Some users have reported that in polled mode the driver fails randomly
to read the last word of the transfer.

The end condition used for the transmissions (in polled and irq mode)
has been the TX_EMPTY flag. But Lars-Peter Clausen has identified a delay
from the TX_EMPTY to the actual end of the data rx.

I believe that this race condition has not been detected until now
because of the latency added by the IRQ handler or the PCIe bridge.
This bugs affects setups with low latency access to the spi core.

This patch replaces the readout logic:

For all the words, except the last one, the TX_EMPTY flag is used (and
cached).

If !TX_EMPY or is the last word. The status register is read and the
RX_EMPTY flag is used.

The performance is not affected: there is an extra read of the
Status Register, but the readout can start as soon as there is a word
in the buffer.
Reported-by: NEdward Kigwana <ekigwana@scires.com>
Initial-fix-by: NLars-Peter Clausen <lars@metafoo.de>
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org

The ACK of an inexistent IRQ can trigger an spurious IRQ that breaks the
txrx logic. This has been observed on axi_quad_spi:3.2 core.

This patch only ACKs IRQs that have not been Acknowledge jet.
Reported-by: NEdward Kigwana <ekigwana@scires.com>
Tested-by: NEdward Kigwana <ekigwana@scires.com>
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org

Once the module process a transfer in irq mode, the next poll transfer
will not work because the transmitter is left in inhibited state.

Fixes: 22417352 (Use polling mode on small transfers)
Reported-by: NEdward Kigwana <ekigwana@scires.com>
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org

On 1 and 2 bytes per word, the transfer of the 3 last bytes will access
memory outside tx_ptr.

Although this has not trigger any error on real hardware, we should
better fix this.

Fixes: 24ba5e59 (Remove rx_fn and tx_fn pointer)
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

This reverts commit 99082eab since it
breaks the build due to differing implementations of iowrite() and
ioread().
Signed-off-by: NMark Brown <broonie@kernel.org>

The core only supports up to 32 slaves, and the chipselect function
expects the same.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Small transfers generally can be accomplished faster in polling mode.
This patch select the transfer which size is bellow the buffer size to
be done on polling mode
Suggested-by: NMark Brown <broonie@kernel.org>
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

The variable never leaves the scope of txrx_bufs.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Save a stack level and cleanup code.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Simplify the code by using the unit used on most of the code logic.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Simplify the code by using the unit used on most of the code logic.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

spi_rx handles the case where the buffer is null. Nevertheless spi_tx
did not handle it, and was handled by the caller function.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Simplify the code by removing the tx and and rx function pointers and
substitute them by a single function.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

The core controls the chip select lines individually.

By default, all the lines are consider active_low. After
spi_setup_transfer, it has its real value.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

When no irq is used, there is no need to inhibit the transmission for
every transaction. This inhibition was implemented to avoid a race
condition with the irq handler.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

The core can run in polling mode. In fact, the performance of the core
is similar (or even better), due to the fact most of the spi
transactions are just a couple of bytes and there is one irq per
transactions.

When an mtd device is connected via spi, reading 8MB of data produces
more than 80K interrupts (with irq disabling, context swith....)
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

The control register has not changed since the previous access.
Therefore we can use the cached value and safe one bus access.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

On the transmission loop, check for remaining bytes at the loop
condition.

This way we can handle transmissions of 0 bytes and clean the code.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Instead of enabling the IRQ and disabling it for every transaction.

Specially the small transactions (1,2 words) benefit from removing 3 bus
accesses.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Instead of checking the TX_FULL flag for every transaction, find out the
size of the buffer at probe time and use it.

To avoid situations where the core had some data on the buffer before
initialization, the core is reseted before the buffer size is detected
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

The number of words in the read buffer will be exactly the same as the
number of words written on write buffer, once the transaction has
finished.

Instead of cheking the rx_empty flags for every word simply save the
number of words written by fill_tx_fifo.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Hardware supports LOOP mode. Support it also in the driver.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Hardware supports LSB_FIRST mode. Support it also in the driver.
Signed-off-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

There is no need to init .owner field.

Based on the patch from Peter Griffin <peter.griffin@linaro.org>
"mmc: remove .owner field for drivers using module_platform_driver"

This patch removes the superflous .owner field for drivers which
use the module_platform_driver API, as this is overriden in
platform_driver_register anyway."
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

The Xilinx SPI driver initializes the bus_num field of the SPI master to
pdev->dev.id. pdev->dev.id is always 0 for platform devices which causes bus
number conflicts for the SPI controller when creating multiple device instances
of the driver. Instead use pdev->id which will have the assignt platform device
id. If the device is instantiated via devicetree pdev->id is -1 which will cause
the SPI core to dynamically assign a id (or use the id defined via a alias in
the devicetree).
Signed-off-by: NLars-Peter Clausen <lars@metafoo.de>
Signed-off-by: NMark Brown <broonie@linaro.org>

Remove unnecessary space in order to fix the following
checkpatch issues.

  WARNING: Unnecessary space after function pointer name
Signed-off-by: NJingoo Han <jg1.han@samsung.com>
Reviewed-by: NMichal Simek <monstr@monstr.eu>
Signed-off-by: NMark Brown <broonie@linaro.org>

Set bits_per_word_mask so spi core will reject transfers that attempt to use
an unsupported bits_per_word value.

spi_bitbang requires custom setup_transfer() to be defined if there is a
custom txrx_bufs(). Thus keep the empty xilinx_spi_setup_transfer() function
in the code.
Signed-off-by: NAxel Lin <axel.lin@ingics.com>
Acked-by: NMichal Simek <michal.simek@xilinx.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

None of these files are actually using any __init type directives
and hence don't need to include <linux/init.h>.  Most are just a
left over from __devinit and __cpuinit removal, or simply due to
code getting copied from one driver to the next.
Signed-off-by: NPaul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

Use this new function to make code more comprehensible, since we are
reinitialzing the completion, not initializing.

[akpm@linux-foundation.org: linux-next resyncs]
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>
Acked-by: Linus Walleij <linus.walleij@linaro.org> (personally at LCE13)
Cc: Ingo Molnar <mingo@kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Many drivers that use bitbang library have a leak on probe error paths.
This is because once a spi_master_get() call succeeds, we need an additional
spi_master_put() call to free the memory.

Fix this issue by moving the code taking a reference to master to
spi_bitbang_start(), so spi_bitbang_start() will take a reference to master on
success. With this change, the caller is responsible for calling
spi_bitbang_stop() to decrement the reference and spi_master_put() as
counterpart of spi_alloc_master() to prevent a memory leak.

So now we have below patten for drivers using bitbang library:

probe:
spi_alloc_master        -> Init reference count to 1
spi_bitbang_start       -> Increment reference count
remove:
spi_bitbang_stop        -> Decrement reference count
spi_master_put          -> Decrement reference count (reference count reaches 0)

Fixup all users accordingly.
Signed-off-by: NAxel Lin <axel.lin@ingics.com>
Suggested-by: NUwe Kleine-Koenig <u.kleine-koenig@pengutronix.de>
Acked-by: NUwe Kleine-Koenig <u.kleine-koenig@pengutronix.de>
Signed-off-by: NMark Brown <broonie@linaro.org>

Many drivers that use bitbang library have a leak on probe error paths.
This is because once a spi_master_get() call succeeds, we need an additional
spi_master_put() call to free the memory.

Fix this issue by moving the code taking a reference to master to
spi_bitbang_start(), so spi_bitbang_start() will take a reference to master on
success. With this change, the caller is responsible for calling
spi_bitbang_stop() to decrement the reference and spi_master_put() as
counterpart of spi_alloc_master() to prevent a memory leak.

So now we have below patten for drivers using bitbang library:

probe:
spi_alloc_master        -> Init reference count to 1
spi_bitbang_start       -> Increment reference count
remove:
spi_bitbang_stop        -> Decrement reference count
spi_master_put          -> Decrement reference count (reference count reaches 0)

Fixup all users accordingly.
Signed-off-by: NAxel Lin <axel.lin@ingics.com>
Suggested-by: NUwe Kleine-Koenig <u.kleine-koenig@pengutronix.de>
Acked-by: NUwe Kleine-Koenig <u.kleine-koenig@pengutronix.de>
Signed-off-by: NMark Brown <broonie@linaro.org>

Use the wrapper function for retrieving the platform data instead of
accessing dev->platform_data directly.
Signed-off-by: NJingoo Han <jg1.han@samsung.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

Now that the bitbang core does not require a setup() function we can
drop the check in the altera, nuc900 and xilinx drivers.
Signed-off-by: NMark Brown <broonie@linaro.org>

In xilinx_spi_probe() we use xspi->irq to store negative error codes so
it has to be signed.  We weren't going to use the upper bit any way so
this is fine.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

Use devm_request_irq() for irq allocation which
simplify driver code.
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

It simplifies driver probing.
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

devm_ioremap_resource() automatically checks that
struct resource is initialized.
Also group platform_get_resource() and devm_ioremap_resource()
together.
And remove mem resource from struct xilinx_spi.
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

dev.of_node is in struct device all the time.
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>
Signed-off-by: NMark Brown <broonie@linaro.org>

This is a more traditional name and makes things a bit clearer when
referring to actual struct devices as we do frequently during probe().
Signed-off-by: NMark Brown <broonie@linaro.org>
Acked-by: NMichal Simek <michal.simek@xilinx.com>

In the past there used to be a separate platform device type for device
tree systems so the probe and removal functions were split into generic
and bus sections. Since this is no longer the case simplify the code (and
remove some unprototyped exports) by factoring everything into the bus
probe() and remove().
Signed-off-by: NMark Brown <broonie@linaro.org>