netdev - Re: [PATCH] ipv4: fix a race in ip4_datagram_release

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:	Tue, 10 Jun 2014 18:12:53 -0700
From:	Eric Dumazet <eric.dumazet@...il.com>
To:	dormando <dormando@...ia.net>
Cc:	Alexey Preobrazhensky <preobr@...gle.com>,
	Steffen Klassert <steffen.klassert@...unet.com>,
	David Miller <davem@...emloft.net>, paulmck@...ux.vnet.ibm.com,
	netdev@...r.kernel.org, Kostya Serebryany <kcc@...gle.com>,
	Dmitry Vyukov <dvyukov@...gle.com>,
	Lars Bull <larsbull@...gle.com>,
	Eric Dumazet <edumazet@...gle.com>,
	Bruce Curtis <brutus@...gle.com>,
	Maciej Żenczykowski <maze@...gle.com>,
	Alexei Starovoitov <alexei.starovoitov@...il.com>
Subject: Re: [PATCH] ipv4: fix a race in ip4_datagram_release_cb()

On Tue, 2014-06-10 at 17:55 -0700, Eric Dumazet wrote:
> On Tue, 2014-06-10 at 17:32 -0700, dormando wrote:
> 
> > >
> > 
> > This is neat. So just confirming that I understand (thanks for re-cc'ing
> > me btw <3):
> > 
> > Because of the UDP glitch, it can cause either the kernel panic in the UDP
> > code, but also elsewhere in the TCP stack based on if it's fiddling with
> > the same DST being hosed by a UDP packet?
> > 
> > Unfortunately we're observing a pretty severe sintr cpu increase under
> > 3.14, so I'm going to try this on 3.10.latest. Any potential issues there?
> > 
> > Thankfully it seems like our udpkill utility will reproduce it quickly.
> > Lets see how it goes!
> 
> This is a start, because we have other bugs of the same kind.
> 
> I was working on follow up
> 
> 

For the curious, another problem is in ipv4_sk_update_pmtu()

This can be called on UDP sockets, but from softirq context.

We cannot use sk_dst_lock because this lock is not softirq safe.

I guess we should use xchg() for sk_dst_set() and sk_dst_reset()



--
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