svn commit: r240427 - head/sys/dev/virtio

[Bryan Venteicher]

Hi,

----- Original Message -----
> From: "John Baldwin" <jhb at freebsd.org>
> To: "Peter Grehan" <grehan at freebsd.org>
> Cc: svn-src-head at freebsd.org, svn-src-all at freebsd.org, src-committers at freebsd.org
> Sent: Thursday, September 13, 2012 7:59:38 AM
> Subject: Re: svn commit: r240427 - head/sys/dev/virtio
> 
> On Wednesday, September 12, 2012 8:36:47 pm Peter Grehan wrote:
> > Author: grehan
> > Date: Thu Sep 13 00:36:46 2012
> > New Revision: 240427
> > URL: http://svn.freebsd.org/changeset/base/240427
> > 
> > Log:
> >   Relax requirement of certain mb()s
> >   
> >   Submitted by:	Bryan Venteicher bryanv at daemoninthecloset org
> > 
> > Modified:
> >   head/sys/dev/virtio/virtqueue.c
> > 
> > Modified: head/sys/dev/virtio/virtqueue.c
> > 
> ==============================================================================
> > --- head/sys/dev/virtio/virtqueue.c	Wed Sep 12 22:54:11 2012
> > 	(r240426)
> > +++ head/sys/dev/virtio/virtqueue.c	Thu Sep 13 00:36:46 2012
> > 	(r240427)
> > @@ -525,7 +525,7 @@ virtqueue_dequeue(struct virtqueue *vq,
> >  	used_idx = vq->vq_used_cons_idx++ & (vq->vq_nentries - 1);
> >  	uep = &vq->vq_ring.used->ring[used_idx];
> >  
> > -	mb();
> > +	rmb();
> >  	desc_idx = (uint16_t) uep->id;
> >  	if (len != NULL)
> >  		*len = uep->len;
> > @@ -623,7 +623,7 @@ vq_ring_update_avail(struct virtqueue *v
> >  	avail_idx = vq->vq_ring.avail->idx & (vq->vq_nentries - 1);
> >  	vq->vq_ring.avail->ring[avail_idx] = desc_idx;
> >  
> > -	mb();
> > +	wmb();
> >  	vq->vq_ring.avail->idx++;
> >  
> >  	/* Keep pending count until virtqueue_notify(). */
> 
> Would it be possible to use atomic_load/store() instead of direct
> memory barriers?  For example:
> 

I've been sitting on a (lightly tested) patch [1] for awhile that
does just that, but am not very happy with it. A lot of the fields
are 16-bit, which not all architectures have atomic(9) support for. 
And I think the atomic(9) behavior on UP kernels does not provide
the same guarantees as on an SMP kernel (could have an UP kernel
on an SMP host). 

I also found myself wanting an atomic_load_rel_*() type function.

> 	desc_idx = (uint16_t)atomic_load_acq_int(&uep->id);
> 
> and
> 
> 	atomic_store_rel_int(&vq->vq_ring.avail->ring[avail_idx], desc_idx);

avail->ring is an array of uint16's so I'm a bit leery of using a
(presumably) 32bit op on it. 

[1] http://www.daemoninthecloset.org/~bryanv/patches/freebsd/vq_atomic.patch

> 
> --
> John Baldwin
> _______________________________________________
> svn-src-head at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/svn-src-head
> To unsubscribe, send any mail to
> "svn-src-head-unsubscribe at freebsd.org"
>