Re: [PATCH] dmaengine: add dmanegine slave map api's

[Russell King - ARM Linux <linux@arm.linux.org.uk>]

On Mon, Sep 17, 2012 at 03:20:29PM +0530, Vinod Koul wrote:
> On Mon, 2012-09-17 at 09:36 +0100, Russell King - ARM Linux wrote:
> > > > I'm not saying take the slave_id out of the map.  I'm saying, let the
> > > > DMA engine driver itself figure out what dma_chan to return. 
> > > But wont that assume the dma controller knows which channel to allocate.
> > > And how would it know this information? This can be problematic for hard
> > > wired muxes, but can be easily done for controller which have
> > > programmable mux.
> > 
> > Well, as I have already said, at the moment you're returning the _first_
> > _free_ _channel_ on a DMA device, which almost certainly will always be
> > the wrong one. 
> Yes I overlooked, the continue is wrong. It needs to move to next
> available channel. I have fixed it.
> 
> Now on the question if we should allow dmaengine to select channel or
> let dma engine driver do that, I don't see how that helps for hard wired
> muxes where dma engine driver doesn't know anything of mapping.
> 
> For your OMAP dma this wont matter as I think you have a programmable
> mux so you maybe able to use any channel with rightly programmed mux,
> right?

Except that we expose one 'channel' per mux setting, so as far as DMA
engine goes, the mux number _is_ the channel number - which is the same
approach taken by the PL08x and sa11x0 DMA engine drivers.  It is the
only sane approach to dealing with N hardware channels vs >>N clients.