Re: [PATCH] driver core: platform: Allow using a dedicated dma_mask for platform_device

[Christoph Hellwig]

I'd much rather bite the bullet and make dev->dma_mask a scalar
instead of a pointer.  The pointer causes way to much boiler plate code,
and the semantics are way to subtile.  Below is a POV patch that
compiles and boots with my usual x86 test config, and at least compiles
with the arm and pmac32 defconfigs.  It probably breaks just about
everything else, but should give us an idea what is involve in the
switch:

---