[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Mon, 30 Apr 2007 11:09:44 +0300 (EEST)
From: "Ilpo Järvinen" <ilpo.jarvinen@...sinki.fi>
To: David Miller <davem@...emloft.net>
cc: Netdev <netdev@...r.kernel.org>
Subject: Re: [PATCH] [RFC] [TCP]: Catch skb with S+L bugs earlier
On Mon, 30 Apr 2007, David Miller wrote:
> From: "Ilpo_Järvinen" <ilpo.jarvinen@...sinki.fi>
> Date: Mon, 30 Apr 2007 10:49:21 +0300 (EEST)
>
> > On Mon, 30 Apr 2007, Ilpo Järvinen wrote:
> >
> > > On Mon, 30 Apr 2007, David Miller wrote:
> > >
> > > > From: "Ilpo_Järvinen" <ilpo.jarvinen@...sinki.fi>
> > > > Date: Wed, 28 Mar 2007 17:02:47 +0300 (EEST)
> > > >
> > > > > SACKED_ACKED and LOST are mutually exclusive, thus this
> > > > > condition is bug with SACK (IMHO). NewReno, however, could get
> >
> > [...snip...]
> >
> > > > I've applied this, thanks for your patience.
> >
> > ...heh... I was getting a bit unsure whether you still had them...
> >
> > > I think I sent an updated version later (hopefully I reach you before
> > > you push these out :-)), which made the BUG_ON unconditional (I used it
> > > instead of BUG_TRAP as it seems to be generic machinery for handling
> > > these).
> >
> > Here:
> > http://marc.info/?l=linux-netdev&m=117648826715609&w=2
>
> Thanks, I'll swap in that version of the patch.
Hmm, just looked a bit more, is the left_out really that necessary as a
separate variable? It's always possible to calculate it in the places
where it is needed and users are quite few after all (basically the
packets_in_flight callers)... ...it just would save some bytes in
tcp_sock... ;-)
--
i.
Powered by blists - more mailing lists