Re: [PATCH 2/5] dmaengine: Add support for custom data mapping

[Vinod Koul <vinod.koul@intel.com>]

On Thu, Jan 19, 2017 at 08:13:17AM -0600, Andy Gross wrote:
> On Thu, Jan 19, 2017 at 10:31:50AM +0530, Vinod Koul wrote:
> 
> <snip>
> 
> > > > >
> > > > >I really think that we need some additional API that allows for the flag
> > > > >munging
> > > > >for the descriptors instead of overriding the prep_slave_sg.  We already
> > > > >needed
> > > > >to change the way the flags are passed anyway.  And instead of building up
> > > > >a
> > > > >special sg list, the API should take a structure that has a 1:1 mapping of
> > > > >the
> > > > >flags to the descriptors.  And you would call this API on your descriptor
> > > > >before
> > > > >issuing it.
> > 
> > Munging wont be a good idea, but for some of the cases current flags can be
> > used, and if need be, we can add additional flags
> 
> Is adding flags a possibility?  I tried to match up BAM flags to ones that made
> sense that were currently defined, but adding a CMD flag would be kind of odd.

Matching flags is a good idea wherever they match, overriding is not :)

> It was kind of a stretch to use the PREP_FENCE for the notify when done flag.

For that, we should use PREP_INTERUPT. DMAengine should assert interrupt
only when this flag is set and continue to next transaction.

> > > > >
> > > > >So build up the sglist.  Call the prep_slave_sg.  You get back a tx
> > > > >descriptor
> > > > >that underneath is a bam descriptor.  Then call the API giving the
> > > > >descriptor
> > > > >and the structure that defines the flags for the descriptors.  Then submit
> > > > >the
> > > > >descriptor.
> > > > >
> > > > >Something like:
> > > > >int qcom_bam_apply_descriptor_flags(struct dma_async_tx_descriptor *tx,
> > > > >				    u16 *flags)
> > > > >{
> > > > >	struct bam_async_desc async_desc = container_of(tx,
> > > > >							struct bam_async_desc,
> > > > >							vd.tx);
> > > > >	int i;
> > > > >
> > > > >	for (i = 0; i < async_desc->num_desc; i++)
> > > > >		async_desc->desc[i].flags = cpu_to_le16(flags[i]);
> > > > >}
> > > > >
> > > > >EXPORT_SYMBOL(qcom_bam_apply_descriptor_flags)
> > 
> > This makes it bam specific and causes issues if we want to use this code in
> > generic libs, but yes this is an option but should be last resort.
> 
> If adding flags is a possibility (which it didn't seem to be in the past), that
> would make things much easier.

Yes if we can describe them generically then yes. So with this and resuing
existing flags without overriding, how many flags do we need..

-- 
~Vinod