Signed-off-by: NAlberto Panizzo <maramaopercheseimorto@gmail.com>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Based on Arjan's suggestion, extend the list of ops structures that should
be const.
Signed-off-by: NEmese Revfy <re.emese@gmail.com>
Cc: Andy Whitcroft <apw@shadowen.org>
Cc: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Here is a small code snippet, which will be complained about by
checkpatch.pl:

#define __STRUCT_KFIFO_COMMON(recsize, ptrtype) \
	union { \
		struct { \
			unsigned int	in; \
			unsigned int	out; \
		}; \
		char		rectype[recsize]; \
		ptrtype		*ptr; \
		const ptrtype	*ptr_const; \
	};

This construct is legal and safe, so checkpatch.pl should accept this.  It
should be also true for struct defined in a macro.

Add the `struct' and `union' keywords to the exceptions list of the
checkpatch.pl script, to prevent error message "Macros with multiple
statements should be enclosed in a do - while loop".  Otherwise it is not
possible to build a struct or union with a macro.
Signed-off-by: NStefani Seibold <stefani@seibold.net>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

checkpatch falsely complained about '__initconst' because it thought the
'const' needed a space before.  Fix this by changing the list of
attributes:

- add '__initconst'
- force plain 'init' to contain a word-boundary at the end
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NJoe Perches <joe@perches.com>
Cc: Andy Whitcroft <apw@canonical.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

In case if the statement and the conditional are in one line, the line
appears in the report doubly.

And items of this check have no blank line before the next item.

This patch fixes these trivial problems, to improve readability of the
report.

[sample.c]
  > if (cond1
  >        && cond2
  >        && cond3) func_foo();
  >
  > if (cond4) func_bar();

Before:
  > ERROR: trailing statements should be on next line
  > #1: FILE: sample.c:1:
  > +if (cond1
  > [...]
  > +       && cond3) func_foo();
  > ERROR: trailing statements should be on next line
  > #5: FILE: sample.c:5:
  > +if (cond4) func_bar();
  > +if (cond4) func_bar();
  > total: 2 errors, 0 warnings, 5 lines checked

After:
  > ERROR: trailing statements should be on next line
  > #1: FILE: sample.c:1:
  > +if (cond1
  > [...]
  > +       && cond3) func_foo();
  >
  > ERROR: trailing statements should be on next line
  > #5: FILE: sample.c:5:
  > +if (cond4) func_bar();
  >
  > total: 2 errors, 0 warnings, 5 lines checked
Signed-off-by: NHidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Cc: Andy Whitcroft <apw@canonical.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

sizeof(&foo) is frequently an error.  Warn on its use.
Signed-off-by: NJoe Perches <joe@perches.com>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This card reader doesn't advertise, however DMA works well.  Probably
windows SDHCI driver assumes that all readers support DMA and thus we see
that bug.
Signed-off-by: NVasily Khoruzhick <anarsoul@gmail.com>
Tested-by: NMaxim Levitsky <maximlevitsky@gmail.com>
Signed-off-by: NMaxim Levitsky <maximlevitsky@gmail.com>
Cc: Harald Welte <HaraldWelte@viatech.com>
Cc: Norbert Preining <preining@logic.at>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

kunmap_atomic() accepts a pointer to any location in the page so we do not
need the subtraction and cast.
Signed-off-by: NNicolas Ferre <nicolas.ferre@atmel.com>
Cc: Wolfgang Muees <wolfgang.mues@auerswald.de>
Cc: Andrew Victor <avictor.za@gmail.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

We used to manage features and differences on a per-cpu basis.  As several
cpus share the same mci revision, this patch aggregates cpus that have the
same IP revision in one defined constant.  We use the
at91mci_is_mci1rev2xx() funtion name not to mess with newer Atmel sd/mmc
IP called "MCI2".  _rev2 naming could have been confusing...
Signed-off-by: NNicolas Ferre <nicolas.ferre@atmel.com>
Cc: Wolfgang Muees <wolfgang.mues@auerswald.de>
Cc: Andrew Victor <avictor.za@gmail.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

According to the datasheets AT91SAM9261 does not support SDIO interrupts,
and AT91SAM9260/9263 have an erratum requiring 4bit mode while using slot
B for the interrupt to work.
Signed-off-by: NNicolas Ferre <nicolas.ferre@atmel.com>
Cc: Wolfgang Muees <wolfgang.mues@auerswald.de>
Cc: Andrew Victor <avictor.za@gmail.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This patch is setting some max_ variables for the IO elevator, so the
elevator will put requests for large data blocks to the driver.  This is
critical for

a) speed

and

b) wear leveling of the flash chip controller: Otherwise the controller
   will treat the SD card badly with millions of single 4 KByte write
   commands.  This will lead to a shorter life time for the SD cards.
Signed-off-by: NWolfgang Muees <wolfgang.mues@auerswald.de>
Signed-off-by: NNicolas Ferre <nicolas.ferre@atmel.com>
Cc: Andrew Victor <avictor.za@gmail.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Convert the read to use the DMA buffer as well.  The old code was doing
double-buffering DMA with the PDC; no way to make it work.  Replace it
with a single-PDC approach.  It also simplify things removing the need for
a pre_dma_read() function.

[nicolas.ferre@atmel.com coding style modifications]
Signed-off-by: NWolfgang Muees <wolfgang.mues@auerswald.de>
Signed-off-by: NNicolas Ferre <nicolas.ferre@atmel.com>
Cc: Andrew Victor <avictor.za@gmail.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The TX DMA buffer is allocated only once, because the
allocation/deallocation of the buffer for EACH chunk of data is
time-consuming and prone to memory fragmentation.

Using a coherent DMA buffer avoids extra data cache calls.

[nicolas.ferre@atmel.com: coding style modifications]
Signed-off-by: NWolfgang Muees <wolfgang.mues@auerswald.de>
Signed-off-by: NNicolas Ferre <nicolas.ferre@atmel.com>
Cc: Andrew Victor <avictor.za@gmail.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Fix two timeout errors, one for slow SDHC cards and one for slow users
while inserting SD cards.
Signed-off-by: NWolfgang Muees <wolfgang.mues@auerswald.de>
Signed-off-by: NNicolas Ferre <nicolas.ferre@atmel.com>
Cc: Andrew Victor <avictor.za@gmail.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Fixes two pointer errors, one which leads to memory overwrites if used
with large chunks of data.
Signed-off-by: NWolfgang Muees <wolfgang.mues@auerswald.de>
Signed-off-by: NNicolas Ferre <nicolas.ferre@atmel.com>
Cc: Andrew Victor <avictor.za@gmail.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NLars-Peter Clausen <lars@metafoo.de>
Cc: Ben Dooks <ben-linux@fluff.org>
Cc: <linux-mmc@vger.kernel.org>
Cc: <stable@kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

If no platform_data was givin to the device it's going to use it's default
platform data struct which has all fields initialized to zero.  As a
result the driver is going to try to request gpio0 both as write protect
and card detect pin.  Which of course will fail and makes the driver
unusable

Previously to the introduction of no_wprotect and no_detect the behavior
was to assume that if no platform data was given there is no write protect
or card detect pin.  This patch restores that behavior.
Signed-off-by: NLars-Peter Clausen <lars@metafoo.de>
Cc: Ben Dooks <ben-linux@fluff.org>
Cc: <linux-mmc@vger.kernel.org>
Cc: <stable@kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

And bring them back to 4-bit mode during resume.
Signed-off-by: NDaniel Drake <dsd@laptop.org>
Signed-off-by: NNicolas Pitre <nico@marvell.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Some SDIO cards may suspend while keeping function interrupts active
especially in the powered suspend case.  Upon resume we need to kick the
SDIO interrupt thread to check for pending interrupts and to restart card
IRQ detection at the host controller level.
Signed-off-by: NNicolas Pitre <nico@marvell.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Seen on a Marvell 8686 SDIO card and Via VX855 controller: we must avoid
sending CMD3/5/7 on a resume where power has been maintained, because the
8686 will refuse to respond to them and the MMC stack will give up on the
card.
Signed-off-by: NChris Ball <cjb@laptop.org>
Signed-off-by: NNicolas Pitre <nico@marvell.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Tested with an XO v1.5 from OLPC.
Signed-off-by: NNicolas Pitre <nico@marvell.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This patch series provides the core changes needed to allow SDIO cards to
remain powered and active while the host system is suspended, and let them
wake up the host system when needed.  This is used to implement
wake-on-lan with SDIO wireless cards at the moment.  Patches to add that
support to the libertas driver will be posted separately.

This patch:

Some SDIO cards have the ability to keep on running autonomously when the
host system is suspended, and wake it up when needed.  This however
requires that the host controller preserve power to the card, and
configure itself appropriately for wake-up.

There is however 4 layers of abstractions involved: the host controller
driver, the MMC core code, the SDIO card management code, and the actual
SDIO function driver.  To make things simple and manageable, host drivers
must advertise their PM capabilities with a feature bitmask, then function
drivers can query and set those features from their suspend method.  Then
each layer in the suspend call chain is expected to act upon those bits
accordingly.

[akpm@linux-foundation.org: fix typo in comment]
Signed-off-by: NNicolas Pitre <nico@marvell.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

sdhci_set_adma_desc() is using byte-writes to write data in a specified
order into memory.  Change to using __le16 for the two byte and __le32 for
the four byte cases and use the cpu_to_{le16,le32} to do the conversion
before writing.

This will reduce the size of the code and the number of writes as we no
longer need to chop the data up before writing.

As an example on ARM S3C64XX SoC, in little-endian configuration:

 000000d4 <sdhci_set_adma_desc>:
-      d8:	e1a0c423 	lsr	ip, r3, #8
-      dc:	e1a0ec21 	lsr	lr, r1, #24
-      e0:	e1a04821 	lsr	r4, r1, #16
-      e4:	e1a05421 	lsr	r5, r1, #8
-      e8:	e1a06442 	asr	r6, r2, #8
-      ec:	e5c0c001 	strb	ip, [r0, #1]
-      f0:	e5c0e007 	strb	lr, [r0, #7]
-      f4:	e5c04006 	strb	r4, [r0, #6]
-      f8:	e5c05005 	strb	r5, [r0, #5]
-      fc:	e5c01004 	strb	r1, [r0, #4]
-     100:	e5c06003 	strb	r6, [r0, #3]
-     104:	e5c02002 	strb	r2, [r0, #2]
-     108:	e5c03000 	strb	r3, [r0]
+      d4:	e5801004 	str	r1, [r0, #4]
+      d8:	e1c030b0 	strh	r3, [r0]
+      dc:	e1c020b2 	strh	r2, [r0, #2]
Signed-off-by: NBen Dooks <ben-linux@fluff.org>
Cc: Pierre Ossman <pierre@ossman.eu>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The code to write the ADMA descriptor into memory is repeated several
times throughout sdhci_adma_table_pre, and thus should be moved into a
common function.  This will also be useful if the patch to make the write
more efficient is accepted.
Signed-off-by: NBen Dooks <ben-linux@fluff.org>
Cc: Pierre Ossman <pierre@ossman.eu>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Some SDIO cards expect byte transfers not to exceed the configured block
transfer size.  Add a quirk to that effect.

Patches to make use of this quirk will be sent separately.
Signed-off-by: NBing Zhao <bzhao@marvell.com>
Signed-off-by: NNicolas Pitre <nico@marvell.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The hardcoded value doesn't really work for all cards.
Signed-off-by: NCliff Cai <cliff.cai@analog.com>
Signed-off-by: NMike Frysinger <vapier@gentoo.org>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The host/Kconfig file is only included when MMC is selected.
Signed-off-by: NMike Frysinger <vapier@gentoo.org>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The local sg variable is only used with BF54x code.
Signed-off-by: NMike Frysinger <vapier@gentoo.org>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This prevents those without an Atmel chip having a line in kernel
configuration which says "Atmel SD/MMC Driver" without any option.
Signed-off-by: NJonathan Cameron <jic23@cam.ac.uk>
Signed-off-by: NNicolas Ferre <nicolas.ferre@atmel.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Add support for 8bit MMC cards.  The controller data width is configurable
depending on the wires setting in the platform data structure.

MMC 8bit is tested on OMAPL137 and MMC 4bit is tested on OMAPL138 EVM.
Signed-off-by: NVipin Bhandari <vipin.bhandari@ti.com>
Cc: David Brownell <david-b@pacbell.net>
Cc: Chaithrika U S <chaithrika@ti.com>
Cc: Sudhakar Rajashekhara <sudhakar.raj@ti.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This patch solves nasty problem original driver has.

Original goal of the ricoh_mmc was to disable this device because then,
mmc cards can be read using standard SDHCI controller, thus avoiding
writing of yet another driver.

However, the act of disablement, makes other pci functions that belong to
this controller (xD and memstick) shift up one level, thus pci core has
now wrong idea about these devices.

To fix this issue, this patch moves the driver into the pci quirk section,
thus it is executes before the pci is enumerated, and therefore solving
that issue, also same sequence of commands is performed on resume for same
reasons.

Also regardless of the above, this way is cleaner.  You still need to set
CONFIG_MMC_RICOH_MMC to enable this quirk
Signed-off-by: NMaxim Levitsky <maximlevitsky@gmail.com>
Acked-by: NPhilip Langdale <philipl@overt.org>
Acked-by: NWolfram Sang <w.sang@pengutronix.de>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

fs/compat_ioctl.c: In function 'do_ioctl_trans':
fs/compat_ioctl.c:534: warning: 'karg' may be used uninitialized in this function
fs/compat_ioctl.c:533: warning: 'kcmd' may be used uninitialized in this function
fs/compat_ioctl.c:656: warning: 'ret' may be used uninitialized in this function

Reduces text size by 44 bytes.

If someone calls one of these functions with an unexpected argument, the
code's buggy as-is.

Amerigo Wang <amwang@redhat.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Acked-by: NArnd Bergmann <arnd@arndb.de>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Replace open-coded loop with for_each_set_bit().
Signed-off-by: NAkinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The function name must be followed by a space, hypen, space, and a short
description.
Signed-off-by: NBen Hutchings <ben@decadent.org.uk>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Build list_sort() only for configs that need it -- those that don't save
~581 bytes (i386).
Signed-off-by: NDon Mullis <don.mullis@gmail.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Artem Bityutskiy <dedekind@infradead.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Clarify and correct header comment of list_sort().
Signed-off-by: NDon Mullis <don.mullis@gmail.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Artem Bityutskiy <dedekind@infradead.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

XFS and UBIFS can pass long lists to list_sort(); this alternative
implementation scales better, reaching ~3x performance gain when list
length exceeds the L2 cache size.

Stand-alone program timings were run on a Core 2 duo L1=32KB L2=4MB,
gcc-4.4, with flags extracted from an Ubuntu kernel build.  Object size is
581 bytes compared to 455 for Mark J.  Roberts' code.

Worst case for either implementation is a list length just over a power of
two, and to roughly the same degree, so here are timing results for a
range of 2^N+1 lengths.  List elements were 16 bytes each including malloc
overhead; initial order was random.

                      time (msec)
                      Tatham-Roberts
                      |       generic-Mullis-v2
loop_count  length    |       |    ratio
4000000       2     206     294    1.427
2000000       3     176     227    1.289
1000000       5     199     172    0.864
 500000       9     235     178    0.757
 250000      17     243     182    0.748
 125000      33     261     196    0.750
  62500      65     277     209    0.754
  31250     129     292     219    0.75
  15625     257     317     235    0.741
   7812     513     340     252    0.741
   3906    1025     362     267    0.737
   1953    2049     388     283    0.729  ~ L1 size
    976    4097     556     323    0.580
    488    8193     678     361    0.532
    244   16385     773     395    0.510
    122   32769     844     418    0.495
     61   65537     917     454    0.495
     30  131073    1128     543    0.481
     15  262145    2355     869    0.369  ~ L2 size
      7  524289    5597    1714    0.306
      3 1048577    6218    2022    0.325

Mark's code does not actually implement the usual or generic mergesort,
but rather a variant from Simon Tatham described here:

    http://www.chiark.greenend.org.uk/~sgtatham/algorithms/listsort.html

Simon's algorithm performs O(log N) passes over the entire input list,
doing merges of sublists that double in size on each pass.  The generic
algorithm instead merges pairs of equal length lists as early as possible,
in recursive order.  For either algorithm, the elements that extend the
list beyond power-of-two length are a special case, handled as nearly as
possible as a "rounding-up" to a full POT.

Some intuition for the locality of reference implications of merge order
may be gotten by watching this animation:

    http://www.sorting-algorithms.com/merge-sort

Simon's algorithm requires only O(1) extra space rather than the generic
algorithm's O(log N), but in my non-recursive implementation the actual
O(log N) data is merely a vector of ~20 pointers, which I've put on the
stack.

Long-running list_sort() calls: If the list passed in may be long, or the
client's cmp() callback function is slow, the client's cmp() may
periodically invoke cond_resched() to voluntarily yield the CPU.  All
inner loops of list_sort() call back to cmp().

Stability of the sort: distinct elements that compare equal emerge from
the sort in the same order as with Mark's code, for simple test cases.  A
boot-time test is provided to verify this and other correctness
requirements.

A kernel that uses drm.ko appears to run normally with this change; I have
no suitable hardware to similarly test the use by UBIFS.

[akpm@linux-foundation.org: style tweaks, fix comment, make list_sort_test __init]
Signed-off-by: NDon Mullis <don.mullis@gmail.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Artem Bityutskiy <dedekind@infradead.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NAndré Goddard Rosa <andre.goddard@gmail.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Joe Perches <joe@perches.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Removes 32 bytes on core2 with gcc 4.4.1:
   text    data     bss     dec     hex filename
   3196       0       0    3196     c7c lib/string-BEFORE.o
   3164       0       0    3164     c5c lib/string-AFTER.o
Signed-off-by: NAndré Goddard Rosa <andre.goddard@gmail.com>
Cc: Joe Perches <joe@perches.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>