~shefty/rdma-dev.git
8 years agobacklight: ld9040: use spi_get_drvdata and spi_set_drvdata
Jingoo Han [Fri, 22 Feb 2013 00:43:33 +0000 (16:43 -0800)]
backlight: ld9040: use spi_get_drvdata and spi_set_drvdata

Use the wrapper functions for getting and setting the driver data using
spi_device instead of using dev_{get|set}_drvdata with &spi->dev, so we
can directly pass a struct spi_device.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: l4f00242t03: use spi_get_drvdata and spi_set_drvdata
Jingoo Han [Fri, 22 Feb 2013 00:43:32 +0000 (16:43 -0800)]
backlight: l4f00242t03: use spi_get_drvdata and spi_set_drvdata

Use the wrapper functions for getting and setting the driver data using
spi_device instead of using dev_{get|set}_drvdata with &spi->dev, so we
can directly pass a struct spi_device.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: ams369fg06: reorder inclusions of <linux/xxx.h>
Jingoo Han [Fri, 22 Feb 2013 00:43:31 +0000 (16:43 -0800)]
backlight: ams369fg06: reorder inclusions of <linux/xxx.h>

Reorder inclusions of <linux/xxx.h> for redability, according to
alphabetical ordering.  Also, unnecessary header comments are removed.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: ams369fg06: remove redundant variable 'before_power'
Jingoo Han [Fri, 22 Feb 2013 00:43:30 +0000 (16:43 -0800)]
backlight: ams369fg06: remove redundant variable 'before_power'

'before_power' was used to check the previous status when resume() is
called.  However, FB_BLANK_POWERDOWN was used in suspend() all the time,
so there is no need to check the previous status.  Also, redundant return
variables are removed to reduce the code.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: ams369fg06: replace EFAULT with EINVAL
Jingoo Han [Fri, 22 Feb 2013 00:43:29 +0000 (16:43 -0800)]
backlight: ams369fg06: replace EFAULT with EINVAL

Replace EFAULT with EINVAL, because EFAULT tends to be for the invalid
memory addresses.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: ams369fg06: remove unnecessary NULL deference check
Jingoo Han [Fri, 22 Feb 2013 00:43:28 +0000 (16:43 -0800)]
backlight: ams369fg06: remove unnecessary NULL deference check

Remove unnecessary NULL deference check, because it was already checked in
ams369fg06_probe().  Also, unnecessary parentheses are removed in
ams369fg06_power_is_on().

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: ams369fg06: use sleep instead of delay
Jingoo Han [Fri, 22 Feb 2013 00:43:27 +0000 (16:43 -0800)]
backlight: ams369fg06: use sleep instead of delay

Replace mdelay with msleep to remove the busy loop waiting.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: s6e63m0: reorder inclusions of <linux/xxx.h>
Jingoo Han [Fri, 22 Feb 2013 00:43:26 +0000 (16:43 -0800)]
backlight: s6e63m0: reorder inclusions of <linux/xxx.h>

Reorder inclusions of <linux/xxx.h> for redability, according to
alphabetical ordering.  Also, unnecessary header comments are removed.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: s6e63m0: remove redundant variable 'before_power'
Jingoo Han [Fri, 22 Feb 2013 00:43:25 +0000 (16:43 -0800)]
backlight: s6e63m0: remove redundant variable 'before_power'

'before_power' was used to check the previous status when resume() is
called.  However, FB_BLANK_POWERDOWN was used in suspend() all the time,
so there is no need to check the previous status.  Also, redundant return
variables are removed to reduce the code.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: s6e63m0: replace EFAULT with EINVAL
Jingoo Han [Fri, 22 Feb 2013 00:43:24 +0000 (16:43 -0800)]
backlight: s6e63m0: replace EFAULT with EINVAL

Replace EFAULT with EINVAL, because EFAULT tends to be for the invalid
memory addresses.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: s6e63m0: remove unnecessary NULL deference check
Jingoo Han [Fri, 22 Feb 2013 00:43:23 +0000 (16:43 -0800)]
backlight: s6e63m0: remove unnecessary NULL deference check

Remove unnecessary NULL deference check, because it was already checked in
s6e63m0_probe().  Also, POWER_IS_ON is replaced with
s6e63m0_power_is_on().

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: s6e63m0: use sleep instead of delay
Jingoo Han [Fri, 22 Feb 2013 00:43:21 +0000 (16:43 -0800)]
backlight: s6e63m0: use sleep instead of delay

Replace mdelay with msleep to remove the busy loop waiting.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: s6e63m0: use lowercase names of structs
Jingoo Han [Fri, 22 Feb 2013 00:43:20 +0000 (16:43 -0800)]
backlight: s6e63m0: use lowercase names of structs

Lowercase names of structs should be used, because they are
not preprocessor macros.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: ld9040: reorder inclusions of <linux/xxx.h>
Jingoo Han [Fri, 22 Feb 2013 00:43:19 +0000 (16:43 -0800)]
backlight: ld9040: reorder inclusions of <linux/xxx.h>

Reorder inclusions of <linux/xxx.h> for redability, according to
alphabetical ordering.  Also, unnecessary header comments are removed.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: ld9040: remove redundant return variables
Jingoo Han [Fri, 22 Feb 2013 00:43:18 +0000 (16:43 -0800)]
backlight: ld9040: remove redundant return variables

Redundant return variables are removed to reduce the code.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: ld9040: replace EFAULT with EINVAL
Jingoo Han [Fri, 22 Feb 2013 00:43:17 +0000 (16:43 -0800)]
backlight: ld9040: replace EFAULT with EINVAL

Replace EFAULT with EINVAL, because EFAULT tends to be for the invalid
memory addresses.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: ld9040: remove unnecessary NULL deference check
Jingoo Han [Fri, 22 Feb 2013 00:43:16 +0000 (16:43 -0800)]
backlight: ld9040: remove unnecessary NULL deference check

Removee unnecessary NULL deference check, because it was already checked
in ld9040_probe().  Also, power_is_on is replaced with
ld9040_power_is_on().

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: ld9040: use sleep instead of delay
Jingoo Han [Fri, 22 Feb 2013 00:43:15 +0000 (16:43 -0800)]
backlight: ld9040: use sleep instead of delay

Replace mdelay with msleep to remove the busy loop waiting.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobacklight: add lms501kf03 LCD driver
Jingoo Han [Fri, 22 Feb 2013 00:43:14 +0000 (16:43 -0800)]
backlight: add lms501kf03 LCD driver

Add the lms501kf03 LCD panel driver.  The lms501kf03 LCD panel (800 x 480)
driver uses 3-wired SPI inteface.

[akpm@linux-foundation.org: remove unused variable `before_power']
[akpm@linux-foundation.org: make lms501kf03_shutdown() static, per Fengguang]
Signed-off-by: Ilho Lee <Ilho215.lee@samsung.com>
Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Cc: "devendra.aaru" <devendra.aaru@gmail.com>
Cc: Sachin Kamat <sachin.kamat@linaro.org>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoMAINTAINERS: mm: add additional include files to listing
Cody P Schafer [Fri, 22 Feb 2013 00:43:13 +0000 (16:43 -0800)]
MAINTAINERS: mm: add additional include files to listing

Add gfp.h, mmzone.h, memory_hotplug.h & vmalloc.h to the "MEMORY
MANAGMENT" section so scripts/get_maintainer.pl can do a better job of
making recommendations.

Signed-off-by: Cody P Schafer <cody@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoget_maintainer.pl: find maintainers for removed files
Geert Uytterhoeven [Fri, 22 Feb 2013 00:43:12 +0000 (16:43 -0800)]
get_maintainer.pl: find maintainers for removed files

For removed files, get_maintainer.pl doesn't find any maintainers (besides
the default linux-kernel@vger.kernel.org), as it only looks at the "+++"
lines, which are "/dev/null" for removals.  Fix this by extending the
parsing to the "---" lines.

E.g. for the two line test patch below the real score maintainers will now
be found:

    --- a/arch/score/include/asm/dma-mapping.h
    +++ /dev/null

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoprintk: add pr_devel_once and pr_devel_ratelimited
Mikhail Gruzdev [Fri, 22 Feb 2013 00:43:10 +0000 (16:43 -0800)]
printk: add pr_devel_once and pr_devel_ratelimited

Standardize pr_devel logging macros family by adding pr_devel_once and
pr_devel_ratelimited.

Signed-off-by: Mikhail Gruzdev <michail.gruzdev@gmail.com>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agolib/vsprintf.c: add %pa format specifier for phys_addr_t types
Stepan Moskovchenko [Fri, 22 Feb 2013 00:43:09 +0000 (16:43 -0800)]
lib/vsprintf.c: add %pa format specifier for phys_addr_t types

Add the %pa format specifier for printing a phys_addr_t type and its
derivative types (such as resource_size_t), since the physical address
size on some platforms can vary based on build options, regardless of
the native integer type.

Signed-off-by: Stepan Moskovchenko <stepanm@codeaurora.org>
Cc: Rob Landley <rob@landley.net>
Cc: George Spelvin <linux@horizon.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agosys_prctl(): coding-style cleanup
Andrew Morton [Fri, 22 Feb 2013 00:43:07 +0000 (16:43 -0800)]
sys_prctl(): coding-style cleanup

Remove a tabstop from the switch statement, in the usual fashion.  A few
instances of weirdwrapping were removed as a result.

Cc: Chen Gang <gang.chen@asianux.com>
Cc: Cyrill Gorcunov <gorcunov@openvz.org>
Acked-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agosys_prctl(): arg2 is unsigned long which is never < 0
Chen Gang [Fri, 22 Feb 2013 00:43:06 +0000 (16:43 -0800)]
sys_prctl(): arg2 is unsigned long which is never < 0

arg2 will never < 0, for its type is 'unsigned long'

Also, use the provided macros.

Signed-off-by: Chen Gang <gang.chen@asianux.com>
Reported-by: Cyrill Gorcunov <gorcunov@openvz.org>
Acked-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agosun.com documentation fixes
Christian Kujau [Fri, 22 Feb 2013 00:43:05 +0000 (16:43 -0800)]
sun.com documentation fixes

After I came across a help text for SUNGEM mentioning a broken sun.com
URL, I felt like fixing those up, as they are now pointing to oracle.com
URLs.

Signed-off-by: Christian Kujau <lists@nerdbynature.de>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agolib/Kconfig.debug: unhide CONFIG_PANIC_ON_OOPS
Kyle McMartin [Fri, 22 Feb 2013 00:43:04 +0000 (16:43 -0800)]
lib/Kconfig.debug: unhide CONFIG_PANIC_ON_OOPS

CONFIG_EXPERT doesn't really make sense, and hides it unintentionally.
Remove superfluous "default n" pointed out by Ingo as well.

Signed-off-by: Kyle McMartin <kyle@redhat.com>
Acked-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agosmp: make smp_call_function_many() use logic similar to smp_call_function_single()
Shaohua Li [Fri, 22 Feb 2013 00:43:03 +0000 (16:43 -0800)]
smp: make smp_call_function_many() use logic similar to smp_call_function_single()

I'm testing swapout workload in a two-socket Xeon machine.  The workload
has 10 threads, each thread sequentially accesses separate memory
region.  TLB flush overhead is very big in the workload.  For each page,
page reclaim need move it from active lru list and then unmap it.  Both
need a TLB flush.  And this is a multthread workload, TLB flush happens
in 10 CPUs.  In X86, TLB flush uses generic smp_call)function.  So this
workload stress smp_call_function_many heavily.

Without patch, perf shows:
+  24.49%  [k] generic_smp_call_function_interrupt
-  21.72%  [k] _raw_spin_lock
   - _raw_spin_lock
      + 79.80% __page_check_address
      + 6.42% generic_smp_call_function_interrupt
      + 3.31% get_swap_page
      + 2.37% free_pcppages_bulk
      + 1.75% handle_pte_fault
      + 1.54% put_super
      + 1.41% grab_super_passive
      + 1.36% __swap_duplicate
      + 0.68% blk_flush_plug_list
      + 0.62% swap_info_get
+   6.55%  [k] flush_tlb_func
+   6.46%  [k] smp_call_function_many
+   5.09%  [k] call_function_interrupt
+   4.75%  [k] default_send_IPI_mask_sequence_phys
+   2.18%  [k] find_next_bit

swapout throughput is around 1300M/s.

With the patch, perf shows:
-  27.23%  [k] _raw_spin_lock
   - _raw_spin_lock
      + 80.53% __page_check_address
      + 8.39% generic_smp_call_function_single_interrupt
      + 2.44% get_swap_page
      + 1.76% free_pcppages_bulk
      + 1.40% handle_pte_fault
      + 1.15% __swap_duplicate
      + 1.05% put_super
      + 0.98% grab_super_passive
      + 0.86% blk_flush_plug_list
      + 0.57% swap_info_get
+   8.25%  [k] default_send_IPI_mask_sequence_phys
+   7.55%  [k] call_function_interrupt
+   7.47%  [k] smp_call_function_many
+   7.25%  [k] flush_tlb_func
+   3.81%  [k] _raw_spin_lock_irqsave
+   3.78%  [k] generic_smp_call_function_single_interrupt

swapout throughput is around 1400M/s.  So there is around a 7%
improvement, and total cpu utilization doesn't change.

Without the patch, cfd_data is shared by all CPUs.
generic_smp_call_function_interrupt does read/write cfd_data several times
which will create a lot of cache ping-pong.  With the patch, the data
becomes per-cpu.  The ping-pong is avoided.  And from the perf data, this
doesn't make call_single_queue lock contend.

Next step is to remove generic_smp_call_function_interrupt() from arch
code.

Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm: use vm_unmapped_area() on alpha architecture
Michel Lespinasse [Fri, 22 Feb 2013 00:43:01 +0000 (16:43 -0800)]
mm: use vm_unmapped_area() on alpha architecture

Update the alpha arch_get_unmapped_area function to make use of
vm_unmapped_area() instead of implementing a brute force search.

Signed-off-by: Michel Lespinasse <walken@google.com>
Acked-by: Rik van Riel <riel@redhat.com>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Matt Turner <mattst88@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoubifs: wait for page writeback to provide stable pages
Jan Kara [Fri, 22 Feb 2013 00:42:59 +0000 (16:42 -0800)]
ubifs: wait for page writeback to provide stable pages

When stable pages are required, we have to wait if the page is just
going to disk and we want to modify it.  Add proper callback to
ubifs_vm_page_mkwrite().

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Cc: Artem Bityutskiy <dedekind1@gmail.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Eric Van Hensbergen <ericvh@gmail.com>
Cc: Ron Minnich <rminnich@sandia.gov>
Cc: Latchesar Ionkov <lucho@ionkov.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoocfs2: wait for page writeback to provide stable pages
Jan Kara [Fri, 22 Feb 2013 00:42:57 +0000 (16:42 -0800)]
ocfs2: wait for page writeback to provide stable pages

When stable pages are required, we have to wait if the page is just
going to disk and we want to modify it.  Add proper callback to
ocfs2_grab_pages_for_write().

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Acked-by: Joel Becker <jlbec@evilplan.org>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Artem Bityutskiy <dedekind1@gmail.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Eric Van Hensbergen <ericvh@gmail.com>
Cc: Ron Minnich <rminnich@sandia.gov>
Cc: Latchesar Ionkov <lucho@ionkov.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoblock: optionally snapshot page contents to provide stable pages during write
Darrick J. Wong [Fri, 22 Feb 2013 00:42:55 +0000 (16:42 -0800)]
block: optionally snapshot page contents to provide stable pages during write

This provides a band-aid to provide stable page writes on jbd without
needing to backport the fixed locking and page writeback bit handling
schemes of jbd2.  The band-aid works by using bounce buffers to snapshot
page contents instead of waiting.

For those wondering about the ext3 bandage -- fixing the jbd locking
(which was done as part of ext4dev years ago) is a lot of surgery, and
setting PG_writeback on data pages when we actually hold the page lock
dropped ext3 performance by nearly an order of magnitude.  If we're
going to migrate iscsi and raid to use stable page writes, the
complaints about high latency will likely return.  We might as well
centralize their page snapshotting thing to one place.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Tested-by: Andy Lutomirski <luto@amacapital.net>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Artem Bityutskiy <dedekind1@gmail.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Eric Van Hensbergen <ericvh@gmail.com>
Cc: Ron Minnich <rminnich@sandia.gov>
Cc: Latchesar Ionkov <lucho@ionkov.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years ago9pfs: fix filesystem to wait for stable page writeback
Darrick J. Wong [Fri, 22 Feb 2013 00:42:53 +0000 (16:42 -0800)]
9pfs: fix filesystem to wait for stable page writeback

Fix up the ->page_mkwrite handler to provide stable page writes if necessary.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Artem Bityutskiy <dedekind1@gmail.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Eric Van Hensbergen <ericvh@gmail.com>
Cc: Ron Minnich <rminnich@sandia.gov>
Cc: Latchesar Ionkov <lucho@ionkov.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm: only enforce stable page writes if the backing device requires it
Darrick J. Wong [Fri, 22 Feb 2013 00:42:51 +0000 (16:42 -0800)]
mm: only enforce stable page writes if the backing device requires it

Create a helper function to check if a backing device requires stable
page writes and, if so, performs the necessary wait.  Then, make it so
that all points in the memory manager that handle making pages writable
use the helper function.  This should provide stable page write support
to most filesystems, while eliminating unnecessary waiting for devices
that don't require the feature.

Before this patchset, all filesystems would block, regardless of whether
or not it was necessary.  ext3 would wait, but still generate occasional
checksum errors.  The network filesystems were left to do their own
thing, so they'd wait too.

After this patchset, all the disk filesystems except ext3 and btrfs will
wait only if the hardware requires it.  ext3 (if necessary) snapshots
pages instead of blocking, and btrfs provides its own bdi so the mm will
never wait.  Network filesystems haven't been touched, so either they
provide their own stable page guarantees or they don't block at all.
The blocking behavior is back to what it was before 3.0 if you don't
have a disk requiring stable page writes.

Here's the result of using dbench to test latency on ext2:

3.8.0-rc3:
 Operation      Count    AvgLat    MaxLat
 ----------------------------------------
 WriteX        109347     0.028    59.817
 ReadX         347180     0.004     3.391
 Flush          15514    29.828   287.283

Throughput 57.429 MB/sec  4 clients  4 procs  max_latency=287.290 ms

3.8.0-rc3 + patches:
 WriteX        105556     0.029     4.273
 ReadX         335004     0.005     4.112
 Flush          14982    30.540   298.634

Throughput 55.4496 MB/sec  4 clients  4 procs  max_latency=298.650 ms

As you can see, the maximum write latency drops considerably with this
patch enabled.  The other filesystems (ext3/ext4/xfs/btrfs) behave
similarly, but see the cover letter for those results.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Acked-by: Steven Whitehouse <swhiteho@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Artem Bityutskiy <dedekind1@gmail.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Eric Van Hensbergen <ericvh@gmail.com>
Cc: Ron Minnich <rminnich@sandia.gov>
Cc: Latchesar Ionkov <lucho@ionkov.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobdi: allow block devices to say that they require stable page writes
Darrick J. Wong [Fri, 22 Feb 2013 00:42:48 +0000 (16:42 -0800)]
bdi: allow block devices to say that they require stable page writes

This patchset ("stable page writes, part 2") makes some key
modifications to the original 'stable page writes' patchset.  First, it
provides creators (devices and filesystems) of a backing_dev_info a flag
that declares whether or not it is necessary to ensure that page
contents cannot change during writeout.  It is no longer assumed that
this is true of all devices (which was never true anyway).  Second, the
flag is used to relaxed the wait_on_page_writeback calls so that wait
only occurs if the device needs it.  Third, it fixes up the remaining
disk-backed filesystems to use this improved conditional-wait logic to
provide stable page writes on those filesystems.

It is hoped that (for people not using checksumming devices, anyway)
this patchset will give back unnecessary performance decreases since the
original stable page write patchset went into 3.0.  Sorry about not
fixing it sooner.

Complaints were registered by several people about the long write
latencies introduced by the original stable page write patchset.
Generally speaking, the kernel ought to allocate as little extra memory
as possible to facilitate writeout, but for people who simply cannot
wait, a second page stability strategy is (re)introduced: snapshotting
page contents.  The waiting behavior is still the default strategy; to
enable page snapshotting, a superblock flag (MS_SNAP_STABLE) must be
set.  This flag is used to bandaid^Henable stable page writeback on
ext3[1], and is not used anywhere else.

Given that there are already a few storage devices and network FSes that
have rolled their own page stability wait/page snapshot code, it would
be nice to move towards consolidating all of these.  It seems possible
that iscsi and raid5 may wish to use the new stable page write support
to enable zero-copy writeout.

Thank you to Jan Kara for helping fix a couple more filesystems.

Per Andrew Morton's request, here are the result of using dbench to measure
latencies on ext2:

3.8.0-rc3:
   Operation      Count    AvgLat    MaxLat
   ----------------------------------------
   WriteX        109347     0.028    59.817
   ReadX         347180     0.004     3.391
   Flush          15514    29.828   287.283

  Throughput 57.429 MB/sec  4 clients  4 procs  max_latency=287.290 ms

3.8.0-rc3 + patches:
   WriteX        105556     0.029     4.273
   ReadX         335004     0.005     4.112
   Flush          14982    30.540   298.634

  Throughput 55.4496 MB/sec  4 clients  4 procs  max_latency=298.650 ms

As you can see, for ext2 the maximum write latency decreases from ~60ms
on a laptop hard disk to ~4ms.  I'm not sure why the flush latencies
increase, though I suspect that being able to dirty pages faster gives
the flusher more work to do.

On ext4, the average write latency decreases as well as all the maximum
latencies:

3.8.0-rc3:
   WriteX         85624     0.152    33.078
   ReadX         272090     0.010    61.210
   Flush          12129    36.219   168.260

  Throughput 44.8618 MB/sec  4 clients  4 procs  max_latency=168.276 ms

3.8.0-rc3 + patches:
   WriteX         86082     0.141    30.928
   ReadX         273358     0.010    36.124
   Flush          12214    34.800   165.689

  Throughput 44.9941 MB/sec  4 clients  4 procs  max_latency=165.722 ms

XFS seems to exhibit similar latency improvements as ext2:

3.8.0-rc3:
   WriteX        125739     0.028   104.343
   ReadX         399070     0.005     4.115
   Flush          17851    25.004   131.390

  Throughput 66.0024 MB/sec  4 clients  4 procs  max_latency=131.406 ms

3.8.0-rc3 + patches:
   WriteX        123529     0.028     6.299
   ReadX         392434     0.005     4.287
   Flush          17549    25.120   188.687

  Throughput 64.9113 MB/sec  4 clients  4 procs  max_latency=188.704 ms

...and btrfs, just to round things out, also shows some latency
decreases:

3.8.0-rc3:
   WriteX         67122     0.083    82.355
   ReadX         212719     0.005     2.828
   Flush           9547    47.561   147.418

  Throughput 35.3391 MB/sec  4 clients  4 procs  max_latency=147.433 ms

3.8.0-rc3 + patches:
   WriteX         64898     0.101    71.631
   ReadX         206673     0.005     7.123
   Flush           9190    47.963   219.034

  Throughput 34.0795 MB/sec  4 clients  4 procs  max_latency=219.044 ms

Before this patchset, all filesystems would block, regardless of whether
or not it was necessary.  ext3 would wait, but still generate occasional
checksum errors.  The network filesystems were left to do their own
thing, so they'd wait too.

After this patchset, all the disk filesystems except ext3 and btrfs will
wait only if the hardware requires it.  ext3 (if necessary) snapshots
pages instead of blocking, and btrfs provides its own bdi so the mm will
never wait.  Network filesystems haven't been touched, so either they
provide their own wait code, or they don't block at all.  The blocking
behavior is back to what it was before 3.0 if you don't have a disk
requiring stable page writes.

This patchset has been tested on 3.8.0-rc3 on x64 with ext3, ext4, and
xfs.  I've spot-checked 3.8.0-rc4 and seem to be getting the same
results as -rc3.

[1] The alternative fixes to ext3 include fixing the locking order and
page bit handling like we did for ext4 (but then why not just use
ext4?), or setting PG_writeback so early that ext3 becomes extremely
slow.  I tried that, but the number of write()s I could initiate dropped
by nearly an order of magnitude.  That was a bit much even for the
author of the stable page series! :)

This patch:

Creates a per-backing-device flag that tracks whether or not pages must
be held immutable during writeout.  Eventually it will be used to waive
wait_for_page_writeback() if nothing requires stable pages.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Artem Bityutskiy <dedekind1@gmail.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Eric Van Hensbergen <ericvh@gmail.com>
Cc: Ron Minnich <rminnich@sandia.gov>
Cc: Latchesar Ionkov <lucho@ionkov.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agolockdep: make lockdep_assert_held() not have a return value
Johannes Berg [Fri, 22 Feb 2013 00:42:47 +0000 (16:42 -0800)]
lockdep: make lockdep_assert_held() not have a return value

I recently made the mistake of writing:

  foo = lockdep_dereference_protected(..., lockdep_assert_held(...));

which is clearly bogus.  If lockdep is disabled in the config this would
cause a compile failure, if it is enabled then it compiles and causes a
puzzling warning about dereferencing without the correct protection.

Wrap the macro in "do { ...  } while (0)" to also fail compile for this
when lockdep is enabled.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoocfs2: unlock super lock if lockres refresh failed
Junxiao Bi [Fri, 22 Feb 2013 00:42:45 +0000 (16:42 -0800)]
ocfs2: unlock super lock if lockres refresh failed

If lockres refresh failed, the super lock will never be released which
will cause some processes on other cluster nodes hung forever.

Signed-off-by: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoocfs2: remove kfree() redundant null checks
Tim Gardner [Fri, 22 Feb 2013 00:42:44 +0000 (16:42 -0800)]
ocfs2: remove kfree() redundant null checks

smatch analysis indicates a number of redundant NULL checks before
calling kfree(), eg:

  fs/ocfs2/alloc.c:6138 ocfs2_begin_truncate_log_recovery() info:
   redundant null check on *tl_copy calling kfree()

  fs/ocfs2/alloc.c:6755 ocfs2_zero_range_for_truncate() info:
   redundant null check on pages calling kfree()

etc....

[akpm@linux-foundation.org: revert dubious change in ocfs2_begin_truncate_log_recovery()]
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Acked-by: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoconfigfs: move the dereference below the NULL test
Wei Yongjun [Fri, 22 Feb 2013 00:42:43 +0000 (16:42 -0800)]
configfs: move the dereference below the NULL test

The dereference should be moved below the NULL test.

spatch with a semantic match is used to found this.
(http://coccinelle.lip6.fr/)

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoscripts/tags.sh: add ctags magic for declarations of popular kernel type
Kirill Tkhai [Fri, 22 Feb 2013 00:42:42 +0000 (16:42 -0800)]
scripts/tags.sh: add ctags magic for declarations of popular kernel type

- Add magic for declarations of variables of popular kernel type like
  spinlock_t, list_head, wait_queue_head_t and other.

- Add a set of specially handled declaration extentions like
  __attribute, __aligned and other.

- Simplify pci_bus_* magic

Signed-off-by: Kirill V Tkhai <tkhai@yandex.ru>
Cc: Michal Marek <mmarek@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agotime: don't inline EXPORT_SYMBOL functions
Greg Kroah-Hartman [Fri, 22 Feb 2013 00:42:40 +0000 (16:42 -0800)]
time: don't inline EXPORT_SYMBOL functions

How is the compiler even handling exported functions that are marked
inline? Anyway, these shouldn't be inline because of that, so remove
that marking.

Based on a larger patch by Mark Charlebois to get LLVM to build the
kernel.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Mark Charlebois <mcharleb@qualcomm.com>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: hank <pyu@redhat.com>
Cc: John Stultz <john.stultz@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agodrivers/video/mx3fb.c: use NULL for pointer
Fabio Estevam [Fri, 22 Feb 2013 00:42:39 +0000 (16:42 -0800)]
drivers/video/mx3fb.c: use NULL for pointer

Fix the following sparse error:

  drivers/video/mx3fb.c:1309:28: warning: Using plain integer as NULL pointer

Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agovideo: exynos_dp: move disable_irq() to exynos_dp_suspend()
Ajay Kumar [Fri, 22 Feb 2013 00:42:38 +0000 (16:42 -0800)]
video: exynos_dp: move disable_irq() to exynos_dp_suspend()

disable_irq() should be moved to exynos_dp_suspend(), because enable_irq()
is called at exynos_dp_resume().

Signed-off-by: Ajay Kumar <ajaykumar.rs@samsung.com>
Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agovideo: exynos_dp: add missing of_node_put()
Jingoo Han [Fri, 22 Feb 2013 00:42:37 +0000 (16:42 -0800)]
video: exynos_dp: add missing of_node_put()

of_find_node_by_name() returns a node pointer with refcount incremented,
use of_node_put() on it when done.

of_find_node_by_name() will call of_node_put() against the node pass to
from parameter, thus we also need to call of_node_get(from) before calling
of_find_node_by_name().

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Ajay Kumar <ajaykumar.rs@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agovideo: s3c-fb: fix typo in definition of VIDCON1_VSTATUS_FRONTPORCH value
Tomasz Figa [Fri, 22 Feb 2013 00:42:36 +0000 (16:42 -0800)]
video: s3c-fb: fix typo in definition of VIDCON1_VSTATUS_FRONTPORCH value

The correct value for VIDCON1_VSTATUS_FRONTPORCH is 3, not 0.

Signed-off-by: Tomasz Figa <t.figa@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agovideo: s3c-fb: add the bit definitions for CSC EQ709 and EQ601
Jingoo Han [Fri, 22 Feb 2013 00:42:34 +0000 (16:42 -0800)]
video: s3c-fb: add the bit definitions for CSC EQ709 and EQ601

Add the bit definitions for CSC EQ709 and EQ601.  These definitons are
used to control the CSC parameter such as equation 709 and equation 601.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Tomasz Figa <t.figa@samsung.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agovideo: s3c-fb: remove unnecessary brackets
Jingoo Han [Fri, 22 Feb 2013 00:42:33 +0000 (16:42 -0800)]
video: s3c-fb: remove unnecessary brackets

Remove unnecessary brackets and the duplicated VIDTCON2 definition.

Also, header comment is modified, because EXYNOS series is supported and
<mach/regs-fb.h> is not available.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Tomasz Figa <t.figa@samsung.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agovideo: s3c-fb: remove duplicated S3C_FB_MAX_WIN
Jingoo Han [Fri, 22 Feb 2013 00:42:31 +0000 (16:42 -0800)]
video: s3c-fb: remove duplicated S3C_FB_MAX_WIN

S3C_FB_MAX_WIN is already defined in 'plat-samsung/include/plat/fb.h'.
So, this definition in 'include/video/samsung_fimd.h' should be removed to
avoid the duplication.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Tomasz Figa <t.figa@samsung.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agovideo: s3c-fb: use ARCH_ dependancy
Jingoo Han [Fri, 22 Feb 2013 00:42:30 +0000 (16:42 -0800)]
video: s3c-fb: use ARCH_ dependancy

Use ARCH_ dependancy when using s3c-fb.  S3C_DEV_FB, S5P_DEV_FIMD0 cannot
be enabled on EXYNOS5.  So, ARCH_ should be used as dependancy for s3c-fb.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Tomasz Figa <t.figa@samsung.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agodrivers/video/exynos/exynos_mipi_dsi.c: use devm_* APIs
Sachin Kamat [Fri, 22 Feb 2013 00:42:28 +0000 (16:42 -0800)]
drivers/video/exynos/exynos_mipi_dsi.c: use devm_* APIs

devm_* APIs are device managed and make exit and cleanup code simpler.
While at it also remove some unused labels and fix an error path.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Acked-by: Donghwa Lee <dh09.lee@samsung.com>
Cc: Inki Dae <inki.dae@samsung.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agodrivers/video/exynos/exynos_mipi_dsi.c: fix an error check condition
Sachin Kamat [Fri, 22 Feb 2013 00:42:27 +0000 (16:42 -0800)]
drivers/video/exynos/exynos_mipi_dsi.c: fix an error check condition

Checking an unsigned variable for negative value returns false.  Hence use
the macro to fix it.

Fixes the following smatch warning:

  drivers/video/exynos/exynos_mipi_dsi.c:417 exynos_mipi_dsi_probe() warn: unsigned 'dsim->irq' is never less than zero.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Acked-by: Donghwa Lee <dh09.lee@samsung.com>
Cc: Inki Dae <inki.dae@samsung.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agodrivers/video/exynos/s6e8ax0.c: use devm_* APIs in s6e8ax0.c
Sachin Kamat [Fri, 22 Feb 2013 00:42:25 +0000 (16:42 -0800)]
drivers/video/exynos/s6e8ax0.c: use devm_* APIs in s6e8ax0.c

devm_* APIs are device managed and make error handling and code cleanup
simpler.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Acked-by: Donghwa Lee <dh09.lee@samsung.com>
Cc: Inki Dae <inki.dae@samsung.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agodrivers/video/Kconfig: specify the SoCs that make use of FB_IMX
Fabio Estevam [Fri, 22 Feb 2013 00:42:24 +0000 (16:42 -0800)]
drivers/video/Kconfig: specify the SoCs that make use of FB_IMX

FB_IMX is the framebuffer driver used by MX1, MX21, MX25 and MX27 processors.

Pass this information to the Kconfig text to make it clear.

Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoARM: mmp: add display and fb support in pxa910 defconfig
Zhou Zhu [Fri, 22 Feb 2013 00:42:23 +0000 (16:42 -0800)]
ARM: mmp: add display and fb support in pxa910 defconfig

Add display and fb support in pxa910 defconfig.
Add tpohvga panel, spi support.
Add logo support.

Signed-off-by: Zhou Zhu <zzhu3@marvell.com>
Acked-by: Haojian Zhuang <haojian.zhuang@gmail.com>
Cc: Lisa Du <cldu@marvell.com>
Cc: Guoqing Li <ligq@marvell.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoARM: mmp: enable display in ttc_dkb
Zhou Zhu [Fri, 22 Feb 2013 00:42:21 +0000 (16:42 -0800)]
ARM: mmp: enable display in ttc_dkb

Enable display in ttc_dkb.

Signed-off-by: Zhou Zhu <zzhu3@marvell.com>
Acked-by: Haojian Zhuang <haojian.zhuang@gmail.com>
Cc: Lisa Du <cldu@marvell.com>
Cc: Guoqing Li <ligq@marvell.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoARM: mmp: added device for display controller
Zhou Zhu [Fri, 22 Feb 2013 00:42:20 +0000 (16:42 -0800)]
ARM: mmp: added device for display controller

Add device for display controller and fb support

Signed-off-by: Zhou Zhu <zzhu3@marvell.com>
Acked-by: Haojian Zhuang <haojian.zhuang@gmail.com>
Cc: Lisa Du <cldu@marvell.com>
Cc: Guoqing Li <ligq@marvell.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agovideo: mmpdisp: add spi port in display controller
Zhou Zhu [Fri, 22 Feb 2013 00:42:18 +0000 (16:42 -0800)]
video: mmpdisp: add spi port in display controller

Add spi port support in mmp display controller.  This port is from display
controller and for panel usage.  This driver implemented and registered as
a spi master.

Signed-off-by: Zhou Zhu <zzhu3@marvell.com>
Acked-by: Haojian Zhuang <haojian.zhuang@gmail.com>
Cc: Lisa Du <cldu@marvell.com>
Cc: Guoqing Li <ligq@marvell.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agovideo: mmp: add tpo hvga panel supported
Lisa Du [Fri, 22 Feb 2013 00:42:17 +0000 (16:42 -0800)]
video: mmp: add tpo hvga panel supported

Add tpo hvga panel support in marvell display framework.  This panel
driver implements modes query and power on/off.

This panel driver gets panel config/ plat power on/off/ connected path
name from machine-info and registered as a spi device.  This panel
driver uses mmp_disp supplied register_panel function to register panel
to path as machine-info defined.

Signed-off-by: Lisa Du <cldu@marvell.com>
Signed-off-by: Zhou Zhu <zzhu3@marvell.com>
Acked-by: Haojian Zhuang <haojian.zhuang@gmail.com>
Cc: Guoqing Li <ligq@marvell.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agovideo: mmp display controller support
Guoqing Li [Fri, 22 Feb 2013 00:42:15 +0000 (16:42 -0800)]
video: mmp display controller support

Marvell mmp series display controller support in mmpdisp subsystem.

This driver focus on implementation of hardware operations of
path/overlay, which is defined in mmp display subsystem interface.  This
driver registers all pathes to mmp display framework.

Signed-off-by: Guoqing Li <ligq@marvell.com>
Signed-off-by: Lisa Du <cldu@marvell.com>
Signed-off-by: Zhou Zhu <zzhu3@marvell.com>
Acked-by: Haojian Zhuang <haojian.zhuang@gmail.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agofb: mmp: include linux/platform_device.h
Arnd Bergmann [Fri, 22 Feb 2013 00:42:14 +0000 (16:42 -0800)]
fb: mmp: include linux/platform_device.h

Commit 16559ae48c76 ("kgdb: remove #include <linux/serial_8250.h> from
kgdb.h") changes the kgdb.h file so that drivers including it do not
implicitly include linux/platform_device.h.  The mmp framebuffer driver
is new, so Greg did not have a chance to fix it up when introducing his
change.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Zhou Zhu <zzhu3@marvell.com>
Cc: Lisa Du <cldu@marvell.com>
Cc: Guoqing Li <ligq@marvell.com>
Acked-by: Haojian Zhuang <haojian.zhuang@gmail.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agovideo: mmp fb support
Zhou Zhu [Fri, 22 Feb 2013 00:42:12 +0000 (16:42 -0800)]
video: mmp fb support

Add fb support for Marvell mmp display subsystem.  This driver is
configured using "buffer driver mach info".  With configured name of path,
this driver get path using using exported interface of mmp display driver.
Then this driver get overlay using configured id and operates on this
overlay to show buffers on display devices.

Signed-off-by: Zhou Zhu <zzhu3@marvell.com>
Signed-off-by: Lisa Du <cldu@marvell.com>
Cc: Guoqing Li <ligq@marvell.com>
Acked-by: Haojian Zhuang <haojian.zhuang@gmail.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agovideo: mmp display subsystem
Zhou Zhu [Fri, 22 Feb 2013 00:42:11 +0000 (16:42 -0800)]
video: mmp display subsystem

Add mmp display subsystem to support Marvell MMP display controllers.

This subsystem contains 4 parts:
--fb folder
--core.c
--hw folder
--panel folder

1. fb folder contains implementation of fb.  fb get path and overlay
   from common interface and operates on these structures.

2. core.c provides common interface for a hardware abstraction.  Major
   parts of this interface are:

   a) Path: path is a output device connected to a panel or HDMI TV.  Main
      operations of the path is set/get timing/output color.  fb operates
      output device through path structure.

   b) Ovly: Ovly is a buffer shown on the path.

      Ovly describes frame buffer and its source/destination size, offset,
      input color, buffer address, z-order, and so on.  Each fb device maps
      to one overlay.

3. hw folder contains implementation of hardware operations defined by
   core.c.  It registers paths for fb use.

4. panel folder contains implementation of panels.  It's connected to
   path.  Panel drivers would also regiester panels and linked to path
   when probe.

Signed-off-by: Zhou Zhu <zzhu3@marvell.com>
Signed-off-by: Lisa Du <cldu@marvell.com>
Cc: Guoqing Li <ligq@marvell.com>
Acked-by: Haojian Zhuang <haojian.zhuang@gmail.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agogoldfish: framebuffer driver
Arve Hjønnevåg [Fri, 22 Feb 2013 00:42:09 +0000 (16:42 -0800)]
goldfish: framebuffer driver

Framebuffer support for the Goldfish emulator.  This takes the Google
emulator and applies the x86 cleanups as well as moving the blank
methods to the usual Linux place and dropping the Android early suspend
logic (for now at least, that can be looked at as Android and upstream
converge).  Dropped various oddities like setting MTRRs on a virtual
frame buffer emulation...

With the drivers so far you can now boot a Linux initrd and have fun.

[sheng@linux.intel.com: cleaned up to handle x86]
[thomas.keel@intel.com: ported to 3.4]
[alan@linux.intel.com: cleaned up for style and 3.7, moved blank methods]
[akpm@linux-foundation.org: fix (silly) sparse warnings]
Signed-off-by: Mike A. Chan <mikechan@google.com>
Signed-off-by: Arve Hjønnevåg <arve@android.com>
Signed-off-by: Sheng Yang <sheng@linux.intel.com>
Signed-off-by: Yunhong Jiang <yunhong.jiang@intel.com>
Signed-off-by: Xiaohui Xin <xiaohui.xin@intel.com>
Signed-off-by: Jun Nakajima <jun.nakajima@intel.com>
Signed-off-by: Bruce Beare <bruce.j.beare@intel.com>
Signed-off-by: Tom Keel <thomas.keel@intel.com>
Signed-off-by: Alan Cox <alan@linux.intel.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agofbcon: clear the logo bitmap from the margin area
Kamal Mostafa [Fri, 22 Feb 2013 00:42:07 +0000 (16:42 -0800)]
fbcon: clear the logo bitmap from the margin area

Explicitly clear_margins when clearing the logo, in case the font dimensions
are non-integral to the framebuffer dimensions.

Signed-off-by: Kamal Mostafa <kamal@whence.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agopcmcia: move unbind/rebind into dev_pm_ops.complete
Christian Lamparter [Fri, 22 Feb 2013 00:42:06 +0000 (16:42 -0800)]
pcmcia: move unbind/rebind into dev_pm_ops.complete

Move the device rebind procedures for cardbus devices from the pm.resume
into the pm.complete callback.

The reason for moving the code is: "[...] The PM code needs to send
suspend and resume messages to every device in the right order, and it
can't do that if new devices are being added at the same time.  [...]"

However the situation really isn't quite that rigid.  In particular,
adding new children during a resume callback shouldn't cause much of
problem because the children don't need to be resumed anyway (since they
were never suspended).  On the other hand, if you do it you will get a
dev_warn() from the PM core, something like 'parent should not be
sleeping'.

Still, it is considered bad form and should be avoided if possible."

(Alan Stern's full comment about the topic can
be found here: <https://lkml.org/lkml/2012/7/10/254>)

Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Greg KH <greg@kroah.com>
Acked-by: "Rafael J. Wysocki" <rjw@sisk.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agocris: use "int" for ssize_t to match size_t
Geert Uytterhoeven [Fri, 22 Feb 2013 00:42:04 +0000 (16:42 -0800)]
cris: use "int" for ssize_t to match size_t

On cris-linux-gcc, __SIZE_TYPE__ expands to "unsigned int", as
gcc-4.6.3-nolibc/cris-linux/lib/gcc/cris-linux/4.6.3/plugin/include/config/cris/linux.h
has

    #define SIZE_TYPE "unsigned int"

Hence __kernel_size_t is also "unsigned int".  But __kernel_ssize_t is
"long", which has a different base type, causing compiler warnings like:

    fs/quota/quota_tree.c:372:4: warning: format '%zd' expects argument of type 'signed size_t', but argument 4 has type 'ssize_t' [-Wformat]

To fix this, __kernel_ssize_t should be changed to "int". Hence cris can
just use the generic 32-bit versions from include/asm-generic/posix_types.h
for all size-related types.

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Mikael Starvik <starvik@axis.com>
Acked-by: Jesper Nilsson <jesper.nilsson@axis.com>
Cc: Hans-Peter Nilsson <hans-peter.nilsson@axis.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomn10300: use for_each_pci_dev to simplify the code
Wei Yongjun [Fri, 22 Feb 2013 00:42:03 +0000 (16:42 -0800)]
mn10300: use for_each_pci_dev to simplify the code

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Cc: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agofs/block_dev.c: page cache wrongly left invalidated after revalidate_disk()
MITSUNARI Shigeo [Fri, 22 Feb 2013 00:42:01 +0000 (16:42 -0800)]
fs/block_dev.c: page cache wrongly left invalidated after revalidate_disk()

We found that bdev->bd_invalidated was left set once revalidate_disk()
is called, which results in page cache flush every time that device is
open.

Specifically, we found this problem in MD block device.  Once we resize
a MD device, mdadm --monitor periodically flush all page cache for that
device every 60 or 1000 seconds when it opens the device.

This bug lies since at least 3.2.0 till the latest kernel(3.6.2).  Patch
is attached.

The following steps will reproduce the problem.

1. prepair a block device (eg /dev/sdb).

2. create two partitions:

   sudo parted /dev/sdb
   mklabel gpt
   mkpart primary 0% 50%
   mkpart primary 50% 100%

3. create a md device.

   sudo mdadm -C /dev/md/hoge -l 1 -n 2 -e 1.2 --assume-clean --auto=md --symlink=no /dev/sdb1 /dev/sdb2

4. create file system and mount it

   sudo mkfs.ext3 /dev/md/hoge
   sudo mkdir /mnt/test
   sudo mount /dev/md/hoge /mnt/test

5. try to resize the device

   sudo mdadm -G /dev/md/hoge --size=max

6. create a file to fill file cache.

  sudo dd if=/dev/urandom of=/mnt/test/data bs=1M count=10

and verify the current status of file by free command.

7. mdadm monitor will open the md device every 1000 seconds and you
   will find all file cache on the device are cleared.

The timing can be reduced by the following steps.

a) kill mdadm and restart it with --delay option

   /sbin/mdadm --monitor --delay=30 --pid-file /var/run/mdadm/monitor.pid --daemonise --scan --syslog

or open the md device directly.

   sudo dd if=/dev/md/hoge of=/dev/null bs=4096 count=1

Signed-off-by: MITSUNARI Shigeo <herumi@nifty.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoinotify: remove broken mask checks causing unmount to be EINVAL
Jim Somerville [Fri, 22 Feb 2013 00:41:59 +0000 (16:41 -0800)]
inotify: remove broken mask checks causing unmount to be EINVAL

Running the command:

inotifywait -e unmount /mnt/disk

immediately aborts with a -EINVAL return code.  This is however a valid
parameter.  This abort occurs only if unmount is the sole event
parameter.  If other event parameters are supplied, then the unmount
event wait will work.

The problem was introduced by commit 44b350fc23e ("inotify: Fix mask
checks").  In that commit, it states:

The mask checks in inotify_update_existing_watch() and
inotify_new_watch() are useless because inotify_arg_to_mask()
sets FS_IN_IGNORED and FS_EVENT_ON_CHILD bits anyway.

But instead of removing the useless checks, it did this:

        mask = inotify_arg_to_mask(arg);
-       if (unlikely(!mask))
+       if (unlikely(!(mask & IN_ALL_EVENTS)))
                return -EINVAL;

The problem is that IN_ALL_EVENTS doesn't include IN_UNMOUNT, and other
parts of the code keep IN_UNMOUNT separate from IN_ALL_EVENTS.  So the
check should be:

if (unlikely(!(mask & (IN_ALL_EVENTS | IN_UNMOUNT))))

But inotify_arg_to_mask(arg) always sets the IN_UNMOUNT bit in the mask
anyway, so the check is always going to pass and thus should simply be
removed.  Also note that inotify_arg_to_mask completely controls what
mask bits get set from arg, there's no way for invalid bits to get
enabled there.

Lets fix it by simply removing the useless broken checks.

Signed-off-by: Jim Somerville <Jim.Somerville@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: Jerome Marchand <jmarchan@redhat.com>
Cc: John McCutchan <john@johnmccutchan.com>
Cc: Robert Love <rlove@rlove.org>
Cc: Eric Paris <eparis@parisplace.org>
Cc: <stable@vger.kernel.org> [2.6.37+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agocompat: return -EFAULT on error in waitid()
Dan Carpenter [Fri, 22 Feb 2013 00:41:57 +0000 (16:41 -0800)]
compat: return -EFAULT on error in waitid()

The copy_to_user() call returns the number of bytes remaining but we
want to return -EFAULT on error.

Fixes "x32: fix waitid()"

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobug.h, compiler.h: introduce compiletime_assert & BUILD_BUG_ON_MSG
Daniel Santos [Fri, 22 Feb 2013 00:41:55 +0000 (16:41 -0800)]
bug.h, compiler.h: introduce compiletime_assert & BUILD_BUG_ON_MSG

Introduce compiletime_assert to compiler.h, which moves the details of
how to break a build and emit an error message for a specific compiler
to the headers where these details should be.  Following in the
tradition of the POSIX assert macro, compiletime_assert creates a
build-time error when the supplied condition is *false*.

Next, we add BUILD_BUG_ON_MSG to bug.h which simply wraps
compiletime_assert, inverting the logic, so that it fails when the
condition is *true*, consistent with the language "build bug on." This
macro allows you to specify the error message you want emitted when the
supplied condition is true.

Finally, we remove all other code from bug.h that mucks with these
details (BUILD_BUG & BUILD_BUG_ON), and have them all call
BUILD_BUG_ON_MSG.  This not only reduces source code bloat, but also
prevents the possibility of code being changed for one macro and not for
the other (which was previously the case for BUILD_BUG and
BUILD_BUG_ON).

Since __compiletime_error_fallback is now only used in compiler.h, I'm
considering it a private macro and removing the double negation that's
now extraneous.

[akpm@linux-foundation.org: checkpatch fixes]
Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: David Rientjes <rientjes@google.com>
Cc: Joe Perches <joe@perches.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agocompiler.h, bug.h: prevent double error messages with BUILD_BUG{,_ON}
Daniel Santos [Fri, 22 Feb 2013 00:41:54 +0000 (16:41 -0800)]
compiler.h, bug.h: prevent double error messages with BUILD_BUG{,_ON}

Prior to the introduction of __attribute__((error("msg"))) in gcc 4.3,
creating compile-time errors required a little trickery.
BUILD_BUG{,_ON} uses this attribute when available to generate
compile-time errors, but also uses the negative-sized array trick for
older compilers, resulting in two error messages in some cases.  The
reason it's "some" cases is that as of gcc 4.4, the negative-sized array
will not create an error in some situations, like inline functions.

This patch replaces the negative-sized array code with the new
__compiletime_error_fallback() macro which expands to the same thing
unless the the error attribute is available, in which case it expands to
do{}while(0), resulting in exactly one compile-time error on all
versions of gcc.

Note that we are not changing the negative-sized array code for the
unoptimized version of BUILD_BUG_ON, since it has the potential to catch
problems that would be disabled in later versions of gcc were
__compiletime_error_fallback used.  The reason is that that an
unoptimized build can't always remove calls to an error-attributed
function call (like we are using) that should effectively become dead
code if it were optimized.  However, using a negative-sized array with a
similar value will not result in an false-positive (error).  The only
caveat being that it will also fail to catch valid conditions, which we
should be expecting in an unoptimized build anyway.

Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: David Rientjes <rientjes@google.com>
Cc: Joe Perches <joe@perches.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobug.h: make BUILD_BUG_ON generate compile-time error
Daniel Santos [Fri, 22 Feb 2013 00:41:52 +0000 (16:41 -0800)]
bug.h: make BUILD_BUG_ON generate compile-time error

Negative sized arrays wont create a compile-time error in some cases
starting with gcc 4.4 (e.g., inlined functions), but gcc 4.3 introduced
the error function attribute that will.

This patch modifies BUILD_BUG_ON to behave like BUILD_BUG already does,
using the error function attribute so that you don't have to build the
entire kernel to discover that you have a problem, and then enjoy trying
to track it down from a link-time error.

Also, we are only including asm/bug.h and then expecting that
linux/compiler.h will eventually be included to define __linktime_error
(used in BUILD_BUG_ON).  This patch includes it directly for clarity and
to avoid the possibility of changes in <arch>/*/include/asm/bug.h being
changed or not including linux/compiler.h for some reason.

Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Joe Perches <joe@perches.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobug.h: prevent double evaulation of `condition' in BUILD_BUG_ON
Daniel Santos [Fri, 22 Feb 2013 00:41:45 +0000 (16:41 -0800)]
bug.h: prevent double evaulation of `condition' in BUILD_BUG_ON

When calling BUILD_BUG_ON in an optimized build using gcc 4.3 and later,
the condition will be evaulated twice, possibily with side-effects.  This
patch eliminates that error.

[akpm@linux-foundation.org: tweak code layout]
Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: David Rientjes <rientjes@google.com>
Cc: Joe Perches <joe@perches.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agobug.h: fix BUILD_BUG_ON macro in __CHECKER__
Daniel Santos [Fri, 22 Feb 2013 00:41:44 +0000 (16:41 -0800)]
bug.h: fix BUILD_BUG_ON macro in __CHECKER__

When __CHECKER__ is defined, we disable all of the BUILD_BUG.* macros.
However, both BUILD_BUG_ON_NOT_POWER_OF_2 and BUILD_BUG_ON was evaluating
to nothing in this case, and we want (0) since this is a function-like
macro that will be followed by a semicolon.

Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Joe Perches <joe@perches.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agocompiler{,-gcc4}.h, bug.h: Remove duplicate macros
Daniel Santos [Fri, 22 Feb 2013 00:41:42 +0000 (16:41 -0800)]
compiler{,-gcc4}.h, bug.h: Remove duplicate macros

__linktime_error() does the same thing as __compiletime_error() and is
only used in bug.h.  Since the macro defines a function attribute that
will cause a failure at compile-time (not link-time), it makes more sense
to keep __compiletime_error(), which is also neatly mated with
__compiletime_warning().

Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Joe Perches <joe@perches.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agocompiler-gcc{3,4}.h: Use GCC_VERSION macro
Daniel Santos [Fri, 22 Feb 2013 00:41:41 +0000 (16:41 -0800)]
compiler-gcc{3,4}.h: Use GCC_VERSION macro

Using GCC_VERSION reduces complexity, is easier to read and is GCC's
recommended mechanism for doing version checks.  (Just don't ask me why
they didn't define it in the first place.) This also makes it easy to
merge compiler-gcc{,3,4}.h should somebody want to.

Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Joe Perches <joe@perches.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agocompiler-gcc.h: Add gcc-recommended GCC_VERSION macro
Daniel Santos [Fri, 22 Feb 2013 00:41:39 +0000 (16:41 -0800)]
compiler-gcc.h: Add gcc-recommended GCC_VERSION macro

Throughout compiler*.h, many version checks are made.  These can be
simplified by using the macro that gcc's documentation recommends.
However, my primary reason for adding this is that I need bug-check
macros that are enabled at certain gcc versions and it's cleaner to use
this macro than the tradition method:

  #if __GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ => 2)

If you add patch level, it gets this ugly:

  #if __GNUC__ > 4 || (__GNUC__ == 4 && (__GNUC_MINOR__ > 2 || \
      __GNUC_MINOR__ == 2 __GNUC_PATCHLEVEL__ >= 1))

As opposed to:

  #if GCC_VERSION >= 40201

While having separate headers for gcc 3 & 4 eliminates some of this
verbosity, they can still be cleaned up by this.

See also:

  http://gcc.gnu.org/onlinedocs/cpp/Common-Predefined-Macros.html

Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Joe Perches <joe@perches.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agocompiler-gcc4.h: Reorder macros based upon gcc ver
Daniel Santos [Fri, 22 Feb 2013 00:41:32 +0000 (16:41 -0800)]
compiler-gcc4.h: Reorder macros based upon gcc ver

This helps to keep the file from getting confusing, removes one
duplicate version check and should encourage future editors to put new
macros where they belong.

Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Joe Perches <joe@perches.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agodevice_cgroup: don't grab mutex in rcu callback
Jerry Snitselaar [Fri, 22 Feb 2013 00:41:31 +0000 (16:41 -0800)]
device_cgroup: don't grab mutex in rcu callback

Commit 103a197c0c4e ("security/device_cgroup: lock assert fails in
dev_exception_clean()") grabs devcgroup_mutex to fix assert failure, but
a mutex can't be grabbed in rcu callback.  Since there shouldn't be any
other references when css_free is called, mutex isn't needed for list
cleanup in devcgroup_css_free().

Signed-off-by: Jerry Snitselaar <jerry.snitselaar@oracle.com>
Acked-by: Tejun Heo <tj@kernel.org>
Acked-by: Aristeu Rozanski <aris@redhat.com>
Cc: James Morris <james.l.morris@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoMerge tag 'please-pull-pstore' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl...
Linus Torvalds [Thu, 21 Feb 2013 17:38:18 +0000 (09:38 -0800)]
Merge tag 'please-pull-pstore' of git://git./linux/kernel/git/aegl/linux

Pull pstore patches from Tony Luck:
 "A few fixes to reduce places where pstore might hang a system in the
  crash path.  Plus a new mountpoint (/sys/fs/pstore ...  makes more
  sense then /dev/pstore)."

Fix up trivial conflict in drivers/firmware/efivars.c

* tag 'please-pull-pstore' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux:
  pstore: Create a convenient mount point for pstore
  efi_pstore: Introducing workqueue updating sysfs
  efivars: Disable external interrupt while holding efivars->lock
  efi_pstore: Avoid deadlock in non-blocking paths
  pstore: Avoid deadlock in panic and emergency-restart path

8 years agoMerge tag 'dlm-3.9' of git://git.kernel.org/pub/scm/linux/kernel/git/teigland/linux-dlm
Linus Torvalds [Thu, 21 Feb 2013 17:25:23 +0000 (09:25 -0800)]
Merge tag 'dlm-3.9' of git://git./linux/kernel/git/teigland/linux-dlm

Pull dlm update from David Teigland:
 "This includes a single patch to avoid excessive and unnecessary
  scanning of rsbs to free."

* tag 'dlm-3.9' of git://git.kernel.org/pub/scm/linux/kernel/git/teigland/linux-dlm:
  dlm: avoid scanning unchanged toss lists

8 years agoMerge tag 'nfs-for-3.9-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Linus Torvalds [Thu, 21 Feb 2013 17:23:01 +0000 (09:23 -0800)]
Merge tag 'nfs-for-3.9-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client bugfixes from Trond Myklebust:

 - Fix an Oops in the pNFS layoutget code

 - Fix a number of NFSv4 and v4.1 state recovery deadlocks and hangs due
   to the interaction of the session drain lock and state management
   locks.

 - Remove task->tk_xprt, which was hiding a lot of RCU dereferencing
   bugs

 - Fix a long standing NFSv3 posix lock recovery bug.

 - Revert commit 324d003b0cd8 ("NFS: add nfs_sb_deactive_async to avoid
   deadlock").  It turned out that the root cause of the deadlock was
   due to interactions with the workqueues that have now been resolved.

* tag 'nfs-for-3.9-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: (22 commits)
  NLM: Ensure that we resend all pending blocking locks after a reclaim
  umount oops when remove blocklayoutdriver first
  sunrpc: silence build warning in gss_fill_context
  nfs: remove kfree() redundant null checks
  NFSv4.1: Don't decode skipped layoutgets
  NFSv4.1: Fix bulk recall and destroy of layouts
  NFSv4.1: Fix an ABBA locking issue with session and state serialisation
  NFSv4: Fix a reboot recovery race when opening a file
  NFSv4: Ensure delegation recall and byte range lock removal don't conflict
  NFSv4: Fix up the return values of nfs4_open_delegation_recall
  NFSv4.1: Don't lose locks when a server reboots during delegation return
  NFSv4.1: Prevent deadlocks between state recovery and file locking
  NFSv4: Allow the state manager to mark an open_owner as being recovered
  SUNRPC: Add missing static declaration to _gss_mech_get_by_name
  Revert "NFS: add nfs_sb_deactive_async to avoid deadlock"
  SUNRPC: Nuke the tk_xprt macro
  SUNRPC: Avoid RCU dereferences in the transport bind and connect code
  SUNRPC: Fix an RCU dereference in xprt_reserve
  SUNRPC: Pass pointers to struct rpc_xprt to the congestion window
  SUNRPC: Fix an RCU dereference in xs_local_rpcbind
  ...

8 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-3.0-nmw
Linus Torvalds [Thu, 21 Feb 2013 17:21:23 +0000 (09:21 -0800)]
Merge git://git./linux/kernel/git/steve/gfs2-3.0-nmw

Pull GFS2 updates from Steven Whitehouse:
 "This is one of the smallest collections of patches for the merge
  window for some time.  There are some clean ups relating to the
  transaction code and the shrinker, which are mostly in preparation for
  further development, but also make the code much easier to follow in
  these areas.

  There is a patch which allows the use of ->writepages even in the
  default ordered write mode for all writebacks.  This results in
  sending larger i/os to the block layer, and a subsequent increase in
  performance.  It also reduces the number of different i/o paths by
  one.

  There is also a bug fix reinstating the withdraw ack system which
  somehow got lost when the lock modules were merged into GFS2."

* git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-3.0-nmw:
  GFS2: Reinstate withdraw ack system
  GFS2: Get a block reservation before resizing a file
  GFS2: Split glock lru processing into two parts
  GFS2: Use ->writepages for ordered writes
  GFS2: Clean up freeze code
  GFS2: Merge gfs2_attach_bufdata() into trans.c
  GFS2: Copy gfs2_trans_add_bh into new data/meta functions
  GFS2: Split gfs2_trans_add_bh() into two
  GFS2: Merge revoke adding functions
  GFS2: Separate LRU scanning from shrinker

8 years agoMerge tag 'for-linus-v3.9-rc1' of git://oss.sgi.com/xfs/xfs
Linus Torvalds [Thu, 21 Feb 2013 17:08:45 +0000 (09:08 -0800)]
Merge tag 'for-linus-v3.9-rc1' of git://oss.sgi.com/xfs/xfs

Pull xfs update from Ben Myers:
 "Primarily bugfixes and a few cleanups:

   - fix(es) for compound buffers

   - remove unused XFS_TRANS_DEBUG routines

   - fix for dquot soft timer asserts due to overflow of d_blk_softlimit

   - don't zero allocation args structure members after they are memset(0)

   - fix for regression in dir v2 code introduced in commit 20f7e9f3

   - remove obsolete simple_strto<foo>

   - fix return value when filesystem probe finds no XFS magic, a
     regression introduced in 9802182.

   - remove boolean_t typedef completely

   - fix stack switch in __xfs_bmapi_allocate by moving the check for
     stack switch up into xfs_bmapi_write.

   - fix build error due to incomplete boolean_t removal

   - fix oops in _xfs_buf_find by validating that the requested block is
     within the filesystem bounds.

   - limit speculative preallocation near ENOSPC.

   - fix an unmount hang in xfs_wait_buftarg by freeing the
     xfs_buf_log_item in xfs_buf_item_unlock.

   - fix a possible use after free with AIO.

   - fix xfs_swap_extents after removal of xfs_flushinval_pages, a
     regression introduced in fb59581404a.

   - replace hardcoded 128 with log header size

   - add memory barrier before wake_up_bit in xfs_ifunlock

   - limit speculative preallocation on sparse files

   - fix xa_lock recursion bug introduced in 90810b9e82a3

   - fix write verifier for symlinks"

Fixed up conflicts in fs/xfs/xfs_buf_item.c (due to bli_format rename in
commit 0f22f9d0cd8a affecting the removed XFS_TRANS_DEBUG routines in
commit ec47eb6b0b45).

* tag 'for-linus-v3.9-rc1' of git://oss.sgi.com/xfs/xfs: (36 commits)
  xfs: xfs_bmap_add_attrfork_local is too generic
  xfs: remove log force from xfs_buf_trylock()
  xfs: recheck buffer pinned status after push trylock failure
  xfs: limit speculative prealloc size on sparse files
  xfs: memory barrier before wake_up_bit()
  xfs: refactor space log reservation for XFS_TRANS_ATTR_SET
  xfs: make use of XFS_SB_LOG_RES() at xfs_fs_log_dummy()
  xfs: make use of XFS_SB_LOG_RES() at xfs_mount_log_sb()
  xfs: make use of XFS_SB_LOG_RES() at xfs_log_sbcount()
  xfs: introduce XFS_SB_LOG_RES() for transactions that modify sb on disk
  xfs: calculate XFS_TRANS_QM_QUOTAOFF_END space log reservation at mount time
  xfs: calculate XFS_TRANS_QM_QUOTAOFF space log reservation at mount time
  xfs: calculate XFS_TRANS_QM_DQALLOC space log reservation at mount time
  xfs: calcuate XFS_TRANS_QM_SETQLIM space log reservation at mount time
  xfs: calculate xfs_qm_write_sb_changes() space log reservation at mount time
  xfs: calculate XFS_TRANS_QM_SBCHANGE space log reservation at mount time
  xfs: make use of xfs_calc_buf_res() in xfs_trans.c
  xfs: add a helper to figure out the space log reservation per item
  xfs: Fix xfs_swap_extents() after removal of xfs_flushinval_pages()
  xfs: Fix possible use-after-free with AIO
  ...

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi...
Linus Torvalds [Thu, 21 Feb 2013 17:03:54 +0000 (09:03 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/mszeredi/fuse

Pull fuse updates from Miklos Szeredi:
 "The biggest part of this pull request is a patch series from Maxim
  Patlasov to optimize scatter-gather direct IO.  There's also the
  addition of a "readdirplus" API, poll events and various fixes and
  cleanups.

  There's a one line change outside of fuse to mm/filemap.c which makes
  the argument of iov_iter_single_seg_count() const, required by Maxim's
  patches."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse: (22 commits)
  fuse: allow control of adaptive readdirplus use
  Synchronize fuse header with one used in library
  fuse: send poll events
  fuse: don't WARN when nlink is zero
  fuse: avoid out-of-scope stack access
  fuse: bump version for READDIRPLUS
  FUSE: Adapt readdirplus to application usage patterns
  Do not use RCU for current process credentials
  fuse: cleanup fuse_direct_io()
  fuse: optimize __fuse_direct_io()
  fuse: optimize fuse_get_user_pages()
  fuse: pass iov[] to fuse_get_user_pages()
  mm: minor cleanup of iov_iter_single_seg_count()
  fuse: use req->page_descs[] for argpages cases
  fuse: add per-page descriptor <offset, length> to fuse_req
  fuse: rework fuse_do_ioctl()
  fuse: rework fuse_perform_write()
  fuse: rework fuse_readpages()
  fuse: rework fuse_retrieve()
  fuse: categorize fuse_get_req()
  ...

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh...
Linus Torvalds [Thu, 21 Feb 2013 17:02:13 +0000 (09:02 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/ericvh/v9fs

Pull v9fs updates from Eric Van Hensbergen:
 "Just fixes and simplifications"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh/v9fs:
  fs/9p: Fix atomic_open
  fs/9p: Don't use O_TRUNC flag in TOPEN and TLOPEN request
  locking in fs/9p ->readdir()

8 years agoMerge branch 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/linux...
Linus Torvalds [Thu, 21 Feb 2013 16:18:12 +0000 (08:18 -0800)]
Merge branch 'next' of git://git./linux/kernel/git/jmorris/linux-security

Pull security subsystem updates from James Morris:
 "This is basically a maintenance update for the TPM driver and EVM/IMA"

Fix up conflicts in lib/digsig.c and security/integrity/ima/ima_main.c

* 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/linux-security: (45 commits)
  tpm/ibmvtpm: build only when IBM pseries is configured
  ima: digital signature verification using asymmetric keys
  ima: rename hash calculation functions
  ima: use new crypto_shash API instead of old crypto_hash
  ima: add policy support for file system uuid
  evm: add file system uuid to EVM hmac
  tpm_tis: check pnp_acpi_device return code
  char/tpm/tpm_i2c_stm_st33: drop temporary variable for return value
  char/tpm/tpm_i2c_stm_st33: remove dead assignment in tpm_st33_i2c_probe
  char/tpm/tpm_i2c_stm_st33: Remove __devexit attribute
  char/tpm/tpm_i2c_stm_st33: Don't use memcpy for one byte assignment
  tpm_i2c_stm_st33: removed unused variables/code
  TPM: Wait for TPM_ACCESS tpmRegValidSts to go high at startup
  tpm: Fix cancellation of TPM commands (interrupt mode)
  tpm: Fix cancellation of TPM commands (polling mode)
  tpm: Store TPM vendor ID
  TPM: Work around buggy TPMs that block during continue self test
  tpm_i2c_stm_st33: fix oops when i2c client is unavailable
  char/tpm: Use struct dev_pm_ops for power management
  TPM: STMicroelectronics ST33 I2C BUILD STUFF
  ...

8 years agoKEYS: Revert one application of "Fix unreachable code" patch
David Howells [Thu, 21 Feb 2013 12:00:25 +0000 (12:00 +0000)]
KEYS: Revert one application of "Fix unreachable code" patch

A patch to fix some unreachable code in search_my_process_keyrings() got
applied twice by two different routes upstream as commits e67eab39bee2
and b010520ab3d2 (both "fix unreachable code").

Unfortunately, the second application removed something it shouldn't
have and this wasn't detected by GIT.  This is due to the patch not
having sufficient lines of context to distinguish the two places of
application.

The effect of this is relatively minor: inside the kernel, the keyring
search routines may search multiple keyrings and then prioritise the
errors if no keys or negative keys are found in any of them.  With the
extra deletion, the presence of a negative key in the thread keyring
(causing ENOKEY) is incorrectly overridden by an error searching the
process keyring.

So revert the second application of the patch.

Signed-off-by: David Howells <dhowells@redhat.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next
Linus Torvalds [Thu, 21 Feb 2013 02:58:50 +0000 (18:58 -0800)]
Merge git://git./linux/kernel/git/davem/net-next

Pull networking update from David Miller:

 1) Checkpoint/restarted TCP sockets now can properly propagate the TCP
    timestamp offset.  From Andrey Vagin.

 2) VMWARE VM VSOCK layer, from Andy King.

 3) Much improved support for virtual functions and SR-IOV in bnx2x,
    from Ariel ELior.

 4) All protocols on ipv4 and ipv6 are now network namespace aware, and
    all the compatability checks for initial-namespace-only protocols is
    removed.  Thanks to Tom Parkin for helping deal with the last major
    holdout, L2TP.

 5) IPV6 support in netpoll and network namespace support in pktgen,
    from Cong Wang.

 6) Multiple Registration Protocol (MRP) and Multiple VLAN Registration
    Protocol (MVRP) support, from David Ward.

 7) Compute packet lengths more accurately in the packet scheduler, from
    Eric Dumazet.

 8) Use per-task page fragment allocator in skb_append_datato_frags(),
    also from Eric Dumazet.

 9) Add support for connection tracking labels in netfilter, from
    Florian Westphal.

10) Fix default multicast group joining on ipv6, and add anti-spoofing
    checks to 6to4 and 6rd.  From Hannes Frederic Sowa.

11) Make ipv4/ipv6 fragmentation memory limits more reasonable in modern
    times, rearrange inet frag datastructures for better cacheline
    locality, and move more operations outside of locking.  From Jesper
    Dangaard Brouer.

12) Instead of strict master <--> slave relationships, allow arbitrary
    scenerios with "upper device lists".  From Jiri Pirko.

13) Improve rate limiting accuracy in TBF and act_police, also from Jiri
    Pirko.

14) Add a BPF filter netfilter match target, from Willem de Bruijn.

15) Orphan and delete a bunch of pre-historic networking drivers from
    Paul Gortmaker.

16) Add TSO support for GRE tunnels, from Pravin B SHelar.  Although
    this still needs some minor bug fixing before it's %100 correct in
    all cases.

17) Handle unresolved IPSEC states like ARP, with a resolution packet
    queue.  From Steffen Klassert.

18) Remove TCP Appropriate Byte Count support (ABC), from Stephen
    Hemminger.  This was long overdue.

19) Support SO_REUSEPORT, from Tom Herbert.

20) Allow locking a socket BPF filter, so that it cannot change after a
    process drops capabilities.

21) Add VLAN filtering to bridge, from Vlad Yasevich.

22) Bring ipv6 on-par with ipv4 and do not cache neighbour entries in
    the ipv6 routes, from YOSHIFUJI Hideaki.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1538 commits)
  ipv6: fix race condition regarding dst->expires and dst->from.
  net: fix a wrong assignment in skb_split()
  ip_gre: remove an extra dst_release()
  ppp: set qdisc_tx_busylock to avoid LOCKDEP splat
  atl1c: restore buffer state
  net: fix a build failure when !CONFIG_PROC_FS
  net: ipv4: fix waring -Wunused-variable
  net: proc: fix build failed when procfs is not configured
  Revert "xen: netback: remove redundant xenvif_put"
  net: move procfs code to net/core/net-procfs.c
  qmi_wwan, cdc-ether: add ADU960S
  bonding: set sysfs device_type to 'bond'
  bonding: fix bond_release_all inconsistencies
  b44: use netdev_alloc_skb_ip_align()
  xen: netback: remove redundant xenvif_put
  net: fec: Do a sanity check on the gpio number
  ip_gre: propogate target device GSO capability to the tunnel device
  ip_gre: allow CSUM capable devices to handle packets
  bonding: Fix initialize after use for 3ad machine state spinlock
  bonding: Fix race condition between bond_enslave() and bond_3ad_update_lacp_rate()
  ...

8 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc
Linus Torvalds [Wed, 20 Feb 2013 22:31:26 +0000 (14:31 -0800)]
Merge git://git./linux/kernel/git/davem/sparc

Pull sparc updates from David Miller:
 "Mostly more sparc64 THP bug fixes, and a refactoring of SMP bootup on
  sparc32 from Sam Ravnborg."

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc:
  sparc32: refactor smp boot
  sparc64: Fix huge PMD to PTE translation for sun4u in TLB miss handler.
  sparc64: Fix tsb_grow() in atomic context.
  sparc64: Handle hugepage TSB being NULL.
  sparc64: Fix gfp_flags setting in tsb_grow().

8 years agoMerge tag 'arm64-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cmarinas...
Linus Torvalds [Wed, 20 Feb 2013 22:30:33 +0000 (14:30 -0800)]
Merge tag 'arm64-for-linus' of git://git./linux/kernel/git/cmarinas/linux-aarch64

Pull arm64 patches from Catalin Marinas:

 - SMP support for the PSCI booting protocol (power state coordination
   interface).

 - Simple earlyprintk support.

 - Platform devices populated by default from the DT (SoC-agnostic).

 - CONTEXTIDR support (used by external trace tools).

* tag 'arm64-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cmarinas/linux-aarch64:
  arm64: mm: update CONTEXTIDR register to contain PID of current process
  arm64: atomics: fix grossly inconsistent asm constraints for exclusives
  arm64: compat: use compat_uptr_t type for compat_ucontext.uc_link
  arm64: Select ARCH_WANT_FRAME_POINTERS
  arm64: Add kvm_para.h and xor.h generic headers
  arm64: SMP: enable PSCI boot method
  arm64: psci: add support for PSCI invocations from the kernel
  arm64: SMP: rework the SMP code to be enabling method agnostic
  arm64: perf: add guest vs host discrimination
  arm64: add COMPAT_PSR_*_BIT flags
  arm64: Add simple earlyprintk support
  arm64: Populate the platform devices

8 years agoMerge branch 'for-linus-2' of git://git.linaro.org/people/rmk/linux-arm
Linus Torvalds [Wed, 20 Feb 2013 22:29:37 +0000 (14:29 -0800)]
Merge branch 'for-linus-2' of git://git.linaro.org/people/rmk/linux-arm

Pull ARM updates (part two) from Russell King:

 - breakpoint and perf updates from Will Deacon.

 - hypervisor boot mode updates from Will.

 - support for Power State Coordination Interface via the Hypervisor

 - core ARM support for KVM

* 'for-linus-2' of git://git.linaro.org/people/rmk/linux-arm: (32 commits)
  KVM: ARM: Add maintainer entry for KVM/ARM
  KVM: ARM: Power State Coordination Interface implementation
  KVM: ARM: Handle I/O aborts
  KVM: ARM: Handle guest faults in KVM
  KVM: ARM: VFP userspace interface
  KVM: ARM: Demux CCSIDR in the userspace API
  KVM: ARM: User space API for getting/setting co-proc registers
  KVM: ARM: Emulation framework and CP15 emulation
  KVM: ARM: World-switch implementation
  KVM: ARM: Inject IRQs and FIQs from userspace
  KVM: ARM: Memory virtualization setup
  KVM: ARM: Hypervisor initialization
  KVM: ARM: Initial skeleton to compile KVM support
  ARM: Section based HYP idmap
  ARM: Add page table and page defines needed by KVM
  ARM: perf: simplify __hw_perf_event_init err handling
  ARM: perf: remove unnecessary checks for idx < 0
  ARM: perf: handle armpmu_register failing
  ARM: perf: don't pretend to support counting of L1I writes
  ARM: perf: remove redundant NULL check on cpu_pmu
  ...

8 years agoMerge branch 'for-linus' of git://git.linaro.org/people/rmk/linux-arm
Linus Torvalds [Wed, 20 Feb 2013 22:27:58 +0000 (14:27 -0800)]
Merge branch 'for-linus' of git://git.linaro.org/people/rmk/linux-arm

Pull ARM updates (part one) from Russell King:

 - MMC patches from Ulf Hansson and Pawel Moll.  These add support for
   DDR mode and the latest variant found on ARM Versatile Express, as
   well as a number of cleanups.

 - A fix for to improve the behaviour of ARMs sched_clock()

 - Changes to the ARM ioremap() code.  I'm not convinced with the
   primary arguments for this, but it's been around for a while, and
   people seem happy with it - and the "other" justification for this is
   at

      http://lkml.org/lkml/2012/12/6/184

 - Add SCHED_HRTICK to ARMs Kconfig

 - Making the ARM SHA/AES code Thumb-2 compatible

 - A collection of other small updates.

* 'for-linus' of git://git.linaro.org/people/rmk/linux-arm: (26 commits)
  ARM: add SCHED_HRTICK config option
  ARM: 7650/1: mm: replace direct access to mm->context.id with new macro
  ARM: 7649/1: mm: mm->context.id fix for big-endian
  ARM: 7648/1: pci: Allow passing per-controller private data
  ARM: 7647/1: pci: Keep pci_common_init() around after init
  ARM: fix warnings introduced by previous patch
  ARM: 7646/1: mm: use static_vm for managing static mapped areas
  ARM: 7645/1: ioremap: introduce an infrastructure for static mapped area
  ARM: 7644/1: vmregion: remove vmregion code entirely
  MAINTAINERS: Re-assert MMCI driver maintainer status
  MAINTAINERS: add additional file for MMCI driver
  MAINTAINERS: add maintainer entry for AMBA serial drivers
  ARM: 7637/1: memory: use SZ_ constants for defining the virtual memory layout
  ARM: 7643/1: sched: correct update_sched_clock()
  ARM: 7635/1: versatile: fix the PCI IRQ regression
  ARM: 7639/1: cache-l2x0: add missed dummy outer_resume entry
  ARM: 7630/1: mmc: mmci: Fixup and cleanup code for DMA handling
  ARM: 7632/1: spinlock: avoid exclusive accesses on unlock() path
  ARM: 7631/1: mmc: mmci: Add new VE MMCI variant
  ARM: 7623/1: mmc: mmci: Fixup clock gating when freq is 0 for ST-variants
  ...

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/geert/linux...
Linus Torvalds [Wed, 20 Feb 2013 22:27:00 +0000 (14:27 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/geert/linux-m68k

Pull m68k update from Geert Uytterhoeven.

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/geert/linux-m68k:
  m68k: Sort out !CONFIG_MMU_SUN3 vs. CONFIG_HAS_DMA
  swim: Add missing spinlock init

8 years agosparc32: refactor smp boot
Sam Ravnborg [Fri, 15 Feb 2013 14:52:06 +0000 (15:52 +0100)]
sparc32: refactor smp boot

Introduce a common smp_callin() function to call
from trampoline_32.S.
Add platform specific functions to handle the
platform details.

This is in preparation for a patch that will
unify the smp boot stuff for all architectures.
sparc32 was significantly different to warrant
this patch in preparation.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: David S. Miller <davem@davemloft.net>
Cc: "Srivatsa S. Bhat" <srivatsa.bhat@linux.vnet.ibm.com>
Acked-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agosparc64: Fix huge PMD to PTE translation for sun4u in TLB miss handler.
David S. Miller [Wed, 20 Feb 2013 20:38:40 +0000 (12:38 -0800)]
sparc64: Fix huge PMD to PTE translation for sun4u in TLB miss handler.

When we set the sun4u version of the PTE execute bit, it's:

or REG, _PAGE_EXEC_4U, REG

_PAGE_EXEC_4U is 0x1000, unfortunately the immedate field of the
'or' instruction is a signed 13-bit value.  So the above actually
assembles into:

or REG, -4096, REG

completely corrupting the final PTE value.

Set it with a:

sethi %hi(_PAGE_EXEC_4U), TMP
or REG, TMP, REG

sequence instead.

This fixes "git gc" crashes on sun4u machines.

Reported-by: Meelis Roos <mroos@linux.ee>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoipv6: fix race condition regarding dst->expires and dst->from.
YOSHIFUJI Hideaki / 吉藤英明 [Wed, 20 Feb 2013 00:29:08 +0000 (00:29 +0000)]
ipv6: fix race condition regarding dst->expires and dst->from.

Eric Dumazet wrote:
| Some strange crashes happen in rt6_check_expired(), with access
| to random addresses.
|
| At first glance, it looks like the RTF_EXPIRES and
| stuff added in commit 1716a96101c49186b
| (ipv6: fix problem with expired dst cache)
| are racy : same dst could be manipulated at the same time
| on different cpus.
|
| At some point, our stack believes rt->dst.from contains a dst pointer,
| while its really a jiffie value (as rt->dst.expires shares the same area
| of memory)
|
| rt6_update_expires() should be fixed, or am I missing something ?
|
| CC Neil because of https://bugzilla.redhat.com/show_bug.cgi?id=892060

Because we do not have any locks for dst_entry, we cannot change
essential structure in the entry; e.g., we cannot change reference
to other entity.

To fix this issue, split 'from' and 'expires' field in dst_entry
out of union.  Once it is 'from' is assigned in the constructor,
keep the reference until the very last stage of the life time of
the object.

Of course, it is unsafe to change 'from', so make rt6_set_from simple
just for fresh entries.

Reported-by: Eric Dumazet <eric.dumazet@gmail.com>
Reported-by: Neil Horman <nhorman@tuxdriver.com>
CC: Gao Feng <gaofeng@cn.fujitsu.com>
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reported-by: Steinar H. Gunderson <sesse@google.com>
Reviewed-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: fix a wrong assignment in skb_split()
Amerigo Wang [Tue, 19 Feb 2013 22:51:30 +0000 (22:51 +0000)]
net: fix a wrong assignment in skb_split()

commit c9af6db4c11ccc6c3e7f1 (net: Fix possible wrong checksum generation)
has a suspicous piece:

-       skb_shinfo(skb1)->gso_type = skb_shinfo(skb)->gso_type;
-
+       skb_shinfo(skb)->tx_flags = skb_shinfo(skb1)->tx_flags & SKBTX_SHARED_FRAG;

skb1 is the new skb, therefore should be on the left side of the assignment.
This patch fixes it.

Cc: Pravin B Shelar <pshelar@nicira.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Cong Wang <amwang@redhat.com>
Acked-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge tag 'edac_3.9' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp
Linus Torvalds [Wed, 20 Feb 2013 19:28:50 +0000 (11:28 -0800)]
Merge tag 'edac_3.9' of git://git./linux/kernel/git/bp/bp

Pull EDAC updates from Borislav Petkov:
 "Mostly AMD's side of EDAC.  It is basically a new family enablement
  stuff: AMD F16h MCE decoding enablement from Jacob Shin.  The rest is
  trivial cleanups."

* tag 'edac_3.9' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp:
  mpc85xx_edac: Fix typo
  EDAC, MCE, AMD: Remove unneeded exports
  EDAC, MCE, AMD: Add MCE decoding support for Family 16h
  EDAC, MCE, AMD: Make MC2 decoding per-family
  amd64_edac: Remove dead code