libata/for-3.10-fixes never got submitted during v3.10 cycle. Merge
it into for-3.11 so that it can be routed together with other changes
scheduled for v3.11.
Three trivial conflicts in drivers/ata/sata_rcar.c. All are caused by
1b20f6a9ad ("sata_rcar: add 'base' local variable to some functions")
conflicting with logic updates in for-3.10-fixes. The offending
commit simply adds local variable @base on functions which
dereferences sata_rcar_priv->base multiple times. The resolutions are
trivial - applying s/priv->base/base/ in the conflicting logic
updates.
Signed-off-by: Tejun Heo <tj@kernel.org>
When compiling the driver with gcc 4.8, it gives the following warning:
drivers/ata/sata_rcar.c: In function `sata_rcar_thaw':
drivers/ata/sata_rcar.c:183:2: warning: large integer implicitly truncated to unsigned type [-Woverflow]
Fix the warning by explicit cast of the 'unsigned long' value to 'u32'.
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
The driver's interrupt handling code is too picky in deciding whether it should
handle an interrupt or not which causes completely unneeded spurious interrupts.
Thus make sata_rcar_{ata|serr}_interrupt() *void*; add ATA status register read
to sata_rcar_ata_interrupt() to clear an unexpected ATA interrupt -- it doesn't
get cleared by writing to the SATAINTSTAT register in the interrupt mode we use.
Also, in sata_rcar_ata_interrupt() we should check SATAINTSTAT register only for
enabled interrupts and we should clear only those interrupts that we have read
as active first time around, because else we have a race and risk clearing an
interrupt that can occur between read and write of the SATAINTSTAT register
and never registering it...
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: stable@vger.kernel.org
The 'base' field of 'struct sata_rcar_priv' is used very often
throughout the driver, so it seems worth loading it into a local
variable if it's used more than once in a function.
While at it, put some unitialized variables after intialized ones for
aesthetic reasons. :-)
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Using ATA_BMDMA_SHT() to intialize 'sata_rcar_sht' was suboptimal as
the R-Car descriptor table transfer counter is 28 bits wide (bit 1 to
bit 28), so that the 'dma_boundary' field of 0xFFFF is just too small,
as well as the 'sg_tablesize' field of 128. Use ATA_BASE_SHT() to
initialize 'sata_rcar_sht' instead and give proper values to the
'dma_boundary' and 'sg_tablesize' fields explicitly.
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
I've modified sata_rcar_bmdma_fill_sg() to take care of splitting long
scatter/ gather segments due to the descriptor table transfer counter
being only 28 bits wide (bit 1 to bit 28) but that was in vain as even
if 'sata_rcar_sht' specified a correct 'dma_boundary' field, the DMA
and block layers would have split the S/G segments on the necassary
boundaries. Since the driver uses ATA_BMDMA_SHT() to initilaize
'sata_rcar_sht', the boundary is much smaller, only 0xFFFF, so the
code I've added is even more useless, and it's better to just remove
it.
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Use the wrapper functions for getting and setting the driver data using
platform_device instead of using dev_{get,set}_drvdata() with &pdev->dev,
so we can directly pass a struct platform_device.
Also, unnecessary dev_set_drvdata() is removed, because the driver core
clears the driver data to NULL after device_release or on probe failure.
Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Iff bmdma_setup() has to stop a DMA transfer before starting a new
one, then the STOP bit in the ATAPI_CONTROL1 register will remain set
(it's only cleared when setting the START bit to 1) and then
bmdma_start() method will set both START and STOP bits simultaneously
which should abort the transfer being just started. Avoid that by
explicitly clearing the STOP bit in bmdma_start() method (in this case
it will be ignored on write).
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: stable@vger.kernel.org
Use the newly introduced devm_ioremap_resource() instead of
devm_request_and_ioremap() which provides more consistent error handling.
Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Acked-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Tejun Heo