Re: [PATCH 4/4] virtio_ring: unify direct/indirect code paths.

[Michael S. Tsirkin]

On Thu, May 29, 2014 at 08:35:58PM +0930, Rusty Russell wrote:
> Peter Zijlstra <peterz@xxxxxxxxxxxxx> writes:
> > On Thu, May 29, 2014 at 04:56:45PM +0930, Rusty Russell wrote:
> >> Before:
> >> 	gcc 4.8.2: virtio_blk: stack used = 392
> >> 	gcc 4.6.4: virtio_blk: stack used = 480
> >> 
> >> After:
> >> 	gcc 4.8.2: virtio_blk: stack used = 408
> >> 	gcc 4.6.4: virtio_blk: stack used = 432
> >
> > Is it worth it to make the good compiler worse? People are going to use
> > the newer GCC more as time goes on anyhow.
> 
> No, but it's only 16 bytes of stack loss for a simplicity win:
> 
>  virtio_ring.c |  120 +++++++++++++++++++++-------------------------------------
>  1 file changed, 45 insertions(+), 75 deletions(-)
> 
> Cheers,
> Rusty.

I'm concerned that we are doing an extra descriptor walk now though.
And desc == &vq.desc at the end is kind of ugly too.

How about
		if (indirect)
                        vq->vring.desc[i].next = i + 1;
		else
                        i = vq->vring.desc[i].next;

or something like this?

--
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/