epoll: optimize EPOLL_CTL_DEL using rcu

Nathan Zimmer found that once we get over 10+ cpus, the scalability of
SPECjbb falls over due to the contention on the global 'epmutex', which is
taken in on EPOLL_CTL_ADD and EPOLL_CTL_DEL operations.

Patch #1 removes the 'epmutex' lock completely from the EPOLL_CTL_DEL path
by using rcu to guard against any concurrent traversals.

Patch #2 remove the 'epmutex' lock from EPOLL_CTL_ADD operations for
simple topologies.  IE when adding a link from an epoll file descriptor to
a wakeup source, where the epoll file descriptor is not nested.

This patch (of 2):

Optimize EPOLL_CTL_DEL such that it does not require the 'epmutex' by
converting the file->f_ep_links list into an rcu one.  In this way, we can
traverse the epoll network on the add path in parallel with deletes.
Since deletes can't create loops or worse wakeup paths, this is safe.

This patch in combination with the patch "epoll: Do not take global 'epmutex'
for simple topologies", shows a dramatic performance improvement in
scalability for SPECjbb.

Signed-off-by: Jason Baron <[email protected]>
Tested-by: Nathan Zimmer <[email protected]>
Cc: Eric Wong <[email protected]>
Cc: Nelson Elhage <[email protected]>
Cc: Al Viro <[email protected]>
Cc: Davide Libenzi <[email protected]>
Cc: "Paul E. McKenney" <[email protected]>
CC: Wu Fengguang <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

checkpatch: add check for sscanf without return use

Naked use sscanf can be troublesome because the pointed to variables may
not have been set.

Add a warning when the sscanf return value is not used.

For now, do not add __must_check to the sscanf prototype because that will
cause a couple of hundred new warnings when compiling a kernel.

Signed-off-by: Joe Perches <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

checkpatch: don't require kernel style __attribute__ shortcuts in uapi paths

Avoid prescribing kernel styled shortcuts for gcc extensions of
__attribute__((foo)) in the uapi include paths.

Fix $realfile filename when using -f/--file to not remove first level
directory as if the filename was used in a -P1 patch.  Only strip the
first level directory (typically a or b) for P1 patches.

Signed-off-by: Joe Perches <[email protected]>
Cc: "Dixit, Ashutosh" <[email protected]>
Cc: Andy Whitcroft <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

checkpatch: improve "return is not a function" test

Find a few more cases where parentheses are used around the value of a
return statement.

This now uses the "$balanced_parens" test and also makes the test depend
on perl v5.10 and higher.

This now finds return with parenthesis uses the old code did not find
like:

ERROR: return is not a function, parentheses are not required
#211: FILE: arch/m68k/include/asm/sun3xflop.h:211:
+ return ((error == 0) ? 0 : -1);

Signed-off-by: Joe Perches <[email protected]>
Tested-by: David Cohen <[email protected]>
Acked-by: David Cohen <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

checkpatch.pl: check for the FSF mailing address

Kernel maintainers reject new instances of the GPL boilerplate paragraph
directing people to write to the FSF for a copy of the GPL, since the FSF
has moved in the past and may do so again.

Make this an error for new code, but just a --strict CHK in --file mode;
anyone interested in doing tree-wide cleanups of this form can enable this
test explicitly.

Signed-off-by: Josh Triplett <[email protected]>
Acked-by: Greg Kroah-Hartman <[email protected]>
Acked-by: Joe Perches <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

checkpatch: make the memory barrier test noisier

Peter Zijlstra prefers that comments be required near uses of memory
barriers.

Change the message level for memory barrier uses from a --strict test only
to a normal WARN so it's always emitted.

This might produce false positives around insertions of memory barriers
when a comment is outside the patch context block.

And checkpatch is still stupid, it only looks for existence of any
comment, not at the comment content.

Signed-off-by: Joe Perches <[email protected]>
Suggested-by: Peter Zijlstra <[email protected]>
Acked-by: Peter Zijlstra <[email protected]>
Acked-by: Paul E. McKenney <[email protected]>
Cc: Oliver Neukum <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

checkpatch: add rules to check init attribute and const defects

People get this regularly wrong and it breaks the LTO builds, as it causes
a section attribute conflict.

Add --fix capability too.

Based on a patch from Andi Kleen.

Signed-off-by: Joe Perches <[email protected]>
Cc: Andi Kleen <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

checkpatch: add test for #defines of ARCH_HAS_<foo>

Add a test for these #defines

Additionally, moved string_find_replace sub as it screws up subsequent
formatting when placed inside another sub.

Signed-off-by: Joe Perches <[email protected]>
Suggested-by: Andrew Morton <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

checkpatch: find CamelCase definitions of struct/union/enum

Checkpatch doesn't currently find CamelCase definitions of structs, unions
or enums.

Add that ability.

Signed-off-by: Joe Perches <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

checkpatch: update seq_<foo> tests

seq_vprintf, seq_printf and seq_puts are logging functions and should be
allowed to exceed the maximium line length.

Add maximum line length exceptions for these functions.

Also, suggesting seq_printf conversions to seq_puts should be tested for
arguments after the format.

Signed-off-by: Joe Perches <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

checkpatch: extend CamelCase types and ignore existing CamelCase uses in a patch

Extend the CamelCase words found to include structure members.

In https://lkml.org/lkml/2013/9/3/318 Sarah Sharp (mostly) wrote:

"In general, if checkpatch.pl complains about a variable a patch
introduces that's CamelCase, you should pay attention to it.  Otherwise,
[] ignore it."

So, if checking a patch, scan the original patched file if it's available
and add any preexisting CamelCase types so reuses do not generate
CamelCase messages.

That also means Andrew's not so cruelly spurned anymore.
https://lkml.org/lkml/2013/2/22/426

Signed-off-by: Joe Perches <[email protected]>
Suggested-by: Sarah Sharp <[email protected]>
Suggested-by: Andrew Morton <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

checkpatch: report missing spaces around trigraphs with --strict

Spaces around trigraphs are specified by CodingStyle but checkpatch is
currently silent about them because there are many current instances
without them.

Make missing spaces around trigraphs a --strict message.

Signed-off-by: Joe Perches <[email protected]>
Reviewed-by: Josh Triplett <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

debugfs: use list_next_entry() in debugfs_remove_recursive()

Change debugfs_remove_recursive() to use list_next_entry(child), no
changes in generated code.

Signed-off-by: Oleg Nesterov <[email protected]>
Cc: Eilon Greenstein <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

list: introduce list_last_entry(), use list_{first,last}_entry()

We already have list_first_entry(), it makes sense to also add
list_last_entry() for consistency.  And we use both helpers in
list_for_each_*().

Signed-off-by: Oleg Nesterov <[email protected]>
Cc: Eilon Greenstein <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

list: change list_for_each_entry*() to use list_*_entry()

Now that we have list_{next,prev}_entry() we can change
list_for_each_entry*() and list_safe_reset_next() to use the new helpers
to improve the readability.

Signed-off-by: Oleg Nesterov <[email protected]>
Cc: Eilon Greenstein <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

list: introduce list_next_entry() and list_prev_entry()

Add two trivial helpers list_next_entry() and list_prev_entry(), they
can have a lot of users including list.h itself.  In fact the 1st one is
already defined in events/core.c and bnx2x_sp.c, so the patch simply
moves the definition to list.h.

Signed-off-by: Oleg Nesterov <[email protected]>
Cc: Eilon Greenstein <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sound/soc/pxa/mmp-pcm.c: use gen_pool_dma_alloc() to allocate dma buffer

Since gen_pool_dma_alloc() is introduced, we implement it to simplify code.

Signed-off-by: Nicolin Chen <[email protected]>
Cc: Eric Miao <[email protected]>
Cc: Russell King <[email protected]>
Cc: Haojian Zhuang <[email protected]>
Cc: Liam Girdwood <[email protected]>
Cc: Mark Brown <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sound/soc/davinci/davinci-pcm.c: use gen_pool_dma_alloc() in davinci-pcm.c

Since gen_pool_dma_alloc() is introduced, we implement it to simplify code.

Signed-off-by: Nicolin Chen <[email protected]>
Cc: Liam Girdwood <[email protected]>
Cc: Mark Brown <[email protected]>
Cc: Jaroslav Kysela <[email protected]>
Cc: Takashi Iwai <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/uio/uio_pruss.c: use gen_pool_dma_alloc() to allocate sram memory

Since gen_pool_dma_alloc() is introduced, we implement it to simplify code.

Signed-off-by: Nicolin Chen <[email protected]>
Cc: "Hans J. Koch" <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/media/platform/coda.c: use gen_pool_dma_alloc() to allocate iram buffer

Since gen_pool_dma_alloc() is introduced, we implement it to simplify code.

Signed-off-by: Nicolin Chen <[email protected]>
Acked-by: Mauro Carvalho Chehab <[email protected]>
Cc: Grant Likely <[email protected]>
Cc: Rob Herring <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/dma/mmp_tdma.c: use gen_pool_dma_alloc() to allocate descriptor

Since gen_pool_dma_alloc() is introduced, we implement it to simplify code.

Signed-off-by: Nicolin Chen <[email protected]>
Cc: Dan Williams <[email protected]>
Cc: Vinod Koul <[email protected]>
Cc: Grant Likely <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

arch/arm/mach-davinci/sram.c: use gen_pool_dma_alloc() to sram.c

Since gen_pool_dma_alloc() is introduced, we implement it to simplify code.

Signed-off-by: Nicolin Chen <[email protected]>
Cc: Sekhar Nori <[email protected]>
Cc: Kevin Hilman <[email protected]>
Cc: Russell King <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

lib/genalloc: add a helper function for DMA buffer allocation

When using pool space for DMA buffer, there might be duplicated calling of
gen_pool_alloc() and gen_pool_virt_to_phys() in each implementation.

Thus it's better to add a simple helper function, a compatible one to the
common dma_alloc_coherent(), to save some code.

Signed-off-by: Nicolin Chen <[email protected]>
Cc: "Hans J. Koch" <[email protected]>
Cc: Dan Williams <[email protected]>
Cc: Eric Miao <[email protected]>
Cc: Grant Likely <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Haojian Zhuang <[email protected]>
Cc: Jaroslav Kysela <[email protected]>
Cc: Kevin Hilman <[email protected]>
Cc: Liam Girdwood <[email protected]>
Cc: Mark Brown <[email protected]>
Cc: Mauro Carvalho Chehab <[email protected]>
Cc: Rob Herring <[email protected]>
Cc: Russell King <[email protected]>
Cc: Sekhar Nori <[email protected]>
Cc: Takashi Iwai <[email protected]>
Cc: Vinod Koul <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

lib/digsig.c: use ERR_CAST inlined function instead of ERR_PTR(PTR_ERR(...))

Signed-off-by: Duan Jiong <[email protected]>
Cc: James Morris <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

lib/vsprintf.c: document formats for dentry and struct file

Looks like these were added to Documentation/printk-formats.txt but
not the in-file table.

Signed-off-by: Olof Johansson <[email protected]>
Cc: Al Viro <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

lib/debugobjects.c: remove unnecessary work pending test

Remove unnecessary work pending test before calling schedule_work().  It
has been tested in queue_work_on() already.  No functional changed.

Signed-off-by: Xie XiuQi <[email protected]>
Reviewed-by: Tejun Heo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

bitops/find: clarify and extend documentation

Add return value documentation and clarify the units of the @size
parameter.

Signed-off-by: Cody P Schafer <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Cody P Schafer <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/video/backlight/hx8357.c: remove redundant of_match_ptr

'hx8357_dt_ids' is always compiled in.  Hence of_match_ptr is not needed.

Signed-off-by: Sachin Kamat <[email protected]>
Acked-by: Maxime Ripard <[email protected]>
Acked-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: atmel-pwm-bl: use gpio_request_one

Use devm_gpio_request_one rather than requesting and setting direction
in two calls.

Signed-off-by: Johan Hovold <[email protected]>
Cc: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: atmel-pwm-bl: refactor gpio_on handling

Add helper function to control the gpio_on signal.

Signed-off-by: Johan Hovold <[email protected]>
Acked-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: atmel-pwm-bl: use gpio_is_valid

Use gpio_is_valid rather than open coding the more restrictive != -1
test.

Signed-off-by: Johan Hovold <[email protected]>
Acked-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: atmel-pwm-bl: remove unused include

Remove unused include of clk.h.

Signed-off-by: Johan Hovold <[email protected]>
Acked-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: atmel-pwm-bl: clean up get_intensity

Clean up get_intensity to increase readability.

Signed-off-by: Johan Hovold <[email protected]>
Acked-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: atmel-pwm-bl: clean up probe error handling

Clean up probe error handling by checking parameters before any
allocations and removing an obsolete error label.  Also remove unnecessary
reset of private gpio number.

Signed-off-by: Johan Hovold <[email protected]>
Acked-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: atmel-pwm-bl: fix module autoload

Add missing module alias which is needed for module autoloading.

Signed-off-by: Johan Hovold <[email protected]>
Acked-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: atmel-pwm-bl: fix gpio polarity in remove

Make sure to honour gpio polarity also at remove so that the backlight is
actually disabled on boards with active-low enable pin.

Signed-off-by: Johan Hovold <[email protected]>
Acked-by: Jingoo Han <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: atmel-pwm-bl: fix reported brightness

The driver supports 16-bit brightness values, but the value returned
from get_brightness was truncated to eight bits.

Signed-off-by: Johan Hovold <[email protected]>
Cc: Jingoo Han <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/video/backlight/lm3630a_bl.c: add missing destroy_workqueue() on error in lm3630a_intr_config()

Add the missing destroy_workqueue() before return from
lm3630a_intr_config() in the error handling case.

Signed-off-by: Wei Yongjun <[email protected]>
Acked-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: s6e63m0: use devm_{backlight,lcd}_device_register()

Use devm_backlight_device_register() and devm_lcd_device_register() to
make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: cr_bllcd: use devm_{backlight,lcd}_device_register()

Use devm_backlight_device_register() and devm_lcd_device_register() to
make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: corgi_lcd: use devm_{backlight,lcd}_device_register()

Use devm_backlight_device_register() and devm_lcd_device_register() to
make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: ld9040: use devm_{backlight,lcd}_device_register()

Use devm_backlight_device_register() and devm_lcd_device_register() to
make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: ams369fg06: use devm_{backlight,lcd}_device_register()

Use devm_backlight_device_register() and devm_lcd_device_register() to
make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: tdo24m: use devm_lcd_device_register()

Use devm_lcd_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: platform_lcd: use devm_lcd_device_register()

Use devm_lcd_device_register() to make cleanup paths simpler, and remove
unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: ltv350qv: use devm_lcd_device_register()

Use devm_lcd_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: lms501kf03: use devm_lcd_device_register()

Use devm_lcd_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: lms283gf05: use devm_lcd_device_register()

Use devm_lcd_device_register() to make cleanup paths simpler, and remove
unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: ili9320: use devm_lcd_device_register()

Use devm_lcd_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: ili922x: use devm_lcd_device_register()

Use devm_lcd_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: hx8357: use devm_lcd_device_register()

Use devm_lcd_device_register() to make cleanup paths simpler, and remove
unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: wm831x_bl: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler, and
remove unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: tps65217_bl: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler, and
remove unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: pcf50633: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler, and
remove unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: pandora_bl: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler, and
remove unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: max8925_bl: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler, and
remove unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: lv5207lp: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: lp855x: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler.  Also,
unnecessary lp855x_backlight_unregister() is removed.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: lm3533_bl: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: kb3886_bl: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler, and
remove unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: gpio_backlight: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler, and
remove unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: generic_bl: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: ep93xx: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler, and
remove unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: da9052_bl: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: da903x_bl: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler, and
remove unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: bd6107: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: atmel-pwm-bl: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: as3711_bl: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler, and
remove unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: adp8870: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: adp8860: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: adp5520: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: aat2870: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: 88pm860x_bl: use devm_backlight_device_register()

Use devm_backlight_device_register() to make cleanup paths simpler, and
remove unnecessary remove().

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: use dev_get_platdata()

Use the wrapper function for retrieving the platform data instead of
accessing dev->platform_data directly.  This is a cosmetic change to
make the code simpler and enhance the readability.

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: lp8788: staticize 'default_bl_config'

Fix the following sparse warning:

  drivers/video/backlight/lp8788_bl.c:55:25: warning: symbol 'default_bl_config' was not declared. Should it be static?

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: lm3639: don't mix different enum types

Don't mix different enum types to fix the sparse warnings.

  drivers/video/backlight/lm3639_bl.c:80:51: warning: mixing different enum types
  drivers/video/backlight/lm3639_bl.c:80:51:     int enum lm3639_fleds  versus
  drivers/video/backlight/lm3639_bl.c:80:51:     int enum lm3639_bleds
  drivers/video/backlight/lm3639_bl.c:82:51: warning: mixing different enum types
  drivers/video/backlight/lm3639_bl.c:82:51:     int enum lm3639_fleds  versus
  drivers/video/backlight/lm3639_bl.c:82:51:     int enum lm3639_bleds

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: ld9040: Staticize local variable gamma_table

Fix the following sparse warnings:

  drivers/video/backlight/ld9040_gamma.h:172:3: warning: symbol 'gamma_table' was not declared. Should it be static?

Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: lm3630: fix sparse warning

This patch is to fix sparse warning due to mixing different enum type.

Signed-off-by: Daniel Jeong <[email protected]>
Cc: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/video/backlight/lm3630a_bl.c: potential NULL deref in probe()

We dereference "pdata" later in the function so we can't leave it as NULL.

Signed-off-by: Dan Carpenter <[email protected]>
Acked-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/video/backlight/lm3630a_bl.c: fix signedness bug in lm3630a_chip_init()

"rval" needs to be signed for the error handling to work.

Signed-off-by: Dan Carpenter <[email protected]>
Acked-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: lm3630: apply chip revision

The LM3630 chip was revised by TI and chip name was also changed to
LM3630A.  And register map, default values and initial sequences are
changed.  The files, lm3630_bl.{c,h} are replaced by lm3630a_bl.{c,h} You
can find more information about LM3630A(datasheet, evm etc) at
http://www.ti.com/product/lm3630a

Signed-off-by: Daniel Jeong <[email protected]>
Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

backlight: lp855x_bl: support new LP8555 device

LP8555 is one of the LP855x family devices.

This device needs pre_init_device() and post_init_device() driver
structure.  It's same as LP8557, so the device configuration code is
shared with LP8557.  Backlight outputs are generated from dual DC-DC boost
converters.  It's configurable EPROM settings which are defined in the
platform data.

Driver documentation and device tree bindings are updated.

Signed-off-by: Milo Kim <[email protected]>
Signed-off-by: Jingoo Han <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

MAINTAINERS: update Zwane Mwaikambo's e-mail address

Zwane Mwaikambo's @arm.linux.org.uk address no longer works.  In February
2013 he asked for his gmail address to be used instead [1] so let's just
do that.

[1] http://marc.info/?l=linux-kernel&m=136079068903214&w=2

Signed-off-by: Jean Delvare <[email protected]>
Cc: Russell King <[email protected]>
Cc: Zwane Mwaikambo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

MAINTAINERS: remove Richard Purdie as backlight maintainer

Remove Richard Purdie as backlight subsystem maintainer,

akpm: Richard is still responsive and is reviewing some of the patches,
but appears to agree that listing him as the maintainer is no longer
appropriate.

[[email protected]: s/USA/UK/]
Signed-off-by: Jingoo Han <[email protected]>
Acked-by: Richard Purdie <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

printk.c: comments should refer to /proc/vmcore instead of /proc/vmcoreinfo

In one of those comments a typo was fixed, too.

Signed-off-by: Dirk Gouders <[email protected]>
Cc: Joe Perches <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

kernel/printk/printk.c: enable boot delay for earlyprintk

boot_delay does not work for earlyprintk because the kernel cmdline
parsing is late.

Change to use early_param so early kernel messages can also be delayed.

Signed-off-by: Dave Young <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

vsprintf: check real user/group id for %pK

Some setuid binaries will allow reading of files which have read
permission by the real user id.  This is problematic with files which
use %pK because the file access permission is checked at open() time,
but the kptr_restrict setting is checked at read() time.  If a setuid
binary opens a %pK file as an unprivileged user, and then elevates
permissions before reading the file, then kernel pointer values may be
leaked.

This happens for example with the setuid pppd application on Ubuntu 12.04:

  $ head -1 /proc/kallsyms
  00000000 T startup_32

  $ pppd file /proc/kallsyms
  pppd: In file /proc/kallsyms: unrecognized option 'c1000000'

This will only leak the pointer value from the first line, but other
setuid binaries may leak more information.

Fix this by adding a check that in addition to the current process having
CAP_SYSLOG, that effective user and group ids are equal to the real ids.
If a setuid binary reads the contents of a file which uses %pK then the
pointer values will be printed as NULL if the real user is unprivileged.

Update the sysctl documentation to reflect the changes, and also correct
the documentation to state the kptr_restrict=0 is the default.

This is a only temporary solution to the issue.  The correct solution is
to do the permission check at open() time on files, and to replace %pK
with a function which checks the open() time permission.  %pK uses in
printk should be removed since no sane permission check can be done, and
instead protected by using dmesg_restrict.

Signed-off-by: Ryan Mallon <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Alexander Viro <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: "Eric W. Biederman" <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

kernel/printk/printk.c: convert to pr_foo()

It was half-and-half.

Cc: Frederic Weisbecker <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Tejun Heo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

printk: report console names during cut-over

This reports the names of consoles as they're being disabled to help
identify which is which during cut-over.  Helps answer the question
"which boot console actually got activated?" once the regular console is
running, mostly when debugging boot console failures.

Signed-off-by: Kees Cook <[email protected]>
Acked-by: Frederic Weisbecker <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Tejun Heo <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

init/main.c: remove prototype for softirq_init()

It's already available in <linux/interrupt.h>

Signed-off-by: Geert Uytterhoeven <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sched: remove ARCH specific fpu_counter from task_struct

fpu_counter in task_struct was used only by sh/x86.  Both of these now
carry it in ARCH specific thread_struct, hence this can now be removed
from generic task_struct, shrinking it slightly for other arches.

Signed-off-by: Vineet Gupta <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Paul Mundt <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

x86: move fpu_counter into ARCH specific thread_struct

Only a couple of arches (sh/x86) use fpu_counter in task_struct so it can
be moved out into ARCH specific thread_struct, reducing the size of
task_struct for other arches.

Compile tested i386_defconfig + gcc 4.7.3

Signed-off-by: Vineet Gupta <[email protected]>
Acked-by: Ingo Molnar <[email protected]>
Cc: Paul Mundt <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sh: move fpu_counter into ARCH specific thread_struct

Only a couple of arches (sh/x86) use fpu_counter in task_struct so it can
be moved out into ARCH specific thread_struct, reducing the size of
task_struct for other arches.

Compile tested sh defconfig + sh4-linux-gcc (4.6.3)

Signed-off-by: Vineet Gupta <[email protected]>
Cc: Paul Mundt <[email protected]>
Cc: Ingo Molnar <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

jump_label: unlikely(x) > 0

if (unlikely(x) > 0) doesn't seem to help branch prediction

Signed-off-by: Roel Kluin <[email protected]>
Cc: Raghavendra K T <[email protected]>
Cc: Konrad Rzeszutek Wilk <[email protected]>
Cc: "H. Peter Anvin" <[email protected]>
Cc: Ingo Molnar <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

kernel/sys.c: remove obsolete #include <linux/kexec.h>

Commit 15d94b82565e ("reboot: move shutdown/reboot related functions to
kernel/reboot.c") moved all kexec-related functionality to
kernel/reboot.c, so kernel/sys.c no longer needs to include
<linux/kexec.h>.

Signed-off-by: Geert Uytterhoeven <[email protected]>
Cc: Robin Holt <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

kernel/delayacct.c: remove redundant checking in __delayacct_add_tsk()

The wrapper function delayacct_add_tsk() already checked 'tsk->delays',
and __delayacct_add_tsk() has no another direct callers, so can remove the
redundancy checking code.

And the label 'done' is also useless, so remove it, too.

Signed-off-by: Chen Gang <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

gen_init_cpio: avoid NULL pointer dereference and rework env expanding

getenv() may return NULL if given environment variable does not exist
which leads to NULL dereference when calling strncat.

Besides that, the environment variable name was copied to a temporary
env_var buffer, but this copying can be avoided by simply using the input
string.

Lastly, the whole loop can be greatly simplified by using the snprintf
function instead of the playing with strncat.

 By the way, the current implementation allows a recursive variable
 expansion, as in:

   $ echo 'out ${A} out ' | A='a ${B} a' B=b /tmp/a
   out a b a out

 I'm assuming this is just a side effect and not a conscious decision
 (especially as this may lead to infinite loop), but I didn't want to
 change this behaviour without consulting.

 If the current behaviour is deamed incorrect, I'll be happy to send
 a patch without recursive processing.

Signed-off-by: Michal Nazarewicz <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Jiri Kosina <[email protected]>
Cc: Jesper Juhl <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

errno.h: remove "NFS" from descriptions in comments

glibc recently changed the error string for ESTALE to remove "NFS" -

https://sourceware.org/git/?p=glibc.git;a=commitdiff;h=96945714ec61951cc748da2b4b8a80cf02127ee9

from: [ERR_REMAP (ESTALE)] = N_("Stale NFS file handle"),
to:   [ERR_REMAP (ESTALE)] = N_("Stale file handle"),

And some have expressed concern that the kernel's errno.h
comments still refer to NFS.

So make that change... note that this is a comment-only change,
and has no functional difference.

Signed-off-by: Eric Sandeen <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

init/do_mounts.c: add maj:min syntax comment

The name_to_dev_t function has a comment block which lists the supported
syntaxes for the device name.  Add a bullet for the <major>:<minor>
syntax, which is already supported in the code

Signed-off-by: Sebastian Capella <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

scripts/mod/modpost.c: handle non ABS crc symbols

For some reason I managed to trick gcc into create CRC symbols that are
not absolute anymore, but weak.

Make modpost handle this case.

Signed-off-by: Andi Kleen <[email protected]>
Cc: Al Viro <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Tetsuo Handa <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>