
Ben Hutchings [Mon, 13 May 2013 11:58:31 +0000 (11:58 +0000)]
sfc: Delete EFX_PAGE_IP_ALIGN, equivalent to NET_IP_ALIGN

The two architectures that define CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS
(powerpc and x86) now both define NET_IP_ALIGN as 0, so there is no
need for this optimisation any more.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotg3: Fix data corruption on 5725 with TSO
Michael Chan [Mon, 13 May 2013 11:04:16 +0000 (11:04 +0000)]
tg3: Fix data corruption on 5725 with TSO

The 5725 family of devices (asic rev 5762), corrupts TSO packets where
the buffer is within MSS bytes of a 4G boundary (4G, 8G etc.). Detect
this condition and trigger the workaround path.

Cc: <stable@vger.kernel.org>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: Nithin Nayak Sujir <nsujir@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotg3: Skip powering down function 0 on certain serdes devices
Nithin Sujir [Mon, 13 May 2013 11:04:15 +0000 (11:04 +0000)]
tg3: Skip powering down function 0 on certain serdes devices

On the 5718, 5719 and 5720 serdes devices, powering down function 0
results in all the other ports being powered down. Add code to skip
function 0 power down.

v2:
 - Modify tg3_phy_power_bug() function to use a switch instead of a
   complicated if statement. Suggested by Joe Perches.

Cc: <stable@vger.kernel.org>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: Nithin Nayak Sujir <nsujir@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agolibata: update "Maintained by:" tags
Tejun Heo [Tue, 14 May 2013 18:09:50 +0000 (11:09 -0700)]
libata: update "Maintained by:" tags

Jeff moved on to a greener pasture.

 s/Maintained by: Jeff Garzik/Maintained by: Tejun Heo/g

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Jeff Garzik <jgarzik@pobox.com>
12 years agorcu: Fix comparison sense in rcu_needs_cpu()
Paul E. McKenney [Mon, 13 May 2013 13:53:37 +0000 (06:53 -0700)]
rcu: Fix comparison sense in rcu_needs_cpu()

Commit c0f4dfd4f (rcu: Make RCU_FAST_NO_HZ take advantage of numbered
callbacks) introduced a bug that can result in excessively long grace
periods.  This bug reverse the senes of the "if" statement checking
for lazy callbacks, so that RCU takes a lazy approach when there are
in fact non-lazy callbacks.  This can result in excessive boot, suspend,
and resume times.

This commit therefore fixes the sense of this "if" statement.

Reported-by: Borislav Petkov <bp@alien8.de>
Reported-by: Bjørn Mork <bjorn@mork.no>
Reported-by: Joerg Roedel <joro@8bytes.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Bjørn Mork <bjorn@mork.no>
Tested-by: Joerg Roedel <joro@8bytes.org>
12 years agoMerge tag 'ext4_for_linus_stable' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 14 May 2013 16:30:54 +0000 (09:30 -0700)]
Merge tag 'ext4_for_linus_stable' of git://git./linux/kernel/git/tytso/ext4

Pull ext4 update from Ted Ts'o:
 "Fixed regressions (two stability regressions and a performance
  regression) introduced during the 3.10-rc1 merge window.

  Also included is a bug fix relating to allocating blocks after
  resizing an ext3 file system when using the ext4 file system driver"

* tag 'ext4_for_linus_stable' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  jbd,jbd2: fix oops in jbd2_journal_put_journal_head()
  ext4: revert "ext4: use io_end for multiple bios"
  ext4: limit group search loop for non-extent files
  ext4: fix fio regression

12 years agofuse: allocate for_background dio requests based on io->async state
Brian Foster [Tue, 14 May 2013 15:25:39 +0000 (11:25 -0400)]
fuse: allocate for_background dio requests based on io->async state

Commit 8b41e671 introduced explicit background checking for fuse_req
structures with BUG_ON() checks for the appropriate type of request in
in the associated send functions. Commit bcba24cc introduced the ability
to send dio requests as background requests but does not update the
request allocation based on the type of I/O request. As a result, a
BUG_ON() triggers in the fuse_request_send_background() background path if
an async I/O is sent.

Allocate a request based on the async state of the fuse_io_priv to avoid
the BUG.

Signed-off-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
12 years agoMerge branch 'for-3.10-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq
Linus Torvalds [Tue, 14 May 2013 16:06:29 +0000 (09:06 -0700)]
Merge branch 'for-3.10-fixes' of git://git./linux/kernel/git/tj/wq

Pull workqueue fix from Tejun Heo:
 "A fix for a workqueue_congested() regression that broke fscache"

* 'for-3.10-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq:
  workqueue: workqueue_congested() shouldn't translate WORK_CPU_UNBOUND into node number

12 years agotimer: Don't reinitialize the cpu base lock during CPU_UP_PREPARE
Tirupathi Reddy [Tue, 14 May 2013 08:29:02 +0000 (13:59 +0530)]
timer: Don't reinitialize the cpu base lock during CPU_UP_PREPARE

An inactive timer's base can refer to a offline cpu's base.

In the current code, cpu_base's lock is blindly reinitialized each
time a CPU is brought up. If a CPU is brought online during the period
that another thread is trying to modify an inactive timer on that CPU
with holding its timer base lock, then the lock will be reinitialized
under its feet. This leads to following SPIN_BUG().

<0> BUG: spinlock already unlocked on CPU#3, kworker/u:3/1466
<0> lock: 0xe3ebe000, .magic: dead4ead, .owner: kworker/u:3/1466, .owner_cpu: 1
<4> [<c0013dc4>] (unwind_backtrace+0x0/0x11c) from [<c026e794>] (do_raw_spin_unlock+0x40/0xcc)
<4> [<c026e794>] (do_raw_spin_unlock+0x40/0xcc) from [<c076c160>] (_raw_spin_unlock+0x8/0x30)
<4> [<c076c160>] (_raw_spin_unlock+0x8/0x30) from [<c009b858>] (mod_timer+0x294/0x310)
<4> [<c009b858>] (mod_timer+0x294/0x310) from [<c00a5e04>] (queue_delayed_work_on+0x104/0x120)
<4> [<c00a5e04>] (queue_delayed_work_on+0x104/0x120) from [<c04eae00>] (sdhci_msm_bus_voting+0x88/0x9c)
<4> [<c04eae00>] (sdhci_msm_bus_voting+0x88/0x9c) from [<c04d8780>] (sdhci_disable+0x40/0x48)
<4> [<c04d8780>] (sdhci_disable+0x40/0x48) from [<c04bf300>] (mmc_release_host+0x4c/0xb0)
<4> [<c04bf300>] (mmc_release_host+0x4c/0xb0) from [<c04c7aac>] (mmc_sd_detect+0x90/0xfc)
<4> [<c04c7aac>] (mmc_sd_detect+0x90/0xfc) from [<c04c2504>] (mmc_rescan+0x7c/0x2c4)
<4> [<c04c2504>] (mmc_rescan+0x7c/0x2c4) from [<c00a6a7c>] (process_one_work+0x27c/0x484)
<4> [<c00a6a7c>] (process_one_work+0x27c/0x484) from [<c00a6e94>] (worker_thread+0x210/0x3b0)
<4> [<c00a6e94>] (worker_thread+0x210/0x3b0) from [<c00aad9c>] (kthread+0x80/0x8c)
<4> [<c00aad9c>] (kthread+0x80/0x8c) from [<c000ea80>] (kernel_thread_exit+0x0/0x8)

As an example, this particular crash occurred when CPU #3 is executing
mod_timer() on an inactive timer whose base is refered to offlined CPU
#2.  The code locked the timer_base corresponding to CPU #2. Before it
could proceed, CPU #2 came online and reinitialized the spinlock
corresponding to its base. Thus now CPU #3 held a lock which was
reinitialized. When CPU #3 finally ended up unlocking the old cpu_base
corresponding to CPU #2, we hit the above SPIN_BUG().

CPU #0 CPU #3        CPU #2
------ -------        -------
.....  ......       <Offline>
mod_timer()
 lock_timer_base
   spin_lock_irqsave(&base->lock)

cpu_up(2)  .....         ......
init_timers_cpu()
....  .....      spin_lock_init(&base->lock)
.....    spin_unlock_irqrestore(&base->lock)  ......
   <spin_bug>

Allocation of per_cpu timer vector bases is done only once under
"tvec_base_done[]" check. In the current code, spinlock_initialization
of base->lock isn't under this check. When a CPU is up each time the
base lock is reinitialized. Move base spinlock initialization under
the check.

Signed-off-by: Tirupathi Reddy <tirupath@codeaurora.org>
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/1368520142-4136-1-git-send-email-tirupath@codeaurora.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
12 years agorcu/idle: Wrap cpu-idle poll mode within rcu_idle_enter/exit
Srivatsa S. Bhat [Mon, 13 May 2013 22:31:27 +0000 (04:01 +0530)]
rcu/idle: Wrap cpu-idle poll mode within rcu_idle_enter/exit

Bjørn Mork reported the following warning when running powertop.

[   49.289034] ------------[ cut here ]------------
[   49.289055] WARNING: at kernel/rcutree.c:502 rcu_eqs_exit_common.isra.48+0x3d/0x125()
[   49.289244] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.10.0-bisect-rcu-warn+ #107
[   49.289251]  ffffffff8157d8c8 ffffffff81801e28 ffffffff8137e4e3 ffffffff81801e68
[   49.289260]  ffffffff8103094f ffffffff81801e68 0000000000000000 ffff88023afcd9b0
[   49.289268]  0000000000000000 0140000000000000 ffff88023bee7700 ffffffff81801e78
[   49.289276] Call Trace:
[   49.289285]  [<ffffffff8137e4e3>] dump_stack+0x19/0x1b
[   49.289293]  [<ffffffff8103094f>] warn_slowpath_common+0x62/0x7b
[   49.289300]  [<ffffffff8103097d>] warn_slowpath_null+0x15/0x17
[   49.289306]  [<ffffffff810a9006>] rcu_eqs_exit_common.isra.48+0x3d/0x125
[   49.289314]  [<ffffffff81079b49>] ? trace_hardirqs_off_caller+0x37/0xa6
[   49.289320]  [<ffffffff810a9692>] rcu_idle_exit+0x85/0xa8
[   49.289327]  [<ffffffff8107076e>] trace_cpu_idle_rcuidle+0xae/0xff
[   49.289334]  [<ffffffff810708b1>] cpu_startup_entry+0x72/0x115
[   49.289341]  [<ffffffff813689e5>] rest_init+0x149/0x150
[   49.289347]  [<ffffffff8136889c>] ? csum_partial_copy_generic+0x16c/0x16c
[   49.289355]  [<ffffffff81a82d34>] start_kernel+0x3f0/0x3fd
[   49.289362]  [<ffffffff81a8274c>] ? repair_env_string+0x5a/0x5a
[   49.289368]  [<ffffffff81a82481>] x86_64_start_reservations+0x2a/0x2c
[   49.289375]  [<ffffffff81a82550>] x86_64_start_kernel+0xcd/0xd1
[   49.289379] ---[ end trace 07a1cc95e29e9036 ]---

The warning is that 'rdtp->dynticks' has an unexpected value, which roughly
translates to - the calls to rcu_idle_enter() and rcu_idle_exit() were not
made in the correct order, or otherwise messed up.

And Bjørn's painstaking debugging indicated that this happens when the idle
loop enters the poll mode. Looking at the poll function cpu_idle_poll(), and
the implementation of trace_cpu_idle_rcuidle(), the problem becomes very clear:
cpu_idle_poll() lacks calls to rcu_idle_enter/exit(), and trace_cpu_idle_rcuidle()
calls them in the reverse order - first rcu_idle_exit(), and then rcu_idle_enter().
Hence the even/odd alternative sequencing of rdtp->dynticks goes for a toss.

And powertop readily triggers this because powertop uses the idle-tracing
infrastructure extensively.

So, to fix this, wrap the code in cpu_idle_poll() within rcu_idle_enter/exit(),
so that it blends properly with the calls inside trace_cpu_idle_rcuidle() and
thus get the function ordering right.

Reported-and-tested-by: Bjørn Mork <bjorn@mork.no>
Cc: Paul McKenney <paulmck@linux.vnet.ibm.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Link: http://lkml.kernel.org/r/519169BF.4080208@linux.vnet.ibm.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
12 years agotick: Don't invoke tick_nohz_stop_sched_tick() if the cpu is offline
Thomas Gleixner [Mon, 13 May 2013 19:40:27 +0000 (21:40 +0200)]
tick: Don't invoke tick_nohz_stop_sched_tick() if the cpu is offline

commit 5b39939a4 (nohz: Move ts->idle_calls incrementation into strict
idle logic) moved code out of tick_nohz_stop_sched_tick() and missed
to bail out when the cpu is offline. That's causing subsequent
failures as an offline CPU is supposed to die and not to fiddle with
nohz magic.

Return false in can_stop_idle_tick() if the cpu is offline.

Reported-and-tested-by: Jiri Kosina <jkosina@suse.cz>
Reported-and-tested-by: Prarit Bhargava <prarit@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Tony Luck <tony.luck@intel.com>
Cc: x86@kernel.org
Link: http://lkml.kernel.org/r/alpine.LFD.2.02.1305132138160.2863@ionos
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
12 years agocgroup: initialize xattr before calling d_instantiate()
Li Zefan [Tue, 14 May 2013 11:44:20 +0000 (19:44 +0800)]
cgroup: initialize xattr before calling d_instantiate()

cgroup_create_file() calls d_instantiate(), which may decide to look
at the xattrs on the file. Smack always does this and SELinux can be
configured to do so.

But cgroup_add_file() didn't initialize xattrs before calling
cgroup_create_file(), which finally leads to dereferencing NULL
dentry->d_fsdata.

This bug has been there since cgroup xattr was introduced.

Cc: <stable@vger.kernel.org> # 3.8.x
Reported-by: Ivan Bulatovic <combuster@archlinux.us>
Reported-by: Casey Schaufler <casey@schaufler-ca.com>
Signed-off-by: Li Zefan <lizefan@huawei.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
12 years agoarm64: Do not source kernel/time/Kconfig explicitly
Catalin Marinas [Tue, 14 May 2013 11:22:06 +0000 (12:22 +0100)]
arm64: Do not source kernel/time/Kconfig explicitly

As per commit 764e0da1 (timers: Fixup the Kconfig consolidation
fallout), init/Kconfig already includes kernel/time/Kconfig, so no need
to do it explicitly for arm64.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
12 years agoarm64: mm: Fix operands of clz in __flush_dcache_all
Sukanto Ghosh [Tue, 14 May 2013 09:26:54 +0000 (10:26 +0100)]
arm64: mm: Fix operands of clz in __flush_dcache_all

The format of the lower 32-bits of the 64-bit operand to 'dc cisw' is
unchanged from ARMv7 architecture and the upper bits are RES0. This
implies that the 'way' field of the operand of 'dc cisw' occupies the
bit-positions [31 .. (32-A)]. Due to the use of 64-bit extended operands
to 'clz', the existing implementation of __flush_dcache_all is incorrectly
placing the 'way' field in the bit-positions [63 .. (64-A)].

Signed-off-by: Sukanto Ghosh <sghosh@apm.com>
Tested-by: Anup Patel <anup.patel@linaro.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: stable@vger.kernel.org
12 years agoarm64: Invoke the of_platform_populate() at arch_initcall() level
Catalin Marinas [Tue, 14 May 2013 09:51:18 +0000 (10:51 +0100)]
arm64: Invoke the of_platform_populate() at arch_initcall() level

The of_platform_populate() is currently invoked at device_initcall()
level. There are however drivers that use platform_driver_probe()
directly and they need the devices to be populated. This patch makes the
of_platform_populate() and arch_initcall().

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Reported-by: Benoit Lecardonnel <Benoit.Lecardonnel@synopsys.com>
Tested-by: Benoit Lecardonnel <Benoit.Lecardonnel@synopsys.com>
12 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc
Linus Torvalds [Tue, 14 May 2013 14:43:11 +0000 (07:43 -0700)]
Merge branch 'merge' of git://git./linux/kernel/git/benh/powerpc

Pull powerpc fixes from Benjamin Herrenschmidt:
 "This is mostly bug fixes (some of them regressions, some of them I
  deemed worth merging now) along with some patches from Li Zhong
  hooking up the new context tracking stuff (for the new full NO_HZ)"

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc: (25 commits)
  powerpc: Set show_unhandled_signals to 1 by default
  powerpc/perf: Fix setting of "to" addresses for BHRB
  powerpc/pmu: Fix order of interpreting BHRB target entries
  powerpc/perf: Move BHRB code into CONFIG_PPC64 region
  powerpc: select HAVE_CONTEXT_TRACKING for pSeries
  powerpc: Use the new schedule_user API on userspace preemption
  powerpc: Exit user context on notify resume
  powerpc: Exception hooks for context tracking subsystem
  powerpc: Syscall hooks for context tracking subsystem
  powerpc/booke64: Fix kernel hangs at kernel_dbg_exc
  powerpc: Fix irq_set_affinity() return values
  powerpc: Provide __bswapdi2
  powerpc/powernv: Fix starting of secondary CPUs on OPALv2 and v3
  powerpc/powernv: Detect OPAL v3 API version
  powerpc: Fix MAX_STACK_TRACE_ENTRIES too low warning again
  powerpc: Make CONFIG_RTAS_PROC depend on CONFIG_PROC_FS
  powerpc: Bring all threads online prior to migration/hibernation
  powerpc/rtas_flash: Fix validate_flash buffer overflow issue
  powerpc/kexec: Fix kexec when using VMX optimised memcpy
  powerpc: Fix build errors STRICT_MM_TYPECHECKS
  ...

12 years agoARM: at91: dts: request only spi cs-gpios used on sama5d3x cpu module
Ludovic Desroches [Wed, 24 Apr 2013 13:56:48 +0000 (15:56 +0200)]
ARM: at91: dts: request only spi cs-gpios used on sama5d3x cpu module

Instead of requesting all available spi cs-gpios, request only the ones used on
the board, in our case on the cpu module.

Signed-off-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Acked-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
12 years agoARM: at91/trivial: typo in GEM compatible string
Nicolas Ferre [Tue, 14 May 2013 13:14:49 +0000 (15:14 +0200)]
ARM: at91/trivial: typo in GEM compatible string

Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
12 years agoARM: at91/trivial: fix model name for SAM9X25-EK
Jean-Christophe PLAGNIOL-VILLARD [Tue, 7 May 2013 20:06:57 +0000 (22:06 +0200)]
ARM: at91/trivial: fix model name for SAM9X25-EK

Signed-off-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Cc: stable <stable@vger.kernel.org> # 3.8+
12 years agoMerge branch 'pm-fixes'
Rafael J. Wysocki [Tue, 14 May 2013 13:12:17 +0000 (15:12 +0200)]
Merge branch 'pm-fixes'

* pm-fixes:
  PM / hibernate: Correct documentation
  PM / Documentation: remove inaccurate suspend/hibernate transition lantency statement
  PM: Documentation update for freeze state
  PM: Avoid calling kfree() under spinlock in dev_pm_put_subsys_data()

12 years agoMerge branch 'acpi-fixes'
Rafael J. Wysocki [Tue, 14 May 2013 13:12:10 +0000 (15:12 +0200)]
Merge branch 'acpi-fixes'

* acpi-fixes:
  ACPI / AC: Add sleep quirk for Thinkpad e530
  ACPI / EC: Restart transaction even when the IBF flag set
  ACPI video: ignore BIOS initial backlight value for HP 1000
  ACPI: Fix section to __init. Align with usage in acpixf.h
  ACPI / PM: Move processor suspend/resume to syscore_ops

12 years agoMerge branch 'pm-cpufreq'
Rafael J. Wysocki [Tue, 14 May 2013 13:11:55 +0000 (15:11 +0200)]
Merge branch 'pm-cpufreq'

* pm-cpufreq:
  cpufreq / intel_pstate: use vzalloc() instead of vmalloc()/memset(0)
  cpufreq, ondemand: Remove leftover debug line
  cpufreq / kirkwood: don't check resource with devm_ioremap_resource
  cpufreq / intel_pstate: remove #ifdef MODULE compile fence
  cpufreq / intel_pstate: Remove idle mode PID
  cpufreq / intel_pstate: fix ffmpeg regression
  cpufreq / intel_pstate: use lowest requested max performance
  cpufreq / intel_pstate: remove idle time and duration from sample and calculations
  cpufreq: Fix incorrect dependecies for ARM SA11xx drivers
  cpufreq: ARM big LITTLE: Fix Kconfig entries
  cpufreq: cpufreq-cpu0: Free parent node for error cases
  cpufreq: cpufreq-cpu0: defer probe when regulator is not ready
  cpufreq: Issue CPUFREQ_GOV_POLICY_EXIT notifier before dropping policy refcount
  cpufreq: governors: Fix CPUFREQ_GOV_POLICY_{INIT|EXIT} notifiers
  cpufreq: ARM big LITTLE: Improve print message
  cpufreq: ARM big LITTLE: Move cpu_to_cluster() to arm_big_little.h
  cpufreq: ARM big LITTLE DT: Return CPUFREQ_ETERNAL if clock-latency isn't found
  cpufreq: ARM big LITTLE DT: Return correct transition latency
  cpufreq: ARM big LITTLE: Select PM_OPP

12 years agoMerge remote-tracking branch 'asoc/fix/wm0010' into asoc-linus
Mark Brown [Tue, 14 May 2013 12:53:21 +0000 (16:53 +0400)]
Merge remote-tracking branch 'asoc/fix/wm0010' into asoc-linus

12 years agoMerge remote-tracking branch 'asoc/fix/ux500' into asoc-linus
Mark Brown [Tue, 14 May 2013 12:53:20 +0000 (16:53 +0400)]
Merge remote-tracking branch 'asoc/fix/ux500' into asoc-linus

12 years agoMerge remote-tracking branch 'asoc/fix/da7213' into asoc-linus
Mark Brown [Tue, 14 May 2013 12:53:19 +0000 (16:53 +0400)]
Merge remote-tracking branch 'asoc/fix/da7213' into asoc-linus

12 years agopinctrl: single: fix error return code in pcs_parse_one_pinctrl_entry()
Wei Yongjun [Tue, 7 May 2013 12:06:19 +0000 (20:06 +0800)]
pinctrl: single: fix error return code in pcs_parse_one_pinctrl_entry()

Fix to return a negative error code from the error handling
case instead of 0, as done elsewhere in this function.

Introduced by commit 9dddb4df90d136429b6d6ddefceb49a9b93f6cd1
(pinctrl: single: support generic pinconf)

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Acked-by: Haojian Zhuang <haojian.zhuang@linaro.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
12 years agopinctrl: generic: Fix typos and clarify comments
Laurent Pinchart [Tue, 23 Apr 2013 13:28:38 +0000 (15:28 +0200)]
pinctrl: generic: Fix typos and clarify comments

Drive strength controls both sink and source currents, clarify the
description accordingly.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
12 years agopinctrl: vt8500: Fix incorrect data in WM8750 pinctrl table
Tony Prisk [Fri, 10 May 2013 17:45:56 +0000 (05:45 +1200)]
pinctrl: vt8500: Fix incorrect data in WM8750 pinctrl table

WMT_PIN_WAKEUP1 should be declared as WMT_PIN(0, 17) rather than
WMT_PIN(0, 16). This currently generates a runtime warning because
WMT_PIN_WAKEUP0 is already defined as WMT_PIN(0, 16).

Signed-off-by: Tony Prisk <linux@prisktech.co.nz>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
12 years agostaging: nvec: cleanup childs on remove
Marc Dietrich [Mon, 29 Apr 2013 21:14:52 +0000 (23:14 +0200)]
staging: nvec: cleanup childs on remove

Disable device functions and unregister notifier if available. The
serio device must not be "kzallocated". Otherwise serio_unregister_port
will fail because the device is already freed.

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
12 years agostaging: nvec: implement unregistering of notifiers
Marc Dietrich [Mon, 29 Apr 2013 21:14:51 +0000 (23:14 +0200)]
staging: nvec: implement unregistering of notifiers

This implements the unregistering of notifiers so kernel modules
can be unloaded.

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
12 years agostaging: nvec: add missing module aliases
Marc Dietrich [Mon, 29 Apr 2013 21:14:50 +0000 (23:14 +0200)]
staging: nvec: add missing module aliases

Keyboard and mouse drivers were missing MODULE_ALIAS
definitions. This fixes auto module loading of these
drivers.

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
12 years agopinctrl: abx500: Rejiggle platform data and DT initialisation
Lee Jones [Wed, 8 May 2013 13:29:08 +0000 (14:29 +0100)]
pinctrl: abx500: Rejiggle platform data and DT initialisation

Platform Data is invariably populated for this driver, even when
booting with Device Tree. Thus the Device Tree probing code encased
within the first check for Platform Data will never executed, causing
the driver to fail when DT is enabled.

This patch fixes the aforementioned regression by rejigging the
probe() semantics to attempt to extract a platform ID from Device Tree
if one can not be sourced from platform data.

A pointer to GPIO platform data is always passed to the driver now, so
there's little point in checking for 'pdata' and executing the DT case if
it's not there. The difference between booting with DT and !DT is when
booting with DT, plat_id is not populated. Thus, in the DT case we have
to use a DT match table in order to find out which platform we're
executing on. So, we're changing the semantics here to only use the
match table if no plat_id is supplied though platform data.

Signed-off-by: Lee Jones <lee.jones@linaro.org>
[edited commit message]
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
12 years agopinctrl: abx500: Specify failed sub-driver by ID instead of driver_data
Lee Jones [Wed, 8 May 2013 13:29:07 +0000 (14:29 +0100)]
pinctrl: abx500: Specify failed sub-driver by ID instead of driver_data

If a sub-driver has not been specified correctly, there is a good chance
that plat_id is NULL, hence using an attribute of plat_id in the error
message is likely to not only fail the driver but Oops the kernel. Use
the failed ID instead.

Signed-off-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
12 years agoARM: ux500: Provide device enumeration number suffix for SMSC911x
Lee Jones [Wed, 8 May 2013 13:29:06 +0000 (14:29 +0100)]
ARM: ux500: Provide device enumeration number suffix for SMSC911x

First Ethernet device has a ".0" appended onto the device name.
Since on a non-DT boot the ethernet will be named "smsc911x.0"
and since the clocks are not converted to device tree these
names need to be matched when providing the name from
auxdata.

Signed-off-by: Lee Jones <lee.jones@linaro.org>
[edited commit message]
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
12 years agoARM: ux500: Fix incorrect DEBUG UART virtual addresses
Lee Jones [Wed, 8 May 2013 13:29:05 +0000 (14:29 +0100)]
ARM: ux500: Fix incorrect DEBUG UART virtual addresses

A recent move to rid header files which were hindering multiplatform
support forced address allocations out of the headers and into the
files which were using them. We also lost some useful macros such as
IO_ADDRESS(), so physical -> virtual addressing has been carried out
manually in this case. Unfortunately the incorrect value was converted.
This patch rectifies the error and ensures earlyprintk works again.

Signed-off-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
12 years agoARM: ux500: Remove duplicated assignment of ab8500_platdata
Lee Jones [Wed, 8 May 2013 13:29:01 +0000 (14:29 +0100)]
ARM: ux500: Remove duplicated assignment of ab8500_platdata

Since: "05ec260 mfd: db8500-prcmu: update resource passing", the AB8500's
platform data 'ab8500_platdata' is passed directly as an attribute to
'db8500_prcmu_pdata', so there's no requirement to assign it a second
time. In fact, it's only due to an ordering issue that the entire
'db8500_prcmu_pdata' data structure isn't completely over-written by the
assignment in u8500_init_devices().

Signed-off-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
12 years agopowerpc: Set show_unhandled_signals to 1 by default
Benjamin Herrenschmidt [Tue, 14 May 2013 07:02:11 +0000 (17:02 +1000)]
powerpc: Set show_unhandled_signals to 1 by default

Just like other architectures

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agox86, efi: initial the local variable of DataSize to zero
Lee, Chun-Yi [Thu, 2 May 2013 14:07:01 +0000 (22:07 +0800)]
x86, efi: initial the local variable of DataSize to zero

That will be better initial the value of DataSize to zero for the input of
GetVariable(), otherwise we will feed a random value. The debug log of input
DataSize like this:

...
[  195.915612] EFI Variables Facility v0.08 2004-May-17
[  195.915819] efi: size: 18446744071581821342
[  195.915969] efi:  size': 18446744071581821342
[  195.916324] efi: size: 18446612150714306560
[  195.916632] efi:  size': 18446612150714306560
[  195.917159] efi: size: 18446612150714306560
[  195.917453] efi:  size': 18446612150714306560
...

The size' is value that was returned by BIOS.

After applied this patch:
[   82.442042] EFI Variables Facility v0.08 2004-May-17
[   82.442202] efi: size: 0
[   82.442360] efi:  size': 1039
[   82.443828] efi: size: 0
[   82.444127] efi:  size': 2616
[   82.447057] efi: size: 0
[   82.447356] efi:  size': 5832
...

Found on Acer Aspire V3 BIOS, it will not return the size of data if we input a
non-zero DataSize.

Cc: Matthew Garrett <mjg59@srcf.ucam.org>
Cc: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Lee, Chun-Yi <jlee@suse.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
12 years agomicroblaze: Initialize temp variable to remove compilation warning
Michal Simek [Tue, 14 May 2013 07:06:17 +0000 (09:06 +0200)]
microblaze: Initialize temp variable to remove compilation warning

Compilation warning:
arch/microblaze/kernel/cpu/cache.c:148:2: warning:
 'temp' is used uninitialized in this function [-Wuninitialized]

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
12 years agopowerpc/perf: Fix setting of "to" addresses for BHRB
Michael Neuling [Mon, 13 May 2013 18:44:58 +0000 (18:44 +0000)]
powerpc/perf: Fix setting of "to" addresses for BHRB

Currently we only set the "to" address in the branch stack when the CPU
explicitly gives us a value.  Unfortunately it only does this for XL form
branches (eg blr, bctr, bctar) and not I and B form branches (eg b, bc).

Fortunately if we read the instruction from memory we can extract the offset of
a branch and calculate the target address.

This adds a function power_pmu_bhrb_to() to calculate the target/to address of
the corresponding I and B form branches.  It handles branches in both user and
kernel spaces.  It also plumbs this into the perf brhb reading code.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/pmu: Fix order of interpreting BHRB target entries
Michael Neuling [Mon, 13 May 2013 18:44:57 +0000 (18:44 +0000)]
powerpc/pmu: Fix order of interpreting BHRB target entries

The current Branch History Rolling Buffer (BHRB) code misinterprets the order
of entries in the hardware buffer.  It assumes that a branch target address
will be read _after_ its corresponding branch.  In reality the branch target
comes before (lower mfbhrb entry) it's corresponding branch.

This is a rewrite of the code to take this into account.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/perf: Move BHRB code into CONFIG_PPC64 region
Michael Neuling [Mon, 13 May 2013 18:44:56 +0000 (18:44 +0000)]
powerpc/perf: Move BHRB code into CONFIG_PPC64 region

The new Branch History Rolling buffer (BHRB) code is only useful on 64bit
processors, so move it into the #ifdef CONFIG_PPC64 region.

This avoids code bloat on 32bit systems.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: select HAVE_CONTEXT_TRACKING for pSeries
Li Zhong [Mon, 13 May 2013 16:16:44 +0000 (16:16 +0000)]
powerpc: select HAVE_CONTEXT_TRACKING for pSeries

Start context tracking support from pSeries.

Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Use the new schedule_user API on userspace preemption
Li Zhong [Mon, 13 May 2013 16:16:43 +0000 (16:16 +0000)]
powerpc: Use the new schedule_user API on userspace preemption

This patch corresponds to
[PATCH] x86: Use the new schedule_user API on userspace preemption
  commit 0430499ce9d78691f3985962021b16bf8f8a8048

Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Exit user context on notify resume
Li Zhong [Mon, 13 May 2013 16:16:42 +0000 (16:16 +0000)]
powerpc: Exit user context on notify resume

This patch allows RCU usage in do_notify_resume, e.g. signal handling.
It corresponds to
[PATCH] x86: Exit RCU extended QS on notify resume
  commit edf55fda35c7dc7f2d9241c3abaddaf759b457c6

Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Exception hooks for context tracking subsystem
Li Zhong [Mon, 13 May 2013 16:16:41 +0000 (16:16 +0000)]
powerpc: Exception hooks for context tracking subsystem

This is the exception hooks for context tracking subsystem, including
data access, program check, single step, instruction breakpoint, machine check,
alignment, fp unavailable, altivec assist, unknown exception, whose handlers
might use RCU.

This patch corresponds to
[PATCH] x86: Exception hooks for userspace RCU extended QS
  commit 6ba3c97a38803883c2eee489505796cb0a727122

But after the exception handling moved to generic code, and some changes in
following two commits:
56dd9470d7c8734f055da2a6bac553caf4a468eb
  context_tracking: Move exception handling to generic code
6c1e0256fad84a843d915414e4b5973b7443d48d
  context_tracking: Restore correct previous context state on exception exit

it is able for exception hooks to use the generic code above instead of a
redundant arch implementation.

Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Syscall hooks for context tracking subsystem
Li Zhong [Mon, 13 May 2013 16:16:40 +0000 (16:16 +0000)]
powerpc: Syscall hooks for context tracking subsystem

This is the syscall slow path hooks for context tracking subsystem,
corresponding to
[PATCH] x86: Syscall hooks for userspace RCU extended QS
  commit bf5a3c13b939813d28ce26c01425054c740d6731

TIF_MEMDIE is moved to the second 16-bits (with value 17), as it seems there
is no asm code using it. TIF_NOHZ is added to _TIF_SYCALL_T_OR_A, so it is
better for it to be in the same 16 bits with others in the group, so in the
asm code, andi. with this group could work.

Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/booke64: Fix kernel hangs at kernel_dbg_exc
Scott Wood [Mon, 13 May 2013 14:14:53 +0000 (14:14 +0000)]
powerpc/booke64: Fix kernel hangs at kernel_dbg_exc

MSR_DE is not cleared on entry to the kernel, and we don't clear it
explicitly outside of debug code.  If we have MSR_DE set in
prime_debug_regs(), and the new thread has events enabled in DBCR0
(e.g.  ICMP is set in thread->dbsr0, even though it was cleared in the
real DBCR0 when the thread got scheduled out), we'll end up taking a
debug exception in the kernel when DBCR0 is loaded.  DSRR0 will not
point to an exception vector, and the kernel ends up hanging at
kernel_dbg_exc.  Fix this by always clearing MSR_DE when we load new
debug state.

Another observed source of kernel_dbg_exc hangs is with the branch
taken event.  If this event is active, but we take a non-debug trap
(e.g. a TLB miss or an asynchronous interrupt) before the next branch.
We end up taking a branch-taken debug exception on the initial branch
instruction of the exception vector, but because the debug exception is
DBSR_BT rather than DBSR_IC we branch to kernel_dbg_exc before even
checking the DSRR0 address.  Fix this by checking for DBSR_BT as well
as DBSR_IC, which is what 32-bit does and what the comments suggest was
intended in the 64-bit code as well.

Signed-off-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Fix irq_set_affinity() return values
Alexander Gordeev [Mon, 13 May 2013 00:57:49 +0000 (00:57 +0000)]
powerpc: Fix irq_set_affinity() return values

Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Provide __bswapdi2
David Woodhouse [Mon, 13 May 2013 00:23:38 +0000 (00:23 +0000)]
powerpc: Provide __bswapdi2

Some versions of GCC apparently expect this to be provided by libgcc.

Updates from Mikey to fix 32 bit version and adding "r" to registers.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/powernv: Fix starting of secondary CPUs on OPALv2 and v3
Benjamin Herrenschmidt [Tue, 14 May 2013 05:12:31 +0000 (15:12 +1000)]
powerpc/powernv: Fix starting of secondary CPUs on OPALv2 and v3

The current code fails to handle kexec on OPALv2. This fixes it
and adds code to improve the situation on OPALv3 where we can
query the CPU status from the firmware and decide what to do
based on that.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/powernv: Detect OPAL v3 API version
Benjamin Herrenschmidt [Tue, 14 May 2013 05:10:02 +0000 (15:10 +1000)]
powerpc/powernv: Detect OPAL v3 API version

Future firmwares will support that new version

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agoACPI / LPSS: register clock device for Lynxpoint DMA properly
Rafael J. Wysocki [Wed, 8 May 2013 08:55:49 +0000 (11:55 +0300)]
ACPI / LPSS: register clock device for Lynxpoint DMA properly

The DMA controller in Lynxpoint is enumerated as a regular ACPI device now. To
work properly it is using the LPSS root clock as a functional clock. That's why
we have to register the clock device accordingly to the ACPI ID of the DMA
controller. The acpi_lpss.c module is responsible to do the job.

This patch also removes hardcoded name of the DMA device in clk-lpt.c and the
name of the root clock in acpi_lpss.c.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
12 years agodma: acpi-dma: parse CSRT to extract additional resources
Andy Shevchenko [Wed, 8 May 2013 08:55:48 +0000 (11:55 +0300)]
dma: acpi-dma: parse CSRT to extract additional resources

Since we have CSRT only to get additional DMA controller resources, let's get
rid of drivers/acpi/csrt.c and move its logic inside ACPI DMA helpers code.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
12 years agopowerpc: Fix MAX_STACK_TRACE_ENTRIES too low warning again
Li Zhong [Mon, 6 May 2013 22:44:41 +0000 (22:44 +0000)]
powerpc: Fix MAX_STACK_TRACE_ENTRIES too low warning again

Saw this warning again, and this time from the ret_from_fork path.

It seems we could clear the back chain earlier in copy_thread(), which
could cover both path, and also fix potential lockdep usage in
schedule_tail(), or exception occurred before we clear the back chain.

Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Make CONFIG_RTAS_PROC depend on CONFIG_PROC_FS
Michael Ellerman [Mon, 6 May 2013 18:43:39 +0000 (18:43 +0000)]
powerpc: Make CONFIG_RTAS_PROC depend on CONFIG_PROC_FS

We are getting build errors with CONFIG_PROC_FS=n:

arch/powerpc/kernel/rtas_flash.c
   In function 'rtas_flash_init':
  745:33: error: unused variable 'f' [-Werror=unused-variable]

But rtas_flash.c should not be built when CONFIG_PROC_FS=n, beacause all
it does is provide a /proc interface to the RTAS flash routines.

CONFIG_RTAS_FLASH already depends on CONFIG_RTAS_PROC, to indicate that
it depends on the RTAS proc support, but CONFIG_RTAS_PROC does not
depend on CONFIG_PROC_FS. So fix that.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Bring all threads online prior to migration/hibernation
Robert Jennings [Tue, 7 May 2013 04:34:11 +0000 (04:34 +0000)]
powerpc: Bring all threads online prior to migration/hibernation

This patch brings online all threads which are present but not online
prior to migration/hibernation.  After migration/hibernation those
threads are taken back offline.

During migration/hibernation all online CPUs must call H_JOIN, this is
required by the hypervisor.  Without this patch, threads that are offline
(H_CEDE'd) will not be woken to make the H_JOIN call and the OS will be
deadlocked (all threads either JOIN'd or CEDE'd).

Cc: <stable@kernel.org>
Signed-off-by: Robert Jennings <rcj@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/rtas_flash: Fix validate_flash buffer overflow issue
Vasant Hegde [Tue, 7 May 2013 16:54:47 +0000 (16:54 +0000)]
powerpc/rtas_flash: Fix validate_flash buffer overflow issue

ibm,validate-flash-image RTAS call output buffer contains 150 - 200
bytes of data on latest system. Presently we have output
buffer size as 64 bytes and we use sprintf to copy data from
RTAS buffer to local buffer. This causes kernel oops (see below
call trace).

This patch increases local buffer size to 256 and also uses
snprintf instead of sprintf to copy data from RTAS buffer.

Kernel call trace :
-------------------
Oops: Kernel access of bad area, sig: 11 [#1]
SMP NR_CPUS=1024 NUMA pSeries
Modules linked in: nfs fscache lockd auth_rpcgss nfs_acl sunrpc fuse loop dm_mod ipv6 ipv6_lib usb_storage ehea(X) sr_mod qlge ses cdrom enclosure st be2net sg ext3 jbd mbcache usbhid hid ohci_hcd ehci_hcd usbcore qla2xxx usb_common sd_mod crc_t10dif scsi_dh_hp_sw scsi_dh_rdac scsi_dh_alua scsi_dh_emc scsi_dh lpfc scsi_transport_fc scsi_tgt ipr(X) libata scsi_mod
Supported: Yes
NIP: 4520323031333130 LR: 4520323031333130 CTR: 0000000000000000
REGS: c0000001b91779b0 TRAP: 0400   Tainted: G            X  (3.0.13-0.27-ppc64)
MSR: 8000000040009032 <EE,ME,IR,DR>  CR: 44022488  XER: 20000018
TASK = c0000001bca1aba0[4736] 'cat' THREAD: c0000001b9174000 CPU: 36
GPR00: 4520323031333130 c0000001b9177c30 c000000000f87c98 000000000000009b
GPR04: c0000001b9177c4a 000000000000000b 3520323031333130 2032303133313031
GPR08: 3133313031350a4d 000000000000009b 0000000000000000 c0000000003664a4
GPR12: 0000000022022448 c000000003ee6c00 0000000000000002 00000000100e8a90
GPR16: 00000000100cb9d8 0000000010093370 000000001001d310 0000000000000000
GPR20: 0000000000008000 00000000100fae60 000000000000005e 0000000000000000
GPR24: 0000000010129350 46573738302e3030 2046573738302e30 300a4d4720323031
GPR28: 333130313520554e 4b4e4f574e0a4d47 2032303133313031 3520323031333130
NIP [4520323031333130] 0x4520323031333130
LR [4520323031333130] 0x4520323031333130
Call Trace:
[c0000001b9177c30] [4520323031333130] 0x4520323031333130 (unreliable)
Instruction dump:
XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX
XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX

Signed-off-by: Vasant Hegde <hegdevasant@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/kexec: Fix kexec when using VMX optimised memcpy
Anton Blanchard [Sun, 12 May 2013 15:04:53 +0000 (15:04 +0000)]
powerpc/kexec: Fix kexec when using VMX optimised memcpy

commit b3f271e86e5a (powerpc: POWER7 optimised memcpy using VMX and
enhanced prefetch) uses VMX when it is safe to do so (ie not in
interrupt). It also looks at the task struct to decide if we have to
save the current tasks' VMX state.

kexec calls memcpy() at a point where the task struct may have been
overwritten by the new kexec segments. If it has been overwritten
then when memcpy -> enable_altivec looks up current->thread.regs->msr
we get a cryptic oops or lockup.

I also notice we aren't initialising thread_info->cpu, which means
smp_processor_id is broken. Fix that too.

Signed-off-by: Anton Blanchard <anton@samba.org>
Cc: <stable@vger.kernel.org> # 3.6+
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Fix build errors STRICT_MM_TYPECHECKS
Aneesh Kumar K.V [Mon, 6 May 2013 10:51:00 +0000 (10:51 +0000)]
powerpc: Fix build errors STRICT_MM_TYPECHECKS

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/mm: Use the correct mask value when looking at pgtable address
Aneesh Kumar K.V [Sat, 11 May 2013 22:33:19 +0000 (22:33 +0000)]
powerpc/mm: Use the correct mask value when looking at pgtable address

Our pgtable are 2*sizeof(pte_t)*PTRS_PER_PTE which is PTE_FRAG_SIZE.
Instead of depending on frag size, mask with PMD_MASKED_BITS.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agocrypto: caam - fix inconsistent assoc dma mapping direction
Horia Geanta [Fri, 10 May 2013 12:08:39 +0000 (15:08 +0300)]
crypto: caam - fix inconsistent assoc dma mapping direction

req->assoc is dma mapped BIDIRECTIONAL and unmapped TO_DEVICE.
Since it is read-only for the device, use TO_DEVICE both for mapping
and unmapping.

Cc: <stable@vger.kernel.org> # 3.9, 3.8
Signed-off-by: Horia Geanta <horia.geanta@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
12 years agoMerge tag 'fixes-for-3.10-rc2-tag' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 14 May 2013 02:03:49 +0000 (19:03 -0700)]
Merge tag 'fixes-for-3.10-rc2-tag' of git://git./linux/kernel/git/sstabellini/xen

Pull Xen/arm fixes from Stefano Stabellini:
 "This contains a couple of Xen on ARM initialization fixes and a patch
  to improve error handling"

* tag 'fixes-for-3.10-rc2-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/sstabellini/xen:
  xen/arm: rename xen_secondary_init and run it on every online cpu
  xen/arm: do not handle VCPUOP_register_vcpu_info failures
  xen/arm: initialize pm functions later

12 years agoPM / hibernate: Correct documentation
Borislav Petkov [Mon, 13 May 2013 12:00:03 +0000 (12:00 +0000)]
PM / hibernate: Correct documentation

Correct the meaning of PM_HIBERNATION_PREPARE in the docs.

References: http://lkml.kernel.org/r/20130512162717.GA6305@pd.tnic
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
12 years agoMerge branch 'parisc-for-3.10' of git://git.kernel.org/pub/scm/linux/kernel/git/delle...
Linus Torvalds [Mon, 13 May 2013 23:49:59 +0000 (16:49 -0700)]
Merge branch 'parisc-for-3.10' of git://git./linux/kernel/git/deller/parisc-linux

Pull parisc update from Helge Deller:
 "The second round of parisc updates for 3.10 includes build fixes and
  enhancements to utilize irq stacks, fixes SMP races when updating PTE
  and TLB entries by proper locking and makes the search for the correct
  cross compiler more robust on Debian and Gentoo."

* 'parisc-for-3.10' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: make default cross compiler search more robust (v3)
  parisc: fix SMP races when updating PTE and TLB entries in entry.S
  parisc: implement irq stacks - part 2 (v2)

12 years agoPM / Documentation: remove inaccurate suspend/hibernate transition lantency statement
Zhang Rui [Mon, 13 May 2013 02:42:12 +0000 (02:42 +0000)]
PM / Documentation: remove inaccurate suspend/hibernate transition lantency statement

The lantency of the transition from suspend and hibernate is
platform-dependent.  Thus we should not refer the lantency in the
documentation.

Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
12 years agoPM: Documentation update for freeze state
Zhang Rui [Mon, 13 May 2013 02:42:11 +0000 (02:42 +0000)]
PM: Documentation update for freeze state

Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
12 years agocpufreq / intel_pstate: use vzalloc() instead of vmalloc()/memset(0)
Wei Yongjun [Mon, 13 May 2013 08:03:43 +0000 (08:03 +0000)]
cpufreq / intel_pstate: use vzalloc() instead of vmalloc()/memset(0)

Use vzalloc() instead of vmalloc() and memset(0).

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
12 years agoARM: 7720/1: ARM v6/v7 cmpxchg64 shouldn't clear upper 32 bits of the old/new value
Jaccon Bastiaansen [Mon, 13 May 2013 16:28:27 +0000 (17:28 +0100)]
ARM: 7720/1: ARM v6/v7 cmpxchg64 shouldn't clear upper 32 bits of the old/new value

The implementation of cmpxchg64() for the ARM v6 and v7 architecture
casts parameter 2 and 3 (the old and new 64bit values) to an unsigned
long before calling the atomic_cmpxchg64() function. This clears
the top 32 bits of the old and new values, resulting in the wrong
values being compare-exchanged. Luckily, this only appears to be used
for 64-bit sched_clock, which we don't (yet) have on ARM.

This bug was introduced by commit 3e0f5a15f500 ("ARM: 7404/1: cmpxchg64:
use atomic64 and local64 routines for cmpxchg64").

Cc: <stable@vger.kernel.org>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Jaccon Bastiaansen <jaccon.bastiaansen@gmail.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
12 years agostaging: nvec: remove inline marking of EXPORT_SYMBOL functions
Denis Efremov [Thu, 9 May 2013 09:19:37 +0000 (13:19 +0400)]
staging: nvec: remove inline marking of EXPORT_SYMBOL functions

EXPORT_SYMBOL and inline directives are contradictory to each other.
The patch fixes this inconsistency.

Found by Linux Driver Verification project (linuxtesting.org).

Signed-off-by: Denis Efremov <yefremov.denis@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
12 years agostaging: comedi: complain if dma buffer allocation not supported
Ian Abbott [Fri, 10 May 2013 13:07:16 +0000 (14:07 +0100)]
staging: comedi: complain if dma buffer allocation not supported

When allocating a buffer to support asynchronous comedi commands, if a
DMA coherent buffer was requested but `CONFIG_HAS_DMA` is undefined,
bail out of local helper function `__comedi_buf_alloc()` with an error
message.

Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
12 years agostaging: comedi: ni_mio_common: only do counter commands for ni_pcimio
Ian Abbott [Fri, 10 May 2013 13:07:17 +0000 (14:07 +0100)]
staging: comedi: ni_mio_common: only do counter commands for ni_pcimio

"ni_mio_common.c" holds common code included by "ni_pcimio.c",
"ni_atmio.c" and "ni_mio_cs.c", including a common initialization
function `ni_E_init()`.  Amongst other things, this initializes some
counter subdevices to support comedi instructions and asynchronous
commands.  However, even though it sets up the handlers to support
asynchronous commands on these subdevices, the handlers will return an
error unless the `PCIDMA` macro is defined (which is defined only in
"ni_pcimio.c").  If the `PCIDMA` macro is not defined, the comedi core
will needlessly allocate buffers to support the asynchronous commands.
Also, `s->async_dma_dir` is set to `DMA_BIDIRECTIONAL`, causing the
physical pages for the buffers to be allocated using
`dma_alloc_coherent()`.

If the comedi core cannot call `dma_alloc_coherent()` because
`CONFIG_HAS_DMA` is not defined, it will fail to allocate the buffers,
which ultimately causes `ni_E_init()` to fail.

Avoid the wastage and prevent the failure by only setting up
asynchronous command support for the counter subdevices if the `PCIDMA`
macro is defined.

Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
12 years agostaging: comedi: work without HAS_DMA
Ian Abbott [Fri, 10 May 2013 13:07:15 +0000 (14:07 +0100)]
staging: comedi: work without HAS_DMA

The core "comedi" module and the "mite" helper module for NI PCI devices
both have calls to `dma_alloc_coherent()` and `dma_free_coherent()`.
Those functions are only available if `CONFIG_HAS_DMA` is defined.

Apart from the "mite" module, the functions are only called for comedi
drivers that set `s->async_dma_dir` (where `s` is a pointer to a `struct
comedi_subdevice`) to anything other than `DMA_NONE`.

Change local helper functions `__comedi_buf_alloc()` and
`__comedi_buf_free()` to only call `dma_alloc_coherent()` and
`dma_free_coherent()` if `CONFIG_HAS_DMA` is defined.

Change the "Kconfig" to make the following configuration options depend
on `HAS_DMA`:

`COMEDI_MITE` - builds the "mite" module.
`COMEDI_NI_6527` - selects `COMEDI_MITE`.
`COMEDI_NI_65XX` - selects `COMEDI_MITE`.
`COMEDI_NI_670X` - selects `COMEDI_MITE`.
`COMEDI_NI_LABPC_PCI` - selects `COMEDI_MITE`.
`COMEDI_NI_PCIDIO` - selects `COMEDI_MITE`.
`COMEDI_NI_TIOCMD` - selects `COMEDI_MITE`.
`COMEDI_NI_660X` - selects `COMEDI_NI_TIOCMD`,
                   sets `s->async_dma_dir`.
`COMEDI_NI_PCIMIO` - selects `COMEDI_NI_TIOCMD`,
                     sets `s->async_dma_dir`.

Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>