[PATCH v2] dmaengine: add CSR SiRFprimaII DMAC driver

[Vinod Koul]

On Mon, 2011-09-19 at 19:25 +0800, Barry Song wrote:
> 2011/9/19 Vinod Koul <vinod.koul at intel.com>:
> > On Mon, 2011-09-19 at 09:56 +0000, Barry Song wrote:
> >> > > > Not sure why you support this, there seem to be no DMA_SLAVE
> >> support in
> >> > > > this version ate least
> >> > >
> >> > > Not. I support dma_slave. But I have no prep_slave_sg function
> >> since I can
> >> > use the gen xfer to replace it.
> >> > Yes thats okay...
> >> >
> >> > Then I have questions on genxfer function...
> >> > where are you copying either src or dstn_start address, you seem to
> >> > completely ignore them?
> >>
> >> Since I only support memory->device or device ->memory, and channel
> >> number is fixed to every device. Then I actually don't care device
> >> address at all. Either src or dst is fixed to the device's address.
> > peripheral address can be fixed, not the memory, where do you copy the
> > memory address?
> 
> +static int sirfsoc_dma_slave_config(struct sirfsoc_dma_chan *schan,
> +       struct dma_slave_config *config)
> +{
> +       u32 addr, direction;
> +       unsigned long flags;
> +
> +       switch (config->direction) {
> +       case DMA_FROM_DEVICE:
> +               direction = 0;
> +               addr = config->dst_addr;
> +               break;
> +
> +       case DMA_TO_DEVICE:
> +               direction = 1;
> +               addr = config->src_addr;
> +               break;
> +
> +       default:
> +               return -EINVAL;
> +       }
> +  ...
This is wrong. You are treating addresses passed in dma_slave_config as
memory address. These should be treating this as peripheral address
(destination in DMA_TO_DEVICE....) and take the memory address from your
prepare function.

-- 
~Vinod