kernel/kernel-generic.git
15 years agoceph: nfs re-export support
Sage Weil [Tue, 6 Oct 2009 18:31:13 +0000 (11:31 -0700)]
ceph: nfs re-export support

Basic NFS re-export support is included.  This mostly works.  However,
Ceph's MDS design precludes the ability to generate a (small)
filehandle that will be valid forever, so this is of limited utility.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: message pools
Sage Weil [Tue, 6 Oct 2009 18:31:13 +0000 (11:31 -0700)]
ceph: message pools

The msgpool is a basic mempool_t-like structure to preallocate
messages we expect to receive over the wire.  This ensures we have the
necessary memory preallocated to process replies to requests, or to
process unsolicited messages from various servers.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: messenger library
Sage Weil [Tue, 6 Oct 2009 18:31:13 +0000 (11:31 -0700)]
ceph: messenger library

A generic message passing library is used to communicate with all
other components in the Ceph file system.  The messenger library
provides ordered, reliable delivery of messages between two nodes in
the system.

This implementation is based on TCP.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: snapshot management
Sage Weil [Tue, 6 Oct 2009 18:31:12 +0000 (11:31 -0700)]
ceph: snapshot management

Ceph snapshots rely on client cooperation in determining which
operations apply to which snapshots, and appropriately flushing
snapshotted data and metadata back to the OSD and MDS clusters.
Because snapshots apply to subtrees of the file hierarchy and can be
created at any time, there is a fair bit of bookkeeping required to
make this work.

Portions of the hierarchy that belong to the same set of snapshots
are described by a single 'snap realm.'  A 'snap context' describes
the set of snapshots that exist for a given file or directory.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: capability management
Sage Weil [Tue, 6 Oct 2009 18:31:12 +0000 (11:31 -0700)]
ceph: capability management

The Ceph metadata servers control client access to inode metadata and
file data by issuing capabilities, granting clients permission to read
and/or write both inode field and file data to OSDs (storage nodes).
Each capability consists of a set of bits indicating which operations
are allowed.

If the client holds a *_SHARED cap, the client has a coherent value
that can be safely read from the cached inode.

In the case of a *_EXCL (exclusive) or FILE_WR capabilities, the client
is allowed to change inode attributes (e.g., file size, mtime), note
its dirty state in the ceph_cap, and asynchronously flush that
metadata change to the MDS.

In the event of a conflicting operation (perhaps by another client),
the MDS will revoke the conflicting client capabilities.

In order for a client to cache an inode, it must hold a capability
with at least one MDS server.  When inodes are released, release
notifications are batched and periodically sent en masse to the MDS
cluster to release server state.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: monitor client
Sage Weil [Tue, 6 Oct 2009 18:31:11 +0000 (11:31 -0700)]
ceph: monitor client

The monitor cluster is responsible for managing cluster membership
and state.  The monitor client handles what minimal interaction
the Ceph client has with it: checking for updated versions of the
MDS and OSD maps, getting statfs() information, and unmounting.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: CRUSH mapping algorithm
Sage Weil [Tue, 6 Oct 2009 18:31:11 +0000 (11:31 -0700)]
ceph: CRUSH mapping algorithm

CRUSH is a pseudorandom data distribution function designed to map
inputs onto a dynamic hierarchy of devices, while minimizing the
extent to which inputs are remapped when the devices are added or
removed.  It includes some features that are specifically useful for
storage, most notably the ability to map each input onto a set of N
devices that are separated across administrator-defined failure
domains.  CRUSH is used to distribute data across the cluster of Ceph
storage nodes.

More information about CRUSH can be found in this paper:

    http://www.ssrc.ucsc.edu/Papers/weil-sc06.pdf

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: OSD client
Sage Weil [Tue, 6 Oct 2009 18:31:10 +0000 (11:31 -0700)]
ceph: OSD client

The OSD client is responsible for reading and writing data from/to the
object storage pool.  This includes determining where objects are
stored in the cluster, and ensuring that requests are retried or
redirected in the event of a node failure or data migration.

If an OSD does not respond before a timeout expires, keepalive
messages are sent across the lossless, ordered communications channel
to ensure that any break in the TCP is discovered.  If the session
does reset, a reconnection is attempted and affected requests are
resent (by the message transport layer).

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: MDS client
Sage Weil [Tue, 6 Oct 2009 18:31:09 +0000 (11:31 -0700)]
ceph: MDS client

The MDS (metadata server) client is responsible for submitting
requests to the MDS cluster and parsing the response.  We decide which
MDS to submit each request to based on cached information about the
current partition of the directory hierarchy across the cluster.  A
stateful session is opened with each MDS before we submit requests to
it, and a mutex is used to control the ordering of messages within
each session.

An MDS request may generate two responses.  The first indicates the
operation was a success and returns any result.  A second reply is
sent when the operation commits to disk.  Note that locking on the MDS
ensures that the results of updates are visible only to the updating
client before the operation commits.  Requests are linked to the
containing directory so that an fsync will wait for them to commit.

If an MDS fails and/or recovers, we resubmit requests as needed.  We
also reconnect existing capabilities to a recovering MDS to
reestablish that shared session state.  Old dentry leases are
invalidated.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: address space operations
Sage Weil [Tue, 6 Oct 2009 18:31:09 +0000 (11:31 -0700)]
ceph: address space operations

The ceph address space methods are concerned primarily with managing
the dirty page accounting in the inode, which (among other things)
must keep track of which snapshot context each page was dirtied in,
and ensure that dirty data is written out to the OSDs in snapshort
order.

A writepage() on a page that is not currently writeable due to
snapshot writeback ordering constraints is ignored (it was presumably
called from kswapd).

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: file operations
Sage Weil [Tue, 6 Oct 2009 18:31:08 +0000 (11:31 -0700)]
ceph: file operations

File open and close operations, and read and write methods that ensure
we have obtained the proper capabilities from the MDS cluster before
performing IO on a file.  We take references on held capabilities for
the duration of the read/write to avoid prematurely releasing them
back to the MDS.

We implement two main paths for read and write: one that is buffered
(and uses generic_aio_{read,write}), and one that is fully synchronous
and blocking (operating either on a __user pointer or, if O_DIRECT,
directly on user pages).

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: directory operations
Sage Weil [Tue, 6 Oct 2009 18:31:08 +0000 (11:31 -0700)]
ceph: directory operations

Directory operations, including lookup, are defined here.  We take
advantage of lookup intents when possible.  For the most part, we just
need to build the proper requests for the metadata server(s) and
pass things off to the mds_client.

The results of most operations are normally incorporated into the
client's cache when the reply is parsed by ceph_fill_trace().
However, if the MDS replies without a trace (e.g., when retrying an
update after an MDS failure recovery), some operation-specific cleanup
may be needed.

We can validate cached dentries in two ways.  A per-dentry lease may
be issued by the MDS, or a per-directory cap may be issued that acts
as a lease on the entire directory.  In the latter case, a 'gen' value
is used to determine which dentries belong to the currently leased
directory contents.

We normally prepopulate the dcache and icache with readdir results.
This makes subsequent lookups and getattrs avoid any server
interaction.  It also lets us satisfy readdir operation by peeking at
the dcache IFF we hold the per-directory cap/lease, previously
performed a readdir, and haven't dropped any of the resulting
dentries.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: inode operations
Sage Weil [Tue, 6 Oct 2009 18:31:08 +0000 (11:31 -0700)]
ceph: inode operations

Inode cache and inode operations.  We also include routines to
incorporate metadata structures returned by the MDS into the client
cache, and some helpers to deal with file capabilities and metadata
leases.  The bulk of that work is done by fill_inode() and
fill_trace().

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: super.c
Sage Weil [Tue, 6 Oct 2009 18:31:07 +0000 (11:31 -0700)]
ceph: super.c

Mount option parsing, client setup and teardown, and a few odds and
ends (e.g., statfs).

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: ref counted buffer
Sage Weil [Tue, 6 Oct 2009 18:31:07 +0000 (11:31 -0700)]
ceph: ref counted buffer

struct ceph_buffer is a simple ref-counted buffer.  We transparently
choose between kmalloc for small buffers and vmalloc for large ones.

This is currently used only for allocating memory for xattr data.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: client types
Sage Weil [Tue, 6 Oct 2009 18:31:07 +0000 (11:31 -0700)]
ceph: client types

We first define constants, types, and prototypes for the kernel client
proper.

A few subsystems are defined separately later: the MDS, OSD, and
monitor clients, and the messaging layer.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: on-wire types
Sage Weil [Tue, 6 Oct 2009 18:31:06 +0000 (11:31 -0700)]
ceph: on-wire types

These headers describe the types used to exchange messages between the
Ceph client and various servers.  All types are little-endian and
packed.  These headers are shared between the kernel and userspace, so
all types are in terms of e.g. __u32.

Additionally, we define a few magic values to identify the current
version of the protocol(s) in use, so that discrepancies to be
detected on mount.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph: documentation
Sage Weil [Tue, 6 Oct 2009 18:31:05 +0000 (11:31 -0700)]
ceph: documentation

Mount options, syntax.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoLinux 2.6.32-rc1
Linus Torvalds [Sun, 27 Sep 2009 21:57:48 +0000 (14:57 -0700)]
Linux 2.6.32-rc1

15 years agoalpha: Fix duplicate <asm/thread_info.h> include
Linus Torvalds [Sun, 27 Sep 2009 21:46:05 +0000 (14:46 -0700)]
alpha: Fix duplicate <asm/thread_info.h> include

.. duplicated by merging the same fix twice, for details see commit
0d9df2515dbceb67d343c0f10fd3ff218380d524 ("Merge
git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-fixes")

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agotty: Fix regressions caused by commit b50989dc
Dave Young [Sun, 27 Sep 2009 16:00:42 +0000 (16:00 +0000)]
tty: Fix regressions caused by commit b50989dc

The following commit made console open fails while booting:

commit b50989dc444599c8b21edc23536fc305f4e9b7d5
Author: Alan Cox <alan@linux.intel.com>
Date:   Sat Sep 19 13:13:22 2009 -0700

tty: make the kref destructor occur asynchronously

Due to tty release routines run in a workqueue now, error like the
following will be reported while booting:

INIT open /dev/console Input/output error

It also causes hibernation regression to appear as reported at
http://bugzilla.kernel.org/show_bug.cgi?id=14229

The reason is that now there's latency issue with closing, but when
we open a "closing not finished" tty, -EIO will be returned.

Fix it as per the following Alan's suggestion:

  Fun but it's actually not a bug and the fix is wrong in itself as
  the port may be closing but not yet being destructed, in which case
  it seems to do the wrong thing.  Opening a tty that is closing (and
  could be closing for long periods) is supposed to return -EIO.

  I suspect a better way to deal with this and keep the old console
  timing is to split tty->shutdown into two functions.

  tty->shutdown() - called synchronously just before we dump the tty
  onto the waitqueue for destruction

  tty->cleanup() - called when the destructor runs.

  We would then do the shutdown part which can occur in IRQ context
  fine, before queueing the rest of the release (from tty->magic = 0
  ...  the end) to occur asynchronously

  The USB update in -next would then need a call like

       if (tty->cleanup)
               tty->cleanup(tty);

  at the top of the async function and the USB shutdown to be split
  between shutdown and cleanup as the USB resource cleanup and final
  tidy cannot occur synchronously as it needs to sleep.

  In other words the logic becomes

       final kref put
               make object unfindable

       async
               clean it up

Signed-off-by: Dave Young <hidave.darkstar@gmail.com>
[ rjw: Rebased on top of 2.6.31-git, reworked the changelog. ]
Signed-off-by: "Rafael J. Wysocki" <rjw@sisk.pl>
[ Changed serial naming to match new rules, dropped tty_shutdown as per
  comments from Alan Stern  - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoACPI: kill "unused variable ‘i’" warning
Linus Torvalds [Sun, 27 Sep 2009 18:58:36 +0000 (11:58 -0700)]
ACPI: kill "unused variable ‘i’" warning

Commit 3d5b6fb47a8e68fa311ca2c3447e7f8a7c3a9cf3 ("ACPI: Kill overly
verbose "power state" log messages") removed the actual use of this
variable, but didn't remove the variable itself, resulting in build
warnings like

  drivers/acpi/processor_idle.c: In function ‘acpi_processor_power_init’:
  drivers/acpi/processor_idle.c:1169: warning: unused variable ‘i’

Just get rid of the now unused variable.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoconst: mark struct vm_struct_operations
Alexey Dobriyan [Sun, 27 Sep 2009 18:29:37 +0000 (22:29 +0400)]
const: mark struct vm_struct_operations

* mark struct vm_area_struct::vm_ops as const
* mark vm_ops in AGP code

But leave TTM code alone, something is fishy there with global vm_ops
being used.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoMerge branch 'timers-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 27 Sep 2009 17:39:04 +0000 (10:39 -0700)]
Merge branch 'timers-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'timers-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  hrtimer: Eliminate needless reprogramming of clock events device

15 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Linus Torvalds [Sun, 27 Sep 2009 17:38:48 +0000 (10:38 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/lenb/linux-acpi-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6:
  ACPI: IA64=y ACPI=n build fix
  ACPI: Kill overly verbose "power state" log messages
  ACPI: fix Compaq Evo N800c (Pentium 4m) boot hang regression
  ACPI: Clarify resource conflict message
  thinkpad-acpi: fix CONFIG_THINKPAD_ACPI_HOTKEY_POLL build problem

15 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 27 Sep 2009 17:38:34 +0000 (10:38 -0700)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86: Fix hwpoison code related build failure on 32-bit NUMAQ

15 years agoACPI: IA64=y ACPI=n build fix
Len Brown [Sun, 27 Sep 2009 08:17:21 +0000 (04:17 -0400)]
ACPI: IA64=y ACPI=n build fix

ia64's sim_defconfig uses CONFIG_ACPI=n
which now #define's acpi_disabled in <linux/acpi.h>

So we shouldn't re-define it here in <asm/acpi.h>

Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: Kill overly verbose "power state" log messages
Roland Dreier [Thu, 24 Sep 2009 21:52:36 +0000 (14:52 -0700)]
ACPI: Kill overly verbose "power state" log messages

I was recently lucky enough to get a 64-CPU system, so my kernel log
ends up with 64 lines like:

    ACPI: CPU0 (power states: C1[C1] C2[C3])

This is pretty useless clutter because this info is already available
after boot from both /sys/devices/system/cpu/cpu*/cpuidle/state?/ as
well as /proc/acpi/processor/CPU*/power.

So just delete the code that prints the C-states in processor_idle.c.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agox86: Fix hwpoison code related build failure on 32-bit NUMAQ
Linus Torvalds [Sat, 26 Sep 2009 16:35:07 +0000 (09:35 -0700)]
x86: Fix hwpoison code related build failure on 32-bit NUMAQ

This build failure triggers:

 In file included from include/linux/suspend.h:8,
                 from arch/x86/kernel/asm-offsets_32.c:11,
                 from arch/x86/kernel/asm-offsets.c:2:
 include/linux/mm.h:503:2: error: #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS

Because due to the hwpoison page flag we ran out of page
flags on 32-bit.

Dont turn on hwpoison on 32-bit NUMA (it's rare in any
case).

Also clean up the Kconfig dependencies in the generic MM
code by introducing ARCH_SUPPORTS_MEMORY_FAILURE.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoACPI: fix Compaq Evo N800c (Pentium 4m) boot hang regression
Zhao Yakui [Sun, 27 Sep 2009 07:30:51 +0000 (03:30 -0400)]
ACPI: fix Compaq Evo N800c (Pentium 4m) boot hang regression

Don't disable ARB_DISABLE when the familary ID is 0x0F.

http://bugzilla.kernel.org/show_bug.cgi?id=14211

This was a 2.6.31 regression, and so this patch
needs to be applied to 2.6.31.stable

Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: Clarify resource conflict message
Jean Delvare [Tue, 8 Sep 2009 13:31:46 +0000 (15:31 +0200)]
ACPI: Clarify resource conflict message

The message "ACPI: Device needs an ACPI driver" is misleading. The
device _may_ need an ACPI driver, if the BIOS implemented a custom
API for the device in question (which, AFAIK, can't be checked.) If
not, then either a generic ACPI driver may be used (for example
"thermal"), or nothing can be done (other than a white list).

I propose to reword the message to:

ACPI: If an ACPI driver is available for this device, you should use
it instead of the native driver

which I think is more correct. Comments and suggestions welcome.

I also added a message warning about possible problems and system
instability when users pass acpi_enforce_resources=lax, as suggested
by Len.

Signed-off-by: Jean Delvare <jdelvare@suse.de>
Cc: Thomas Renninger <trenn@suse.de>
Cc: Alan Jenkins <sourcejedi.lkml@googlemail.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agothinkpad-acpi: fix CONFIG_THINKPAD_ACPI_HOTKEY_POLL build problem
Henrique de Moraes Holschuh [Sun, 27 Sep 2009 00:42:49 +0000 (21:42 -0300)]
thinkpad-acpi: fix CONFIG_THINKPAD_ACPI_HOTKEY_POLL build problem

Fix this problem when CONFIG_THINKPAD_ACPI_HOTKEY_POLL is undefined:

  CHECK   drivers/platform/x86/thinkpad_acpi.c
drivers/platform/x86/thinkpad_acpi.c:1968:21: error: not an lvalue
  CC [M]  drivers/platform/x86/thinkpad_acpi.o
drivers/platform/x86/thinkpad_acpi.c: In function 'tpacpi_hotkey_driver_mask_set':
drivers/platform/x86/thinkpad_acpi.c:1968: error: lvalue required as left operand of assignment

Reported-by: Noah Dain <noahdain@gmail.com>
Reported-by: Audrius Kazukauskas <audrius@neutrino.lt>
Signed-off-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6
Linus Torvalds [Sat, 26 Sep 2009 20:39:21 +0000 (13:39 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6:
  [IA64] implement ticket locks for Itanium

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-fixes
Linus Torvalds [Sat, 26 Sep 2009 17:51:54 +0000 (10:51 -0700)]
Merge git://git./linux/kernel/git/sam/kbuild-fixes

* git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-fixes:
  alpha: fix build after vmlinux.lds.S cleanup
  mips: fix build of vmlinux.lds

15 years agoMerge branch 'for-linus' of git://git.o-hand.com/linux-rpurdie-leds
Linus Torvalds [Sat, 26 Sep 2009 17:50:47 +0000 (10:50 -0700)]
Merge branch 'for-linus' of git://git.o-hand.com/linux-rpurdie-leds

* 'for-linus' of git://git.o-hand.com/linux-rpurdie-leds:
  leds: move leds-clevo-mail's probe function to .devinit.text
  leds: Fix indentation in LEDS_LP3944 Kconfig entry
  leds: Fix LED names
  leds: Fix leds-pca9532 whitespace issues
  leds: fix coding style in worker thread code for ledtrig-gpio.
  leds: gpio-leds: fix typographics fault
  leds: Add WM831x status LED driver

15 years agoMerge branch 'for-linus' of git://git.o-hand.com/linux-rpurdie-backlight
Linus Torvalds [Sat, 26 Sep 2009 17:49:42 +0000 (10:49 -0700)]
Merge branch 'for-linus' of git://git.o-hand.com/linux-rpurdie-backlight

* 'for-linus' of git://git.o-hand.com/linux-rpurdie-backlight:
  backlight: new driver for ADP5520/ADP5501 MFD PMICs
  backlight: extend event support to also support poll()
  backlight/eeepc-laptop: Update the backlight state when we change brightness
  backlight/acpi: Update the backlight state when we change brightness
  backlight: Allow drivers to update the core, and generate events on changes
  backlight: switch to da903x driver to dev_pm_ops
  backlight: Add support for the Avionic Design Xanthos backlight device.
  backlight: spi driver for LMS283GF05 LCD
  backlight: move hp680-bl's probe function to .devinit.text
  backlight: Add support for new Apple machines.
  backlight: mbp_nvidia_bl: add support for MacBookAir 1,1
  backlight: Add WM831x backlight driver

Trivial conflicts due to '#ifdef CONFIG_PM' differences in
drivers/video/backlight/da903x_bl.c

15 years agoheaders: kref.h redux
Alexey Dobriyan [Sat, 26 Sep 2009 15:37:22 +0000 (19:37 +0400)]
headers: kref.h redux

* remove asm/atomic.h inclusion from kref.h -- not needed, linux/types.h
  is enough for atomic_t
* remove linux/kref.h inclusion from files which do not need it.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoMerge branch 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 26 Sep 2009 17:15:53 +0000 (10:15 -0700)]
Merge branch 'core-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  futex: Add memory barrier commentary to futex_wait_queue_me()
  futex: Fix wakeup race by setting TASK_INTERRUPTIBLE before queue_me()
  futex: Correct futex_q woken state commentary
  futex: Make function kernel-doc commentary consistent
  futex: Correct queue_me and unqueue_me commentary
  futex: Correct futex_wait_requeue_pi() commentary

15 years agoMerge branch 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 26 Sep 2009 17:15:33 +0000 (10:15 -0700)]
Merge branch 'perf-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  perf tools: Dont use openat()
  perf tools: Fix buffer allocation
  perf tools: .gitignore += perf*.html
  perf tools: Handle relative paths while loading module symbols
  perf tools: Fix module symbol loading bug
  perf_event, x86: Fix 'perf sched record' crashing the machine
  perf_event: Update PERF_EVENT_FORK header definition
  perf stat: Fix zero total printouts

15 years agoMerge branch 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 26 Sep 2009 17:14:41 +0000 (10:14 -0700)]
Merge branch 'timers-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  clocksource: Resume clocksource without taking the clocksource mutex

15 years agoMerge branch 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 26 Sep 2009 17:13:54 +0000 (10:13 -0700)]
Merge branch 'tracing-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  modules, tracing: Remove stale struct marker signature from module_layout()
  tracing/workqueue: Use %pf in workqueue trace events
  tracing: Fix a comment and a trivial format issue in tracepoint.h
  tracing: Fix failure path in ftrace_regex_open()
  tracing: Fix failure path in ftrace_graph_write()
  tracing: Check the return value of trace_get_user()
  tracing: Fix off-by-one in trace_get_user()

15 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 26 Sep 2009 17:13:35 +0000 (10:13 -0700)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86: Remove redundant non-NUMA topology functions
  x86: early_printk: Protect against using the same device twice
  x86: Reduce verbosity of "PAT enabled" kernel message
  x86: Reduce verbosity of "TSC is reliable" message
  x86: mce: Use safer ways to access MCE registers
  x86: mce, inject: Use real inject-msg in raise_local
  x86: mce: Fix thermal throttling message storm
  x86: mce: Clean up thermal throttling state tracking code
  x86: split NX setup into separate file to limit unstack-protected code
  xen: check EFER for NX before setting up GDT mapping
  x86: Cleanup linker script using new linker script macros.
  x86: Use section .data.page_aligned for the idt_table.
  x86: convert to use __HEAD and HEAD_TEXT macros.
  x86: convert compressed loader to use __HEAD and HEAD_TEXT macros.
  x86: fix fragile computation of vsyscall address

15 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Linus Torvalds [Sat, 26 Sep 2009 17:12:03 +0000 (10:12 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/lenb/linux-acpi-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6: (32 commits)
  ACPI: i2c-scmi: don't use acpi_device_uid()
  ACPI: simplify building device HID/CID list
  ACPI: remove acpi_device_uid() and related stuff
  ACPI: remove acpi_device.flags.hardware_id
  ACPI: remove acpi_device.flags.compatible_ids
  ACPI: maintain a single list of _HID and _CID IDs
  ACPI: make sure every acpi_device has an ID
  ACPI: use acpi_device_hid() when possible
  ACPI: fix synthetic HID for \_SB_
  ACPI: handle re-enumeration, when acpi_devices might already exist
  ACPI: factor out device type and status checking
  ACPI: add acpi_bus_get_status_handle()
  ACPI: use acpi_walk_namespace() to enumerate devices
  ACPI: identify device tree root by null parent pointer, not ACPI_BUS_TYPE
  ACPI: enumerate namespace before adding functional fixed hardware devices
  ACPI: convert acpi_bus_scan() to operate on an acpi_handle
  ACPI: add acpi_bus_get_parent() and remove "parent" arguments
  ACPI: remove unnecessary argument checking
  ACPI: remove redundant "type" arguments
  ACPI: remove acpi_device_set_context() "type" argument
  ...

15 years agoMerge branch 'writeback' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Sat, 26 Sep 2009 17:11:13 +0000 (10:11 -0700)]
Merge branch 'writeback' of git://git.kernel.dk/linux-2.6-block

* 'writeback' of git://git.kernel.dk/linux-2.6-block:
  writeback: pass in super_block to bdi_start_writeback()

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Sat, 26 Sep 2009 17:10:35 +0000 (10:10 -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: fix locking and list handling code in cifs_open and its helper
  [CIFS] Remove build warning
  cifs: fix problems with last two commits
  [CIFS] Fix build break when keys support turned off
  cifs: eliminate cifs_init_private
  cifs: convert oplock breaks to use slow_work facility (try #4)
  cifs: have cifsFileInfo hold an extra inode reference
  cifs: take read lock on GlobalSMBSes_lock in is_valid_oplock_break
  cifs: remove cifsInodeInfo.oplockPending flag
  cifs: fix oplock request handling in posix codepath
  [CIFS] Re-enable Lanman security

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Sat, 26 Sep 2009 17:09:39 +0000 (10:09 -0700)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
  at91_can: Forgotten git 'add' of at91_can.c
  TI Davinci EMAC: Fix in vector definition for EMAC_VERSION_2
  ax25: Fix ax25_cb refcounting in ax25_ctl_ioctl
  virtio_net: Check for room in the vq before adding buffer
  virtio_net: avoid (most) NETDEV_TX_BUSY by stopping queue early.
  virtio_net: formalize skb_vnet_hdr
  virtio_net: don't free buffers in xmit ring
  virtio_net: return NETDEV_TX_BUSY instead of queueing an extra skb.
  virtio_net: skb_orphan() and nf_reset() in xmit path.

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Sat, 26 Sep 2009 17:08:23 +0000 (10:08 -0700)]
Merge git://git./linux/kernel/git/davem/sparc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  sparc64: vio: Kill BUILD_BUG_ON() in vio_dring_avail().

Trivial conflict in arch/sparc/include/asm/vio.h due to David removing
the whole messy BUG_ON that was confused.

15 years agomake Linux bootable on ARM again
Nicolas Pitre [Fri, 25 Sep 2009 20:28:02 +0000 (16:28 -0400)]
make Linux bootable on ARM again

Commit 200b812d00 "Clear the exclusive monitor when returning from an
exception" broke the vast majority of ARM systems in the wild which are
still pre ARMv6.  The kernel is crashing on the first occurrence of an
exception due to the removal of the actual return instruction for them.
Let's add it back.

Signed-off-by: Nicolas Pitre <nico@marvell.com>
Acked-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agobacklight: new driver for ADP5520/ADP5501 MFD PMICs
Michael Hennerich [Thu, 17 Sep 2009 18:29:52 +0000 (14:29 -0400)]
backlight: new driver for ADP5520/ADP5501 MFD PMICs

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>
15 years agobacklight: extend event support to also support poll()
Henrique de Moraes Holschuh [Sun, 20 Sep 2009 17:44:47 +0000 (14:44 -0300)]
backlight: extend event support to also support poll()

Extend the backlight event support to also allow the use of
poll()/select() on actual_brightness.

We already have the entire event hookup anyway, adding a single
function call in one line to get functionality like that is a really
good deal.

Signed-off-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>
15 years agoACPI: i2c-scmi: don't use acpi_device_uid()
Bjorn Helgaas [Sat, 26 Sep 2009 03:11:43 +0000 (21:11 -0600)]
ACPI: i2c-scmi: don't use acpi_device_uid()

We recently removed the acpi_device_uid() interface because nobody
used it.  I don't think it's essential here either.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoMerge branch 'thinkpad-2.6.32-part2' into release
Len Brown [Sat, 26 Sep 2009 05:08:55 +0000 (01:08 -0400)]
Merge branch 'thinkpad-2.6.32-part2' into release

15 years agoMerge branch 'bjorn-HID' into release
Len Brown [Sat, 26 Sep 2009 05:08:43 +0000 (01:08 -0400)]
Merge branch 'bjorn-HID' into release

15 years agowriteback: pass in super_block to bdi_start_writeback()
Jens Axboe [Fri, 25 Sep 2009 22:07:46 +0000 (00:07 +0200)]
writeback: pass in super_block to bdi_start_writeback()

Sometimes we only want to write pages from a specific super_block,
so allow that to be passed in.

This fixes a problem with commit 56a131dcf7ed36c3c6e36bea448b674ea85ed5bb
causing writeback on all super_blocks on a bdi, where we only really
want to sync a specific sb from writeback_inodes_sb().

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agoat91_can: Forgotten git 'add' of at91_can.c
David S. Miller [Fri, 25 Sep 2009 19:14:43 +0000 (12:14 -0700)]
at91_can: Forgotten git 'add' of at91_can.c

Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoTI Davinci EMAC: Fix in vector definition for EMAC_VERSION_2
Sriram [Thu, 24 Sep 2009 19:15:18 +0000 (19:15 +0000)]
TI Davinci EMAC: Fix in vector definition for EMAC_VERSION_2

In the emac_poll function when looking for interrupt status masks
correct definition must be chosen based on EMAC_VERSION(the bit
mask has changed from version 1 to version 2).

Signed-off-by: Sriram <srk@ti.com>
Acked-by: Chaithrika U S <chaithrika@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoax25: Fix ax25_cb refcounting in ax25_ctl_ioctl
Jarek Poplawski [Fri, 25 Sep 2009 03:10:38 +0000 (03:10 +0000)]
ax25: Fix ax25_cb refcounting in ax25_ctl_ioctl

Use ax25_cb_put after ax25_find_cb in ax25_ctl_ioctl.

Reported-by: Bernard Pidoux F6BVP <f6bvp@free.fr>
Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Reviewed-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoACPI: simplify building device HID/CID list
Bjorn Helgaas [Mon, 21 Sep 2009 19:35:40 +0000 (13:35 -0600)]
ACPI: simplify building device HID/CID list

Minor code cleanup, no functional change.  Instead of remembering
what HIDs & CIDs to add later, just add them immediately.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: remove acpi_device_uid() and related stuff
Bjorn Helgaas [Mon, 21 Sep 2009 19:35:35 +0000 (13:35 -0600)]
ACPI: remove acpi_device_uid() and related stuff

Nobody uses acpi_device_uid(), so this patch removes it.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: remove acpi_device.flags.hardware_id
Bjorn Helgaas [Mon, 21 Sep 2009 19:35:29 +0000 (13:35 -0600)]
ACPI: remove acpi_device.flags.hardware_id

Every acpi_device has at least one ID (if there's no _HID or _CID, we
give it a synthetic or default ID).  So there's no longer a need to
check whether an ID exists; we can just use it.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: remove acpi_device.flags.compatible_ids
Bjorn Helgaas [Mon, 21 Sep 2009 19:35:24 +0000 (13:35 -0600)]
ACPI: remove acpi_device.flags.compatible_ids

We now keep a single list of IDs that includes both the _HID and any
_CIDs.  We no longer need to keep track of whether the device has a _CID.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: maintain a single list of _HID and _CID IDs
Bjorn Helgaas [Mon, 21 Sep 2009 19:35:19 +0000 (13:35 -0600)]
ACPI: maintain a single list of _HID and _CID IDs

There's no need to treat _HID and _CID differently.  Keeping them in
a single list makes code that uses the IDs a little simpler because it
can just traverse the list rather than checking "do we have a HID?",
"do we have any CIDs?"

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Reviewed-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: make sure every acpi_device has an ID
Bjorn Helgaas [Mon, 21 Sep 2009 19:35:14 +0000 (13:35 -0600)]
ACPI: make sure every acpi_device has an ID

This makes sure every acpi_device has at least one ID.  If we build an
acpi_device for a namespace node with no _HID or _CID, we sometimes
synthesize an ID like "LNXCPU" or "LNXVIDEO".  If we don't even have
that, give it a default "device" ID.

Note that this means things like:
    /sys/devices/LNXSYSTM:00/LNXSYBUS:00/HWP0001:00/HWP0002:04/device:00
(a PCI slot SxFy device) will have "hid" and "modprobe" entries, where
they didn't before.  These aren't very useful (a HID of "device" doesn't
tell you what *kind* of device it is, so it doesn't help find a driver),
but I don't think they're harmful.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: use acpi_device_hid() when possible
Bjorn Helgaas [Mon, 21 Sep 2009 19:35:09 +0000 (13:35 -0600)]
ACPI: use acpi_device_hid() when possible

Use acpi_device_hid() rather than accessing acpi_device.pnp.hardware_id
directly.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: fix synthetic HID for \_SB_
Bjorn Helgaas [Mon, 21 Sep 2009 19:35:04 +0000 (13:35 -0600)]
ACPI: fix synthetic HID for \_SB_

This makes \_SB_ show up as /sys/devices/LNXSYSTM:00/LNXSYBUS:00
rather than "device:00".  This has been broken for a loooong time
(at least since 2.6.13) because device->parent is an acpi_device
pointer, not a handle.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: handle re-enumeration, when acpi_devices might already exist
Bjorn Helgaas [Mon, 21 Sep 2009 19:30:11 +0000 (19:30 +0000)]
ACPI: handle re-enumeration, when acpi_devices might already exist

acpi_bus_scan() traverses the namespace to enumerate devices and uses
acpi_add_single_object() to create acpi_devices.  When the platform
notifies us of a hot-plug event, we need to traverse part of the namespace
again to figure out what appeared or disappeared.  (We don't yet call
acpi_bus_scan() during hot-plug, but I plan to do that in the future.)

This patch makes acpi_add_single_object() notice when we already have
an acpi_device, so we don't need to make a new one.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: factor out device type and status checking
Bjorn Helgaas [Mon, 21 Sep 2009 19:30:06 +0000 (19:30 +0000)]
ACPI: factor out device type and status checking

This patch adds acpi_bus_type_and_status(), which determines the type
of the object and whether we want to build an acpi_device for it.  If
it is acpi_device-worthy, it returns the type and the device's current
status.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: add acpi_bus_get_status_handle()
Bjorn Helgaas [Mon, 21 Sep 2009 19:30:01 +0000 (19:30 +0000)]
ACPI: add acpi_bus_get_status_handle()

Add acpi_bus_get_status_handle() so we can get the status of a namespace
object before building a struct acpi_device.

This removes a use of "device->flags.dynamic_status", a cached indicator of
whether _STA exists.  It seems simpler and more reliable to just evaluate
_STA and catch AE_NOT_FOUND errors.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: use acpi_walk_namespace() to enumerate devices
Bjorn Helgaas [Mon, 21 Sep 2009 19:29:56 +0000 (19:29 +0000)]
ACPI: use acpi_walk_namespace() to enumerate devices

acpi_bus_scan() currently walks the namespace manually.  This patch changes
it to use acpi_walk_namespace() instead.

Besides removing some complicated code, this means we take advantage of the
namespace locking done by acpi_walk_namespace().  The locking isn't so
important at boot-time, but I hope to eventually use this same path to
handle hot-addition of devices, when it will be important.

Note that acpi_walk_namespace() does not actually visit the starting node
first, so we need to do that by hand first.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: identify device tree root by null parent pointer, not ACPI_BUS_TYPE
Bjorn Helgaas [Mon, 21 Sep 2009 19:29:50 +0000 (19:29 +0000)]
ACPI: identify device tree root by null parent pointer, not ACPI_BUS_TYPE

We can identify the root of the ACPI device tree by the fact that it
has no parent.  This is simpler than passing around ACPI_BUS_TYPE_SYSTEM
and will help remove special treatment of the device tree root.

Currently, we add the root by hand with ACPI_BUS_TYPE_SYSTEM.  If we
traverse the tree treating the root as just another device and use
acpi_get_type(), the root shows up as ACPI_TYPE_DEVICE.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: enumerate namespace before adding functional fixed hardware devices
Bjorn Helgaas [Mon, 21 Sep 2009 19:29:45 +0000 (19:29 +0000)]
ACPI: enumerate namespace before adding functional fixed hardware devices

This patch changes the order so we enumerate in the "root, namespace,
functional fixed" order instead of the "root, functional fixed, namespace"
order.  When I change acpi_bus_scan() to use acpi_walk_namespace(), it
will use the former order, so this patch isolates the order change for
bisectability.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: convert acpi_bus_scan() to operate on an acpi_handle
Bjorn Helgaas [Mon, 21 Sep 2009 19:29:40 +0000 (19:29 +0000)]
ACPI: convert acpi_bus_scan() to operate on an acpi_handle

This patch changes acpi_bus_scan() to take an acpi_handle rather than an
acpi_device pointer.  I plan to use acpi_bus_scan() in the hotplug path,
and I'd rather not assume that notifications only go to nodes that already
have acpi_devices.

This will also help remove the special case for adding the root node.  We
currently add the root by hand before acpi_bus_scan(), but using a handle
here means we can start the acpi_bus_scan() directly with the root even
though it doesn't have an acpi_device yet.

Note that acpi_bus_scan() currently adds and/or starts the *children* of
its device argument.  It doesn't do anything with the device itself.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: add acpi_bus_get_parent() and remove "parent" arguments
Bjorn Helgaas [Mon, 21 Sep 2009 19:29:35 +0000 (19:29 +0000)]
ACPI: add acpi_bus_get_parent() and remove "parent" arguments

This patch adds acpi_bus_get_parent(), which ascends the namespace until
it finds a parent with an acpi_device.

Then we use acpi_bus_get_parent() in acpi_add_single_object(), so callers
don't have to figure out or keep track of the parent acpi_device.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: remove unnecessary argument checking
Bjorn Helgaas [Mon, 21 Sep 2009 19:29:30 +0000 (19:29 +0000)]
ACPI: remove unnecessary argument checking

acpi_add_single_object() is static, and all callers supply a valid "child"
argument, so we don't need to check it.  This patch also remove some
unnecessary initializations.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: remove redundant "type" arguments
Bjorn Helgaas [Mon, 21 Sep 2009 19:29:25 +0000 (19:29 +0000)]
ACPI: remove redundant "type" arguments

We now save the ACPI bus "device_type" in the acpi_device structure, so
we don't need to pass it around explicitly anymore.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: remove acpi_device_set_context() "type" argument
Bjorn Helgaas [Mon, 21 Sep 2009 19:29:20 +0000 (19:29 +0000)]
ACPI: remove acpi_device_set_context() "type" argument

We only pass the "type" to acpi_device_set_context() so we know whether
the device has a handle to which we can attach the acpi_device pointer.
But it's safer to just check for the handle directly, since it's in the
acpi_device already.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: use device_type rather than comparing HID
Bjorn Helgaas [Mon, 21 Sep 2009 19:29:15 +0000 (19:29 +0000)]
ACPI: use device_type rather than comparing HID

Check the acpi_device device_type rather than the HID.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: save device_type in acpi_device
Bjorn Helgaas [Mon, 21 Sep 2009 19:29:10 +0000 (19:29 +0000)]
ACPI: save device_type in acpi_device

Most uses of the ACPI bus device_type (ACPI_BUS_TYPE_DEVICE,
ACPI_BUS_TYPE_POWER, etc) are during device initialization, but
we do need it later for notify handler installation, since that
is different for fixed hardware devices vs. namespace devices.

This patch saves the device_type in the acpi_device structure,
so we can check that rather than comparing against the _HID string.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: remove redundant "handle" and "parent" arguments
Bjorn Helgaas [Mon, 21 Sep 2009 19:29:05 +0000 (19:29 +0000)]
ACPI: remove redundant "handle" and "parent" arguments

In several cases, functions take handle and parent device pointers in
addition to acpi_device pointers.  But the acpi_device structure contains
both the handle and the parent pointer, so it's pointless and error-prone
to pass them all.  This patch removes the unnecessary "handle" and "parent"
arguments.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: remove unused acpi_bus_scan_fixed() argument
Bjorn Helgaas [Mon, 21 Sep 2009 19:28:59 +0000 (19:28 +0000)]
ACPI: remove unused acpi_bus_scan_fixed() argument

We never use the "root" argument, so just remove it.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPI: add debug for device addition
Bjorn Helgaas [Mon, 21 Sep 2009 19:28:54 +0000 (19:28 +0000)]
ACPI: add debug for device addition

Add debug output for adding an ACPI device.  Enable this with
"acpi.debug_layer=0x00010000" (ACPI_BUS_COMPONENT).

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoACPICA: fixup after acpi_get_object_info() change
Bjorn Helgaas [Mon, 21 Sep 2009 19:28:49 +0000 (19:28 +0000)]
ACPICA: fixup after acpi_get_object_info() change

Commit 15b8dd53f5ffa changed info->hardware_id from a static array to
a pointer.  If hardware_id is non-NULL, it points to a NULL-terminated
string, so we don't need to terminate it explicitly.  However, it may
be NULL; in that case, we *can't* add a NULL terminator.

This causes a NULL pointer dereference oops for devices without _HID.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
CC: Lin Ming <ming.m.lin@intel.com>
CC: Bob Moore <robert.moore@intel.com>
CC: Gary Hade <garyhade@us.ibm.com>
Signed-off-by: Len Brown <len.brown@intel.com>
15 years agoMerge branch 'master' of ssh://master.kernel.org/pub/scm/linux/kernel/git/rusty/linux...
David S. Miller [Fri, 25 Sep 2009 18:09:08 +0000 (11:09 -0700)]
Merge branch 'master' of ssh:///linux/kernel/git/rusty/linux-2.6-for-davem

15 years agocifs: fix locking and list handling code in cifs_open and its helper
Jeff Layton [Fri, 25 Sep 2009 13:53:37 +0000 (09:53 -0400)]
cifs: fix locking and list handling code in cifs_open and its helper

The patch to remove cifs_init_private introduced a locking imbalance. It
didn't remove the leftover list addition code and the unlocking in that
function. cifs_new_fileinfo does the list addition now, so there should
be no need to do it outside of that function.

pCifsInode will never be NULL, so we don't need to check for that. This
patch also gets rid of the ugly locking and unlocking across function
calls.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Acked-by: Steve French <sfrench@us.ibm.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
15 years agoalpha: fix build after vmlinux.lds.S cleanup
Sam Ravnborg [Fri, 25 Sep 2009 17:53:43 +0000 (19:53 +0200)]
alpha: fix build after vmlinux.lds.S cleanup

Add include to get missing THREAD_SIZE definition

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Tim Abbott <tabbott@ksplice.com>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Richard Henderson <rth@twiddle.net>
15 years agomips: fix build of vmlinux.lds
Manuel Lauss [Thu, 24 Sep 2009 19:44:24 +0000 (21:44 +0200)]
mips: fix build of vmlinux.lds

Commit 51b563fc93c8cb5bff1d67a0a71c374e4a4ea049 ("arm, cris, mips,
sparc, powerpc, um, xtensa: fix build with bash 4.0") removed a few
CPPFLAGS with vital include paths necessary to build vmlinux.lds
on MIPS, and moved the calculation of the 'jiffies' symbol
directly to vmlinux.lds.S but forgot to change make ifdef/... to
cpp macros.

Signed-off-by: Manuel Lauss <manuel.lauss@gmail.com>
[sam: moved assignment of CPPFLAGS arch/mips/kernel/Makefile]
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Dmitri Vorobiev <dmitri.vorobiev@movial.com>
15 years agoMerge branch 'writeback' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Fri, 25 Sep 2009 16:27:30 +0000 (09:27 -0700)]
Merge branch 'writeback' of git://git.kernel.dk/linux-2.6-block

* 'writeback' of git://git.kernel.dk/linux-2.6-block:
  writeback: writeback_inodes_sb() should use bdi_start_writeback()
  writeback: don't delay inodes redirtied by a fast dirtier
  writeback: make the super_block pinning more efficient
  writeback: don't resort for a single super_block in move_expired_inodes()
  writeback: move inodes from one super_block together
  writeback: get rid to incorrect references to pdflush in comments
  writeback: improve readability of the wb_writeback() continue/break logic
  writeback: cleanup writeback_single_inode()
  writeback: kupdate writeback shall not stop when more io is possible
  writeback: stop background writeback when below background threshold
  writeback: balance_dirty_pages() shall write more than dirtied pages
  fs: Fix busyloop in wb_writeback()

15 years agowriteback: writeback_inodes_sb() should use bdi_start_writeback()
Jens Axboe [Fri, 25 Sep 2009 15:15:03 +0000 (17:15 +0200)]
writeback: writeback_inodes_sb() should use bdi_start_writeback()

Pointless to iterate other devices looking for a super, when
we have a bdi mapping.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agowriteback: don't delay inodes redirtied by a fast dirtier
Wu Fengguang [Fri, 25 Sep 2009 04:04:10 +0000 (06:04 +0200)]
writeback: don't delay inodes redirtied by a fast dirtier

Debug traces show that in per-bdi writeback, the inode under writeback
almost always get redirtied by a busy dirtier.  We used to call
redirty_tail() in this case, which could delay inode for up to 30s.

This is unacceptable because it now happens so frequently for plain cp/dd,
that the accumulated delays could make writeback of big files very slow.

So let's distinguish between data redirty and metadata only redirty.
The first one is caused by a busy dirtier, while the latter one could
happen in XFS, NFS, etc. when they are doing delalloc or updating isize.

The inode being busy dirtied will now be requeued for next io, while
the inode being redirtied by fs will continue to be delayed to avoid
repeated IO.

CC: Jan Kara <jack@suse.cz>
CC: Theodore Ts'o <tytso@mit.edu>
CC: Dave Chinner <david@fromorbit.com>
CC: Chris Mason <chris.mason@oracle.com>
CC: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agowriteback: make the super_block pinning more efficient
Jens Axboe [Thu, 24 Sep 2009 13:25:11 +0000 (15:25 +0200)]
writeback: make the super_block pinning more efficient

Currently we pin the inode->i_sb for every single inode. This
increases cache traffic on sb->s_umount sem. Lets instead
cache the inode sb pin state and keep the super_block pinned
for as long as keep writing out inodes from the same
super_block.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agowriteback: don't resort for a single super_block in move_expired_inodes()
Jens Axboe [Thu, 24 Sep 2009 13:12:57 +0000 (15:12 +0200)]
writeback: don't resort for a single super_block in move_expired_inodes()

If we only moved inodes from a single super_block to the temporary
list, there's no point in doing a resort for multiple super_blocks.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agowriteback: move inodes from one super_block together
Shaohua Li [Thu, 24 Sep 2009 12:42:33 +0000 (14:42 +0200)]
writeback: move inodes from one super_block together

__mark_inode_dirty adds inode to wb dirty list in random order. If a disk has
several partitions, writeback might keep spindle moving between partitions.
To reduce the move, better write big chunk of one partition and then move to
another. Inodes from one fs usually are in one partion, so idealy move indoes
from one fs together should reduce spindle move. This patch tries to address
this. Before per-bdi writeback is added, the behavior is write indoes
from one fs first and then another, so the patch restores previous behavior.
The loop in the patch is a bit ugly, should we add a dirty list for each
superblock in bdi_writeback?

Test in a two partition disk with attached fio script shows about 3% ~ 6%
improvement.

Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Reviewed-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agowriteback: get rid to incorrect references to pdflush in comments
Jens Axboe [Wed, 23 Sep 2009 17:37:09 +0000 (19:37 +0200)]
writeback: get rid to incorrect references to pdflush in comments

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agowriteback: improve readability of the wb_writeback() continue/break logic
Jens Axboe [Wed, 23 Sep 2009 17:32:26 +0000 (19:32 +0200)]
writeback: improve readability of the wb_writeback() continue/break logic

And throw some comments in there, too.

Reviewed-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agowriteback: cleanup writeback_single_inode()
Wu Fengguang [Wed, 23 Sep 2009 12:33:42 +0000 (20:33 +0800)]
writeback: cleanup writeback_single_inode()

Make the if-else straight in writeback_single_inode().
No behavior change.

Cc: Jan Kara <jack@suse.cz>
Cc: Michael Rubin <mrubin@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Fengguang Wu <wfg@mail.ustc.edu.cn>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agowriteback: kupdate writeback shall not stop when more io is possible
Wu Fengguang [Wed, 23 Sep 2009 12:33:41 +0000 (20:33 +0800)]
writeback: kupdate writeback shall not stop when more io is possible

Fix the kupdate case, which disregards wbc.more_io and stop writeback
prematurely even when there are more inodes to be synced.

wbc.more_io should always be respected.

Also remove the pages_skipped check. It will set when some page(s) of some
inode(s) cannot be written for now. Such inodes will be delayed for a while.
This variable has nothing to do with whether there are other writeable inodes.

CC: Jan Kara <jack@suse.cz>
CC: Dave Chinner <david@fromorbit.com>
CC: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agowriteback: stop background writeback when below background threshold
Wu Fengguang [Wed, 23 Sep 2009 12:33:40 +0000 (20:33 +0800)]
writeback: stop background writeback when below background threshold

Treat bdi_start_writeback(0) as a special request to do background write,
and stop such work when we are below the background dirty threshold.

Also simplify the (nr_pages <= 0) checks. Since we already pass in
nr_pages=LONG_MAX for WB_SYNC_ALL and background writes, we don't
need to worry about it being decreased to zero.

Reported-by: Richard Kennedy <richard@rsk.demon.co.uk>
CC: Jan Kara <jack@suse.cz>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agowriteback: balance_dirty_pages() shall write more than dirtied pages
Wu Fengguang [Wed, 23 Sep 2009 13:56:00 +0000 (21:56 +0800)]
writeback: balance_dirty_pages() shall write more than dirtied pages

Some filesystem may choose to write much more than ratelimit_pages
before calling balance_dirty_pages_ratelimited_nr(). So it is safer to
determine number to write based on real number of dirtied pages.

Otherwise it is possible that
  loop {
    btrfs_file_write():     dirty 1024 pages
    balance_dirty_pages():  write up to 48 pages (= ratelimit_pages * 1.5)
  }
in which the writeback rate cannot keep up with dirty rate, and the
dirty pages go all the way beyond dirty_thresh.

The increased write_chunk may make the dirtier more bumpy.
So filesystems shall be take care not to dirty too much at
a time (eg. > 4MB) without checking the ratelimit.

Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agofs: Fix busyloop in wb_writeback()
Jan Kara [Wed, 16 Sep 2009 17:22:48 +0000 (19:22 +0200)]
fs: Fix busyloop in wb_writeback()

If all inodes are under writeback (e.g. in case when there's only one inode
with dirty pages), wb_writeback() with WB_SYNC_NONE work basically degrades
to busylooping until I_SYNC flags of the inode is cleared. Fix the problem by
waiting on I_SYNC flags of an inode on b_more_io list in case we failed to
write anything.

Tested-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years ago[IA64] implement ticket locks for Itanium
Tony Luck [Fri, 25 Sep 2009 15:42:16 +0000 (08:42 -0700)]
[IA64] implement ticket locks for Itanium

Back in January 2008 Nick Piggin implemented "ticket" spinlocks
for X86 (See commit 314cdbefd1fd0a7acf3780e9628465b77ea6a836).

IA64 implementation has a couple of differences because of the
available atomic operations ... e.g. we have no fetchadd2 instruction
that operates on a 16-bit quantity so we make ticket locks use
a 32-bit word for each of the current ticket and now-serving values.

Performance on uncontended locks is about 8% worse than the previous
implementation, but this seems a good trade for determinism in the
contended case. Performance impact on macro-level benchmarks is in
the noise.

Signed-off-by: Tony Luck <tony.luck@intel.com>