[v9,1/6] dmaengine: Add API function dmaengine_prep_peripheral_dma_vec()

This function can be used to initiate a scatter-gather DMA transfer,
where the address and size of each segment is located in one entry of
the dma_vec array.

The major difference with dmaengine_prep_slave_sg() is that it supports
specifying the lengths of each DMA transfer; as trying to override the
length of the transfer with dmaengine_prep_slave_sg() is a very tedious
process. The introduction of a new API function is also justified by the
fact that scatterlists are on their way out.

Note that dmaengine_prep_interleaved_dma() is not helpful either in that
case, as it assumes that the address of each segment will be higher than
the one of the previous segment, which we just cannot guarantee in case
of a scatter-gather transfer.

Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Signed-off-by: Nuno Sa <nuno.sa@analog.com>

---
v3: New patch

v5: Replace with function dmaengine_prep_slave_dma_vec(), and struct
    'dma_vec'.
    Note that at some point we will need to support cyclic transfers
    using dmaengine_prep_slave_dma_vec(). Maybe with a new "flags"
    parameter to the function?

v7:
  - Renamed *device_prep_slave_dma_vec() -> device_prep_peripheral_dma_vec();
  - Added a new flag parameter to the function as agreed between Paul
    and Vinod. I renamed the first parameter to prep_flags as it's supposed to
    be used (I think) with enum dma_ctrl_flags. I'm not really sure how that API
    can grow but I was thinking in just having a bool cyclic parameter (as the
    first intention of the flags is to support cyclic transfers) but ended up
    "respecting" the previously agreed approach.
---
 include/linux/dmaengine.h | 27 +++++++++++++++++++++++++++
 1 file changed, 27 insertions(+)

On 10-03-24, 13:48, Paul Cercueil wrote:
> This function can be used to initiate a scatter-gather DMA transfer,
> where the address and size of each segment is located in one entry of
> the dma_vec array.
> 
> The major difference with dmaengine_prep_slave_sg() is that it supports
> specifying the lengths of each DMA transfer; as trying to override the
> length of the transfer with dmaengine_prep_slave_sg() is a very tedious
> process. The introduction of a new API function is also justified by the
> fact that scatterlists are on their way out.
> 
> Note that dmaengine_prep_interleaved_dma() is not helpful either in that
> case, as it assumes that the address of each segment will be higher than
> the one of the previous segment, which we just cannot guarantee in case
> of a scatter-gather transfer.
> 
> Signed-off-by: Paul Cercueil <paul@crapouillou.net>
> Signed-off-by: Nuno Sa <nuno.sa@analog.com>
> 
> ---
> v3: New patch
> 
> v5: Replace with function dmaengine_prep_slave_dma_vec(), and struct
>     'dma_vec'.
>     Note that at some point we will need to support cyclic transfers
>     using dmaengine_prep_slave_dma_vec(). Maybe with a new "flags"
>     parameter to the function?
> 
> v7:
>   - Renamed *device_prep_slave_dma_vec() -> device_prep_peripheral_dma_vec();
>   - Added a new flag parameter to the function as agreed between Paul
>     and Vinod. I renamed the first parameter to prep_flags as it's supposed to
>     be used (I think) with enum dma_ctrl_flags. I'm not really sure how that API
>     can grow but I was thinking in just having a bool cyclic parameter (as the
>     first intention of the flags is to support cyclic transfers) but ended up
>     "respecting" the previously agreed approach.
> ---
>  include/linux/dmaengine.h | 27 +++++++++++++++++++++++++++
>  1 file changed, 27 insertions(+)
> 
> diff --git a/include/linux/dmaengine.h b/include/linux/dmaengine.h
> index 752dbde4cec1..856df8cd9a4e 100644
> --- a/include/linux/dmaengine.h
> +++ b/include/linux/dmaengine.h
> @@ -160,6 +160,16 @@ struct dma_interleaved_template {
>  	struct data_chunk sgl[];
>  };
>  
> +/**
> + * struct dma_vec - DMA vector
> + * @addr: Bus address of the start of the vector
> + * @len: Length in bytes of the DMA vector
> + */
> +struct dma_vec {
> +	dma_addr_t addr;
> +	size_t len;
> +};
> +
>  /**
>   * enum dma_ctrl_flags - DMA flags to augment operation preparation,
>   *  control completion, and communicate status.
> @@ -910,6 +920,10 @@ struct dma_device {
>  	struct dma_async_tx_descriptor *(*device_prep_dma_interrupt)(
>  		struct dma_chan *chan, unsigned long flags);
>  
> +	struct dma_async_tx_descriptor *(*device_prep_peripheral_dma_vec)(
> +		struct dma_chan *chan, const struct dma_vec *vecs,
> +		size_t nents, enum dma_transfer_direction direction,
> +		unsigned long prep_flags, unsigned long flags);
>  	struct dma_async_tx_descriptor *(*device_prep_slave_sg)(
>  		struct dma_chan *chan, struct scatterlist *sgl,
>  		unsigned int sg_len, enum dma_transfer_direction direction,
> @@ -973,6 +987,19 @@ static inline struct dma_async_tx_descriptor *dmaengine_prep_slave_single(
>  						  dir, flags, NULL);
>  }
>  
> +static inline struct dma_async_tx_descriptor *dmaengine_prep_peripheral_dma_vec(
> +	struct dma_chan *chan, const struct dma_vec *vecs, size_t nents,
> +	enum dma_transfer_direction dir, unsigned long prep_flags,
> +	unsigned long flags)
> +{
> +	if (!chan || !chan->device || !chan->device->device_prep_peripheral_dma_vec)
> +		return NULL;
> +
> +	return chan->device->device_prep_peripheral_dma_vec(chan, vecs, nents,
> +							    dir, prep_flags,
> +							    flags);
> +}

API looks good to me, thanks
Few nits though:
- Can we add kernel-doc for this new API please
- Also update the documentation adding this new api
- Lastly, we seem to have two flags, I know you have added a comment but
  I dont seem to recall the discussion (looked at old threads for clue
  as well), can you please remind me why we need both? And in your case,
  what is the intended usage of these flags, i would prefer single
  clean one...

Hi Vinod,

Le jeudi 28 mars 2024 à 11:53 +0530, Vinod Koul a écrit :
> On 10-03-24, 13:48, Paul Cercueil wrote:
> > This function can be used to initiate a scatter-gather DMA
> > transfer,
> > where the address and size of each segment is located in one entry
> > of
> > the dma_vec array.
> > 
> > The major difference with dmaengine_prep_slave_sg() is that it
> > supports
> > specifying the lengths of each DMA transfer; as trying to override
> > the
> > length of the transfer with dmaengine_prep_slave_sg() is a very
> > tedious
> > process. The introduction of a new API function is also justified
> > by the
> > fact that scatterlists are on their way out.
> > 
> > Note that dmaengine_prep_interleaved_dma() is not helpful either in
> > that
> > case, as it assumes that the address of each segment will be higher
> > than
> > the one of the previous segment, which we just cannot guarantee in
> > case
> > of a scatter-gather transfer.
> > 
> > Signed-off-by: Paul Cercueil <paul@crapouillou.net>
> > Signed-off-by: Nuno Sa <nuno.sa@analog.com>
> > 
> > ---
> > v3: New patch
> > 
> > v5: Replace with function dmaengine_prep_slave_dma_vec(), and
> > struct
> >     'dma_vec'.
> >     Note that at some point we will need to support cyclic
> > transfers
> >     using dmaengine_prep_slave_dma_vec(). Maybe with a new "flags"
> >     parameter to the function?
> > 
> > v7:
> >   - Renamed *device_prep_slave_dma_vec() ->
> > device_prep_peripheral_dma_vec();
> >   - Added a new flag parameter to the function as agreed between
> > Paul
> >     and Vinod. I renamed the first parameter to prep_flags as it's
> > supposed to
> >     be used (I think) with enum dma_ctrl_flags. I'm not really sure
> > how that API
> >     can grow but I was thinking in just having a bool cyclic
> > parameter (as the
> >     first intention of the flags is to support cyclic transfers)
> > but ended up
> >     "respecting" the previously agreed approach.
> > ---
> >  include/linux/dmaengine.h | 27 +++++++++++++++++++++++++++
> >  1 file changed, 27 insertions(+)
> > 
> > diff --git a/include/linux/dmaengine.h b/include/linux/dmaengine.h
> > index 752dbde4cec1..856df8cd9a4e 100644
> > --- a/include/linux/dmaengine.h
> > +++ b/include/linux/dmaengine.h
> > @@ -160,6 +160,16 @@ struct dma_interleaved_template {
> >  	struct data_chunk sgl[];
> >  };
> >  
> > +/**
> > + * struct dma_vec - DMA vector
> > + * @addr: Bus address of the start of the vector
> > + * @len: Length in bytes of the DMA vector
> > + */
> > +struct dma_vec {
> > +	dma_addr_t addr;
> > +	size_t len;
> > +};
> > +
> >  /**
> >   * enum dma_ctrl_flags - DMA flags to augment operation
> > preparation,
> >   *  control completion, and communicate status.
> > @@ -910,6 +920,10 @@ struct dma_device {
> >  	struct dma_async_tx_descriptor
> > *(*device_prep_dma_interrupt)(
> >  		struct dma_chan *chan, unsigned long flags);
> >  
> > +	struct dma_async_tx_descriptor
> > *(*device_prep_peripheral_dma_vec)(
> > +		struct dma_chan *chan, const struct dma_vec *vecs,
> > +		size_t nents, enum dma_transfer_direction
> > direction,
> > +		unsigned long prep_flags, unsigned long flags);
> >  	struct dma_async_tx_descriptor *(*device_prep_slave_sg)(
> >  		struct dma_chan *chan, struct scatterlist *sgl,
> >  		unsigned int sg_len, enum dma_transfer_direction
> > direction,
> > @@ -973,6 +987,19 @@ static inline struct dma_async_tx_descriptor
> > *dmaengine_prep_slave_single(
> >  						  dir, flags,
> > NULL);
> >  }
> >  
> > +static inline struct dma_async_tx_descriptor
> > *dmaengine_prep_peripheral_dma_vec(
> > +	struct dma_chan *chan, const struct dma_vec *vecs, size_t
> > nents,
> > +	enum dma_transfer_direction dir, unsigned long prep_flags,
> > +	unsigned long flags)
> > +{
> > +	if (!chan || !chan->device || !chan->device-
> > >device_prep_peripheral_dma_vec)
> > +		return NULL;
> > +
> > +	return chan->device->device_prep_peripheral_dma_vec(chan,
> > vecs, nents,
> > +							    dir,
> > prep_flags,
> > +							   
> > flags);
> > +}
> 
> API looks good to me, thanks
> Few nits though:
> - Can we add kernel-doc for this new API please
> - Also update the documentation adding this new api
> - Lastly, we seem to have two flags, I know you have added a comment
> but
>   I dont seem to recall the discussion (looked at old threads for
> clue
>   as well), can you please remind me why we need both? And in your
> case,
>   what is the intended usage of these flags, i would prefer single
>   clean one...
> 

The "prep_flags" is a mask of "enum dma_ctrl_flags".

The second "flags" was supposed to be specific to this function, and
was to future-proof the API as we eventually want to have a "cyclic"
flag, which would emulate a cyclic transfer by linking the SG hardware
descriptors accordingly.

However - I think we can already do that with DMA_PREP_REPEAT and
DMA_PREP_LOAD_EOT, right? So we can probably drop the second "flags".

Cheers,
-Paul