linux-kernel - Re: [RFC PATCH] sched/eevdf: Return leftmost entity in pick

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZiJBqRafMdBi+wCV@chenyu5-mobl2>
Date: Fri, 19 Apr 2024 18:04:25 +0800
From: Chen Yu <yu.c.chen@...el.com>
To: Peter Zijlstra <peterz@...radead.org>
CC: Xuewen Yan <xuewen.yan94@...il.com>, Abel Wu <wuyun.abel@...edance.com>,
	Ingo Molnar <mingo@...hat.com>, Vincent Guittot <vincent.guittot@...aro.org>,
	Juri Lelli <juri.lelli@...hat.com>, Tim Chen <tim.c.chen@...el.com>, "Tiwei
 Bie" <tiwei.btw@...group.com>, Honglei Wang <wanghonglei@...ichuxing.com>,
	Aaron Lu <aaron.lu@...el.com>, Chen Yu <yu.chen.surf@...il.com>, Yujie Liu
	<yujie.liu@...el.com>, <linux-kernel@...r.kernel.org>, kernel test robot
	<oliver.sang@...el.com>, Tianchen Ding <dtcccc@...ux.alibaba.com>
Subject: Re: [RFC PATCH] sched/eevdf: Return leftmost entity in pick_eevdf()
 if no eligible entity is found

On 2024-04-19 at 10:24:40 +0200, Peter Zijlstra wrote:
> On Thu, Apr 18, 2024 at 09:03:36PM +0800, Chen Yu wrote:
> 
> > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > index 31bca05c3612..9f203012e8f5 100644
> > --- a/kernel/sched/fair.c
> > +++ b/kernel/sched/fair.c
> > @@ -696,15 +696,23 @@ u64 avg_vruntime(struct cfs_rq *cfs_rq)
> >   *
> >   * XXX could add max_slice to the augmented data to track this.
> >   */
> > +
> > +static s64 limit_entity_lag(struct sched_entity *se, s64 lag)
> > +{
> > +	s64 limit;
> > +
> > +	limit = calc_delta_fair(max_t(u64, 2*se->slice, TICK_NSEC), se);
> > +	return clamp(lag, -limit, limit);
> > +}
> 
> Right, helper makes sense.
> 
> > @@ -3721,6 +3729,7 @@ static void reweight_eevdf(struct cfs_rq *cfs_rq, struct sched_entity *se,
> >  	if (avruntime != se->vruntime) {
> >  		vlag = (s64)(avruntime - se->vruntime);
> >  		vlag = div_s64(vlag * old_weight, weight);
> > +		vlag = limit_entity_lag(se, vlag);
> >  		se->vruntime = avruntime - vlag;
> 
> So the !on_rq case has clamping in update_entity_lag() which is before
> scaling. And that makes more sense to me, because putting a limit on
> vlag before the multiplication *should* ensure the multiplication itself
> doesn't overflow.
>
> But now you allow it to compute garbage and then clip the garbage.
> 

Yes, there is possibility to get multiplication overflow. Clamp first should
be better.
(BTW for !on_rq case in this patch, it is also scaled before the clamp).

> >  	}
> >  
> > @@ -3768,6 +3777,9 @@ static void reweight_entity(struct cfs_rq *cfs_rq, struct sched_entity *se,
> >  
> >  	update_load_set(&se->load, weight);
> >  
> > +	if (!se->on_rq)
> > +		se->vlag = limit_entity_lag(se, se->vlag);
> > +
> 
> Except you now add clamping after scaling too, but in a really weird
> place. Should this not go right after the div_s64() that scales?
>

The reason to put this after update_load_set(&se->load, weight) is because
we want to clamp the vlag based on the latest load, although for reweight_eevdf(),
it uses the old load to clamp it, unless we add new parameter to the calc_delta_fair() to
use the new load rather than the current se->load.
 
> Unlike the reweight_eevdf() case, there might be an argument for doing
> it after scaling in this case. Namely, you can have multiple reweights
> stacking their scale ops.
> 
>

Yes, I saw that your patch which clamp the vlag before scaling, I'll have a try
on that patch. Xuewen should post that v2 if everything works well.
 
> Also, could you put a few words in on how often these clips are hit? I
> suspect it's fairly rare (but crucial when it does).

Everytime it happens a subsequent NULL pointer exception usually happens.
The trace data was posted here: 

https://lore.kernel.org/lkml/ZiAWTU5xb%2FJMn%2FHs@chenyu5-mobl2/

"
Here is the debug log printed by place_entity():


[  397.597268]cfs_rq:0xe75f7100
              cfs_rq.avg_vruntime:-1111846207333767
              cfs_rq.min_vruntime:810640668779
              avg_vruntime():686982466017
              curr(0xc59f4f20 rb_producer weight:15 vruntime:1447773196654 sum_exec_ns:187707021870 ctx(0 73)
              leftmost(0xeacb6e00 vruntime:332464705486 sum_exec_ns:78776125437 load:677)
..

[  397.877251]cfs_rq:0xe75f7100
              cfs_rq.avg_vruntime:-759390883821798
              cfs_rq.min_vruntime:810640668779
              avg_vruntime(): 689577229374
              curr(0xc59f4f20 rb_producer weight:15 vruntime:1453640907998 sum_ns:187792974673 ctx(0 73)
              leftmost(0xeacb6e00 vruntime:-59752941080010 sum_ns:78776125437 load:4)


The leftmost se is a task group, its vruntime reduces from 332464705486 to
-59752941080010, because its load reduced from 677 to 4 due to update_cfs_group()
on the tree entities.

Back to reweight_entity():
vlag = avruntime - se->vruntime = 689577229374 - 332464705486 = 357112523888;
vlag = vlag * old_weight / weight = 357112523888 * 677 / 4 = 60441294668044;        <-------- ouch!
se->vruntime = avruntime - vlag = -59751717438670;

the new se vruntime -59751717438670 is close to what we printed -59752941080010,
consider that the avg_vruntime() vary.

Then later this leftmost se has changed its load back and forth, and when the load is 2,
the vuntime has reached a dangerous threshold to trigger the s64 overflow in
eligible check:

[  398.011991]cfs_rq:0xe75f7100
              cfs_rq.avg_vruntime:-11875977385353427
              cfs_rq.min_vruntime:810640668779
              cfs_rq.avg_load:96985
              leftmost(0xeacb6e00 vruntime:18446623907344963655 load:2)

vruntime_eligible()
{
   key = se.vruntime - cfs_rq.min_vruntime = -120977005256740;
   key * avg_load                        <--------------------- OVERFLOW  s64
}
"