Re: [PATCH 11/11] DIO: optimize cache misses in the submission path

[Andi Kleen]

> I bet we could default to using the smaller block size all the time, and
> still be able to detect when we don't have to do the sub-block zeroing.
> Maybe that would be a good follow-on patch.

It doesn't really matter because it's out of the fast path now.

> > +	/* 
> > +	 * Avoid references to bdev if not absolutely needed to give
> > +	 * the early prefetch in the caller enough time.
> > +	 */
> >  
> > -	if (offset & blocksize_mask) {
> > +	if (unlikely(offset & blocksize_mask)) {
> 
> You can't make this assumption.  Userspace controls what size/alignment
> of blocks to send in.

What assumption do you mean?

The code semantics are identical, just the place where I fetch
the block size is different (unless I missed something of course)


-Andi
-- 
ak@xxxxxxxxxxxxxxx -- Speaking for myself only.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/