Merge branch 'akpm' (Andrew's incoming - part two)

Says Andrew:

 "60 patches.  That's good enough for -rc1 I guess.  I have quite a lot
  of detritus to be rechecked, work through maintainers, etc.

 - most of the remains of MM
 - rtc
 - various misc
 - cgroups
 - memcg
 - cpusets
 - procfs
 - ipc
 - rapidio
 - sysctl
 - pps
 - w1
 - drivers/misc
 - aio"

* akpm: (60 commits)
  memcg: replace ss->id_lock with a rwlock
  aio: allocate kiocbs in batches
  drivers/misc/vmw_balloon.c: fix typo in code comment
  drivers/misc/vmw_balloon.c: determine page allocation flag can_sleep outside loop
  w1: disable irqs in critical section
  drivers/w1/w1_int.c: multiple masters used same init_name
  drivers/power/ds2780_battery.c: fix deadlock upon insertion and removal
  drivers/power/ds2780_battery.c: add a nolock function to w1 interface
  drivers/power/ds2780_battery.c: create central point for calling w1 interface
  w1: ds2760 and ds2780, use ida for id and ida_simple_get() to get it
  pps gpio client: add missing dependency
  pps: new client driver using GPIO
  pps: default echo function
  include/linux/dma-mapping.h: add dma_zalloc_coherent()
  sysctl: make CONFIG_SYSCTL_SYSCALL default to n
  sysctl: add support for poll()
  RapidIO: documentation update
  drivers/net/rionet.c: fix ethernet address macros for LE platforms
  RapidIO: fix potential null deref in rio_setup_device()
  RapidIO: add mport driver for Tsi721 bridge
  ...

memcg: replace ss->id_lock with a rwlock

While back-porting Johannes Weiner's patch "mm: memcg-aware global
reclaim" for an internal effort, we noticed a significant performance
regression during page-reclaim heavy workloads due to high contention of
the ss->id_lock.  This lock protects idr map, and serializes calls to
idr_get_next() in css_get_next() (which is used during the memcg hierarchy
walk).

Since idr_get_next() is just doing a look up, we need only serialize it
with respect to idr_remove()/idr_get_new().  By making the ss->id_lock a
rwlock, contention is greatly reduced and performance improves.

Tested: cat a 256m file from a ramdisk in a 128m container 50 times on
each core (one file + container per core) in parallel on a NUMA machine.
Result is the time for the test to complete in 1 of the containers.
Both kernels included Johannes' memcg-aware global reclaim patches.

Before rwlock patch: 1710.778s
After rwlock patch: 152.227s

Signed-off-by: Andrew Bresticker <abrestic@google.com>
Cc: Paul Menage <menage@gmail.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

aio: allocate kiocbs in batches

In testing aio on a fast storage device, I found that the context lock
takes up a fair amount of cpu time in the I/O submission path.  The reason
is that we take it for every I/O submitted (see __aio_get_req).  Since we
know how many I/Os are passed to io_submit, we can preallocate the kiocbs
in batches, reducing the number of times we take and release the lock.

In my testing, I was able to reduce the amount of time spent in
_raw_spin_lock_irq by .56% (average of 3 runs).  The command I used to
test this was:

   aio-stress -O -o 2 -o 3 -r 8 -d 128 -b 32 -i 32 -s 16384 <dev>

I also tested the patch with various numbers of events passed to
io_submit, and I ran the xfstests aio group of tests to ensure I didn't
break anything.

Signed-off-by: Jeff Moyer <jmoyer@redhat.com>
Cc: Daniel Ehrenberg <dehrenberg@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/misc/vmw_balloon.c: fix typo in code comment

Fix typo in code comment.

Signed-off-by: Rakib Mullick <rakib.mullick@gmail.com>
Acked-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/misc/vmw_balloon.c: determine page allocation flag can_sleep outside loop

In vmballoon_reserve_page(), flags has been passed from the callee
function (vmballoon_inflate here).  So, we can determine can_sleep outside
the loop.

Signed-off-by: Rakib Mullick <rakib.mullick@gmail.com>
Acked-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

w1: disable irqs in critical section

Interrupting w1_delay() in w1_read_bit() results in missing the low level
on the w1 line and receiving "1" instead of "0".

Add local_irq_save()/local_irq_restore() around the critical section

Signed-off-by: Jan Weitzel <j.weitzel@phytec.de>
Acked-by: Evgeniy Polyakov <zbr@ioremap.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/w1/w1_int.c: multiple masters used same init_name

When using multiple masters, w1_int.c would use the .init_name from w1.c
for all entities, which will fail when creating a corresponding sysfs
entry.  This patch uses the unique name previously generated.

  WARNING: at fs/sysfs/dir.c:451 sysfs_add_one+0x48/0x64()
  sysfs: cannot create duplicate filename '/devices/w1 bus master'
  Modules linked in:
  Call trace:
   [<9001a604>] warn_slowpath_common+0x34/0x44
   [<9001a64c>] warn_slowpath_fmt+0x14/0x18
   [<90078020>] sysfs_add_one+0x48/0x64
   [<900784ec>] create_dir+0x40/0x68
   [<9007857a>] sysfs_create_dir+0x66/0x78
   [<900c1a8a>] kobject_add_internal+0x6e/0x104
   [<900c1bc0>] kobject_add_varg+0x20/0x2c
   [<900c1c1c>] kobject_add+0x30/0x3c
   [<900dbd66>] device_add+0x6a/0x378
   [<900dbb4a>] device_initialize+0x12/0x48
   [<900dc080>] device_register+0xc/0x10
   [<900f99be>] w1_add_master_device+0x162/0x274
   [<90008e7a>] w1_gpio_probe+0x66/0xb4
   [<9000030c>] kernel_init+0x0/0xe8
   [<900dde54>] platform_drv_probe+0xc/0xe
   [<9000030c>] kernel_init+0x0/0xe8
   [<900dd4f8>] driver_probe_device+0x6c/0xdc
   [<900dd5fc>] __driver_attach+0x34/0x48
   [<900dcce8>] bus_for_each_dev+0x2c/0x48
   [<900dd5c8>] __driver_attach+0x0/0x48
   [<900dd38c>] driver_attach+0x10/0x14
   [<900dd16a>] bus_add_driver+0x6a/0x18c
   [<900dd768>] driver_register+0x60/0xb8
   [<90011594>] __initcall_w1_therm_init6+0x0/0x4
   [<90008e00>] w1_gpio_init+0x0/0x14
   [<9000030c>] kernel_init+0x0/0xe8
   [<900ddf48>] platform_driver_register+0x30/0x38
   [<90011594>] __initcall_w1_therm_init6+0x0/0x4
   [<90008e00>] w1_gpio_init+0x0/0x14
   [<9000030c>] kernel_init+0x0/0xe8
   [<900ddf5e>] platform_driver_probe+0xe/0x3c
   [<90008e0c>] w1_gpio_init+0xc/0x14
   [<90011594>] __initcall_w1_therm_init6+0x0/0x4
   [<90008e00>] w1_gpio_init+0x0/0x14
   [<900126d4>] do_one_initcall+0x34/0x130
   [<90000372>] kernel_init+0x66/0xe8
   [<90011594>] __initcall_w1_therm_init6+0x0/0x4
   [<9001ca3e>] do_exit+0x0/0x3a6
   [<9000030c>] kernel_init+0x0/0xe8
   [<9001ca3e>] do_exit+0x0/0x3a6

  ---[ end trace 5a9233884fead918 ]---
  kobject_add_internal failed for w1 bus master with -EEXIST, don't try to register things with the same name in the same directory.

Signed-off-by: Florian Faber <faber@faberman.de>
Cc: Evgeniy Polyakov <zbr@ioremap.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/power/ds2780_battery.c: fix deadlock upon insertion and removal

Fixes the deadlock when inserting and removing the ds2780.

Signed-off-by: Clifton Barnes <cabarnes@indesign-llc.com>
Cc: Evgeniy Polyakov <zbr@ioremap.net>
Cc: <stable@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/power/ds2780_battery.c: add a nolock function to w1 interface

Adds a nolock function to the w1 interface to avoid locking the
mutex if needed.

Signed-off-by: Clifton Barnes <cabarnes@indesign-llc.com>
Cc: Evgeniy Polyakov <zbr@ioremap.net>
Cc: <stable@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/power/ds2780_battery.c: create central point for calling w1 interface

Simply creates one point to call the w1 interface.

Signed-off-by: Clifton Barnes <cabarnes@indesign-llc.com>
Cc: Evgeniy Polyakov <zbr@ioremap.net>
Cc: <stable@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

w1: ds2760 and ds2780, use ida for id and ida_simple_get() to get it

Straightforward.  As an aside, the ida_init calls are not needed as far as
I can see needed.  (DEFINE_IDA does the same already).

Signed-off-by: Jonathan Cameron <jic23@cam.ac.uk>
Cc: Evgeniy Polyakov <zbr@ioremap.net>
Acked-by: Clifton Barnes <cabarnes@indesign-llc.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pps gpio client: add missing dependency

Add "depends on GENERIC_HARDIRQS" to avoid compile breakage on s390:

drivers/built-in.o: In function `pps_gpio_remove':
linux-next/drivers/pps/clients/pps-gpio.c:189: undefined reference to `free_irq'

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: James Nuss <jamesnuss@nanometrics.ca>
Cc: Rodolfo Giometti <giometti@enneenne.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pps: new client driver using GPIO

This client driver allows you to use a GPIO pin as a source for PPS
signals.  Platform data [1] are used to specify the GPIO pin number,
label, assert event edge type, and whether clear events are captured.

This driver is based on the work by Ricardo Martins who submitted an
initial implementation [2] of a PPS IRQ client driver to the linuxpps
mailing-list on Dec 3 2010.

[1] include/linux/pps-gpio.h
[2] http://ml.enneenne.com/pipermail/linuxpps/2010-December/004155.html

[akpm@linux-foundation.org: remove unneeded cast of void*]
Signed-off-by: James Nuss <jamesnuss@nanometrics.ca>
Cc: Ricardo Martins <rasm@fe.up.pt>
Acked-by: Rodolfo Giometti <giometti@linux.it>
Signed-off-by: Ricardo Martins <rasm@fe.up.pt>
Cc: Alexander Gordeev <lasaine@lvk.cs.msu.su>
Cc: Igor Plyatov <plyatov@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

pps: default echo function

A default echo function has been provided so it is no longer an error when
you specify PPS_ECHOASSERT or PPS_ECHOCLEAR without an explicit echo
function.  This allows some code re-use and also makes it easier to write
client drivers since the default echo function does not normally need to
change.

Signed-off-by: James Nuss <jamesnuss@nanometrics.ca>
Reviewed-by: Ben Gardiner <bengardiner@nanometrics.ca>
Acked-by: Rodolfo Giometti <giometti@linux.it>
Cc: Ricardo Martins <rasm@fe.up.pt>
Cc: Alexander Gordeev <lasaine@lvk.cs.msu.su>
Cc: Igor Plyatov <plyatov@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

include/linux/dma-mapping.h: add dma_zalloc_coherent()

Lots of driver code does a dma_alloc_coherent() and then zeroes out the
memory with a memset.  Make it easy for them.

Cc: Alexandre Bounine <alexandre.bounine@idt.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

sysctl: make CONFIG_SYSCTL_SYSCALL default to n

When I tried to send a patch to remove it, Andi told me we still need to
keep compabitlies for old libc, so we can't remove this completely.  Then
just make it default to n and remove the doc from
feature-removal-schedule.txt.

Signed-off-by: WANG Cong <amwang@redhat.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

sysctl: add support for poll()

Adding support for poll() in sysctl fs allows userspace to receive
notifications of changes in sysctl entries.  This adds a infrastructure to
allow files in sysctl fs to be pollable and implements it for hostname and
domainname.

[akpm@linux-foundation.org: s/declare/define/ for definitions]
Signed-off-by: Lucas De Marchi <lucas.demarchi@profusion.mobi>
Cc: Greg KH <gregkh@suse.de>
Cc: Kay Sievers <kay.sievers@vrfy.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

RapidIO: documentation update

Update rapidio.txt to reflect changes from recent patch.
See http://marc.info/?l=linux-kernel&m=131285620113589&w=2 for details.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Liu Gang <Gang.Liu@freescale.com>
Cc: Micha Nelissen <micha@neli.hopto.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/net/rionet.c: fix ethernet address macros for LE platforms

Modify Ethernet addess macros to be compatible with BE/LE platforms

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Chul Kim <chul.kim@idt.com>
Cc: Kumar Gala <galak@kernel.crashing.org>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Li Yang <leoli@freescale.com>
Cc: <stable@kernel.org> [2.6.39+]
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

RapidIO: fix potential null deref in rio_setup_device()

The "goto cleanup" path can deference "rswitch" when it is NULL.

Reported-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Dan Carpenter <error27@gmail.com>
Cc: Kumar Gala <galak@kernel.crashing.org>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Chul Kim <chul.kim@idt.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

RapidIO: add mport driver for Tsi721 bridge

Add RapidIO mport driver for IDT TSI721 PCI Express-to-SRIO bridge device.
 The driver provides full set of callback functions defined for mport
devices in RapidIO subsystem.  It also is compatible with current version
of RIONET driver (Ethernet over RapidIO messaging services).

This patch is applicable to kernel versions starting from 2.6.39.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Signed-off-by: Chul Kim <chul.kim@idt.com>
Cc: Kumar Gala <galak@kernel.crashing.org>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Li Yang <leoli@freescale.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

arch/powerpc/sysdev/fsl_rio.c: release rapidio port I/O region resource if port failed to initialize

The "struct rio_mport" contains a member of master port I/O memory
resource structure "struct resource iores".  This resource will be read
from device tree and be used for rapidio R/W transaction memory space.
Rapidio requests the port I/O memory resource under the root resource
"iomem_resource".

struct rio_mport *port;
port = kzalloc(sizeof(struct rio_mport), GFP_KERNEL);

request_resource(&iomem_resource, &port->iores);

When port failed to initialize, allocated "rio_mport" structure memory
will be freed, and the port I/O memory resource structure pointer
"&port->iores" will be invalid.  If other requests resource under
"iomem_resource", "&port->iores" node may be operated in the child
resources list and this will cause the system to crash.

So the requested port I/O memory resource should be released before
freeing allocated "rio_mport" structure.

Signed-off-by: Liu Gang <Gang.Liu@freescale.com>
Acked-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/rapidio/rio-scan.c: use discovered bit to test if enumeration is complete

The discovered bit in PGCCSR register indicates if the device has been
discovered by system host.  In Rapidio systems, some agent devices can also
be master devices.  They can issue requests into the system.

Signed-off-by: Liu Gang <Gang.Liu@freescale.com>
Acked-by: Alexandre Bounine <alexandre.bounine@idt.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

init: add root=PARTUUID=UUID/PARTNROFF=%d support

Expand root=PARTUUID=UUID syntax to support selecting a root partition by
integer offset from a known, unique partition.  This approach provides
similar properties to specifying a device and partition number, but using
the UUID as the unique path prior to evaluating the offset.

For example,
  root=PARTUUID=99DE9194-FC15-4223-9192-FC243948F88B/PARTNROFF=1
selects the partition with UUID 99DE.. then select the next
partition.

This change is motivated by a particular usecase in Chromium OS where the
bootloader can easily determine what partition it is on (by UUID) but
doesn't perform general partition table walking.

That said, support for this model provides a direct mechanism for the user
to modify the root partition to boot without specifically needing to
extract each UUID or update the bootloader explicitly when the root
partition UUID is changed (if it is recreated to be larger, for instance).
 Pinning to a /boot-style partition UUID allows the arbitrary root
partition reconfiguration/modifications with slightly less ambiguity than
just [dev][partition] and less stringency than the specific root partition
UUID.

[sfr@canb.auug.org.au: fix init sections warning]
Signed-off-by: Will Drewry <wad@chromium.org>
Cc: Kay Sievers <kay.sievers@vrfy.org>
Cc: Randy Dunlap <rdunlap@xenotime.net>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

include/linux/sem.h: make sysv_sem empty if SYSVIPC is disabled

For the sysvsem undo, each task struct contains a sysv_sem structure with
a pointer to the undo information.

This pointer is only necessary if sysvipc is enabled - thus the pointer
can be made conditional on CONFIG_SYSVIPC.

Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ipc/sem.c: remove private structures from public header file

include/linux/sem.h contains several structures that are only used within
ipc/sem.c.

The patch moves them into ipc/sem.c - there is no need to expose the
structures to the whole kernel.

No functional changes, only whitespace cleanups and 80-char per line
fixes.

Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ipc/sem.c: handle spurious wakeups

semtimedop() does not handle spurious wakeups, it returns -EINTR to user
space.  Most other schedule() users would just loop and not return to user
space.  The patch adds such a loop to semtimedop()

Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Reported-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ipc/sem.c: fix return code race with semop vs. semop +semctl(IPC_RMID)

sys_semtimedop() may return -EIDRM although the semaphore operation
completed successfully:

thread 1: thread 2:
semtimedop(), sleeps
semop():
* acquires sem_lock()
semtimedop() woken up due to timeout
sem_lock() loops
* notices that thread 2 could be completed.
* performs the operations that thread 2 is sleeping on.
* marks the semaphore operation as IN_WAKEUP
* drops sem_lock(), does wakeup, sets return code to 0
* thread delayed due to interrupt, whatever
* returns to user space
* thread still delayed
semctl(IPC_RMID)
* acquires sem_lock()
* ipc_rmid(), ipcp->deleted=1
* drops sem_lock()
* thread finally continues - but seem_lock()
  now fails due to ipcp->deleted == 1
* returns -EIDRM instead of 0

The fix is trivial: Always use the return code in queue.status.

In real world, the race probably doesn't matter:
If the semaphore array is destroyed, the app is probably not interested
if the last operation succeeded or was already cancelled.

Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Mike Galbraith <efault@gmx.de>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ida: make ida_simple_get/put() IRQ safe

It's often convenient to be able to release resource from IRQ context.
Make ida_simple_*() use irqsave/restore spin ops so that they are IRQ
safe.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

proc: fix races against execve() of /proc/PID/fd**

fd* files are restricted to the task's owner, and other users may not get
direct access to them.  But one may open any of these files and run any
setuid program, keeping opened file descriptors.  As there are permission
checks on open(), but not on readdir() and read(), operations on the kept
file descriptors will not be checked.  It makes it possible to violate
procfs permission model.

Reading fdinfo/* may disclosure current fds' position and flags, reading
directory contents of fdinfo/ and fd/ may disclosure the number of opened
files by the target task.  This information is not sensible per se, but it
can reveal some private information (like length of a password stored in a
file) under certain conditions.

Used existing (un)lock_trace functions to check for ptrace_may_access(),
but instead of using EPERM return code from it use EACCES to be consistent
with existing proc_pid_follow_link()/proc_pid_readlink() return code.  If
they differ, attacker can guess what fds exist by analyzing stat() return
code.  Patched handlers: stat() for fd/*, stat() and read() for fdindo/*,
readdir() and lookup() for fd/ and fdinfo/.

Signed-off-by: Vasiliy Kulikov <segoon@openwall.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: <stable@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

procfs: report EISDIR when reading sysctl dirs in proc

On reading sysctl dirs we should return -EISDIR instead of -EINVAL.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cpusets: avoid looping when storing to mems_allowed if one node remains set

{get,put}_mems_allowed() exist so that general kernel code may locklessly
access a task's set of allowable nodes without having the chance that a
concurrent write will cause the nodemask to be empty on configurations
where MAX_NUMNODES > BITS_PER_LONG.

This could incur a significant delay, however, especially in low memory
conditions because the page allocator is blocking and reclaim requires
get_mems_allowed() itself.  It is not atypical to see writes to
cpuset.mems take over 2 seconds to complete, for example.  In low memory
conditions, this is problematic because it's one of the most imporant
times to change cpuset.mems in the first place!

The only way a task's set of allowable nodes may change is through cpusets
by writing to cpuset.mems and when attaching a task to a generic code is
not reading the nodemask with get_mems_allowed() at the same time, and
then clearing all the old nodes.  This prevents the possibility that a
reader will see an empty nodemask at the same time the writer is storing a
new nodemask.

If at least one node remains unchanged, though, it's possible to simply
set all new nodes and then clear all the old nodes.  Changing a task's
nodemask is protected by cgroup_mutex so it's guaranteed that two threads
are not changing the same task's nodemask at the same time, so the
nodemask is guaranteed to be stored before another thread changes it and
determines whether a node remains set or not.

Signed-off-by: David Rientjes <rientjes@google.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Paul Menage <paul@paulmenage.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/page_cgroup.c: quiet sparse noise

warning: symbol 'swap_cgroup_ctrl' was not declared. Should it be static?

Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Paul Menage <paul@paulmenage.org>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Balbir Singh <bsingharora@gmail.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: Fix race condition in memcg_check_events() with this_cpu usage

Various code in memcontrol.c () calls this_cpu_read() on the calculations
to be done from two different percpu variables, or does an open-coded
read-modify-write on a single percpu variable.

Disable preemption throughout these operations so that the writes go to
the correct palces.

[hannes@cmpxchg.org: added this_cpu to __this_cpu conversion]
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Greg Thelen <gthelen@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: close race between charge and putback

There is a potential race between a thread charging a page and another
thread putting it back to the LRU list:

  charge:                         putback:
  SetPageCgroupUsed               SetPageLRU
  PageLRU && add to memcg LRU     PageCgroupUsed && add to memcg LRU

The order of setting one flag and checking the other is crucial, otherwise
the charge may observe !PageLRU while the putback observes !PageCgroupUsed
and the page is not linked to the memcg LRU at all.

Global memory pressure may fix this by trying to isolate and putback the
page for reclaim, where that putback would link it to the memcg LRU again.
 Without that, the memory cgroup is undeletable due to a charge whose
physical page can not be found and moved out.

Signed-off-by: Johannes Weiner <jweiner@redhat.com>
Cc: Ying Han <yinghan@google.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <bsingharora@gmail.com>
Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: skip scanning active lists based on individual size

Reclaim decides to skip scanning an active list when the corresponding
inactive list is above a certain size in comparison to leave the assumed
working set alone while there are still enough reclaim candidates around.

The memcg implementation of comparing those lists instead reports whether
the whole memcg is low on the requested type of inactive pages,
considering all nodes and zones.

This can lead to an oversized active list not being scanned because of the
state of the other lists in the memcg, as well as an active list being
scanned while its corresponding inactive list has enough pages.

Not only is this wrong, it's also a scalability hazard, because the global
memory state over all nodes and zones has to be gathered for each memcg
and zone scanned.

Make these calculations purely based on the size of the two LRU lists
that are actually affected by the outcome of the decision.

Signed-off-by: Johannes Weiner <jweiner@redhat.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <bsingharora@gmail.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Reviewed-by: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: do not expose uninitialized mem_cgroup_per_node to world

If somebody is touching data too early, it might be easier to diagnose a
problem when dereferencing NULL at mem->info.nodeinfo[node] than trying to
understand why mem_cgroup_per_zone is [un|partly]initialized.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: fix oom schedule_timeout()

Before calling schedule_timeout(), task state should be changed.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

memcg: rename mem variable to memcg

The memcg code sometimes uses "struct mem_cgroup *mem" and sometimes uses
"struct mem_cgroup *memcg".  Rename all mem variables to memcg in source
file.

Signed-off-by: Raghavendra K T <raghavendra.kt@linux.vnet.ibm.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroup/kmemleak: Annotate alloc_page() for cgroup allocations

When the cgroup base was allocated with kmalloc, it was necessary to
annotate the variable with kmemleak_not_leak().  But because it has
recently been changed to be allocated with alloc_page() (which skips
kmemleak checks) causes a warning on boot up.

I was triggering this output:

 allocated 8388608 bytes of page_cgroup
 please try 'cgroup_disable=memory' option if you don't want memory cgroups
 kmemleak: Trying to color unknown object at 0xf5840000 as Grey
 Pid: 0, comm: swapper Not tainted 3.0.0-test #12
 Call Trace:
  [<c17e34e6>] ? printk+0x1d/0x1f^M
  [<c10e2941>] paint_ptr+0x4f/0x78
  [<c178ab57>] kmemleak_not_leak+0x58/0x7d
  [<c108ae9f>] ? __rcu_read_unlock+0x9/0x7d
  [<c1cdb462>] kmemleak_init+0x19d/0x1e9
  [<c1cbf771>] start_kernel+0x346/0x3ec
  [<c1cbf1b4>] ? loglevel+0x18/0x18
  [<c1cbf0aa>] i386_start_kernel+0xaa/0xb0

After a bit of debugging I tracked the object 0xf840000 (and others) down
to the cgroup code.  The change from allocating base with kmalloc to
alloc_page() has the base not calling kmemleak_alloc() which adds the
pointer to the object_tree_root, but kmemleak_not_leak() adds it to the
crt_early_log[] table.  On kmemleak_init(), the entry is found in the
early_log[] but not the object_tree_root, and this error message is
displayed.

If alloc_page() fails then it defaults back to vmalloc() which still uses
the kmemleak_alloc() which makes us still need the kmemleak_not_leak()
call.  The solution is to call the kmemleak_alloc() directly if the
alloc_page() succeeds.

Reviewed-by: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroups: don't attach task to subsystem if migration failed

If a task has exited to the point it has called cgroup_exit() already,
then we can't migrate it to another cgroup anymore.

This can happen when we are attaching a task to a new cgroup between the
call to ->can_attach_task() on subsystems and the migration that is
eventually tried in cgroup_task_migrate().

In this case cgroup_task_migrate() returns -ESRCH and we don't want to
attach the task to the subsystems because the attachment to the new cgroup
itself failed.

Fix this by only calling ->attach_task() on the subsystems if the cgroup
migration succeeded.

Reported-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Ben Blum <bblum@andrew.cmu.edu>
Acked-by: Paul Menage <paul@paulmenage.org>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroups: more safe tasklist locking in cgroup_attach_proc

Fix unstable tasklist locking in cgroup_attach_proc.

According to this thread - https://lkml.org/lkml/2011/7/27/243 - RCU is
not sufficient to guarantee the tasklist is stable w.r.t.  de_thread and
exit.  Taking tasklist_lock for reading, instead of rcu_read_lock, ensures
proper exclusion.

Signed-off-by: Ben Blum <bblum@andrew.cmu.edu>
Acked-by: Paul Menage <paul@paulmenage.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Cc: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

hfs: fix hfs_find_init() sb->ext_tree NULL ptr oops

Clement Lecigne reports a filesystem which causes a kernel oops in
hfs_find_init() trying to dereference sb->ext_tree which is NULL.

This proves to be because the filesystem has a corrupted MDB extent
record, where the extents file does not fit into the first three extents
in the file record (the first blocks).

In hfs_get_block() when looking up the blocks for the extent file
(HFS_EXT_CNID), it fails the first blocks special case, and falls
through to the extent code (which ultimately calls hfs_find_init())
which is in the process of being initialised.

Hfs avoids this scenario by always having the extents b-tree fitting
into the first blocks (the extents B-tree can't have overflow extents).

The fix is to check at mount time that the B-tree fits into first
blocks, i.e.  fail if HFS_I(inode)->alloc_blocks >=
HFS_I(inode)->first_blocks

Note, the existing commit 47f365eb57573 ("hfs: fix oops on mount with
corrupted btree extent records") becomes subsumed into this as a special
case, but only for the extents B-tree (HFS_EXT_CNID), it is perfectly
acceptable for the catalog B-Tree file to grow beyond three extents,
with the remaining extent descriptors in the extents overfow.

This fixes CVE-2011-2203

Reported-by: Clement LECIGNE <clement.lecigne@netasq.com>
Signed-off-by: Phillip Lougher <plougher@redhat.com>
Cc: Jeff Mahoney <jeffm@suse.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

isofs: add readpages support

Use mpage_readpages() instead of multiple calls to isofs_readpage() to
reduce the CPU utilization and make performance higher.

Signed-off-by: Namjae Jeon <linkinjeon@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

minix: describe usage of different magic numbers

One can get this information from minix/inode.c, but adding the
explanations at the definition sites is more appropriate.

Signed-off-by: Sami Kerola <kerolasa@iki.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/rtc/rtc-mc13xxx.c: move probe and remove callbacks to .init.text and .exit.text

The driver is added using platform_driver_probe(), so the callbacks can be
discarded more aggessively.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

rtc: add initial support for mcp7941x parts

Add initial support for the microchip mcp7941x series of real time clocks.

The mcp7941x series is generally compatible with the ds1307 and ds1337 rtc
devices from dallas semiconductor.  minor differences include a backup
battery enable bit, and the polarity of the oscillator enable bit.

Signed-off-by: David Anders <danders.dev@gmail.com>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Reviewed-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/rtc/class.c: convert idr to ida and use ida_simple_get()

This is the one use of an ida that doesn't retry on receiving -EAGAIN.
I'm assuming do so will cause no harm and may help on a rare occasion.

Signed-off-by: Jonathan Cameron <jic23@cam.ac.uk>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

init/do_mounts_rd.c: fix ramdisk identification for padded cramfs

When a cramfs ramdisk padded with 512 bytes is given to the kernel, the
current identify_ramdisk_image function fails to identify it.

Tested with a padded cramfs image on an ARM based board.

Signed-off-by: Neil Armstrong <narmstrong@neotion.com>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Davidlohr Bueso <dave@gnu.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ramfs: remove module leftovers

Since ramfs is hard-selected to "y", the module leftovers make no sense.

Signed-off-by: Richard Weinberger <richard@nod.at>
Reviewed-by: WANG Cong <xiyou.wangcong@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

binfmt_elf: fix PIE execution with randomization disabled

The case of address space randomization being disabled in runtime through
randomize_va_space sysctl is not treated properly in load_elf_binary(),
resulting in SIGKILL coming at exec() time for certain PIE-linked binaries
in case the randomization has been disabled at runtime prior to calling
exec().

Handle the randomize_va_space == 0 case the same way as if we were not
supporting .text randomization at all.

Based on original patch by H.J. Lu and Josh Boyer.

Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: H.J. Lu <hongjiu.lu@intel.com>
Cc: <stable@kernel.org>
Tested-by: Josh Boyer <jwboyer@redhat.com>
Acked-by: Nicolas Pitre <nicolas.pitre@linaro.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

thp: share get_huge_page_tail()

This avoids duplicating the function in every arch gup_fast.

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: David Gibson <david@gibson.dropbear.id.au>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

sparc: gup_pte_range() support THP based tail recounting

Up to this point the code assumed old refcounting for hugepages (pre-thp).
 This updates the code directly to the thp mapcount tail page refcounting.

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: David Gibson <david@gibson.dropbear.id.au>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Acked-by: David Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

s390: gup_huge_pmd() return 0 if pte changes

s390 didn't return 0 in that case, if it's rolling back the *nr pointer it
should also return zero to avoid adding pages to the array at the wrong
offset.

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: David Gibson <david@gibson.dropbear.id.au>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

s390: gup_huge_pmd() support THP tail recounting

Up to this point the code assumed old refcounting for hugepages (pre-thp).
This updates the code directly to the thp mapcount tail page refcounting.

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: David Gibson <david@gibson.dropbear.id.au>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

powerpc: gup_huge_pmd() return 0 if pte changes

powerpc didn't return 0 in that case, if it's rolling back the *nr pointer
it should also return zero to avoid adding pages to the array at the wrong
offset.

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

powerpc: gup_hugepte() support THP based tail recounting

Up to this point the code assumed old refcounting for hugepages (pre-thp).
This updates the code directly to the thp mapcount tail page refcounting.

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

powerpc: gup_hugepte() avoid freeing the head page too many times

We only taken "refs" pins on the head page not "*nr" pins.

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

powerpc: get_hugepte() don't put_page() the wrong page

"page" may have changed to point to the next hugepage after the loop
completed, The references have been taken on the head page, so the
put_page must happen there too.

This is a longstanding issue pre-thp inclusion.

It's totally unclear how these page_cache_add_speculative and
pte_val(pte) != pte_val(*ptep) checks are necessary across all the
powerpc gup_fast code, when x86 doesn't need any of that: there's no way
the page can be freed with irq disabled so we're guaranteed the
atomic_inc will happen on a page with page_count > 0 (so not needing the
speculative check).

The pte check is also meaningless on x86: no need to rollback on x86 if
the pte changed, because the pte can still change a CPU tick after the
check succeeded and it won't be rolled back in that case.  The important
thing is we got a reference on a valid page that was mapped there a CPU
tick ago.  So not knowing the soft tlb refill code of ppc64 in great
detail I'm not removing the "speculative" page_count increase and the
pte checks across all the code, but unless there's a strong reason for
it they should be later cleaned up too.

If a pte can change from huge to non-huge (like it could happen with
THP) passing a pte_t *ptep to gup_hugepte() would also require to repeat
the is_hugepd in gup_hugepte(), but that shouldn't happen with hugetlbfs
only so I'm not altering that.

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

powerpc: remove superfluous PageTail checks on the pte gup_fast

This part of gup_fast doesn't seem capable of handling hugetlbfs ptes,
those should be handled by gup_hugepd only, so these checks are
superfluous.

Plus if this wasn't a noop, it would have oopsed because, the insistence
of using the speculative refcounting would trigger a VM_BUG_ON if a tail
page was encountered in the page_cache_get_speculative().

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: thp: tail page refcounting fix

Michel while working on the working set estimation code, noticed that
calling get_page_unless_zero() on a random pfn_to_page(random_pfn)
wasn't safe, if the pfn ended up being a tail page of a transparent
hugepage under splitting by __split_huge_page_refcount().

He then found the problem could also theoretically materialize with
page_cache_get_speculative() during the speculative radix tree lookups
that uses get_page_unless_zero() in SMP if the radix tree page is freed
and reallocated and get_user_pages is called on it before
page_cache_get_speculative has a chance to call get_page_unless_zero().

So the best way to fix the problem is to keep page_tail->_count zero at
all times.  This will guarantee that get_page_unless_zero() can never
succeed on any tail page.  page_tail->_mapcount is guaranteed zero and
is unused for all tail pages of a compound page, so we can simply
account the tail page references there and transfer them to
tail_page->_count in __split_huge_page_refcount() (in addition to the
head_page->_mapcount).

While debugging this s/_count/_mapcount/ change I also noticed get_page is
called by direct-io.c on pages returned by get_user_pages.  That wasn't
entirely safe because the two atomic_inc in get_page weren't atomic.  As
opposed to other get_user_page users like secondary-MMU page fault to
establish the shadow pagetables would never call any superflous get_page
after get_user_page returns.  It's safer to make get_page universally safe
for tail pages and to use get_page_foll() within follow_page (inside
get_user_pages()).  get_page_foll() is safe to do the refcounting for tail
pages without taking any locks because it is run within PT lock protected
critical sections (PT lock for pte and page_table_lock for
pmd_trans_huge).

The standard get_page() as invoked by direct-io instead will now take
the compound_lock but still only for tail pages.  The direct-io paths
are usually I/O bound and the compound_lock is per THP so very
finegrined, so there's no risk of scalability issues with it.  A simple
direct-io benchmarks with all lockdep prove locking and spinlock
debugging infrastructure enabled shows identical performance and no
overhead.  So it's worth it.  Ideally direct-io should stop calling
get_page() on pages returned by get_user_pages().  The spinlock in
get_page() is already optimized away for no-THP builds but doing
get_page() on tail pages returned by GUP is generally a rare operation
and usually only run in I/O paths.

This new refcounting on page_tail->_mapcount in addition to avoiding new
RCU critical sections will also allow the working set estimation code to
work without any further complexity associated to the tail page
refcounting with THP.

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-by: Michel Lespinasse <walken@google.com>
Reviewed-by: Michel Lespinasse <walken@google.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: David Gibson <david@gibson.dropbear.id.au>
Cc: <stable@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Merge git://github.com/rustyrussell/linux

* git://github.com/rustyrussell/linux:
  virtio-blk: use ida to allocate disk index
  virtio: Add platform bus driver for memory mapped virtio device
  virtio: Dont add "config" to list for !per_vq_vector
  virtio: console: wait for first console port for early console output
  virtio: console: add port stats for bytes received, sent and discarded
  virtio: console: make discard_port_data() use get_inbuf()
  virtio: console: rename variable
  virtio: console: make get_inbuf() return port->inbuf if present
  virtio: console: Fix return type for get_inbuf()
  virtio: console: Use wait_event_freezable instead of _interruptible
  virtio: console: Ignore port name update request if name already set
  virtio: console: Fix indentation
  virtio: modify vring_init and vring_size to take account of the layout containing *_event_idx
  virtio.h: correct comment for struct virtio_driver
  virtio-net: Use virtio_config_val() for retrieving config
  virtio_config: Add virtio_config_val_len()
  virtio-console: Use virtio_config_val() for retrieving config

Merge branch 'for-next' of git://git./linux/kernel/git/hch/vfs-queue

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/hch/vfs-queue:
  vfs: add d_prune dentry operation
  vfs: protect i_nlink
  filesystems: add set_nlink()
  filesystems: add missing nlink wrappers
  logfs: remove unnecessary nlink setting
  ocfs2: remove unnecessary nlink setting
  jfs: remove unnecessary nlink setting
  hypfs: remove unnecessary nlink setting
  vfs: ignore error on forced remount
  readlinkat: ensure we return ENOENT for the empty pathname for normal lookups
  vfs: fix dentry leak in simple_fill_super()

Merge branch 'for_linus' of git://git./linux/kernel/git/tytso/ext4

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4: (97 commits)
  jbd2: Unify log messages in jbd2 code
  jbd/jbd2: validate sb->s_first in journal_get_superblock()
  ext4: let ext4_ext_rm_leaf work with EXT_DEBUG defined
  ext4: fix a syntax error in ext4_ext_insert_extent when debugging enabled
  ext4: fix a typo in struct ext4_allocation_context
  ext4: Don't normalize an falloc request if it can fit in 1 extent.
  ext4: remove comments about extent mount option in ext4_new_inode()
  ext4: let ext4_discard_partial_buffers handle unaligned range correctly
  ext4: return ENOMEM if find_or_create_pages fails
  ext4: move vars to local scope in ext4_discard_partial_page_buffers_no_lock()
  ext4: Create helper function for EXT4_IO_END_UNWRITTEN and i_aiodio_unwritten
  ext4: optimize locking for end_io extent conversion
  ext4: remove unnecessary call to waitqueue_active()
  ext4: Use correct locking for ext4_end_io_nolock()
  ext4: fix race in xattr block allocation path
  ext4: trace punch_hole correctly in ext4_ext_map_blocks
  ext4: clean up AGGRESSIVE_TEST code
  ext4: move variables to their scope
  ext4: fix quota accounting during migration
  ext4: migrate cleanup
  ...

Merge branch 'for_linus' of git://git./linux/kernel/git/jack/linux-fs

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs:
  udf: Cleanup metadata flags handling
  udf: Skip mirror metadata FE loading when metadata FE is ok
  ext3: Allow quota file use root reservation
  udf: Remove web reference from UDF MAINTAINERS entry
  quota: Drop path reference on error exit from quotactl
  udf: Neaten udf_debug uses
  udf: Neaten logging output, use vsprintf extension %pV
  udf: Convert printks to pr_<level>
  udf: Rename udf_warning to udf_warn
  udf: Rename udf_error to udf_err
  udf: Promote some debugging messages to udf_error
  ext3: Remove the obsolete broken EXT3_IOC32_WAIT_FOR_READONLY.
  udf: Add readpages support for udf.
  ext3/balloc.c: local functions should be static
  ext2: fix the outdated comment in ext2_nfs_get_inode()
  ext3: remove deprecated oldalloc
  fs/ext3/balloc.c: delete useless initialization
  fs/ext2/balloc.c: delete useless initialization
  ext3: fix message in ext3_remount for rw-remount case
  ext3: Remove i_mutex from ext3_sync_file()

Fix up trivial (printf format cleanup) conflicts in fs/udf/udfdecl.h

Merge branch 'for-linus' of git://github.com/richardweinberger/linux

* 'for-linus' of git://github.com/richardweinberger/linux: (90 commits)
  um: fix ubd cow size
  um: Fix kmalloc argument order in um/vdso/vma.c
  um: switch to use of drivers/Kconfig
  UserModeLinux-HOWTO.txt: fix a typo
  UserModeLinux-HOWTO.txt: remove ^H characters
  um: we need sys/user.h only on i386
  um: merge delay_{32,64}.c
  um: distribute exports to where exported stuff is defined
  um: kill system-um.h
  um: generic ftrace.h will do...
  um: segment.h is x86-only and needed only there
  um: asm/pda.h is not needed anymore
  um: hw_irq.h can go generic as well
  um: switch to generic-y
  um: clean Kconfig up a bit
  um: a couple of missing dependencies...
  um: kill useless argument of free_chan() and free_one_chan()
  um: unify ptrace_user.h
  um: unify KSTK_...
  um: fix gcov build breakage
  ...

um: fix ubd cow size

ubd_file_size() cannot use ubd_dev->cow.file because at this time
ubd_dev->cow.file is not initialized.
Therefore, ubd_file_size() will always report a wrong disk size when
COW files are used.
Reading from /dev/ubd* would crash the kernel.

We have to read the correct disk size from the COW file's backing
file.

Signed-off-by: Richard Weinberger <richard@nod.at>
CC: stable@kernel.org

um: Fix kmalloc argument order in um/vdso/vma.c

kmalloc size is 1st arg, not second.

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
Cc: <stable@kernel.org> # 3.0.x
[richard@nod.at: on 3.0 the to be patched file is
arch/um/sys-x86_64/vdso/vma.c]

um: switch to use of drivers/Kconfig

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

UserModeLinux-HOWTO.txt: fix a typo

Signed-off-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
Signed-off-by: Richard Weinberger <richard@nod.at>

UserModeLinux-HOWTO.txt: remove ^H characters

If you can't read this patch, please run:

sed -i -e "s/[^\o10]\o10//g" \
Documentation/virtual/uml/UserModeLinux-HOWTO.txt

Signed-off-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: we need sys/user.h only on i386

Signed-off-by: Richard Weinberger <richard@nod.at>

um: merge delay_{32,64}.c

Signed-off-by: Richard Weinberger <richard@nod.at>

um: distribute exports to where exported stuff is defined

ksyms.c is down to the stuff defined in various USER_OBJS

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: kill system-um.h

most of it belonged in irqflags.h, actually

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: generic ftrace.h will do...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: segment.h is x86-only and needed only there

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: asm/pda.h is not needed anymore

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: hw_irq.h can go generic as well

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: switch to generic-y

kill wrapper headers

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: clean Kconfig up a bit

* kill duplicates with drivers/char/Kconfig
* take watchdog one into drivers/watchdog/Kconfig
* take mmapper to arch/um/Kconfig.um
* rename Kconfig.char menu to "UML Character Devices"

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: a couple of missing dependencies...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: kill useless argument of free_chan() and free_one_chan()

delay_free_irq is always 0 for those...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: unify ptrace_user.h

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: unify KSTK_...

... and switch get_thread_register() to HOST_... for register numbers

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: fix gcov build breakage

a) exports in gmon_syms.c duplicate kernel/gcov/* ones
b) excluding -pg in vdso compile is not enough - -fprofile-arcs
and -ftest-coverage also needs to be excluded

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: page_offset.h is never used

... and neither is the only define in it

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: irq_vectors.h just shadows x86 one

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: required-features.h is there only to shadow x86 one...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: asm/apic.h is there only to shadow the x86 one...

... so take it to arch/um/x86/asm.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: take ubd_user.h to its users...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: take ldt.h to arch/x86/um/asm/mm_context.h

it's x86-only and we have no business playing with it in asm/mmu.h; make
the latter have
struct uml_arch_mm_context arch;
instead of
struct uml_ldt ldt;
and let arch/<subarch>/um/asm/mm_context.h decide what'll be in there.
While we are at it, kill host_ldt.h - it's not needed in part of places
that include it (we want asm/ldt.h in those) and it can be trivially
expanded into the single remaining one.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: merge signal_{32,64}.c

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: no need to play with save_sp in signal frame setup anymore

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: increase stack growth cushion in pagefault

analog of [PATCH] i386: let usermode execute the "enter" instruction from
circa 2006.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: merge HOST_... of registers common on i386 and amd64

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: sanitize paths in sys_call_table* includes

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: merge os-Linux/tls.c into arch/x86/um/os-Linux/tls.c

it's i386-specific; moreover, analogs on other targets have
incompatible interface - PTRACE_GET_THREAD_AREA does exist
elsewhere, but struct user_desc does *not*

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: move asm/desc.h into arch/x86/um/asm

its only purpose is to shadow the x86 asm/desc.h

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>

um: merge host_ldt_{32,64}.h

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Richard Weinberger <richard@nod.at>