On 11/19/2013 08:33 PM, Lee Jones wrote:
On Tue, 19 Nov 2013, Mark Brown wrote:
On Tue, Nov 19, 2013 at 11:07:47AM +0000, Lee Jones wrote:
require slightly different flags to inform the core that we 'are' booting with DT.
Is there some situation when we would want to say we're booting from DT when we aren't? Just wondering about the quotes.
The quotes do 'not' mean anything special. :)
+static const struct snd_dmaengine_pcm_config ux500_dmaengine_of_pcm_config = {
- .pcm_hardware = &ux500_pcm_hw,
- .prealloc_buffer_size = 128 * 1024,
You shouldn't need to set this explicitly, the generic code should be able to pick a number for you - if you do need this number please explain why the number was chosen in the comments (or fix the core to guess better). At the minute the core just makes up a number too but at least then it's a consistent random number between platforms.
Can you also get away without the pcm_hardware - the core should also have support for discovering this by querying the DMA controller?
Despite the '+'s, I'm not actually adding these parameters, I'm duplicating the pdata version and removing the stuff I 'know' that's not required. I don't know what happens when/if these two parameters are removed. I can add this to my TODO when I rip out platform data support, which will happen when this stuff lands.
I think the patch is fine for now. Once non-DT support has been removed for ux500 we should be able to remove the whole ux500_pcm.c file (Assuming that the ux500 DMA engine driver gains dma_slave_caps support).
- Lars