Commit Graph

14 Commits

Author SHA1 Message Date
Peter Ujfalusi 1c7f072d94 dmaengine: virt-dma: Support for race free transfer termination
Even with the introduced vchan_synchronize() we can face race when
terminating a cyclic transfer.

If the terminate_all is called after the interrupt handler called
vchan_cyclic_callback(), but before the vchan_complete tasklet is called:
vc->cyclic is set to the cyclic descriptor, but the descriptor itself was
freed up in the driver's terminate_all() callback.
When the vhan_complete() is executed it will try to fetch the vc->cyclic
vdesc, but the pointer is pointing now to uninitialized memory leading to
(hard to reproduce) kernel crash.

In order to fix this, drivers should:
- call vchan_terminate_vdesc() from their terminate_all callback instead
calling their free_desc function to free up the descriptor.
- implement device_synchronize callback and call vchan_synchronize().

This way we can make sure that the descriptor is only going to be freed up
after the vchan_callback was executed in a safe manner.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2017-12-04 22:33:51 +05:30
Peter Ujfalusi 6af149d2b1 dmaengine: virt-dma: Add helper to free/reuse a descriptor
The vchan_vdesc_fini() can be used to free or reuse a given descriptor
after it has been marked as completed.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2017-12-04 22:33:51 +05:30
Baoyou Xie 02aa84860c dmaengine: virt-dma: move function declarations
We get 2 warnings when building kernel with W=1:
drivers/dma/virt-dma.c:22:14: warning: no previous prototype for 'vchan_tx_submit' [-Wmissing-prototypes]
drivers/dma/virt-dma.c:52:5: warning: no previous prototype for 'vchan_tx_desc_free' [-Wmissing-prototypes]

In fact, these two functions are incorrectly declared in a function.

So this patch moves function declarations out of this function.

Signed-off-by: Baoyou Xie <baoyou.xie@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2016-10-01 11:41:39 +05:30
Masahiro Yamada 360af35b08 dmaengine: cleanup with list_first_entry_or_null()
The combo of list_empty() check and return list_first_entry()
can be replaced with list_first_entry_or_null().

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2016-09-14 19:10:46 +05:30
Vinod Koul d3f1e93ce8 Merge branch 'topic/async' into for-linus 2016-01-06 15:17:47 +05:30
Robert Jarzmik 13bb26ae88 dmaengine: virt-dma: don't always free descriptor upon completion
This patch attempts to enhance the case of a transfer submitted multiple
times, and where the cost of creating the descriptors chain is not
negligible.

This happens with big video buffers (several megabytes, ie. several
thousands of linked descriptors in one scatter-gather list). In these
cases, a video driver would want to do :
 - tx = dmaengine_prep_slave_sg()
 - dma_engine_submit(tx);
 - dma_async_issue_pending()
 - wait for video completion
 - read video data (or not, skipping a frame is also possible)
 - dma_engine_submit(tx)
   => here, the descriptors chain recalculation will take time
   => the dma coherent allocation over and over might create holes in
      the dma pool, which is counter-productive.
 - dma_async_issue_pending()
 - etc ...

In order to cope with this case, virt-dma is modified to prevent freeing
the descriptors upon completion if DMA_CTRL_REUSE flag is set in the
transfer.

This patch is a respin of the former DMA_CTRL_ACK approach, which was
reverted due to a regression in audio drivers.

Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-11-16 08:32:16 +05:30
Lars-Peter Clausen 2ed086296e dmaengine: virt-dma: Add synchronization helper function
Add a synchronize helper function for the virt-dma library. The function
makes sure that any scheduled descriptor complete callbacks have finished
running before the function returns.

This needs to be called by drivers using virt-dma in their
device_synchronize() callback. Depending on the driver additional
operations might be necessary in addition to calling vchan_synchronize() to
ensure proper synchronization.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-11-16 08:28:52 +05:30
Lars-Peter Clausen 28ca3e8556 dmaengine: virt-dma: Fix kernel-doc annotations
In kernel-doc annotations parameters need to start with a @ for them to be
properly recognized. Add those where missing for virt-dma.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-10-27 11:12:09 +09:00
Jun Nie 8c8fe97b2b Revert "dmaengine: virt-dma: don't always free descriptor upon completion"
This reverts commit b9855f03d5.
The patch break existing DMA usage case. For example, audio SOC
dmaengine never release channel and cause virt-dma to cache too
much memory in descriptor to exhaust system memory.

Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-07-31 20:33:43 +05:30
Robert Jarzmik b9855f03d5 dmaengine: virt-dma: don't always free descriptor upon completion
This patch attempts to enhance the case of a transfer submitted multiple
times, and where the cost of creating the descriptors chain is not
negligible.

This happens with big video buffers (several megabytes, ie. several
thousands of linked descriptors in one scatter-gather list). In these
cases, a video driver would want to do :
 - tx = dmaengine_prep_slave_sg()
 - dma_engine_submit(tx);
 - dma_async_issue_pending()
 - wait for video completion
 - read video data (or not, skipping a frame is also possible)
 - dma_engine_submit(tx)
   => here, the descriptors chain recalculation will take time
   => the dma coherent allocation over and over might create holes in
      the dma pool, which is counter-productive.
 - dma_async_issue_pending()
 - etc ...

In order to cope with this case, virt-dma is modified to prevent freeing
the descriptors upon completion if DMA_CTRL_ACK flag is set in the
transfer.

Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-06-17 22:15:59 +05:30
Jonas Jensen af58652a3e dma: fix vchan_cookie_complete() debug print
vd->tx.cookie is set zero on dma_cookie_complete(),
save to local before printing it.

Signed-off-by: Jonas Jensen <jonas.jensen@gmail.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2014-01-26 17:33:45 +05:30
Russell King 571fa74034 dmaengine: virt-dma: add support for cyclic DMA periodic callbacks
Add support for cyclic DMA's periodic callbacks.  Drivers are expected
to call vchan_cyclic_callback() when a period has completed, which will
schedule the tasklet to make the callback into the driver.

As callbacks are made from tasklet context, it is important to realise
that we don't guarantee a callback for each completed period, but for
N completed periods where N may be greater than one.

Tested-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
2012-07-01 14:15:23 +01:00
Russell King fe045874aa dmaengine: virt-dma: vchan_find_desc()
Add a function to find a descriptor within the depths of the
virtualized DMA channel support.  Needed for tx_status functionality.

Tested-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
2012-07-01 14:15:22 +01:00
Russell King 50437bff7f dmaengine: split out virtual channel DMA support from sa11x0 driver
Split the virtual slave channel DMA support from the sa11x0 driver so
this code can be shared with other slave DMA engine drivers.

Acked-by: Linus Walleij <linus.walleij@linaro.org>
Tested-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
2012-07-01 14:15:21 +01:00