netdev - Re: [PATCH] tcp: fix potential corner case issue in segmentation (Was: Re: [PATCH] Do not use TSO/GSO when there is urgent data)

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:	Fri, 21 Nov 2008 16:51:51 +0100
From:	Petr Tesarik <ptesarik@...e.cz>
To:	"Ilpo Järvinen" <ilpo.jarvinen@...sinki.fi>
Cc:	Netdev <netdev@...r.kernel.org>,
	LKML <linux-kernel@...r.kernel.org>,
	"David S. Miller" <davem@...emloft.net>,
	"Jan Šembera" <jsembera@...e.cz>
Subject: Re: [PATCH] tcp: fix potential corner case issue in segmentation (Was: Re: [PATCH] Do not use TSO/GSO when there is urgent data)

Dne Friday 21 of November 2008 14:07:32 Ilpo Järvinen napsal(a):
> On Fri, 21 Nov 2008, Petr Tesarik wrote:
> > This patch fixes http://bugzilla.kernel.org/show_bug.cgi?id=12014
> >
> > Since most (if not all) implementations of TSO and even the in-kernel
> > software GSO do not update the urgent pointer when splitting a large
> > segment, it is necessary to turn off TSO/GSO for all outgoing traffic
> > with the URG pointer set.
>
> Good observation, I totally missed this possibility of T/GSO while
> looking.
>
> > Looking at tcp_current_mss (and the preceding comment) I even think
> > this was the original intention. However, this approach is insufficient,
> > because TSO/GSO is turned off only for newly created frames, not for
> > frames which were already pending at the arrival of a message with
> > MSG_OOB set. These frames were created when TSO/GSO was enabled,
> > so they may be large, and they will have the urgent pointer set
> > in tcp_transmit_skb().
> >
> > With this patch, such large packets will be fragmented again before
> > going to the transmit routine.
>
> I wonder if there's some corner case which still fails to fragment
> in tcp_retransmit_xmit's in skb->len <= cur_mss case if cur_mss
> grew very recently (and therefore skb-len now fits to a single segment).

This shouldn't be a problem, because TSO only applies to packets which are 
larger than MSS, so the problematic case is when cur_mss gets smaller, not 
when it grows. In other words, the original implementation of 
tcp_retransmit_xmit() could never make use of TSO/GSO, anyway...

>[...]
> > Signed-off-by: Petr Tesarik <ptesarik@...e.cz>
> > CC: Jan Sembera <jsembera@...e.cz>
> > CC: Ilpo Jarvinen <ilpo.jarvinen@...sinki.fi>
> >
> > --
> >  tcp_output.c |    7 +++++--
> >  1 file changed, 5 insertions(+), 2 deletions(-)
> >
> > diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
> > --- a/net/ipv4/tcp_output.c
> > +++ b/net/ipv4/tcp_output.c
> > @@ -722,7 +722,8 @@ static void tcp_queue_skb(struct sock *sk, struct
> > sk_buff *skb)
> >  static void tcp_set_skb_tso_segs(struct sock *sk, struct sk_buff *skb,
> >  				 unsigned int mss_now)
> >  {
> > -	if (skb->len <= mss_now || !sk_can_gso(sk)) {
> > +	if (skb->len <= mss_now || !sk_can_gso(sk) ||
> > +	    tcp_urg_mode(tcp_sk(sk))) {
> >  		/* Avoid the costly divide in the normal
> >  		 * non-TSO case.
> >  		 */
> > @@ -1163,7 +1164,9 @@ static int tcp_init_tso_segs(struct sock *sk,
> > struct sk_buff *skb,
> >  {
> >  	int tso_segs = tcp_skb_pcount(skb);
> >
> > -	if (!tso_segs || (tso_segs > 1 && tcp_skb_mss(skb) != mss_now)) {
> > +	if (!tso_segs ||
> > +	    (tso_segs > 1 && (tcp_skb_mss(skb) != mss_now ||
> > +			      tcp_urg_mode(tcp_sk(sk))))) {
> >  		tcp_set_skb_tso_segs(sk, skb, mss_now);
> >  		tso_segs = tcp_skb_pcount(skb);
> >  	}
>
> It's a bit intrusive but I couldn't immediately come up with alternative
> that would have worked (came up with some not working ones :-)).

Yes, I also noticed that. We could add some more code to tcp_mark_urg(), e.g. 
walk sk_write_queue and adjust the pending SKBs there...

Is it OK to simply set all skb->gso_segs to zero, and let the next call to 
tcp_init_tso_segs redo them? I mean, will tcp_init_tso_segs() be always 
called on all SKBs which are in the write queue at the time tcp_mark_urg() is 
called?

Petr
--
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