A significant increase (10-20%) in performance throughput for
USB mass storage is the reason for incrementing the value.
By some reason the USB driver allocates buffers which requires
a scattergather list to contain a lot more than 16 elements to
get optimal performance. This change sets the maximum elements
to 128.

Tests with large reads and large writes (100 MiB) show that the
throughput increase is significant for write (10% for this test)
but not for read. Tests are run on a Linux host with ext4 FS on
the gadget mass storage device. The sg-len still exceeds 16 for
the read tests but the performance gain is low or nothing.
Tested-by: NLinus Walleij <linus.walleij@linaro.org>
Signed-off-by: NPer Forlin <per.forlin@stericsson.com>
Signed-off-by: NUlf Hansson <ulf.hansson@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

pre_req() runs dma_map_sg() and prepares the dma descriptor for the next
mmc data transfer. post_req() runs dma_unmap_sg.  If not calling pre_req()
before mmci_request(), mmci_request() will prepare the cache and dma just
like it did it before.  It is optional to use pre_req() and post_req()
for mmci.
Signed-off-by: NPer Forlin <per.forlin@linaro.org>
Tested-by: NLinus Walleij <linus.walleij@linaro.org>
Signed-off-by: NChris Ball <cjb@laptop.org>

Stresstesting insert/remove of SD-cards can trigger
a StartBitErr. This made the driver to hang in forever
waiting for a non ocurring data timeout.

This bit and interrupt is documented in the original
PL180 TRM, just never implemented until now.
Signed-off-by: NUlf Hansson <ulf.hansson@stericsson.com>
Reviewed-by: NLinus Walleij <linus.walleij@stericsson.com>
Reviewed-by: NJonas Aberg <jonas.aberg@stericsson.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The mmci.h header contained a few registers not clearly marked
as ST Micro only, rectify this and remove the HWFC magic in the
process. The idea is to make the mmci.h header file more ordered
so other vendors with PL180 derivates can see where to put in
their custom register defines.

Includes portions of an earlier patch from Sebastian Rasmussen.
Acked-by: NSebastian Rasmussen <sebastian.rasmussen@stericsson.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Based on a patch from Linus Walleij.

Add dmaengine based support for DMA to the MMCI driver, using the
Primecell DMA engine interface.  The changes over Linus' driver are:

- rename txsize_threshold to dmasize_threshold, as this reflects the
  purpose more.
- use 'mmci_dma_' as the function prefix rather than 'dma_mmci_'.
- clean up requesting of dma channels.
- don't release a single channel twice when it's shared between tx and rx.
- get rid of 'dma_enable' bool - instead check whether the channel is NULL.
- detect incomplete DMA at the end of a transfer.  Some DMA controllers
  (eg, PL08x) are unable to be configured for scatter DMA and also listen
  to all four DMA request signals [BREQ,SREQ,LBREQ,LSREQ] from the MMCI.
  They can do one or other but not both.  As MMCI uses LBREQ/LSREQ for the
  final burst/words, PL08x does not transfer the last few words.
- map and unmap DMA buffers using the DMA engine struct device, not the
  MMCI struct device - the DMA engine is doing the DMA transfer, not us.
- avoid double-unmapping of the DMA buffers on MMCI data errors.
- don't check for negative values from the dmaengine tx submission
  function - Dan says this must never fail.
- use new dmaengine helper functions rather than using the ugly function
  pointers directly.
- allow DMA code to be fully optimized away using dma_inprogress() which
  is defined to constant 0 if DMA engine support is disabled.
- request maximum segment size from the DMA engine struct device and
  set this appropriately.
- removed checking of buffer alignment - the DMA engine should deal with
  its own restrictions on buffer alignment, not the individual DMA engine
  users.
- removed setting DMAREQCTL - this confuses some DMA controllers as it
  causes LBREQ to be asserted for the last seven transfers, rather than
  six SREQ and one LSREQ.
- removed burst setting - the DMA controller should not burst past the
  transfer size required to complete the DMA operation.
Tested-by: NLinus Walleij <linus.walleij@linaro.org>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

We don't need to store the number of bytes transferred in our host
structure - we can store this directly in data->bytes_xfered.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Implement a suggestion from Russell to drop the use of blockend
interrupts altogether and instead rely on the data counter.

Tested with error-free cards on U300, U8500 and RealView PB1176.
Signed-off-by: NUlf Hansson <ulf.hansson@stericsson.com>
Signed-off-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

On the U300 the MCI_DATAEND and MCI_DATABLOCKEND IRQs can arrive
out-of-order. Replace an ugly #ifdef hack with a proper runtime
solution which models what is really happening.

In the U300 DMA mode and on all Ux500 models, the MCI_DATABLOCKEND
flag isn't properly cleared in hardware following and ACK leading
to all kind of weird behaviour when the flag is still up in
subsequent interrupts, so we add two flags indicating the
error and handle this runtime.

Cc: Rabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The DBx500 variants have only one IRQ line hooked up.  Allow these (and
any other implementations which choose to use only one irq) to work by
directing the PIO interrupts also to the first IRQ line.
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

This adds a few registers to the MMCI/PL180 derivates that
is used for some odd control stuff like SDIO.
Signed-off-by: NMarcin Mielczarczyk <marcin.mielczarczyk@tieto.com>
Signed-off-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

If an IRQ can be requested on the card detected GPIO, use it instead of
polling.
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The Ux500 variant has a 32-word FIFO (TXFIFOEMPTY is asserted when it
has 2 left) and TXFIFOHALFEMPTY is repurposed as TXFIFOBURSTWRITEABLE,
with a burst being defined as 8-words.  Likewise for RX.
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Although both the U300 and Ux500 use ST variants, the HWFCEN bits are at
different positions, so use the variant_data to store the information.
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Add a variant_data structure to handle the differences between the
various variants of this peripheral.  Add a first quirk for a default
MCICLOCK value, required on the Ux500 variant where the enable bit needs
to be always set, since it controls access to some registers.
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The mmci driver's SG list iteration logic assumes that each SG entry
spans only one page, and only maps and flushes one page of the sg.  This
is not a valid assumption.  Fix it by converting the driver to the
sg_miter API, which correctly handles sgs which span multiple pages.
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

This adds support for an 8bit wide bus to the card (data lines
MCIDAT0 through 7 exist) on the ST Micro version and alters the
U300 platform to support this. Also add some ST_ prefix to the
ST-specific registers.
Signed-off-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

This enables the use of a regulator to power the MMCI/PL180
PrimeCell. The OCR mask is calculated and voltage is set using
the new MMC core functions for discovering voltage ranges
in regulators. The platform translate_vdd function which basically
controls the 4 lines out of the PL180 is disabled if you use a
regulator instead.
Signed-off-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

This moves the mmci platform data definition struct away from
arch/arm/include/asm/mach/mmc.h into the more proper place among
the other primecells in include/linux/amba/mmci.h and at the same
time renames it to "mmci.h", and also the struct in this file
confusingly named mmc_platform_data has been renamed
mmci_platform_data for clarity.

Cc: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Use gpiolib where available (and when valid GPIOs are provided) for
write protect/card detect status reporting.  We fall back to the old
'status' method where gpiolib support is not available.
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

This patch adds support for the ST Microelectronics version of
the PL180 PrimeCell. They use designer ID 0x80 and have a few
alterations/bugfixes related to open drain and HW flow control.
They also add some SDIO registers, I am unsure if these are
in ST HW only or if this is things also added in later ARM
revisions, but they are included in the mmci.h file for
completeness.
Signed-off-by: NLinus Walleij <linus.walleij@ericsson.com>
Signed-off-by: NPierre Ossman <drzeus@drzeus.cx>

Signed-off-by: NEmil Medve <Emilian.Medve@Freescale.com>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

Make sure all headers in the files reflect their true position
in the tree.
Signed-off-by: NPierre Ossman <drzeus@drzeus.cx>

Clean up the drivers/mmc directory by moving card and host drivers
into subdirectories.
Signed-off-by: NPierre Ossman <drzeus@drzeus.cx>

Signed-off-by: NEvgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.

Let it rip!