[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20170921110429.2qx7jn32f6nqnuwf@hirez.programming.kicks-ass.net>
Date: Thu, 21 Sep 2017 13:04:29 +0200
From: Peter Zijlstra <peterz@...radead.org>
To: Steven Rostedt <rostedt@...dmis.org>
Cc: Thomas Gleixner <tglx@...utronix.de>,
LKML <linux-kernel@...r.kernel.org>,
Ingo Molnar <mingo@...nel.org>,
"H. Peter Anvin" <hpa@...or.com>
Subject: Re: [LOCKDEP SPLAT 4.14-rc1] Conflict between smpboot thread create
completions and get_online_cpus
On Mon, Sep 18, 2017 at 03:24:59PM -0400, Steven Rostedt wrote:
> And adding some debugging where the lockdep splat happened:
>
> watchdog-27 3.... 764857us : cpus_read_lock+0x5/0xc0: lock cpu
> watchdog-27 3.... 764858us : <stack trace>
> => x86_release_hardware+0x8f/0xa0
> => hw_perf_event_destroy+0xe/0x20
> => _free_event+0xca/0x3d0
> => put_event+0x14/0x20
> => perf_event_release_kernel+0x236/0x330
> => watchdog_nmi_disable+0x49/0x70
> => watchdog_disable+0x4d/0x60
> => smpboot_thread_fn+0x138/0x280
> => kthread+0x152/0x190
> => ret_from_fork+0x2a/0x40
> watchdog-27 3.... 764917us : cpus_read_unlock+0x5/0xa0: unlock cpu
> watchdog-27 3d..1 764918us : lock_commit_crosslock+0xc1/0x5b0: xhlock=132
Yeah, known issue. Thomas has already been prodding at the watchdog
stuff.
Powered by blists - more mailing lists