This switches the mxcmmc driver to use the dmaengine API. Unlike
the old one this one is always present in the tree, even if no DMA
is implemented, hence we can remove all the #ifdefs in from the driver.
The driver automatically switches to PIO mode if no DMA support or no
suitable channel is available.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Chris Ball <cjb@laptop.org>

This patch adds support for 8-bit buswidth.
dw_mmc can use 8-bit buswidth and set to CTYPE_8BIT in card-type register.

Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Acked-by: Will Newton <will.newton@imgtec.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

If we need some quirks, maybe add quirks in future
But now, quirks value set to integer..later we should be confused..
So I think that need bit-shift control.

And If we need not any quirks, we didn't set anything..
(Need not DW_MCI_QUIRK_NONE)

Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
Acked-by: Will Newton <will.newton@imgtec.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

Use the new dmaengine helper functions, and drop the error check
on the returned cookier from the dmaengine - we recently
established that this is really not allowed to fail.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

This variable doesn't seem to be used for anything after the
other patches so just drop it.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

According to the DMA-API you shall unmap the sglists with the same
sglist length as passed into the mapping function, not the
returned value from the mapping function.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

As established for the MMCI, it is proper to map the DMA buffers
on the DMA engine which is the one actually performing the DMA.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

Use the new dmaengine helper functions, and drop the error check
on the returned cookier from the dmaengine - we recently
established that this is really not allowed to fail.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

The host_sglen is now actually used to keep track of whether DMA
is active or not, so rename and retype it to bool.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

According to the DMA-API you shall unmap the sglists with the same
sglist length as passed into the mapping function, not the
returned value from the mapping function.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

As established for the MMCI, it is proper to map the DMA buffers
on the DMA engine which is the one actually performing the DMA.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

Use the new dmaengine helpers to make the code more readable.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

Fixes the following:
- It is perfectly legal for the dma_map_sg() to return fewer
  entries than were passed in.
- Supply the returned numer of (possibly coalesced) entries to
  the device_pre_slave_sg() function.
- Use the proper original sg_len when unmapping the sglist
  in the error path.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

As established for the MMCI, it is proper to map the DMA buffers
on the DMA engine which is the one actually performing the DMA.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

Signed-off-by: Manoj Iyer <manoj.iyer@canonical.com>
Cc: <stable@kernel.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

So we know the implementation and prototypes agree with each other.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Reviewed-by: Chris Ball <cjb@laptop.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

The test file is created under debugfs, not sysfs. Also remove
the unnecessary default n.

Signed-off-by: Will Newton <will.newton@imgtec.com>
Reviewed-by: Chris Ball <cjb@laptop.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

Setting this bit in the clock enable register will stop the clock
when the card is in the IDLE state.

Signed-off-by: Will Newton <will.newton@imgtec.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

We need to run the card detect tasklet at the end of slot initialisation
as it is possible that a card has been inserted prior to boot, so we don't
see an insertion interrupt and now the card is sitting there inserted but
with no power to it.

Signed-off-by: Neil Jones <neil.jones@imgtec.com>
Signed-off-by: Will Newton <will.newton@imgtec.com>
Reviewed-by: Matt Fleming <matt@console-pimps.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

Currently MMC_MXC driver can be selected by all i.MX devices.

Restrict its use only for the appropriate processors.

Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Chris Ball <cjb@laptop.org>

Reviewed-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Chris Ball <cjb@laptop.org>

Add two large sequential I/O performance tests:
        35. Large sequential read into scattered pages
        36. Large sequential write from scattered pages

The tests measure transfer times for 10MiB, 100MiB, 1000MiB.

Signed-off-by: Adrian Hunter <adrian.hunter@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

Existing performance tests measure single or sequential I/O speed.
Add two random I/O tests:
        33. Random read performance by transfer size
        34. Random write performance by transfer size

Signed-off-by: Adrian Hunter <adrian.hunter@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

The test area size was set to the preferred erase size but for comparison
purposes it is better if it is the same size for different devices.  Make
it a multiple of preferred erase size that is greater than or equal to 4MiB.

Signed-off-by: Adrian Hunter <adrian.hunter@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

This sdio card supports having its sdio clock shutdown.
It is also not using the SDIO IRQ, but rather uses a side gpio irq.

Signed-off-by: Pierre Tardy <tardyp@gmail.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

Some sdio card are not following sdio standard, and do not work
when the sdio bus's clock is gated.

To keep functionnality for all legacy driver, we turn this quirk on
for every sdio card.
Drivers needs to disable the quirk manually when someone verifies that
their supported card works with clock gating.

Signed-off-by: Pierre Tardy <tardyp@gmail.com>
Acked-by: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

Some cards have quirks valid for every platforms using current
platform quirk hooks leads to a lot of code and debug duplication.

So we inspire a bit from what exists in PCI subsystem and do our own
per vendorid/deviceid quirk.  We still drop the complexity of the pci
quirk system (with special section tables, and so on).
That can be added later if needed.

Signed-off-by: Pierre Tardy <pierre.tardy@intel.com>
Acked-by: Linus Walleij <linus.walleij@stericsson.com>
Acked-by: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

Since mmc clock gating can also be used as a power gating
tip, it's better to put the led blinking after having
ungated the clock.

Signed-off-by: Pierre Tardy <pierre.tardy@intel.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

If the MMC host controller does not support waiting for card signaling
busy state (MMC_CAP_WAIT_WHILE_BUSY cap), there is no point in prining
the relevant warning message.

Signed-off-by: Pawel Moll <pawel.moll@arm.com>
Reviewed-by: Matt Fleming <matt@console-pimps.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

Samsung SDHCI host controller supports the Auto CMD12.

Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

While the MMC handled the card's read only flag correctly on open,
it did not setup the flag in the allocated disk structure. The
consequence being that probing the /sys/class/block/mmcblkX/ro
attribute always reported 0.

Signed-off-by: Marc-Andre Hebert <hebert.marcandre@gmail.com>
Reviewed-by: Chris Ball <cjb@laptop.org>
Tested-by: Chris Ball <cjb@laptop.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

Enhanced area feature is a new feature defined in eMMC4.4 standard. This
user data area provides higher performance/reliability, at the expense
of using twice the effective media space due to the area using SLC.

The MMC driver now reads out the enhanced area offset and size and adds
them to the device attributes in sysfs. Enabling the enhanced area can
only be done once, and should be done in manufacturing. To use this
feature, bit ERASE_GRP_DEF should also be set.

Documentation/ABI/testing/sysfs-devices-mmc describes the two new
attributes.

Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.com>
Reviewed-by: Chris Ball <cjb@laptop.org>
Signed-off-by: Chris Ball <cjb@laptop.org>

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq:
  [CPUFREQ] pcc-cpufreq: don't load driver if get_freq fails during init.

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cjb/mmc:
  mmc: fix CONFIG_MMC_UNSAFE_RESUME regression

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6:
  nd->inode is not set on the second attempt in path_walk()
  unfuck proc_sysctl ->d_compare()
  minimal fix for do_filp_open() race

Return 0 on failure. This will cause the initialization of the driver
to fail and prevent the driver from loading if the BIOS cannot handle
the PCC interface command to "get frequency". Otherwise, the driver
will load and display a very high value like "4294967274" (which is
actually -EINVAL) for frequency:

# cat /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq
4294967274

Signed-off-by: Naga Chumbalkar <nagananda.chumbalkar@hp.com>
CC: stable@kernel.org
Signed-off-by: Dave Jones <davej@redhat.com>

We leave it at whatever it had been pointing to after the
first link_path_walk() had failed with -ESTALE.  Things
do not work well after that...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

30201e7f

 ("mmc: skip detection of nonremovable cards on rescan")
allowed skipping detection of nonremovable cards on mmc_rescan().
The intention was to only skip detection of hardwired cards that
cannot be removed, so make sure this is indeed the case by directly
checking for (lack of) MMC_CAP_NONREMOVABLE, instead of using
mmc_card_is_removable(), which is overloaded with
CONFIG_MMC_UNSAFE_RESUME semantics.

The user-visible symptom of the bug this patch fixes is that no
"mmc: card XXXX removed" message appears in dmesg when a card is
removed and CONFIG_MMC_UNSAFE_RESUME=y.

Reported-and-tested-by: Dmitry Shmidt <dimitrysh@google.com>
Reported-and-tested-by: Maxim Levitsky <maximlevitsky@gmail.com>
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Chris Ball <cjb@laptop.org>

a) struct inode is not going to be freed under ->d_compare();
however, the thing PROC_I(inode)->sysctl points to just might.
Fortunately, it's enough to make freeing that sucker delayed,
provided that we don't step on its ->unregistering, clear
the pointer to it in PROC_I(inode) before dropping the reference
and check if it's NULL in ->d_compare().

b) I'm not sure that we *can* walk into NULL inode here (we recheck
dentry->seq between verifying that it's still hashed / fetching
dentry->d_inode and passing it to ->d_compare() and there's no
negative hashed dentries in /proc/sys/*), but if we can walk into
that, we really should not have ->d_compare() return 0 on it!
Said that, I really suspect that this check can be simply killed.
Nick?

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>