Re: problems in commit 2d4dc890b5c8 (block: add helpers to run flush_dcache_page() against a bio and a request's pages)

[Ilya Loginov <isloginov@gmail.com>]

On Thu, 10 Dec 2009 12:20:55 -0600
James Bottomley <James.Bottomley@HansenPartnership.com> wrote:

> I'm not religiously opposed to the separation into a flush and a non
> flush case ... although I think if we have to do this, it's equivalent
> to just forcing users to add the flush_kernel_dcache_page() ... but if
> we can do it so that the users don't need to know the details, I think
> the API is much better.

I wrote that flush_kernel_dcache_page() was exist only in parisc(and sh).
There is flush_dcache_page() in other architectures. And I don't think,
that flush_kernel_dcache_page() should be introduced into other
architectures. Actually, I don't know a lot about many architectures.

So, may be. We should have a varian of rq_flush... call for architectures
where ARCH_HAS_FLUSH_KERNEL_DCACHE_PAGE was defined. I wrote this before.
And I continue to think such manner since I've read about ARM.

-- 
Ilya Loginov <isloginov@gmail.com>