It turns out that in order for the SPI DMA to properly support
continuous transfers longer than 65280 bytes, there are some very
important parts that were left out from the documentation.
Firstly, the XFER_SIZE register is not written with the whole length
of a transfer, but is written by each and every chained descriptor
with the length of the descriptors data buffer.
Next, unlike the demo code supplied by FSL, which only writes one PIO
word per descriptor, this does not apply if the descriptors are chained,
since the XFER_SIZE register must be written. Therefore, it is essential
to use four PIO words, CTRL0, CMD0, CMD1, XFER_SIZE. CMD0 and CMD1 are
written with zero, since they don't apply. The DMA programs the PIO words
in an incrementing order, so four PIO words.
Finally, unlike the demo code supplied by FSL, the SSP_CTRL0_IGNORE_CRC
must not be set during the whole transfer, but it must be set only on the
last descriptor in the chain.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Fabio Estevam <festevam@gmail.com>
Cc: Otavio Salvador <otavio@ossystems.com.br>
Cc: Stefano Babic <sbabic@denx.de>
This patch fixes dcache-related problem. The problem manifested
when dcache was enabled and the following command issued twice:
mw 0x42000000 0 0x4000 ; sf probe ; sf read 0x42000000 0x0 0x10000 ; sha1sum 0x42000000 0x10000
The SHA1 checksum was correct during the first call. Yet with
every subsequent call of the above command, it differed and was
wrong.
It turns out this was because of a race condition. On the first
time the command was called, no cacheline contained any data from
the destination memory location. The DMA transfered data into the
location and the cache above the location was invalidated. Then the
checksum was computed, but that meant the data were loaded into data
cache.
On any subsequent call, the DMA again transfered data into the same
destination. Yet during the transfer, some of the DCache lines were
evicted and written back into the main memory. Once the DMA transfer
completed, the data cache was invalidated over the memory location as
usual. But the data that were to be loaded back into the data cache
by subsequent SHA1 checksuming were corrupted.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Fabio Estevam <festevam@gmail.com>
Cc: Otavio Salvador <otavio@ossystems.com.br>
Cc: Stefano Babic <sbabic@denx.de>
This change implements DMA chaining into SPI driver. This allows
the transfers to go much faster, while also fixing SF issues.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Fabio Estevam <festevam@gmail.com>
Cc: Otavio Salvador <otavio@ossystems.com.br>
Cc: Stefano Babic <sbabic@denx.de>
Use calloc() instead of malloc() to allocate the mxs_spi_slave structure.
Clearing the memory is necessary since most of the time this gets done
super early in boot, but on warm reboots, and when SPI probing is done
long after the init stages it could actually pick up previously used memory,
and things like the chipselect polarity and other data end up being filled
with trash data if not explicitly set by the board files.
This solves a semi-random, almost unreproducable error whereby SPI devices
act very, very strangly on boot.
Signed-off-by: Matt Sealey <matt@genesi-usa.com>
Acked-by: Stefano Babic <sbabic@denx.de>
The sys_proto.h functions (except the boot modes) are compatible with
i.MX233 and i.MX28 so we use 'mxs' prefix for its methods.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
The DMA transfers happen only if the transfered data are larger
than 512 bytes. Otherwise PIO is used. This is a small speed
optimization.
The DMA transfer doesn't work if unaligned transfer is requested
due to the limitation of the DMA controller. This has to be fixed
by introducing generic bounce buffer. Therefore the DMA feature
is now disabled by default.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Fabio Estevam <festevam@gmail.com>
Cc: Otavio Salvador <otavio@ossystems.com.br>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Pull out all the PIO transfer logic into separate function,
so DMA can be added.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Fabio Estevam <festevam@gmail.com>
Cc: Otavio Salvador <otavio@ossystems.com.br>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
This makes it easier to adapt for addition of DMA support.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Fabio Estevam <festevam@gmail.com>
Cc: Otavio Salvador <otavio@ossystems.com.br>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
MXS SSP controller may have up to three chip selects per port: SS0, SS1 and SS2.
Currently only SS0 is supported in the mxs_spi driver.
Allow all the three chip select to work by selecting the desired one
in bits 20 and 21 of the HW_SSP_CTRL0 register.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Marek Vasut <marex@denx.de>
Introduce spi_cs_is_valid() for validating spi bus and chip select numbers.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Marek Vasut <marex@denx.de>
Acked-by: Mike Frysinger <vapier@gentoo.org>
Instead of returning -1, it is preferred to return -ETIMEDOUT in case of timeouts.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Marek Vasut <marex@denx.de>
The generic spi flash driver (drivers/mtd/spi/spi_flash.c) uses the
spi low level driver's spi_xfer() function with len=0 to deassert the
SPI flash' chip select. But the i.MX28 spi driver rejects this call
due to len=0.
This patch implements an exception for len=0 with the SPI_XFER_END
flag set. This results in an extra read with the chip select being
deasserted afterwards. There seems to be no way to deassert the signal
by hand.
Signed-off-by: Matthias Fuchs <matthias.fuchs@esd.eu>
Tested-by: Fabio Estevam <fabio.estevam@freescale.com>
