This commit fixes the three following races in PIO code:

- The CTRL0 register is racy in itself, when programming transfer state and
  run bit in the same cycle the hardware sometimes ends up using the state
  from the last transfer. Fix this by programming state in one cycle, make
  sure the write is flushed down APBX bus by reading back the reg and only
  then trigger the run bit.

- Only clear the DMAREQ bit in DEBUG0 after the read/write to the data reg
  happened. Otherwise we are racing with the hardware about who touches
  the data reg first.

- When checking for completion of a transfer it's not sufficient to check
  if the data engine finished, but also a check for i2c bus idle is needed.
  In PIO mode we are really fast to program the next transfer after a finished
  one, so the controller possibly tries to start a new transfer while the
  clkgen engine is still busy writing the NAK/STOP from the last transfer to
  the bus.
Signed-off-by: NLucas Stach <l.stach@pengutronix.de>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

i2c_del_adapter() always returns 0. So all checks testing whether it will be
non zero will always evaluate to false and the conditional code is dead code.
This patch updates all callers of i2c_del_mux_adapter() to ignore the return
value and assume that it will always succeed (which it will). In a subsequent
patch the return type of i2c_del_adapter() will be made void.
Signed-off-by: NLars-Peter Clausen <lars@metafoo.de>
Acked-by: NBen Hutchings <bhutchings@solarflare.com>
Reviewed-by: NJean Delvare <khali@linux-fr.org>
Acked-by: NShawn Guo <shawn.guo@linaro.org>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

There is simply no reason to be manually setting the private driver
data to NULL in the remove/fail to probe cases.  This is just extra
cruft code that can be removed.

A few notes:
* Nothing relies on drvdata being set to NULL.
* The __device_release_driver() function eventually calls
  dev_set_drvdata(dev, NULL) anyway, so there's no need to do it
  twice.
* I verified that there were no cases where xxx_get_drvdata() was
  being called in these drivers and checking for / relying on the NULL
  return value.

This could be cleaned up kernel-wide but for now just take the baby
step and remove from the i2c subsystem.
Reported-by: NWolfram Sang <wsa@the-dreams.de>
Reported-by: NStephen Warren <swarren@wwwdotorg.org>
Signed-off-by: NDoug Anderson <dianders@chromium.org>
Reviewed-by: NJean Delvare <khali@linux-fr.org>
Acked-by: NPeter Korsgaard <jacmet@sunsite.dk>
Reviewed-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: NMarek Vasut <marex@denx.de>
Signed-off-by: NWolfram Sang <wolfram@the-dreams.de>

This patch drops the i2c timing tables from this driver and instead
derives the timing based from the requested clock sleep. The timing
tables were completely wrong anyway when observed on a scope.

This new algorithm is also only derived by using a scope, but it seems
to produce much more accurate result.
Signed-off-by: NMarek Vasut <marex@denx.de>
Tested-by: NShawn Guo <shawn.guo@linaro.org>
[wsa: changed messages from dev_err to dev_warn]
Signed-off-by: NWolfram Sang <wolfram@the-dreams.de>

Add support for the PIO mode and mixed PIO/DMA mode support. The mixed
PIO/DMA is the default mode of operation. This shall leverage overhead
that the driver creates due to setting up DMA descriptors even for very
short transfers.

The current boundary between PIO/DMA 8 bytes, transfers shorter than 8
bytes are transfered by PIO, longer transfers use DMA. The performance
of write transfers remains unchanged, while there is a minor improvement
of read performance. Reading 16KB EEPROM with DMA-only operations gives
a read speed of 39.5KB/s, while with then new mixed-mode the speed is
blazing 40.6KB/s.
Signed-off-by: NMarek Vasut <marex@denx.de>
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>

cmd_err is used to handle error code, so it should not be unsigned.

This fixes the following warning when building with W=1 option:

drivers/i2c/busses/i2c-mxs.c: In function 'mxs_i2c_xfer_msg':
drivers/i2c/busses/i2c-mxs.c:331:19: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Acked-by: NMarek Vasut <marex@denx.de>
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>
Cc: stable@kernel.org

The init_completion() call does reinit not only the variable carrying
the flag that the completion finished, but also initialized the
waitqueue associated with the completion. On the contrary, the
INIT_COMPLETION() call only reinits the flag.

In case there was anything still stuck in the waitqueue, subsequent call
to init_completion() would be able to create possible race condition. This
patch uses the proper function and moves init_completion() into .probe() call
of the driver, to be issued only once.

Note that such scenario is impossible, since two threads can never enter the
mxs_i2c_xfer_msg(), since whole this section is protected by mutex in I2C core.
This by no means allows this issue to exit though.
Signed-off-by: NMarek Vasut <marex@denx.de>
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>

CONFIG_HOTPLUG is going away as an option.  As result the __dev*
markings will be going away.

Remove use of __devinit, __devexit_p, __devinitdata, __devinitconst,
and __devexit.
Signed-off-by: NBill Pemberton <wfp5p@virginia.edu>
Acked-by: Peter Korsgaard <peter.korsgaard@barco.com> (for ocores and mux-gpio)
Acked-by: Havard Skinnemoen <hskinnemoen@gmail.com> (for i2c-gpio)
Acked-by: Guan Xuetao <gxt@mprc.pku.edu.cn> (for puf3)
Acked-by: Barry Song <baohua.song@csr.com> (for sirf)
Reviewed-by: NJean Delvare <khali@linux-fr.org>
[wsa: Fixed "foo* bar" flaws while we are here]
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>

Properly terminate the DMA transfer in case the DMA PIO transfer
or setup fails for any reason.
Signed-off-by: NMarek Vasut <marex@denx.de>
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>

There is no reason to disable the I2C SMBus quick mode on this
IP block. Enable it. This essentially fixes the problem with the
"i2c-detect" command for probing the bus.
Signed-off-by: NMarek Vasut <marex@denx.de>
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>

Properly terminate the DMA transfer in case the DMA PIO transfer
or setup fails for any reason.
Signed-off-by: NMarek Vasut <marex@denx.de>
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>

This I2C master can do DMA and PIOQUEUE (PIO with FIFO). Originally,
only PIOQUEUE was supported and it had issues, then DMA support was added
this cycle. The original intention was to keep PIOQUEUE since it has
less overhead what is nice for small transfers. However, runtime
switching between PIOQEUE and DMA depending on the transfer size never
worked despite a lot of trying. Since PIOQUEUE mode itself was flaky
(polling at places where interrupts failed to work) and the
implementation also imposed a size limit for transfers, it is best to
remove the support, so users don't fall over its limitations. It also
makes the driver a lot cleaner and more robust. If somebody really wants
less overhead, plain PIO mode could still be implemented with the
addidtional advantage that this mode is also available on MX23, too.
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>
Reviewed-by: NMarek Vasut <marex@denx.de>

This patch implements DMA support into mxs-i2c. DMA transfers are now enabled
via DT. The DMA operation is enabled by default.
Signed-off-by: NMarek Vasut <marex@denx.de>
Tested-by: NFabio Estevam <fabio.estevam@freescale.com>

[wsa: rebased to 3.6-rc7]
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>

Commit cd4f2d4a (i2c: mxs: Set I2C timing registers for mxs-i2c) only
covered the case for devicetree and made platform_data based boards
bail out with -EINVAL. Correctly support the latter one, too.
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>

This patch configures the I2C bus timing registers according
to information passed via DT. Currently, 100kHz and 400kHz
modes are supported.

The TIMING2 register value is wrong in the documentation for
i.MX28! This was found and fixed by:
  Shawn Guo <shawn.guo@linaro.org>
Signed-off-by: NMarek Vasut <marex@denx.de>
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>

This patch mutes the false positive compiler warning:

drivers/i2c/busses/i2c-mxs.c: In function 'mxs_i2c_xfer_msg':
drivers/i2c/busses/i2c-mxs.c:206:8: warning: 'data' may be used uninitialized in this function [-Wuninitialized]
drivers/i2c/busses/i2c-mxs.c:196:6: note: 'data' was declared here
Signed-off-by: NMarc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>

Add device tree probe support for i2c-mxs driver.  So far, it's only
been tested on imx28.
Signed-off-by: NShawn Guo <shawn.guo@linaro.org>
Acked-by: NWolfram Sang <w.sang@pengutronix.de>

Cc: linux-i2c@vger.kernel.org
Cc: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: NShawn Guo <shawn.guo@linaro.org>

Since the last fixes to this driver ensure now the queue termination is
done correctly, we can finally disable the queue after a transfer
without problems. The gain is that it will only be reenabled after the
next transfer is fully set up. Before, the queue was running all the
time and if the setup of the next message was interrupted by another
thread, an incomplete buffer could have been sent, padded with zeroes.
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>

After an error interrupt setting cmd->err, I see another interrupt that
the data engine is empty which clears cmd->err before being processed.
So, clear cmd->err at the beginning of a transfer only to handle these
consecutive interrupts.
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>

The former mach specific reset_block function has been converted to a global
one. Use the new one to remove mach dependency from the driver.
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>

The hardware generates an interrupt for every completed command in the
queue while the code assumed that it will only generate one interrupt
when the queue is empty. So, explicitly check if the queue is really
empty. This patch fixed problems which occurred due to high traffic on
the bus. While we are here, move the completion-initialization after the
parameter error checking.
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>
Cc: Shawn Guo <shawn.guo@linaro.org>
Cc: Marek Vasut <marek.vasut@gmail.com>
Cc: Lothar Waßmann <LW@KARO-electronics.de>
Cc: stable@kernel.org

Fixes generated by 'codespell' and manually reviewed.
Signed-off-by: NLucas De Marchi <lucas.demarchi@profusion.mobi>

mxs_reset_block() clears the PIO_QUEUE_MODE bit. So we have
to set it again after a controller reset.
Signed-off-by: NSascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>
Cc: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: NBen Dooks <ben-linux@fluff.org>

Currently only supporting the PIOQUEUE-mode, because DMA-support for
this platform is not yet in mainline. When it becomes available and
support has been added to this driver, it will also be suitable for
i.MX23 and STMP3xxx.
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>
Cc: Ben Dooks <ben-linux@fluff.org>
Cc: Shawn Guo <shawn.guo@freescale.com>
Signed-off-by: NBen Dooks <ben-linux@fluff.org>