From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-5.5 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, INCLUDES_PATCH,MAILING_LIST_MULTI,SPF_PASS,USER_AGENT_MUTT autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 14438C43441 for ; Wed, 21 Nov 2018 02:41:18 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id A6BB82146D for ; Wed, 21 Nov 2018 02:41:17 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org A6BB82146D Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.ibm.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-kernel-owner@vger.kernel.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1726483AbeKUNNh (ORCPT ); Wed, 21 Nov 2018 08:13:37 -0500 Received: from mx0a-001b2d01.pphosted.com ([148.163.156.1]:45602 "EHLO mx0a-001b2d01.pphosted.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1725926AbeKUNNg (ORCPT ); Wed, 21 Nov 2018 08:13:36 -0500 Received: from pps.filterd (m0098399.ppops.net [127.0.0.1]) by mx0a-001b2d01.pphosted.com (8.16.0.22/8.16.0.22) with SMTP id wAL2YIKA049972 for ; Tue, 20 Nov 2018 21:41:13 -0500 Received: from e12.ny.us.ibm.com (e12.ny.us.ibm.com [129.33.205.202]) by mx0a-001b2d01.pphosted.com with ESMTP id 2nvw2hubh2-1 (version=TLSv1.2 cipher=AES256-GCM-SHA384 bits=256 verify=NOT) for ; Tue, 20 Nov 2018 21:41:13 -0500 Received: from localhost by e12.ny.us.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Wed, 21 Nov 2018 02:41:12 -0000 Received: from b01cxnp22034.gho.pok.ibm.com (9.57.198.24) by e12.ny.us.ibm.com (146.89.104.199) with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted; (version=TLSv1/SSLv3 cipher=AES256-GCM-SHA384 bits=256/256) Wed, 21 Nov 2018 02:41:09 -0000 Received: from b01ledav003.gho.pok.ibm.com (b01ledav003.gho.pok.ibm.com [9.57.199.108]) by b01cxnp22034.gho.pok.ibm.com (8.14.9/8.14.9/NCO v10.0) with ESMTP id wAL2f8ai64553140 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-GCM-SHA384 bits=256 verify=FAIL); Wed, 21 Nov 2018 02:41:08 GMT Received: from b01ledav003.gho.pok.ibm.com (unknown [127.0.0.1]) by IMSVA (Postfix) with ESMTP id 98D9DB205F; Wed, 21 Nov 2018 02:41:08 +0000 (GMT) Received: from b01ledav003.gho.pok.ibm.com (unknown [127.0.0.1]) by IMSVA (Postfix) with ESMTP id 6E22EB2068; Wed, 21 Nov 2018 02:41:08 +0000 (GMT) Received: from paulmck-ThinkPad-W541 (unknown [9.85.189.19]) by b01ledav003.gho.pok.ibm.com (Postfix) with ESMTP; Wed, 21 Nov 2018 02:41:08 +0000 (GMT) Received: by paulmck-ThinkPad-W541 (Postfix, from userid 1000) id E27CB16C05C0; Tue, 20 Nov 2018 18:41:07 -0800 (PST) Date: Tue, 20 Nov 2018 18:41:07 -0800 From: "Paul E. McKenney" To: Joel Fernandes Cc: linux-kernel@vger.kernel.org, josh@joshtriplett.org, rostedt@goodmis.org, mathieu.desnoyers@efficios.com, jiangshanlai@gmail.com Subject: Re: dyntick-idle CPU and node's qsmask Reply-To: paulmck@linux.ibm.com References: <20181110214659.GA96924@google.com> <20181110230436.GL4170@linux.ibm.com> <20181111030925.GA182908@google.com> <20181111042210.GN4170@linux.ibm.com> <20181111180916.GA25327@google.com> <20181111183618.GY4170@linux.ibm.com> <20181120204243.GA22801@google.com> <20181120222813.GE4170@linux.ibm.com> <20181121020612.GB60896@google.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20181121020612.GB60896@google.com> User-Agent: Mutt/1.5.21 (2010-09-15) X-TM-AS-GCONF: 00 x-cbid: 18112102-0060-0000-0000-000002D638EF X-IBM-SpamModules-Scores: X-IBM-SpamModules-Versions: BY=3.00010090; HX=3.00000242; KW=3.00000007; PH=3.00000004; SC=3.00000270; SDB=6.01120454; UDB=6.00581407; IPR=6.00900576; MB=3.00024256; MTD=3.00000008; XFM=3.00000015; UTC=2018-11-21 02:41:11 X-IBM-AV-DETECTION: SAVI=unused REMOTE=unused XFE=unused x-cbparentid: 18112102-0061-0000-0000-000047439C14 Message-Id: <20181121024107.GI4170@linux.ibm.com> X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10434:,, definitions=2018-11-21_01:,, signatures=0 X-Proofpoint-Spam-Details: rule=outbound_notspam policy=outbound score=0 priorityscore=1501 malwarescore=0 suspectscore=0 phishscore=0 bulkscore=0 spamscore=0 clxscore=1015 lowpriorityscore=0 mlxscore=0 impostorscore=0 mlxlogscore=999 adultscore=0 classifier=spam adjust=0 reason=mlx scancount=1 engine=8.0.1-1810050000 definitions=main-1811210022 Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue, Nov 20, 2018 at 06:06:12PM -0800, Joel Fernandes wrote: > On Tue, Nov 20, 2018 at 02:28:14PM -0800, Paul E. McKenney wrote: > > On Tue, Nov 20, 2018 at 12:42:43PM -0800, Joel Fernandes wrote: > > > On Sun, Nov 11, 2018 at 10:36:18AM -0800, Paul E. McKenney wrote: > > > > On Sun, Nov 11, 2018 at 10:09:16AM -0800, Joel Fernandes wrote: > > > > > On Sat, Nov 10, 2018 at 08:22:10PM -0800, Paul E. McKenney wrote: > > > > > > On Sat, Nov 10, 2018 at 07:09:25PM -0800, Joel Fernandes wrote: > > > > > > > On Sat, Nov 10, 2018 at 03:04:36PM -0800, Paul E. McKenney wrote: > > > > > > > > On Sat, Nov 10, 2018 at 01:46:59PM -0800, Joel Fernandes wrote: > > > > > > > > > Hi Paul and everyone, > > > > > > > > > > > > > > > > > > I was tracing/studying the RCU code today in paul/dev branch and noticed that > > > > > > > > > for dyntick-idle CPUs, the RCU GP thread is clearing the rnp->qsmask > > > > > > > > > corresponding to the leaf node for the idle CPU, and reporting a QS on their > > > > > > > > > behalf. > > > > > > > > > > > > > > > > > > rcu_sched-10 [003] 40.008039: rcu_fqs: rcu_sched 792 0 dti > > > > > > > > > rcu_sched-10 [003] 40.008039: rcu_fqs: rcu_sched 801 2 dti > > > > > > > > > rcu_sched-10 [003] 40.008041: rcu_quiescent_state_report: rcu_sched 805 5>0 0 0 3 0 > > > > > > > > > > > > > > > > > > That's all good but I was wondering if we can do better for the idle CPUs if > > > > > > > > > we can some how not set the qsmask of the node in the first place. Then no > > > > > > > > > reporting would be needed of quiescent state is needed for idle CPUs right? > > > > > > > > > And we would also not need to acquire the rnp lock I think. > > > > > > > > > > > > > > > > > > At least for a single node tree RCU system, it seems that would avoid needing > > > > > > > > > to acquire the lock without complications. Anyway let me know your thoughts > > > > > > > > > and happy to discuss this at the hallways of the LPC as well for folks > > > > > > > > > attending :) > > > > > > > > > > > > > > > > We could, but that would require consulting the rcu_data structure for > > > > > > > > each CPU while initializing the grace period, thus increasing the number > > > > > > > > of cache misses during grace-period initialization and also shortly after > > > > > > > > for any non-idle CPUs. This seems backwards on busy systems where each > > > > > > > > > > > > > > When I traced, it appears to me that rcu_data structure of a remote CPU was > > > > > > > being consulted anyway by the rcu_sched thread. So it seems like such cache > > > > > > > miss would happen anyway whether it is during grace-period initialization or > > > > > > > during the fqs stage? I guess I'm trying to say, the consultation of remote > > > > > > > CPU's rcu_data happens anyway. > > > > > > > > > > > > Hmmm... > > > > > > > > > > > > The rcu_gp_init() function does access an rcu_data structure, but it is > > > > > > that of the current CPU, so shouldn't involve a communications cache miss, > > > > > > at least not in the common case. > > > > > > > > > > > > Or are you seeing these cross-CPU rcu_data accesses in rcu_gp_fqs() or > > > > > > functions that it calls? In that case, please see below. > > > > > > > > > > Yes, it was rcu_implicit_dynticks_qs called from rcu_gp_fqs. > > > > > > > > > > > > > CPU will with high probability report its own quiescent state before three > > > > > > > > jiffies pass, in which case the cache misses on the rcu_data structures > > > > > > > > would be wasted motion. > > > > > > > > > > > > > > If all the CPUs are busy and reporting their QS themselves, then I think the > > > > > > > qsmask is likely 0 so then rcu_implicit_dynticks_qs (called from > > > > > > > force_qs_rnp) wouldn't be called and so there would no cache misses on > > > > > > > rcu_data right? > > > > > > > > > > > > Yes, but assuming that all CPUs report their quiescent states before > > > > > > the first call to rcu_gp_fqs(). One exception is when some CPU is > > > > > > looping in the kernel for many milliseconds without passing through a > > > > > > quiescent state. This is because for recent kernels, cond_resched() > > > > > > is not a quiescent state until the grace period is something like 100 > > > > > > milliseconds old. (For older kernels, cond_resched() was never an RCU > > > > > > quiescent state unless it actually scheduled.) > > > > > > > > > > > > Why wait 100 milliseconds? Because otherwise the increase in > > > > > > cond_resched() overhead shows up all too well, causing 0day test robot > > > > > > to complain bitterly. Besides, I would expect that in the common case, > > > > > > CPUs would be executing usermode code. > > > > > > > > > > Makes sense. I was also wondering about this other thing you mentioned about > > > > > waiting for 3 jiffies before reporting the idle CPU's quiescent state. Does > > > > > that mean that even if a single CPU is dyntick-idle for a long period of > > > > > time, then the minimum grace period duration would be atleast 3 jiffies? In > > > > > our mobile embedded devices, jiffies is set to 3.33ms (HZ=300) to keep power > > > > > consumption low. Not that I'm saying its an issue or anything (since IIUC if > > > > > someone wants shorter grace periods, they should just use expedited GPs), but > > > > > it sounds like it would be shorter GP if we just set the qsmask early on some > > > > > how and we can manage the overhead of doing so. > > > > > > > > First, there is some autotuning of the delay based on HZ: > > > > > > > > #define RCU_JIFFIES_TILL_FORCE_QS (1 + (HZ > 250) + (HZ > 500)) > > > > > > > > So at HZ=300, you should be seeing a two-jiffy delay rather than the > > > > usual HZ=1000 three-jiffy delay. Of course, this means that the delay > > > > is 6.67ms rather than the usual 3ms, but the theory is that lower HZ > > > > rates often mean slower instruction execution and thus a desire for > > > > lower RCU overhead. There is further autotuning based on number of > > > > CPUs, but this does not kick in until you have 256 CPUs on your system, > > > > and I bet that smartphones aren't there yet. Nevertheless, check out > > > > RCU_JIFFIES_FQS_DIV for more info on this. > > > > > > > > But you can always override this autotuning using the following kernel > > > > boot paramters: > > > > > > > > rcutree.jiffies_till_first_fqs > > > > rcutree.jiffies_till_next_fqs > > > > > > Slightly related, I was just going through your patch in the dev branch "doc: > > > Now jiffies_till_sched_qs solicits from cond_resched()". > > > > > > If I understand correctly, what you're trying to do is set > > > rcu_data.rcu_urgent_qs if you've not heard from the CPU long enough from > > > rcu_implicit_dynticks_qs. > > > > > > Then in the other paths, you are reading this value and similuating a dyntick > > > idle transition even though you may not be really going into dyntick-idle. > > > Actually in the scheduler-tick, you are also using it to set NEED_RESCHED > > > appropriately. > > > > > > Did I get it right so far? > > > > Partially. > > > > The simulated dyntick-idle transition happens if the grace period extends > > for even longer, so that ->rcu_need_heavy_qs gets set. Up to that point, > > all that is asked for is a local-to-the-CPU report of a quiescent state. > > Right, that's true. My feeling was the whole "fake a dyntick idle transition" > seems to me a bit of a hack and I was thinking if not depending on that would > simplify the code so we don't need the rcu_momentary_dyntick_idle. It is there in case scheduling-clock interrupts are for whatever reason not happening on that CPU. > > > I was thinking if we could simplify rcu_note_context_switch (the parts that > > > call rcu_momentary_dyntick_idle), if we did the following in > > > rcu_implicit_dynticks_qs. > > > > > > Since we already call rcu_qs in rcu_note_context_switch, that would clear the > > > rdp->cpu_no_qs flag. Then there should be no need to call > > > rcu_momentary_dyntick_idle from rcu_note_context switch. > > > > But does this also work for the rcu_all_qs() code path? > > Could we not do something like this in rcu_all_qs? as some over-simplified > pseudo code: > > rcu_all_qs() { > if (!urgent_qs || !heavy_qs) > return; > > rcu_qs(); // This clears the rdp->cpu_no_qs flags which we can monitor in > // the diff in my last email (from rcu_implicit_dynticks_qs) > } Except that rcu_qs() doesn't necessarily report the quiescent state to the RCU core. Keeping down context-switch overhead and all that. > > > I think this would simplify cond_resched as well. Could this avoid the need > > > for having an rcu_all_qs at all? Hopefully I didn't some Tasks-RCU corner cases.. > > > > There is also the code path from cond_resched() in PREEMPT=n kernels. > > This needs rcu_all_qs(). Though it is quite possible that some additional > > code collapsing is possible. > > > > > Basically for some background, I was thinking can we simplify the code that > > > calls "rcu_momentary_dyntick_idle" since we already register a qs in other > > > ways (like by resetting cpu_no_qs). > > > > One complication is that rcu_all_qs() is invoked with interrupts > > and preemption enabled, while rcu_note_context_switch() is > > invoked with interrupts disabled. Also, as you say, Tasks RCU. > > Plus rcu_all_qs() wants to exit immediately if there is nothing to > > do, while rcu_note_context_switch() must unconditionally do rcu_qs() > > -- yes, it could check, but that would be redundant with the checks > > This immediate exit is taken care off in the above psuedo code, would that > help the cond_resched performance? It look like you are cautiously edging towards the two wrapper functions calling common code, relying on inlining and simplification. Why not just try doing it? ;-) > > > diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c > > > index c818e0c91a81..5aa0259c014d 100644 > > > --- a/kernel/rcu/tree.c > > > +++ b/kernel/rcu/tree.c > > > @@ -1063,7 +1063,7 @@ static int rcu_implicit_dynticks_qs(struct rcu_data *rdp) > > > * read-side critical section that started before the beginning > > > * of the current RCU grace period. > > > */ > > > - if (rcu_dynticks_in_eqs_since(rdp, rdp->dynticks_snap)) { > > > + if (rcu_dynticks_in_eqs_since(rdp, rdp->dynticks_snap) || !rdp->cpu_no_qs.b.norm) { > > > > If I am not too confused, this change could cause trouble for > > nohz_full CPUs looping in the kernel. Such CPUs don't necessarily take > > scheduler-clock interrupts, last I checked, and this could prevent the > > CPU from reporting its quiescent state to core RCU. > > Would that still be a problem if rcu_all_qs called rcu_qs? Also the above > diff is an OR condition so it is more relaxed than before. Yes, because rcu_qs() is only guaranteed to capture the quiescent state on the current CPU, not necessarily report it to the RCU core. > Assuming the NOHZ_FULL CPUs call cond_resched during their looping, that > would trigger the rcu_all_qs -> rcu_qs path which would clear cpu_no_qs flag > for that CPU right? This would result in the above diff causing a return of 1 > for that CPU (from rcu_implicit_dynticks_qs). One problem is that cond_resched() is a complete no-op in PREEMPT=y kernels. > > Or am I missing something here? > > I think I might be the one missing something but I'm glad we are almost on > the same page ;-) There are a lot of moving parts in this area, to be sure. Thanx, Paul