memcg: avoid css_get()

Now, memory cgroup increments css(cgroup subsys state)'s reference count
per a charged page.  And the reference count is kept until the page is
uncharged.  But this has 2 bad effect.

 1. Because css_get/put calls atomic_inc()/dec, heavy call of them
    on large smp will not scale well.
 2. Because css's refcnt cannot be in a state as "ready-to-release",
    cgroup's notify_on_release handler can't work with memcg.
 3. css's refcnt is atomic_t, it means smaller than 32bit. Maybe too small.

This has been a problem since the 1st merge of memcg.

This is a trial to remove css's refcnt per a page. Even if we remove
refcnt, pre_destroy() does enough synchronization as
  - check res->usage == 0.
  - check no pages on LRU.

This patch removes css's refcnt per page.  Even after this patch, at the
1st look, it seems css_get() is still called in try_charge().

But the logic is.

  - If a memcg of mm->owner is cached one, consume_stock() will work.
    At success, return immediately.
  - If consume_stock returns false, css_get() is called and go to
    slow path which may be blocked. At the end of slow path,
    css_put() is called and restart from the start if necessary.

So, in the fast path, we don't call css_get() and can avoid access to
shared counter. This patch can make the most possible case fast.

Here is a result of multi-threaded page fault benchmark.

[Before]
    25.32%  multi-fault-all  [kernel.kallsyms]      [k] clear_page_c
     9.30%  multi-fault-all  [kernel.kallsyms]      [k] _raw_spin_lock_irqsave
     8.02%  multi-fault-all  [kernel.kallsyms]      [k] try_get_mem_cgroup_from_mm <=====(*)
     7.83%  multi-fault-all  [kernel.kallsyms]      [k] down_read_trylock
     5.38%  multi-fault-all  [kernel.kallsyms]      [k] __css_put
     5.29%  multi-fault-all  [kernel.kallsyms]      [k] __alloc_pages_nodemask
     4.92%  multi-fault-all  [kernel.kallsyms]      [k] _raw_spin_lock_irq
     4.24%  multi-fault-all  [kernel.kallsyms]      [k] up_read
     3.53%  multi-fault-all  [kernel.kallsyms]      [k] css_put
     2.11%  multi-fault-all  [kernel.kallsyms]      [k] handle_mm_fault
     1.76%  multi-fault-all  [kernel.kallsyms]      [k] __rmqueue
     1.64%  multi-fault-all  [kernel.kallsyms]      [k] __mem_cgroup_commit_charge

[After]
    28.41%  multi-fault-all  [kernel.kallsyms]      [k] clear_page_c
    10.08%  multi-fault-all  [kernel.kallsyms]      [k] _raw_spin_lock_irq
     9.58%  multi-fault-all  [kernel.kallsyms]      [k] down_read_trylock
     9.38%  multi-fault-all  [kernel.kallsyms]      [k] _raw_spin_lock_irqsave
     5.86%  multi-fault-all  [kernel.kallsyms]      [k] __alloc_pages_nodemask
     5.65%  multi-fault-all  [kernel.kallsyms]      [k] up_read
     2.82%  multi-fault-all  [kernel.kallsyms]      [k] handle_mm_fault
     2.64%  multi-fault-all  [kernel.kallsyms]      [k] mem_cgroup_add_lru_list
     2.48%  multi-fault-all  [kernel.kallsyms]      [k] __mem_cgroup_commit_charge

Then, 8.02% of try_get_mem_cgroup_from_mm() disappears because this patch
removes css_tryget() in it. (But yes, this is an extreme case.)

Signed-off-by: Daisuke Nishimura <[email protected]>
Signed-off-by: KAMEZAWA Hiroyuki <[email protected]>
Cc: Daisuke Nishimura <[email protected]>
Cc: Balbir Singh <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

memcg: use find_lock_task_mm() in memory cgroups oom

When the OOM killer scans task, it check a task is under memcg or
not when it's called via memcg's context.

But, as Oleg pointed out, a thread group leader may have NULL ->mm
and task_in_mem_cgroup() may do wrong decision. We have to use
find_lock_task_mm() in memcg as generic OOM-Killer does.

Signed-off-by: KAMEZAWA Hiroyuki <[email protected]>
Cc: Oleg Nesterov <[email protected]>
Cc: Daisuke Nishimura <[email protected]>
Cc: Balbir Singh <[email protected]>
Reviewed-by: Minchan Kim <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

memcg: remove mem from arg of charge_common

mem_cgroup_charge_common() is always called with @mem = NULL, so it's
meaningless.  This patch removes it.

Signed-off-by: Daisuke Nishimura <[email protected]>
Cc: Balbir Singh <[email protected]>
Acked-by: KAMEZAWA Hiroyuki <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

memcg: remove redundant code

- try_get_mem_cgroup_from_mm() calls rcu_read_lock/unlock by itself, so we
  don't have to call them in task_in_mem_cgroup().
- *mz is not used in __mem_cgroup_uncharge_common().
- we don't have to call lookup_page_cgroup() in mem_cgroup_end_migration()
  after we've cleared PCG_MIGRATION of @oldpage.
- remove empty comment.
- remove redundant empty line in mem_cgroup_cache_charge().

Signed-off-by: Daisuke Nishimura <[email protected]>
Acked-by: Balbir Singh <[email protected]>
Acked-by: KAMEZAWA Hiroyuki <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

memcg: clean up waiting move acct

Now, for checking a memcg is under task-account-moving, we do css_tryget()
against mc.to and mc.from.  But this is just complicating things.  This
patch makes the check easier.

This patch adds a spinlock to move_charge_struct and guard modification of
mc.to and mc.from.  By this, we don't have to think about complicated
races arount this not-critical path.

[[email protected]: don't crash on a null memcg being passed]
Signed-off-by: KAMEZAWA Hiroyuki <[email protected]>
Signed-off-by: Balbir Singh <[email protected]>
Cc: Daisuke Nishimura <[email protected]>
Cc: Balbir Singh <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

memcg: clean up try_charge main loop

mem_cgroup_try_charge() has a big loop in it and seems to be hard to read.
 Most of routines are for slow path.  This patch moves codes out from the
loop and make it clear what's done.

Summary:
 - refactoring a function to detect a memcg is under acccount move or not.
 - refactoring a function to wait for the end of moving task acct.
 - refactoring a main loop('s slow path) as a function and make it clear
   why we retry or quit by return code.
 - add fatal_signal_pending() check for bypassing charge loops.

Signed-off-by: KAMEZAWA Hiroyuki <[email protected]>
Cc: Daisuke Nishimura <[email protected]>
Cc: Balbir Singh <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

memcg: remove experimental from swap account config

It's 11 months since we changed swap_map[] to indicates SWAP_HAS_CACHE.
Since that, memcg's swap accounting has been very stable and it seems
it can be maintained.

So, I'd like to remove EXPERIMENTAL from the config.

Acked-by: Balbir Singh <[email protected]>
Acked-by: Daisuke Nishimura <[email protected]>
Signed-off-by: KAMEZAWA Hiroyuki <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

blkdev: cgroup whitelist permission fix

The cgroup device whitelist code gets confused when trying to grant
permission to a disk partition that is not currently open.  Part of
blkdev_open() includes __blkdev_get() on the whole disk.

Basically, the only ways to reliably allow a cgroup access to a partition
on a block device when using the whitelist are to 1) also give it access
to the whole block device or 2) make sure the partition is already open in
a different context.

The patch avoids the cgroup check for the whole disk case when opening a
partition.

Addresses https://bugzilla.redhat.com/show_bug.cgi?id=589662

Signed-off-by: Chris Wright <[email protected]>
Acked-by: Serge E. Hallyn <[email protected]>
Tested-by: Serge E. Hallyn <[email protected]>
Reported-by: Vivek Goyal <[email protected]>
Cc: Al Viro <[email protected]>
Cc: Christoph Hellwig <[email protected]>
Cc: "Daniel P. Berrange" <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

cgroups: save space for the terminator

The original code didn't leave enough space for a NULL terminator.  These
strings are copied with strcpy() into fixed length buffers in
cgroup_root_from_opts().

Signed-off-by: Dan Carpenter <[email protected]>
Acked-by: Serge E. Hallyn <[email protected]>
Reviewd-by: KAMEZAWA Hiroyuki <[email protected]>
Cc: Paul Menage <[email protected]>
Cc: Li Zefan <[email protected]>
Cc: Ben Blum <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

Documentation/padata.txt: fix typos etc.

Fix typos & grammar.
Use CPU instead of cpu in text.

Signed-off-by: Randy Dunlap <[email protected]>
Acked-by: Steffen Klassert <[email protected]>
Cc: Herbert Xu <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

Documentation/00-INDEX: remove reference to exception.txt

The exception.txt has been removed from the Documentation directory.  So
update the index file for it.

Signed-off-by: Huang Shijie <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

docbook: need xmldoclinks for all doc types

$ rm -rf build
$ mkdir build
$ cp .config build
$ make O=build htmldocs
...
xmlto: linux-2.6/build/Documentation/DocBook/media.xml
does not validate (status 3)
xmlto: Fix document syntax or use --skip-validation option
linux-2.6/build/Documentation/DocBook/media.xml:4:
warning: failed to load external entity
"linux-2.6/build/Documentation/DocBook/media-entities.tmpl"

We need the xmldoclinks built for any document types built from the
XML sources.

Signed-off-by: Ben Hutchings <[email protected]>
Acked-by: Andy Whitcroft <[email protected]>
Cc: Randy Dunlap <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

Documentation/networking/wavelan.txt: deleted, not in tree

Commit 1d794e3b353b ("Staging: wavelan: delete the driver") removed the
source, so remove the documentation as well.

Signed-off-by: Joe Perches <[email protected]>
Cc: Jean Tourrilhes <[email protected]>
Acked-by: Randy Dunlap <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

mtd/nand_base: fix kernel-doc warnings & typos

Fix mtd/nand_base.c kernel-doc warnings and typos.

Warning(drivers/mtd/nand/nand_base.c:893): No description found for parameter 'mtd'
Warning(drivers/mtd/nand/nand_base.c:893): No description found for parameter 'ofs'
Warning(drivers/mtd/nand/nand_base.c:893): No description found for parameter 'len'
Warning(drivers/mtd/nand/nand_base.c:893): No description found for parameter 'invert'
Warning(drivers/mtd/nand/nand_base.c:930): No description found for parameter 'mtd'
Warning(drivers/mtd/nand/nand_base.c:930): No description found for parameter 'ofs'
Warning(drivers/mtd/nand/nand_base.c:930): No description found for parameter 'len'
Warning(drivers/mtd/nand/nand_base.c:987): No description found for parameter 'mtd'
Warning(drivers/mtd/nand/nand_base.c:987): No description found for parameter 'ofs'
Warning(drivers/mtd/nand/nand_base.c:987): No description found for parameter 'len'
Warning(drivers/mtd/nand/nand_base.c:2087): No description found for parameter 'len'

Signed-off-by: Randy Dunlap <[email protected]>
Cc: David Woodhouse <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

fusion: fix kernel-doc warnings

Fix (delete) empty kernel-doc lines/warnings:
Warning(drivers/message/fusion/mptbase.c:6916): bad line:
Warning(drivers/message/fusion/mptbase.c:7060): bad line:

Signed-off-by: Randy Dunlap <[email protected]>
Cc: Eric Moore <[email protected]>
Cc: James Bottomley <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

reiserfs: remove unused local `wait'

Signed-off-by: Changli Gao <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/video/igafb.c: make igafb_setup() and igafb_init() static

Cc: Kulikov Vasiliy <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/video/igafb.c: introduce lost 'return'

If iga_init() fails, code releases resources and continues to use it.  It
seems that after releasing resources 'return' should be.

Signed-off-by: Kulikov Vasiliy <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

fbmem: VM_IO set, but not propagated

When we setup up the VMA flags for the mmap flag and we end up using the
fallback mmap functionality we set the vma->vm_flags |= VM_IO.  However we
neglect to propagate the flag to the vma->vm_page_prot.

This bug was found when Linux kernel was running under Xen.  In that
scenario, any page that has VM_IO flag to it, means that it MUST be a
MMIO/VRAM backend memory , _not_ System RAM.  That is what the fbmem.c
does: sets VM_IO, ioremaps the region - everything is peachy.

Well, not exactly.  The vm_page_prot does not get the relevant PTE flags
set (_PAGE_IOMAP) which under Xen is a death-kneel to pages that are
referencing real physical devices but don't have that flag set.

This patch fixes this.

Signed-off-by: Konrad Rzeszutek Wilk <[email protected]>
Signed-off-by: Daniel De Graaf <[email protected]>
Tested-by: Eamon Walsh <[email protected]>
Cc: Florian Tobias Schandinat <[email protected]>
Cc: Dave Airlie <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

arm: samsung: remove pixclock from several boards

Since "s3c-fb: Automatically calculate pixel clock when none is given",
there's no need for manually calculating the pixel clock anymore so remove
these lines and add the correct refresh rate where appropriately.

Signed-off-by: Maurus Cuelenaere <[email protected]>
Cc: Pawel Osciak <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Cc: Kyungmin Park <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Russell King <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: automatically calculate pixel clock when none is given

Add a simple algorithm which calculates the pixel clock based on the video
mode parameters.  This is only done when no pixel clock is supplied
through the platform data.

This allows drivers to omit the pixel clock data and thus share the
algorithm used for calculating it.

Signed-off-by: Maurus Cuelenaere <[email protected]>
Cc: Pawel Osciak <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Cc: Kyungmin Park <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Russell King <[email protected]>
Tested-by: Donghwa Lee <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: add support for DMA channel control on S5PV210

S5PV210 SoCs allow enabling/disabling DMA channels per window.  For a
window to display data from framebuffer memory, its channel has to be
enabled.

Signed-off-by: Pawel Osciak <[email protected]>
Signed-off-by: Marek Szyprowski <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Ben Dooks <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: fix section mismatch

This patch fixes the following section mismatch errors:

WARNING: vmlinux.o(.data+0x20b40): Section mismatch in reference from the variable s3c_fb_driver_ids to the (unknown reference) .devinit.data:(unknown)
The variable s3c_fb_driver_ids references
the (unknown reference) __devinitdata (unknown)
If the reference is valid then annotate the
variable with __init* or __refdata (see linux/init.h) or name the variable:
*driver, *_template, *_timer, *_sht, *_ops, *_probe, *_probe_one, *_console,

WARNING: vmlinux.o(.data+0x20b58): Section mismatch in reference from the variable s3c_fb_driver_ids to the (unknown reference) .devinit.data:(unknown)
The variable s3c_fb_driver_ids references
the (unknown reference) __devinitdata (unknown)
If the reference is valid then annotate the
variable with __init* or __refdata (see linux/init.h) or name the variable:
*driver, *_template, *_timer, *_sht, *_ops, *_probe, *_probe_one, *_console,

WARNING: vmlinux.o(.data+0x20b70): Section mismatch in reference from the variable s3c_fb_driver_ids to the (unknown reference) .devinit.data:(unknown)
The variable s3c_fb_driver_ids references
the (unknown reference) __devinitdata (unknown)
If the reference is valid then annotate the
variable with __init* or __refdata (see linux/init.h) or name the variable:
*driver, *_template, *_timer, *_sht, *_ops, *_probe, *_probe_one, *_console,

Signed-off-by: Marek Szyprowski <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: protect window-specific registers during updates

Newer hardware (S3C6410, S5P) have the ability to block updates from
shadow registers during reconfiguration.  Add protect calls for set_par
and clear protection when resetting.

Signed-off-by: Pawel Osciak <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: correct window osd size and alpha register handling

S3C64xx and S5P OSD registers for OSD size and alpha are as follows:
VIDOSDC: win 0 - size, win 1-4: alpha
VIDOSDD: win 1-2 - size; not present for windows 0, 3 and 4

Signed-off-by: Pawel Osciak <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: add SHADOWCON shadow register locking support for S5PV210

S5PV210 allows per-window locking of register value updates from shadow
registers.

Signed-off-by: Pawel Osciak <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: window 3 of 64xx+ does not have an osd_d register

Signed-off-by: Pawel Osciak <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: add wait for VSYNC ioctl

Add VSYNC interrupt support and an ioctl that allows waiting for it.
Interrupts are turned on only when needed.

Signed-off-by: Pawel Osciak <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: add support for display panning

Supports all bpp modes.

The PRTCON register is used to disable in-hardware updates of registers
that store start and end addresses of framebuffer memory.  This prevents
display corruption in case we do not make it before VSYNC with updating
them atomically.  With this feature there is no need to wait for a VSYNC
interrupt before each such update.

Signed-off-by: Pawel Osciak <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: add device name initialization

Add framebuffer device name initialization calls for S3C2443, S3C64xx and
S5P machines.

Signed-off-by: Pawel Osciak <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: separate S5PC100 and S5PV210 framebuffer driver data structures

S5PC100 and S5PV210 framebuffer devices differ slightly in terms of
available registers and their driver data structures have to be separate.
Those differences include dissimilar ways to control shadow register
updates.

Signed-off-by: Pawel Osciak <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: correct FRAMESEL1 bitfield defines for VIDINTCON0 register

FRAMESEL1 bitfield starts on 13th bit, not on 14th.

Signed-off-by: Pawel Osciak <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Acked-by: Ben Dooks <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: fix various null references on framebuffer memory alloc failure

The following problems were found in the above situation:

sfb->windows[win] was being assigned at the end of s3c_fb_probe_win only.
This resulted in passing a NULL to s3c_fb_release_win if probe_win
returned early and a memory leak.

dma_free_writecombine does not allow its third argument to be NULL.

fb_dealloc_cmap does not verify whether its argument is not NULL.

Signed-off-by: Pawel Osciak <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: integrate palette setup code into main driver

Remove the palette setup code from the header files and put it into the
main driver.

Signed-off-by: Ben Dooks <[email protected]>
Signed-off-by: Pawel Osciak <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: KyungMin Park <kyungmin.park.samsung.com>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: udpate to support s3c2416/s3c2443 style hardware

Update the variant and window variant structures with the necessary
changes to support the older style of hardware where these are not in the
same place.

Add the support for the s3c2443/s3c2416 hardware by using the
platform-device s3c2443 to cover both, and add the initialisation data for
these.

Also change to including just the v4 header files for the moment until the
last of the merging of these is sorted out.

Signed-off-by: Ben Dooks <[email protected]>
Signed-off-by: Pawel Osciak <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: KyungMin Park <kyungmin.park.samsung.com>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: initial move to unifying the header files

Remove the various header files that configure this driver and use the
platform device name to select the correct configuration at probe time.

Currently this does not remove the header files, only updates the driver
and the relevant platform files.

Signed-off-by: Ben Dooks <[email protected]>
Signed-off-by: Pawel Osciak <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: KyungMin Park <kyungmin.park.samsung.com>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: only init window colour key controls for windows with blending

The driver clears all windows, but also sets the windows' colour key
controls at the same time.  However, the last window does not have these
registers as it is always blended into the previous window.

Move the colour key initialisation into the probe, and run it for only
nr_win-1 windows.

Signed-off-by: Ben Dooks <[email protected]>
Signed-off-by: Pawel Osciak <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: KyungMin Park <kyungmin.park.samsung.com>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: fix distortedness situation for the mode more then 24bpp

It has been working fine at 16bpp but in case of pixel format more then
24bpp it would occur distortedness situation on that mode.  so this patch
set the word swap control bit of WINCONx to 1 as default value.  but it
should be set to 0 in case that each ENLOCAL bit of WINCON0 ~ 2 registers
is enabled.  this issue would be solved with local path feature soon.

Signed-off-by: InKi Dae <[email protected]>
Reviewed-by: KyungMin Park <kyungmin.park.samsung.com>
Signed-off-by: Ben Dooks <[email protected]>
Signed-off-by: Pawel Osciak <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: add default window feature.

s5pv210 has five window layers (window0 ~ 4), among them, window0 ~ 2
could be used for local path with fimc(capture device) and fimd writeback
feature so this patch makes default window layer for UI to be set at
machine code.

Signed-off-by: InKi Dae <[email protected]>
Reviewed-by: KyungMin Park <kyungmin.park.samsung.com>
Signed-off-by: Ben Dooks <[email protected]>
Signed-off-by: Pawel Osciak <[email protected]>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

s3c-fb: change to depending on CONFIG_S3C_FB_DEV

As suggested by Marek Szyprowski, we should make the driver depend on the
configuration currently being used to build the platform device into the
kernel.

Signed-off-by: Ben Dooks <[email protected]>
Signed-off-by: Pawel Osciak <[email protected]>
Cc: InKi Dae <[email protected]>
Cc: KyungMin Park <kyungmin.park.samsung.com>
Cc: Marek Szyprowski <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/video/fbmem.c: simplify strlen()==0 check in fb_get_options()

Replaced !strlen(str) check with !str[0].  Removed the variable which was
used solely to store strlen result.

Signed-off-by: Denys Vlasenko <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

fbcon: uninline four foo_update_attr() functions

This patch uninlines four similar functions, foo_update_attr(), in four
fbcon-related files.

These functions contain loops, two of theam have _nested_ loops, and they
have more than one callsite each.  I think they should not be inlined.

Signed-off-by: Denys Vlasenko <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

fbcon: uninline large static function get_color()

This function's body is good two screenfuls and it has six callsites.  No
apparent reason why it is marked inline.

Signed-off-by: Denys Vlasenko <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

fbdev: efifb: section fixes

Remove 43 section mismatches by moving the two structures efifb_defined
and efifb_fix from .init.data to .devinit.data.

Also the two structure arrays dmi_system_table[] and dmi_list[] have been
moved from .data to .init.rodata and .init.data, which saves, if built-in,
some space.

On x86_64 'size -A' showed that these sections changed size:

efifb.o:
section                     size-old        size-new
.data                       1200            688
.init.data                  7840            512
.init.rodata                0               7568
.devinit.data               0               256

Total                       11927           11911

Signed-off-by: Henrik Kretzschmar <[email protected]>
Cc: Peter Jones <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

gpio: sx150x: add Semtech I2C sx150x gpio expander driver

Add support for Semtech SX150-series I2C GPIO expanders.  Compatible
models include:

8 bits:  sx1508q
16 bits: sx1509q

Signed-off-by: Gregory Bean <[email protected]>
Cc: David Brownell <[email protected]>
Cc: Jean Delvare <[email protected]>
Cc: Trilok Soni <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

gpiolib: allow nested threaded irqs for poll(2)

The pca953x driver requires the use of threaded irqs as its irq
demultiplexer can sleep.  Our irq handler can be called from any context,
so use request_any_context_irq to allow threaded irqs as well.

Signed-off-by: Daniel Gl?ckner <[email protected]>
Reported-by: Ian Jeffray <[email protected]>
Cc: David Brownell <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

gpiolib: get rid of struct poll_desc and worklet

As sysfs_notify_dirent has been made irq safe, there is no reason to not
call it directly from irq.  With the work_struct removed, the remaining
element in poll_desc is a sysfs_dirent pointer which may not be NULL.  We
can therefore store it directly in the idr and pass it as context to the
irq handler.

Most part of the patch deals with renaming defines and variables to
reflect their new use without functional change.

I also took the opportunity to initialize the idr statically.

Signed-off-by: Daniel Gl?ckner <[email protected]>
Cc: David Brownell <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

gpiolib: decouple might_sleep_if() from DEBUG

Be more consistent about runtime programming interface abuse warnings,
which can reduce some confusion and trigger bugfixes.  Based on an
observation and patch from Jani Nikula.

Also update doc to highlight some sleeping-call issues and to match some
recent changes.

Signed-off-by: David Brownell <[email protected]>
Cc: Jani Nikula <[email protected]>
Cc: "Ryan Mallon" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

pcf857x: support working w/o platform data

Provide sane defaults for pcf857x, so the driver can be used w/o providing
platform data (and thus can be simply bound via OF tree).

Signed-off-by: Dmitry Eremin-Solenikov <[email protected]>
Cc: Jean Delvare <[email protected]>
Cc: David Brownell <[email protected]>
Cc: Grant Likely <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

gpio: max730x: make pullups configurable via platformdata

The gpios on the max730x chips have support for internal pullups while in
input mode.

This patch adds support for configuring these pullups via platform data.
A new member ("input_pullup_active") to the platform data struct is
introduced.  A set bit in this variable activates the pullups while the
respective port is in input mode.  This is a compatible enhancement since
unset bits lead to disables pullups which was the default in the original
driver.

_Note_: the 4 lowest bits in "input_pullup_active" are unused because the
first 4 ports of the controller are not used, too.

Signed-off-by: Marc Kleine-Budde <[email protected]>
Reviewed-by: Wolfram Sang <[email protected]>
Cc: David Brownell <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc: rp5c01: add NVRAM support

The Ricoh RP5C01 RTC contains 26 x 4 bits of NVRAM.  Provide access to it
via a sysfs "nvram" attribute file.

Signed-off-by: Geert Uytterhoeven <[email protected]>
Cc: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/rtc: remove unneeded ifdef CONFIG_PM

Because CONFIG_PM is a precondition to CONFIG_ACPI, the ifdef CONFIG_PM
within ifdef CONFIG_ACPI is redundant.

Signed-off-by: Christian Dietrich <[email protected]>
Acked-by: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc-isl12022: properly handle military hour format

Mask out PM flag when reading the hour, always set MIL bit when
writing the hour.

Signed-off-by: Roman Fietze <[email protected]>
Acked-by: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc: add support for DS3232 RTC

Add a driver for the DS3232 RTC chip via the I2C bus.  Alarms are not
supported in this version of the driver.

[[email protected]: fix Kconfig help text]
Signed-off-by: Mingkai Hu <[email protected]>
Signed-off-by: Jingchang Lu <[email protected]>
Signed-off-by: Srikanth Srinivasan <[email protected]>
Signed-off-by: Roy Zang <[email protected]>
Acked-by: Wan ZongShun <[email protected]>
Cc: Kumar Gala <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc/nuc900: fix build warning

Remove unused local variable.

Signed-off-by: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc/nuc900: modify enable/disable IRQs and driver data setting location

This patch does two modifications:

(1) Adjust enable/disable IRQs location,enable it after rtc
    registration and disable it prior to unregistration.

(2) Put 'platform_set_drvdata(pdev, nuc900_rtc)' in front of rtc
    registration still be safety, though there is no need to do this, when
    I move enable irq after rtc registration, I think still put
    'platform_set_drvdata' before rtc registration that would be a good
    habit.

Signed-off-by: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc/nuc900: make returning time checking function valid

Make returning time checking function valid.  In spite of using the
'rtc_valid_tm', nevertheless, the read time function omits its returning
value, that means the 'rtc_valid_tm' is useless here.

Signed-off-by: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc/m48t86: use rtc_valid_tm() to check returned tm

Use rtc_valid_tm() to check the returned struct rtc_time *tm, to avoid
returning a wrong tm value.

Signed-off-by: Wan ZongShun <[email protected]>
Cc: Paul Gortmaker <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc/m48t59: use rtc_valid_tm() to check returned tm

Use rtc_valid_tm to check the returned struct rtc_time *tm, to avoid
returning a wrong tm value.

Signed-off-by: Wan ZongShun <[email protected]>
Cc: Paul Gortmaker <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc/max6900: use rtc_valid_tm() to check returning tm

Use rtc_valid_tm() to check returning tm for max6900, it can avoid
returning wrong tm value.

Signed-off-by: Wan ZongShun <[email protected]>
Acked-by: Dale Farnsworth <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc/m41t80: use rtc_valid_tm() to check returned tm

Use rtc_valid_tm() to check returned struct rtc_time *tm - it can avoid
returning wrong tm value.

Signed-off-by: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc: add Intersil ISL12022 RTC driver

- derived from rtc-pcf8563

- no SRAM driver

Signed-off-by: Roman Fietze <[email protected]>
Cc: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc: driver for the DryIce block found in i.MX25 chips

This driver is based on code from Freescale which accompanies their i.MX25
PDK board, with some cleanup.

Signed-off-by: Baruch Siach <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Cc: Sascha Hauer <[email protected]>
Acked-by: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/rtc/rtc-pl031.c: remove unused #include <linux/version.h>

Signed-off-by: Huang Weiyi <[email protected]>
Acked-by: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc/pxa: remove unnecessary private ops->ioctl()

We shouldn't implement private ops->ioctl() unless absolutely necessary.
pxa series RTC driver's ioctl() is unnecessary, since RTC subsystem has
implement the ioctl() very well,so we can only use the API of
'.alarm_irq_enable' and '.update_irq_enable' to do enable irq action.

Signed-off-by: Wan ZongShun <[email protected]>
Acked-by: Robert Jarzmik <[email protected]>
Acked-by: Eric Miao <[email protected]>
Cc: Paul Gortmaker <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc: fixes and new functionality for fm3130

- add sanity check for alarm data in fm3130_probe

- fix fm3130_set_alarm.

  According to the datasheet, setting match bit '0' indicates that the
  corresponding alarm field will be used in the match process

- add operation alarm_irq_enable operation which is responsible for
  handling RTC_AIE_ON, RTC_AIE_OFF ioctls

- remove clearing of AF bit after reading rtc/alarm control register:
  according to datasheet this bit is cleared anyway when rtc/alarm control
  register is read

[[email protected]: make fm3130_alarm_irq_enable() static, fix comment layout]
Signed-off-by: Sergey Matyukevich <[email protected]>
Acked-by: Wan ZongShun <[email protected]>
Acked-by: Sergey Lapin <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc/rtc-mxc: remove six unused fields

Remove six unused fields from `struct rtc_plat_data'.

Signed-off-by: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Cc: Paul Gortmaker <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc/m48t59: kfree(NULL) is OK

Signed-off-by: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Cc: Paul Gortmaker <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc/pxa: remove unused field

pxa_rtc.rtc_alarm is unused.

Signed-off-by: Wan ZongShun <[email protected]>
Cc: Eric Miao <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/rtc/rtc-pcf8563.c: remove unused struct

Signed-off-by: Graham Gower <[email protected]>
Acked-by: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

rtc/nuc900: fix checking of args during time-setting

When a user application wants to set the rtc time, the RTC subsystem takes
advantage of 'rtc_valid_tm(tm)' to check 'rtc_time *tm' value validity, it
make sure the 'tm->tm_year' is larger than 70,so if '70< tm_year < 100',
the '(settm->tm_year - 100)' will be negative.  ' Setting the negative
value to hardware register will be invalid, so I add the 'if' condition to
make sure set a valid value to register.

Signed-off-by: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

nuc900/rtc: change the waiting for device ready implement

- add an mdelay(1) to the polling loop to cause less frequent access to
  the hardware register.

- change the return value from ENODEV to EPERM if the loop timed out.  I
  think the 'Operation not permitted' description is more suitable for the
  meaning of 'check_rtc_access_enable()' function, it just be used to
  judge rtc access operation is permitted or not.

Signed-off-by: Wan ZongShun <[email protected]>
Cc: Alessandro Zummo <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

autofs4: remove unneeded null check in try_to_fill_dentry()

After 97e7449a7ad: "autofs4: fix indirect mount pending expire race" we no
longer assumed that "ino" can be null.  The other null checks got removed
but this was one was missed.

Signed-off-by: Dan Carpenter <[email protected]>
Cc: Ian Kent <[email protected]>
Cc: Al Viro <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

i2o: check return code from put_user()

Check return value of put_user() and return -EFAULT if it failed.
Original comment "We did a get user...so assuming mem is ok...is this
bad?" is incorrect because memory can be read only.

Signed-off-by: Kulikov Vasiliy <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

i2o: fix overflow of copy_to_user()

If (len > reslen) we must not call copy_to_user() since kernel buffer is
smaller than we want to copy.  Similar code in this file is correct, so
this bug was a typo.

Signed-off-by: Kulikov Vasiliy <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/message/i2o/exec-osm.c: add missing mutex_unlock

Add a mutex_unlock missing on the error path.

The semantic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@
expression E1;
@@

* mutex_lock(E1,...);
  <+... when != E1
  if (...) {
    ... when != E1
*   return ...;
  }
  ...+>
* mutex_unlock(E1,...);
// </smpl>

Signed-off-by: Julia Lawall <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

mmc: add ricoh e822 pci id

Signed-off-by: Pablo Castillo <[email protected]>
Signed-off-by: Stanislaw Gruszka <[email protected]>
Tested-by: Gregg Lebovitz <[email protected]>
Cc: Maxim Levitsky <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

dts: add ESDHC weird voltage bits workaround

P4080 ESDHC controller does not support 1.8V and 3.0V voltage.  but the
host controller capabilities register wrongly set the bits.  This patch
adds the workaround to correct the weird voltage setting bits.  Only 3.3V
voltage is supported for P4080 ESDHC controller.

Signed-off-by: Roy Zang <[email protected]>
Cc: Jerry Huang <[email protected]>
Cc: Benjamin Herrenschmidt <[email protected]>
Cc: Kumar Gala <[email protected]>
Cc: Grant Likely <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

dts: add sdhci,auto-cmd12 field for p4080 device tree

Signed-off-by: Roy Zang <[email protected]>
Cc: Jerry Huang <[email protected]>
Cc: Benjamin Herrenschmidt <[email protected]>
Cc: Kumar Gala <[email protected]>
Cc: Grant Likely <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sdhci: add auto CMD12 support for eSDHC driver

Add auto CMD12 command support for eSDHC driver.  This is needed by P4080
and P1022 for block read/write.  Manual asynchronous CMD12 abort operation
causes protocol violations on these silicons.

Signed-off-by: Jerry Huang <[email protected]>
Signed-off-by: Roy Zang <[email protected]>
Cc: Benjamin Herrenschmidt <[email protected]>
Cc: Kumar Gala <[email protected]>
Cc: Grant Likely <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sdhci: add regulator support

This patch adds support for regulator API to sdhci core driver.
Regulators can be used to disable power in suspended state to reduce
dissipated energy.

Signed-off-by: Marek Szyprowski <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Acked-by: Mark Brown <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sdhci-s3c: add support for new card detection methods

On some Samsung SoCs not all SDHCI controllers have card detect (CD) line.
 For some embedded designs it is not even needed, because ususally the
device (like SDIO flash memory or wifi controller) is permanently wired to
the controller.  There are also systems which have a card detect line
connected to some of the external interrupt lines or the presence of the
card depends on some other actions (like enabling a power regulator).

This patch adds support for all these cases.  The following card detection
methods are possible:

1. internal sdhci host card detect line
2. external event
3. external gpio interrupt
4. no card detect line, controller will poll for the card
5. no card detect line, card is permanently wired to the controller
(once detected host won't poll it any more)

By default, all existing code would use method #1, what is compatible with
the previous version of the driver.

In case of external event, two callbacks must be provided in platdata:
ext_cd_init and ext_cd_cleanup.  Both of them get a callback to a function
that notifies the s3c-sdhci host contoller as their argument.  That
callback function should be called from the even dispatcher to let host
notice the card insertion/removal.

In case of external gpio interrupt, a gpio pin number must be provided in
platdata (ext_cd_gpio parameter), as well as the information about the
polarity of that gpio pin (ext_cd_gpio_invert).  By default
(ext_cd_gpio_invert == 0) gpio value 0 means 'card has been removed', but
this can be changed to 'card has been removed' when ext_cd_gpio_invert ==
1.

This patch adds all required changes to sdhci-s3c driver.

Signed-off-by: Marek Szyprowski <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sdhci-s3c: enable SDHCI_QUIRK_NO_HISPD_BIT quirk

This patch enables SDHCI_QUIRK_NO_HISPD_BIT on Samsung SDHCI driver.  This
solves detection problems with some external SD cards.  This change has
been tested on S5PC100 and S5PC110.  It has no inpact on driver speed.

Signed-off-by: Kyungmin Park <[email protected]>
Signed-off-by: Marek Szyprowski <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sdhci-s3c: add support for the non standard minimal clock value

S3C SDHCI host controller can change the source for generating mmc clock.
By default host bus clock is used, what causes some problems on machines
with 133MHz bus, because the SDHCI divider cannot be as high get proper
clock value for identification mode.  This is not a problem for the
controller, because it can generate lower frequencies from other clock
sources.  This patch changes sdhci driver to use get_min_clock() call if
it has been provided.

This fixes the flood of the following warnings on Samsung S5PV210 SoCs:
mmc0: Minimum clock frequency too high for identification mode

Signed-off-by: Marek Szyprowski <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sdhci-s3c: add SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK quirk

On Samsung's SDMMC hosts the timeout clock is derivied from the SD Clock
which is set dynamically.  So checked SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK
quirk and removed 'sdhci_s3c_get_timeout_clk' callback which doesn't need
any more.

Signed-off-by: Hyuk Lee <[email protected]>
Signed-off-by: Kukjin Kim <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

omap hsmmc: fix processing of all dma interrupts as block completion

If other informative interrupts are enabled for the DMA channel used by
hsmmc, those are incorrectly treated as block completion.  This patch lets
only the block completion interrupt to be processed.

Signed-off-by: Venkatraman S <[email protected]>
Acked-by: Madhusudhan Chikkature <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

drivers/mmc/host/msm-sdcc: remove dead config options

CONFIG_MMC_MSM7X00A_RESUME_IN_WQ and CONFIG_MMC_EMBEDDED_SDIO don't exist
in Kconfig and is never defined anywhere else, therefore removing all
references for it from the source code.

Signed-off-by: Christian Dietrich <[email protected]>
Acked-by: David Brown <[email protected]>
Cc: Daniel Walker <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

omap: pandora: pass wl1251 information to SDIO core

Pandora has TI WL1251 attached on MMC3, which is non-standard SDIO chip.
Make use MMC_QUIRK_NONSTD_SDIO to tell SDIO core about it.

Signed-off-by: Grazvydas Ignotas <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Tony Lindgren <[email protected]>
Cc: Bob Copeland <[email protected]>
Cc: Kalle Valo <[email protected]>
Cc: Madhusudhan Chikkature <[email protected]>
Cc: Kishore Kadiyala <[email protected]>
Cc: Russell King <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

omap_hsmmc: add init_card pass-through callback

This will allow us to set up special cards in machine drivers just after
they are detected by MMC core.

Signed-off-by: Grazvydas Ignotas <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Tony Lindgren <[email protected]>
Cc: Bob Copeland <[email protected]>
Cc: Kalle Valo <[email protected]>
Cc: Madhusudhan Chikkature <[email protected]>
Cc: Kishore Kadiyala <[email protected]>
Cc: Russell King <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sdio: allow non-standard SDIO cards

There are some chips (like TI WL12xx series) that can be interfaced over
SDIO but don't support the SDIO specification, meaning that they are
missing CIA (Common I/O Area) with all it's registers.  Current Linux SDIO
implementation relies on those registers to identify and configure the
card, so non-standard cards can not function and cause lots of warnings
from the core when it reads invalid data from non-existent registers.

After this patch, init_card() host callback can now set new quirk
MMC_QUIRK_NONSTD_SDIO, which means that SDIO core should not try to access
any standard SDIO registers and rely on init_card() to fill all SDIO
structures instead.  As those cards are usually embedded chips, all the
required information can be obtained from machine board files by the host
driver when it's called through init_card() callback.

Signed-off-by: Grazvydas Ignotas <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Tony Lindgren <[email protected]>
Cc: Bob Copeland <[email protected]>
Cc: Kalle Valo <[email protected]>
Cc: Madhusudhan Chikkature <[email protected]>
Cc: Kishore Kadiyala <[email protected]>
Cc: Russell King <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sdhci-pltfm: add support for CNS3xxx SoC devices

There's nothing special, just SoC-specific ops and quirks.

Signed-off-by: Anton Vorontsov <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Richard R?jfors <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sdhci-pltfm: reorganize Makefile entries to support SoC devices

Due to build system limitations, intermediate and final objects can't have
the same names.  And as we're going to start building SoC-specific
objects, let's rename the module to sdhci-platform, into which we'll link
sdhci-pltfm and SoC-specifc objects.

There should be no issue in renaming as the driver uses modalias
mechanism.

This is exactly the same approach as in sdhci-of driver.

Signed-off-by: Anton Vorontsov <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Richard R?jfors <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sdhci-pltfm: switch to module device table matching

Sometimes want to place SoC-specific parts alongside with the generic
driver, and to do so, we have to switch the driver over to the module
device table matching.

Note that drivers/mmc/host/sdhci-pltfm.h is so far empty, but it'll hold
SoC-specific driver data handlers soon.

Signed-off-by: Anton Vorontsov <[email protected]>
Cc: Ben Dooks <[email protected]>
Cc: Richard R?jfors <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

mmc: omap: fix for bus width which improves SD card's peformance.

This patch improves low speeds for SD cards.

OMAP-MMC controller's can support maximum bus width of '8'.  when bus
width is mentioned as "8" in controller data,the SD stack will check
whether bus width is "4" and if not it will set bus width to "1" and there
by degrading performance.  This patch fixes the issue and improves the
performance of SD cards.

Signed-off-by: Kishore Kadiyala <[email protected]>
Signed-off-by: Venkatraman S <[email protected]>
Signed-off-by: Nishanth Menon <[email protected]>
Acked-by: Madhusudhan Chikkature <[email protected]>
Tested-by: Jarkko Nikula <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Matt Fleming <[email protected]>
Cc: Tony Lindgren <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

mmc: only set blockaddressed for > 2GiB cards

A non-zero value of SEC_COUNT does not indicate that the card is sector
addressed.  According to the MMC specification, cards with a density
greater than 2GiB are sector addressed.

Acked-by: Linus Walleij <[email protected]>
Signed-off-by: Hanumath Prasad <[email protected]>
Signed-off-by: Rabin Vincent <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sdhci: don't assign mmc->caps at SDHCI directly

Some host controllers can set mmc->caps before sdhci_add_host().

Signed-off-by: Kyungmin Park <[email protected]>
Cc: Grant Likely <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sdhci: 8-bit data transfer width support

Some host controllers such as s5pc110 support the WIDE8 feature.

Signed-off-by: Kyungmin Park <[email protected]>
Cc: Grant Likely <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

sdhci: remove useless set_clock() check

When using QUIRK_NONSTANDARD_CLOCK, it checks the set_clock() function
which is not used actually.  So delete it.

Signed-off-by: Kyungmin Park <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

mmc: make sdhci work with ricoh mmc controller

The current way of disabling it is not well tested by vendor and has all
kinds of bugs that show up on resume from ram/disk.  A very good example
is a dead SDHCI controller.

Old way of disabling is still supported by continuing to use
CONFIG_MMC_RICOH_MMC.

Based on 'http://list.drzeus.cx/pipermail/sdhci-devel/2007-December/002085.html'
Therefore most of the credit for this goes to Andrew de Quincey

Signed-off-by: Maxim Levitsky <[email protected]>
Cc: Andrew de Quincey <[email protected]>
Acked-by: Philip Langdale <[email protected]>
Cc: "Rafael J. Wysocki" <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

mmc: fix all hangs related to mmc/sd card insert/removal during suspend/resume

If you don't use CONFIG_MMC_UNSAFE_RESUME, as soon as you attempt to
suspend, the card will be removed, therefore this patch doesn't change the
behavior of this option.

However the removal will be done by pm notifier, which runs while
userspace is still not frozen and thus can freely use del_gendisk, without
the risk of deadlock which would happen otherwise.

Card detect workqueue is now disabled while userspace is frozen, Therefore
if you do use CONFIG_MMC_UNSAFE_RESUME, and remove the card during
suspend, the removal will be detected as soon as userspace is unfrozen,
again at the moment it is safe to call del_gendisk.

Tested with and without CONFIG_MMC_UNSAFE_RESUME with suspend and hibernate.

[[email protected]: clean up function prototype]
[[email protected]: fix CONFIG_PM-n linkage, small cleanups]
[[email protected]: coding-style fixes]
Signed-off-by: Maxim Levitsky <[email protected]>
Cc: David Brownell <[email protected]>
Cc: Alan Stern <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>