Re: [Patch] dma_sync_to_device

["David S. Miller" <davem@redhat.com>]

On Wed, 11 Feb 2004 12:23:19 -0700
Matt Porter <mporter@kernel.crashing.org> wrote:

> On Wed, Feb 11, 2004 at 10:30:56AM -0800, David S. Miller wrote:
> > 1) pci_map_single(), device DMA's from the buffer.
> > 
> > 2) pci_dma_sync_single().  Cpu writes some new command or
> >    status flag into the buffer.
> > 
> > 3) pci_dma_sync_to_device_single(), now device is asked to DMA from the buffer
> >    again.
> 
> Actually, not yet.  Is it not possible for MIPS to determine the correct
> cache operation to use if step #3 used a pci_dma_sync_single() with a
> TO_DEVICE direction?  

It should do a writeback of dirty data from the cpu cache, so that the device
may see it after pci_dma_sync_to_device_single() completes.

> I'm guessing that MIPS must have some kind of bridge cache in order to
> require the pci_dma_sync_to_device_single() if I'm starting to follow
> this.

Some platforms do, some don't.  Sparc64 has PCI controller DMA caches but it's
cpu caches are fully coherent, for example.  On such a platform this new interface
is going to be a NOP, but on things like MIPS it will not be.