Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 15 Aug 2023 18:21:41 +0200
From:      Mateusz Guzik <mjguzik@gmail.com>
To:        =?UTF-8?Q?Dag=2DErling_Sm=C3=B8rgrav?= <des@freebsd.org>
Cc:        current@freebsd.org
Subject:   Re: ZFS deadlock in 14
Message-ID:  <CAGudoHHGeC4qaZpmTc15-Rimo78qVUmg8-oYveMfo0_JO45TSw@mail.gmail.com>
In-Reply-To: <86350kqokl.fsf@ltc.des.no>
References:  <86leeltqcb.fsf@ltc.des.no> <86h6p4s64h.fsf@ltc.des.no> <86a5utrafp.fsf@ltc.des.no> <86350kqokl.fsf@ltc.des.no>

next in thread | previous in thread | raw e-mail | index | archive | help
On 8/15/23, Dag-Erling Sm=C3=B8rgrav <des@freebsd.org> wrote:
> Dag-Erling Sm=C3=B8rgrav <des@FreeBSD.org> writes:
>> I managed to geat a deadlock with 4e8d558c9d1c.  Its predecessor
>> 5ca7f02946 appears to be working.  I'm going to try to come up with a
>> more efficient way to reproduce the deadlock than running poudriere.
>
> I wrote a script that creates multiple filesystems, snapshots them,
> populates them and rolls them back continuously but so far I have not
> succeeded in triggering the deadlock without poudriere.  I guess my
> script doesn't consume enough vnodes.
>
> Also, 9228ac3a69c4 (9 August, last commit before the contrib/googletest
> breakage) still deadlocks.
>

Given that the custom reproducer failed I think the most prudent
course of action is to reproduce again with poudriere, but this time
arrange to have all stacktraces dumped.

this should do it:
sbin/ddb/ddb.conf:script kdb.enter.panic=3Dtextdump set; capture on; run
lockinfo; show pcpu; bt; ps; alltrace; capture off; textdump dump;
reset

it is a slightly finicky beast so I would trigger a panic by hand
first to validate it works as expected.

--=20
Mateusz Guzik <mjguzik gmail.com>



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAGudoHHGeC4qaZpmTc15-Rimo78qVUmg8-oYveMfo0_JO45TSw>