1                        DMA Engine API Guide
   2                        ====================
   4                 Vinod Koul <vinod dot koul at>
   6NOTE: For DMA Engine usage in async_tx please see:
   7        Documentation/crypto/async-tx-api.txt
  10Below is a guide to device driver writers on how to use the Slave-DMA API of the
  11DMA Engine. This is applicable only for slave DMA usage only.
  13The slave DMA usage consists of following steps:
  141. Allocate a DMA slave channel
  152. Set slave and controller specific parameters
  163. Get a descriptor for transaction
  174. Submit the transaction
  185. Issue pending requests and wait for callback notification
  201. Allocate a DMA slave channel
  22   Channel allocation is slightly different in the slave DMA context,
  23   client drivers typically need a channel from a particular DMA
  24   controller only and even in some cases a specific channel is desired.
  25   To request a channel dma_request_channel() API is used.
  27   Interface:
  28        struct dma_chan *dma_request_channel(dma_cap_mask_t mask,
  29                        dma_filter_fn filter_fn,
  30                        void *filter_param);
  31   where dma_filter_fn is defined as:
  32        typedef bool (*dma_filter_fn)(struct dma_chan *chan, void *filter_param);
  34   The 'filter_fn' parameter is optional, but highly recommended for
  35   slave and cyclic channels as they typically need to obtain a specific
  36   DMA channel.
  38   When the optional 'filter_fn' parameter is NULL, dma_request_channel()
  39   simply returns the first channel that satisfies the capability mask.
  41   Otherwise, the 'filter_fn' routine will be called once for each free
  42   channel which has a capability in 'mask'.  'filter_fn' is expected to
  43   return 'true' when the desired DMA channel is found.
  45   A channel allocated via this interface is exclusive to the caller,
  46   until dma_release_channel() is called.
  482. Set slave and controller specific parameters
  50   Next step is always to pass some specific information to the DMA
  51   driver.  Most of the generic information which a slave DMA can use
  52   is in struct dma_slave_config.  This allows the clients to specify
  53   DMA direction, DMA addresses, bus widths, DMA burst lengths etc
  54   for the peripheral.
  56   If some DMA controllers have more parameters to be sent then they
  57   should try to embed struct dma_slave_config in their controller
  58   specific structure. That gives flexibility to client to pass more
  59   parameters, if required.
  61   Interface:
  62        int dmaengine_slave_config(struct dma_chan *chan,
  63                                  struct dma_slave_config *config)
  65   Please see the dma_slave_config structure definition in dmaengine.h
  66   for a detailed explanation of the struct members.  Please note
  67   that the 'direction' member will be going away as it duplicates the
  68   direction given in the prepare call.
  703. Get a descriptor for transaction
  72   For slave usage the various modes of slave transfers supported by the
  73   DMA-engine are:
  75   slave_sg     - DMA a list of scatter gather buffers from/to a peripheral
  76   dma_cyclic   - Perform a cyclic DMA operation from/to a peripheral till the
  77                  operation is explicitly stopped.
  78   interleaved_dma - This is common to Slave as well as M2M clients. For slave
  79                 address of devices' fifo could be already known to the driver.
  80                 Various types of operations could be expressed by setting
  81                 appropriate values to the 'dma_interleaved_template' members.
  83   A non-NULL return of this transfer API represents a "descriptor" for
  84   the given transaction.
  86   Interface:
  87        struct dma_async_tx_descriptor *(*chan->device->device_prep_slave_sg)(
  88                struct dma_chan *chan, struct scatterlist *sgl,
  89                unsigned int sg_len, enum dma_data_direction direction,
  90                unsigned long flags);
  92        struct dma_async_tx_descriptor *(*chan->device->device_prep_dma_cyclic)(
  93                struct dma_chan *chan, dma_addr_t buf_addr, size_t buf_len,
  94                size_t period_len, enum dma_data_direction direction);
  96        struct dma_async_tx_descriptor *(*device_prep_interleaved_dma)(
  97                struct dma_chan *chan, struct dma_interleaved_template *xt,
  98                unsigned long flags);
 100   The peripheral driver is expected to have mapped the scatterlist for
 101   the DMA operation prior to calling device_prep_slave_sg, and must
 102   keep the scatterlist mapped until the DMA operation has completed.
 103   The scatterlist must be mapped using the DMA struct device.  So,
 104   normal setup should look like this:
 106        nr_sg = dma_map_sg(chan->device->dev, sgl, sg_len);
 107        if (nr_sg == 0)
 108                /* error */
 110        desc = chan->device->device_prep_slave_sg(chan, sgl, nr_sg,
 111                        direction, flags);
 113   Once a descriptor has been obtained, the callback information can be
 114   added and the descriptor must then be submitted.  Some DMA engine
 115   drivers may hold a spinlock between a successful preparation and
 116   submission so it is important that these two operations are closely
 117   paired.
 119   Note:
 120        Although the async_tx API specifies that completion callback
 121        routines cannot submit any new operations, this is not the
 122        case for slave/cyclic DMA.
 124        For slave DMA, the subsequent transaction may not be available
 125        for submission prior to callback function being invoked, so
 126        slave DMA callbacks are permitted to prepare and submit a new
 127        transaction.
 129        For cyclic DMA, a callback function may wish to terminate the
 130        DMA via dmaengine_terminate_all().
 132        Therefore, it is important that DMA engine drivers drop any
 133        locks before calling the callback function which may cause a
 134        deadlock.
 136        Note that callbacks will always be invoked from the DMA
 137        engines tasklet, never from interrupt context.
 1394. Submit the transaction
 141   Once the descriptor has been prepared and the callback information
 142   added, it must be placed on the DMA engine drivers pending queue.
 144   Interface:
 145        dma_cookie_t dmaengine_submit(struct dma_async_tx_descriptor *desc)
 147   This returns a cookie can be used to check the progress of DMA engine
 148   activity via other DMA engine calls not covered in this document.
 150   dmaengine_submit() will not start the DMA operation, it merely adds
 151   it to the pending queue.  For this, see step 5, dma_async_issue_pending.
 1535. Issue pending DMA requests and wait for callback notification
 155   The transactions in the pending queue can be activated by calling the
 156   issue_pending API. If channel is idle then the first transaction in
 157   queue is started and subsequent ones queued up.
 159   On completion of each DMA operation, the next in queue is started and
 160   a tasklet triggered. The tasklet will then call the client driver
 161   completion callback routine for notification, if set.
 163   Interface:
 164        void dma_async_issue_pending(struct dma_chan *chan);
 166Further APIs:
 1681. int dmaengine_terminate_all(struct dma_chan *chan)
 170   This causes all activity for the DMA channel to be stopped, and may
 171   discard data in the DMA FIFO which hasn't been fully transferred.
 172   No callback functions will be called for any incomplete transfers.
 1742. int dmaengine_pause(struct dma_chan *chan)
 176   This pauses activity on the DMA channel without data loss.
 1783. int dmaengine_resume(struct dma_chan *chan)
 180   Resume a previously paused DMA channel.  It is invalid to resume a
 181   channel which is not currently paused.
 1834. enum dma_status dma_async_is_tx_complete(struct dma_chan *chan,
 184        dma_cookie_t cookie, dma_cookie_t *last, dma_cookie_t *used)
 186   This can be used to check the status of the channel.  Please see
 187   the documentation in include/linux/dmaengine.h for a more complete
 188   description of this API.
 190   This can be used in conjunction with dma_async_is_complete() and
 191   the cookie returned from 'descriptor->submit()' to check for
 192   completion of a specific DMA transaction.
 194   Note:
 195        Not all DMA engine drivers can return reliable information for
 196        a running DMA channel.  It is recommended that DMA engine users
 197        pause or stop (via dmaengine_terminate_all) the channel before
 198        using this API.