Add support for providing device to filter_fn mapping so client drivers
can switch to use the dma_request_chan() API.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Add support for providing device to filter_fn mapping so client drivers
can switch to use the dma_request_chan() API.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

The two API function can cover most, if not all current APIs used to
request a channel. With minimal effort dmaengine drivers, platforms and
dmaengine user drivers can be converted to use the two function.

struct dma_chan *dma_request_chan_by_mask(const dma_cap_mask_t *mask);

To request any channel matching with the requested capabilities, can be
used to request channel for memcpy, memset, xor, etc where no hardware
synchronization is needed.

struct dma_chan *dma_request_chan(struct device *dev, const char *name);
To request a slave channel. The dma_request_chan() will try to find the
channel via DT, ACPI or in case if the kernel booted in non DT/ACPI mode
it will use a filter lookup table and retrieves the needed information from
the dma_slave_map provided by the DMA drivers.
This legacy mode needs changes in platform code, in dmaengine drivers and
finally the dmaengine user drivers can be converted:

For each dmaengine driver an array of DMA device, slave and the parameter
for the filter function needs to be added:

static const struct dma_slave_map da830_edma_map[] = {
	{ "davinci-mcasp.0", "rx", EDMA_FILTER_PARAM(0, 0) },
	{ "davinci-mcasp.0", "tx", EDMA_FILTER_PARAM(0, 1) },
	{ "davinci-mcasp.1", "rx", EDMA_FILTER_PARAM(0, 2) },
	{ "davinci-mcasp.1", "tx", EDMA_FILTER_PARAM(0, 3) },
	{ "davinci-mcasp.2", "rx", EDMA_FILTER_PARAM(0, 4) },
	{ "davinci-mcasp.2", "tx", EDMA_FILTER_PARAM(0, 5) },
	{ "spi_davinci.0", "rx", EDMA_FILTER_PARAM(0, 14) },
	{ "spi_davinci.0", "tx", EDMA_FILTER_PARAM(0, 15) },
	{ "da830-mmc.0", "rx", EDMA_FILTER_PARAM(0, 16) },
	{ "da830-mmc.0", "tx", EDMA_FILTER_PARAM(0, 17) },
	{ "spi_davinci.1", "rx", EDMA_FILTER_PARAM(0, 18) },
	{ "spi_davinci.1", "tx", EDMA_FILTER_PARAM(0, 19) },
};

This information is going to be needed by the dmaengine driver, so
modification to the platform_data is needed, and the driver map should be
added to the pdata of the DMA driver:

da8xx_edma0_pdata.slave_map = da830_edma_map;
da8xx_edma0_pdata.slavecnt = ARRAY_SIZE(da830_edma_map);

The DMA driver then needs to configure the needed device -> filter_fn
mapping before it registers with dma_async_device_register() :

ecc->dma_slave.filter_map.map = info->slave_map;
ecc->dma_slave.filter_map.mapcnt = info->slavecnt;
ecc->dma_slave.filter_map.fn = edma_filter_fn;

When neither DT or ACPI lookup is available the dma_request_chan() will
try to match the requester's device name with the filter_map's list of
device names, when a match found it will use the information from the
dma_slave_map to get the channel with the dma_get_channel() internal
function.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Channel matching with private_candidate() is used in two paths, the error
checking is slightly different in them and they are duplicating code also.
Move the code under find_candidate() to provide consistent execution and
going to allow us to reuse this mode of channel lookup later.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

If mask is NULL skip the mask matching against the DMA device capabilities.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Use of the CANCEL bit in mdc_terminate_all creates an
additional 'command done' to appear in the registers (in
addition to an interrupt).

In addition, there is a potential race between
mdc_terminate_all and the irq handler if a transfer
completes at the same time as the terminate all (presently
this results in an inappropriate warning).

To handle these issues, any outstanding 'command done'
events are cleared during mdc_terminate_all and the irq
handler takes no action when there are no new 'command done'
events.
Signed-off-by: Damien.Horsley <Damien.Horsley@imgtec.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Due to changes in device and platform code drivers w/o probe will fail to
load. This means that the devices for eDMA TPTCs are goign to be without
driver and omap hwmod code will turn them off after the kernel finished
loading:
[    3.015900] platform 49800000.tptc: omap_device_late_idle: enabled but no driver.  Idling
[    3.024671] platform 49a00000.tptc: omap_device_late_idle: enabled but no driver.  Idling

This will prevent eDMA to work since the TPTCs are not enabled.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Fixes: 34635b1a ("dmaengine: edma: Add dummy driver skeleton for edma3-tptc")
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

If the "dma-channels" DT property is missing, the dw_dma_parse_dt()
function return NULL, but not before allocating memory for a struct
dw_dma_platform_data through devres.  If the device supports parameter
detection, the probe still succeeds and the allocated memory is not
released until the device is removed.

Fix this by deferring the allocation until after checking the
"dma-channels" property.
Signed-off-by: Mans Rullgard <mans@mansr.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Calling synchronize_irq() right before devm_free_irq() is quite useless. On
one hand the IRQ can easily fire again before devm_free_irq() is entered,
on the other hand devm_free_irq() itself calls synchronize_irq() internally
(in a race condition free way), before any state associated with the IRQ is
freed.

Patch was generated using the following semantic patch:
// <smpl>
@@
expression irq, dev;
@@
-synchronize_irq(irq);
 devm_free_irq(dev, irq, ...);
// </smpl>
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Calling synchronize_irq() right before free_irq() is quite useless. On one
hand the IRQ can easily fire again before free_irq() is entered, on the
other hand free_irq() itself calls synchronize_irq() internally (in a race
condition free way), before any state associated with the IRQ is freed.

Patch was generated using the following semantic patch:
// <smpl>
@@
expression irq;
@@
-synchronize_irq(irq);
 free_irq(irq, ...);
// </smpl>
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Acked-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

This add power management suspend/resume support for the fsl-edma
driver.

eDMA acted as a basic function used by others. What it needs to do
is the two steps below to support power management.

In fsl_edma_suspend_late:
Check whether the DMA chan is idle, if it is not idle disable DMA
request.

In fsl_edma_resume_early:
Enable the eDMA and wait for being used.
Signed-off-by: Yuan Yao <yao.yuan@freescale.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

There is no need to calculate an overall length of the descriptor each time we
call for DMA transfer status. Instead we do this at descriptor allocation stage
and keep the stored length for further usage.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Currently the match DMA controller is done only for lower 32 bits of
address which might be not true on 64-bit platform. Check upper portion
as well.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Since a local variable contains the number of hardware desriptors at the
beginning of idma64_desc_fill() we may use it to index the last descriptor as
well.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Explicitly show in idma64_desc_fill() how we link the hardware
descriptors.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

This tells, for example, IOMMU what the maximum size of a segment
the DMA controller can send.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

There is no need to disable interrupts in the IRQ handler. The driver
guarantess that at one time only one descriptor is active, besides the fact
that each call to the same channel will be serialized in idma64_chan_irq()
handler anyway.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

When setting the channel configuration register, the perid field is not
set to 0 since it is useless for mem2mem transfers. Unfortunately, a
device has 0 as perid. It could cause spurious flags status because
the controller could mix some events from the two channels.
For that reason, use the highest perid value for mem2mem transfers since it
doesn't match the perid of other devices.
Signed-off-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Implementations of dmaengine_synchronize() are allowed to sleep, hence the
function must not be called to from atomic context. Add might_sleep() to
dmaengine_synchronize() to make it easier to detect non-compliant callers.
Suggested-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

This patch fixes an issue that list_for_each_entry() in
usb_dmac_chan_terminate_all() is possible to cause endless loop because
this will move own desc to the desc_freed. So, this driver should use
list_for_each_entry_safe() instead of list_for_each_entry().
Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

When a DMA client driver decides that it is not providing callback for
completion of a transfer (and/or does not set the DMA_PREP_INTERRUPT) but
it will poll the status of the transfer (in case of short memcpy for
example) we will not get interrupt for the completion of the transfer and
will not mark the transaction as done.
Check the channel enable bit in the CCR when the status is queried and if
the channel is no longer active, we call the omap_dma_callback() to handle
the transfer completion.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

The use of tasklet to actually start the DMA transfer slightly decreases the
DMA throughput since it adds small scheduling delay when the transfer is
started. In normal use, even with high I/O load the tasklet would start
one transaction at a time, however running the DMAtest for memcpy on all
available channels will cause the tasklet to start about 15 transfers.
The performance numbers on OMAP4 PandaBoard-es (test_buf_size = 6553):
With the tasklet:
dmatest: dma0chan30-copy: summary 5000 tests, 0 failures 186 iops 593 KB/s (0)
dmatest: dma0chan8-copy0: summary 5000 tests, 0 failures 184 iops 584 KB/s (0)
dmatest: dma0chan13-copy: summary 5000 tests, 0 failures 184 iops 585 KB/s (0)
dmatest: dma0chan12-copy: summary 5000 tests, 0 failures 184 iops 585 KB/s (0)
dmatest: dma0chan7-copy0: summary 5000 tests, 0 failures 183 iops 581 KB/s (0)

With this patch (no tasklet):
dmatest: dma0chan4-copy0: summary 5000 tests, 0 failures 199 iops 644 KB/s (0)
dmatest: dma0chan5-copy0: summary 5000 tests, 0 failures 199 iops 645 KB/s (0)
dmatest: dma0chan6-copy0: summary 5000 tests, 0 failures 199 iops 637 KB/s (0)
dmatest: dma0chan24-copy: summary 5000 tests, 0 failures 199 iops 638 KB/s (0)
dmatest: dma0chan16-copy: summary 5000 tests, 0 failures 199 iops 638 KB/s (0)
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

The for_each_sg() macro's last parameter is inteded to be used as counter.
We can use 'i' instead of 'j' within the loop for indexes.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

During mem copy both src and dst position moves at the same pace. Check the
dst position for progress reporting.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Signed-off-by: Jyri Sarha <jsarha@ti.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

The dca_ops structure is never modified, so declare it as const.

Done with the help of Coccinelle.
Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Acked-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Remove pre-3.0 channel status reads. 3.0 and later chansts register
is 64bit and can be read 64bit. This was clarified with the hardware
architects and since the driver now only support 3.0+ we don't need the
legacy support
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

As this driver provides a mechanism to reuse transfers, declare it in
its probe function.
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

In the current state, the capability of transfer reuse can neither be
set by a slave dmaengine driver, nor used by a client driver, because
the capability is not available to dma_get_slave_caps().

Fix this by adding a way to declare the capability.

Fixes: 27242021 ("dmaengine: Add DMA_CTRL_REUSE")
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

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>

Use the new dmaengine_synchronize() function to make sure that all complete
callbacks have finished running before the runtime data, which is accessed
in the completed callback, is freed.

This fixes a long standing use-after-free race condition that has been
observed on some systems.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Implement the new device_synchronize() callback to allow proper
synchronization when stopping a channel. Since the driver already makes
sure that no new complete callbacks are scheduled after the
device_terminate_all() callback has been called, all left to do in the
device_synchronize() callback is to wait for all currently running complete
callbacks to finish.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

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>

The DMAengine API has a long standing race condition that is inherent to
the API itself. Calling dmaengine_terminate_all() is supposed to stop and
abort any pending or active transfers that have previously been submitted.
Unfortunately it is possible that this operation races against a currently
running (or with some drivers also scheduled) completion callback.

Since the API allows dmaengine_terminate_all() to be called from atomic
context as well as from within a completion callback it is not possible to
synchronize to the execution of the completion callback from within
dmaengine_terminate_all() itself.

This means that a user of the DMAengine API does not know when it is safe
to free resources used in the completion callback, which can result in a
use-after-free race condition.

This patch addresses the issue by introducing an explicit synchronization
primitive to the DMAengine API called dmaengine_synchronize().

The existing dmaengine_terminate_all() is deprecated in favor of
dmaengine_terminate_sync() and dmaengine_terminate_async(). The former
aborts all pending and active transfers and synchronizes to the current
context, meaning it will wait until all running completion callbacks have
finished. This means it is only possible to call this function from
non-atomic context. The later function does not synchronize, but can still
be used in atomic context or from within a complete callback. It has to be
followed up by dmaengine_synchronize() before a client can free the
resources used in a completion callback.

In addition to this the semantics of the device_terminate_all() callback
are slightly relaxed by this patch. It is now OK for a driver to only
schedule the termination of the active transfer, but does not necessarily
have to wait until the DMA controller has completely stopped. The driver
must ensure though that the controller has stopped and no longer accesses
any memory when the device_synchronize() callback returns.

This was in part done since most drivers do not pay attention to this
anyway at the moment and to emphasize that this needs to be done when the
device_synchronize() callback is implemented. But it also helps with
implementing support for devices where stopping the controller can require
operations that may sleep.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>