Signed-off-by: Pierre Ossman <pierre@ossman.eu>

The SDHCI controller found in the VX855ES requires 10ms
delay between applying power and applying clock.

This issue has been discovered and documented by the OLPC XO1.5 team.

Signed-off-by: Harald Welte <HaraldWelte@viatech.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

This was missed in the DMA changes during the s3c24xx
updates in commit 8970ef47.

Signed-off-by: Ben Dooks <ben@simtec.co.uk>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

This adds the via-sdmmc driver for the SD/MMC-controller of VIA,
which is found in a number of recent integrated VIA chipset
products.

Signed-off-by: Harald Welte <HaraldWelte@viatech.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Some hosts (hardware configurations, or particular SD/MMC slots) may
not support 4-bit bus. For example, on MPC8569E-MDS boards we can
switch between serial (1-bit only) and nibble (4-bit) modes, thought
we have to disable more peripherals to work in 4-bit mode.

Along with some small core changes, this patch modifies sdhci-of
driver, so that now it looks for "sdhci,1-bit-only" property in the
device-tree, and if specified we enable a proper quirk.

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Add quirk to show the controller cannot do multi-block IO.

This is mainly for the Samsung SDHCI controller that currently
cannot manage to do multi-block PIO without timing out.

Signed-off-by: Ben Dooks <ben@simtec.co.uk>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Update the ADMA error reporting to not only show the
overall controller state but also to print the ADMA
descriptor list.

Signed-off-by: Ben Dooks <ben@simtec.co.uk>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Add support for the 'HSMMC' block(s) in the Samsung SoC
line. These are compatible with the SDHCI driver so add
the necessary setup and driver binding for the platform
devices.

Signed-off-by: Ben Dooks <ben@simtec.co.uk>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

This was missed in the DMA changes during the s3c24xx
updates in commit 8970ef47.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>

New revision of Atmel MCI interface adds new features. This is a update of
register definition in header file. This new MCI IP is called MCI2.

Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Acked-by: Haavard Skinnemoen <haavard.skinnemoen@atmel.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

The MCI IP is shared among AVR32 and AT91 SOCs.
AT91 has specific bit definitions in the user interface of MCI SD/MMC IP.

Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Acked-by: Haavard Skinnemoen <haavard.skinnemoen@atmel.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Now tmio_mmc is able to drive the MMC/SD cell in ASIC3.

Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Acked-by: Ian Molton <spyro@f2s.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

In the write recovery routine, the data to get from the card
is allocated from the stack. The DMA mapping documentation says
explicitly stack memory is not mappable by any of the DMA calls.

Change to using kmalloc() to allocate the memory for the result
from the card and then free it once we've finished with the
transaction.

[ Changed to GFP_KERNEL allocation - Pierre Ossman ]

Signed-off-by: Ben Dooks <ben@simtec.co.uk>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

If using ADMA, then we should print the ADMA error
and current pointer in sdhci_dumpregs() when any
debug is requested.

Signed-off-by: Ben Dooks <ben@simtec.co.uk>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

This patch fixes the clock setup in tmio_mmc.

  * Incorrect divider setting
  * Cruft written to the clock registers (seemingly harmless but Not
Good (tm))

It also eliminates some unnecessary ifs and tidies the loop syntax.

Thanks to Philipp Zabel who discovered the divider issue, commenting

   "Except for the SDCLK = HCLK (divider bypassed) case, the clock
    setting resulted in double the requested frequency.
    The smallest possible frequency (f_max/512) is configured with
    a divider setting 0x80, not 0x40."

Signed-off-by: Ian Molton <ian@mnementh.co.uk>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

ASIC3 can disable the memory, so we need to wait for mfd_cell->enable
to enable the memory before we can map the SD control registers.

Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Acked-by: Ian Molton <ian@mnementh.co.uk>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Acked-by: Ian Molton <ian@mnementh.co.uk>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Use an IRQF_TRIGGER_ flag in request_irq instead.

Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Acked-by: Ian Molton <ian@mnementh.co.uk>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Some ASIC3 devices in the wild are connected with the address bus shifted
by one line, so that its 16-bit registers appear 32-bit aligned in host
memory space.

Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Acked-by: Ian Molton <ian@mnementh.co.uk>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

The Toshiba parts all have a 24 MHz HCLK, but HTC ASIC3 has a 24.576 MHz HCLK
and AMD Imageon w228x's HCLK is 80 MHz. With this patch, the MFD driver
provides the HCLK frequency to tmio_mmc via mfd_cell->driver_data.

Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Acked-by: Ian Molton <ian@mnementh.co.uk>
Acked-by: Samuel Ortiz <sameo@openedhand.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

This patch changes the reported error code for the responses
to a command from EINVAL to EFAULT/ENOSYS, as EINVAL is reserved
for non-recoverable host errors, and the responses from
the SD/MMC card may be because of recoverable transmission
errors in the command or in the response. Response codes
in SPI mode are NOT protected by a checksum, so don't trust them.

Signed-off-by: Wolfgang Muees <wolfgang.mues@auerswald.de>
Acked-by: Matt Fleming <matt@console-pimps.org>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Added a platform driver which uses the SDHCI core.

Signed-off-by: Richard Röjfors <richard.rojfors.ext@mocean-labs.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

The MMC core has now been fixed to not send silly frequencies to the
drivers which means we can remove this workaround.

Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Signed-off-by: Pierre Ossman <pierre@ossman.eu>

The code is divided in two parts. There is a virtual 'bus' driver
that handles PCI device and registers three new devices one per card
reader type. The other driver handles SD/MMC part of the reader.

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Changes pxamci.c to use the regulator subsystem. Uses the regulator case
CONFIG_REGULATOR is defined and a matching is regulator is provided, or
falls back to pdata->setpower otherwise. A warning is displayed case
both a valid regulator and pdata is set, and the regulator is used.

Signed-off-by: Daniel Ribeiro <drwyrm@gmail.com>
Acked-by: Eric Miao <eric.miao@marvell.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Some controllers allow a much lower frequency than 400kHz.
Keep the minimum frequency within sensible limits.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Because of granularity issues, sometimes we told the hardware to change
to the voltage we were already at. Rework the logic so this doesn't
happen.

Signed-off-by: Pierre Ossman <pierre@ossman.eu>

A pointer to mmc_omap_probe which lives in .init.text is passed to the
core via platform_driver_register and so the kernel might oops if probe
is called after the init code is discarded.

As requested by David Brownell platform_driver_probe is used instead of
moving the probe function to .devinit.text.  This saves some memory, but
might have the downside that a device being registered after the call to
mmc_omap_init but before the init sections are discarded will not be
bound anymore to the driver.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Speedup for slow cards by transfering more data at once.
This patch also reduces the amount of wear-out of the flash
blocks because fewer partial blocks are written.

Signed-off-by: Wolfgang Muees <wolfgang.mues@auerswald.de>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

With this patch, mmc_rescan can detect the removal of an mmc card and
the insertion of (possibly another) card in the same run. This means
that a card change can be detected without having to call
mmc_detect_change multiple times.

This change generalises the core such that it can be easily used by
hosts which provide a mechanism to detect only the presence of a card
reader cover, which has to be taken off in order to insert a card. Other
hosts ("card detect" or "MMC_CAP_NEEDS_POLL") each receive an event when
a card is removed and when a card is inserted, so it is sufficient for
them if mmc_rescan handles only one event at a time. "Cover detect"
hosts, however, only receive events about the cover status. This means
that between 2 subsequent events, both a card removal and a card
insertion can occur. In this case, the pre-patch version of mmc_rescan
would only detect the removal of the previous card but not the insertion
of the new card.

Signed-off-by: Jorg Schummer <ext-jorg.2.schummer@nokia.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

I found the PrimeCell/AMBA Bus drivers distrusting the resource
passed in as part of the struct amba_device abstraction. This
patch removes all hard coded resource sizes found in the PrimeCell
drivers and move the responsibility of this definition back to
the platform/board device definition, which already exist and
appear to be correct for all in-tree users of these drivers.
We do this using the resource_size() inline function which was
also replicated in the only driver using the resource size, so
that has been changed too. The KMI_SIZE was left in kmi.h in case
someone likes it. Test-compiled against Versatile and Integrator
defconfigs, seems to work but I don't posess these boards and
cannot test them.

Signed-off-by: Linus Walleij <linus.walleij@stericsson.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

Freescale eSDHC controller has the special order for
the HOST version register. that is not same as the other's
registers. The address of HOSTVER in spec is 0xFE, and
we need use the in_be16(0xFE) to access it, not in_be16(0xFC).

Signed-off-by: Dave Liu <daveliu@freescale.com>
Acked-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Especially with Sandisk SDHC cards, the second SWITCH command was failing
with a timeout and the card was not recognized at all.  However if the
system was busy, or debugging was enabled, or a udelay(100) was inserted
before the second SWITCH command in the core code, then the timing was
so that the card started to work.

With some unusual block sizes, the data FIFO status doesn't indicate a
"empty" state right away when the data transfer is done.  Queuing
another data transfer in that condition results in a transfer timeout.

The empty FIFO bit eventually get set by itself in less than 50 usecs
when it is not set right away. So let's just poll for that bit before
configuring the controller with a new data transfer.

Signed-off-by: Nicolas Pitre <nico@marvell.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Empirical evidences show that this is causing far more problems than it
solves when this mode is enabled in the host hardware.  Amongst those
cards that are known to be non functional when this bit is set are:

	A-Data "Speedy" 2GB SD card
	Kodak 512MB SD card
	Ativa 1GB MicroSD card
	Marvell 8688 (WIFI/Bluetooth) SDIO card

Since those cards do work on other host controllers which do honnor the
hs timing, the issue must be with this particular host hardware.

Signed-off-by: Nicolas Pitre <nico@marvell.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

disable_irq() should wait for all running handlers to complete
before returning.  As such, if it's used to disable an interrupt
from that interrupt's handler it will deadlock.  This replaces
the dangerous instances with the _nosync() variant which doesn't
have this problem.

Signed-off-by: Ben Nizette <bn@niasdigital.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

We plan to use fsl,esdhc going forward as the base compatible so update
the driver to bind against it.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Signed-off-by: Nicolas Pitre <nico@marvell.com>
Tested-by: Martin Michlmayr <tbm@cyrius.com>
Signed-off-by: Pierre Ossman <pierre@ossman.eu>

Signed-off-by: Martin Fuzzey <mfuzzey@gmail.com>