From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Paul E. McKenney" Subject: Re: blk_mq: suspicious RCU usage, rcu_read_lock() used illegally while idle! Date: Fri, 7 Nov 2014 15:42:27 -0800 Message-ID: <20141107234227.GD4901@linux.vnet.ibm.com> References: <20141106171110.GJ5718@linux.vnet.ibm.com> <20141107150628.GB4901@linux.vnet.ibm.com> <20141107.124558.1965251829673123349.davem@davemloft.net> Reply-To: paulmck@linux.vnet.ibm.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20141107.124558.1965251829673123349.davem@davemloft.net> Sender: sparclinux-owner@vger.kernel.org To: David Miller Cc: mroos@linux.ee, hch@infradead.org, linux-scsi@vger.kernel.org, axboe@kernel.dk, sparclinux@vger.kernel.org List-Id: linux-scsi@vger.kernel.org On Fri, Nov 07, 2014 at 12:45:58PM -0500, David Miller wrote: > From: "Paul E. McKenney" > Date: Fri, 7 Nov 2014 07:06:28 -0800 > > > On Fri, Nov 07, 2014 at 12:50:02PM +0200, Meelis Roos wrote: > >> Perhaps DaveM can tell which one is coreect or if there is any related > >> problem. > > > > Happy to push either of them along, if appropriate. > > We should use irq_enter()/irq_exit(), and we need to also add it to > smp_call_function_single_client(). > > I'll code up a final patch for sparc64 and push it upstream, thanks! Even better! ;-) > Some other architectures appear to have the same problem, for example > alpha and IA64. Someone should grep around generic_smp_call_function* > under arch/ and do an audit. I will send something along in a couple days if someone else doesn't beat me to it. ;-) Thanx, Paul From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Paul E. McKenney" Date: Fri, 07 Nov 2014 23:42:27 +0000 Subject: Re: blk_mq: suspicious RCU usage, rcu_read_lock() used illegally while idle! Message-Id: <20141107234227.GD4901@linux.vnet.ibm.com> List-Id: References: <20141106171110.GJ5718@linux.vnet.ibm.com> <20141107150628.GB4901@linux.vnet.ibm.com> <20141107.124558.1965251829673123349.davem@davemloft.net> In-Reply-To: <20141107.124558.1965251829673123349.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, hch@infradead.org, linux-scsi@vger.kernel.org, axboe@kernel.dk, sparclinux@vger.kernel.org On Fri, Nov 07, 2014 at 12:45:58PM -0500, David Miller wrote: > From: "Paul E. McKenney" > Date: Fri, 7 Nov 2014 07:06:28 -0800 > > > On Fri, Nov 07, 2014 at 12:50:02PM +0200, Meelis Roos wrote: > >> Perhaps DaveM can tell which one is coreect or if there is any related > >> problem. > > > > Happy to push either of them along, if appropriate. > > We should use irq_enter()/irq_exit(), and we need to also add it to > smp_call_function_single_client(). > > I'll code up a final patch for sparc64 and push it upstream, thanks! Even better! ;-) > Some other architectures appear to have the same problem, for example > alpha and IA64. Someone should grep around generic_smp_call_function* > under arch/ and do an audit. I will send something along in a couple days if someone else doesn't beat me to it. ;-) Thanx, Paul