From: "Paul E. McKenney" Date: Mon, 14 Dec 2009 20:25:56 +0000 Subject: Re: RCU detected CPU 0 stall after panic on sparc64 Message-Id: <20091214202556.GC6679@linux.vnet.ibm.com> List-Id: References: <20091212.014002.173528710.davem@davemloft.net> In-Reply-To: <20091212.014002.173528710.davem@davemloft.net> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: David Miller Cc: mroos@linux.ee, linux-kernel@vger.kernel.org, sparclinux@vger.kernel.org On Sat, Dec 12, 2009 at 01:40:02AM -0800, David Miller wrote: > From: Meelis Roos > Date: Sat, 12 Dec 2009 10:54:19 +0200 (EET) > > > I tried 2.6.32 git gaad3bf0 on a SMP sparc64 machine (Ultra Enterprise > > 250). For some reason my disks were not found (not important in this > > bugreport) and this resulted in panic (cannnot mount root). However, the > > machine kept going and got the below messages (RCU detected CPU 0 stall > > (t00 jiffies)) from timer interrupts. > > > > This seems to be like a bug in RCU vs panic. > > It's normal actually. A panic() just loops forever and the cpu never > goes through an RCU grace period again as a result, and this triggers > the debugging timer that detects this condition. > > Probably the panic() code should disable that assertion check. Hmmm... At first glance, it looks like RCU should put a notifier onto the panic_notifier_list to disable this check. Or is there some global variable that I can check? Thanx, Paul