Re: [STABLE 2.6.32 PATCH] net: release dst entry while cache-hot for GSO case too

[Eric Dumazet <eric.dumazet@gmail.com>]

Le lundi 11 octobre 2010 à 19:40 +0400, Michael Tokarev a écrit :
> Andrey Vagin wrote:
> > From: Krishna Kumar <krkumar2@in.ibm.com>
> > 
> > commit 068a2de57ddf4f472e32e7af868613c574ad1d88 upstream.
> > 
> > Non-GSO code drops dst entry for performance reasons, but
> > the same is missing for GSO code. Drop dst while cache-hot
> > for GSO case too.
> > 
> > Note: Without this patch the kernel may oops if used bridged veth
> > devices. A bridge set skb->dst = fake_dst_ops, veth transfers this skb
> > to netif_receive_skb...ip_rcv_finish and it calls dst_input(skb), but
> > fake_dst_ops->input = NULL -> Oops
> 
> Hmm.  Isn't this the reason for my mysterious OOPSes (jump to
> NULL) which I concluded are due to stack overflow?  See f.e.
> http://www.spinics.net/lists/netdev/msg142104.html
> 
> This started happening when I updated virtio drivers in a windows
> virtual machne to the ones which supports GSO, and my config
> involves bridging veth devices, and this is where the prob
> actually occurs - when doing guest => virtio => tap => bridge => veth
> route....
> 
> Thanks!

This patch was an optimization, not a bug fix.

If something gets better, then a bug is somewhere else ?