Re: [PATCH 02/20] dma-mapping: provide a generic dma-noncoherent implementation

From: Alexey Brodkin
Date: Fri May 18 2018 - 09:17:46 EST


Hi Christoph,

On Fri, 2018-05-18 at 15:27 +0200, hch@xxxxxx wrote:
> On Fri, May 18, 2018 at 01:03:46PM +0000, Alexey Brodkin wrote:
> > Note mmc_get_dma_dir() is just "data->flags & MMC_DATA_WRITE ? DMA_TO_DEVICE : DMA_FROM_DEVICE".
> > I.e. if we're preparing for sending data dma_noncoherent_map_sg() will have DMA_TO_DEVICE which
> > is quite OK for passing to dma_noncoherent_sync_sg_for_device() but in case of reading we'll have
> > DMA_FROM_DEVICE which we'll pass to dma_noncoherent_sync_sg_for_device() in dma_noncoherent_map_sg().
> >
> > I'd say this is not entirely correct because IMHO arch_sync_dma_for_cpu() is supposed to only be used
> > in case of DMA_FROM_DEVICE and arch_sync_dma_for_device() only in case of DMA_TO_DEVICE.
>
> arc overrides the dir paramter of the dma_sync_single_for_device/
> dma_sync_single_for_cpu calls. My patches dropped that, and I have
> restored that, and audit for the other architectures is pending.

Well at least for me that's a confusion what is a reason to pass direction
to function which purpose is already known.

I'd say that XXX_sync_for_device() doesn't need _variable_ direction as an argument,
otherwise what does that mean if we pass DMA_FROM_DEVICE to that function?

> That being said the existing arc code still looks rather odd as it
> didn't do the same thing for the scatterlist versions of the calls.

That might easily be the case so good we caught that now and it will be fixed :)

> I've thrown in a few patches into my new tree to make the sg versions
> make the normal calls, and to clean up the area a bit.

I'll try your newer series now, thanks!

-Alexey