When current request is running on the bus and if next request fetched
by mmcqd is NULL, mmc context (mmcqd thread) gets blocked until the
current request completes. This means that if new request comes in while
the mmcqd thread is blocked, this new request can not be prepared in
parallel to current ongoing request. This may result in delaying the new
request execution and increase it's latency.

This change allows to wake up the MMC thread on new request arrival.
Now once the MMC thread is woken up, a new request can be fetched and
prepared in parallel to the current running request which means this new
request can be started immediately after the current running request
completes.

With this change read throughput is improved by 16%.
Signed-off-by: NKonstantin Dorfman <kdorfman@codeaurora.org>
Reviewed-by: NSeungwon Jeon <tgih.jun@samsung.com>
Signed-off-by: NChris Ball <cjb@laptop.org>

Unlike normal r/w request, special requests(discard, flush)
is finished with a one-time issue_fn. Request change to
mqrq_prev makes unnecessary call.
Signed-off-by: NSeungwon Jeon <tgih.jun@samsung.com>
Reviewed-by: NKonstantin Dorfman <kdorfman@codeaurora.org>
Signed-off-by: NChris Ball <cjb@laptop.org>

The classical way to process IRQs is read out the status, ack all triggered
IRQs, possibly mask them, then process them. Follow this simple procesure
instead of the current complex custom algorithm.
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

Make error reporting in the driver more verbose. This patch is based on
an earlier work by Teppei Kamijou, but we try to not add any new error
messages to the log in the normal case to avoid confusing the user, and
also add a few more dev_dbg() calls.
Signed-off-by: NTeppei Kamijou <teppei.kamijou.yb@renesas.com>
Signed-off-by: NShinya Kuribayashi <shinya.kuribayashi.px@renesas.com>
[g.liakhovetski@gmx.de: avoid producing new errors in normal case]
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

The INT_BUFWEN IRQ often arrives with other bits set too. If they are not
cleared, an additional IRQ can be triggered, sometimes also after the MMC
request has already been completed. This leads to block I/O errors. Earlier
Teppei Kamijou also observed these additional interrupts and proposed to
explicitly wait for them. This patch chooses an alternative approach of
clearing all active bits immediately, when processing the main interrupt.
Reported-by: NTeppei Kamijou <teppei.kamijou.yb@renesas.com>
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

DMA completion can be signalled from the DMA callback and from the error
handler. If both are called, the completion struct can enter an
inconsistent state. To prevent this move completion initialisation
immediately before activating DMA.
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

If a command execution has produced an error, it has to be reset as a part
of the error handling.
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

Oopses have been observed on SMP in the sh-mmcif IRQ thread, when the two
IRQ threads run simultaneously on two CPUs. Also take care to guard the
timeout work and the DMA completion callback from possible NULL-pointer
dereferences and races.
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

Use "return condition" instead of "if (condition) return true; return false"
in functions, returning bool.
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

If a DMA transaction fails, terminate all outstanding DMA transfers and
unmap buffers.
Signed-off-by: NTeppei Kamijou <teppei.kamijou.yb@renesas.com>
Signed-off-by: NShinya Kuribayashi <shinya.kuribayashi.px@renesas.com>
[g.liakhovetski@gmx.de: forward-port, add dma_unmap_sg() in error cases]
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

Read block and write block operations are currently missing completion
timeouts. Add missing timeouts and consolidate them at one location.
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

SH/R-Mobile MMCIF host controller can wait while the card signals busy.
Set MMC_CAP_WAIT_WHILE_BUSY to inform an upper layer (core/mmc_ops.c)
not to insert unnecessary mmc_delay().
Signed-off-by: NTeppei Kamijou <teppei.kamijou.yb@renesas.com>
Signed-off-by: NShinya Kuribayashi <shinya.kuribayashi.px@renesas.com>
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

Timeout period should be properly normalized using msecs_to_jiffies().
Signed-off-by: NTeppei Kamijou <teppei.kamijou.yb@renesas.com>
Signed-off-by: NShinya Kuribayashi <shinya.kuribayashi.px@renesas.com>
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

Some MMCIF implementations support the Dual Data Rate. With this patch,
platforms can set the MMC_CAP_UHS_DDR50 capability flag in MMCIF platform
data. This will let the MMC core to actually use the DDR mode.
Signed-off-by: NTeppei Kamijou <teppei.kamijou.yb@renesas.com>
Signed-off-by: NShinya Kuribayashi <shinya.kuribayashi.px@renesas.com>
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

On newer SoCs like R-Mobile U2, MMCIF interrupts are bundled.
Signed-off-by: NShinya Kuribayashi <shinya.kuribayashi.px@renesas.com>
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

With this post-v2.6.35 change applied:

commit a0a1a5fd
Author: Tejun Heo <tj@kernel.org>
Date:   Tue Jun 29 10:07:12 2010 +0200
    workqueue: reimplement workqueue freeze using max_active

freeze_workqueues_begin() was introduced and workqueue now gets frozen
before device drivers suspend operations.

We have to ensure that run-time PM suspend operation completes before
system-wide suspend is started.
Signed-off-by: NTeppei Kamijou <teppei.kamijou.yb@renesas.com>
Signed-off-by: NShinya Kuribayashi <shinya.kuribayashi.px@renesas.com>
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

mmc_rescan() sends CMD52 (SD_IO_RW_DIRECT) to reset SDIO card during
card detection. CMD52 should be ignored by SD/eMMC cards, but we can
also abort it in the driver immediately, since MMCIF doesn't support
SDIO cards anyway.
Signed-off-by: NTeppei Kamijou <teppei.kamijou.yb@renesas.com>
Signed-off-by: NShinya Kuribayashi <shinya.kuribayashi.px@renesas.com>
Signed-off-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

Use slot-gpio helpers to save some code in the driver.
Signed-off-by: NShawn Guo <shawn.guo@linaro.org>
Signed-off-by: NChris Ball <cjb@laptop.org>

Since slot-gpio uses devm_* managed functions in mmc_gpio_request_cd()
now, we can remove those mmc_gpio_free_cd() call from host drivers'
.probe() error path and .remove().
Signed-off-by: NShawn Guo <shawn.guo@linaro.org>
Acked-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

Use devm_* managed functions, so that slot-gpio users do not have to
call mmc_gpio_free_ro/cd to free up resources requested in
mmc_gpio_request_ro/cd.
Signed-off-by: NShawn Guo <shawn.guo@linaro.org>
Acked-by: NGuennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: NChris Ball <cjb@laptop.org>

Call mmc_gpio_get_cd() to query card presence from cd-gpio before
asking SDHCI.  The rationale behind this change is that flag
SDHCI_QUIRK_BROKEN_CARD_DETECTION is designed for SDHCI controller to
tell that SDHCI_PRESENT_STATE is broken, and it should be used for this
case only.  So when cd-gpio is being used, the controller should set
the flag to tell that SDHCI_PRESENT_STATE is not available.

However, the existing code will skip checking cd-gpio as long as flag
SDHCI_QUIRK_BROKEN_CARD_DETECTION is set.  Change the querying order
between cd-gpio and SDHCI to support the rationale above.
Signed-off-by: NShawn Guo <shawn.guo@linaro.org>
Signed-off-by: NChris Ball <cjb@laptop.org>

When SDIO3.0 card is detected, incorrect bus speed mode
is printed as part of card detection print in kernel logs.

This change fixes it so that user won't be confused by
looking at incorrect card detection message in logs.
Signed-off-by: NSubhash Jadavani <subhashj@codeaurora.org>
Tested-by: NJackey Shen <Jackey.Shen@amd.com>
Acked-by: NUlf Hansson <ulf.hansson@linaro.org>
Signed-off-by: NChris Ball <cjb@laptop.org>

According to UHS-I initialization sequence for SDIO 3.0 cards,
the host must set bit[24] (S18R) of OCR register during OCR
handshake to know whether the SDIO card is capable of doing
1.8V I/O.
Signed-off-by: NSujit Reddy Thumma <sthumma@codeaurora.org>
Signed-off-by: NSubhash Jadavani <subhashj@codeaurora.org>
Reviewed-by: NJohan Rudholm <johan.rudholm@stericsson.com>
Reviewed-by: NUlf Hansson <ulf.hansson@linaro.org>
Signed-off-by: NChris Ball <cjb@laptop.org>

If SDIO keep power flag (MMC_PM_KEEP_POWER) is not set, card would
be reinitialized during resume but as we are not resetting
(CMD52 reset) the SDIO card during this reinitialization, card may
fail to respond back to subsequent commands (CMD5 etc...).

This change resets the card before the reinitialization of card
during resume.
Signed-off-by: NSubhash Jadavani <subhashj@codeaurora.org>
Acked-by: NUlf Hansson <ulf.hansson@linaro.org>
Signed-off-by: NChris Ball <cjb@laptop.org>

Pull btrfs fixes from Chris Mason:
 "It turns out that we had two crc bugs when running fsx-linux in a
  loop.  Many thanks to Josef, Miao Xie, and Dave Sterba for nailing it
  all down.  Miao also has a new OOM fix in this v2 pull as well.

  Ilya fixed a regression Liu Bo found in the balance ioctls for pausing
  and resuming a running balance across drives.

  Josef's orphan truncate patch fixes an obscure corruption we'd see
  during xfstests.

  Arne's patches address problems with subvolume quotas.  If the user
  destroys quota groups incorrectly the FS will refuse to mount.

  The rest are smaller fixes and plugs for memory leaks."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (30 commits)
  Btrfs: fix repeated delalloc work allocation
  Btrfs: fix wrong max device number for single profile
  Btrfs: fix missed transaction->aborted check
  Btrfs: Add ACCESS_ONCE() to transaction->abort accesses
  Btrfs: put csums on the right ordered extent
  Btrfs: use right range to find checksum for compressed extents
  Btrfs: fix panic when recovering tree log
  Btrfs: do not allow logged extents to be merged or removed
  Btrfs: fix a regression in balance usage filter
  Btrfs: prevent qgroup destroy when there are still relations
  Btrfs: ignore orphan qgroup relations
  Btrfs: reorder locks and sanity checks in btrfs_ioctl_defrag
  Btrfs: fix unlock order in btrfs_ioctl_rm_dev
  Btrfs: fix unlock order in btrfs_ioctl_resize
  Btrfs: fix "mutually exclusive op is running" error code
  Btrfs: bring back balance pause/resume logic
  btrfs: update timestamps on truncate()
  btrfs: fix btrfs_cont_expand() freeing IS_ERR em
  Btrfs: fix a bug when llseek for delalloc bytes behind prealloc extents
  Btrfs: fix off-by-one in lseek
  ...

Pull cifs fixes from Steve French:
 "Two small cifs fixes"

* 'for-next' of git://git.samba.org/sfrench/cifs-2.6:
  fs/cifs/cifs_dfs_ref.c: fix potential memory leakage
  cifs: fix srcip_matches() for ipv6

Pull kvm fixlet from Marcelo Tosatti.

* git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: PPC: Emulate dcbf

Pull ARM fixes from Russell King:
 "A number of fixes:

  Patrik found a problem with preempt counting in the VFP assembly
  functions which can cause the preempt count to be upset.

  Nicolas fixed a problem with the parsing of the DT when it straddles a
  1MB boundary.

  Subhash Jadavani reported a problem with sparsemem and our highmem
  support for cache maintanence for DMA areas, and TI found a bug in
  their strongly ordered memory mapping type.

  Also, three fixes by way of Will Deacon's tree from Dave Martin for
  instruction compatibility and Marc Zyngier to fix hypervisor boot mode
  issues."

* 'fixes' of git://git.linaro.org/people/rmk/linux-arm:
  ARM: 7629/1: mm: Fix missing XN flag for for MT_MEMORY_SO
  ARM: DMA: Fix struct page iterator in dma_cache_maint() to work with sparsemem
  ARM: 7628/1: head.S: map one extra section for the ATAG/DTB area
  ARM: 7627/1: Predicate preempt logic on PREEMP_COUNT not PREEMPT alone
  ARM: virt: simplify __hyp_stub_install epilog
  ARM: virt: boot secondary CPUs through the right entry point
  ARM: virt: Avoid bx instruction for compatibility with <=ARMv4

Pull ARM SoC fixes from Olof Johansson:
 "Here's a long-pending fixes pull request for arm-soc (I didn't send
  one in the -rc4 cycle).

  The larger deltas are from:

   - A fixup of error paths in the mvsdio driver

   - Header file move for a driver that hadn't been properly converted
     to multiplatform on i.MX, which was causing build failures when
     included

   - Device tree updates for at91 dealing mostly with their new pinctrl
     setup merged in 3.8 and mistakes in those initial configs

  The rest are the normal mix of small fixes all over the place; sunxi,
  omap, imx, mvebu, etc, etc."

* tag 'fixes-for-linus2' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (40 commits)
  mfd: vexpress-sysreg: Don't skip initialization on probe
  ARM: vexpress: Enable A7 cores in V2P-CA15_A7's Device Tree
  ARM: vexpress: extend the MPIDR range used for pen release check
  ARM: at91/dts: correct comment in at91sam9x5.dtsi for mii
  ARM: at91/at91_dt_defconfig: add at91sam9n12 SoC to DT defconfig
  ARM: at91/at91_dt_defconfig: remove memory specification to cmdline
  ARM: at91/dts: add macb mii pinctrl config for kizbox
  ARM: at91: rm9200: remake the BGA as default version
  ARM: at91: fix gpios on i2c-gpio for RM9200 DT
  ARM: at91/at91sam9x5 DTS: add SCK USART pins
  ARM: at91/at91sam9x5 DTS: correct wrong PIO BANK values on u(s)arts
  ARM: at91/at91-pinctrl documentation: fix typo and add some details
  ARM: kirkwood: fix missing #interrupt-cells property
  mmc: mvsdio: use devm_ API to simplify/correct error paths.
  clk: mvebu/clk-cpu.c: fix memory leakage
  ARM: OMAP2+: omap4-panda: add UART2 muxing for WiLink shared transport
  ARM: OMAP2+: DT node Timer iteration fix
  ARM: OMAP2+: Fix section warning for omap_init_ocp2scp()
  ARM: OMAP2+: fix build break for omapdrm
  ARM: OMAP2: Fix missing omap2xxx_clkt_vps_late_init function calls
  ...

Pull ACPI and power management fixes from Rafael Wysocki:

 - Two cpuidle initialization fixes from Konrad Rzeszutek Wilk.

 - cpufreq regression fixes for AMD processors from Borislav Petkov,
   Stefan Bader, and Matthew Garrett.

 - ACPI cpufreq fix from Thomas Schlichter.

 - cpufreq and devfreq fixes related to incorrect usage of operating
   performance points (OPP) framework and RCU from Nishanth Menon.

 - APEI workaround for incorrect BIOS information from Lans Zhang.

* tag 'pm+acpi-for-3.8-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpufreq: Add module aliases for acpi-cpufreq
  ACPI: Check MSR valid bit before using P-state frequencies
  PM / devfreq: exynos4_bus: honor RCU lock usage
  PM / devfreq: add locking documentation for recommended_opp
  cpufreq: cpufreq-cpu0: use RCU locks around usage of OPP
  cpufreq: OMAP: use RCU locks around usage of OPP
  ACPI, APEI: Fixup incorrect 64-bit access width firmware bug
  ACPI / processor: Get power info before updating the C-states
  powernow-k8: Add a kconfig dependency on acpi-cpufreq
  ACPI / cpuidle: Fix NULL pointer issues when cpuidle is disabled
  intel_idle: Don't register CPU notifier if we are not running.

Pull regmap fixes from Mark Brown:
 "One more oversight in the debugfs code was reported and fixed, plus a
  documentation fix."

* tag 'regmap-fix-3.8-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap:
  regmap: fix small typo in regmap_bulk_write comment
  regmap: debugfs: Fix seeking from the cache

Pull slave-dmaengine fixes from Vinod Koul:
 "A few fixes on slave dmanengine.  There are trivial fixes in imx-dma,
  tegra-dma & ioat driver"

* 'fixes' of git://git.infradead.org/users/vkoul/slave-dma:
  dma: tegra: implement flags parameters for cyclic transfer
  dmaengine: imx-dma: Disable use of hw_chain to fix sg_dma transfers.
  ioat: Fix DMA memory sync direction correct flag

Pill i2c fixes from Wolfram Sang:
 "Here are a few, typical driver fixes for the I2C subsystem"

* 'i2c-embedded/for-current' of git://git.pengutronix.de/git/wsa/linux:
  i2c-designware: add missing MODULE_LICENSE
  i2c: omap: fix draining irq handling
  i2c: omap: errata i462: fix incorrect ack for arbitration lost interrupt
  i2c: muxes: fix wrong use of sizeof(ptr)
  i2c: sirf: register i2c_client from dt child-nodes in probe entry
  i2c: mxs: Fix type of error code
  i2c: mxs: Fix misuse init_completion

btrfs_start_delalloc_inodes() locks the delalloc_inodes list, fetches the
first inode, unlocks the list, triggers btrfs_alloc_delalloc_work/
btrfs_queue_worker for this inode, and then it locks the list, checks the
head of the list again. But because we don't delete the first inode that it
deals with before, it will fetch the same inode. As a result, this function
allocates a huge amount of btrfs_delalloc_work structures, and OOM happens.

Fix this problem by splice this delalloc list.
Reported-by: NAlex Lyakas <alex.btrfs@zadarastorage.com>
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

The max device number of single profile is 1, not 0 (0 means 'as many as
possible'). Fix it.

Cc: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Reviewed-by: NLiu Bo <bo.li.liu@oracle.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

First, though the current transaction->aborted check can stop the commit early
and avoid unnecessary operations, it is too early, and some transaction handles
don't end, those handles may set transaction->aborted after the check.

Second, when we commit the transaction, we will wake up some worker threads to
flush the space cache and inode cache. Those threads also allocate some transaction
handles and may set transaction->aborted if some serious error happens.

So we need more check for ->aborted when committing the transaction. Fix it.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

We may access and update transaction->aborted on the different CPUs without
lock, so we need ACCESS_ONCE() wrapper to prevent the compiler from creating
unsolicited accesses and make sure we can get the right value.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

I noticed a WARN_ON going off when adding csums because we were going over
the amount of csum bytes that should have been allowed for an ordered
extent.  This is a leftover from when we used to hold the csums privately
for direct io, but now we use the normal ordered sum stuff so we need to
make sure and check if we've moved on to another extent so that the csums
are added to the right extent.  Without this we could end up with csums for
bytenrs that don't have extents to cover them yet.  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

For compressed extents, the range of checksum is covered by disk length,
and the disk length is different with ram length, so we need to use disk
length instead to get us the right checksum.
Signed-off-by: NLiu Bo <bo.li.liu@oracle.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>