[RFC 0/2] fix dma_map_sg not to do barriers for each buffer

[Russell King - ARM Linux]

On Thu, Feb 11, 2010 at 10:45:01AM +0000, Catalin Marinas wrote:
> Alternatively we could use the dsb() macro. I don't think we need more
> than this and we would not (well, not easily) compile ARMv5 and ARMv6 in
> the same kernel.

That doesn't work - ARMv3 and some ARMv4 don't have a 'drain write
buffer' instruction but others do - executing that instruction on
older CPUs which don't have a write buffer causes an illegal
instruction fault.

> The ___dma_single_cpu_to_dev() covers both inner and outer caches but I
> haven't seen it touched by this patch (nor the other you posted). When
> you clean the L1 cache, you need to make sure that there is a barrier
> (DSB) so that it completes before cleaning the L2, otherwise you clean
> the L2 but data keeps coming from L1.
> 
> For the *_sg functions, you either use barrier between L1 and L2 for
> each page or you do the for_each_sg() loop twice, once for L1 and
> another for L2.

Okay, that's a fundamental problem with this approach.  Spanner in the
works kind of thing.  I think that's a problem for Abhijeet's patch
as well - since the same comment appears to apply there too.

Sounds like it needs a totally different approach then.