On Fri, 2012-01-06 at 18:17 +0530, Ravi Kumar V wrote:
> Qualcomm MSM have a feature to pass command configuration and
> control data along with source,destination and length of transfer
> for every transaction, as of now struct scatterlist has no support
> to send private data related to each transaction we added private_data
> variable for supporting this type of archictures.
this looks quite similar to what RIO [1] folks were asking, ie ability
to pass specific parameters for each transaction which are device
specific.
>
> Qualcomm MSM also supports BOX mode of dma, currently as there is no
> API in dmaengine to support this type of dma we added new API.
>
> Change-Id: Ia9ee19f2c253e68b8e5ff254a57478dcc51014ca
> Signed-off-by: Ravi Kumar V <kumarrav@xxxxxxxxxxxxxx>
> ---
> include/asm-generic/scatterlist.h | 1 +
> include/linux/dmaengine.h | 17 ++++++++++++++++-
> 2 files changed, 17 insertions(+), 1 deletions(-)
>
> diff --git a/include/asm-generic/scatterlist.h b/include/asm-generic/scatterlist.h
> index 5de0735..e66dfcb 100644
> --- a/include/asm-generic/scatterlist.h
> +++ b/include/asm-generic/scatterlist.h
> @@ -14,6 +14,7 @@ struct scatterlist {
> #ifdef CONFIG_NEED_SG_DMA_LENGTH
> unsigned int dma_length;
> #endif
> + unsigned long private_data;
what do you plan to pass here. Please keep in mind this is a generic
scatterlist structure, and modifying it for your purposes doesn't seem
to be a great one!
Also why can't you pass this as addition argument in your new "BOX API"?
> };
>
> /*
> diff --git a/include/linux/dmaengine.h b/include/linux/dmaengine.h
> index 75f53f8..ea29e73 100644
> --- a/include/linux/dmaengine.h
> +++ b/include/linux/dmaengine.h
> @@ -72,10 +72,11 @@ enum dma_transaction_type {
> DMA_ASYNC_TX,
> DMA_SLAVE,
> DMA_CYCLIC,
> + DMA_BOX,
> };
>
> /* last transaction type for creation of the capabilities mask */
> -#define DMA_TX_TYPE_END (DMA_CYCLIC + 1)
> +#define DMA_TX_TYPE_END (DMA_BOX + 1)
>
>
> /**
> @@ -404,6 +405,15 @@ struct dma_tx_state {
> u32 residue;
> };
>
> +
> +struct dma_box_list {
> + dma_addr_t dma_row_address;
> + unsigned int dma_row_len;
> + unsigned int dma_row_num;
> + unsigned int dma_row_offset;
> + unsigned long private_data;
> +};
again a private data here?
is this some kind of interleaved pattern?
> +
> /**
> * struct dma_device - info on the entity supplying DMA services
> * @chancnt: how many DMA channels are supported
> @@ -497,6 +507,11 @@ struct dma_device {
> struct dma_async_tx_descriptor *(*device_prep_dma_cyclic)(
> struct dma_chan *chan, dma_addr_t buf_addr, size_t buf_len,
> size_t period_len, enum dma_data_direction direction);
> + struct dma_async_tx_descriptor *(*device_prep_dma_box)(
> + struct dma_chan *chan, struct dma_box_list *dst_box,
> + struct dma_box_list *src_box, unsigned int num_list,
> + unsigned long flags);
still not clean what kind of transfer do you want to do here
> +
> int (*device_control)(struct dma_chan *chan, enum dma_ctrl_cmd cmd,
> unsigned long arg);
>
--
~Vinod
--
To unsubscribe from this list: send the line "unsubscribe linux-arch" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
