From mboxrd@z Thu Jan 1 00:00:00 1970 From: Waiman Long Subject: Re: [PATCH v3 1/2] qspinlock: Introducing a 4-byte queue spinlock implementation Date: Fri, 31 Jan 2014 13:16:21 -0500 Message-ID: <52EBE875.3070107@hp.com> References: <1390933151-1797-1-git-send-email-Waiman.Long@hp.com> <1390933151-1797-2-git-send-email-Waiman.Long@hp.com> <1391108430.3138.86.camel@schen9-DESK> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from g1t0029.austin.hp.com ([15.216.28.36]:37874 "EHLO g1t0029.austin.hp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932331AbaAaSQ3 (ORCPT ); Fri, 31 Jan 2014 13:16:29 -0500 In-Reply-To: <1391108430.3138.86.camel@schen9-DESK> Sender: linux-arch-owner@vger.kernel.org List-ID: To: Tim Chen Cc: Thomas Gleixner , Ingo Molnar , "H. Peter Anvin" , Arnd Bergmann , linux-arch@vger.kernel.org, x86@kernel.org, linux-kernel@vger.kernel.org, Peter Zijlstra , Steven Rostedt , Andrew Morton , Michel Lespinasse , Andi Kleen , Rik van Riel , "Paul E. McKenney" , Linus Torvalds , Raghavendra K T , George Spelvin , Daniel J Blueman , Alexander Fyodorov , Aswin Chandramouleeswaran , Scott J Norton , Thavatchai Makphaibulchoke On 01/30/2014 02:00 PM, Tim Chen wrote: > On Tue, 2014-01-28 at 13:19 -0500, Waiman Long wrote: > >> +/** >> + * queue_spin_lock_slowpath - acquire the queue spinlock >> + * @lock: Pointer to queue spinlock structure >> + */ >> +void queue_spin_lock_slowpath(struct qspinlock *lock) >> +{ >> + unsigned int cpu_nr, qn_idx; >> + struct qnode *node, *next = NULL; >> + u32 prev_qcode, my_qcode; >> + >> + /* >> + * Get the queue node >> + */ >> + cpu_nr = smp_processor_id(); >> + node = this_cpu_ptr(&qnodes[0]); >> + qn_idx = 0; >> + >> + if (unlikely(node->used)) { >> + /* >> + * This node has been used, try to find an empty queue >> + * node entry. >> + */ >> + for (qn_idx = 1; qn_idx< MAX_QNODES; qn_idx++) >> + if (!node[qn_idx].used) >> + break; >> + if (unlikely(qn_idx == MAX_QNODES)) { >> + /* >> + * This shouldn't happen, print a warning message >> + *& busy spinning on the lock. >> + */ >> + printk_sched( >> + "qspinlock: queue node table exhausted at cpu %d!\n", >> + cpu_nr); >> + while (!unfair_trylock(lock)) >> + arch_mutex_cpu_relax(); >> + return; >> + } >> + /* Adjust node pointer */ >> + node += qn_idx; >> + } >> + >> + /* >> + * Set up the new cpu code to be exchanged >> + */ >> + my_qcode = SET_QCODE(cpu_nr, qn_idx); >> + > If we get interrupted here before we have a chance to set the used flag, > the interrupt handler could pick up the same qnode if it tries to > acquire queued spin lock. Then we could overwrite the qcode we have set > here. > > Perhaps an exchange operation for the used flag to prevent this race > condition? > > Tim That actually is fine. I am assuming that whenever an interrupt handler needs to acquire a spinlock, it can use the same queue node as the interrupted function as long as it can finish the lock acquisition and release queue node back to the pool before returning to the interrupted function. The only case where an interrupt handler cannot use the queue node is when useful data were already there indicated by the setting of the used flag. I will add comment to clarify this possible scenario. -Longman