Date:	Wed, 18 May 2011 09:45:40 -0700
From:	Shirley Ma <mashirle@...ibm.com>
To:	"Michael S. Tsirkin" <mst@...hat.com>
Cc:	Michał Mirosław <mirqus@...il.com>,
	Ben Hutchings <bhutchings@...arflare.com>,
	David Miller <davem@...emloft.net>,
	Eric Dumazet <eric.dumazet@...il.com>,
	Avi Kivity <avi@...hat.com>, Arnd Bergmann <arnd@...db.de>,
	netdev@...r.kernel.org, kvm@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH V5 2/6 net-next] netdevice.h: Add zero-copy flag in
 netdevice

On Wed, 2011-05-18 at 19:36 +0300, Michael S. Tsirkin wrote:
> On Wed, May 18, 2011 at 09:07:37AM -0700, Shirley Ma wrote:
> > On Wed, 2011-05-18 at 18:47 +0300, Michael S. Tsirkin wrote:
> > > On Wed, May 18, 2011 at 07:38:27AM -0700, Shirley Ma wrote:
> > > > On Wed, 2011-05-18 at 13:40 +0200, Michał Mirosław wrote:
> > > > > >> >> Not more other restrictions, skb clone is OK.
> > > pskb_expand_head()
> > > > > looks
> > > > > >> >> OK to me from code review.
> > > > > >> > Hmm. pskb_expand_head calls skb_release_data while
> keeping
> > > > > >> > references to pages. How is that ok? What do I miss?
> > > > > >> It's making copy of the skb_shinfo earlier, so the pages
> > > refcount
> > > > > >> stays the same.
> > > > > > Exactly. But the callback is invoked so the guest thinks
> it's ok
> > > to
> > > > > > change this memory. If it does a corrupted packet will be
> sent
> > > out.
> > > > > 
> > > > > Hmm. I tool a quick look at skb_clone(), and it looks like
> this
> > > > > sequence will break this scheme:
> > > > > 
> > > > > skb2 = skb_clone(skb...);
> > > > > kfree_skb(skb) or pskb_expand_head(skb);  /* callback called
> */
> > > > > [use skb2, pages still referenced]
> > > > > kfree_skb(skb); /* callback called again */
> > > > > 
> > > > > This sequence is common in bridge, might be in other places.
> > > > > 
> > > > > Maybe this ubuf thing should just track clones? This will make
> it
> > > work
> > > > > on all devices then.
> > > > 
> > > > The callback was only invoked when last reference of skb was
> gone.
> > > > skb_clone does increase skb refcnt. I tested tcpdump on lower
> > > device, it
> > > > worked.
> > > 
> > > Right, it will normally work, but two issues I think you miss:
> > > 1. malicious guest can change the memory between when it is sent
> out
> > > by
> > >    device and consumed by tcpdump, so you will see different
> things
> > >    (not sure how important this is).
> > > 2. if tcpdump stops consuming stuff from the packet socket (it's
> > >    userspace, can't be trusted) then we won't get a callback for
> > >    page potentially forever, guest networking will get blocked
> etc.
> > > > For the sequence of:
> > > > 
> > > > skb_clone  -> last refcnt + 1
> > > > kfree_skb() or pskb_expand_head -> callback not called
> > > > kfree_skb() -> callback called
> > > > 
> > > > I will check page refcount to see whether it's balanced.
> > > > 
> > > > Thanks
> > > > shirley
> > > 
> > > 
> > > pskb_expand_head is a problem anyway I think as it
> > > can hang on to pages after it calls release_data.
> > > Then guest will modify these pages and you get trash there. 
> > 
> > This can be avoid by allowing pskb_expand_head in fastpath only, I
> > think. But not sure whether tcpdump can still work with this.
> > 
> > Thanks
> > Shirley
> 
> Yes, I agree.  I think for tcpdump, we really need to copy the data
> anyway, to avoid guest changing it in between.  So we do that and then
> use the copy everywhere, release the old one. Hmm? 

Yes. Old one use zerocopy, new one use copy data.

Thanks
Shirley

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Powered by blists - more mailing lists