[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <87lea4qqun.ffs@tglx>
Date: Fri, 08 Dec 2023 23:20:00 +0100
From: Thomas Gleixner <tglx@...utronix.de>
To: Martin Zaharinov <micron10@...il.com>, peterz@...radead.org
Cc: netdev <netdev@...r.kernel.org>, Paolo Abeni <pabeni@...hat.com>,
patchwork-bot+netdevbpf@...nel.org, Jakub Kicinski <kuba@...nel.org>,
Stephen Hemminger <stephen@...workplumber.org>, kuba+netdrv@...nel.org,
dsahern@...il.com, Eric Dumazet <edumazet@...gle.com>
Subject: Re: Urgent Bug Report Kernel crash 6.5.2
On Thu, Dec 07 2023 at 00:38, Martin Zaharinov wrote:
>> On 7 Dec 2023, at 0:26, Martin Zaharinov <micron10@...il.com> wrote:
>>
>> in this line is :
>>
>>
>> /*
>> * If the reference count was already in the dead zone, then this
>> * put() operation is imbalanced. Warn, put the reference count back to
>> * DEAD and tell the caller to not deconstruct the object.
>> */
>> if (WARN_ONCE(cnt >= RCUREF_RELEASED, "rcuref - imbalanced put()")) {
>> atomic_set(&ref->refcnt, RCUREF_DEAD);
>> return false;
>> }
So a rcuref_put() operation triggers the warning because the reference
count is already dead, which means the rcuref_put() operation is
imbalanced.
>> [529520.875413] CPU: 13 PID: 0 Comm: swapper/13 Tainted: G O 6.6.3 #1
Can you reproduce this without the Out of Tree module?
>> [529520.875653] RIP: 0010:rcuref_put_slowpath+0x5f/0x70
>> [529520.878136] dst_release+0x1c/0x40
>> [529520.878229] __dev_queue_xmit+0x594/0xcd0
>> [529520.878324] ? eth_header+0x25/0xc0
>> [529520.878417] ip_finish_output2+0x1a0/0x530
>> [529520.878514] process_backlog+0x107/0x210
>> [529520.878610] __napi_poll+0x20/0x180
>> [529520.878702] net_rx_action+0x29f/0x380
>> [529520.878935] __do_softirq+0xd0/0x202
>> [529520.879033] do_softirq+0x3a/0x50
So this is one call chain triggering the issue...
>>> report same problem with kernel 6.6.1 - i think problem is in rcu
>>> but … if have options to add people from RCU here.
That's definitely not a RCU problem. It's a simple refcount fail.
Thanks,
tglx
Powered by blists - more mailing lists