Re: [PATCH] tcp: splice as many packets as possible at once

[David Miller <davem@davemloft.net>]

From: Willy Tarreau <w@1wt.eu>
Date: Mon, 19 Jan 2009 07:14:20 +0100

> On Sun, Jan 18, 2009 at 07:27:19PM -0800, David Miller wrote:
> > From: Herbert Xu <herbert@gondor.apana.org.au>
> > Date: Mon, 19 Jan 2009 14:08:44 +1100
> > 
> > > One thing to note is that Myricom's driver probably uses page
> > > frags which means that you're not actually triggering the copy.
> 
> So does this mean that the corruption problem should still there for
> such a driver ? I'm asking before testing, because at these speeds,
> validity tests are not that easy ;-)

It ought not to, but it seems that is the case where you
saw the original corruptions, so hmmm...

Actually, I see, the myri10ge driver does put up to
64 bytes of the initial packet into the linear area.
If the IPV4 + TCP headers are less than this, you will
hit the corruption case even with the myri10ge driver.

So everything checks out.

> > And this is also the only reason why jumbo MTU worked :-)
> 
> What should we expect from other drivers with jumbo frames ? Hangs,
> corruption, errors, packet loss ?

Upon recent review I think jumbo frames in such drivers should
actually be fine.