Bytes are being copied from/to a single page.  The intent
of the warning is to warn if the page boundary is crossed.
There are two problems.  First, PAGE_MASK is mistaken for
(PAGE_SIZE - 1).  Secondly, instead of using the number
of bytes to copy, the warning is using the maximum that
that value could be.  Fix both.
Signed-off-by: NAdrian Hunter <adrian.hunter@intel.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The ADMA2 descriptor table size was being calculated incorrectly
Fix it.

Note that it has been wrong for a long time and likely has not
caused any problems because of a combination of 1) not needing
alignment descriptors for block operations 2) more memory being
allocated than was requested 3) the use of
SDHCI_QUIRK_NO_ENDATTR_IN_NOPDESC which does not use an extra
descriptor for the end marker.
Signed-off-by: NAdrian Hunter <adrian.hunter@intel.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Add support for non-removable slots which have no card detection GPIO
and which should not be polled for a card change.
Signed-off-by: NTimo Kokkonen <timo.kokkonen@offcode.fi>
Acked-by: NLudovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Now that sdhci-pxav3 driver allows to have more than one IP clock defined,
document both clocks and clock-names properties.
Signed-off-by: NSebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Besides the I/O clock, some PXAv3 SDHCI IP also requires a core clock to
be enabled. Add an optional core clock to the corresponding driver.
Signed-off-by: NSebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

With support for more than one clock, we'll need to distinguish between
the clock by name. Change clock probing to first try to get "io" clock
before falling back to unnamed clock.
Signed-off-by: NSebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

As we are using references to the I/O clock throughout the driver,
move it to the private data. Also, in preparation for core clock,
rename it to clk_io.
Signed-off-by: NSebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

sdhci_add_host and sdhci_platfm_init already report failure,
so don't emit error messages when a failure occurs. This prevents
occurences of "deferred" messages when required power supplies
are not ready for operation yet.
Signed-off-by: NMike Looijmans <mike.looijmans@topic.nl>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

While allocating buffers for CXD data, let's use kzalloc() to make sure
those are zeroed.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Due to previous patches, all callers of mmc_send_cxd_data() now
allocates their buffers from the heap. This enables us to simplify
mmc_send_cxd_data() by removing the support of handling buffers, which
are allocated from the stack.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Previous patches has replaced the calls to mmc_send_ext_csd() into
mmc_get_ext_csd(), thus mmc_send_ext_csd() has become redundant. Let's
remove it.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

By using mmc_get_ext_csd() in favor of mmc_send_ext_csd, we decrease
code duplication.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

By using mmc_get_ext_csd() in favor of mmc_send_ext_csd, we decrease
code duplication.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Callers of mmc_send_ext_csd() will be able to decrease code duplication
by using mmc_get_ext_csd() instead. Let's make it available.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Instead of doing BUG_ON(), return an error code.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The callers of mmc_get_ext_csd() need the flexibility to handle errors
themselves, because they behave differently.

Let's clean up mmc_get_ext_csd() with its friends and adopt the error
handling as stated above.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

As a step in cleaning up code around reading/decoding EXT_CSD, convert
the current mmc_read_ext_csd(), to handle both fetching the EXT_CSD
and decoding its data.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The helper function mmc_can_ext_csd() will return a positive value if
the card supports the EXT_CSD register. Start using it at relavant
places in the mmc core.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Rely on the prints handled internally by kmalloc().
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

If the MMC spec version is < CSD_SPEC_VER_4, there aren't support for
the EXT_CSD register. Since max_dtr is fetched from there, it will
default to zero, which thus isn't needed to verify.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The validation of the buswidth and the MMC spec version in
__mmc_select_powerclass() is redundant, let's remove it.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

mmc_send_ext_csd() is an exported function used by both the mmc core
and the mmc block layer. Let's remove the local duplicated definition
in the mmc core.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Let callers of mmc_free_ext_csd() do kfree() directly instead.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The dw_mmc driver had a bunch of code that ran whenever a card was
ejected and inserted.  However, this code was old and crufty and
should be removed.  Some evidence that it's really not needed:

1. Is is supposed to be legal to use 'cd-gpio' on dw_mmc instead of
   using the built-in card detect mechanism.  The 'cd-gpio' code
   doesn't run any of the crufty old code but yet still works.

2. While looking at this, I realized that my old change (369ac861 mmc:
   dw_mmc: don't queue up a card detect at slot startup) actually
   castrated the old code a little bit already and nobody noticed.
   Specifically "last_detect_state" was left as 0 at bootup.  That
   means that on the first card removal none of the crufty code ran.

3. I can run "while true; do dd if=/dev/mmcblk1 of=/dev/null; done"
   while ejecting and inserting an SD Card and the world doesn't
   explode.

If some of the crufty old code is actually needed, we should justify
it and also put it in some place where it will be run even with
"cd-gpio".

Note that in my case I'm using the "cd-gpio" mechanism but for various
reasons the hardware triggers a dw_mmc "card detect" at bootup.  That
was actually causing a real bug.  The card detect workqueue was
running while the system was trying to enumerate the card.  The
"present != slot->last_detect_state" triggered and we were doing all
kinds of crazy stuff and messing up enumeration.  The new mechanism of
just asking the core to check the card is much safer and then the
bogus interrupt doesn't hurt.
Signed-off-by: NDoug Anderson <dianders@chromium.org>
Tested-by: NJaehoon Chung <jh80.chung@samsung.com>
Acked-by: NJaehoon Chung <jh80.chung@samsung.com>
Tested-by: Nalim.akhtar <alim.akhtar@samsung.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Presumably ->slotno is normally fairly small and the shift doesn't wrap
but static checkers will complain about it.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

"ret" is a signed int, so use "%d" in format strings instead of "%u".
This prevents cryptic codes in error messages like this:
sdhci-arasan e0101000.sdhci: platform register failed (4294966779)
Signed-off-by: NMike Looijmans <mike.looijmans@topic.nl>
Reviewed-by: NMichal Simek <michal.simek@xilinx.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Use the generic platform_data header file instead of mach/atmel-mci.h
Signed-off-by: NAlexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: NArnd Bergmann <arnd@arndb.de>
Acked-by: NLudovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Use the generic platform_data header file instead of mach/atmel-mci.h
Signed-off-by: NAlexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: NArnd Bergmann <arnd@arndb.de>
Acked-by: NLudovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Move the mach header that can come either from arm/mach-at91 or avr32 to
platform_data to be able to switch the AT91 platforms to multiplatform.
Signed-off-by: NAlexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: NArnd Bergmann <arnd@arndb.de>
Acked-by: NLudovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>
[Ulf: Fixed compile error]

For eMMC 5.0 compliant device, firmware version is stored in ext_csd.
Report firmware as a 64bit hexa decimal. Vendor can use hexa or ascii
string to report firmware version.
Also add FFU related EXT_CSD register and note if the device is FFU capable.
Signed-off-by: NGwendal Grignou <gwendal@chromium.org>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The "set_ios" function is called with a clock of 0 when the clock is
turning off.  There's no reason to go through all the extra Rockchip
logic (whose goal is to make sure DIV is 0 or 1) in that case.  The
Rockchip logic happened to work because the CCF will pick the lowest
possible rate when you ask it for a clock of 0, but it's silly to go
through all the remuxing and adjusting for no reason.
Signed-off-by: NDoug Anderson <dianders@chromium.org>
Acked-by: NJaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

In (28f92b5 mmc: core: Try other signal levels during power up) we can
see that there are times when it's valid to try several signal
voltages.  Don't print an ugly error in the logs when that happens.
Signed-off-by: NDoug Anderson <dianders@chromium.org>
Reviewed-by: NAlim Akhtar <alim.akhtar@samsung.com>
Acked-by: NJaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

We've already got a reset of DMA after it's done.  Add one before we
start DMA too.  This fixes a data corruption on Rockchip SoCs which
will get bad data when doing a DMA transfer after doing a PIO transfer.

We tested this on an Exynos 5800 with HS200 and didn't notice any
difference in sequential read throughput.
Signed-off-by: NSonny Rao <sonnyrao@chromium.org>
Signed-off-by: NDoug Anderson <dianders@chromium.org>
Tested-by: NDoug Anderson <dianders@chromium.org>
Acked-by: NJaehoon Chung <jh80.chung@samsung.com>
Tested-by: NJaehoon Chung <jh80.chung@samsung.com>
Reviewed-by: NAlim Akhtar <alim.akhtar@samsung.com>
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

In most of the cases mmc_get|set_drvdata() didn't simplify code, which
should be the primary reason for such macros.

Let's remove them and convert to the common device_driver macros,
dev_set|get_drvdata() instead.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The msm_sdcc host shall not use mmc core specific macros to handle its
driver data. Instead, convert to use the platform device driver macros.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The struct mmc_driver adds an extra layer on top of the struct
device_driver. That would be fine, if there were a good reason, but
that's not the case.

Let's simplify code by converting to the common struct device_driver
instead and thus also removing superfluous overhead.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Instead of having specific mmc system PM callbacks for the mmc driver,
let's convert to use the common ones.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

The macro is only used by the mmc core, so let's move it in there.
Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>

Signed-off-by: NUlf Hansson <ulf.hansson@linaro.org>