netdev - Re: inconsistent lock state in rxrpc_put_client

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>] [day] [month] [year] [list]

Message-ID: <2183052.1580994576@warthog.procyon.org.uk>
Date:   Thu, 06 Feb 2020 13:09:36 +0000
From:   David Howells <dhowells@...hat.com>
To:     Hillf Danton <hdanton@...a.com>
Cc:     dhowells@...hat.com,
        syzbot <syzbot+3f1fd6b8cbf8702d134e@...kaller.appspotmail.com>,
        davem@...emloft.net, kuba@...nel.org,
        linux-afs@...ts.infradead.org, linux-kernel@...r.kernel.org,
        netdev@...r.kernel.org, syzkaller-bugs@...glegroups.com
Subject: Re: inconsistent lock state in rxrpc_put_client_conn

Hillf Danton <hdanton@...a.com> wrote:

> Take lock with irq quiesced.

I think that's overkill.  It only needs _bh annotations, not _irqsave/restore
- but even that is probably not the best way.

The best way is to offload the stuff done by rxrpc_rcu_destroy_call() to a
workqueue if called in softirq mode.  I'm not sure whether rcu callbacks are
done in softirq mode - if they are, then it can just call rxrpc_queue_work().

David