[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20171205.112213.619683454960802411.davem@davemloft.net>
Date: Tue, 05 Dec 2017 11:22:13 -0500 (EST)
From: David Miller <davem@...emloft.net>
To: code@...pl.cz
Cc: netdev@...r.kernel.org, xiyou.wangcong@...il.com, jiri@...nulli.us
Subject: Re: [PATCH v3] net: sched: crash on blocks with goto chain action
From: Roman Kapl <code@...pl.cz>
Date: Wed, 29 Nov 2017 20:20:14 +0100
> tcf_block_put_ext has assumed that all filters (and thus their goto
> actions) are destroyed in RCU callback and so can not race with our
> list iteration. However, that is not true during netns cleanup (see
> tcf_exts_get_net comment). The assumption was broken by the patch series
> c7e460ce5572..623859ae06b8 ("Merge branch 'net-sched-race-fix'").
>
> Prevent the user after free by holding all chains (except 0, that one is
> already held) as it was done before
> 822e86d997e4 ("net_sched: remove tcf_block_put_deferred()").
>
> To reproduce, run the following in a netns and then delete the ns:
> ip link add dtest type dummy
> tc qdisc add dev dtest ingress
> tc filter add dev dtest chain 1 parent ffff: handle 1 prio 1 flower action goto chain 2
>
> Fixes: 623859ae06b8 ("Merge branch 'net-sched-race-fix'")
> Signed-off-by: Roman Kapl <code@...pl.cz>
This doesn't apply cleanly to 'net'.
Powered by blists - more mailing lists