GitHub/MotorolaMobilityLLC/kernel-slsi.git
13 years agoatomic: use <linux/atomic.h>
Arun Sharma [Tue, 26 Jul 2011 23:09:06 +0000 (16:09 -0700)]
atomic: use <linux/atomic.h>

This allows us to move duplicated code in <asm/atomic.h>
(atomic_inc_not_zero() for now) to <linux/atomic.h>

Signed-off-by: Arun Sharma <asharma@fb.com>
Reviewed-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoasm-generic: add another generic ext2 atomic bitops
Akinobu Mita [Tue, 26 Jul 2011 23:09:04 +0000 (16:09 -0700)]
asm-generic: add another generic ext2 atomic bitops

The majority of architectures implement ext2 atomic bitops as
test_and_{set,clear}_bit() without spinlock.

This adds this type of generic implementation in ext2-atomic-setbit.h and
use it wherever possible.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Suggested-by: Andreas Dilger <adilger@dilger.ca>
Suggested-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agofail_make_request: cleanup should_fail_request
Akinobu Mita [Tue, 26 Jul 2011 23:09:03 +0000 (16:09 -0700)]
fail_make_request: cleanup should_fail_request

This changes should_fail_request() to more usable wrapper function of
should_fail().  It can avoid putting #ifdef CONFIG_FAIL_MAKE_REQUEST in
the middle of a function.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agofail_page_alloc: simplify debugfs initialization
Akinobu Mita [Tue, 26 Jul 2011 23:09:03 +0000 (16:09 -0700)]
fail_page_alloc: simplify debugfs initialization

Now cleanup_fault_attr_dentries() recursively removes a directory, So we
can simplify the error handling in the initialization code and no need
to hold dentry structs for each debugfs file.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agofailslab: simplify debugfs initialization
Akinobu Mita [Tue, 26 Jul 2011 23:09:02 +0000 (16:09 -0700)]
failslab: simplify debugfs initialization

Now cleanup_fault_attr_dentries() recursively removes a directory, So we
can simplify the error handling in the initialization code and no need
to hold dentry structs for each debugfs file.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Matt Mackall <mpm@selenic.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agofault-injection: use debugfs_remove_recursive
Akinobu Mita [Tue, 26 Jul 2011 23:09:02 +0000 (16:09 -0700)]
fault-injection: use debugfs_remove_recursive

Use debugfs_remove_recursive() to simplify initialization and
deinitialization of fault injection debugfs files.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agofault-injection: cleanup simple attribute of stacktrace_depth
Akinobu Mita [Tue, 26 Jul 2011 23:09:01 +0000 (16:09 -0700)]
fault-injection: cleanup simple attribute of stacktrace_depth

Minor cosmetic changes for simple attribute of stacktrace_depth:

 - use min_t()
 - reduce #ifdef by moving a function
 - do not use partly capitalized function name

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agofault-injection: remove nonexistent function extern
Akinobu Mita [Tue, 26 Jul 2011 23:09:00 +0000 (16:09 -0700)]
fault-injection: remove nonexistent function extern

should_fail_srandom() does not exist.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agofault-injection: do not include unneeded header
Akinobu Mita [Tue, 26 Jul 2011 23:09:00 +0000 (16:09 -0700)]
fault-injection: do not include unneeded header

No need to include linux/kallsyms.h.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoramoops: make record_size a module parameter
Sergiu Iordache [Tue, 26 Jul 2011 23:08:59 +0000 (16:08 -0700)]
ramoops: make record_size a module parameter

The size of the dump is currently set using the RECORD_SIZE macro which
is set to a page size.  This patch makes the record size a module
parameter and allows it to be set through platform data as well to allow
larger dumps if needed.

Signed-off-by: Sergiu Iordache <sergiu@chromium.org>
Acked-by: Marco Stornelli <marco.stornelli@gmail.com>
Cc: "Ahmed S. Darwish" <darwish.07@gmail.com>
Cc: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoramoops: move dump_oops into platform data
Sergiu Iordache [Tue, 26 Jul 2011 23:08:58 +0000 (16:08 -0700)]
ramoops: move dump_oops into platform data

The platform driver currently allows setting the mem_size and
mem_address.

ince dump_oops is also a module parameter it would be more consistent if
it could be set through platform data as well.

Signed-off-by: Sergiu Iordache <sergiu@chromium.org>
Acked-by: Marco Stornelli <marco.stornelli@gmail.com>
Cc: "Ahmed S. Darwish" <darwish.07@gmail.com>
Cc: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoramoops: add new line to each print
Marco Stornelli [Tue, 26 Jul 2011 23:08:57 +0000 (16:08 -0700)]
ramoops: add new line to each print

Add new line to each print.

Signed-off-by: Marco Stornelli <marco.stornelli@gmail.com>
Reported-by: Stevie Trujillo <stevie.trujillo@gmail.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Américo Wang <xiyou.wangcong@gmail.com>
Cc: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoramoops: use module parameters instead of platform data if not available
Marco Stornelli [Tue, 26 Jul 2011 23:08:57 +0000 (16:08 -0700)]
ramoops: use module parameters instead of platform data if not available

Use generic module parameters instead of platform data, if platform data
are not available.  This limitation has been introduced with commit
c3b92ce9e75 ("ramoops: use the platform data structure instead of module
params").

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Marco Stornelli <marco.stornelli@gmail.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Américo Wang <xiyou.wangcong@gmail.com>
Reported-by: Stevie Trujillo <stevie.trujillo@gmail.com>
Cc: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoVmware balloon: switch to using sysem-wide freezable workqueue
Dmitry Torokhov [Tue, 26 Jul 2011 23:08:56 +0000 (16:08 -0700)]
Vmware balloon: switch to using sysem-wide freezable workqueue

With the arrival of concurrency-managed workqueues there is no need for
our driver to use dedicated workqueue; system-wide one should suffice just
fine.

[akpm@linux-foundation.org: fix comment layout & grammar]
Signed-off-by: Dmitry Torokhov <dtor@vmware.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agodrivers/w1/slaves/w1_therm.c: add support for DS28EA00
Christian Glindkamp [Tue, 26 Jul 2011 23:08:55 +0000 (16:08 -0700)]
drivers/w1/slaves/w1_therm.c: add support for DS28EA00

Signed-off-by: Christian Glindkamp <christian.glindkamp@taskit.de>
Cc: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agopanic, vt: do not force oops output when panic_timeout < 0
Mandeep Singh Baines [Tue, 26 Jul 2011 23:08:54 +0000 (16:08 -0700)]
panic, vt: do not force oops output when panic_timeout < 0

Don't force output if you intend to reboot immediately.

In this patch, I'm disabling the functionality enabled by
vc->vc_panic_force_write if panic_timeout < 0 (i.e.  no timeout).
vc_panic_force_write is only enabled for fb video consoles if the
FBINFO_CAN_FORCE_OUTPUT flag is set.

For our application, we're using ram_oops to preserved the panic in
memory.  We want to reliably, and as fast as possible, machine_restart.
The vc_panic_force_write flag results in a bunch of graphics driver code
to be invoked which slows down restart and decreases reliability.  Since
we're already storing the panic in RAM and are going to reboot
immediately, there is no benefit in mode switching back to the vc in
order to display the panic output.  The log buffer will get flushed by
the console_unblank() call so remote management consoles should see all
output.

Signed-off-by: Mandeep Singh Baines <msb@chromium.org>
Cc: Huang Ying <ying.huang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Olaf Hering <olaf@aepfle.de>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: Dave Airlie <airlied@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Acked-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agopanic: panic=-1 for immediate reboot
Hugh Dickins [Tue, 26 Jul 2011 23:08:52 +0000 (16:08 -0700)]
panic: panic=-1 for immediate reboot

When a kernel BUG or oops occurs, ChromeOS intends to panic and
immediately reboot, with stacktrace and other messages preserved in RAM
across reboot.

But the longer we delay, the more likely the user is to poweroff and
lose the info.

panic_timeout (seconds before rebooting) is set by panic= boot option or
sysctl or /proc/sys/kernel/panic; but 0 means wait forever, so at
present we have to delay at least 1 second.

Let a negative number mean reboot immediately (with the small cosmetic
benefit of suppressing that newline-less "Rebooting in %d seconds.."
message).

Signed-off-by: Hugh Dickins <hughd@chromium.org>
Signed-off-by: Mandeep Singh Baines <msb@chromium.org>
Cc: Huang Ying <ying.huang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Olaf Hering <olaf@aepfle.de>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: Dave Airlie <airlied@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoDocumentation/DMA-API-HOWTO.txt: fix misleading example
Michal Miroslaw [Tue, 26 Jul 2011 23:08:51 +0000 (16:08 -0700)]
Documentation/DMA-API-HOWTO.txt: fix misleading example

See: DMA-API.txt, part Id, DMA_FROM_DEVICE description.

Signed-off-by: Michal Miroslaw <mirq-linux@rere.qmqm.pl>
Cc: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoinclude/linux/dma-mapping.h: remove DMA_xxBIT_MASK macros
WANG Cong [Tue, 26 Jul 2011 23:08:50 +0000 (16:08 -0700)]
include/linux/dma-mapping.h: remove DMA_xxBIT_MASK macros

git grep shows there are no users in tree, so we can remove them safely.

Signed-off-by: WANG Cong <xiyou.wangcong@gmail.com>
Acked-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: Jiri Slaby <jslaby@suse.cz>
Acked-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agogcov: disable CONSTRUCTORS for UML
Vitaliy Ivanov [Tue, 26 Jul 2011 23:08:49 +0000 (16:08 -0700)]
gcov: disable CONSTRUCTORS for UML

Selecting GCOV for UML causing configuration mismatch:

  warning: (GCOV_KERNEL) selects CONSTRUCTORS which has unmet direct dependencies (!UML)

Constructors are not needed for UML.

Signed-off-by: Vitaliy Ivanov <vitalivanov@gmail.com>
Cc: Peter Oberparleiter <oberpar@linux.vnet.ibm.com>
Acked-by: Richard Weinberger <richard@nod.at>
Acked-by: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agodrivers/edac/mpc85xx_edac.c: correct offset_in_page mask bits in edac_mc_handle_ce()
Kai.Jiang [Tue, 26 Jul 2011 23:08:49 +0000 (16:08 -0700)]
drivers/edac/mpc85xx_edac.c: correct offset_in_page mask bits in edac_mc_handle_ce()

Parameter offset_in_page in edac_mc_handle_ce() should mask the higher
bits above the page size, not the lower bits.  The original input
sometimes causes a crash.

Signed-off-by: Kai.Jiang <Kai.Jiang@freescale.com>
Signed-off-by: Shaohui Xie <Shaohui.Xie@freescale.com>
Cc: Anton Vorontsov <avorontsov@mvista.com>
Cc: Grant Likely <grant.likely@secretlab.ca>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoipc: introduce shm_rmid_forced sysctl
Vasiliy Kulikov [Tue, 26 Jul 2011 23:08:48 +0000 (16:08 -0700)]
ipc: introduce shm_rmid_forced sysctl

Add support for the shm_rmid_forced sysctl.  If set to 1, all shared
memory objects in current ipc namespace will be automatically forced to
use IPC_RMID.

The POSIX way of handling shmem allows one to create shm objects and
call shmdt(), leaving shm object associated with no process, thus
consuming memory not counted via rlimits.

With shm_rmid_forced=1 the shared memory object is counted at least for
one process, so OOM killer may effectively kill the fat process holding
the shared memory.

It obviously breaks POSIX - some programs relying on the feature would
stop working.  So set shm_rmid_forced=1 only if you're sure nobody uses
"orphaned" memory.  Use shm_rmid_forced=0 by default for compatability
reasons.

The feature was previously impemented in -ow as a configure option.

[akpm@linux-foundation.org: fix documentation, per Randy]
[akpm@linux-foundation.org: fix warning]
[akpm@linux-foundation.org: readability/conventionality tweaks]
[akpm@linux-foundation.org: fix shm_rmid_forced/shm_forced_rmid confusion, use standard comment layout]
Signed-off-by: Vasiliy Kulikov <segoon@openwall.com>
Cc: Randy Dunlap <rdunlap@xenotime.net>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: "Serge E. Hallyn" <serge.hallyn@canonical.com>
Cc: Daniel Lezcano <daniel.lezcano@free.fr>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Solar Designer <solar@openwall.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoipc/mqueue.c: fix mq_open() return value
Jiri Slaby [Tue, 26 Jul 2011 23:08:47 +0000 (16:08 -0700)]
ipc/mqueue.c: fix mq_open() return value

We return ENOMEM from mqueue_get_inode even when we have enough memory.
Namely in case the system rlimit of mqueue was reached.  This error
propagates to mq_queue and user sees the error unexpectedly.  So fix
this up to properly return EMFILE as described in the manpage:

EMFILE The process already has the maximum number of files and
       message queues open.

instead of:

ENOMEM Insufficient memory.

With the previous patch we just switch to ERR_PTR/PTR_ERR/IS_ERR error
handling here.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoipc/mqueue.c: refactor failure handling
Jiri Slaby [Tue, 26 Jul 2011 23:08:46 +0000 (16:08 -0700)]
ipc/mqueue.c: refactor failure handling

If new_inode fails to allocate an inode we need only to return with
NULL.  But now we test the opposite and have all the work in a nested
block.  So do the opposite to save one indentation level (and remove
unnecessary line breaks).

This is only a preparation/cleanup for the next patch where we fix up
return values from mqueue_get_inode.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agocpumask: add cpumask_var_t documentation
KOSAKI Motohiro [Tue, 26 Jul 2011 23:08:45 +0000 (16:08 -0700)]
cpumask: add cpumask_var_t documentation

cpumask_var_t has one notable difference from cpumask_t.  Add the
explanation.

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Thiago Farina <tfransosi@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agocpumask: alloc_cpumask_var() use NUMA_NO_NODE
KOSAKI Motohiro [Tue, 26 Jul 2011 23:08:44 +0000 (16:08 -0700)]
cpumask: alloc_cpumask_var() use NUMA_NO_NODE

NUMA_NO_NODE and numa_node_id() have different meanings.  NUMA_NO_NODE is
obviously the recommended fallback.

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agocpumask: convert for_each_cpumask() with for_each_cpu()
KOSAKI Motohiro [Tue, 26 Jul 2011 23:08:44 +0000 (16:08 -0700)]
cpumask: convert for_each_cpumask() with for_each_cpu()

Adapt new API fashion.

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agofs/exec.c:acct_arg_size(): ptl is no longer needed for add_mm_counter()
Oleg Nesterov [Tue, 26 Jul 2011 23:08:43 +0000 (16:08 -0700)]
fs/exec.c:acct_arg_size(): ptl is no longer needed for add_mm_counter()

acct_arg_size() takes ->page_table_lock around add_mm_counter() if
!SPLIT_RSS_COUNTING.  This is not needed after commit 172703b08cd0 ("mm:
delete non-atomic mm counter implementation").

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reviewed-by: Matt Fleming <matt.fleming@linux.intel.com>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
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>
13 years agoexec: do not retry load_binary method if CONFIG_MODULES=n
Tetsuo Handa [Tue, 26 Jul 2011 23:08:42 +0000 (16:08 -0700)]
exec: do not retry load_binary method if CONFIG_MODULES=n

If CONFIG_MODULES=n, it makes no sense to retry the list of binary formats
handler because the list will not be modified by request_module().

Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Richard Weinberger <richard@nod.at>
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>
13 years agoexec: do not call request_module() twice from search_binary_handler()
Tetsuo Handa [Tue, 26 Jul 2011 23:08:41 +0000 (16:08 -0700)]
exec: do not call request_module() twice from search_binary_handler()

Currently, search_binary_handler() tries to load binary loader module
using request_module() if a loader for the requested program is not yet
loaded.  But second attempt of request_module() does not affect the result
of search_binary_handler().

If request_module() triggered recursion, calling request_module() twice
causes 2 to the power of MAX_KMOD_CONCURRENT (= 50) repetitions.  It is
not an infinite loop but is sufficient for users to consider as a hang up.

Therefore, this patch changes not to call request_module() twice, making 1
to the power of MAX_KMOD_CONCURRENT repetitions in case of recursion.

Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Reported-by: Richard Weinberger <richard@nod.at>
Tested-by: Richard Weinberger <richard@nod.at>
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>
13 years agofs/exec.c: use BUILD_BUG_ON for VM_STACK_FLAGS & VM_STACK_INCOMPLETE_SETUP
Michal Hocko [Tue, 26 Jul 2011 23:08:40 +0000 (16:08 -0700)]
fs/exec.c: use BUILD_BUG_ON for VM_STACK_FLAGS & VM_STACK_INCOMPLETE_SETUP

Commit a8bef8ff6ea1 ("mm: migration: avoid race between
shift_arg_pages() and rmap_walk() during migration by not migrating
temporary stacks") introduced a BUG_ON() to ensure that VM_STACK_FLAGS
and VM_STACK_INCOMPLETE_SETUP do not overlap.  The check is a compile
time one, so BUILD_BUG_ON is more appropriate.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Richard Weinberger <richard@nod.at>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agokernel/fork.c: fix a few coding style issues
Daniel Rebelo de Oliveira [Tue, 26 Jul 2011 23:08:39 +0000 (16:08 -0700)]
kernel/fork.c: fix a few coding style issues

Signed-off-by: Daniel Rebelo de Oliveira <psykon@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoproc: fix a race in do_io_accounting()
Vasiliy Kulikov [Tue, 26 Jul 2011 23:08:38 +0000 (16:08 -0700)]
proc: fix a race in do_io_accounting()

If an inode's mode permits opening /proc/PID/io and the resulting file
descriptor is kept across execve() of a setuid or similar binary, the
ptrace_may_access() check tries to prevent using this fd against the
task with escalated privileges.

Unfortunately, there is a race in the check against execve().  If
execve() is processed after the ptrace check, but before the actual io
information gathering, io statistics will be gathered from the
privileged process.  At least in theory this might lead to gathering
sensible information (like ssh/ftp password length) that wouldn't be
available otherwise.

Holding task->signal->cred_guard_mutex while gathering the io
information should protect against the race.

The order of locking is similar to the one inside of ptrace_attach():
first goes cred_guard_mutex, then lock_task_sighand().

Signed-off-by: Vasiliy Kulikov <segoon@openwall.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoprocfs: return ENOENT on opening a being-removed proc entry
Daisuke Ogino [Tue, 26 Jul 2011 23:08:37 +0000 (16:08 -0700)]
procfs: return ENOENT on opening a being-removed proc entry

Change the return value to ENOENT.  This return value is then returned
when opening the proc entry that have been removed.  For example,
open("/proc/bus/pci/XX/YY") when the corresponding device is being
hot-removed.

Signed-off-by: Daisuke Ogino <ogino.daisuke@jp.fujitsu.com>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Acked-by: Alexey Dobriyan <adobriyan@gmail.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>
13 years agoh8300/m68k/xtensa: __FD_ISSET should return 0/1
Andrew Morton [Tue, 26 Jul 2011 23:08:35 +0000 (16:08 -0700)]
h8300/m68k/xtensa: __FD_ISSET should return 0/1

Harmonise these return values with other architectures.  In some cases
this affects all compilers and in other cases non-gcc compilers only.

Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Chris Zankel <chris@zankel.net>
Cc: Ulrich Drepper <drepper@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agodo_coredump: fix the "ispipe" error check
Oleg Nesterov [Tue, 26 Jul 2011 23:08:34 +0000 (16:08 -0700)]
do_coredump: fix the "ispipe" error check

do_coredump() assumes that if format_corename() fails it should return
-ENOMEM.  This is not true, for example cn_print_exe_file() can propagate
the error from d_path.  Even if it was true, this is too fragile.  Change
the code to check "ispipe < 0".

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Reviewed-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agocoredump: escape / in hostname and comm
Jiri Slaby [Tue, 26 Jul 2011 23:08:33 +0000 (16:08 -0700)]
coredump: escape / in hostname and comm

Change every occurence of / in comm and hostname to !.  If the process
changes its name to contain /, the core is not dumped (if the directory
tree doesn't exist like that).  The same with hostname being something
like myhost/3.  Fix this behaviour by using the escape loop used in %E.
(We extract it to a separate function.)

Now both with comm == myprocess/1 and hostname == myhost/1, the core is
dumped like (kernel.core_pattern='core.%p.%e.%h):
core.2349.myprocess!1.myhost!1

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agocoredump: use task comm instead of (unknown)
Jiri Slaby [Tue, 26 Jul 2011 23:08:32 +0000 (16:08 -0700)]
coredump: use task comm instead of (unknown)

If we don't know the file corresponding to the binary (i.e.  exe_file is
unknown), use "task->comm (path unknown)" instead of simple "(unknown)"
as suggested by ak.

The fallback is the same as %e except it will append "(path unknown)".

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoptrace: unify show_regs() prototype
Mike Frysinger [Tue, 26 Jul 2011 23:08:31 +0000 (16:08 -0700)]
ptrace: unify show_regs() prototype

[ poleg@redhat.com: no need to declare show_regs() in ptrace.h, sched.h does this ]
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agocpusets: randomize node rotor used in cpuset_mem_spread_node()
Michal Hocko [Tue, 26 Jul 2011 23:08:30 +0000 (16:08 -0700)]
cpusets: randomize node rotor used in cpuset_mem_spread_node()

[ This patch has already been accepted as commit 0ac0c0d0f837 but later
  reverted (commit 35926ff5fba8) because it itroduced arch specific
  __node_random which was defined only for x86 code so it broke other
  archs.  This is a followup without any arch specific code.  Other than
  that there are no functional changes.]

Some workloads that create a large number of small files tend to assign
too many pages to node 0 (multi-node systems).  Part of the reason is
that the rotor (in cpuset_mem_spread_node()) used to assign nodes starts
at node 0 for newly created tasks.

This patch changes the rotor to be initialized to a random node number
of the cpuset.

[akpm@linux-foundation.org: fix layout]
[Lee.Schermerhorn@hp.com: Define stub numa_random() for !NUMA configuration]
[mhocko@suse.cz: Make it arch independent]
[akpm@linux-foundation.org: fix CONFIG_NUMA=y, MAX_NUMNODES>1 build]
Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Michal Hocko <mhocko@suse.cz>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Paul Menage <menage@google.com>
Cc: Jack Steiner <steiner@sgi.com>
Cc: Robin Holt <holt@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Jack Steiner <steiner@sgi.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Paul Menage <menage@google.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Robin Holt <holt@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: get rid of percpu_charge_mutex lock
Michal Hocko [Tue, 26 Jul 2011 23:08:29 +0000 (16:08 -0700)]
memcg: get rid of percpu_charge_mutex lock

percpu_charge_mutex protects from multiple simultaneous per-cpu charge
caches draining because we might end up having too many work items.  At
least this was the case until commit 26fe61684449 ("memcg: fix percpu
cached charge draining frequency") when we introduced a more targeted
draining for async mode.

Now that also sync draining is targeted we can safely remove mutex
because we will not send more work than the current number of CPUs.
FLUSHING_CACHED_CHARGE protects from sending the same work multiple
times and stock->nr_pages == 0 protects from pointless sending a work if
there is obviously nothing to be done.  This is of course racy but we
can live with it as the race window is really small (we would have to
see FLUSHING_CACHED_CHARGE cleared while nr_pages would be still
non-zero).

The only remaining place where we can race is synchronous mode when we
rely on FLUSHING_CACHED_CHARGE test which might have been set by other
drainer on the same group but we should wait in that case as well.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Cc: Balbir Singh <bsingharora@gmail.com>
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>
13 years agomemcg: add mem_cgroup_same_or_subtree() helper
Michal Hocko [Tue, 26 Jul 2011 23:08:29 +0000 (16:08 -0700)]
memcg: add mem_cgroup_same_or_subtree() helper

We are checking whether a given two groups are same or at least in the
same subtree of a hierarchy at several places.  Let's make a helper for
it to make code easier to read.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Balbir Singh <bsingharora@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: unify sync and async per-cpu charge cache draining
Michal Hocko [Tue, 26 Jul 2011 23:08:28 +0000 (16:08 -0700)]
memcg: unify sync and async per-cpu charge cache draining

Currently we have two ways how to drain per-CPU caches for charges.
drain_all_stock_sync will synchronously drain all caches while
drain_all_stock_async will asynchronously drain only those that refer to
a given memory cgroup or its subtree in hierarchy.  Targeted async
draining has been introduced by 26fe6168 (memcg: fix percpu cached
charge draining frequency) to reduce the cpu workers number.

sync draining is currently triggered only from mem_cgroup_force_empty
which is triggered only by userspace (mem_cgroup_force_empty_write) or
when a cgroup is removed (mem_cgroup_pre_destroy).  Although these are
not usually frequent operations it still makes some sense to do targeted
draining as well, especially if the box has many CPUs.

This patch unifies both methods to use the single code (drain_all_stock)
which relies on the original async implementation and just adds
flush_work to wait on all caches that are still under work for the sync
mode.  We are using FLUSHING_CACHED_CHARGE bit check to prevent from
waiting on a work that we haven't triggered.  Please note that both sync
and async functions are currently protected by percpu_charge_mutex so we
cannot race with other drainers.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Balbir Singh <bsingharora@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: do not try to drain per-cpu caches without pages
Michal Hocko [Tue, 26 Jul 2011 23:08:27 +0000 (16:08 -0700)]
memcg: do not try to drain per-cpu caches without pages

drain_all_stock_async tries to optimize a work to be done on the work
queue by excluding any work for the current CPU because it assumes that
the context we are called from already tried to charge from that cache
and it's failed so it must be empty already.

While the assumption is correct we can optimize it even more by checking
the current number of pages in the cache.  This will also reduce a work
on other CPUs with an empty stock.

For the current CPU we can simply call drain_local_stock rather than
deferring it to the work queue.

[kamezawa.hiroyu@jp.fujitsu.com: use drain_local_stock for current CPU optimization]
Signed-off-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>
13 years agomemcg: add memory.vmscan_stat
KAMEZAWA Hiroyuki [Tue, 26 Jul 2011 23:08:26 +0000 (16:08 -0700)]
memcg: add memory.vmscan_stat

The commit log of 0ae5e89c60c9 ("memcg: count the soft_limit reclaim
in...") says it adds scanning stats to memory.stat file.  But it doesn't
because we considered we needed to make a concensus for such new APIs.

This patch is a trial to add memory.scan_stat. This shows
  - the number of scanned pages(total, anon, file)
  - the number of rotated pages(total, anon, file)
  - the number of freed pages(total, anon, file)
  - the number of elaplsed time (including sleep/pause time)

  for both of direct/soft reclaim.

The biggest difference with oringinal Ying's one is that this file
can be reset by some write, as

  # echo 0 ...../memory.scan_stat

Example of output is here. This is a result after make -j 6 kernel
under 300M limit.

  [kamezawa@bluextal ~]$ cat /cgroup/memory/A/memory.scan_stat
  [kamezawa@bluextal ~]$ cat /cgroup/memory/A/memory.vmscan_stat
  scanned_pages_by_limit 9471864
  scanned_anon_pages_by_limit 6640629
  scanned_file_pages_by_limit 2831235
  rotated_pages_by_limit 4243974
  rotated_anon_pages_by_limit 3971968
  rotated_file_pages_by_limit 272006
  freed_pages_by_limit 2318492
  freed_anon_pages_by_limit 962052
  freed_file_pages_by_limit 1356440
  elapsed_ns_by_limit 351386416101
  scanned_pages_by_system 0
  scanned_anon_pages_by_system 0
  scanned_file_pages_by_system 0
  rotated_pages_by_system 0
  rotated_anon_pages_by_system 0
  rotated_file_pages_by_system 0
  freed_pages_by_system 0
  freed_anon_pages_by_system 0
  freed_file_pages_by_system 0
  elapsed_ns_by_system 0
  scanned_pages_by_limit_under_hierarchy 9471864
  scanned_anon_pages_by_limit_under_hierarchy 6640629
  scanned_file_pages_by_limit_under_hierarchy 2831235
  rotated_pages_by_limit_under_hierarchy 4243974
  rotated_anon_pages_by_limit_under_hierarchy 3971968
  rotated_file_pages_by_limit_under_hierarchy 272006
  freed_pages_by_limit_under_hierarchy 2318492
  freed_anon_pages_by_limit_under_hierarchy 962052
  freed_file_pages_by_limit_under_hierarchy 1356440
  elapsed_ns_by_limit_under_hierarchy 351386416101
  scanned_pages_by_system_under_hierarchy 0
  scanned_anon_pages_by_system_under_hierarchy 0
  scanned_file_pages_by_system_under_hierarchy 0
  rotated_pages_by_system_under_hierarchy 0
  rotated_anon_pages_by_system_under_hierarchy 0
  rotated_file_pages_by_system_under_hierarchy 0
  freed_pages_by_system_under_hierarchy 0
  freed_anon_pages_by_system_under_hierarchy 0
  freed_file_pages_by_system_under_hierarchy 0
  elapsed_ns_by_system_under_hierarchy 0

total_xxxx is for hierarchy management.

This will be useful for further memcg developments and need to be
developped before we do some complicated rework on LRU/softlimit
management.

This patch adds a new struct memcg_scanrecord into scan_control struct.
sc->nr_scanned at el is not designed for exporting information.  For
example, nr_scanned is reset frequentrly and incremented +2 at scanning
mapped pages.

To avoid complexity, I added a new param in scan_control which is for
exporting scanning score.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Ying Han <yinghan@google.com>
Cc: Andrew Bresticker <abrestic@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: fix behavior of mem_cgroup_resize_limit()
Daisuke Nishimura [Tue, 26 Jul 2011 23:08:25 +0000 (16:08 -0700)]
memcg: fix behavior of mem_cgroup_resize_limit()

Commit 22a668d7c3ef ("memcg: fix behavior under memory.limit equals to
memsw.limit") introduced "memsw_is_minimum" flag, which becomes true
when mem_limit == memsw_limit.  The flag is checked at the beginning of
reclaim, and "noswap" is set if the flag is true, because using swap is
meaningless in this case.

This works well in most cases, but when we try to shrink mem_limit,
which is the same as memsw_limit now, we might fail to shrink mem_limit
because swap doesn't used.

This patch fixes this behavior by:
 - check MEM_CGROUP_RECLAIM_SHRINK at the begining of reclaim
 - If it is set, don't set "noswap" flag even if memsw_is_minimum is true.

Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <bsingharora@gmail.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Ying Han <yinghan@google.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: fix vmscan count in small memcgs
KAMEZAWA Hiroyuki [Tue, 26 Jul 2011 23:08:24 +0000 (16:08 -0700)]
memcg: fix vmscan count in small memcgs

Commit 246e87a93934 ("memcg: fix get_scan_count() for small targets")
fixes the memcg/kswapd behavior against small targets and prevent vmscan
priority too high.

But the implementation is too naive and adds another problem to small
memcg.  It always force scan to 32 pages of file/anon and doesn't handle
swappiness and other rotate_info.  It makes vmscan to scan anon LRU
regardless of swappiness and make reclaim bad.  This patch fixes it by
adjusting scanning count with regard to swappiness at el.

At a test "cat 1G file under 300M limit." (swappiness=20)
 before patch
        scanned_pages_by_limit 360919
        scanned_anon_pages_by_limit 180469
        scanned_file_pages_by_limit 180450
        rotated_pages_by_limit 31
        rotated_anon_pages_by_limit 25
        rotated_file_pages_by_limit 6
        freed_pages_by_limit 180458
        freed_anon_pages_by_limit 19
        freed_file_pages_by_limit 180439
        elapsed_ns_by_limit 429758872
 after patch
        scanned_pages_by_limit 180674
        scanned_anon_pages_by_limit 24
        scanned_file_pages_by_limit 180650
        rotated_pages_by_limit 35
        rotated_anon_pages_by_limit 24
        rotated_file_pages_by_limit 11
        freed_pages_by_limit 180634
        freed_anon_pages_by_limit 0
        freed_file_pages_by_limit 180634
        elapsed_ns_by_limit 367119089
        scanned_pages_by_system 0

the numbers of scanning anon are decreased(as expected), and elapsed time
reduced. By this patch, small memcgs will work better.
(*) Because the amount of file-cache is much bigger than anon,
    recalaim_stat's rotate-scan counter make scanning files more.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: change memcg_oom_mutex to spinlock
Michal Hocko [Tue, 26 Jul 2011 23:08:24 +0000 (16:08 -0700)]
memcg: change memcg_oom_mutex to spinlock

memcg_oom_mutex is used to protect memcg OOM path and eventfd interface
for oom_control.  None of the critical sections which it protects sleep
(eventfd_signal works from atomic context and the rest are simple linked
list resp.  oom_lock atomic operations).

Mutex is also too heavyweight for those code paths because it triggers a
lot of scheduling.  It also makes makes convoying effects more visible
when we have a big number of oom killing because we take the lock
mutliple times during mem_cgroup_handle_oom so we have multiple places
where many processes can sleep.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Cc: Balbir Singh <bsingharora@gmail.com>
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>
13 years agomemcg: make oom_lock 0 and 1 based rather than counter
Michal Hocko [Tue, 26 Jul 2011 23:08:23 +0000 (16:08 -0700)]
memcg: make oom_lock 0 and 1 based rather than counter

Commit 867578cb ("memcg: fix oom kill behavior") introduced a oom_lock
counter which is incremented by mem_cgroup_oom_lock when we are about to
handle memcg OOM situation.  mem_cgroup_handle_oom falls back to a sleep
if oom_lock > 1 to prevent from multiple oom kills at the same time.
The counter is then decremented by mem_cgroup_oom_unlock called from the
same function.

This works correctly but it can lead to serious starvations when we have
many processes triggering OOM and many CPUs available for them (I have
tested with 16 CPUs).

Consider a process (call it A) which gets the oom_lock (the first one
that got to mem_cgroup_handle_oom and grabbed memcg_oom_mutex) and other
processes that are blocked on the mutex.  While A releases the mutex and
calls mem_cgroup_out_of_memory others will wake up (one after another)
and increase the counter and fall into sleep (memcg_oom_waitq).

Once A finishes mem_cgroup_out_of_memory it takes the mutex again and
decreases oom_lock and wakes other tasks (if releasing memory by
somebody else - e.g.  killed process - hasn't done it yet).

A testcase would look like:
  Assume malloc XXX is a program allocating XXX Megabytes of memory
  which touches all allocated pages in a tight loop
  # swapoff SWAP_DEVICE
  # cgcreate -g memory:A
  # cgset -r memory.oom_control=0   A
  # cgset -r memory.limit_in_bytes= 200M
  # for i in `seq 100`
  # do
  #     cgexec -g memory:A   malloc 10 &
  # done

The main problem here is that all processes still race for the mutex and
there is no guarantee that we will get counter back to 0 for those that
got back to mem_cgroup_handle_oom.  In the end the whole convoy
in/decreases the counter but we do not get to 1 that would enable
killing so nothing useful can be done.  The time is basically unbounded
because it highly depends on scheduling and ordering on mutex (I have
seen this taking hours...).

This patch replaces the counter by a simple {un}lock semantic.  As
mem_cgroup_oom_{un}lock works on the a subtree of a hierarchy we have to
make sure that nobody else races with us which is guaranteed by the
memcg_oom_mutex.

We have to be careful while locking subtrees because we can encounter a
subtree which is already locked: hierarchy:

          A
        /   \
       B     \
      /\      \
     C  D     E

B - C - D tree might be already locked.  While we want to enable locking
E subtree because OOM situations cannot influence each other we
definitely do not want to allow locking A.

Therefore we have to refuse lock if any subtree is already locked and
clear up the lock for all nodes that have been set up to the failure
point.

On the other hand we have to make sure that the rest of the world will
recognize that a group is under OOM even though it doesn't have a lock.
Therefore we have to introduce under_oom variable which is incremented
and decremented for the whole subtree when we enter resp.  leave
mem_cgroup_handle_oom.  under_oom, unlike oom_lock, doesn't need be
updated under memcg_oom_mutex because its users only check a single
group and they use atomic operations for that.

This can be checked easily by the following test case:

  # cgcreate -g memory:A
  # cgset -r memory.use_hierarchy=1 A
  # cgset -r memory.oom_control=1   A
  # cgset -r memory.limit_in_bytes= 100M
  # cgset -r memory.memsw.limit_in_bytes= 100M
  # cgcreate -g memory:A/B
  # cgset -r memory.oom_control=1 A/B
  # cgset -r memory.limit_in_bytes=20M
  # cgset -r memory.memsw.limit_in_bytes=20M
  # cgexec -g memory:A/B malloc 30  &    #->this will be blocked by OOM of group B
  # cgexec -g memory:A   malloc 80  &    #->this will be blocked by OOM of group A

While B gets oom_lock A will not get it.  Both of them go into sleep and
wait for an external action.  We can make the limit higher for A to
enforce waking it up

  # cgset -r memory.memsw.limit_in_bytes=300M A
  # cgset -r memory.limit_in_bytes=300M A

malloc in A has to wake up even though it doesn't have oom_lock.

Finally, the unlock path is very easy because we always unlock only the
subtree we have locked previously while we always decrement under_oom.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Balbir Singh <bsingharora@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: consolidate memory cgroup lru stat functions
KAMEZAWA Hiroyuki [Tue, 26 Jul 2011 23:08:22 +0000 (16:08 -0700)]
memcg: consolidate memory cgroup lru stat functions

In mm/memcontrol.c, there are many lru stat functions as..

  mem_cgroup_zone_nr_lru_pages
  mem_cgroup_node_nr_file_lru_pages
  mem_cgroup_nr_file_lru_pages
  mem_cgroup_node_nr_anon_lru_pages
  mem_cgroup_nr_anon_lru_pages
  mem_cgroup_node_nr_unevictable_lru_pages
  mem_cgroup_nr_unevictable_lru_pages
  mem_cgroup_node_nr_lru_pages
  mem_cgroup_nr_lru_pages
  mem_cgroup_get_local_zonestat

Some of them are under #ifdef MAX_NUMNODES >1 and others are not.
This seems bad. This patch consolidates all functions into

  mem_cgroup_zone_nr_lru_pages()
  mem_cgroup_node_nr_lru_pages()
  mem_cgroup_nr_lru_pages()

For these functions, "which LRU?" information is passed by a mask.

example:
  mem_cgroup_nr_lru_pages(mem, BIT(LRU_ACTIVE_ANON))

And I added some macro as ALL_LRU, ALL_LRU_FILE, ALL_LRU_ANON.

example:
  mem_cgroup_nr_lru_pages(mem, ALL_LRU)

BTW, considering layout of NUMA memory placement of counters, this patch seems
to be better.

Now, when we gather all LRU information, we scan in following orer
    for_each_lru -> for_each_node -> for_each_zone.

This means we'll touch cache lines in different node in turn.

After patch, we'll scan
    for_each_node -> for_each_zone -> for_each_lru(mask)

Then, we'll gather information in the same cacheline at once.

[akpm@linux-foundation.org: fix warnigns, build error]
Signed-off-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>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: export memory cgroup's swappiness with mem_cgroup_swappiness()
KAMEZAWA Hiroyuki [Tue, 26 Jul 2011 23:08:21 +0000 (16:08 -0700)]
memcg: export memory cgroup's swappiness with mem_cgroup_swappiness()

Each memory cgroup has a 'swappiness' value which can be accessed by
get_swappiness(memcg).  The major user is try_to_free_mem_cgroup_pages()
and swappiness is passed by argument.  It's propagated by scan_control.

get_swappiness() is a static function but some planned updates will need
to get swappiness from files other than memcontrol.c This patch exports
get_swappiness() as mem_cgroup_swappiness().  With this, we can remove the
argument of swapiness from try_to_free...  and drop swappiness from
scan_control.  only memcg uses it.

Signed-off-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>
Cc: Ying Han <yinghan@google.com>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agortc: fix hrtimer deadlock
Thomas Gleixner [Tue, 26 Jul 2011 23:08:20 +0000 (16:08 -0700)]
rtc: fix hrtimer deadlock

Ben reported a lockup related to rtc. The lockup happens due to:

CPU0                                        CPU1

rtc_irq_set_state()     __run_hrtimer()
  spin_lock_irqsave(&rtc->irq_task_lock)    rtc_handle_legacy_irq();
      spin_lock(&rtc->irq_task_lock);
  hrtimer_cancel()
    while (callback_running);

So the running callback never finishes as it's blocked on
rtc->irq_task_lock.

Use hrtimer_try_to_cancel() instead and drop rtc->irq_task_lock while
waiting for the callback.  Fix this for both rtc_irq_set_state() and
rtc_irq_set_freq().

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reported-by: Ben Greear <greearb@candelatech.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agortc: limit frequency
Thomas Gleixner [Tue, 26 Jul 2011 23:08:19 +0000 (16:08 -0700)]
rtc: limit frequency

Due to the hrtimer self rearming mode a user can DoS the machine simply
because it's starved by hrtimer events.

The RTC hrtimer is self rearming.  We really need to limit the frequency
to something sensible.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Ben Greear <greearb@candelatech.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agortc: handle errors correctly in rtc_irq_set_state()
Thomas Gleixner [Tue, 26 Jul 2011 23:08:18 +0000 (16:08 -0700)]
rtc: handle errors correctly in rtc_irq_set_state()

The code checks the correctness of the parameters, but unconditionally
arms/disarms the hrtimer.

The result is that a random task might arm/disarm rtc timer and surprise
the real owner by either generating events or by stopping them.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Ben Greear <greearb@candelatech.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomn10300, exec: remove redundant set_fs(USER_DS)
Mathias Krause [Tue, 26 Jul 2011 23:08:17 +0000 (16:08 -0700)]
mn10300, exec: remove redundant set_fs(USER_DS)

The address limit is already set in flush_old_exec() so this
set_fs(USER_DS) is redundant.

Signed-off-by: Mathias Krause <minipli@googlemail.com>
Cc: Koichi Yasutake <yasutake.koichi@jp.panasonic.com>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agodrivers/base/power/opp.c: fix dev_opp initial value
Jonghwan Choi [Tue, 26 Jul 2011 23:08:16 +0000 (16:08 -0700)]
drivers/base/power/opp.c: fix dev_opp initial value

Dev_opp initial value shoule be ERR_PTR(), IS_ERR() is used to check
error.

Signed-off-by: Jonghwan Choi <jhbird.choi@samsung.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agofrv, exec: remove redundant set_fs(USER_DS)
Mathias Krause [Tue, 26 Jul 2011 23:08:15 +0000 (16:08 -0700)]
frv, exec: remove redundant set_fs(USER_DS)

The address limit is already set in flush_old_exec() so those calls to
set_fs(USER_DS) are redundant.

Also removed the dead code in flush_thread().

Signed-off-by: Mathias Krause <minipli@googlemail.com>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge branch 'upstream' of git://git.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Tue, 26 Jul 2011 21:17:28 +0000 (14:17 -0700)]
Merge branch 'upstream' of git://git.linux-mips.org/upstream-linus

* 'upstream' of git://git.linux-mips.org/pub/scm/upstream-linus: (31 commits)
  MIPS: Close races in TLB modify handlers.
  MIPS: Add uasm UASM_i_SRL_SAFE macro.
  MIPS: RB532: Use hex_to_bin()
  MIPS: Enable cpu_has_clo_clz for MIPS Technologies' platforms
  MIPS: PowerTV: Provide cpu-feature-overrides.h
  MIPS: Remove pointless return statement from empty void functions.
  MIPS: Limit fixrange_init() to the FIXMAP region
  MIPS: Install handlers for software IRQs
  MIPS: Move FIXADDR_TOP into spaces.h
  MIPS: Add SYNC after cacheflush
  MIPS: pfn_valid() is broken on low memory HIGHMEM systems
  MIPS: HIGHMEM DMA on noncoherent MIPS32 processors
  MIPS: topdown mmap support
  MIPS: Remove redundant addr_limit assignment on exec.
  MIPS: AR7: Replace __attribute__((__packed__)) with __packed
  MIPS: AR7: Remove 'space before tabs' in platform.c
  MIPS: Lantiq: Add missing clk_enable and clk_disable functions.
  MIPS: AR7: Fix trailing semicolon bug in clock.c
  MAINTAINERS: Update MIPS entry.
  MIPS: BCM63xx: Remove duplicate PERF_IRQSTAT_REG definition
  ...

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph...
Linus Torvalds [Tue, 26 Jul 2011 20:38:50 +0000 (13:38 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/sage/ceph-client

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client: (23 commits)
  ceph: document unlocked d_parent accesses
  ceph: explicitly reference rename old_dentry parent dir in request
  ceph: document locking for ceph_set_dentry_offset
  ceph: avoid d_parent in ceph_dentry_hash; fix ceph_encode_fh() hashing bug
  ceph: protect d_parent access in ceph_d_revalidate
  ceph: protect access to d_parent
  ceph: handle racing calls to ceph_init_dentry
  ceph: set dir complete frag after adding capability
  rbd: set blk_queue request sizes to object size
  ceph: set up readahead size when rsize is not passed
  rbd: cancel watch request when releasing the device
  ceph: ignore lease mask
  ceph: fix ceph_lookup_open intent usage
  ceph: only link open operations to directory unsafe list if O_CREAT|O_TRUNC
  ceph: fix bad parent_inode calc in ceph_lookup_open
  ceph: avoid carrying Fw cap during write into page cache
  libceph: don't time out osd requests that haven't been received
  ceph: report f_bfree based on kb_avail rather than diffing.
  ceph: only queue capsnap if caps are dirty
  ceph: fix snap writeback when racing with writes
  ...

13 years agogma500: udelay(20000) it too long again
Stephen Rothwell [Mon, 25 Jul 2011 05:18:44 +0000 (15:18 +1000)]
gma500: udelay(20000) it too long again

so replace it with mdelay(20).

Fixes build error:

  ERROR: "__bad_udelay" [drivers/staging/gma500/psb_gfx.ko] undefined!

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoUSB / Renesas: Fix build issue related to struct scatterlist
Rafael J. Wysocki [Tue, 26 Jul 2011 18:51:01 +0000 (20:51 +0200)]
USB / Renesas: Fix build issue related to struct scatterlist

Fix build issue caused by undefined struct scatterlist in
drivers/usb/renesas_usbhs/fifo.c.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMMC / TMIO: Fix build issue related to struct scatterlist
Rafael J. Wysocki [Tue, 26 Jul 2011 18:50:23 +0000 (20:50 +0200)]
MMC / TMIO: Fix build issue related to struct scatterlist

Fix build issue caused by undefined struct scatterlist in
drivers/mmc/host/tmio_mmc.c.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux...
Linus Torvalds [Tue, 26 Jul 2011 18:34:40 +0000 (11:34 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/jack/linux-fs-2.6

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs-2.6:
  jbd: change the field "b_cow_tid" of struct journal_head from type unsigned to tid_t
  ext3.txt: update the links in the section "useful links" to the latest ones
  ext3: Fix data corruption in inodes with journalled data
  ext2: check xattr name_len before acquiring xattr_sem in ext2_xattr_get
  ext3: Fix compilation with -DDX_DEBUG
  quota: Remove unused declaration
  jbd: Use WRITE_SYNC in journal checkpoint.
  jbd: Fix oops in journal_remove_journal_head()
  ext3: Return -EINVAL when start is beyond the end of fs in ext3_trim_fs()
  ext3/ioctl.c: silence sparse warnings about different address spaces
  ext3/ext4 Documentation: remove bh/nobh since it has been deprecated
  ext3: Improve truncate error handling
  ext3: use proper little-endian bitops
  ext2: include fs.h into ext2_fs.h
  ext3: Fix oops in ext3_try_to_allocate_with_rsv()
  jbd: fix a bug of leaking jh->b_jcount
  jbd: remove dependency on __GFP_NOFAIL
  ext3: Convert ext3 to new truncate calling convention
  jbd: Add fixed tracepoints
  ext3: Add fixed tracepoints

Resolve conflicts in fs/ext3/fsync.c due to fsync locking push-down and
new fixed tracepoints.

13 years agoceph: document unlocked d_parent accesses
Sage Weil [Tue, 26 Jul 2011 18:31:26 +0000 (11:31 -0700)]
ceph: document unlocked d_parent accesses

For the most part we don't care about racing with rename when directing
MDS requests; either the old or new parent is fine.  Document that, and
do some minor cleanup.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: explicitly reference rename old_dentry parent dir in request
Sage Weil [Tue, 26 Jul 2011 18:31:14 +0000 (11:31 -0700)]
ceph: explicitly reference rename old_dentry parent dir in request

We carry a pin on the parent directory for the rename source and dest
dentries.  For the source it's r_locked_dir; we need to explicitly
reference the old_dentry parent as well, since the dentry's d_parent may
change between when the request was created and pinned and when it is
freed.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: document locking for ceph_set_dentry_offset
Sage Weil [Tue, 26 Jul 2011 18:31:08 +0000 (11:31 -0700)]
ceph: document locking for ceph_set_dentry_offset

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: avoid d_parent in ceph_dentry_hash; fix ceph_encode_fh() hashing bug
Sage Weil [Tue, 26 Jul 2011 18:30:55 +0000 (11:30 -0700)]
ceph: avoid d_parent in ceph_dentry_hash; fix ceph_encode_fh() hashing bug

Have caller pass in a safely-obtained reference to the parent directory
for calculating a dentry's hash valud.

While we're here, simpify the flow through ceph_encode_fh() so that there
is a single exit point and cleanup.

Also fix a bug with the dentry hash calculation: calculate the hash for the
dentry we were given, not its parent.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: protect d_parent access in ceph_d_revalidate
Sage Weil [Tue, 26 Jul 2011 18:30:43 +0000 (11:30 -0700)]
ceph: protect d_parent access in ceph_d_revalidate

Protect d_parent with d_lock.  Carry a reference.  Simplify the flow so
that there is a single exit point and cleanup.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: protect access to d_parent
Sage Weil [Tue, 26 Jul 2011 18:30:29 +0000 (11:30 -0700)]
ceph: protect access to d_parent

d_parent is protected by d_lock: use it when looking up a dentry's parent
directory inode.  Also take a reference and drop it in the caller to avoid
a use-after-free.

Reported-by: Al Viro <viro@ZenIV.linux.org.uk>
Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: handle racing calls to ceph_init_dentry
Sage Weil [Tue, 26 Jul 2011 18:30:15 +0000 (11:30 -0700)]
ceph: handle racing calls to ceph_init_dentry

The ->lookup() and prepopulate_readdir() callers are working with unhashed
dentries, so we don't have to worry.  The export.c callers, though, need
to initialize something they got back from d_obtain_alias() and are
potentially racing with other callers.  Make sure we don't return unless
the dentry is properly initialized (by us or someone else).

Reported-by: Al Viro <viro@ZenIV.linux.org.uk>
Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: set dir complete frag after adding capability
Sage Weil [Tue, 26 Jul 2011 18:30:02 +0000 (11:30 -0700)]
ceph: set dir complete frag after adding capability

Curretly ceph_add_cap clears the complete bit if we are newly issued the
FILE_SHARED cap, which is normally the case for a newly issue cap on a new
directory.  That means we clear the just-set bit.  Move the check that sets
the flag to after the cap is added/updated.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agorbd: set blk_queue request sizes to object size
Josh Durgin [Fri, 22 Jul 2011 18:35:23 +0000 (11:35 -0700)]
rbd: set blk_queue request sizes to object size

This improves performance since more requests can be merged.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Josh Durgin <josh.durgin@dreamhost.com>
13 years agoceph: set up readahead size when rsize is not passed
Yehuda Sadeh [Fri, 22 Jul 2011 18:12:28 +0000 (11:12 -0700)]
ceph: set up readahead size when rsize is not passed

This should improve the default read performance, as without it
readahead is practically disabled.

Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
13 years agorbd: cancel watch request when releasing the device
Yehuda Sadeh [Tue, 12 Jul 2011 23:56:57 +0000 (16:56 -0700)]
rbd: cancel watch request when releasing the device

We were missing this cleanup, so when a device was released
the osd didn't clean up its watchers list, so following notifications
could be slow as osd needed to timeout on the client.

Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
13 years agoceph: ignore lease mask
Sage Weil [Tue, 26 Jul 2011 18:28:25 +0000 (11:28 -0700)]
ceph: ignore lease mask

The lease mask is no longer used (and it changed a while back).  Instead,
use a non-zero duration to indicate that there is a lease being issued.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: fix ceph_lookup_open intent usage
Sage Weil [Tue, 26 Jul 2011 18:28:11 +0000 (11:28 -0700)]
ceph: fix ceph_lookup_open intent usage

We weren't properly calling lookup_instantiate_filp when setting up the
lookup intent, which could lead to file leakage on errors.  So:

 - use separate helper for the hidden snapdir translation, immediately
   following the mds request
 - use ceph_finish_lookup for the final dentry/return value dance in the
   exit path
 - lookup_instantiate_filp on success

Reported-by: Al Viro <viro@ZenIV.linux.org.uk>
Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: only link open operations to directory unsafe list if O_CREAT|O_TRUNC
Sage Weil [Tue, 26 Jul 2011 18:27:59 +0000 (11:27 -0700)]
ceph: only link open operations to directory unsafe list if O_CREAT|O_TRUNC

We only need to put these on the directory unsafe list if they have
side effects that fsync(2) should flush out.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: fix bad parent_inode calc in ceph_lookup_open
Sage Weil [Tue, 26 Jul 2011 18:27:48 +0000 (11:27 -0700)]
ceph: fix bad parent_inode calc in ceph_lookup_open

We were always getting NULL here because the intent file f_dentry is always
NULL at this point, which means we were always passing NULL to
ceph_mdsc_do_request.  In reality, this was fine, since this isn't
currently ever a write operation that needs to get strung on the dir's
unsafe list.

Use the dir explicitly, and only pass it if this open has side-effects that
a dir fsync should flush.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: avoid carrying Fw cap during write into page cache
Sage Weil [Tue, 26 Jul 2011 18:27:34 +0000 (11:27 -0700)]
ceph: avoid carrying Fw cap during write into page cache

The generic_file_aio_write call may block on balance_dirty_pages while we
flush data to the OSDs.  If we hold a reference to the FILE_WR cap during
that interval revocation by the MDS (e.g., to do a stat(2)) may be very
slow.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agolibceph: don't time out osd requests that haven't been received
Sage Weil [Tue, 26 Jul 2011 18:27:24 +0000 (11:27 -0700)]
libceph: don't time out osd requests that haven't been received

Keep track of when an outgoing message is ACKed (i.e., the server fully
received it and, presumably, queued it for processing).  Time out OSD
requests only if it's been too long since they've been received.

This prevents timeouts and connection thrashing when the OSDs are simply
busy and are throttling the requests they read off the network.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: report f_bfree based on kb_avail rather than diffing.
Greg Farnum [Tue, 26 Jul 2011 18:26:54 +0000 (11:26 -0700)]
ceph: report f_bfree based on kb_avail rather than diffing.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
13 years agoceph: only queue capsnap if caps are dirty
Sage Weil [Tue, 26 Jul 2011 18:26:41 +0000 (11:26 -0700)]
ceph: only queue capsnap if caps are dirty

We used to go into this branch if i_wrbuffer_ref_head was non-zero.  This
was an ancient check from before we were careful about dealing with all
kinds of caps (and not just dirty pages).  It is cleaner to only queue a
capsnap if there is an actual dirty cap.  If we are racing with...
something...we will end up here with ci->i_wrbuffer_refs but no dirty
caps.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: fix snap writeback when racing with writes
Sage Weil [Tue, 26 Jul 2011 18:26:31 +0000 (11:26 -0700)]
ceph: fix snap writeback when racing with writes

There are two problems that come up when we try to queue a capsnap while a
write is in progress:

 - The FILE_WR cap is held, but not yet dirty, so we may queue a capsnap
   with dirty == 0.  That will crash later in __ceph_flush_snaps().  Or
   on the FILE_WR cap if a write is in progress.
 - We may not have i_head_snapc set, which causes problems pretty quickly.
   Look to the snaprealm in this case.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: use flag bit for at_end readdir flag
Sage Weil [Tue, 26 Jul 2011 18:26:18 +0000 (11:26 -0700)]
ceph: use flag bit for at_end readdir flag

This saves us a word of memory per file.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: add F_SYNC file flag to force sync (non-O_DIRECT) io
Sage Weil [Tue, 26 Jul 2011 18:26:07 +0000 (11:26 -0700)]
ceph: add F_SYNC file flag to force sync (non-O_DIRECT) io

This allows us to force IO through the sync path which you normally only
get when multiple clients are reading/writing to the same file or by
mounting with -o sync.  Among other things, this lets test programs verify
correctness with a single mount.

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoceph: add flags field to file_info
Sage Weil [Tue, 26 Jul 2011 18:25:27 +0000 (11:25 -0700)]
ceph: add flags field to file_info

Reviewed-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoMerge branch 'x86-olpc-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 26 Jul 2011 18:11:54 +0000 (11:11 -0700)]
Merge branch 'x86-olpc-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-olpc-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86, olpc-xo15-sci: Enable EC wakeup capability
  x86, olpc: Fix dependency on POWER_SUPPLY
  x86, olpc: Add XO-1.5 SCI driver
  x86, olpc: Add XO-1 RTC driver
  x86, olpc-xo1-sci: Propagate power supply/battery events
  x86, olpc-xo1-sci: Add lid switch functionality
  x86, olpc-xo1-sci: Add GPE handler and ebook switch functionality
  x86, olpc: EC SCI wakeup mask functionality
  x86, olpc: Add XO-1 SCI driver and power button control
  x86, olpc: Add XO-1 suspend/resume support
  x86, olpc: Rename olpc-xo1 to olpc-xo1-pm
  x86, olpc: Move CS5536-related constants to cs5535.h
  x86, olpc: Add missing elements to device tree

13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Tue, 26 Jul 2011 18:11:28 +0000 (11:11 -0700)]
Merge git://git./linux/kernel/git/sfrench/cifs-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6:
  cifs: Cleanup: check return codes of crypto api calls
  CIFS: Fix oops while mounting with prefixpath
  [CIFS] Redundant null check after dereference
  cifs: use cifs_dirent in cifs_save_resume_key
  cifs: use cifs_dirent to replace cifs_get_name_from_search_buf
  cifs: introduce cifs_dirent
  cifs: cleanup cifs_filldir

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt...
Linus Torvalds [Tue, 26 Jul 2011 18:10:56 +0000 (11:10 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/rostedt/linux-2.6-ktest

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-ktest:
  ktest: Fix bug when ADD_CONFIG is set but MIN_CONFIG is not
  ktest: Keep fonud configs separate from default configs
  ktest: Add prompt to use OUTPUT_MIN_CONFIG
  ktest: Use Kconfig dependencies to shorten time to make min_config
  ktest: Add test type make_min_config
  ktest: Require one TEST_START in config file
  ktest: Add helper function to avoid duplicate code
  ktest: Add IGNORE_WARNINGS to ignore warnings in some patches
  ktest: Fix tar extracting of modules to target
  ktest: Have the testing tmp dir include machine name
  ktest: Add POST/PRE_BUILD options
  ktest: Allow initrd processing without modules defined
  ktest: Have LOG_FILE evaluate options as well
  ktest: Have wait on stdio honor bug timeout
  ktest: Implement our own force min config
  ktest: Add TEST_NAME option
  ktest: Add CONFIG_BISECT_GOOD option
  ktest: Add detection of triple faults
  ktest: Notify reason to break out of monitoring boot

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/wfg/writeback
Linus Torvalds [Tue, 26 Jul 2011 17:39:54 +0000 (10:39 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/wfg/writeback

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/wfg/writeback: (27 commits)
  mm: properly reflect task dirty limits in dirty_exceeded logic
  writeback: don't busy retry writeback on new/freeing inodes
  writeback: scale IO chunk size up to half device bandwidth
  writeback: trace global_dirty_state
  writeback: introduce max-pause and pass-good dirty limits
  writeback: introduce smoothed global dirty limit
  writeback: consolidate variable names in balance_dirty_pages()
  writeback: show bdi write bandwidth in debugfs
  writeback: bdi write bandwidth estimation
  writeback: account per-bdi accumulated written pages
  writeback: make writeback_control.nr_to_write straight
  writeback: skip tmpfs early in balance_dirty_pages_ratelimited_nr()
  writeback: trace event writeback_queue_io
  writeback: trace event writeback_single_inode
  writeback: remove .nonblocking and .encountered_congestion
  writeback: remove writeback_control.more_io
  writeback: skip balance_dirty_pages() for in-memory fs
  writeback: add bdi_dirty_limit() kernel-doc
  writeback: avoid extra sync work at enqueue time
  writeback: elevate queue_io() into wb_writeback()
  ...

Fix up trivial conflicts in fs/fs-writeback.c and mm/filemap.c

13 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-block
Linus Torvalds [Tue, 26 Jul 2011 17:02:58 +0000 (10:02 -0700)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-block

* 'for-linus' of git://git.kernel.dk/linux-block:
  block: fix warning with calling smp_processor_id() in preemptible section

13 years agoMerge branch 'drm-core-next' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Tue, 26 Jul 2011 16:21:09 +0000 (09:21 -0700)]
Merge branch 'drm-core-next' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-core-next' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (135 commits)
  drm/radeon/kms: fix DP training for DPEncoderService revision bigger than 1.1
  drm/radeon/kms: add missing vddci setting on NI+
  drm/radeon: Add a rmb() in IH processing
  drm/radeon: ATOM Endian fix for atombios_crtc_program_pll()
  drm/radeon: Fix the definition of RADEON_BUF_SWAP_32BIT
  drm/radeon: Do an MMIO read on interrupts when not uisng MSIs
  drm/radeon: Writeback endian fixes
  drm/radeon: Remove a bunch of useless _iomem casts
  drm/gem: add support for private objects
  DRM: clean up and document parsing of video= parameter
  DRM: Radeon: Fix section mismatch.
  drm: really make debug levels match in edid failure code
  drm/radeon/kms: fix i2c map for rv250/280
  drm/nouveau/gr: disable fifo access and idle before suspend ctx unload
  drm/nouveau: pass flag to engine fini() method on suspend
  drm/nouveau: replace nv04_graph_fifo_access() use with direct reg bashing
  drm/nv40/gr: rewrite/split context takedown functions
  drm/nouveau: detect disabled device in irq handler and return IRQ_NONE
  drm/nouveau: ignore connector type when deciding digital/analog on DVI-I
  drm/nouveau: Add a quirk for Gigabyte NX86T
  ...

13 years agoblock: fix warning with calling smp_processor_id() in preemptible section
Jens Axboe [Tue, 26 Jul 2011 13:01:15 +0000 (15:01 +0200)]
block: fix warning with calling smp_processor_id() in preemptible section

After commit 5757a6d7 introduced an unsafe calling of
smp_processor_id(), with preempt debuggin turned on we spew a lot of:

BUG: using smp_processor_id() in preemptible [00000000] code: kjournald/514
caller is __make_request+0x1b8/0x308
[<c0019f44>] (unwind_backtrace+0x0/0xe8) from [<c024b4cc>] (debug_smp_processor_id+0xbc/0xf0)
[<c024b4cc>] (debug_smp_processor_id+0xbc/0xf0) from [<c0223d14>] (__make_request+0x1b8/0x308)
[<c0223d14>] (__make_request+0x1b8/0x308) from [<c02215ac>] (generic_make_request+0x4dc/0x558)
[<c02215ac>] (generic_make_request+0x4dc/0x558) from [<c022173c>] (submit_bio+0x114/0x138)
[<c022173c>] (submit_bio+0x114/0x138) from [<c011f504>] (submit_bh+0x148/0x16c)
[<c011f504>] (submit_bh+0x148/0x16c) from [<c0121ed8>] (__sync_dirty_buffer+0x88/0xd8)
[<c0121ed8>] (__sync_dirty_buffer+0x88/0xd8) from [<c01aff78>] (journal_commit_transaction+0x1198/0x1688)
[<c01aff78>] (journal_commit_transaction+0x1198/0x1688) from [<c01b4034>] (kjournald+0xb4/0x224)
[<c01b4034>] (kjournald+0xb4/0x224) from [<c0069ea0>] (kthread+0x8c/0x94)
[<c0069ea0>] (kthread+0x8c/0x94) from [<c00137f8>] (kernel_thread_exit+0x0/0x8)

Fix this by just using raw_smp_processor_id(), it's just a hint
after all. There's no pinning of the CPU or accessing per-cpu
structures involved.

Reported-by: Ming Lei <tom.leiming@gmail.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
13 years agodrm/radeon/kms: fix DP training for DPEncoderService revision bigger than 1.1
Jerome Glisse [Mon, 25 Jul 2011 15:57:43 +0000 (11:57 -0400)]
drm/radeon/kms: fix DP training for DPEncoderService revision bigger than 1.1

DPEncoderService newer than 1.1 can't properly program the DP (display port)
link training. When facing such version use the DIGxEncoderControl method
instead. Fix DP link training on some R7XX.

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agodrm/radeon/kms: add missing vddci setting on NI+
Alex Deucher [Mon, 25 Jul 2011 22:50:08 +0000 (18:50 -0400)]
drm/radeon/kms: add missing vddci setting on NI+

Need to add vddci setting to pm init as well as
resume.  Fixes hangs on load on some boards.

Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=38754

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agop9: avoid unused variable warning
Linus Torvalds [Tue, 26 Jul 2011 06:43:53 +0000 (23:43 -0700)]
p9: avoid unused variable warning

Commit 4e34e719e457 ("fs: take the ACL checks to common code") removed
the use of the 'acl' variable in v9fs_iop_get_acl(), but left the
variable definition around.  Remove it to get rid of the warning:

  fs/9p/acl.c: In function ‘v9fs_iop_get_acl’:
  fs/9p/acl.c:101:20: warning: unused variable ‘acl’

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge branch 'staging-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Tue, 26 Jul 2011 06:26:34 +0000 (23:26 -0700)]
Merge branch 'staging-next' of git://git./linux/kernel/git/gregkh/staging-2.6

* 'staging-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging-2.6: (741 commits)
  staging:iio:meter:ade7753 should be 16 bit read not 8 bit for mode register.
  staging:iio:kfifo_buf fix double initialization of the ring device structure.
  staging:iio:accel:lis3l02dq: fix incorrect pointer passed to spi_set_drvdata.
  staging:iio:imu fix missing register table index for some channels
  spectra: enable device before poking it
  staging: rts_pstor: Fix a miswriting
  staging/lirc_bt829: Return -ENODEV when no hardware is found.
  staging/lirc_parallel: remove pointless prototypes.
  staging/lirc_parallel: fix panic on rmmod
  staging:iio:adc:ad7476: Incorrect pointer into spi_set_drvdata.
  Staging: zram: Fix kunmapping order
  Revert "gma500: Fix dependencies"
  gma500: Add medfield header
  gma500: wire up the mrst i2c bus from chip_info
  gma500: Fix DPU build
  gma500: Clean up the DPU config and make it runtime
  gma500: resync with Medfield progress
  gma500: Use the mrst helpers and power control for mode commit
  gma500@ Fix backlight range error
  gma500: More Moorestown muddle meddling means MM maybe might modeset
  ...

Fix up fairly trivial conflicts all over, mostly due to header file
cleanup conflicts, but some deleted files and some just context changes:
 - Documentation/feature-removal-schedule.txt
 - drivers/staging/bcm/headers.h
 - drivers/staging/brcm80211/brcmfmac/dhd_linux.c
 - drivers/staging/brcm80211/brcmfmac/dhd_sdio.c
 - drivers/staging/brcm80211/brcmfmac/wl_cfg80211.h
 - drivers/staging/brcm80211/brcmfmac/wl_iw.c
 - drivers/staging/et131x/et131x_netdev.c
 - drivers/staging/rtl8187se/ieee80211/ieee80211_softmac.c
 - drivers/staging/rtl8192e/r8192E.h
 - drivers/staging/usbip/userspace/src/utils.h

13 years agoMerge branch 'tty-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty-2.6
Linus Torvalds [Tue, 26 Jul 2011 06:09:27 +0000 (23:09 -0700)]
Merge branch 'tty-next' of git://git./linux/kernel/git/gregkh/tty-2.6

* 'tty-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty-2.6: (26 commits)
  amba pl011: workaround for uart registers lockup
  n_gsm: fix the wrong FCS handling
  pch_uart: add missing comment about OKI ML7223
  pch_uart: Add MSI support
  tty: fix "IRQ45: nobody cared"
  PTI feature to allow user to name and mark masterchannel request.
  0 for o PTI Makefile bug.
  tty: serial: samsung.c remove legacy PM code.
  SERIAL: SC26xx: Fix link error.
  serial: mrst_max3110: initialize waitqueue earlier
  mrst_max3110: Change max missing message priority.
  tty: s5pv210: Add delay loop on fifo reset function for UART
  tty/serial: Fix XSCALE serial ports, e.g. ce4100
  serial: bfin_5xx: fix off-by-one with resource size
  drivers/tty: use printk_ratelimited() instead of printk_ratelimit()
  tty: n_gsm: Added refcount usage to gsm_mux and gsm_dlci structs
  tty: n_gsm: Add raw-ip support
  tty: n_gsm: expose gsmtty device nodes at ldisc open time
  pch_phub: Fix register miss-setting issue
  serial: 8250, increase PASS_LIMIT
  ...

13 years agoMerge branch 'usb-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6
Linus Torvalds [Tue, 26 Jul 2011 06:08:32 +0000 (23:08 -0700)]
Merge branch 'usb-next' of git://git./linux/kernel/git/gregkh/usb-2.6

* 'usb-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6: (115 commits)
  EHCI: fix direction handling for interrupt data toggles
  USB: serial: add IDs for WinChipHead USB->RS232 adapter
  USB: OHCI: fix another regression for NVIDIA controllers
  usb: gadget: m66592-udc: add pullup function
  usb: gadget: m66592-udc: add function for external controller
  usb: gadget: r8a66597-udc: add pullup function
  usb: renesas_usbhs: support multi driver
  usb: renesas_usbhs: inaccessible pipe is not an error
  usb: renesas_usbhs: care buff alignment when dma handler
  USB: PL2303: correctly handle baudrates above 115200
  usb: r8a66597-hcd: fixup USB_PORT_STAT_C_SUSPEND shift
  usb: renesas_usbhs: compile/config are rescued
  usb: renesas_usbhs: fixup comment-out
  usb: update email address in ohci-sh and r8a66597-hcd
  usb: r8a66597-hcd: add function for external controller
  EHCI: only power off port if over-current is active
  USB: mon: Allow to use usbmon without debugfs
  USB: EHCI: go back to using the system clock for QH unlinks
  ehci: add pci quirk for Ordissimo and RM Slate 100 too
  ehci: refactor pci quirk to use standard dmi_check_system method
  ...

Fix up trivial conflicts in Documentation/feature-removal-schedule.txt

13 years agoMerge branch 'driver-core-next' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 26 Jul 2011 06:06:24 +0000 (23:06 -0700)]
Merge branch 'driver-core-next' of git://git./linux/kernel/git/gregkh/driver-core-2.6

* 'driver-core-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6:
  updated Documentation/ja_JP/SubmittingPatches
  debugfs: add documentation for debugfs_create_x64
  uio: uio_pdrv_genirq: Add OF support
  firmware: gsmi: remove sysfs entries when unload the module
  Documentation/zh_CN: Fix messy code file email-clients.txt
  driver core: add more help description for "path to uevent helper"
  driver-core: modify FIRMWARE_IN_KERNEL help message
  driver-core: Kconfig grammar corrections in firmware configuration
  DOCUMENTATION: Replace create_device() with device_create().
  DOCUMENTATION: Update overview.txt in Doc/driver-model.
  pti: pti_tty_install documentation mispelling.