Re: [PATCH net] tcp: prevent concurrent execution of tcp_sk_exit_batch

[Kuniyuki Iwashima <kuniyu@amazon.com>]

From: Florian Westphal <fw@strlen.de>
Date: Tue, 13 Aug 2024 01:52:59 +0200
> Kuniyuki Iwashima <kuniyu@amazon.com> wrote:
> > > ... because refcount_dec() of tw_refcount unexpectedly dropped to 0.
> > > 
> > > This doesn't seem like an actual bug (no tw sockets got lost and I don't
> > > see a use-after-free) but as erroneous trigger of debug check.
> > 
> > I guess the reason you don't move the check back to tcp_sk_exit() is
> 
> No, it would not work. .exit runs before .exit_batch, we'd splat.

ahh, sorry, somehow I checked the order of exit_batch_rtnl() and
ops_exit_list() ... :/


> 
> Before e9bd0cca09d1 ordering doesn't matter because
> refcount_dec_and_test is used consistently, so it was not relevant
> if the 0-transition occured from .exit or later via inet_twsk_kill.
> 
> > to catch a potential issue explained in solution 4 in the link, right ?
> 
> Yes, it would help to catch such issue wrt. twsk lifetime.
> Having the WARN ensures no twsk can escape .exit_batch completion.
> 
> E.g. if twsk destructors in the future refer to some other
> object that has to be released before netns pointers become invalid
> or something like that.
> 
> Does that make sense?

Yes.


> Otherwise I'm open to alternative approaches.
> Or we can wait until syzbot finds a reproducer, I don't think its
> a real/urgent bug.

Agree, also I guess syzbot will not find a repro as it's been
3 months since the first report.

Thanks!