netdev - Re: [PATCH] xfrm: don't segment UFO packets

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 17 Mar 2016 12:01:31 +0100
From:	Steffen Klassert <steffen.klassert@...unet.com>
To:	Jiri Bohac <jbohac@...e.cz>
CC:	Herbert Xu <herbert@...dor.apana.org.au>,
	"David S. Miller" <davem@...emloft.net>, <netdev@...r.kernel.org>
Subject: Re: [PATCH] xfrm: don't segment UFO packets

On Thu, Mar 17, 2016 at 11:49:53AM +0100, Jiri Bohac wrote:
> On Thu, Mar 17, 2016 at 11:24:59AM +0100, Steffen Klassert wrote:
> > > > On Wed, Mar 16, 2016 at 05:00:26PM +0100, Jiri Bohac wrote:
> > > Fixes my broken case. 
> > 
> > Is this IPv4 or IPv6? IPv4 should not create a GSO skb
> > if IPsec is done. It checks for rt->dst.header_len
> > in __ip_append_data() and does a fallback to the
> > standard case if rt->dst.header_len is non zero.
> 
> It's IPv6.
> 
> > In IPv6 this check is missing, so this could be the
> > problem if this is IPv6.
> 
> Doesn't the check do exactly the opposite of what the RFC says?
> The RFC wants ESP to be performed first and fragmentation after
> that. UDPv4 currently seems to be doing the opposite. 

No, __ip_append_data() only prepares the packets for fragmentation
and enqueues them. Then __ip_make_skb() dequeues and builds
one skb with a fraglist. Then the xfrm layer is called, so
esp linearizes (unfortunately) the skb and applies the
transformation. Fragmentation happens after that.