In order to avoid constantly allocating and deallocating there is a fixed
buffer which spi_write_then_read() uses for transfers, with an early error
check to ensure that the transfer fits within the buffer. This limits the
size of transfers to this size, currently max(32, SMP_CACHE_BYTES).

Since we can dynamically allocate and in fact already have a fallback
to do so when there is contention for the fixed buffer remove this
restriction and instead dynamically allocate a suitably sized buffer if
the transfer won't fit.
Signed-off-by: NMark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

This will allow to use gpio for chip select with no modification in the
driver binding

When use the cs-gpios, the gpio number will be passed via the cs_gpio field
and the number of chip select will automatically increased with max(hw cs, gpio cs).

So if for example the controller has 2 CS lines, and the cs-gpios
property looks like this:

cs-gpios = <&gpio1 0 0> <0> <&gpio1 1 0> <&gpio1 2 0>;

Then it should be configured so that num_chipselect = 4 with the
following mapping:

cs0 : &gpio1 0 0
cs1 : native
cs2 : &gpio1 1 0
cs3 : &gpio1 2 0
Signed-off-by: NJean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
Cc: devicetree-discuss@lists.ozlabs.org
Cc: spi-devel-general@lists.sourceforge.net
Signed-off-by: NRichard Genoud <richard.genoud@gmail.com>
[grant.likely: fixed up type of cs count so min() can do type checking]
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

Currently the master->setup() is called unconditionally.
The assumption is that every driver need to implement this
callback. This encourages drivers to populate empty functions
to prevent crashing.

This patch prevents the call of master->setup() if it is not
populated.
Signed-off-by: NLaxman Dewangan <ldewangan@nvidia.com>
Signed-off-by: NMark Brown <broonie@opensource.wolfsonmicro.com>

When spi client does the spi transfer and does not sets
the bits_per_word for each transfer then set it as default
of spi device in spi core before calling low level transfer.

Removing the similar code from spi-tegra20-slink driver as
it is not required.
Signed-off-by: NLaxman Dewangan <ldewangan@nvidia.com>
Signed-off-by: NMark Brown <broonie@opensource.wolfsonmicro.com>

The modalias attribute of spi devices doesn't have the "spi:" prefix
that is used in the UEVENT and in spi device drivers.  This patch adds
the prefix so the modprobe can correctly match modules to devices.
Reported-by: NDavid Daney <david.daney@cavium.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

Trivial simplification.  Instead of requiring spi master drivers to
always set the bus number (even when a dynamic number is desired),
this patch modifies spi_alloc_master() to initialize num_chipselect to
1 (because there will always be at least one CS) and bus_num to -1 for
dynamic allocation.  This simplifies the code needed to be written for
drivers.
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

Currently the prepare/unprepare transfer are called unconditionally.
The assumption is that every driver using the spi core queue infrastructure
has to populate the prepare and unprepare functions. This encourages
drivers to populate empty functions to prevent crashing.
This patch prevents the call to prepare/unprepare if not populated.
Signed-off-by: NShubhrajyoti D <shubhrajyoti@ti.com>
Acked-by: NLinus Walleij <linus.walleij@linaro.org>
[grant.likely: fix whitespace defect]
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

The SPI device tree support code isn't shared by any other subsystem.  It can
be moved into the core drivers/spi directory and the exported symbol can be
removed.
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>
Cc: Rob Herring <rob.herring@calxeda.com>

The status variable is guaranteed to be 0 at that location anyway.
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

We should release the lock here and enable IRQs before returning.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
[grant.likely: move unlock above dev_err() call]
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

Some systems have SPI devices located on plugin modules which are
enumerated at runtime as devices. The drivers for these plugin modules
need to register their SPI devices at probe() time so want to be able
to call spi_register_board_info() but that function is currently marked
as __init rather than __devinit so this usage isn't legal. Change the
annotation to __devinit to handle this.
Signed-off-by: NMark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

Signed-off-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

This rips the message queue in the PL022 driver out and pushes
it into (optional) common infrastructure. Drivers that want to
use the message pumping thread will need to define the new
per-messags transfer methods and leave the deprecated transfer()
method as NULL.

Most of the design is described in the documentation changes that
are included in this patch.

Since there is a queue that need to be stopped when the system
is suspending/resuming, two new calls are implemented for the
device drivers to call in their suspend()/resume() functions:
spi_master_suspend() and spi_master_resume().

ChangeLog v1->v2:
- Remove Kconfig entry and do not make the queue support optional
  at all, instead be more agressive and have it as part of the
  compulsory infrastructure.
- If the .transfer() method is implemented, delete print a small
  deprecation notice and do not start the transfer pump.
- Fix a bitrotted comment.
ChangeLog v2->v3:
- Fix up a problematic sequence courtesy of Chris Blair.
- Stop rather than destroy the queue on suspend() courtesy of
  Chris Blair.
Signed-off-by: NChris Blair <chris.blair@stericsson.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>
Tested-by: NMark Brown <broonie@opensource.wolfsonmicro.com>
Reviewed-by: NMark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

Device are added as children of the bus master's parent device, but
spi_unregister_master() looks for devices to unregister in the bus
master's children. This results in the child devices not being
unregistered.

Fix this by registering devices as direct children of the bus master.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

This uses both EXPORT_SYMBOL and THIS_MODULE, both which come from
the export.h file now.
Signed-off-by: NPaul Gortmaker <paul.gortmaker@windriver.com>

Sort the SPI makefile and enforce the naming convention spi_*.c for
spi drivers.

This change also rolls the contents of atmel_spi.h into the .c file
since there is only one user of that particular include file.

v2: - Use 'spi-' prefix instead of 'spi_' to match what seems to be
      be the predominant pattern for subsystem prefixes.
    - Clean up filenames in Kconfig and header comment blocks
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>
Acked-by: NWolfram Sang <w.sang@pengutronix.de>
Acked-by: NLinus Walleij <linus.walleij@linaro.org>

Currently the simple SPI I/O operations all take pointers to u8 * buffers
to operate on. This creates needless type compatibility issues and the
underlying spi_transfer structure uses void pointers anyway so convert the
API over to take void pointers too.
Signed-off-by: NMark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

Fixes generated by 'codespell' and manually reviewed.
Signed-off-by: NLucas De Marchi <lucas.demarchi@profusion.mobi>

Allow SPI drivers to use runtime PM and other dev_pm_ops features by
implementing dev_pm_ops for the bus. The existing bus specific suspend
and resume functions will be called if a driver does not provide dev_pm_ops
allowing for transition to the new model.
Signed-off-by: NMark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: NGrant Likely <grant.likely@secretlab.ca>
Signed-off-by: NRafael J. Wysocki <rjw@sisk.pl>

This was fixed by David Lamparter in v2.6.36-rc5 34860089 ("spi: free
children in spi_unregister_master, not siblings") and broken again in
v2.6.37-rc1~2^2~4 during the merge of 2b9603a0 ("spi: enable
spi_board_info to be registered after spi_master").
Signed-off-by: NSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: NDavid Lamparter <equinox@diac24.net>
Cc: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This patch removes convention of passing a static string as a
parameter to another static string.  The convention is intended to
reduce text usage by sharing the common bits of the string, but the
implementation is inherently fragile (a change to one format string
but not the other will nullify any possible advantage), it isn't
necessarily a net win depending on what this compiler does, and it
it reduces code readability.
Signed-off-by: NLinus Walleij <linus.walleij@stericsson.com>
[grant.likely@secretlab.ca: removed dev_dbg->dev_err hunk]
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

Currently spi_register_board_info() has to be called before its related
spi_master be registered, otherwise these board info will be just ignored.

This patch will remove this order limit, it adds a global spi master list
like the existing global board info listr. Whenever a board info or a
spi_master is registered, the spi master list or board info list
will be scanned, and a new spi device will be created if there is a
master-board info match.
Signed-off-by: NFeng Tang <feng.tang@intel.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

This patch adds the OF hook to the spi core so that devices
can automatically be registered based on device tree data.  This fixes
a problem with spi devices not binding to drivers after the cleanup of
the spi & i2c binding code.
Signed-off-by: NSinan Akman <sinan@writeme.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

introduced by 49dce689 ("spi doesn't need class_device") and bad-fixed
by 350d0076 ("spi: fix double-free on spi_unregister_master"),
spi_unregister_master would previously device_unregister all of the spi
master's siblings (instead of its children). hilarity ensues.

fix it to unregister children.
Signed-off-by: NDavid Lamparter <equinox@diac24.net>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

Move of_register_spi_devices() call from drivers to
spi_register_master(). Also change the function to use
the struct device_node pointer from master spi device
instead of passing it as function argument.
Signed-off-by: NAnatolij Gustschin <agust@denx.de>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

SPI bus locking API to allow exclusive access to the SPI bus, especially, but
not limited to, for the mmc_spi driver.

Coded according to an outline from Grant Likely; here is his
specification (accidentally swapped function names corrected):

It requires 3 things to be added to struct spi_master.
- 1 Mutex
- 1 spin lock
- 1 flag.

The mutex protects spi_sync, and provides sleeping "for free"
The spinlock protects the atomic spi_async call.
The flag is set when the lock is obtained, and checked while holding
the spinlock in spi_async().  If the flag is checked, then spi_async()
must fail immediately.

The current runtime API looks like this:
spi_async(struct spi_device*, struct spi_message*);
spi_sync(struct spi_device*, struct spi_message*);

The API needs to be extended to this:
spi_async(struct spi_device*, struct spi_message*)
spi_sync(struct spi_device*, struct spi_message*)
spi_bus_lock(struct spi_master*)  /* although struct spi_device* might
be easier */
spi_bus_unlock(struct spi_master*)
spi_async_locked(struct spi_device*, struct spi_message*)
spi_sync_locked(struct spi_device*, struct spi_message*)

Drivers can only call the last two if they already hold the spi_master_lock().

spi_bus_lock() obtains the mutex, obtains the spin lock, sets the
flag, and releases the spin lock before returning.  It doesn't even
need to sleep while waiting for "in-flight" spi_transactions to
complete because its purpose is to guarantee no additional
transactions are added.  It does not guarantee that the bus is idle.

spi_bus_unlock() clears the flag and releases the mutex, which will
wake up any waiters.

The difference between spi_async() and spi_async_locked() is that the
locked version bypasses the check of the lock flag.  Both versions
need to obtain the spinlock.

The difference between spi_sync() and spi_sync_locked() is that
spi_sync() must hold the mutex while enqueuing a new transfer.
spi_sync_locked() doesn't because the mutex is already held.  Note
however that spi_sync must *not* continue to hold the mutex while
waiting for the transfer to complete, otherwise only one transfer
could be queued up at a time!

Almost no code needs to be written.  The current spi_async() and
spi_sync() can probably be renamed to __spi_async() and __spi_sync()
so that spi_async(), spi_sync(), spi_async_locked() and
spi_sync_locked() can just become wrappers around the common code.

spi_sync() is protected by a mutex because it can sleep
spi_async() needs to be protected with a flag and a spinlock because
it can be called atomically and must not sleep
Signed-off-by: NErnst Schwab <eschwab@online.de>
[grant.likely@secretlab.ca: use spin_lock_irqsave()]
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>
Tested-by: NMatt Fleming <matt@console-pimps.org>
Tested-by: NAntonio Ospite <ospite@studenti.unina.it>

The memory for dev variable is allocated as a part of
spi_device structure memory which the dev belongs to.
Thus when the memory is released the right pointer is used.
Signed-off-by: NRoman Tereshonkov <roman.tereshonkov@nokia.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

In success case the function bus_find_device_by_name calls
get_device. In our context put_device should be called to
decrease the device count usage.
Signed-off-by: NRoman Tereshonkov <roman.tereshonkov@nokia.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h

percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files.  percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.

percpu.h -> slab.h dependency is about to be removed.  Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability.  As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.

  http://userweb.kernel.org/~tj/misc/slabh-sweep.py

The script does the followings.

* Scan files for gfp and slab usages and update includes such that
  only the necessary includes are there.  ie. if only gfp is used,
  gfp.h, if slab is used, slab.h.

* When the script inserts a new include, it looks at the include
  blocks and try to put the new include such that its order conforms
  to its surrounding.  It's put in the include block which contains
  core kernel includes, in the same order that the rest are ordered -
  alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
  doesn't seem to be any matching order.

* If the script can't find a place to put a new include (mostly
  because the file doesn't have fitting include block), it prints out
  an error message indicating which .h file needs to be added to the
  file.

The conversion was done in the following steps.

1. The initial automatic conversion of all .c files updated slightly
   over 4000 files, deleting around 700 includes and adding ~480 gfp.h
   and ~3000 slab.h inclusions.  The script emitted errors for ~400
   files.

2. Each error was manually checked.  Some didn't need the inclusion,
   some needed manual addition while adding it to implementation .h or
   embedding .c file was more appropriate for others.  This step added
   inclusions to around 150 files.

3. The script was run again and the output was compared to the edits
   from #2 to make sure no file was left behind.

4. Several build tests were done and a couple of problems were fixed.
   e.g. lib/decompress_*.c used malloc/free() wrappers around slab
   APIs requiring slab.h to be added manually.

5. The script was run on all .h files but without automatically
   editing them as sprinkling gfp.h and slab.h inclusions around .h
   files could easily lead to inclusion dependency hell.  Most gfp.h
   inclusion directives were ignored as stuff from gfp.h was usually
   wildly available and often used in preprocessor macros.  Each
   slab.h inclusion directive was examined and added manually as
   necessary.

6. percpu.h was updated not to include slab.h.

7. Build test were done on the following configurations and failures
   were fixed.  CONFIG_GCOV_KERNEL was turned off for all tests (as my
   distributed build env didn't work with gcov compiles) and a few
   more options had to be turned off depending on archs to make things
   build (like ipr on powerpc/64 which failed due to missing writeq).

   * x86 and x86_64 UP and SMP allmodconfig and a custom test config.
   * powerpc and powerpc64 SMP allmodconfig
   * sparc and sparc64 SMP allmodconfig
   * ia64 SMP allmodconfig
   * s390 SMP allmodconfig
   * alpha SMP allmodconfig
   * um on x86_64 SMP allmodconfig

8. percpu.h modifications were reverted so that it could be applied as
   a separate patch and serve as bisection point.

Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: NTejun Heo <tj@kernel.org>
Guess-its-ok-by: NChristoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>

Support two new half-duplex SPI implementation restrictions, for links
that talk to TX-only or RX-only devices.  (Existing half-duplex flavors
support both transfer directions, just not at the same time.)

Move spi_async() into the spi.c core, and stop inlining it.  Then make
that function perform error checks and reject messages that demand more
than the underlying controller can support.

Based on a patch from Marek Szyprowski which did this only for the
bitbanged GPIO driver.

Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This makes it consistent with other buses (platform, i2c, vio, ...).  I'm
not sure why we use the prefixes, but there must be a reason.

This was easy enough to do it, and I did it.
Signed-off-by: NAnton Vorontsov <avorontsov@ru.mvista.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: Grant Likely <grant.likely@secretlab.ca>
Cc: Jean Delvare <khali@linux-fr.org>
Cc: Ben Dooks <ben-linux@fluff.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Dmitry Torokhov <dtor@mail.ru>
Cc: Samuel Ortiz <sameo@openedhand.com>
Cc: "John W. Linville" <linville@tuxdriver.com>
Acked-by: NMike Frysinger <vapier.adi@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

With this patch spi drivers can use standard spi_driver.id_table and
MODULE_DEVICE_TABLE() mechanisms to bind against the devices.  Just like
we do with I2C drivers.

This is useful when a single driver supports several variants of devices
but it is not possible to detect them in run-time (like non-JEDEC chips
probing in drivers/mtd/devices/m25p80.c), and when platform_data usage is
overkill.

This patch also makes life a lot easier on OpenFirmware platforms, since
with OF we extensively use proper device IDs in modaliases.
Signed-off-by: NAnton Vorontsov <avorontsov@ru.mvista.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: Grant Likely <grant.likely@secretlab.ca>
Cc: Jean Delvare <khali@linux-fr.org>
Cc: Ben Dooks <ben-linux@fluff.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Buffer needs not be dma-safe, not rx data length.
Signed-off-by: NJiri Pirko <jpirko@redhat.com>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Move some common spi_setup() error checks into the SPI framework from the
spi_master controller drivers:

 - Add a new "mode_bits" field to spi_master

 - Use that in spi_setup to validate the spi->mode value being
   requested.  Setting this new field is now mandatory for any
   controller supporting more than vanilla SPI_MODE_0.

 - Update all spi_master drivers to:

     * Initialize that field
     * Remove current spi_setup() checks using that value.

This is a net minor code shrink.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Start moving some spi_setup() functionality into the SPI core from the
various spi_master controller drivers:

 - Make that function stop being an inline;

 - Move two common idioms from drivers into that new function:
    * Default bits_per_word to 8 if that field isn't set
    * Issue a standardized dev_dbg() message

This is a net minor source code shrink, and supports enhancments found in
some follow-up patches.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The "simplify spi_write_then_read()" patch included two regressions from
the 2.6.27 behaviors:

 - The data it wrote out during the (full duplex) read side
   of the transfer was not zeroed.

 - It fails completely on half duplex hardware, such as
   Microwire and most "3-wire" SPI variants.

So, revert that patch.  A revised version should be submitted at some
point, which can get the speedup on standard hardware (full duplex)
without breaking on less-capable half-duplex stuff.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Cc: <stable@kernel.org>		[2.6.28.x, 2.6.29.x]
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Acked-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NKay Sievers <kay.sievers@vrfy.org>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Make the SPI external GPIO expander drivers register themselves at
subsys_initcall() time when they're statically linked, and make the SPI
core do its driver model initialization earlier so that's safe.

SOC-integrated GPIOs are available starting very early -- often before
initcalls start to run, or earily in arch_initcall() at latest -- so this
improves consistency, letting more subsystems rely on GPIOs being usable
by their own subsys_initcall() code.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Modify spi_write_then_read() to use one transfer.  This speeds up all
callers, and is a minor code shrink.
Signed-off-by: NVernon Sauder <Vernon.Sauder@gmail.com>
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

When reviewing a recent patch I noticed a potential trouble spot in the
registration of new SPI devices.  The SPI master driver is told to set
the device up before adding it to the driver model, so that it's always
properly set up when probe() is called.  (This is important, because in
the case of inverted chipselects, this device can make the bus misbehave
until it's properly deselected.  It's got to be set up even if no driver
binds to the device.)

The trouble spot is that it doesn't first verify that no other device
has been added using that chipselect.  If such a device has been added,
its configuration gets trashed.  (Fortunately this has not been a common
error!)

The fix here adds an explicit check, and a mutex to protect the relevant
critical region.

[akpm@linux-foundation.org: make the lock local to spi_add_device()]
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>