netdev - Re: rtnl_lock deadlock with tg3 driver

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>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CACKFLin0kKuckRf2b7CmoAM3UyzOQZo7fRUg0-9jT5p_LLAhTA@mail.gmail.com>
Date:   Sun, 30 Aug 2020 21:58:04 -0700
From:   Michael Chan <michael.chan@...adcom.com>
To:     Baptiste Covolato <baptiste@...sta.com>
Cc:     David Christensen <drc@...ux.vnet.ibm.com>,
        Michael Chan <mchan@...adcom.com>,
        Siva Reddy Kallam <siva.kallam@...adcom.com>,
        Prashant Sreedharan <prashant@...adcom.com>,
        Netdev <netdev@...r.kernel.org>, Daniel Stodden <dns@...sta.com>
Subject: Re: rtnl_lock deadlock with tg3 driver

On Fri, Aug 28, 2020 at 5:40 PM Baptiste Covolato <baptiste@...sta.com> wrote:
>
> Hi David, Michael,
>
> I am contacting you because I'm experiencing an issue that seems to be
> awfully close to what David attempted to fix related to the tg3 driver
> infinite sleep while holding rtnl_lock
> (https://lkml.org/lkml/2020/6/15/1122).

David's remaining issue was tg3_reset_task() returning failure due to
some hardware error.  This would leave the driver in a limbo state
with netif_running() still true, but NAPI not enabled.  This can
easily lead to a soft lockup with rtnl held when it tries to disable
NAPI again.

I think the proper fix is to close the device when tg3_reset_task()
fails to bring it to a consistent state.  I haven't heard back from
David in a while, so I will propose a patch to do this in the next
day.

Let's see if this patch will also work for you.  Thanks.