firefly-linux-kernel-4.4.55.git
9 years agosched: Start stopper early
Peter Zijlstra [Fri, 9 Oct 2015 16:00:54 +0000 (18:00 +0200)]
sched: Start stopper early

Ensure the stopper thread is active 'early', because the load balancer
pretty much assumes that its available. And when 'online && active' the
load-balancer is fully available.

Not only the numa balancing stop_two_cpus() caller relies on it, but
also the self migration stuff does, and at CPU_ONLINE time the cpu
really is 'free' to run anything.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: heiko.carstens@de.ibm.com
Link: http://lkml.kernel.org/r/20151009160054.GA10176@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agostop_machine: Kill cpu_stop_threads->setup() and cpu_stop_unpark()
Oleg Nesterov [Fri, 9 Oct 2015 16:00:51 +0000 (18:00 +0200)]
stop_machine: Kill cpu_stop_threads->setup() and cpu_stop_unpark()

Now that we always use stop_machine_unpark() to wake the stopper
threas up, we can kill ->setup() and fold cpu_stop_unpark() into
stop_machine_unpark().

And we do not need stopper->lock to set stopper->enabled = true.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: heiko.carstens@de.ibm.com
Link: http://lkml.kernel.org/r/20151009160051.GA10169@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agostop_machine: Kill smp_hotplug_thread->pre_unpark, introduce stop_machine_unpark()
Oleg Nesterov [Fri, 9 Oct 2015 16:00:49 +0000 (18:00 +0200)]
stop_machine: Kill smp_hotplug_thread->pre_unpark, introduce stop_machine_unpark()

1. Change smpboot_unpark_thread() to check ->selfparking, just
   like smpboot_park_thread() does.

2. Introduce stop_machine_unpark() which sets ->enabled and calls
   kthread_unpark().

3. Change smpboot_thread_call() and cpu_stop_init() to call
   stop_machine_unpark() by hand.

This way:

    - IMO the ->selfparking logic becomes more consistent.

    - We can kill the smp_hotplug_thread->pre_unpark() method.

    - We can easily unpark the stopper thread earlier. Say, we
      can move stop_machine_unpark() from smpboot_thread_call()
      to sched_cpu_active() as Peter suggests.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: heiko.carstens@de.ibm.com
Link: http://lkml.kernel.org/r/20151009160049.GA10166@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agostop_machine: Change cpu_stop_queue_two_works() to rely on stopper->enabled
Oleg Nesterov [Thu, 8 Oct 2015 17:01:41 +0000 (19:01 +0200)]
stop_machine: Change cpu_stop_queue_two_works() to rely on stopper->enabled

Change cpu_stop_queue_two_works() to ensure that both CPU's have
stopper->enabled == T or fail otherwise.

This way stop_two_cpus() no longer needs to check cpu_active() to
avoid the deadlock. This patch doesn't remove these checks, we will
do this later.

Note: we need to take both stopper->lock's at the same time, but this
will also help to remove lglock from stop_machine.c, so I hope this
is fine.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: heiko.carstens@de.ibm.com
Link: http://lkml.kernel.org/r/20151008170141.GA25537@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agostop_machine: Introduce __cpu_stop_queue_work() and cpu_stop_queue_two_works()
Oleg Nesterov [Thu, 8 Oct 2015 14:51:34 +0000 (16:51 +0200)]
stop_machine: Introduce __cpu_stop_queue_work() and cpu_stop_queue_two_works()

Preparation to simplify the review of the next change. Add two simple
helpers, __cpu_stop_queue_work() and cpu_stop_queue_two_works() which
simply take a bit of code from their callers.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: heiko.carstens@de.ibm.com
Link: http://lkml.kernel.org/r/20151008145134.GA18146@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agostop_machine: Ensure that a queued callback will be called before cpu_stop_park()
Oleg Nesterov [Thu, 8 Oct 2015 14:51:31 +0000 (16:51 +0200)]
stop_machine: Ensure that a queued callback will be called before cpu_stop_park()

cpu_stop_queue_work() checks stopper->enabled before it queues the
work, but ->enabled == T can only guarantee cpu_stop_signal_done()
if we race with cpu_down().

This is not enough for stop_two_cpus() or stop_machine(), they will
deadlock if multi_cpu_stop() won't be called by one of the target
CPU's. stop_machine/stop_cpus are fine, they rely on stop_cpus_mutex.
But stop_two_cpus() has to check cpu_active() to avoid the same race
with hotplug, and this check is very unobvious and probably not even
correct if we race with cpu_up().

Change cpu_down() pass to clear ->enabled before cpu_stopper_thread()
flushes the pending ->works and returns with KTHREAD_SHOULD_PARK set.

Note also that smpboot_thread_call() calls cpu_stop_unpark() which
sets enabled == T at CPU_ONLINE stage, so this CPU can't go away until
cpu_stopper_thread() is called at least once. This all means that if
cpu_stop_queue_work() succeeds, we know that work->fn() will be called.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: heiko.carstens@de.ibm.com
Link: http://lkml.kernel.org/r/20151008145131.GA18139@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agoMerge branch 'sched/urgent' into sched/core, to pick up fixes and resolve conflicts
Ingo Molnar [Tue, 20 Oct 2015 08:18:16 +0000 (10:18 +0200)]
Merge branch 'sched/urgent' into sched/core, to pick up fixes and resolve conflicts

Conflicts:
kernel/sched/fair.c

Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/deadline: Fix migration of SCHED_DEADLINE tasks
Luca Abeni [Fri, 16 Oct 2015 08:06:21 +0000 (10:06 +0200)]
sched/deadline: Fix migration of SCHED_DEADLINE tasks

Commit:

  9d5142624256 ("sched/deadline: Reduce rq lock contention by eliminating locking of non-feasible target")

broke select_task_rq_dl() and find_lock_later_rq(), because it introduced
a comparison between the local task's deadline and dl.earliest_dl.curr of
the remote queue.

However, if the remote runqueue does not contain any SCHED_DEADLINE
task its earliest_dl.curr is 0 (always smaller than the deadline of
the local task) and the remote runqueue is not selected for pushing.

As a result, if an application creates multiple SCHED_DEADLINE
threads, they will never be pushed to runqueues that do not already
contain SCHED_DEADLINE tasks.

This patch fixes the issue by checking if dl.dl_nr_running == 0.

Signed-off-by: Luca Abeni <luca.abeni@unitn.it>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Juri Lelli <juri.lelli@arm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Wanpeng Li <wanpeng.li@linux.intel.com>
Fixes: 9d5142624256 ("sched/deadline: Reduce rq lock contention by eliminating locking of non-feasible target")
Link: http://lkml.kernel.org/r/1444982781-15608-1-git-send-email-luca.abeni@unitn.it
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agonohz: Revert "nohz: Set isolcpus when nohz_full is set"
Frederic Weisbecker [Mon, 12 Oct 2015 15:21:23 +0000 (17:21 +0200)]
nohz: Revert "nohz: Set isolcpus when nohz_full is set"

This reverts:

  8cb9764fc88b ("nohz: Set isolcpus when nohz_full is set")

We assumed that full-nohz users always want scheduler isolation on full
dynticks CPUs, therefore we included full-nohz CPUs on cpu_isolated_map.

This means that tasks run by default on CPUs outside the nohz_full range
unless their affinity is explicity overwritten.

This suits pure isolation workloads but when the machine is needed to
run common workloads, the available sets of CPUs to run common tasks
becomes reduced.

We reach an extreme case when CONFIG_NO_HZ_FULL_ALL is enabled as it
leaves only CPU 0 for non-isolation tasks, which makes people think that
their supercomputer regressed to 90's UP - which is true in a sense.

Some full-nohz users appear to be interested in running normal workloads
either before or after an isolation workload. Full-nohz isn't optimized
toward normal workloads but it's still better than UP performance.

We are reaching a limitation in kernel presets here. Lets revert this
cpu_isolated_map inclusion and let userspace do its own scheduler
isolation using cpusets or explicit affinity settings.

Reported-by: Ingo Molnar <mingo@kernel.org>
Reported-by: Mike Galbraith <umgwanakikbuti@gmail.com>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Chris Metcalf <cmetcalf@ezchip.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Paul E . McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@redhat.com>
Link: http://lkml.kernel.org/r/1444663283-30068-1-git-send-email-fweisbec@gmail.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/fair: Update task group's load_avg after task migration
Yuyang Du [Tue, 13 Oct 2015 01:18:23 +0000 (09:18 +0800)]
sched/fair: Update task group's load_avg after task migration

When cfs_rq has cfs_rq->removed_load_avg set (when a task migrates from
this cfs_rq), we need to update its contribution to the group's load_avg.

This should not increase tg's update too much, because in most cases, the
cfs_rq has already decayed its load_avg.

Tested-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Signed-off-by: Yuyang Du <yuyang.du@intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1444699103-20272-2-git-send-email-yuyang.du@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/fair: Fix overly small weight for interactive group entities
Yuyang Du [Tue, 13 Oct 2015 01:18:22 +0000 (09:18 +0800)]
sched/fair: Fix overly small weight for interactive group entities

Commit:

  9d89c257dfb9 ("sched/fair: Rewrite runnable load and utilization average tracking")

led to an overly small weight for interactive group entities. The bad case
can be easily reproduced when a number of CPU hogs compete for the CPUs
at the same time (thanks to Mike). This is largly because the task group's
load average tracking cross CPUs lags behind the real changes.

To fix this we accelerate the group share distribution process by using
the load.weight of the cfs_rq. This may increase the entire group's
share, but we have to do so to protect the (fragile) interactive
tasks, especially from CPU hogs.

Reported-by: Mike Galbraith <umgwanakikbuti@gmail.com>
Tested-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Tested-by: Mike Galbraith <umgwanakikbuti@gmail.com>
Signed-off-by: Yuyang Du <yuyang.du@intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1444699103-20272-1-git-send-email-yuyang.du@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/x86: Fix typo in __switch_to() comments
Chuck Ebbert [Wed, 14 Oct 2015 18:31:19 +0000 (14:31 -0400)]
sched/x86: Fix typo in __switch_to() comments

Fix obvious mistake: FS/GS should be DS/ES.

Signed-off-by: Chuck Ebbert <cebbert.lkml@gmail.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20151014143119.78858eeb@r5
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched, tracing: Stop/start critical timings around the idle=poll idle loop
Daniel Bristot de Oliveira [Thu, 8 Oct 2015 18:36:06 +0000 (15:36 -0300)]
sched, tracing: Stop/start critical timings around the idle=poll idle loop

When using idle=poll, the preemptoff tracer is always showing
the idle task as the culprit for long latencies. That happens
because critical timings are not stopped before idle loop. This
patch stops critical timings before entering the idle loop,
starting it again after the idle loop.

This problem does not affect the irqsoff tracer because
interruptions are enabled before entering the idle loop.

Signed-off-by: Daniel Bristot de Oliveira <bristot@redhat.com>
Reviewed-by: Luis Claudio R. Goncalves <lgoncalv@redhat.com>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/10fc3705874aef11dbe152a068b591a7be1899b4.1444314899.git.bristot@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agoLinux 4.3-rc5
Linus Torvalds [Sun, 11 Oct 2015 18:09:45 +0000 (11:09 -0700)]
Linux 4.3-rc5

9 years agoMerge branch 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 11 Oct 2015 17:24:32 +0000 (10:24 -0700)]
Merge branch 'sched-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull scheduler fix from Thomas Gleixner:
 "Fix a long standing state race in finish_task_switch()"

* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched/core: Fix TASK_DEAD race in finish_task_switch()

9 years agoMerge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 11 Oct 2015 17:23:52 +0000 (10:23 -0700)]
Merge branch 'perf-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull perf fix from Thomas Glexiner:
 "Fix build breakage on powerpc in perf tools"

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf tools: Fix build break on powerpc due to sample_reg_masks

9 years agoMerge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 11 Oct 2015 17:23:00 +0000 (10:23 -0700)]
Merge branch 'x86-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull maintainer email update from Thomas Gleixner:
 "Change Matt Fleming's email address in the maintainers file"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  MAINTAINERS: Change Matt Fleming's email address

9 years agoMerge branch 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 11 Oct 2015 17:16:59 +0000 (10:16 -0700)]
Merge branch 'irq-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull irq fixes from Thomas Gleixner:
 "Three trivial commits:

   - Fix a kerneldoc regression

   - Export handle_bad_irq to unbreak a driver in next

   - Add an accessor for the of_node field so refactoring in next does
     not depend on merge ordering"

* 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqdomain: Add an accessor for the of_node field
  genirq: Fix handle_bad_irq kerneldoc comment
  genirq: Export handle_bad_irq

9 years agoMerge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Sun, 11 Oct 2015 17:02:30 +0000 (10:02 -0700)]
Merge tag 'scsi-fixes' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "This is a set of three bug fixes, two of which are regressions from
  recent updates (the 3ware one from 4.1 and the device handler fixes
  from 4.2)"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  3w-9xxx: don't unmap bounce buffered commands
  scsi_dh: Use the correct module name when loading device handler
  libiscsi: Fix iscsi_check_transport_timeouts possible infinite loop

9 years agoMerge tag 'md/4.3-rc4-fix' of git://neil.brown.name/md
Linus Torvalds [Sun, 11 Oct 2015 16:35:51 +0000 (09:35 -0700)]
Merge tag 'md/4.3-rc4-fix' of git://neil.brown.name/md

Pull md bugfix from Neil Brown:
 "One bug fix for raid1/raid10.

  Very careless bug earler in 4.3-rc, now fixed :-)"

* tag 'md/4.3-rc4-fix' of git://neil.brown.name/md:
  crash in md-raid1 and md-raid10 due to incorrect list manipulation

9 years agoMAINTAINERS: Change Matt Fleming's email address
Matt Fleming [Sat, 10 Oct 2015 16:22:16 +0000 (17:22 +0100)]
MAINTAINERS: Change Matt Fleming's email address

My Intel email address will soon expire. Replace it with my
personal address so people still know where to send patches.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1444494136-10333-1-git-send-email-matt@codeblueprint.co.uk
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agoMerge tag 'usb-4.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Sat, 10 Oct 2015 18:17:45 +0000 (11:17 -0700)]
Merge tag 'usb-4.3-rc5' of git://git./linux/kernel/git/gregkh/usb

Pull USB fixes from Greg KH:
 "Here are some small USB and PHY fixes and quirk updates for 4.3-rc5.

  Nothing major here, full details in the shortlog, and all of these
  have been in linux-next for a while"

* tag 'usb-4.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  usb: Add device quirk for Logitech PTZ cameras
  USB: chaoskey read offset bug
  USB: Add reset-resume quirk for two Plantronics usb headphones.
  usb: renesas_usbhs: Add support for R-Car H3
  usb: renesas_usbhs: fix build warning if 64-bit architecture
  usb: gadget: bdc: fix memory leak
  phy: berlin-sata: Fix module autoload for OF platform driver
  phy: rockchip-usb: power down phy when rockchip phy probe
  phy: qcom-ufs: fix build error when the component is built as a module

9 years agoMerge tag 'tty-4.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Linus Torvalds [Sat, 10 Oct 2015 18:09:55 +0000 (11:09 -0700)]
Merge tag 'tty-4.3-rc5' of git://git./linux/kernel/git/gregkh/tty

Pull tty/serial driver fixes from Greg KH:
 "Here are a few bug fixes for the tty core that resolve reported
  issues, and some serial driver fixes as well (including the
  much-reported imx driver problem)

  All of these have been in linux-next with no reported problems"

* tag 'tty-4.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  drivers/tty: require read access for controlling terminal
  serial: 8250: add uart_config entry for PORT_RT2880
  tty: fix data race on tty_buffer.commit
  tty: fix data race in tty_buffer_flush
  tty: fix data race in flush_to_ldisc
  tty: fix stall caused by missing memory barrier in drivers/tty/n_tty.c
  serial: atmel: fix error path of probe function
  tty: don't leak cdev in tty_cdev_add()
  Revert "serial: imx: remove unbalanced clk_prepare"

9 years agoMerge tag 'staging-4.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Sat, 10 Oct 2015 18:03:31 +0000 (11:03 -0700)]
Merge tag 'staging-4.3-rc5' of git://git./linux/kernel/git/gregkh/staging

Pull staging fixes from Greg KH:
 "Here are two tiny staging tree fixes for 4.3-rc5.

  One fixes the broken speakup subsystem as reported by a user, and the
  other removes an entry in the MAINTAINERS file for a developer that
  doesn't want to be listed anymore"

* tag 'staging-4.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
  staging: speakup: fix speakup-r regression
  MAINTAINERS: Remove myself as nvec co-maintainer

9 years agoMerge tag 'char-misc-4.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Sat, 10 Oct 2015 17:58:27 +0000 (10:58 -0700)]
Merge tag 'char-misc-4.3-rc5' of git://git./linux/kernel/git/gregkh/char-misc

Pull char/misc driver fixes from Greg KH:
 "Here are some small fixes for some misc drivers that resolve some
  reported issues.  All of these have been linux-next for a while"

* tag 'char-misc-4.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  mcb: Fix error handling in mcb_pci_probe()
  mei: hbm: fix error in state check logic
  nvmem: sunxi: Check for memory allocation failure
  nvmem: core: Fix memory leak in nvmem_cell_write
  nvmem: core: Handle shift bits in-place if cell->nbits is non-zero
  nvmem: core: fix the out-of-range leak in read/write()

9 years agoMerge git://git.linux-mips.org/pub/scm/ralf/upstream-linus
Linus Torvalds [Sat, 10 Oct 2015 17:51:55 +0000 (10:51 -0700)]
Merge git://git.linux-mips.org/ralf/upstream-linus

Pull MIPS fixes from Ralf Baechle:

 - MIPS didn't define the new ioremap_uc.  Defined it as an alias for
   ioremap_uncached.

 - Replace workaround for MIPS16 build issue with a correct one.

* git://git.linux-mips.org/pub/scm/ralf/upstream-linus:
  MIPS: Define ioremap_uc
  MIPS: UAPI: Ignore __arch_swab{16,32,64} when using MIPS16
  Revert "MIPS: UAPI: Fix unrecognized opcode WSBH/DSBH/DSHD when using MIPS16."

9 years agoMerge branch 'stable/for-linus-4.3' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 10 Oct 2015 17:31:13 +0000 (10:31 -0700)]
Merge branch 'stable/for-linus-4.3' of git://git./linux/kernel/git/konrad/swiotlb

Pull swiotlb fixlet from Konrad Rzeszutek Wilk:
 "Enable the SWIOTLB under 32-bit PAE kernels.

  Nowadays most distros enable this due to CONFIG_HYPERVISOR|XEN=y which
  select SWIOTLB.  But for those that are not interested in
  virtualization and wanting to use 32-bit PAE kernels and wanting to
  have working DMA operations - this configures it for them"

* 'stable/for-linus-4.3' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/swiotlb:
  swiotlb: Enable it under x86 PAE

9 years agonamei: results of d_is_negative() should be checked after dentry revalidation
Trond Myklebust [Fri, 9 Oct 2015 17:44:34 +0000 (13:44 -0400)]
namei: results of d_is_negative() should be checked after dentry revalidation

Leandro Awa writes:
 "After switching to version 4.1.6, our parallelized and distributed
  workflows now fail consistently with errors of the form:

  T34: ./regex.c:39:22: error: config.h: No such file or directory

  From our 'git bisect' testing, the following commit appears to be the
  possible cause of the behavior we've been seeing: commit 766c4cbfacd8"

Al Viro says:
 "What happens is that 766c4cbfacd8 got the things subtly wrong.

  We used to treat d_is_negative() after lookup_fast() as "fall with
  ENOENT".  That was wrong - checking ->d_flags outside of ->d_seq
  protection is unreliable and failing with hard error on what should've
  fallen back to non-RCU pathname resolution is a bug.

  Unfortunately, we'd pulled the test too far up and ran afoul of
  another kind of staleness.  The dentry might have been absolutely
  stable from the RCU point of view (and we might be on UP, etc), but
  stale from the remote fs point of view.  If ->d_revalidate() returns
  "it's actually stale", dentry gets thrown away and the original code
  wouldn't even have looked at its ->d_flags.

  What we need is to check ->d_flags where 766c4cbfacd8 does (prior to
  ->d_seq validation) but only use the result in cases where we do not
  discard this dentry outright"

Reported-by: Leandro Awa <lawa@nvidia.com>
Link: https://bugzilla.kernel.org/show_bug.cgi?id=104911
Fixes: 766c4cbfacd8 ("namei: d_is_negative() should be checked...")
Tested-by: Leandro Awa <lawa@nvidia.com>
Cc: stable@vger.kernel.org # v4.1+
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge tag 'pm+acpi-4.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Sat, 10 Oct 2015 01:39:04 +0000 (18:39 -0700)]
Merge tag 'pm+acpi-4.3-rc5' of git://git./linux/kernel/git/rafael/linux-pm

Pull power management and ACPI fixes from Rafael Wysocki:
 "These are four fixes for bugs in the devfreq and cpufreq subsystems,
  including two regression fixes (one for a recent regression and one
  for a problem introduced in 4.2).

  Specifics:

   - Two fixes for cpufreq regressions, an acpi-cpufreq driver one
     introduced during the 4.2 cycle when we started to preserve cpufreq
     directories for offline CPUs and a general one introduced recently
     (Srinivas Pandruvada).

   - Two devfreq fixes, one for a double kfree() in an error code path
     and one for a confusing sysfs-related failure (Geliang Tang, Tobias
     Jakobi)"

* tag 'pm+acpi-4.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpufreq: prevent lockup on reading scaling_available_frequencies
  cpufreq: acpi_cpufreq: prevent crash on reading freqdomain_cpus
  PM / devfreq: fix double kfree
  PM / devfreq: Fix governor_store()

9 years agoMerge branch 'strscpy' of git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf...
Linus Torvalds [Sat, 10 Oct 2015 01:01:26 +0000 (18:01 -0700)]
Merge branch 'strscpy' of git://git./linux/kernel/git/cmetcalf/linux-tile

Pull strscpy powerpc fix from Chris Metcalf.

Fix powerpc big-endian build.

* 'strscpy' of git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile:
  arch/powerpc: provide zero_bytemask() for big-endian

9 years agoMerge tag 'sound-4.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Sat, 10 Oct 2015 00:56:00 +0000 (17:56 -0700)]
Merge tag 'sound-4.3-rc5' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "We see various small fixes, but nothing looks too scary, all are small
  gentle bug fixes:

   - Most of changes are for ASoC codecs: Realtek, SGTL5000, TAS2552,
     TLV320, WM8962

   - A couple of dwc and imx-ssi fixes

   - Usual oneliner HD-audio quirks

   - An old emux synth code fix"

* tag 'sound-4.3-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  MAINTAINERS: Remove wm97xx entry
  ASoC: tas2552: fix dBscale-min declaration
  ALSA: synth: Fix conflicting OSS device registration on AWE32
  ALSA: hda - Disable power_save_node for IDT 92HD73xx chips
  ALSA: hda - Apply SPDIF pin ctl to MacBookPro 12,1
  ALSA: hda: Add dock support for ThinkPad T550
  ASoC: dwc: fix dma stop transferring issue
  ASoC: dwc: correct irq clear method
  ASoC: tlv320aic3x: Prevent writing reserved registers on tlv320aic3104 CODECs
  ASoC: rt5645: Correct the naming and setting of ADC Boost Volume Control
  ASoC: sgtl5000: fix wrong register MIC_BIAS_VOLTAGE setup on probe
  ASoC: wm8962: balance pm_runtime_enable
  ASoC: imx-ssi: Fix DAI hardware signal inversions
  ASoC: sgtl5000: fix error message output for MicBias voltage
  ASoC: db1200: Fix DAI link format for db1300 and db1550

9 years agoMerge tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 10 Oct 2015 00:33:15 +0000 (17:33 -0700)]
Merge tag 'clk-fixes-for-linus' of git://git./linux/kernel/git/clk/linux

Pull clk fixes from Stephen Boyd:
 "A Samsung driver fix and a handful of TI driver fixes"

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  clk: ti: dflt: fix enable_reg validity check
  clk: ti: fix dual-registration of uart4_ick
  clk: ti: clk-7xx: Remove hardwired ABE clock configuration
  clk: samsung: fix cpu clock's flags checking

9 years agoMerge tag 'mmc-v4.3-rc4' of git://git.linaro.org/people/ulf.hansson/mmc
Linus Torvalds [Sat, 10 Oct 2015 00:25:26 +0000 (17:25 -0700)]
Merge tag 'mmc-v4.3-rc4' of git://git.linaro.org/people/ulf.hansson/mmc

Pull MMC fixes from Ulf Hansson:
 "MMC host:
   - omap_hsmmc: Fix boot regressions for omap4430
   - omap_hsmmc: Fix legacy boot regression for omap3
   - sdhci-pxav3: Fix some clock issues for Armada 38x
   - sdhci-pxav3: Fix error handling at probe
   - sdhci-of-at91: Fix clock stabilization problem"

* tag 'mmc-v4.3-rc4' of git://git.linaro.org/people/ulf.hansson/mmc:
  mmc: sdhci-of-at91: use SDHCI_QUIRK2_NEED_DELAY_AFTER_INT_CLK_RST quirk
  mmc: sdhci: add quirk SDHCI_QUIRK2_NEED_DELAY_AFTER_INT_CLK_RST
  mmc: sdhci-pxav3: fix error handling of armada_38x_quirks
  mmc: sdhci-pxav3: disable clock inversion for HS MMC cards
  mmc: sdhci-pxav3: remove broken clock base quirk for Armada 38x sdhci driver
  mmc: host: omap_hsmmc: Fix MMC for omap3 legacy booting
  Revert "mmc: host: omap_hsmmc: use regulator_is_enabled to find pbias status"

9 years agoMerge tag 'dm-4.3-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/device...
Linus Torvalds [Fri, 9 Oct 2015 23:58:11 +0000 (16:58 -0700)]
Merge tag 'dm-4.3-fixes-2' of git://git./linux/kernel/git/device-mapper/linux-dm

Pull dm fixes from Mike Snitzer:
 "Three stable fixes:

   - DM core AB-BA deadlock fix in the device destruction path (vs
     device creation's DM table swap).

   - DM raid fix to properly round up the region_size to the next
     power-of-2.

   - DM cache fix for a NULL pointer seen while switching from the
     "cleaner" cache policy.

  Two fixes for regressions introduced during the 4.3 merge:

   - request-based DM error propagation regressed due to incorrect
     changes introduced when adding the bi_error field to bio.

   - DM snapshot fix to only support snapshots that overflow if the
     client (e.g. lvm2) is prepared to deal with the associated
     snapshot status interface change"

* tag 'dm-4.3-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm:
  dm snapshot: add new persistent store option to support overflow
  dm cache: fix NULL pointer when switching from cleaner policy
  dm: fix request-based dm error reporting
  dm raid: fix round up of default region size
  dm: fix AB-BA deadlock in __dm_destroy()

9 years agoMerge branch 'for-linus-4.3' of git://git.kernel.org/pub/scm/linux/kernel/git/mason...
Linus Torvalds [Fri, 9 Oct 2015 23:39:35 +0000 (16:39 -0700)]
Merge branch 'for-linus-4.3' of git://git./linux/kernel/git/mason/linux-btrfs

Pull btrfs fixes from Chris Mason:
 "These are small and assorted.  Neil's is the oldest, I dropped the
  ball thinking he was going to send it in"

* 'for-linus-4.3' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  Btrfs: support NFSv2 export
  Btrfs: open_ctree: Fix possible memory leak
  Btrfs: fix deadlock when finalizing block group creation
  Btrfs: update fix for read corruption of compressed and shared extents
  Btrfs: send, fix corner case for reference overwrite detection

9 years agoMerge tag 'nfsd-4.3-1' of git://linux-nfs.org/~bfields/linux
Linus Torvalds [Fri, 9 Oct 2015 23:34:45 +0000 (16:34 -0700)]
Merge tag 'nfsd-4.3-1' of git://linux-nfs.org/~bfields/linux

Pull nfsd bugfix from Bruce Fields:
 "Just one RDMA bugfix"

* tag 'nfsd-4.3-1' of git://linux-nfs.org/~bfields/linux:
  svcrdma: handle rdma read with a non-zero initial page offset

9 years agoMerge tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm...
Linus Torvalds [Fri, 9 Oct 2015 22:54:14 +0000 (15:54 -0700)]
Merge tag 'fixes-for-linus' of git://git./linux/kernel/git/arm/arm-soc

Pull ARM SoC fixes from Arnd Bergmann:
 "The fixes for this week include one small patch that was years in the
  making and that finally fixes using all eight CPUs on exynos542x.

  The rest are lots of minor changes for sunxi, imx, exynos and shmobile

   - fixing the minimum voltage for Allwinner A20
   - thermal boot issue on SMDK5250.
   - invalid clock used for FIMD IOMMU.
   - audio on Renesas r8a7790/r8a7791
   - invalid clock used for FIMD IOMMU
   - LEDs on exynos5422-odroidxu3-common
   - usb pin control for imx-rex
   - imx53: fix PMIC interrupt level
   - a Makefile typo"

* tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  ARM: dts: Fix wrong clock binding for sysmmu_fimd1_1 on exynos5420
  ARM: dts: Fix bootup thermal issue on smdk5250
  ARM: shmobile: r8a7791 dtsi: Add CPG/MSTP Clock Domain for sound
  ARM: shmobile: r8a7790 dtsi: Add CPG/MSTP Clock Domain for sound
  arm-cci500: Don't enable PMU driver by default
  ARM: dts: fix usb pin control for imx-rex dts
  ARM: imx53: qsrb: fix PMIC interrupt level
  ARM: imx53: include IRQ dt-bindings header
  ARM: dts: add suspend opp to exynos4412
  ARM: dts: Fix LEDs on exynos5422-odroidxu3
  ARM: EXYNOS: reset Little cores when cpu is up
  ARM: dts: Fix Makefile target for sun4i-a10-itead-iteaduino-plus
  ARM: dts: sunxi: Raise minimum CPU voltage for sun7i-a20 to meet SoC specifications

9 years agodm snapshot: add new persistent store option to support overflow
Mike Snitzer [Thu, 8 Oct 2015 22:05:41 +0000 (18:05 -0400)]
dm snapshot: add new persistent store option to support overflow

Commit 76c44f6d80 introduced the possibly for "Overflow" to be reported
by the snapshot device's status.  Older userspace (e.g. lvm2) does not
handle the "Overflow" status response.

Fix this incompatibility by requiring newer userspace code, that can
cope with "Overflow", request the persistent store with overflow support
by using "PO" (Persistent with Overflow) for the snapshot store type.

Reported-by: Zdenek Kabelac <zkabelac@redhat.com>
Fixes: 76c44f6d80 ("dm snapshot: don't invalidate on-disk image on snapshot write overflow")
Reviewed-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
9 years agoMerge branches 'pm-devfreq' and 'pm-cpufreq'
Rafael J. Wysocki [Fri, 9 Oct 2015 20:13:30 +0000 (22:13 +0200)]
Merge branches 'pm-devfreq' and 'pm-cpufreq'

* pm-devfreq:
  PM / devfreq: fix double kfree
  PM / devfreq: Fix governor_store()

* pm-cpufreq:
  cpufreq: prevent lockup on reading scaling_available_frequencies
  cpufreq: acpi_cpufreq: prevent crash on reading freqdomain_cpus

9 years agoirqdomain: Add an accessor for the of_node field
Marc Zyngier [Fri, 9 Oct 2015 14:50:11 +0000 (15:50 +0100)]
irqdomain: Add an accessor for the of_node field

As we're about to remove the of_node field from the irqdomain
structure, introduce an accessor for it. Subsequent patches
will take care of the actual repainting.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Cc: Jiang Liu <jiang.liu@linux.intel.com>
Cc: Jason Cooper <jason@lakedaemon.net>
Link: http://lkml.kernel.org/r/1444402211-1141-1-git-send-email-marc.zyngier@arm.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agogenirq: Fix handle_bad_irq kerneldoc comment
Arnd Bergmann [Tue, 6 Oct 2015 20:59:40 +0000 (22:59 +0200)]
genirq: Fix handle_bad_irq kerneldoc comment

A recent cleanup removed the 'irq' parameter from many functions, but
left the documentation for this in place for at least one function.

This removes it.

Fixes: bd0b9ac405e1 ("genirq: Remove irq argument from irq flow handlers")
Reported-by: kbuild test robot <lkp@intel.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: kbuild-all@01.org
Cc: Austin Schuh <austin@peloton-tech.com>
Cc: Santosh Shilimkar <ssantosh@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: http://lkml.kernel.org/r/5400000.cD19rmgWjV@wuerfel
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agogenirq: Export handle_bad_irq
Arnd Bergmann [Tue, 6 Oct 2015 20:24:50 +0000 (22:24 +0200)]
genirq: Export handle_bad_irq

A cleanup of the omap gpio driver introduced a use of the
handle_bad_irq() function in a device driver that can be
a loadable module.

This broke the ARM allmodconfig build:

ERROR: "handle_bad_irq" [drivers/gpio/gpio-omap.ko] undefined!

This patch exports the handle_bad_irq symbol in order to
allow the use in modules.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Cc: Santosh Shilimkar <ssantosh@kernel.org>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Austin Schuh <austin@peloton-tech.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: linux-arm-kernel@lists.infradead.org
Link: http://lkml.kernel.org/r/5847725.4IBopItaOr@wuerfel
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agodm cache: fix NULL pointer when switching from cleaner policy
Joe Thornber [Fri, 9 Oct 2015 13:03:38 +0000 (14:03 +0100)]
dm cache: fix NULL pointer when switching from cleaner policy

The cleaner policy doesn't make use of the per cache block hint space in
the metadata (unlike the other policies).  When switching from the
cleaner policy to mq or smq a NULL pointer crash (in dm_tm_new_block)
was observed.  The crash was caused by bugs in dm-cache-metadata.c
when trying to skip creation of the hint btree.

The minimal fix is to change hint size for the cleaner policy to 4 bytes
(only hint size supported).

Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: stable@vger.kernel.org
9 years agocrash in md-raid1 and md-raid10 due to incorrect list manipulation
Mikulas Patocka [Thu, 1 Oct 2015 19:17:43 +0000 (15:17 -0400)]
crash in md-raid1 and md-raid10 due to incorrect list manipulation

The commit 55ce74d4bfe1b9444436264c637f39a152d1e5ac (md/raid1: ensure
device failure recorded before write request returns) is causing crash in
the LVM2 testsuite test shell/lvchange-raid.sh. For me the crash is 100%
reproducible.

The reason for the crash is that the newly added code in raid1d moves the
list from conf->bio_end_io_list to tmp, then tests if tmp is non-empty and
then incorrectly pops the bio from conf->bio_end_io_list (which is empty
because the list was alrady moved).

Raid-10 has a similar bug.

Kernel Fault: Code=15 regs=000000006ccb8640 (Addr=0000000100000000)
CPU: 3 PID: 1930 Comm: mdX_raid1 Not tainted 4.2.0-rc5-bisect+ #35
task: 000000006cc1f258 ti: 000000006ccb8000 task.ti: 000000006ccb8000

     YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
PSW: 00001000000001001111111000001111 Not tainted
r00-03  000000ff0804fe0f 000000001059d000 000000001059f818 000000007f16be38
r04-07  000000001059d000 000000007f16be08 0000000000200200 0000000000000001
r08-11  000000006ccb8260 000000007b7934d0 0000000000000001 0000000000000000
r12-15  000000004056f320 0000000000000000 0000000000013dd0 0000000000000000
r16-19  00000000f0d00ae0 0000000000000000 0000000000000000 0000000000000001
r20-23  000000000800000f 0000000042200390 0000000000000000 0000000000000000
r24-27  0000000000000001 000000000800000f 000000007f16be08 000000001059d000
r28-31  0000000100000000 000000006ccb8560 000000006ccb8640 0000000000000000
sr00-03  0000000000249800 0000000000000000 0000000000000000 0000000000249800
sr04-07  0000000000000000 0000000000000000 0000000000000000 0000000000000000

IASQ: 0000000000000000 0000000000000000 IAOQ: 000000001059f61c 000000001059f620
 IIR: 0f8010c6    ISR: 0000000000000000  IOR: 0000000100000000
 CPU:        3   CR30: 000000006ccb8000 CR31: 0000000000000000
 ORIG_R28: 000000001059d000
 IAOQ[0]: call_bio_endio+0x34/0x1a8 [raid1]
 IAOQ[1]: call_bio_endio+0x38/0x1a8 [raid1]
 RP(r2): raid_end_bio_io+0x88/0x168 [raid1]
Backtrace:
 [<000000001059f818>] raid_end_bio_io+0x88/0x168 [raid1]
 [<00000000105a4f64>] raid1d+0x144/0x1640 [raid1]
 [<000000004017fd5c>] kthread+0x144/0x160

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Fixes: 55ce74d4bfe1 ("md/raid1: ensure device failure recorded before write request returns.")
Fixes: 95af587e95aa ("md/raid10: ensure device failure recorded before write request returns.")
Signed-off-by: NeilBrown <neilb@suse.com>
9 years agocpufreq: prevent lockup on reading scaling_available_frequencies
Srinivas Pandruvada [Wed, 7 Oct 2015 20:50:44 +0000 (13:50 -0700)]
cpufreq: prevent lockup on reading scaling_available_frequencies

When scaling_available_frequencies is read on an offlined cpu, then
either lockup or junk values are displayed. This is caused by
freed freq_table, which policy is using.

Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
9 years agocpufreq: acpi_cpufreq: prevent crash on reading freqdomain_cpus
Srinivas Pandruvada [Wed, 7 Oct 2015 20:50:43 +0000 (13:50 -0700)]
cpufreq: acpi_cpufreq: prevent crash on reading freqdomain_cpus

When freqdomain_cpus attribute is read from an offlined cpu, it will
cause crash. This change prevents calling cpufreq_show_cpus when
policy driver_data is NULL.

Crash info:

[  170.814949] BUG: unable to handle kernel NULL pointer dereference at 0000000000000018
[  170.814990] IP: [<ffffffff813b2490>] _find_next_bit.part.0+0x10/0x70
[  170.815021] PGD 227d30067 PUD 229e56067 PMD 0
[  170.815043] Oops: 0000 [#2] SMP
[  170.816022] CPU: 3 PID: 3121 Comm: cat Tainted: G      D    OE   4.3.0-rc3+ #33
...
...
[  170.816657] Call Trace:
[  170.816672]  [<ffffffff813b2505>] ? find_next_bit+0x15/0x20
[  170.816696]  [<ffffffff8160e47c>] cpufreq_show_cpus+0x5c/0xd0
[  170.816722]  [<ffffffffa031a409>] show_freqdomain_cpus+0x19/0x20 [acpi_cpufreq]
[  170.816749]  [<ffffffff8160e65b>] show+0x3b/0x60
[  170.816769]  [<ffffffff8129b31c>] sysfs_kf_seq_show+0xbc/0x130
[  170.816793]  [<ffffffff81299be3>] kernfs_seq_show+0x23/0x30
[  170.816816]  [<ffffffff81240f2c>] seq_read+0xec/0x390
[  170.816837]  [<ffffffff8129a64a>] kernfs_fop_read+0x10a/0x160
[  170.816861]  [<ffffffff8121d9b7>] __vfs_read+0x37/0x100
[  170.816883]  [<ffffffff813217c0>] ? security_file_permission+0xa0/0xc0
[  170.816909]  [<ffffffff8121e2e3>] vfs_read+0x83/0x130
[  170.816930]  [<ffffffff8121f035>] SyS_read+0x55/0xc0
...
...
[  170.817185] ---[ end trace bc6eadf82b2b965a ]---

Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Cc: 4.2+ <stable@vger.kernel.org> # 4.2+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
9 years agommc: sdhci-of-at91: use SDHCI_QUIRK2_NEED_DELAY_AFTER_INT_CLK_RST quirk
ludovic.desroches@atmel.com [Thu, 17 Sep 2015 08:16:20 +0000 (10:16 +0200)]
mmc: sdhci-of-at91: use SDHCI_QUIRK2_NEED_DELAY_AFTER_INT_CLK_RST quirk

The Atmel sdhci device needs the
SDHCI_QUIRK2_NEED_DELAY_AFTER_INT_CLK_RST quirk. Without it, the
internal clock could never stabilised when changing the sd clock
frequency.

Signed-off-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci: add quirk SDHCI_QUIRK2_NEED_DELAY_AFTER_INT_CLK_RST
ludovic.desroches@atmel.com [Thu, 17 Sep 2015 08:16:19 +0000 (10:16 +0200)]
mmc: sdhci: add quirk SDHCI_QUIRK2_NEED_DELAY_AFTER_INT_CLK_RST

The Atmel sdhci device needs a new quirk. sdhci_set_clock set the Clock
Control Register to 0 before computing the new value and writing it.
It disables the internal clock which causes a reset mecanism. If we
write the new value before this reset mecanism is done, it will prevent
the stabilisation of the internal clock, so a delay is needed. This
delay is about 2-3 cycles of the base clock. To be safe, a 1 ms delay is
used.

Signed-off-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-pxav3: fix error handling of armada_38x_quirks
Marcin Wojtas [Tue, 6 Oct 2015 01:22:37 +0000 (03:22 +0200)]
mmc: sdhci-pxav3: fix error handling of armada_38x_quirks

In case of armada_38x_quirks error, all clocks should be cleaned-up, same
as after mv_conf_mbus_windows failure.

Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Cc: <stable@vger.kernel.org> # v4.2
Reviewed-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-pxav3: disable clock inversion for HS MMC cards
Nadav Haklai [Tue, 6 Oct 2015 01:22:36 +0000 (03:22 +0200)]
mmc: sdhci-pxav3: disable clock inversion for HS MMC cards

According to 'FE-2946959' erratum the clock inversion option is
needed to support slow frequencies when the card input hold time
requirement is high. This setting is not required for high speed
MMC and might cause timing violation.

Signed-off-by: Nadav Haklai <nadavh@marvell.com>
Cc: <stable@vger.kernel.org> # v4.2
Reviewed-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agommc: sdhci-pxav3: remove broken clock base quirk for Armada 38x sdhci driver
Nadav Haklai [Tue, 6 Oct 2015 01:22:35 +0000 (03:22 +0200)]
mmc: sdhci-pxav3: remove broken clock base quirk for Armada 38x sdhci driver

shci-pxav3 driver is enabling by default the
SDHCI_QUIRK_CAP_CLOCK_BASE_BROKEN quirk. However this quirk is not
required for Armada 38x and leads to wrong clock setting in the divider.

Signed-off-by: Nadav Haklai <nadavh@marvell.com>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Cc: <stable@vger.kernel.org> # v4.2
Reviewed-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agoarch/powerpc: provide zero_bytemask() for big-endian
Chris Metcalf [Wed, 7 Oct 2015 13:29:11 +0000 (09:29 -0400)]
arch/powerpc: provide zero_bytemask() for big-endian

For some reason, only the little-endian flavor of
powerpc provided the zero_bytemask() implementation.

Reported-by: Michal Sojka <sojkam1@fel.cvut.cz>
Acked-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Chris Metcalf <cmetcalf@ezchip.com>
9 years agommc: host: omap_hsmmc: Fix MMC for omap3 legacy booting
Tony Lindgren [Wed, 7 Oct 2015 13:22:25 +0000 (06:22 -0700)]
mmc: host: omap_hsmmc: Fix MMC for omap3 legacy booting

Starting with commit 7d607f917008 ("mmc: host: omap_hsmmc: use
devm_regulator_get_optional() for vmmc") MMC on omap3 stopped working
for legacy booting.

This is because legacy booting sets up some of the resource in the
platform init code, and for optional regulators always seem to
return -EPROBE_DEFER for the legacy booting.

Let's fix the issue by checking for device tree based booting for
now. Then when omap3 boots in device tree only mode, this patch
can be just reverted.

Fixes: 7d607f917008 ("mmc: host: omap_hsmmc: use
devm_regulator_get_optional() for vmmc")
Cc: Felipe Balbi <balbi@ti.com>
Cc: Kishon Vijay Abraham I <kishon@ti.com>
Cc: Nishanth Menon <nm@ti.com>
Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Tested-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agoRevert "mmc: host: omap_hsmmc: use regulator_is_enabled to find pbias status"
Tony Lindgren [Wed, 7 Oct 2015 13:22:24 +0000 (06:22 -0700)]
Revert "mmc: host: omap_hsmmc: use regulator_is_enabled to find pbias status"

This reverts commit c55d7a0553643a7e8f120688b82b594471084d3c.

Without reverting this commit we get "unbalanced disables for pbias_mmc_omap4"
errors on omap4430. It seems that 4430 and 4460 behave in a different way for
the PBIAS regulator registers and until that has been debugged further we
cannot rely on the regulator status registers in hardare on 4430.

Fixes: 7d607f917008 ("mmc: host: omap_hsmmc: use
devm_regulator_get_optional() for vmmc")
Cc: Felipe Balbi <balbi@ti.com>
Cc: Kishon Vijay Abraham I <kishon@ti.com>
Cc: Nishanth Menon <nm@ti.com>
Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Tested-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
9 years agoswiotlb: Enable it under x86 PAE
Christian Melki [Mon, 5 Oct 2015 15:31:33 +0000 (17:31 +0200)]
swiotlb: Enable it under x86 PAE

Most distributions end up enabling SWIOTLB already with 32-bit
kernels due to the combination of CONFIG_HYPERVISOR_GUEST|CONFIG_XEN=y
as those end up requiring the SWIOTLB.

However for those that are not interested in virtualization and
run in 32-bit they will discover that: "32-bit PAE 4.2.0 kernel
(no IOMMU code) would hang when writing to my USB disk. The kernel
spews million(-ish messages per sec) to syslog, effectively
"hanging" userspace with my kernel.

Oct  2 14:33:06 voodoochild kernel: [  223.287447] nommu_map_sg:
overflow 25dcac000+1024 of device mask ffffffff
Oct  2 14:33:06 voodoochild kernel: [  223.287448] nommu_map_sg:
overflow 25dcac000+1024 of device mask ffffffff
Oct  2 14:33:06 voodoochild kernel: [  223.287449] nommu_map_sg:
overflow 25dcac000+1024 of device mask ffffffff
... etc ..."

Enabling it makes the problem go away.

N.B. With a6dfa128ce5c414ab46b1d690f7a1b8decb8526d
"config: Enable NEED_DMA_MAP_STATE by default when SWIOTLB is selected"
we also have the important part of the SG macros enabled to make this
work properly - in case anybody wants to backport this patch.

Reported-and-Tested-by: Christian Melki <christian.melki@t2data.com>
Signed-off-by: Christian Melki <christian.melki@t2data.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
9 years agoMerge tag 'asoc-fix-v4.3-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/brooni...
Takashi Iwai [Wed, 7 Oct 2015 18:11:21 +0000 (20:11 +0200)]
Merge tag 'asoc-fix-v4.3-rc4' of git://git./linux/kernel/git/broonie/sound into for-linus

ASoC: Fixes for v4.3

Quite a few fixes here but they're all very small and driver specific,
none of them really stand out if you aren't using the relevant hardware
but they're all useful if you do happen to have an affected device.

9 years ago3w-9xxx: don't unmap bounce buffered commands
Christoph Hellwig [Sat, 3 Oct 2015 17:16:07 +0000 (19:16 +0200)]
3w-9xxx: don't unmap bounce buffered commands

3w controller don't dma map small single SGL entry commands but instead
bounce buffer them.  Add a helper to identify these commands and don't
call scsi_dma_unmap for them.

Based on an earlier patch from James Bottomley.

Fixes: 118c85 ("3w-9xxx: fix command completion race")
Reported-by: Tóth Attila <atoth@atoth.sote.hu>
Tested-by: Tóth Attila <atoth@atoth.sote.hu>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Adam Radford <aradford@gmail.com>
Signed-off-by: James Bottomley <JBottomley@Odin.com>
9 years agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Wed, 7 Oct 2015 17:17:46 +0000 (18:17 +0100)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fixes from Will Deacon:
 "This addresses a couple of issues found with RT, a broken initrd
  message in the console log and a simple performance fix for some MMC
  workloads.

  Summary:

   - A couple of locking fixes for RT kernels
   - Avoid printing bogus initrd warnings when initrd isn't present
   - Performance fix for random mmap file readahead
   - Typo fix"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: replace read_lock to rcu lock in call_break_hook
  arm64: Don't relocate non-existent initrd
  arm64: convert patch_lock to raw lock
  arm64: readahead: fault retry breaks mmap file read random detection
  arm64: debug: Fix typo in debug-monitors.c

9 years agoMerge tag 'fbdev-fixes-4.3' of git://git.kernel.org/pub/scm/linux/kernel/git/tomba...
Linus Torvalds [Wed, 7 Oct 2015 17:05:09 +0000 (18:05 +0100)]
Merge tag 'fbdev-fixes-4.3' of git://git./linux/kernel/git/tomba/linux

Pull fbdev fixes from Tomi Valkeinen:

 - fbdev: Minor fixes to broadsheetfb, fsl-diu-fb, mb862xxfb, tridentfb,
   omapfb

 - display-timing: Fix memory leak in error path

* tag 'fbdev-fixes-4.3' of git://git.kernel.org/pub/scm/linux/kernel/git/tomba/linux:
  video: of: fix memory leak
  fbdev: broadsheetfb: fix memory leak
  OMAPDSS: panel-sony-acx565akm: Export OF module alias information
  fbdev: omap2: connector-dvi: use of_get_i2c_adapter_by_node interface
  tridentfb: Fix set_lwidth on TGUI9440 and CYBER9320
  tridentfb: fix hang on Blade3D with CONFIG_CC_OPTIMIZE_FOR_SIZE
  video: fbdev: mb862xx: Fix module autoload for OF platform driver
  video: fbdev: fsl: Fix the sleep function for FSL DIU module

9 years agoMerge tag 'perf-urgent-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git...
Ingo Molnar [Wed, 7 Oct 2015 16:33:10 +0000 (18:33 +0200)]
Merge tag 'perf-urgent-for-mingo' of git://git./linux/kernel/git/acme/linux into perf/urgent

Pull perf/urgent fix from Arnaldo Carvalho de Melo:

  - Fix build break on (at least) powerpc due to sample_reg_masks, not being
    available for linking. (Sukadev Bhattiprolu)

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agoMerge remote-tracking branches 'asoc/fix/tlv320aic3x' and 'asoc/fix/wm8962' into...
Mark Brown [Wed, 7 Oct 2015 15:07:50 +0000 (16:07 +0100)]
Merge remote-tracking branches 'asoc/fix/tlv320aic3x' and 'asoc/fix/wm8962' into asoc-linus

9 years agoMerge remote-tracking branches 'asoc/fix/db1200', 'asoc/fix/dwc', 'asoc/fix/imx-ssi...
Mark Brown [Wed, 7 Oct 2015 15:07:16 +0000 (16:07 +0100)]
Merge remote-tracking branches 'asoc/fix/db1200', 'asoc/fix/dwc', 'asoc/fix/imx-ssi', 'asoc/fix/maintainers', 'asoc/fix/rt5645', 'asoc/fix/sgtl5000' and 'asoc/fix/tas2552' into asoc-linus

9 years agoperf tools: Fix build break on powerpc due to sample_reg_masks
Sukadev Bhattiprolu [Thu, 24 Sep 2015 21:53:49 +0000 (17:53 -0400)]
perf tools: Fix build break on powerpc due to sample_reg_masks

perf_regs.c does not get built on Powerpc as CONFIG_PERF_REGS is false.
So the weak definition for 'sample_regs_masks' doesn't get picked up.

Adding perf_regs.o to util/Build unconditionally, exposes a redefinition
error for 'perf_reg_value()' function (due to the static inline version
in util/perf_regs.h). So use #ifdef HAVE_PERF_REGS_SUPPORT' around that
function.

Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: linuxppc-dev@ozlabs.org
Link: http://lkml.kernel.org/r/20150930182836.GA27858@us.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
9 years agoMerge tag 'regmap-fix-v4.3-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 7 Oct 2015 13:15:49 +0000 (14:15 +0100)]
Merge tag 'regmap-fix-v4.3-rc4' of git://git./linux/kernel/git/broonie/regmap

Pull regmap fixes from Mark Brown:
 "A couple of fixes for the debugfs information on the register map,
  fixing issues with very small reads potentially causing underflows and
  wraparounds"

* tag 'regmap-fix-v4.3-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap:
  regmap: debugfs: Don't bother actually printing when calculating max length
  regmap: debugfs: Ensure we don't underflow when printing access masks

9 years agoMerge tag 'spi-fix-v4.3-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi
Linus Torvalds [Wed, 7 Oct 2015 13:13:10 +0000 (14:13 +0100)]
Merge tag 'spi-fix-v4.3-rc4' of git://git./linux/kernel/git/broonie/spi

Pull spi fixes from Mark Brown:
 "A couple of very minor fixes, one for error handling in the Davinci
  driver probe function and another making the Renesas sh-msiof DT
  binding documentation correspond to what's actually implemented"

* tag 'spi-fix-v4.3-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
  spi: sh-msiof: Match renesas,rx-fifo-size in DT bindings doc with driver
  spi: davinci: fix handling platform_get_irq result

9 years agoMerge tag 'regulator-fix-v4.3-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 7 Oct 2015 13:11:47 +0000 (14:11 +0100)]
Merge tag 'regulator-fix-v4.3-rc4' of git://git./linux/kernel/git/broonie/regulator

Pull regulator fixes from Mark Brown:
 "Two fixes here, one device specific fix for axp20x and a core fix for
   cases where one regulator is supplying another which broke probe
  deferral, substituting in a dummy regulator too aggressively"

* tag 'regulator-fix-v4.3-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator:
  regulator: core: Handle probe deferral from DT when resolving supplies
  regulator: axp20x: Fix enable bit indexes for DCDC4 and DCDC5

9 years agovideo: of: fix memory leak
Sudip Mukherjee [Wed, 30 Sep 2015 09:54:08 +0000 (15:24 +0530)]
video: of: fix memory leak

If of_parse_display_timing() fails we are printing an error message and
jumping to the error path but we missed freeing "dt".

Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
9 years agoMerge remote-tracking branches 'spi/fix/davinci' and 'spi/fix/sh-msiof' into spi...
Mark Brown [Wed, 7 Oct 2015 10:43:39 +0000 (11:43 +0100)]
Merge remote-tracking branches 'spi/fix/davinci' and 'spi/fix/sh-msiof' into spi-linus

9 years agoMerge branch 'strscpy' of git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf...
Linus Torvalds [Wed, 7 Oct 2015 08:52:42 +0000 (09:52 +0100)]
Merge branch 'strscpy' of git://git./linux/kernel/git/cmetcalf/linux-tile

Pull strscpy fixes from Chris Metcalf :
 "This patch series fixes up a couple of architecture issues where
  strscpy wasn't configured correctly (missing on h8300, duplicating
  local and asm-generic copies on powerpc and tile).

  It also adds a use of zero_bytemask() to the final store for strscpy
  to avoid writing uninitialized data to the destination.  However, to
  make this work we had to add support for zero_bytemask() to the two
  architectures that didn't have it (alpha and tile), because they were
  providing their own local copies, but didn't provide the
  zero_bytemask() that was previously only required when building with
  CONFIG_DCACHE_WORD_ACCESS"

[ Side note: there is still no actual users of strscpy except for the
  one preexisting use in arch/tile that predates the generic version.
  So this is all about fixing the infrastructure so that we eventually
  can start using it.  - Linus ]

* 'strscpy' of git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile:
  strscpy: zero any trailing garbage bytes in the destination
  word-at-a-time.h: support zero_bytemask() on alpha and tile
  word-at-a-time.h: fix some Kbuild files

9 years agoMerge tag 'for-linus-20151006' of git://git.infradead.org/linux-mtd
Linus Torvalds [Wed, 7 Oct 2015 08:35:15 +0000 (09:35 +0100)]
Merge tag 'for-linus-20151006' of git://git.infradead.org/linux-mtd

Pull MTD fixes from Brian Norris:
 "A few MTD fixes:

   - mxc_nand: a "refactoring only" change in 4.3-rc1 had some bad
     pointer (array) arithmetic.  Fix that

   - sunxi_nand:

   - Fix an old list manipulation / memory management bug in the device
     release() code path

   - Correct a few mistakes in OOB write support"

* tag 'for-linus-20151006' of git://git.infradead.org/linux-mtd:
  mxc_nand: fix copy_spare
  mtd: nand: sunxi: fix sunxi_nand_chips_cleanup()
  mtd: nand: sunxi: fix OOB handling in ->write_xxx() functions

9 years agoMerge tag 'nfs-for-4.3-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Linus Torvalds [Wed, 7 Oct 2015 07:54:22 +0000 (08:54 +0100)]
Merge tag 'nfs-for-4.3-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client bugfixes from Trond Myklebust:
 "Highlights include:

  Bugfixes:
   - Fix a use-after-free bug in the RPC/RDMA client
   - Fix a write performance regression
   - Fix up page writeback accounting
   - Don't try to reclaim unused state owners
   - Fix a NFSv4 nograce recovery hang
   - reset states to use open_stateid when returning delegation
     voluntarily
   - Fix a tracepoint NULL-pointer dereference"

* tag 'nfs-for-4.3-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  NFS: Fix a tracepoint NULL-pointer dereference
  nfs4: reset states to use open_stateid when returning delegation voluntarily
  NFSv4: Fix a nograce recovery hang
  NFSv4.1: nfs4_opendata_check_deleg needs to handle NFS4_OPEN_CLAIM_DELEG_CUR_FH
  NFSv4: Don't try to reclaim unused state owners
  NFS: Fix a write performance regression
  NFS: Fix up page writeback accounting
  xprtrdma: disconnect and flush cqs before freeing buffers

9 years agoRevert "fs: do not prefault sys_write() user buffer pages"
Linus Torvalds [Wed, 7 Oct 2015 07:32:38 +0000 (08:32 +0100)]
Revert "fs: do not prefault sys_write() user buffer pages"

This reverts commit 998ef75ddb5709bbea0bf1506cd2717348a3c647.

The commit itself does not appear to be buggy per se, but it is exposing
a bug in ext4 (and Ted thinks ext3 too, but we solved that by getting
rid of it).  It's too late in the release cycle to really worry about
this, even if Dave Hansen has a patch that may actually fix the
underlying ext4 problem.  We can (and should) revisit this for the next
release.

The problem is that moving the prefaulting later now exposes a special
case with partially successful writes that isn't handled correctly.  And
the prefaulting likely isn't normally even that much of a performance
issue - it looks like at least one reason Dave saw this in his
performance tests is that he also ran them on Skylake that now supports
the new SMAP code, which makes the normally very cheap user space
prefaulting noticeably more expensive.

Bisected-and-acked-by: Ted Ts'o <tytso@mit.edu>
Analyzed-and-acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoNFS: Fix a tracepoint NULL-pointer dereference
Anna Schumaker [Mon, 5 Oct 2015 20:43:26 +0000 (16:43 -0400)]
NFS: Fix a tracepoint NULL-pointer dereference

Running xfstest generic/013 with the tracepoint nfs:nfs4_open_file
enabled produces a NULL-pointer dereference when calculating fileid and
filehandle of the opened file.  Fix this by checking if state is NULL
before trying to use the inode pointer.

Reported-by: Olga Kornievskaia <aglo@umich.edu>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
9 years agostrscpy: zero any trailing garbage bytes in the destination
Chris Metcalf [Tue, 6 Oct 2015 16:37:41 +0000 (12:37 -0400)]
strscpy: zero any trailing garbage bytes in the destination

It's possible that the destination can be shadowed in userspace
(as, for example, the perf buffers are now).  So we should take
care not to leak data that could be inspected by userspace.

Signed-off-by: Chris Metcalf <cmetcalf@ezchip.com>
9 years agoword-at-a-time.h: support zero_bytemask() on alpha and tile
Chris Metcalf [Tue, 6 Oct 2015 18:20:45 +0000 (14:20 -0400)]
word-at-a-time.h: support zero_bytemask() on alpha and tile

Both alpha and tile needed implementations of zero_bytemask.

The alpha version is untested.

Signed-off-by: Chris Metcalf <cmetcalf@ezchip.com>
9 years agoword-at-a-time.h: fix some Kbuild files
Chris Metcalf [Tue, 6 Oct 2015 17:35:10 +0000 (13:35 -0400)]
word-at-a-time.h: fix some Kbuild files

arch/tile added word-at-a-time.h after the patch that added generic-y
entries; the generic-y entry is now stale.

arch/h8300 is newer than the generic-y patch for word-at-a-time.h,
and needs a generic-y entry.

arch/powerpc seems to have gotten a generic-y entry by mistake in
the first patch; this change removes it.

Signed-off-by: Chris Metcalf <cmetcalf@ezchip.com>
9 years agoarm64: replace read_lock to rcu lock in call_break_hook
Yang Shi [Mon, 5 Oct 2015 21:32:51 +0000 (14:32 -0700)]
arm64: replace read_lock to rcu lock in call_break_hook

BUG: sleeping function called from invalid context at kernel/locking/rtmutex.c:917
in_atomic(): 0, irqs_disabled(): 128, pid: 342, name: perf
1 lock held by perf/342:
 #0:  (break_hook_lock){+.+...}, at: [<ffffffc0000851ac>] call_break_hook+0x34/0xd0
irq event stamp: 62224
hardirqs last  enabled at (62223): [<ffffffc00010b7bc>] __call_rcu.constprop.59+0x104/0x270
hardirqs last disabled at (62224): [<ffffffc0000fbe20>] vprintk_emit+0x68/0x640
softirqs last  enabled at (0): [<ffffffc000097928>] copy_process.part.8+0x428/0x17f8
softirqs last disabled at (0): [<          (null)>]           (null)
CPU: 0 PID: 342 Comm: perf Not tainted 4.1.6-rt5 #4
Hardware name: linux,dummy-virt (DT)
Call trace:
[<ffffffc000089968>] dump_backtrace+0x0/0x128
[<ffffffc000089ab0>] show_stack+0x20/0x30
[<ffffffc0007030d0>] dump_stack+0x7c/0xa0
[<ffffffc0000c878c>] ___might_sleep+0x174/0x260
[<ffffffc000708ac8>] __rt_spin_lock+0x28/0x40
[<ffffffc000708db0>] rt_read_lock+0x60/0x80
[<ffffffc0000851a8>] call_break_hook+0x30/0xd0
[<ffffffc000085a70>] brk_handler+0x30/0x98
[<ffffffc000082248>] do_debug_exception+0x50/0xb8
Exception stack(0xffffffc00514fe30 to 0xffffffc00514ff50)
fe20:                                     00000000 00000000 c1594680 0000007f
fe40: ffffffff ffffffff 92063940 0000007f 0550dcd8 ffffffc0 00000000 00000000
fe60: 0514fe70 ffffffc0 000be1f8 ffffffc0 0514feb0 ffffffc0 0008948c ffffffc0
fe80: 00000004 00000000 0514fed0 ffffffc0 ffffffff ffffffff 9282a948 0000007f
fea0: 00000000 00000000 9282b708 0000007f c1592820 0000007f 00083914 ffffffc0
fec0: 00000000 00000000 00000010 00000000 00000064 00000000 00000001 00000000
fee0: 005101e0 00000000 c1594680 0000007f c1594740 0000007f ffffffd8 ffffff80
ff00: 00000000 00000000 00000000 00000000 c1594770 0000007f c1594770 0000007f
ff20: 00665e10 00000000 7f7f7f7f 7f7f7f7f 01010101 01010101 00000000 00000000
ff40: 928e4cc0 0000007f 91ff11e8 0000007f

call_break_hook is called in atomic context (hard irq disabled), so replace
the sleepable lock to rcu lock, replace relevant list operations to rcu
version and call synchronize_rcu() in unregister_break_hook().

And, replace write lock to spinlock in {un}register_break_hook.

Signed-off-by: Yang Shi <yang.shi@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
9 years agoarm64: Don't relocate non-existent initrd
Mark Rutland [Tue, 6 Oct 2015 17:24:37 +0000 (18:24 +0100)]
arm64: Don't relocate non-existent initrd

When booting a kernel without an initrd, the kernel reports that it
moves -1 bytes worth, having gone through the motions with initrd_start
equal to initrd_end:

    Moving initrd from [4080000000-407fffffff] to [9fff49000-9fff48fff]

Prevent this by bailing out early when the initrd size is zero (i.e. we
have no initrd), avoiding the confusing message and other associated
work.

Fixes: 1570f0d7ab425c1e ("arm64: support initrd outside kernel linear map")
Cc: Mark Salter <msalter@redhat.com>
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
9 years agosched/core: Remove a parameter in the migrate_task_rq() function
xiaofeng.yan [Wed, 23 Sep 2015 06:55:59 +0000 (14:55 +0800)]
sched/core: Remove a parameter in the migrate_task_rq() function

The parameter "int next_cpu" in the following function is unused:

  migrate_task_rq(struct task_struct *p, int next_cpu)

Remove it.

Signed-off-by: xiaofeng.yan <yanxiaofeng@inspur.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Link: http://lkml.kernel.org/r/1442991360-31945-1-git-send-email-yanxiaofeng@inspur.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: Drop unlikely behind BUG_ON()
Geliang Tang [Mon, 5 Oct 2015 13:26:05 +0000 (21:26 +0800)]
sched/core: Drop unlikely behind BUG_ON()

(1) For !CONFIG_BUG cases, the bug call is a no-op, so we couldn't care
    less and the change is ok.

(2) PPC and MIPS, which HAVE_ARCH_BUG_ON, do not rely on branch predictions
    as it seems to be pointless [1] and thus callers should not be trying to
    push an optimization in the first place.

(3) For CONFIG_BUG and !HAVE_ARCH_BUG_ON cases, BUG_ON() contains an
    unlikely compiler flag already.

Hence, we can drop unlikely behind BUG_ON().

  [1] http://lkml.iu.edu/hypermail/linux/kernel/1101.3/02289.html

Signed-off-by: Geliang Tang <geliangtang@163.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Link: http://lkml.kernel.org/r/6fa7125979f98bbeac26e268271769b6ca935c8d.1444051018.git.geliangtang@163.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: Fix task and run queue sched_info::run_delay inconsistencies
Peter Zijlstra [Wed, 30 Sep 2015 15:44:13 +0000 (17:44 +0200)]
sched/core: Fix task and run queue sched_info::run_delay inconsistencies

Mike Meyer reported the following bug:

> During evaluation of some performance data, it was discovered thread
> and run queue run_delay accounting data was inconsistent with the other
> accounting data that was collected.  Further investigation found under
> certain circumstances execution time was leaking into the task and
> run queue accounting of run_delay.
>
> Consider the following sequence:
>
>     a. thread is running.
>     b. thread moves beween cgroups, changes scheduling class or priority.
>     c. thread sleeps OR
>     d. thread involuntarily gives up cpu.
>
> a. implies:
>
>     thread->sched_info.last_queued = 0
>
> a. and b. results in the following:
>
>     1. dequeue_task(rq, thread)
>
>            sched_info_dequeued(rq, thread)
>                delta = 0
>
>                sched_info_reset_dequeued(thread)
>                    thread->sched_info.last_queued = 0
>
>                thread->sched_info.run_delay += delta
>
>     2. enqueue_task(rq, thread)
>
>            sched_info_queued(rq, thread)
>
>                /* thread is still on cpu at this point. */
>                thread->sched_info.last_queued = task_rq(thread)->clock;
>
> c. results in:
>
>     dequeue_task(rq, thread)
>
>         sched_info_dequeued(rq, thread)
>
>             /* delta is execution time not run_delay. */
>             delta = task_rq(thread)->clock - thread->sched_info.last_queued
>
>         sched_info_reset_dequeued(thread)
>             thread->sched_info.last_queued = 0
>
>         thread->sched_info.run_delay += delta
>
>     Since thread was running between enqueue_task(rq, thread) and
>     dequeue_task(rq, thread), the delta above is really execution
>     time and not run_delay.
>
> d. results in:
>
>     __sched_info_switch(thread, next_thread)
>
>         sched_info_depart(rq, thread)
>
>             sched_info_queued(rq, thread)
>
>                 /* last_queued not updated due to being non-zero */
>                 return
>
>     Since thread was running between enqueue_task(rq, thread) and
>     __sched_info_switch(thread, next_thread), the execution time
>     between enqueue_task(rq, thread) and
>     __sched_info_switch(thread, next_thread) now will become
>     associated with run_delay due to when last_queued was last updated.
>

This alternative patch solves the problem by not calling
sched_info_{de,}queued() in {de,en}queue_task(). Therefore the
sched_info state is preserved and things work as expected.

By inlining the {de,en}queue_task() functions the new condition
becomes (mostly) a compile-time constant and we'll not emit any new
branch instructions.

It even shrinks the code (due to inlining {en,de}queue_task()):

$ size defconfig-build/kernel/sched/core.o defconfig-build/kernel/sched/core.o.orig
   text    data     bss     dec     hex filename
  64019   23378    2344   89741   15e8d defconfig-build/kernel/sched/core.o
  64149   23378    2344   89871   15f0f defconfig-build/kernel/sched/core.o.orig

Reported-by: Mike Meyer <Mike.Meyer@Teradata.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Link: http://lkml.kernel.org/r/20150930154413.GO3604@twins.programming.kicks-ass.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/numa: Fix task_tick_fair() from disabling numa_balancing
Srikar Dronamraju [Fri, 2 Oct 2015 02:18:25 +0000 (07:48 +0530)]
sched/numa: Fix task_tick_fair() from disabling numa_balancing

If static branch 'sched_numa_balancing' is enabled, it should kickstart
NUMA balancing through task_tick_numa(). However the following commit:

  2a595721a1fa ("sched/numa: Convert sched_numa_balancing to a static_branch")

erroneously disables this.

Fix this anomaly by enabling task_tick_numa() when the static branch
'sched_numa_balancing' is enabled.

Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Link: http://lkml.kernel.org/r/1443752305-27413-1-git-send-email-srikar@linux.vnet.ibm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: Add preempt_count invariant check
Peter Zijlstra [Tue, 29 Sep 2015 10:18:46 +0000 (12:18 +0200)]
sched/core: Add preempt_count invariant check

Ingo requested I keep my debug check for the preempt_count invariant.

Requested-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: More notrace annotations
Peter Zijlstra [Mon, 28 Sep 2015 16:52:36 +0000 (18:52 +0200)]
sched/core: More notrace annotations

preempt_schedule_common() is marked notrace, but it does not use
_notrace() preempt_count functions and __schedule() is also not marked
notrace, which means that its perfectly possible to end up in the
tracer from preempt_schedule_common().

Steve says:

  | Yep, there's some history to this. This was originally the issue that
  | caused function tracing to go into infinite recursion. But now we have
  | preempt_schedule_notrace(), which is used by the function tracer, and
  | that function must not be traced till preemption is disabled.
  |
  | Now if function tracing is running and we take an interrupt when
  | NEED_RESCHED is set, it calls
  |
  |   preempt_schedule_common() (not traced)
  |
  | But then that calls preempt_disable() (traced)
  |
  | function tracer calls preempt_disable_notrace() followed by
  | preempt_enable_notrace() which will see NEED_RESCHED set, and it will
  | call preempt_schedule_notrace(), which stops the recursion, but
  | still calls __schedule() here, and that means when we return, we call
  | the __schedule() from preempt_schedule_common().
  |
  | That said, I prefer this patch. Preemption is disabled before calling
  | __schedule(), and we get rid of a one round recursion with the
  | scheduler.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: Kill PREEMPT_ACTIVE
Peter Zijlstra [Mon, 28 Sep 2015 16:11:35 +0000 (18:11 +0200)]
sched/core: Kill PREEMPT_ACTIVE

Its unused, kill the definition.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
Reviewed-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core, sched/x86: Kill thread_info::saved_preempt_count
Peter Zijlstra [Mon, 28 Sep 2015 16:11:18 +0000 (18:11 +0200)]
sched/core, sched/x86: Kill thread_info::saved_preempt_count

With the introduction of the context switch preempt_count invariant,
and the demise of PREEMPT_ACTIVE, its pointless to save/restore the
per-cpu preemption count, it must always be 2.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: Simplify preempt_count tests
Peter Zijlstra [Mon, 28 Sep 2015 16:11:45 +0000 (18:11 +0200)]
sched/core: Simplify preempt_count tests

Since we stopped setting PREEMPT_ACTIVE, there is no need to mask it
out of preempt_count() tests.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
Reviewed-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: Robustify preemption leak checks
Peter Zijlstra [Mon, 28 Sep 2015 15:57:39 +0000 (17:57 +0200)]
sched/core: Robustify preemption leak checks

When we warn about a preempt_count leak; reset the preempt_count to
the known good value such that the problem does not ripple forward.

This is most important on x86 which has a per cpu preempt_count that is
not saved/restored (after this series). So if you schedule with an
invalid (!2*PREEMPT_DISABLE_OFFSET) preempt_count the next task is
messed up too.

Enforcing this invariant limits the borkage to just the one task.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Frederic Weisbecker <fweisbec@gmail.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: Stop setting PREEMPT_ACTIVE
Peter Zijlstra [Mon, 28 Sep 2015 16:09:19 +0000 (18:09 +0200)]
sched/core: Stop setting PREEMPT_ACTIVE

Now that nothing tests for PREEMPT_ACTIVE anymore, stop setting it.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
Reviewed-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: Fix trace_sched_switch()
Peter Zijlstra [Mon, 28 Sep 2015 16:06:56 +0000 (18:06 +0200)]
sched/core: Fix trace_sched_switch()

__trace_sched_switch_state() is the last remaining PREEMPT_ACTIVE
user, move trace_sched_switch() from prepare_task_switch() to
__schedule() and propagate the @preempt argument.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: Add preempt argument to __schedule()
Peter Zijlstra [Mon, 28 Sep 2015 16:05:34 +0000 (18:05 +0200)]
sched/core: Add preempt argument to __schedule()

There is only a single PREEMPT_ACTIVE use in the regular __schedule()
path and that is to circumvent the task->state check. Since the code
setting PREEMPT_ACTIVE is the immediate caller of __schedule() we can
replace this with a function argument.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Frederic Weisbecker <fweisbec@gmail.com>
Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: Create preempt_count invariant
Peter Zijlstra [Mon, 28 Sep 2015 15:52:18 +0000 (17:52 +0200)]
sched/core: Create preempt_count invariant

Assuming units of PREEMPT_DISABLE_OFFSET for preempt_count() numbers.

Now that TASK_DEAD no longer results in preempt_count() == 3 during
scheduling, we will always call context_switch() with preempt_count()
== 2.

However, we don't always end up with preempt_count() == 2 in
finish_task_switch() because new tasks get created with
preempt_count() == 1.

Create FORK_PREEMPT_COUNT and set it to 2 and use that in the right
places. Note that we cannot use INIT_PREEMPT_COUNT as that serves
another purpose (boot).

After this, preempt_count() is invariant across the context switch,
with exception of PREEMPT_ACTIVE.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: Rework TASK_DEAD preemption exception
Peter Zijlstra [Mon, 28 Sep 2015 16:02:03 +0000 (18:02 +0200)]
sched/core: Rework TASK_DEAD preemption exception

TASK_DEAD is special in that the final schedule call from do_exit()
must be done with preemption disabled.

This means we end up scheduling with a preempt_count() higher than
usual (3 instead of the 'expected' 2).

Since future patches will want to rely on an invariant
preempt_count() value during schedule, fix this up.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Frederic Weisbecker <fweisbec@gmail.com>
Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: Simplify INIT_PREEMPT_COUNT
Peter Zijlstra [Mon, 28 Sep 2015 15:45:40 +0000 (17:45 +0200)]
sched/core: Simplify INIT_PREEMPT_COUNT

As per the following commit:

  d86ee4809d03 ("sched: optimize cond_resched()")

we need PREEMPT_ACTIVE to avoid cond_resched() from working before
the scheduler is set up.

However, keeping preemption disabled should do the same thing
already, making the PREEMPT_ACTIVE part entirely redundant.

The only complication is !PREEMPT_COUNT kernels, where
PREEMPT_DISABLED ends up being 0. Instead we use an unconditional
PREEMPT_OFFSET to set preempt_count() even on !PREEMPT_COUNT
kernels.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agoMerge branch 'sched/urgent' into sched/core, to pick up fixes before applying new...
Ingo Molnar [Tue, 6 Oct 2015 15:05:36 +0000 (17:05 +0200)]
Merge branch 'sched/urgent' into sched/core, to pick up fixes before applying new changes

Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agosched/core: Fix TASK_DEAD race in finish_task_switch()
Peter Zijlstra [Tue, 29 Sep 2015 12:45:09 +0000 (14:45 +0200)]
sched/core: Fix TASK_DEAD race in finish_task_switch()

So the problem this patch is trying to address is as follows:

        CPU0                            CPU1

        context_switch(A, B)
                                        ttwu(A)
                                          LOCK A->pi_lock
                                          A->on_cpu == 0
        finish_task_switch(A)
          prev_state = A->state  <-.
          WMB                      |
          A->on_cpu = 0;           |
          UNLOCK rq0->lock         |
                                   |    context_switch(C, A)
                                   `--  A->state = TASK_DEAD
          prev_state == TASK_DEAD
            put_task_struct(A)
                                        context_switch(A, C)
                                        finish_task_switch(A)
                                          A->state == TASK_DEAD
                                            put_task_struct(A)

The argument being that the WMB will allow the load of A->state on CPU0
to cross over and observe CPU1's store of A->state, which will then
result in a double-drop and use-after-free.

Now the comment states (and this was true once upon a long time ago)
that we need to observe A->state while holding rq->lock because that
will order us against the wakeup; however the wakeup will not in fact
acquire (that) rq->lock; it takes A->pi_lock these days.

We can obviously fix this by upgrading the WMB to an MB, but that is
expensive, so we'd rather avoid that.

The alternative this patch takes is: smp_store_release(&A->on_cpu, 0),
which avoids the MB on some archs, but not important ones like ARM.

Reported-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: <stable@vger.kernel.org> # v3.1+
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Cc: manfred@colorfullife.com
Cc: will.deacon@arm.com
Fixes: e4a52bcb9a18 ("sched: Remove rq->lock from the first half of ttwu()")
Link: http://lkml.kernel.org/r/20150929124509.GG3816@twins.programming.kicks-ass.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agodm: fix request-based dm error reporting
Junichi Nomura [Tue, 6 Oct 2015 04:19:54 +0000 (04:19 +0000)]
dm: fix request-based dm error reporting

end_clone_bio() is a endio callback for clone bio and should check
and save the clone's bi_error for error reporting.  However,
4246a0b63bd8 ("block: add a bi_error field to struct bio") changed
the function to check the original bio's bi_error, which is 0.

Without this fix, clone's error is ignored and reported to the
original request as success.  Thus data corruption will be observed.

Fixes: 4246a0b63bd8 ("block: add a bi_error field to struct bio")
Signed-off-by: Jun'ichi Nomura <j-nomura@ce.jp.nec.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
9 years agoMerge tag 'for-linus-4.3b-rc4-tag' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 6 Oct 2015 14:05:02 +0000 (15:05 +0100)]
Merge tag 'for-linus-4.3b-rc4-tag' of git://git./linux/kernel/git/xen/tip

Pull xen bug fixes from David Vrabel:

 - Fix VM save performance regression with x86 PV guests

 - Make kexec work in x86 PVHVM guests (if Xen has the soft-reset ABI)

 - Other minor fixes.

* tag 'for-linus-4.3b-rc4-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
  x86/xen/p2m: hint at the last populated P2M entry
  x86/xen: Do not clip xen_e820_map to xen_e820_map_entries when sanitizing map
  x86/xen: Support kexec/kdump in HVM guests by doing a soft reset
  xen/x86: Don't try to write syscall-related MSRs for PV guests
  xen: use correct type for HYPERVISOR_memory_op()

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Linus Torvalds [Tue, 6 Oct 2015 13:59:36 +0000 (14:59 +0100)]
Merge branch 'for-linus' of git://git./linux/kernel/git/s390/linux

Pull s390 fixes from Martin Schwidefsky:
 "Three bug fixes and an update to the default configuration"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/defconfig: set SCSI_DH=y
  s390/vtime: correct scaled cputime of partially idle CPUs
  s390/boot/decompression: disable floating point in decompressor
  s390/numa: use correct type for node_to_cpumask_map

9 years agoBTRFS: support NFSv2 export
NeilBrown [Fri, 8 May 2015 00:16:23 +0000 (10:16 +1000)]
BTRFS: support NFSv2 export

The "fh_len" passed to ->fh_to_* is not guaranteed to be that same as
that returned by encode_fh - it may be larger.

With NFSv2, the filehandle is fixed length, so it may appear longer
than expected and be zero-padded.

So we must test that fh_len is at least some value, not exactly equal
to it.

Signed-off-by: NeilBrown <neilb@suse.de>
Acked-by: David Sterba <dsterba@suse.cz>