Commit graph

21 commits

Author SHA1 Message Date
Mark Brown
ae324b270a
Merge remote-tracking branches 'spi/topic/a3700', 'spi/topic/atmel', 'spi/topic/bcm53xx', 'spi/topic/davinci' and 'spi/topic/dw' into spi-next 2018-01-26 17:57:24 +00:00
Maxime Chevallier
34b1fcaeb2
spi: a3700: Remove endianness swapping for full-duplex transfers
Fixes the following sparse warnings :
line 767: warning: incorrect type in assignment (different base types)
line 767:    expected unsigned int [unsigned] [assigned] [usertype] val_out
line 767:    got restricted __le32 [usertype] <noident>
line 776: warning: cast to restricted __le32

This takes advantage of readl/writel to do the endianness reordering,
and removes an extra variable in the function.

Fixes: f68a7dcb91 ("spi: a3700: Add full-duplex support")
Signed-off-by: Maxime Chevallier <maxime.chevallier@smile.fr>
Reviewed-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
2018-01-24 15:03:31 +00:00
Maxime Chevallier
162f8debc0
spi: a3700: Remove endianness swapping functions when accessing FIFOs
Fixes the following sparse warnings :
line 504: warning: incorrect type in assignment (different base types)
line 504:    expected unsigned int [unsigned] [usertype] val
line 504:    got restricted __le32 [usertype] <noident>
line 527: warning: cast to restricted __le32

This is solved by removing endian-converson functions, since the
converted values are going through readl/writel anyway, which take care
of the conversion.

Fixes: 6fd6fd68c9 ("spi: armada-3700: Fix padding when sending not 4-byte aligned data")
Signed-off-by: Maxime Chevallier <maxime.chevallier@smile.fr>
Reviewed-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
2018-01-24 15:03:13 +00:00
Maxime Chevallier
f68a7dcb91
spi: a3700: Add full-duplex support
The armada 3700 SPI controller has support for full-duplex transfers,
but it can only be done without using the hardware FIFOs.

A full duplex transfer is done by shifting 4 bytes at a time, or even
one byte at a time for transfers less than 4 bytes long.

While this method is perfectly suitable for small transfers, it is still
slower than using the FIFOs.

This commit implement full-duplex support, making sure that half-duplex
transfers are still done using the FIFOs with the existing method.

Some setup functions were moved around to make sure the controller is
properly configured before beginning each transfer.

This was tested on EspressoBin with a logical analyser, and a simple
setup where MISO is connected on MOSI. Transfers were made from
userspace using spidev and spi-pipe from the spi-tools project

Signed-off-by: Maxime Chevallier <maxime.chevallier@smile.fr>
Signed-off-by: Mark Brown <broonie@kernel.org>
2018-01-18 11:10:45 +00:00
Maxime Chevallier
a456c9320d
spi: a3700: Allow to enable or disable FIFO mode
The armada 3700 SPI controller allows to make transfers without using
the 32 bytes RFIFO and WFIFO.

This commit enable switching between FIFO and non-FIFO mode, which is
necessary to implement full-duplex transfers.

Signed-off-by: Maxime Chevallier <maxime.chevallier@smile.fr>
Signed-off-by: Mark Brown <broonie@kernel.org>
2018-01-18 10:59:26 +00:00
Maxime Chevallier
abf3a49e50
spi: a3700: Set frequency limits at startup
Armada 3700 SPI controller has an internal clock divider which can
divide the parent clock frequency by up to 30.

This patch sets the limits in the spi_controller fields so that we can
detect when a non-supported frequency is requested by a device for a
transfer.

Signed-off-by: Maxime Chevallier <maxime.chevallier@smile.fr>
Signed-off-by: Mark Brown <broonie@kernel.org>
2018-01-18 10:59:18 +00:00
Maxime Chevallier
44a5f423e7
spi: a3700: Clear DATA_OUT when performing a read
When performing a read using FIFO mode, the spi controller shifts out
the last 2 bytes that were written in a previous transfer on MOSI.

This undocumented behaviour can cause devices to misinterpret the
transfer, so we explicitly clear the WFIFO before each read.

This behaviour was noticed on EspressoBin.

Signed-off-by: Maxime Chevallier <maxime.chevallier@smile.fr>
Signed-off-by: Mark Brown <broonie@kernel.org>
2018-01-18 10:59:12 +00:00
Maxime Chevallier
251c201bf4
spi: a3700: Fix clk prescaling for coefficient over 15
The Armada 3700 SPI controller has 2 ranges of prescaler coefficients.
One ranging from 0 to 15 by steps of 1, and one ranging from 0 to 30 by
steps of 2.

This commit fixes the prescaler coefficients that are over 15 so that it
uses the correct range of values. The prescaling coefficient is rounded
to the upper value if it is odd.

This was tested on Espressobin with spidev and a locigal analyser.

Signed-off-by: Maxime Chevallier <maxime.chevallier@smile.fr>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
2017-11-27 15:23:30 +00:00
Maxime Chevallier
dd7aa8d4b5 spi: a3700: Change SPI mode before asserting chip-select
The spi device mode should be configured in the controller before the
chip-select is asserted, so that a clock polarity configuration change
is not interpreted as a clock tick by the device.

This patch moves the mode setting to the 'prepare_message' function
instead of the 'transfer_one' function.

By doing so, this patch also removes redundant code in
a3700_spi_clock_set.

This was tested on EspressoBin board, with spidev.

Signed-off-by: Maxime Chevallier <maxime.chevallier@smile.fr>
Signed-off-by: Mark Brown <broonie@kernel.org>
2017-10-10 10:17:06 +01:00
Maxime Chevallier
5a866ec001 spi: a3700: Return correct value on timeout detection
When waiting for transfer completion, a3700_spi_wait_completion
returns a boolean indicating if a timeout occurred.

The function was returning 'true' everytime, failing to detect any
timeout.

This patch makes it return 'false' when a timeout is reached.

Signed-off-by: Maxime Chevallier <maxime.chevallier@smile.fr>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
2017-10-10 10:14:32 +01:00
Zachary Zhang
6fd6fd68c9 spi: armada-3700: Fix padding when sending not 4-byte aligned data
In 4-byte transfer mode, extra padding/dummy bytes '0xff' would be
sent in write operation if TX data is not 4-byte aligned since the
SPI data register is always shifted out as whole 4 bytes.

Fix this by using the header count feature that allows to transfer 0 to
4 bytes. Use it to actually send the first 1 to 3 bytes of data before
the rest of the buffer that will hence be 4-byte aligned.

Signed-off-by: Zachary Zhang <zhangzg@marvell.com>
Signed-off-by: Miquel Raynal <miquel.raynal@free-electrons.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
2017-09-13 09:37:49 -07:00
Miquel Raynal
747e1f6047 spi: armada-3700: Fix failing commands with quad-SPI
A3700 SPI controller datasheet states that only the first line (IO0) is
used to receive and send instructions, addresses and dummy bytes,
unless for addresses during an RX operation in a quad SPI configuration
(see p.821 of the Armada-3720-DB datasheet). Otherwise, some commands
such as SPI NOR commands like READ_FROM_CACHE_DUAL_IO(0xeb) and
READ_FROM_CACHE_DUAL_IO(0xbb) will fail because these commands must send
address bytes through the four pins. Data transfer always use the four
bytes with this setup.

Thus, in quad SPI configuration, the A3700_SPI_ADDR_PIN bit must be set
only in this case to inform the controller that it must use the number
of pins indicated in the {A3700_SPI_DATA_PIN1,A3700_SPI_DATA_PIN0} field
during the address cycles of an RX operation.

Suggested-by: Ken Ma <make@marvell.com>
Signed-off-by: Miquel Raynal <miquel.raynal@free-electrons.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
2017-09-13 09:37:26 -07:00
Mark Brown
244a60c28d Merge remote-tracking branches 'spi/topic/armada', 'spi/topic/ath79', 'spi/topic/bcm-qspi' and 'spi/topic/bcm53xx' into spi-next 2017-02-19 16:40:55 +00:00
Wei Yongjun
e92f0051c0 spi: armada-3700: Remove spi_master_put in a3700_spi_remove()
The call to spi_master_put() in a3700_spi_remove() is redundant since
the master is registered using devm_spi_register_master() and no
reference hold by using spi_master_get() in a3700_spi_remove().

This is detected by Coccinelle semantic patch.

Fixes: 5762ab71eb ("spi: Add support for Armada 3700 SPI Controller")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
2017-02-19 16:10:44 +00:00
Wei Yongjun
d9928b4c48 spi: armada-3700: Remove .owner field for driver
Remove .owner field if calls are used which set it automatically.

Generated by: scripts/coccinelle/api/platform_no_drv_owner.cocci

Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
2017-02-09 16:35:26 +00:00
Romain Perier
85798e153e spi: armada-3700: Coding style fixes
The following warning are reported by checkpatch.pl:

CHECK: Alignment should match open parenthesis
+static void a3700_spi_transfer_setup(struct spi_device *spi,
+                                   struct spi_transfer *xfer)

WARNING: Missing a blank line after declarations
+                       u32 data = le32_to_cpu(val);
+                       memcpy(a3700_spi->rx_buf, &data, 4);

total: 0 errors, 1 warnings, 1 checks, 923 lines checked

Signed-off-by: Romain Perier <romain.perier@free-electrons.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
2016-12-31 18:53:16 +00:00
Romain Perier
cfd6693c06 spi: armada-3700: Replaced raw values for nbits by the SPI macros
Currently, function a3700_spi_pin_mode_set() configures the SPI transfer
mode according to the value passed as second argument. This value is
detected using the raw values from a switch case.

This commit replaces these raw values by the corresponding macro
constants in linux/spi/spi.h

Signed-off-by: Romain Perier <romain.perier@free-electrons.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
2016-12-31 18:53:16 +00:00
Dan Carpenter
42cd4ed888 spi: armada-3700: Set mode bits correctly
We set SPI_RX_DUAL twice instead of setting SPI_TX_DUAL.

Fixes: 5762ab71eb ("spi: Add support for Armada 3700 SPI Controller")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Romain Perier <romain.perier@free-electrons.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
2016-12-16 12:28:36 +00:00
Dan Carpenter
0cc059abac spi: armada-3700: Remove unnecessary condition
We checked that "a3700_spi->wait_mask & cause" was set at the beginning
of the function so we don't need to check again here.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Romain Perier <romain.perier@free-electrons.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
2016-12-16 12:28:10 +00:00
Colin Ian King
f6f0083cca spi: armada-3700: fix unsigned compare than zero on irq
spi->irq is an unsigned integer hence the check if status is less than
zero has no effect.  Fix this by replacing spi->irq with an int irq
so the less than zero compare will correctly detect errors.

Issue found with static analysis with CoverityScan, CID1388567

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Romain Perier <romain.perier@free-electrons.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
2016-12-14 17:38:51 +00:00
Romain Perier
5762ab71eb spi: Add support for Armada 3700 SPI Controller
Marvell Armada 3700 SoC comprises an SPI Controller. This Controller
supports up to 4 SPI slave devices, with dedicated chip selects,supports
SPI mode 0/1/2 and 3, CPIO or Fifo mode with DMA transfers and different
SPI transfer mode (Single, Dual or Quad).

This commit adds basic driver support for FIFO mode. In this mode,
dedicated registers are used to store the instruction, the address, the
read mode and the data. Write and Read FIFO are used to store the
outcoming or incoming data. The data FIFOs are accessible via DMA or by
the CPU. Only the CPU is supported for now.

Signed-off-by: Romain Perier <romain.perier@free-electrons.com>
Tested-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
2016-12-08 16:05:34 +00:00