Re: SCSI breakage on non-cache coherent architectures

[David Miller <davem@davemloft.net>]

From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Date: Tue, 20 Nov 2007 11:34:24 +1100

> Do you still think we should introduce this __dma_cacheline_aligned ? Do
> you see other cases of drivers where it would be useful ? It tend to
> agree with your earlier statement that drivers doing that are broken and
> should be using a separate allocator for DMA'ble objects (in fact, on
> non-cache coherent archs, kmalloc is just fine).

I don't care either way.

If we say that the DMA api works on "DMA cacheline boundaries"
(and in reality it does) we do need to either:

1) Require that entire buffers are commited by call sites,
   and thus "embedding" DMA'd within non-DMA stuff isn't allowed

2) Add the __dma_cacheline_aligned tag.

But note that with #2 it could get quite ugly because the
alignment and size both have a minimum that needs to be
enforced, not just the alignment alone.  So either:

struct foo {
	unsigned int other_unrelated_stuff;

	struct object dma_thing __dma_cacheline_aligned;

	unsigned int more_nondma_stuff __dma_cacheline_aligned;
};

or:

struct foo {
	unsigned int other_unrelated_stuff;

	union {
		struct object dma_thing __dma_cacheline_aligned;
		char __pad[(sizeof(object) + DMA_CACHELINE_SIZE &
			   ~DMA_CACHELINE_SIZE)];
	} u;

	unsigned int more_nondma_stuff __dma_cacheline_aligned;
};

I hope you see what I'm trying to say.