Re: hw_random fixes

[Matt Mackall <mpm@selenic.com>]

[cc:ing to linux-kernel, finally]

On Wed, 2009-11-25 at 10:16 +0000, Ian Molton wrote:
> Rusty Russell wrote:
> 
> > Make that:
> > 
> > 	ssize_t (*get_rng_data)(void *buf, size_t max, bool wait);
> > 
> > Then, if driver supplies that hook, use it exclusively.  Otherwise, use old
> > ones.  We can convert them gradually that way.
> 
> This doesn't quite solve things neatly, because it means one of:
> 
> 1) The core has to wait until there is nothing left before requesting
> more data, because it doesnt know the alignment requirements of the driver.

Hmm, this seems to imply you'd be calling get_rng_data multiple times
with different offsets into buf to accumulate data. I think that's more
complex than is needed. Just use buf as a nicely aligned scratch buffer
and empty it completely into the final output buffer before the next
driver request. If you end up with 1 byte of data hanging around until
the next read, that's not a problem - the next read might be 5 bytes.

You'll probably want to use cacheline alignment on buf to make Via
Padlock happy, if anything needs larger alignment (ie page) it should
handle it internally.

-- 
http://selenic.com : development and support for Mercurial and Linux