platform/kernel/linux-exynos.git
8 years agostaging/lustre/ldlm: Fix Multiple Assignments
Nathaniel Clark [Fri, 10 Jun 2016 02:35:13 +0000 (22:35 -0400)]
staging/lustre/ldlm: Fix Multiple Assignments

Fix all multiple assignments on lustre/ldlm directory.

Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Signed-off-by: Oleg Drokin <green@linuxhacker.ru>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging/lustre/fid: Fix Multiple Assignments
Nathaniel Clark [Fri, 10 Jun 2016 02:35:12 +0000 (22:35 -0400)]
staging/lustre/fid: Fix Multiple Assignments

Fix all multiple assignments on lustre/fid directory.

Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Signed-off-by: Oleg Drokin <green@linuxhacker.ru>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging/lustre/osc: Fix Multiple Assignment Warnings
Nathaniel Clark [Fri, 10 Jun 2016 02:35:11 +0000 (22:35 -0400)]
staging/lustre/osc: Fix Multiple Assignment Warnings

Fix all multiple assignments on lustre/osc directory.

Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Signed-off-by: Oleg Drokin <green@linuxhacker.ru>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: lustre: hide call to Posix ACL in ifdef
Arnd Bergmann [Mon, 13 Jun 2016 20:44:57 +0000 (22:44 +0200)]
staging: lustre: hide call to Posix ACL in ifdef

A call to forget_cached_acl() was recently added to the lustre file
system, but this is only available when CONFIG_FS_POSIX_ACL is
enabled, otherwise the build now fails with:

lustre/llite/file.c: In function 'll_get_acl':
lustre/llite/file.c:3134:2: error: implicit declaration of function 'forget_cached_acl' [-Werror=implicit-function-declaration]
  forget_cached_acl(inode, type);

This adds one more #ifdef for this call, corresponding to the
other 22 such checks for ACL in lustre.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: b788dc51e425 ("staging: lustre: llite: drop acl from cache")
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: lustre: lnet: Replace semaphore ln_rc_signal with completion
Binoy Jayan [Wed, 8 Jun 2016 07:50:42 +0000 (13:20 +0530)]
staging: lustre: lnet: Replace semaphore ln_rc_signal with completion

The semaphore ln_rc_signal is used as completion, so convert it to
struct completion.  Semaphores are going away in the future.

Signed-off-by: Binoy Jayan <binoy.jayan@linaro.org>
Acked-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: lustre: llite: remove lloop device
James Simmons [Wed, 8 Jun 2016 22:50:12 +0000 (18:50 -0400)]
staging: lustre: llite: remove lloop device

The lloop device was original developed to work around
the lack of direct I/O for the default loop back device.
Also the lloop device greatly out performed the default
loop back device. The lloop hasn't been worked on for
some time and now it no longer out performs the loop
device and loop now supports direct I/O. Since this is
the case we can delete this device.

Signed-off-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agoMerge tag 'iio-for-4.8a' of git://git.kernel.org/pub/scm/linux/kernel/git/jic23/iio...
Greg Kroah-Hartman [Thu, 9 Jun 2016 16:15:58 +0000 (09:15 -0700)]
Merge tag 'iio-for-4.8a' of git://git./linux/kernel/git/jic23/iio into staging-next

Jonathan writes:

First round of IIO new device support, features and cleanups for the 4.8 cycle.

New device support
* ads1015
  - add ads1115 support
* bma220 accelerometer
  - new driver
  - triggered buffer support.
* bmc150
  - add bmm150 support.
* bmp280
  - bme280 support with addition of humidity channel.
* max5487 potentiometer
  - new driver
* MMA7660FC accelerometer.
  - New driver
* st-pressure
  - support for the lps22hb
* loop trigger.
  - This one is *nasty* but we have real applications (parrot drones) where
  it is useful.  The trigger basically spins as hard as it can firing off
  a new trigger each time all triggered devices come back to say they are
  done.  It doesn't hang a machine even when doing it on a dummy driver.
  A lot nicer than having this implemented within lots of device drivers
  anyway.

Core stuff
* Add support to create IIO devices via configfs (similar to we did for
triggers a while back) + docs.
* New channel types
  - IIO_ELECTRICAL_CONDUCTIVITY
* Couple of MAINTAINERS patches to list the device tree bindings.
* Make trigger ops structure non optional (comment fix). It hasn't been for
an awful long time, but that's not what the description said.

New features
* ak8975
  - support adapters that are limited to byte data only by allowing the
  emulated block read i2c function that was recently introduced.
* atlas-ph
  - support atlas-ec (electrical conductivity sensor)
* bmi160
  - add available frequency and scale attributes to make the driver
  more user friendly (and avoid having to read the datasheet to know
  what will work).
* dummy
  - move creation to configfs interface.  It's not real hardware so we
  are not that worried about the ABI breakage ;)
* mma8452
  - oversampling ration support
* nau7802
  - expose available gains to make life easier for userspace.
* st-sensors
  - allow use of emulation for SMBus block reads as all the st parts support
  it.
* ti-ads1015
  - list datasheet names to allow their use by inkernel consumers.
* Various module alias additions to help auto probing.  Drop one redundant one
as well.

Cleanups
* ad7266, ad7476, ad7887, ad7923, ad799x
  - use direct mode claim function rather than open coding it during sensor
  read (prevents switching on buffers mid read).
* ad7793, ad7791
  - use direct mode claim to prevent frequency changes when buffers running.
* afe440x - These are ABI breaking but the driver requires custom userspace
  code to do anything useful anyway and that is still being written and under
  control of TI.  Ultimately we may have other libraries to do pulse
  oximetry with these devices but we aren't aware of any yet.
  - kernel-doc format fixes
  - drop ifdef fun around of_match_ptr - it's not worth the mess to save
  a tiny amount of space.
  - drop some unnecessary register initializations.
  - drop the weird locked gain modes as they gain us nothing (can just set
  all gains separately).
  - remove handling of offset attributes seeing as no channels actually have
  them (oops)
  - Drop the LED3 input channel as it's an alias for ALED2.
  - *big one* remove channel names - an experiment that turned out to not
  make sense - see patch for details.
  - use regmap fields to clean up code.
  - tie the tia gain stages to appropriate channels in the ABI as that is
  what they really effect. Same with the LED currents.
  - cleanout some unused defines and fix a missnamed one.
* atlas-ph
  - reorganise to allow support of other similar parts.
* bmc150
  - document supported chips in kconfig help.
* jsa1212
  - drop an unneeded i2c functionality check for functionality the driver
  doesn't use.
* mxs-lradc
  - simply touch screen registration code.
  - remove the touch screen unregister as all devm based now.
  - disable only those channels that are masked in hardware stop (others
  are already dealt with elsewhere)
* st-sensors
  - unexport st_sensors_get_buffer_element as nothing outside the st-sensors
  core driver uses it.
  - fix handling of failure to start up regulators.
* tpl0102
  - drop an i2c functionality test for features that aren't needed.
* ti-am335x
  - use variable name rather than type in sizeof for clarity.
  - use SIMPLE_DEV_PM_OPS helper macro to tidy up a bit.

Tools
* Add install / uninstall to makefile.  Someone cares, so presumably
some people will find it useful!
*  generic_buffer
   - rename to iio_generic_buffer to line up with other tools.
   - handle cleanup when receiving signals
   - Add a --device-num option and a --trigger-num option rather than
   relying on naming which doesn't work if you have two of the same part.

8 years agostaging: unisys: visornic: change return statements
David Binder [Sat, 14 May 2016 03:17:22 +0000 (23:17 -0400)]
staging: unisys: visornic: change return statements

Changes return statements in visornic_rx() to use literals instead of a
variable. Also changes function description to reflect the correct return
type.

Signed-off-by: David Binder <david.binder@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: iovmcall_gnuc.h change -1 return values
Erik Arfvidson [Sat, 14 May 2016 03:17:21 +0000 (23:17 -0400)]
staging: unisys: iovmcall_gnuc.h change -1 return values

This patch changes the vague -1 return values to -EPERM.
This operation is not supported is a good alternative
to -1 because the return is basically telling the caller
that the processor doesn't support vmcall operations.

Signed-off-by: Erik Arfvidson <erik.arfvidson@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visorchipset change -1 return value
Erik Arfvidson [Sat, 14 May 2016 03:17:20 +0000 (23:17 -0400)]
staging: unisys: visorchipset change -1 return value

This patch changes the vague -1 return value to -EINVAL

Signed-off-by: Erik Arfvidson <erik.arfvidson@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visorbus change -1 return values
Erik Arfvidson [Sat, 14 May 2016 03:17:19 +0000 (23:17 -0400)]
staging: unisys: visorbus change -1 return values

This patch changes the vague -1 return values to -EFAULT since
it would be the most appropriate, given that this error
would only occur in an unexpected bad offset field.
Resulting in a bad address.

Signed-off-by: Erik Arfvidson <erik.arfvidson@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visorhba change -1 return value
Erik Arfvidson [Sat, 14 May 2016 03:17:18 +0000 (23:17 -0400)]
staging: unisys: visorhba change -1 return value

This patch changes the vague -1 return value to -EBUSY

Signed-off-by: Erik Arfvidson <erik.arfvidson@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visorinput change -1 return value
Erik Arfvidson [Sat, 14 May 2016 03:17:17 +0000 (23:17 -0400)]
staging: unisys: visorinput change -1 return value

This patch changes the vague -1 return value to -EINVAL

Signed-off-by: Erik Arfvidson <erik.arfvidson@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visorhba: "Prefer 'unsigned int'" checkpatch warnings
Tim Sell [Sat, 14 May 2016 03:17:16 +0000 (23:17 -0400)]
staging: unisys: visorhba: "Prefer 'unsigned int'" checkpatch warnings

This patch fixes a few checkpatch warnings in visorhba:

    WARNING: Prefer 'unsigned int' to bare use of 'unsigned'

Signed-off-by: Tim Sell <Timothy.Sell@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visornic: remove extraneous error check
David Binder [Sat, 14 May 2016 03:17:15 +0000 (23:17 -0400)]
staging: unisys: visornic: remove extraneous error check

Removes an extraneous error check in devdata_initialize(), and updates the
function comment accordingly.

Signed-off-by: David Binder <david.binder@unisys.com>
Reviewed-by: Tim Sell <Timothy.Sell@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visornic: check for error instead of success
David Binder [Thu, 12 May 2016 13:14:48 +0000 (09:14 -0400)]
staging: unisys: visornic: check for error instead of success

Changes the conditional logic to check for an error code instead
of a success code.

Signed-off-by: David Binder <david.binder@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visorhba: return 0 literal
David Binder [Thu, 12 May 2016 13:14:47 +0000 (09:14 -0400)]
staging: unisys: visorhba: return 0 literal

Returns 0 instead of variable rc in visorhba_init().

Signed-off-by: David Binder <david.binder@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visornic: cleanup error handling
David Binder [Thu, 12 May 2016 13:14:46 +0000 (09:14 -0400)]
staging: unisys: visornic: cleanup error handling

Adjusts goto labels to prevent attempts to free unallocated resources.

Signed-off-by: David Binder <david.binder@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visornic: simplify visornic if statements
David Binder [Thu, 12 May 2016 13:14:45 +0000 (09:14 -0400)]
staging: unisys: visornic: simplify visornic if statements

Changes the conditional logic by looking for the absence of work
to do, instead of the opposite.

Signed-off-by: David Binder <david.binder@unisys.com>
Reviewed-by: Tim Sell <Timothy.Sell@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visorhba: visorhbas_open[] no longer used, so deleted
Tim Sell [Thu, 12 May 2016 13:14:44 +0000 (09:14 -0400)]
staging: unisys: visorhba: visorhbas_open[] no longer used, so deleted

The prior patch which simplified the visorhba debugfs interface made it so
visorhbas_open[] and VISORHBA_OPEN_MAX were no longer needed, so they have
now been deleted.

Signed-off-by: Tim Sell <Timothy.Sell@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visorhba: simplify and enhance debugfs interface
Tim Sell [Thu, 12 May 2016 13:14:43 +0000 (09:14 -0400)]
staging: unisys: visorhba: simplify and enhance debugfs interface

debugfs info for each visorhba device is now presented by a file named of
the following form within the debugfs tree:

    visorhba/vbus<x>:dev<y>/info

where <x> is the vbus number, and <y> is the relative device number.

Also, the debugfs presentation function was converted to use the seq_file
interface, so that it could access the device context without resorting to
a global array.  This also simplified the function.

Signed-off-by: Tim Sell <Timothy.Sell@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visorhba: remove unused (and broken) logic
Tim Sell [Thu, 12 May 2016 13:14:42 +0000 (09:14 -0400)]
staging: unisys: visorhba: remove unused (and broken) logic

The handling of CMD_NOTIFYGUEST_TYPE messages from the IO partition appears
to be only partially implemented, but fortunately it is never used in our
current environment.  This patch deletes the unused code.

Signed-off-by: Tim Sell <Timothy.Sell@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visorhba: correct scsi task mgmt completion handling
Tim Sell [Thu, 12 May 2016 13:14:41 +0000 (09:14 -0400)]
staging: unisys: visorhba: correct scsi task mgmt completion handling

This patch is necessary to enable ANY task mgmt command to complete
successfully via visorhba.

When issuing a task mgmt command (CMD_SCSITASKMGMT_TYPE) to the IO
partition (back-end), forward_taskmgmt_command() includes pointers
within the command area that will be used to wake up the issuing
process and provide the result when the command completes:

    cmdrsp->scsitaskmgmt.notify_handle = (u64)&notifyevent;
    cmdrsp->scsitaskmgmt.notifyresult_handle = (u64)&notifyresult;

'notify_handle' is a pointer to a 'wait_queue_head_t' variable, and
'notifyresult' is a pointer to an int.  Both of these are just local
stack variables in the issuing process.

The way it's supposed to happen is that when the IO partition completes
the command, in our completion handling we get copies of those pointers
back from the IO partition, where we stash the result of the command at
'*notifyresult' (which should not be 0xffff, because that is the initial
value that the caller is looking to see a change in), and wake up the
wait queue at '*notify_handle'.  There are several places we do that dance,
but prior to this patch, we always do it WRONG, like:

    cmdrsp->scsitaskmgmt.notifyresult_handle = TASK_MGMT_FAILED;
    wake_up_all((wait_queue_head_t *)cmdrsp->scsitaskmgmt.notify_handle);

The wake_up_all() part is correct (albeit with the help of the sloppy
pointer casting, but that's irrelevant to the bug), but the assignment of
'notifyresult_handle' is WRONG, and SHOULD read:

    *(int *)(cmdrsp->scsitaskmgmt.notifyresult_handle) = TASK_MGMT_FAILED;

Without this change, the caller is NEVER going to notice a change in his
local value of 'notifyresult' when he does the:

    if (!wait_event_timeout(notifyevent, notifyresult != 0xffff,
                            msecs_to_jiffies(45000)))

and hence will be timing out EVERY taskmgmt command.

This patch also eliminates the need for sloppy casting of pointers
back-and-forth between u64 values, with the help of idr_alloc() to provide
handles for us.  It is the generated int handles we pass to the IO
partition to denote our completion context, and these are validated and
converted back to the required pointers when the task mgmt commands are
returned back to us by the IO partition.

== Testing ==

You must enable dynamic debugging in visorhba (build kernel with
'CONFIG_DYNAMIC_DEBUG=y', provide kernel parameter 'visorhba.dyndbg=+p')
to see kernel messages involved with visorhba scsi task mgmt commands,
which were added in this patch in the form of a few dev_dbg() / pr_debug()
messages.

In order to inject faults necessary to get visorhba to actully issue scsi
task mgmt commands, you will need to compile a kernel with
CONFIG_FAIL_IO_TIMEOUT and friends, in the "Kernel hacking" section:
* Enable "Fault-injection framework"
  * Enable "Fault-injection capability for disk IO"
  * Enable "Fault-injection capability for faking disk interrupts"
* Enable "Debugfs entries for fault-injection capabilities"

When running a kernel with those options, you can manually inject a fault
that will force a scsi task mgmt command to be issued like this:

    # mount -t debugfs nodev /sys/kernel/debug
    # cd /sys/kernel/debug/fail_io_timeout
    # cat interval
    1
    # cat probability
    0
    # cat times
    1
    # echo 100 >probability
    # cd /sys/block/sda
    # l | grep fail
    -rw-r--r--  1 root root 4096 May  5 10:53 io-timeout-fail
    -rw-r--r--  1 root root 4096 May  5 10:54 make-it-fail
    # echo 1 >io-timeout-fail
    # echo 1 >make-it-fail

To test this patch, after performing the above steps, I did something to
force a block device i/o, then shortly afterwards examined the kernel log.
There I found evidence that visorhba had successfully issued a task mgmt
command, and that it completed successfully:

    [  333.352612] FAULT_INJECTION: forcing a failure.
    name fail_io_timeout, interval 1, probability 100, space 0, times 1
    [  333.352617] CPU: 0 PID: 295 Comm: vhba_incoming Tainted: G         C
                   4.6.0-rc3-ARCH+ #2
    [  333.352619] Hardware name: Dell Inc. PowerEdge T110/ ,
                   BIOS 1.23 12/15/2009
    [  333.352620]  0000000000000000 ffff88001d1a7dd0 ffffffff8125beeb
                    ffffffff818507c0
    [  333.352623]  0000000000000064 ffff88001d1a7df0 ffffffff8128047a
                    ffff8800113462b0
    [  333.352625]  ffff88000e523000 ffff88001d1a7e00 ffffffff81241c79
                    ffff88001d1a7e18
    [  333.352627] Call Trace:
    [  333.352634]  [<ffffffff8125beeb>] dump_stack+0x4d/0x72
    [  333.352637]  [<ffffffff8128047a>] should_fail+0x11a/0x120
    [  333.352641]  [<ffffffff81241c79>] blk_should_fake_timeout+0x29/0x30
    [  333.352643]  [<ffffffff81241c36>] blk_complete_request+0x16/0x30
    [  333.352654]  [<ffffffffa0118b36>] scsi_done+0x26/0x80 [scsi_mod]
    [  333.352657]  [<ffffffffa014a56c>] process_incoming_rsps+0x2bc/0x770
                                         [visorhba]
    [  333.352661]  [<ffffffff81095630>] ? wait_woken+0x80/0x80
    [  333.352663]  [<ffffffffa014a2b0>] ? add_scsipending_entry+0x100/0x100
                                         [visorhba]
    [  333.352666]  [<ffffffff81077759>] kthread+0xc9/0xe0
    [  333.352669]  [<ffffffff814609d2>] ret_from_fork+0x22/0x40
    [  333.352671]  [<ffffffff81077690>] ? kthread_create_on_node+0x180/0x180
    [  364.025672] sd 0:0:1:1: visorhba: initiating type=1 taskmgmt command
    [  364.029721] visorhba: notifying initiator with result=0x1
    [  364.029726] sd 0:0:1:1: visorhba: taskmgmt type=1 success; result=0x1

Signed-off-by: Tim Sell <Timothy.Sell@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: unisys: visorhba: delete processing of vdiskmgmt commands
Tim Sell [Thu, 12 May 2016 13:14:40 +0000 (09:14 -0400)]
staging: unisys: visorhba: delete processing of vdiskmgmt commands

We never issue SCSI commands of type CMD_VDISKMGMT_TYPE, so there is no
need to have code that processes their completions.

Signed-off-by: Tim Sell <Timothy.Sell@unisys.com>
Signed-off-by: David Kershner <david.kershner@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: cleanup file headers
Wolfram Sang [Tue, 31 May 2016 10:56:43 +0000 (12:56 +0200)]
staging: ks7010: cleanup file headers

Remove svn-ids and fix typos in the licence declaration. Add my
copyright to the sdio code which I worked on mainly.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: drop counting sd errors
Wolfram Sang [Tue, 31 May 2016 10:56:42 +0000 (12:56 +0200)]
staging: ks7010: drop counting sd errors

They were counted but never really used anywhere. Also change the printk
to a debug print, since it mostly shows on the expected -ENOMEDIUM on
card removal.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: simplify module_init/exit
Wolfram Sang [Tue, 31 May 2016 10:56:41 +0000 (12:56 +0200)]
staging: ks7010: simplify module_init/exit

The printouts are not needed, the driver core has enough debug output
for this if wanted. So, use a helper to save boilerplate code.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: adapt to new trans_start handling
Wolfram Sang [Tue, 31 May 2016 10:56:40 +0000 (12:56 +0200)]
staging: ks7010: adapt to new trans_start handling

trans_start is gone from netdevice, so use the new helper function to
set the mark.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: fix module annotations
Wolfram Sang [Tue, 31 May 2016 10:56:39 +0000 (12:56 +0200)]
staging: ks7010: fix module annotations

List all authors, beautify description, match license to what is stated
in file headers, add firmware information.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: remove supported card table with one element
Wolfram Sang [Tue, 31 May 2016 10:56:38 +0000 (12:56 +0200)]
staging: ks7010: remove supported card table with one element

There is only this card supported, no need to iterate over the table.
The resulting firmware filename wasn't used anyway, but came from the
config file or hardcoded default.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: indent ks7010_sdio.c
Wolfram Sang [Tue, 31 May 2016 10:56:37 +0000 (12:56 +0200)]
staging: ks7010: indent ks7010_sdio.c

Unlike the previous patches which are plain indent outcomes, this has
some manual fixups to be not overly strict with the 80 char limit.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: indent michael_mic.h
Wolfram Sang [Tue, 31 May 2016 10:56:36 +0000 (12:56 +0200)]
staging: ks7010: indent michael_mic.h

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: indent michael_mic.c
Wolfram Sang [Tue, 31 May 2016 10:56:35 +0000 (12:56 +0200)]
staging: ks7010: indent michael_mic.c

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: indent ks_wlan_net.c
Wolfram Sang [Tue, 31 May 2016 10:56:34 +0000 (12:56 +0200)]
staging: ks7010: indent ks_wlan_net.c

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: indent ks_wlan_ioctl.h
Wolfram Sang [Tue, 31 May 2016 10:56:33 +0000 (12:56 +0200)]
staging: ks7010: indent ks_wlan_ioctl.h

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: indent ks_wlan.h
Wolfram Sang [Tue, 31 May 2016 10:56:32 +0000 (12:56 +0200)]
staging: ks7010: indent ks_wlan.h

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: indent ks_hostif.h
Wolfram Sang [Tue, 31 May 2016 10:56:31 +0000 (12:56 +0200)]
staging: ks7010: indent ks_hostif.h

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: indent ks_hostif.c
Wolfram Sang [Tue, 31 May 2016 10:56:30 +0000 (12:56 +0200)]
staging: ks7010: indent ks_hostif.c

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: indent ks7010_sdio.h
Wolfram Sang [Tue, 31 May 2016 10:56:29 +0000 (12:56 +0200)]
staging: ks7010: indent ks7010_sdio.h

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: indent ks7010_config.c
Wolfram Sang [Tue, 31 May 2016 10:56:28 +0000 (12:56 +0200)]
staging: ks7010: indent ks7010_config.c

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: indent eap_packet.h
Wolfram Sang [Tue, 31 May 2016 10:56:27 +0000 (12:56 +0200)]
staging: ks7010: indent eap_packet.h

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: remove unecessary typedef
Wolfram Sang [Tue, 31 May 2016 10:56:26 +0000 (12:56 +0200)]
staging: ks7010: remove unecessary typedef

Let's simply specify the struct to keep in sync with kernel coding
style.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: add example cfg file as a reference
Wolfram Sang [Tue, 31 May 2016 10:56:25 +0000 (12:56 +0200)]
staging: ks7010: add example cfg file as a reference

We want to remove it, but to do so properly, it is good to have a
working example. Needs to be copied to /lib/firmware in order to be
used.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: fix printk format warnings
Wolfram Sang [Tue, 31 May 2016 10:56:24 +0000 (12:56 +0200)]
staging: ks7010: fix printk format warnings

Use proper type for size_t.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: make loading config file optional
Wolfram Sang [Tue, 31 May 2016 10:56:23 +0000 (12:56 +0200)]
staging: ks7010: make loading config file optional

We have sane defaults, so we don't need to bail out if there is no
config file. Note that the config file should go away completely in
favour of configuration mechanisms already upstream.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: really iterate over multicast addresses
Wolfram Sang [Tue, 31 May 2016 10:56:22 +0000 (12:56 +0200)]
staging: ks7010: really iterate over multicast addresses

The loop variable was defined but not really used. Fix this.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: delete seperate debug header
Wolfram Sang [Tue, 31 May 2016 10:56:21 +0000 (12:56 +0200)]
staging: ks7010: delete seperate debug header

Move the one debug macro to the generic wlan header.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: use kernel helper to print buffer
Wolfram Sang [Tue, 31 May 2016 10:56:20 +0000 (12:56 +0200)]
staging: ks7010: use kernel helper to print buffer

No need for an open coded one.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: use long preamble as default
Wolfram Sang [Tue, 31 May 2016 10:56:19 +0000 (12:56 +0200)]
staging: ks7010: use long preamble as default

I had a problem connecting to a network with a short preamble, so let's
make the safer option the default.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: avoid workqueue races
Wolfram Sang [Tue, 31 May 2016 10:56:18 +0000 (12:56 +0200)]
staging: ks7010: avoid workqueue races

My Spectec SDW823 card oopsed when it was already inserted during boot.
When debugging this, I noticed that the card init was done in a seperate
workqueue which was only activated once in probe. After removing the
workqueue and calling the card init directly from probe, the OOPS went
away. It turned out this is the same OOPS which happened when removing
the card, so this seems possible now. Note: There is still a
not-understood card-removed event during boot, but at least it doesn't
crash anymore and the card will be re-probed right away.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: remove code for old kernel versions
Wolfram Sang [Tue, 31 May 2016 10:56:17 +0000 (12:56 +0200)]
staging: ks7010: remove code for old kernel versions

No need to be backwards compatible.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: remove checks for WIRELESS_EXT version
Wolfram Sang [Tue, 31 May 2016 10:56:16 +0000 (12:56 +0200)]
staging: ks7010: remove checks for WIRELESS_EXT version

We are by far newer than that anyhow.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: remove custom firmware loader
Wolfram Sang [Tue, 31 May 2016 10:56:15 +0000 (12:56 +0200)]
staging: ks7010: remove custom firmware loader

FW_LOADER works fine, no need for a open coded fallback.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: remove non-SDIO code and #ifdefs
Wolfram Sang [Tue, 31 May 2016 10:56:14 +0000 (12:56 +0200)]
staging: ks7010: remove non-SDIO code and #ifdefs

I couldn't find any trace of code or even products using ks7010 with
something else than SDIO. So, remove the conditionals.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: ks7010: add driver from Nanonote extra-repository
Wolfram Sang [Tue, 31 May 2016 10:56:13 +0000 (12:56 +0200)]
staging: ks7010: add driver from Nanonote extra-repository

See the TODO for details where this driver came from. Only a few minor
changes were made to make the driver suitable for staging:

* updated Kconfig help text and dependencies
* added TODO
* removed two __DATE__ and __TIME__ printouts to allow reproducible builds
* added to staging main Kconfig + Makefile

Tested on a Renesas Salvator-X board with a Spectec SDW-823 card. I
could connect to a WPA-protected network.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: lustre: set function scope with static
Tobin C Harding [Mon, 23 May 2016 10:17:46 +0000 (20:17 +1000)]
staging: lustre: set function scope with static

A number of function definitions were found to be candidates for
static scoping. This patch adds static to these functions.

Signed-off-by: Tobin C Harding <me@tobin.cc>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging/lustre/osc: Get rid of osc_page_protected()
Oleg Drokin [Mon, 6 Jun 2016 03:28:53 +0000 (23:28 -0400)]
staging/lustre/osc: Get rid of osc_page_protected()

There was a proper debugging function by that name that's long
gone.
The currently remaining shadow that always returns true is not
really useful so it could be dropped along with all the
asserts it is part of.

Signed-off-by: Oleg Drokin <green@linuxhacker.ru>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging/lustre/osc: Remove ops_temp from osc_page
Oleg Drokin [Mon, 6 Jun 2016 03:28:52 +0000 (23:28 -0400)]
staging/lustre/osc: Remove ops_temp from osc_page

It's no longer used and never set anywhere.

Signed-off-by: Oleg Drokin <green@linuxhacker.ru>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging/lustre/llite: define per open file cache for ll_cl_context
Jinshan Xiong [Mon, 6 Jun 2016 03:28:51 +0000 (23:28 -0400)]
staging/lustre/llite: define per open file cache for ll_cl_context

In ll_readpage and ll_write_begin, it needs to find out the cl_env
and cl_io, a.k.a ll_cl_context, when the IO is initialized. It used
to call cl_env_get() to figure it out but turned out to be contended
if multiple threads are doing IO.

In this patch, a per open file ll_cl_context cache is created. When
IO type of CIT_READ, CIT_WRITE and CIR_FAULT is initialized, it will
add a ll_cl_context into the cache maintained in ll_file_data. In this
case, the ll_cl_context can be found in ll_readpage and ll_write_begin
later.

Signed-off-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-on: http://review.whamcloud.com/10503
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-5108
Reviewed-on: http://review.whamcloud.com/10955
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-5260
Reviewed-by: Lai Siyao <lai.siyao@intel.com>
Reviewed-by: Bobi Jam <bobijam@gmail.com>
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging/lustre/lov: calculate file offset correctly
Jinshan Xiong [Mon, 6 Jun 2016 03:28:50 +0000 (23:28 -0400)]
staging/lustre/lov: calculate file offset correctly

In lov_stripe_pgoff(), it calls lov_stripe_size() to calculate the
file size by ost_size, which will be wrong if the stripe_index
happens to be stripe aligned.

Signed-off-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-on: http://review.whamcloud.com/14462
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-6482
Reviewed-by: Bobi Jam <bobijam@hotmail.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: lustre: llite: drop acl from cache
James Simmons [Tue, 24 May 2016 00:35:08 +0000 (20:35 -0400)]
staging: lustre: llite: drop acl from cache

Commit b8a7a3a6 change get_acl() for posix xattr to always cache
the ACL which increases the reference count. That reference count
can be reduced by have ll_get_acl() call forget_cached_acl() which
it wasn't. When an inode gets deleted by Lustre the POSIX ACL
reference count is tested to ensure its 1 and if not produces an error.
Since forget_cached_acl() was not called Lustre started to complain.
This patch changes ll_get_acl() to call forget_cached_acl().

Signed-off-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: lustre: llite: kzalloc/copy_to_user to memdup_user
Tobin C Harding [Mon, 23 May 2016 00:14:22 +0000 (10:14 +1000)]
staging: lustre: llite: kzalloc/copy_to_user to memdup_user

kzalloc call followed by copy_to_user can be replaced by call to memdup_user.

Signed-off-by: Tobin C Harding <me@tobin.cc>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging/lustre/ptlrpc: Removes potential null dereference
Lidza Louina [Mon, 16 May 2016 18:51:42 +0000 (14:51 -0400)]
staging/lustre/ptlrpc: Removes potential null dereference

The lustre_msg_buf method could return NULL. Subsequent code didn't
check if it's null before using it. This patch adds two checks.

Signed-off-by: Lidza Louina <lidza.louina@oracle.com>
Acked-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: lustre: use setup_timer().
Muhammad Falak R Wani [Wed, 11 May 2016 15:23:31 +0000 (20:53 +0530)]
staging: lustre: use setup_timer().

Use setup_timer() for initializing the timer, instead of structure
assignments.
This is the preferred/standard way.

Signed-off-by: Muhammad Falak R Wani <falakreyaz@gmail.com>
Acked-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: lustre: ko2iblnd: fix memory corruption with fragments
James Simmons [Mon, 9 May 2016 14:53:48 +0000 (10:53 -0400)]
staging: lustre: ko2iblnd: fix memory corruption with fragments

In my test of the upstream client this change exposed a long
standing issues where we have a offset that is not page algined
would causes us to access memory beyond the scatter gather list
which was causing memory corruption when all 256 fragments were
in use.

Signed-off-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: lustre: libcfs: fix memort leak in libcfs crypto layer
James Simmons [Mon, 9 May 2016 14:53:47 +0000 (10:53 -0400)]
staging: lustre: libcfs: fix memort leak in libcfs crypto layer

During code review Boyko discovered a memory leak. This
patch fixes that leak.

Signed-off-by: James Simmons <uja.ornl@yahoo.com>
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-4423
Reviewed-on: http://review.whamcloud.com/19716
Reviewed-by: Alexander Boyko <alexander.boyko@seagate.com>
Reviewed-by: Frank Zago <fzago@cray.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agostaging: lustre: o2iblnd: make rdma_create_id() support containers
Li Dongyang [Mon, 9 May 2016 14:53:46 +0000 (10:53 -0400)]
staging: lustre: o2iblnd: make rdma_create_id() support containers

Add support for lustre's ko2iblnd driver to work with
containers which was requested by Sebastien Buisson.

Signed-off-by: Li Dongyang <dongyang.li@anu.edu.au>
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-6215
Reviewed-on: http://review.whamcloud.com/18759
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Sebastien Buisson <sbuisson@ddn.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8 years agoLinux 4.7-rc2 v4.7-rc2
Linus Torvalds [Sun, 5 Jun 2016 21:31:26 +0000 (14:31 -0700)]
Linux 4.7-rc2

8 years agoMerge branch 'parisc-4.7-2' of git://git.kernel.org/pub/scm/linux/kernel/git/deller...
Linus Torvalds [Sun, 5 Jun 2016 18:15:33 +0000 (11:15 -0700)]
Merge branch 'parisc-4.7-2' of git://git./linux/kernel/git/deller/parisc-linux

Pull parisc fixes from Helge Deller:

 - Fix printk time stamps on SMP systems which got wrong due to a patch
   which was added during the merge window

 - Fix two bugs in the stack backtrace code: Races in module unloading
   and possible invalid accesses to memory due to wrong instruction
   decoding (Mikulas Patocka)

 - Fix userspace crash when syscalls access invalid unaligned userspace
   addresses.  Those syscalls will now return EFAULT as expected.
   (tagged for stable kernel series)

* 'parisc-4.7-2' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Move die_if_kernel() prototype into traps.h header
  parisc: Fix pagefault crash in unaligned __get_user() call
  parisc: Fix printk time during boot
  parisc: Fix backtrace on PA-RISC

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris...
Linus Torvalds [Sun, 5 Jun 2016 18:02:00 +0000 (11:02 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jmorris/linux-security

Pull key handling update from James Morris:
 "This alters a new keyctl function added in the current merge window to
  allow for a future extension planned for the next merge window"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/linux-security:
  KEYS: Add placeholder for KDF usage with DH

8 years agodevpts: Make each mount of devpts an independent filesystem.
Eric W. Biederman [Thu, 2 Jun 2016 15:29:47 +0000 (10:29 -0500)]
devpts: Make each mount of devpts an independent filesystem.

The /dev/ptmx device node is changed to lookup the directory entry "pts"
in the same directory as the /dev/ptmx device node was opened in.  If
there is a "pts" entry and that entry is a devpts filesystem /dev/ptmx
uses that filesystem.  Otherwise the open of /dev/ptmx fails.

The DEVPTS_MULTIPLE_INSTANCES configuration option is removed, so that
userspace can now safely depend on each mount of devpts creating a new
instance of the filesystem.

Each mount of devpts is now a separate and equal filesystem.

Reserved ttys are now available to all instances of devpts where the
mounter is in the initial mount namespace.

A new vfs helper path_pts is introduced that finds a directory entry
named "pts" in the directory of the passed in path, and changes the
passed in path to point to it.  The helper path_pts uses a function
path_parent_directory that was factored out of follow_dotdot.

In the implementation of devpts:
 - devpts_mnt is killed as it is no longer meaningful if all mounts of
   devpts are equal.
 - pts_sb_from_inode is replaced by just inode->i_sb as all cached
   inodes in the tty layer are now from the devpts filesystem.
 - devpts_add_ref is rolled into the new function devpts_ptmx.  And the
   unnecessary inode hold is removed.
 - devpts_del_ref is renamed devpts_release and reduced to just a
   deacrivate_super.
 - The newinstance mount option continues to be accepted but is now
   ignored.

In devpts_fs.h definitions for when !CONFIG_UNIX98_PTYS are removed as
they are never used.

Documentation/filesystems/devices.txt is updated to describe the current
situation.

This has been verified to work properly on openwrt-15.05, centos5,
centos6, centos7, debian-6.0.2, debian-7.9, debian-8.2, ubuntu-14.04.3,
ubuntu-15.10, fedora23, magia-5, mint-17.3, opensuse-42.1,
slackware-14.1, gentoo-20151225 (13.0?), archlinux-2015-12-01.  With the
caveat that on centos6 and on slackware-14.1 that there wind up being
two instances of the devpts filesystem mounted on /dev/pts, the lower
copy does not end up getting used.

Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Greg KH <greg@kroah.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Peter Anvin <hpa@zytor.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Serge Hallyn <serge.hallyn@ubuntu.com>
Cc: Willy Tarreau <w@1wt.eu>
Cc: Aurelien Jarno <aurelien@aurel32.net>
Cc: One Thousand Gnomes <gnomes@lxorguk.ukuu.org.uk>
Cc: Jann Horn <jann@thejh.net>
Cc: Jiri Slaby <jslaby@suse.com>
Cc: Florian Weimer <fw@deneb.enyo.de>
Cc: Konstantin Khlebnikov <koct9i@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoparisc: Move die_if_kernel() prototype into traps.h header
Helge Deller [Sat, 4 Jun 2016 15:38:09 +0000 (17:38 +0200)]
parisc: Move die_if_kernel() prototype into traps.h header

Signed-off-by: Helge Deller <deller@gmx.de>
8 years agoparisc: Fix pagefault crash in unaligned __get_user() call
Helge Deller [Sat, 4 Jun 2016 15:21:33 +0000 (17:21 +0200)]
parisc: Fix pagefault crash in unaligned __get_user() call

One of the debian buildd servers had this crash in the syslog without
any other information:

 Unaligned handler failed, ret = -2
 clock_adjtime (pid 22578): Unaligned data reference (code 28)
 CPU: 1 PID: 22578 Comm: clock_adjtime Tainted: G  E  4.5.0-2-parisc64-smp #1 Debian 4.5.4-1
 task: 000000007d9960f8 ti: 00000001bde7c000 task.ti: 00000001bde7c000

      YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
 PSW: 00001000000001001111100000001111 Tainted: G            E
 r00-03  000000ff0804f80f 00000001bde7c2b0 00000000402d2be8 00000001bde7c2b0
 r04-07  00000000409e1fd0 00000000fa6f7fff 00000001bde7c148 00000000fa6f7fff
 r08-11  0000000000000000 00000000ffffffff 00000000fac9bb7b 000000000002b4d4
 r12-15  000000000015241c 000000000015242c 000000000000002d 00000000fac9bb7b
 r16-19  0000000000028800 0000000000000001 0000000000000070 00000001bde7c218
 r20-23  0000000000000000 00000001bde7c210 0000000000000002 0000000000000000
 r24-27  0000000000000000 0000000000000000 00000001bde7c148 00000000409e1fd0
 r28-31  0000000000000001 00000001bde7c320 00000001bde7c350 00000001bde7c218
 sr00-03  0000000001200000 0000000001200000 0000000000000000 0000000001200000
 sr04-07  0000000000000000 0000000000000000 0000000000000000 0000000000000000

 IASQ: 0000000000000000 0000000000000000 IAOQ: 00000000402d2e84 00000000402d2e88
  IIR: 0ca0d089    ISR: 0000000001200000  IOR: 00000000fa6f7fff
  CPU:        1   CR30: 00000001bde7c000 CR31: ffffffffffffffff
  ORIG_R28: 00000002369fe628
  IAOQ[0]: compat_get_timex+0x2dc/0x3c0
  IAOQ[1]: compat_get_timex+0x2e0/0x3c0
  RP(r2): compat_get_timex+0x40/0x3c0
 Backtrace:
  [<00000000402d4608>] compat_SyS_clock_adjtime+0x40/0xc0
  [<0000000040205024>] syscall_exit+0x0/0x14

This means the userspace program clock_adjtime called the clock_adjtime()
syscall and then crashed inside the compat_get_timex() function.
Syscalls should never crash programs, but instead return EFAULT.

The IIR register contains the executed instruction, which disassebles
into "ldw 0(sr3,r5),r9".
This load-word instruction is part of __get_user() which tried to read the word
at %r5/IOR (0xfa6f7fff). This means the unaligned handler jumped in.  The
unaligned handler is able to emulate all ldw instructions, but it fails if it
fails to read the source e.g. because of page fault.

The following program reproduces the problem:

#define _GNU_SOURCE
#include <unistd.h>
#include <sys/syscall.h>
#include <sys/mman.h>

int main(void) {
        /* allocate 8k */
        char *ptr = mmap(NULL, 2*4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
        /* free second half (upper 4k) and make it invalid. */
        munmap(ptr+4096, 4096);
        /* syscall where first int is unaligned and clobbers into invalid memory region */
        /* syscall should return EFAULT */
        return syscall(__NR_clock_adjtime, 0, ptr+4095);
}

To fix this issue we simply need to check if the faulting instruction address
is in the exception fixup table when the unaligned handler failed. If it
is, call the fixup routine instead of crashing.

While looking at the unaligned handler I found another issue as well: The
target register should not be modified if the handler was unsuccessful.

Signed-off-by: Helge Deller <deller@gmx.de>
Cc: stable@vger.kernel.org
8 years agoparisc: Fix printk time during boot
Helge Deller [Fri, 3 Jun 2016 17:22:31 +0000 (19:22 +0200)]
parisc: Fix printk time during boot

Avoid showing invalid printk time stamps during boot.

Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Aaro Koskinen <aaro.koskinen@iki.fi>
8 years agoparisc: Fix backtrace on PA-RISC
Mikulas Patocka [Tue, 28 Jun 2011 22:48:19 +0000 (00:48 +0200)]
parisc: Fix backtrace on PA-RISC

This patch fixes backtrace on PA-RISC

There were several problems:

1) The code that decodes instructions handles instructions that subtract
from the stack pointer incorrectly. If the instruction subtracts the
number X from the stack pointer the code increases the frame size by
(0x100000000-X).  This results in invalid accesses to memory and
recursive page faults.

2) Because gcc reorders blocks, handling instructions that subtract from
the frame pointer is incorrect. For example, this function
int f(int a)
{
if (__builtin_expect(a, 1))
return a;
g();
return a;
}
is compiled in such a way, that the code that decreases the stack
pointer for the first "return a" is placed before the code for "g" call.
If we recognize this decrement, we mistakenly believe that the frame
size for the "g" call is zero.

To fix problems 1) and 2), the patch doesn't recognize instructions that
decrease the stack pointer at all. To further safeguard the unwind code
against nonsense values, we don't allow frame size larger than
Total_frame_size.

3) The backtrace is not locked. If stack dump races with module unload,
invalid table can be accessed.

This patch adds a spinlock when processing module tables.

Note, that for correct backtrace, you need recent binutils.
Binutils 2.18 from Debian 5 produce garbage unwind tables.
Binutils 2.21 work better (it sometimes forgets function frames, but at
least it doesn't generate garbage).

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Helge Deller <deller@gmx.de>
8 years agoMerge tag 'drm-fixes-for-v4.7-rc2' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Sat, 4 Jun 2016 19:30:36 +0000 (12:30 -0700)]
Merge tag 'drm-fixes-for-v4.7-rc2' of git://people.freedesktop.org/~airlied/linux

Pull drm fixes from Dave Airlie:
 "A bunch of ARM drivers got into the fixes vibe this time around, so
  this contains a bunch of fixes for imx, atmel hlcdc, arm hdlcd (only
  so many combos of hlcd), mediatek and omap drm.

  Other than that there is one mgag200 fix and a few core drm regression
  fixes"

* tag 'drm-fixes-for-v4.7-rc2' of git://people.freedesktop.org/~airlied/linux: (34 commits)
  drm/omap: fix unused variable warning.
  drm: hdlcd: Add information about the underlying framebuffers in debugfs
  drm: hdlcd: Cleanup the atomic plane operations
  drm/hdlcd: Fix up crtc_state->event handling
  drm: hdlcd: Revamp runtime power management
  drm/mediatek: mtk_dsi: Remove spurious drm_connector_unregister
  drm/mediatek: mtk_dpi: remove invalid error message
  drm: atmel-hlcdc: fix a NULL check
  drm: atmel-hlcdc: fix atmel_hlcdc_crtc_reset() implementation
  drm/mgag200: Black screen fix for G200e rev 4
  drm: Wrap direct calls to driver->gem_free_object from CMA
  drm: fix fb refcount issue with atomic modesetting
  drm: make drm_atomic_set_mode_prop_for_crtc() more reliable
  drm/sti: remove extra mode fixup
  drm: add missing drm_mode_set_crtcinfo call
  drm/omap: include gpio/consumer.h where needed
  drm/omap: include linux/seq_file.h where needed
  Revert "drm/omap: no need to select OMAP2_DSS"
  drm/omap: Remove regulator API abuse
  OMAPDSS: HDMI5: Change DDC timings
  ...

8 years agoMerge tag 'vfio-v4.7-rc2' of git://github.com/awilliam/linux-vfio
Linus Torvalds [Sat, 4 Jun 2016 19:25:36 +0000 (12:25 -0700)]
Merge tag 'vfio-v4.7-rc2' of git://github.com/awilliam/linux-vfio

Pull VFIO fixes from Alex Williamson:
 "Fix irqfd shutdown ordering, build warning, and VPD short read"

* tag 'vfio-v4.7-rc2' of git://github.com/awilliam/linux-vfio:
  vfio/pci: Allow VPD short read
  vfio/type1: Fix build warning
  vfio/pci: Fix ordering of eventfd vs virqfd shutdown

8 years agoMerge tag 'mmc-v4.7-rc1-2' of git://git.linaro.org/people/ulf.hansson/mmc
Linus Torvalds [Sat, 4 Jun 2016 19:20:26 +0000 (12:20 -0700)]
Merge tag 'mmc-v4.7-rc1-2' of git://git.linaro.org/people/ulf.hansson/mmc

Pull MMC fixes from Ulf Hansson:
 "MMC core:
   - Fix/restore behaviour when selecting bus width for (e)MMC

  MMC host:
   - sunxi: Fix eMMC HS-DDR modes on Allwinner A80"

* tag 'mmc-v4.7-rc1-2' of git://git.linaro.org/people/ulf.hansson/mmc:
  mmc: sunxi: Re-enable eMMC HS-DDR modes on Allwinner A80
  mmc: sunxi: Fix DDR MMC timings for A80
  mmc: fix mmc mode selection for HS-DDR and higher

8 years agoMerge branch 'for-linus-4.7' of git://git.kernel.org/pub/scm/linux/kernel/git/mason...
Linus Torvalds [Sat, 4 Jun 2016 18:56:28 +0000 (11:56 -0700)]
Merge branch 'for-linus-4.7' of git://git./linux/kernel/git/mason/linux-btrfs

Pull btrfs fixes from Chris Mason:
 "The important part of this pull is Filipe's set of fixes for btrfs
  device replacement.  Filipe fixed a few issues seen on the list and a
  number he found on his own"

* 'for-linus-4.7' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  Btrfs: deal with duplciates during extent_map insertion in btrfs_get_extent
  Btrfs: fix race between device replace and read repair
  Btrfs: fix race between device replace and discard
  Btrfs: fix race between device replace and chunk allocation
  Btrfs: fix race setting block group back to RW mode during device replace
  Btrfs: fix unprotected assignment of the left cursor for device replace
  Btrfs: fix race setting block group readonly during device replace
  Btrfs: fix race between device replace and block group removal
  Btrfs: fix race between readahead and device replace/removal

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph...
Linus Torvalds [Sat, 4 Jun 2016 18:37:53 +0000 (11:37 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/sage/ceph-client

Pull Ceph fixes from Sage Weil:
 "We have a few follow-up fixes for the libceph refactor from Ilya, and
  then some cephfs + fscache fixes from Zheng.

  The first two FS-Cache patches are acked by David Howells and deemed
  trivial enough to go through our tree.  The rest fix some issues with
  the ceph fscache handling (disable cache for inodes opened for write,
  and simplify the revalidation logic accordingly, dropping the
  now-unnecessary work queue)"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  ceph: use i_version to check validity of fscache
  ceph: improve fscache revalidation
  ceph: disable fscache when inode is opened for write
  ceph: avoid unnecessary fscache invalidation/revlidation
  ceph: call __fscache_uncache_page() if readpages fails
  FS-Cache: make check_consistency callback return int
  FS-Cache: wake write waiter after invalidating writes
  libceph: use %s instead of %pE in dout()s
  libceph: put request only if it's done in handle_reply()
  libceph: change ceph_osdmap_flag() to take osdc

8 years agoMerge tag 'acpi-4.7-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Sat, 4 Jun 2016 18:26:49 +0000 (11:26 -0700)]
Merge tag 'acpi-4.7-rc2' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI fixes from Rafael Wysocki:
 "Two fixes for problems introduced recently (ACPICA and the ACPI
  backlight driver) and one fix for an older issue that prevents at
  least one system from booting.

  Specifics:

   - Fix an incorrect check introduced by recent ACPICA changes which
     causes problems with booting KVM guests to happen, among other
     things (Lv Zheng).

   - Fix a backlight issue introduced by recent changes to the ACPI
     video driver (Aaron Lu).

   - Fix the ACPI processor initialization which attempts to register an
     IO region without checking if that really is necessary and
     sometimes prevents drivers loaded subsequently from registering
     their resources which leads to boot issues (Rafael Wysocki)"

* tag 'acpi-4.7-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  ACPI / processor: Avoid reserving IO regions too early
  ACPICA / Hardware: Fix old register check in acpi_hw_get_access_bit_width()
  ACPI / Thermal / video: fix max_level incorrect value

8 years agoMerge tag 'pm-4.7-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Linus Torvalds [Sat, 4 Jun 2016 18:07:57 +0000 (11:07 -0700)]
Merge tag 'pm-4.7-rc2' of git://git./linux/kernel/git/rafael/linux-pm

Pull power management fixes from Rafael Wysocki:
 "Two fixes for problems introduced recently in the cpufreq core and the
  intel_pstate driver.

  Specifics:

   - Fix a silly mistake related to the clamp_val() usage in a function
     added by a recent commit (Rafael Wysocki).

   - Reduce the log level of an annoying message added to intel_pstate
     during the recent merge window (Srinivas Pandruvada)"

* tag 'pm-4.7-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpufreq: Fix clamp_val() usage in cpufreq_driver_fast_switch()
  cpufreq: intel_pstate: Downgrade print level for _PPC

8 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Sat, 4 Jun 2016 17:51:29 +0000 (10:51 -0700)]
Merge branch 'akpm' (patches from Andrew)

Merge various fixes from Andrew Morton:
 "10 fixes"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  mm, page_alloc: recalculate the preferred zoneref if the context can ignore memory policies
  mm, page_alloc: reset zonelist iterator after resetting fair zone allocation policy
  mm, oom_reaper: do not use siglock in try_oom_reaper()
  mm, page_alloc: prevent infinite loop in buffered_rmqueue()
  checkpatch: reduce git commit description style false positives
  mm/z3fold.c: avoid modifying HEADLESS page and minor cleanup
  memcg: add RCU locking around css_for_each_descendant_pre() in memcg_offline_kmem()
  mm: check the return value of lookup_page_ext for all call sites
  kdump: fix dmesg gdbmacro to work with record based printk
  mm: fix overflow in vm_map_ram()

8 years agoMerge branch 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 3 Jun 2016 23:12:35 +0000 (16:12 -0700)]
Merge branch 'irq-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull irq fixes from Thomas Gleixner:
 - a few simple fixes for fallout from the recent gic-v3 changes
 - a workaround for a Cavium thunderX erratum
 - a bugfix for the pic32 irqchip to make external interrupts work proper
 - a missing return value in the generic IPI management code

* 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/irq-pic32-evic: Fix bug with external interrupts.
  irqchip/gicv3-its: numa: Enable workaround for Cavium thunderx erratum 23144
  irqchip/gic-v3: Fix quiescence check in gic_enable_redist
  irqchip/gic-v3: Fix copy+paste mistakes in defines
  irqchip/gic-v3: Fix ICC_SGI1R_EL1.INTID decoding mask
  genirq: Fix missing return value in irq_destroy_ipi()

8 years agomm, page_alloc: recalculate the preferred zoneref if the context can ignore memory...
Mel Gorman [Fri, 3 Jun 2016 21:56:01 +0000 (14:56 -0700)]
mm, page_alloc: recalculate the preferred zoneref if the context can ignore memory policies

The optimistic fast path may use cpuset_current_mems_allowed instead of
of a NULL nodemask supplied by the caller for cpuset allocations.  The
preferred zone is calculated on this basis for statistic purposes and as
a starting point in the zonelist iterator.

However, if the context can ignore memory policies due to being atomic
or being able to ignore watermarks then the starting point in the
zonelist iterator is no longer correct.  This patch resets the zonelist
iterator in the allocator slowpath if the context can ignore memory
policies.  This will alter the zone used for statistics but only after
it is known that it makes sense for that context.  Resetting it before
entering the slowpath would potentially allow an ALLOC_CPUSET allocation
to be accounted for against the wrong zone.  Note that while nodemask is
not explicitly set to the original nodemask, it would only have been
overwritten if cpuset_enabled() and it was reset before the slowpath was
entered.

Link: http://lkml.kernel.org/r/20160602103936.GU2527@techsingularity.net
Fixes: c33d6c06f60f710 ("mm, page_alloc: avoid looking up the first zone in a zonelist twice")
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Tested-by: Geert Uytterhoeven <geert@linux-m68k.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm, page_alloc: reset zonelist iterator after resetting fair zone allocation policy
Mel Gorman [Fri, 3 Jun 2016 21:55:58 +0000 (14:55 -0700)]
mm, page_alloc: reset zonelist iterator after resetting fair zone allocation policy

Geert Uytterhoeven reported the following problem that bisected to
commit c33d6c06f60f ("mm, page_alloc: avoid looking up the first zone
in a zonelist twice") on m68k/ARAnyM

    BUG: scheduling while atomic: cron/668/0x10c9a0c0
    Modules linked in:
    CPU: 0 PID: 668 Comm: cron Not tainted 4.6.0-atari-05133-gc33d6c06f60f710f #364
    Call Trace: [<0003d7d0>] __schedule_bug+0x40/0x54
      __schedule+0x312/0x388
      __schedule+0x0/0x388
      prepare_to_wait+0x0/0x52
      schedule+0x64/0x82
      schedule_timeout+0xda/0x104
      set_next_entity+0x18/0x40
      pick_next_task_fair+0x78/0xda
      io_schedule_timeout+0x36/0x4a
      bit_wait_io+0x0/0x40
      bit_wait_io+0x12/0x40
      __wait_on_bit+0x46/0x76
      wait_on_page_bit_killable+0x64/0x6c
      bit_wait_io+0x0/0x40
      wake_bit_function+0x0/0x4e
      __lock_page_or_retry+0xde/0x124
      do_scan_async+0x114/0x17c
      lookup_swap_cache+0x24/0x4e
      handle_mm_fault+0x626/0x7de
      find_vma+0x0/0x66
      down_read+0x0/0xe
      wait_on_page_bit_killable_timeout+0x77/0x7c
      find_vma+0x16/0x66
      do_page_fault+0xe6/0x23a
      res_func+0xa3c/0x141a
      buserr_c+0x190/0x6d4
      res_func+0xa3c/0x141a
      buserr+0x20/0x28
      res_func+0xa3c/0x141a
      buserr+0x20/0x28

The relationship is not obvious but it's due to a failure to rescan the
full zonelist after the fair zone allocation policy exhausts the batch
count.  While this is a functional problem, it's also a performance
issue.  A page allocator microbenchmark showed the following

                                   4.7.0-rc1                  4.7.0-rc1
                                     vanilla                 reset-v1r2
  Min      alloc-odr0-1     327.00 (  0.00%)           326.00 (  0.31%)
  Min      alloc-odr0-2     235.00 (  0.00%)           235.00 (  0.00%)
  Min      alloc-odr0-4     198.00 (  0.00%)           198.00 (  0.00%)
  Min      alloc-odr0-8     170.00 (  0.00%)           170.00 (  0.00%)
  Min      alloc-odr0-16    156.00 (  0.00%)           156.00 (  0.00%)
  Min      alloc-odr0-32    150.00 (  0.00%)           150.00 (  0.00%)
  Min      alloc-odr0-64    146.00 (  0.00%)           146.00 (  0.00%)
  Min      alloc-odr0-128   145.00 (  0.00%)           145.00 (  0.00%)
  Min      alloc-odr0-256   155.00 (  0.00%)           155.00 (  0.00%)
  Min      alloc-odr0-512   168.00 (  0.00%)           165.00 (  1.79%)
  Min      alloc-odr0-1024  175.00 (  0.00%)           174.00 (  0.57%)
  Min      alloc-odr0-2048  180.00 (  0.00%)           180.00 (  0.00%)
  Min      alloc-odr0-4096  187.00 (  0.00%)           186.00 (  0.53%)
  Min      alloc-odr0-8192  190.00 (  0.00%)           190.00 (  0.00%)
  Min      alloc-odr0-16384 191.00 (  0.00%)           191.00 (  0.00%)
  Min      alloc-odr1-1     736.00 (  0.00%)           445.00 ( 39.54%)
  Min      alloc-odr1-2     343.00 (  0.00%)           335.00 (  2.33%)
  Min      alloc-odr1-4     277.00 (  0.00%)           270.00 (  2.53%)
  Min      alloc-odr1-8     238.00 (  0.00%)           233.00 (  2.10%)
  Min      alloc-odr1-16    224.00 (  0.00%)           218.00 (  2.68%)
  Min      alloc-odr1-32    210.00 (  0.00%)           208.00 (  0.95%)
  Min      alloc-odr1-64    207.00 (  0.00%)           203.00 (  1.93%)
  Min      alloc-odr1-128   276.00 (  0.00%)           202.00 ( 26.81%)
  Min      alloc-odr1-256   206.00 (  0.00%)           202.00 (  1.94%)
  Min      alloc-odr1-512   207.00 (  0.00%)           202.00 (  2.42%)
  Min      alloc-odr1-1024  208.00 (  0.00%)           205.00 (  1.44%)
  Min      alloc-odr1-2048  213.00 (  0.00%)           212.00 (  0.47%)
  Min      alloc-odr1-4096  218.00 (  0.00%)           216.00 (  0.92%)
  Min      alloc-odr1-8192  341.00 (  0.00%)           219.00 ( 35.78%)

Note that order-0 allocations are unaffected but higher orders get a
small boost from this patch and a large reduction in system CPU usage
overall as can be seen here:

             4.7.0-rc1   4.7.0-rc1
               vanilla  reset-v1r2
  User           85.32       86.31
  System       2221.39     2053.36
  Elapsed      2368.89     2202.47

Fixes: c33d6c06f60f ("mm, page_alloc: avoid looking up the first zone in a zonelist twice")
Link: http://lkml.kernel.org/r/20160531100848.GR2527@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Tested-by: Geert Uytterhoeven <geert@linux-m68k.org>
Tested-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm, oom_reaper: do not use siglock in try_oom_reaper()
Michal Hocko [Fri, 3 Jun 2016 21:55:55 +0000 (14:55 -0700)]
mm, oom_reaper: do not use siglock in try_oom_reaper()

Oleg has noted that siglock usage in try_oom_reaper is both pointless
and dangerous.  signal_group_exit can be checked lockless.  The problem
is that sighand becomes NULL in __exit_signal so we can crash.

Fixes: 3ef22dfff239 ("oom, oom_reaper: try to reap tasks which skip regular OOM killer path")
Link: http://lkml.kernel.org/r/1464679423-30218-1-git-send-email-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Suggested-by: Oleg Nesterov <oleg@redhat.com>
Cc: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm, page_alloc: prevent infinite loop in buffered_rmqueue()
Vlastimil Babka [Fri, 3 Jun 2016 21:55:52 +0000 (14:55 -0700)]
mm, page_alloc: prevent infinite loop in buffered_rmqueue()

In DEBUG_VM kernel, we can hit infinite loop for order == 0 in
buffered_rmqueue() when check_new_pcp() returns 1, because the bad page
is never removed from the pcp list.  Fix this by removing the page
before retrying.  Also we don't need to check if page is non-NULL,
because we simply grab it from the list which was just tested for being
non-empty.

Fixes: 479f854a207c ("mm, page_alloc: defer debugging checks of pages allocated from the PCP")
Link: http://lkml.kernel.org/r/20160530090154.GM2527@techsingularity.net
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reported-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agocheckpatch: reduce git commit description style false positives
Joe Perches [Fri, 3 Jun 2016 21:55:49 +0000 (14:55 -0700)]
checkpatch: reduce git commit description style false positives

Some lines in a commit log appear to be commit SHA1 ids like:

  ERROR: Please use git commit description style 'commit <12+ chars of sha1> ("<title line>")' - ie: 'commit 0123456789ab ("commit description")'
Link: http://lkml.kernel.org/r/40e03fd7aaf1f55c75d787128d6d17c5a71226c2.1464358556.git.vdavydov@virtuozzo.com
Reduce the false positives.

Link: http://lkml.kernel.org/r/eda977eaa8328fef42bb3c87935d97e10ea8ff67.1464384023.git.joe@perches.com
Signed-off-by: Joe Perches <joe@perches.com>
Reported-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm/z3fold.c: avoid modifying HEADLESS page and minor cleanup
Vitaly Wool [Fri, 3 Jun 2016 21:55:47 +0000 (14:55 -0700)]
mm/z3fold.c: avoid modifying HEADLESS page and minor cleanup

Fix erroneous z3fold header access in a HEADLESS page in reclaim
function, and change one remaining direct handle-to-buddy conversion to
use the appropriate helper.

Link: http://lkml.kernel.org/r/5748706F.9020208@gmail.com
Signed-off-by: Vitaly Wool <vitalywool@gmail.com>
Reviewed-by: Dan Streetman <ddstreet@ieee.org>
Cc: Seth Jennings <sjenning@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomemcg: add RCU locking around css_for_each_descendant_pre() in memcg_offline_kmem()
Tejun Heo [Fri, 3 Jun 2016 21:55:44 +0000 (14:55 -0700)]
memcg: add RCU locking around css_for_each_descendant_pre() in memcg_offline_kmem()

memcg_offline_kmem() may be called from memcg_free_kmem() after a css
init failure.  memcg_free_kmem() is a ->css_free callback which is
called without cgroup_mutex and memcg_offline_kmem() ends up using
css_for_each_descendant_pre() without any locking.  Fix it by adding rcu
read locking around it.

    mkdir: cannot create directory `65530': No space left on device
    ===============================
    [ INFO: suspicious RCU usage. ]
    4.6.0-work+ #321 Not tainted
    -------------------------------
    kernel/cgroup.c:4008 cgroup_mutex or RCU read lock required!
     [  527.243970] other info that might help us debug this:
     [  527.244715]
    rcu_scheduler_active = 1, debug_locks = 0
    2 locks held by kworker/0:5/1664:
     #0:  ("cgroup_destroy"){.+.+..}, at: [<ffffffff81060ab5>] process_one_work+0x165/0x4a0
     #1:  ((&css->destroy_work)#3){+.+...}, at: [<ffffffff81060ab5>] process_one_work+0x165/0x4a0
     [  527.248098] stack backtrace:
    CPU: 0 PID: 1664 Comm: kworker/0:5 Not tainted 4.6.0-work+ #321
    Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.9.1-1.fc24 04/01/2014
    Workqueue: cgroup_destroy css_free_work_fn
    Call Trace:
      dump_stack+0x68/0xa1
      lockdep_rcu_suspicious+0xd7/0x110
      css_next_descendant_pre+0x7d/0xb0
      memcg_offline_kmem.part.44+0x4a/0xc0
      mem_cgroup_css_free+0x1ec/0x200
      css_free_work_fn+0x49/0x5e0
      process_one_work+0x1c5/0x4a0
      worker_thread+0x49/0x490
      kthread+0xea/0x100
      ret_from_fork+0x1f/0x40

Link: http://lkml.kernel.org/r/20160526203018.GG23194@mtj.duckdns.org
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Vladimir Davydov <vdavydov@virtuozzo.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: <stable@vger.kernel.org> [4.5+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoMerge branch 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 3 Jun 2016 22:37:27 +0000 (15:37 -0700)]
Merge branch 'timers-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull timer bugfix from Thomas Gleixner:
 "A single bugfix for the error check wreckage we introduced in the
  merge window"

* 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  time: Make settimeofday error checking work again

8 years agomm: check the return value of lookup_page_ext for all call sites
Yang Shi [Fri, 3 Jun 2016 21:55:38 +0000 (14:55 -0700)]
mm: check the return value of lookup_page_ext for all call sites

Per the discussion with Joonsoo Kim [1], we need check the return value
of lookup_page_ext() for all call sites since it might return NULL in
some cases, although it is unlikely, i.e.  memory hotplug.

Tested with ltp with "page_owner=0".

[1] http://lkml.kernel.org/r/20160519002809.GA10245@js1304-P5Q-DELUXE

[akpm@linux-foundation.org: fix build-breaking typos]
[arnd@arndb.de: fix build problems from lookup_page_ext]
Link: http://lkml.kernel.org/r/6285269.2CksypHdYp@wuerfel
[akpm@linux-foundation.org: coding-style fixes]
Link: http://lkml.kernel.org/r/1464023768-31025-1-git-send-email-yang.shi@linaro.org
Signed-off-by: Yang Shi <yang.shi@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agokdump: fix dmesg gdbmacro to work with record based printk
Corey Minyard [Fri, 3 Jun 2016 21:55:36 +0000 (14:55 -0700)]
kdump: fix dmesg gdbmacro to work with record based printk

Commit 7ff9554bb578 ("printk: convert byte-buffer to variable-length
record buffer") introduced a record based printk buffer.  Modify
gdbmacros.txt to parse this new structure so dmesg will work properly.

Link: http://lkml.kernel.org/r/1463515794-1599-1-git-send-email-minyard@acm.org
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm: fix overflow in vm_map_ram()
Guillermo Julián Moreno [Fri, 3 Jun 2016 21:55:33 +0000 (14:55 -0700)]
mm: fix overflow in vm_map_ram()

When remapping pages accounting for 4G or more memory space, the
operation 'count << PAGE_SHIFT' overflows as it is performed on an
integer.  Solution: cast before doing the bitshift.

[akpm@linux-foundation.org: fix vm_unmap_ram() also]
[akpm@linux-foundation.org: fix vmap() as well, per Guillermo]
Link: http://lkml.kernel.org/r/etPan.57175fb3.7a271c6b.2bd@naudit.es
Signed-off-by: Guillermo Julián Moreno <guillermo.julian@naudit.es>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoMerge branch 'fixes' of git://git.armlinux.org.uk/~rmk/linux-arm
Linus Torvalds [Fri, 3 Jun 2016 21:39:29 +0000 (14:39 -0700)]
Merge branch 'fixes' of git://git.armlinux.org.uk/~rmk/linux-arm

Pull ARM fix from Russell King:
 "Just one fix to the ptrace code, spotted by Simon Marchi, where if a
  thread migrates to a different CPU and the VFP registers are changed
  through ptrace, the application doesn't see the updated VFP registers"

* 'fixes' of git://git.armlinux.org.uk/~rmk/linux-arm:
  ARM: fix PTRACE_SETVFPREGS on SMP systems

8 years agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Fri, 3 Jun 2016 21:29:47 +0000 (14:29 -0700)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fixes from Will Deacon:
 "The main thing here is reviving hugetlb support using contiguous ptes,
  which we ended up reverting at the last minute in 4.5 pending a fix
  which went into the core mm/ code during the recent merge window.

   - Revert a previous revert and get hugetlb going with contiguous hints
   - Wire up missing compat syscalls
   - Enable CONFIG_SET_MODULE_RONX by default
   - Add missing line to our compat /proc/cpuinfo output
   - Clarify levels in our page table dumps
   - Fix booting with RANDOMIZE_TEXT_OFFSET enabled
   - Misc fixes to the ARM CPU PMU driver (refcounting, probe failure)
   - Remove some dead code and update a comment"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: fix alignment when RANDOMIZE_TEXT_OFFSET is enabled
  arm64: move {PAGE,CONT}_SHIFT into Kconfig
  arm64: mm: dump: log span level
  arm64: update stale PAGE_OFFSET comment
  drivers/perf: arm_pmu: Avoid leaking pmu->irq_affinity on error
  drivers/perf: arm_pmu: Defer the setting of __oprofile_cpu_pmu
  drivers/perf: arm_pmu: Fix reference count of a device_node in of_pmu_irq_cfg
  arm64: report CPU number in bad_mode
  arm64: unistd32.h: wire up missing syscalls for compat tasks
  arm64: Provide "model name" in /proc/cpuinfo for PER_LINUX32 tasks
  arm64: enable CONFIG_SET_MODULE_RONX by default
  arm64: Remove orphaned __addr_ok() definition
  Revert "arm64: hugetlb: partial revert of 66b3923a1a0f"

8 years agoMerge tag 'powerpc-4.7-2' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc...
Linus Torvalds [Fri, 3 Jun 2016 21:20:22 +0000 (14:20 -0700)]
Merge tag 'powerpc-4.7-2' of git://git./linux/kernel/git/powerpc/linux

Pull powerpc fixes from Michael Ellerman:
 - Handle RTAS delay requests in configure_bridge from Russell Currey
 - Refactor the configure_bridge RTAS tokens from Russell Currey
 - Fix definition of SIAR and SDAR registers from Thomas Huth
 - Use privileged SPR number for MMCR2 from Thomas Huth
 - Update LPCR only if it is powernv from Aneesh Kumar K.V
 - Fix the reference bit update when handling hash fault from Aneesh
   Kumar K.V
 - Add missing tlb flush from Aneesh Kumar K.V
 - Add POWER8NVL support to ibm,client-architecture-support call from
   Thomas Huth

* tag 'powerpc-4.7-2' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/pseries: Add POWER8NVL support to ibm,client-architecture-support call
  powerpc/mm/radix: Add missing tlb flush
  powerpc/mm/hash: Fix the reference bit update when handling hash fault
  powerpc/mm/radix: Update LPCR only if it is powernv
  powerpc: Use privileged SPR number for MMCR2
  powerpc: Fix definition of SIAR and SDAR registers
  powerpc/pseries/eeh: Refactor the configure_bridge RTAS tokens
  powerpc/pseries/eeh: Handle RTAS delay requests in configure_bridge

8 years agoMerge branches 'acpica-fixes', 'acpi-video' and 'acpi-processor'
Rafael J. Wysocki [Fri, 3 Jun 2016 20:35:05 +0000 (22:35 +0200)]
Merge branches 'acpica-fixes', 'acpi-video' and 'acpi-processor'

* acpica-fixes:
  ACPICA / Hardware: Fix old register check in acpi_hw_get_access_bit_width()

* acpi-video:
  ACPI / Thermal / video: fix max_level incorrect value

* acpi-processor:
  ACPI / processor: Avoid reserving IO regions too early

8 years agoMerge branch 'pm-cpufreq-fixes'
Rafael J. Wysocki [Fri, 3 Jun 2016 20:34:18 +0000 (22:34 +0200)]
Merge branch 'pm-cpufreq-fixes'

* pm-cpufreq-fixes:
  cpufreq: Fix clamp_val() usage in cpufreq_driver_fast_switch()
  cpufreq: intel_pstate: Downgrade print level for _PPC