Re: [RFC net] net: stream: don't purge sk_error_queue without holding its lock

On Tue, 14 Sep 2021 10:55:49 -0700 Eric Dumazet wrote:

On 9/14/21 9:56 AM, Jakub Kicinski wrote:

quoted

Right, but then inet_sock_destruct() also purges the err queue, again.
I was afraid of regressions but we could just remove the purging 
from sk_stream_kill_queues(), and target net-next?
  

Yes, this would be the safest thing.

Alright, let me do more testing and post a removal to net-next.

quoted

quoted

If you think there is a bug, it must be fixed in another way.

IMO, preventing err packets from a prior session being queued after a tcp_disconnect()
is rather hard. We should not even try (packets could be stuck for hours in a qdisc)  

Indeed, we could rearrange the SOCK_DEAD check in sock_queue_err_skb()
to skip queuing and put it under the err queue lock (provided we make
sk_stream_kill_queues() take that lock as well). But seems like an
overkill. I'd lean towards the existing patch or removing the purge from
sk_stream_kill_queues(). LMK what you prefer, this is not urgent.

The issue would really about the tcp_disconnect() case, 
followed by a reuse of the socket to establish another session.

In order to prevent polluting sk_error_queue with notifications
triggered by old packets (from prior flow), this would require
to record the socket cookie in skb, or something like that :/

Ah, I see what you meant! I'll make a note of the disconnect case 
in the commit message. I just care that we don't corrupt the queue
on close.