Re: [PATCH AUTOSEL 4.19 46/84] tcp/dccp: fix possible race __inet_lookup_established()

From: Sasha Levin
Date: Thu Jan 09 2020 - 10:32:30 EST


On Thu, Jan 02, 2020 at 01:31:22PM +0530, Naresh Kamboju wrote:
On Fri, 27 Dec 2019 at 23:17, Sasha Levin <sashal@xxxxxxxxxx> wrote:

From: Eric Dumazet <edumazet@xxxxxxxxxx>

[ Upstream commit 8dbd76e79a16b45b2ccb01d2f2e08dbf64e71e40 ]

Michal Kubecek and Firo Yang did a very nice analysis of crashes
happening in __inet_lookup_established().

Since a TCP socket can go from TCP_ESTABLISH to TCP_LISTEN
(via a close()/socket()/listen() cycle) without a RCU grace period,
I should not have changed listeners linkage in their hash table.

They must use the nulls protocol (Documentation/RCU/rculist_nulls.txt),
so that a lookup can detect a socket in a hash list was moved in
another one.

Since we added code in commit d296ba60d8e2 ("soreuseport: Resolve
merge conflict for v4/v6 ordering fix"), we have to add
hlist_nulls_add_tail_rcu() helper.

The kernel panic reported on all devices,
While running LTP syscalls accept* test cases on stable-rc-4.19 branch kernel.
This report log extracted from qemu_x86_64.

Reverting this patch re-solved kernel crash.

I'll drop it until we can look into what's happening here, thanks!

--
Thanks,
Sasha