Re: [PATCH 0/4] Add generic DMA DT binding support

[Arnd Bergmann <arnd@arndb.de>]

On Monday 28 January 2013, Vinod Koul wrote:
> On Fri, Jan 18, 2013 at 05:03:40PM +0530, Padmavathi Venna wrote:
> > This patch set adds support for generic dma device tree bindings for
> > Samsung platforms and is dependent on the following patches from
> > Vinod Koul next branch
> > 1)of: Add generic device tree DMA helpers
> > 2)dmaengine: add helper function to request a slave DMA channel
> Changes look fairly decent. I need somone with better knowldge of DT to akc this
> before this is applied. Arnd...?

Thanks for the pointer, it seems everthing is coming together today ;-)

The binding looks good, aside from the wording on some of the properties
in there that could be a little clearer.

The method of calling dma_request_channel() from the xlate() function
is a little bit clumsy IMHO, but it does work and is the easiest way
to retrofit generic DT support to this driver, given that it already
provides a global filter function. It's also apparently what Jon had
in mind, and what Matt is doing on AM33XX.

I'll probably propose something similar on dw_dma, since I was just
(this hour, actually) looking at the same problem there. I believe
you had something smarter in mind when we discussed it in San Diego,
but I don't remember what the idea was.  I would need something
like

	struct dma_chan *dma_channel_get(struct dma_device *);

and I assume that there is a reason for why I can't call this
from the xlate() function but instead have to go through
dma_request_channel(), but I don't completely understand that part
of the puzzle.

	Arnd