linux-kernel - Re: [BUG nohz]: wrong user and system time accounting

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: <20170412131818.GB21309@lerouge>
Date:   Wed, 12 Apr 2017 15:18:19 +0200
From:   Frederic Weisbecker <fweisbec@...il.com>
To:     Thomas Gleixner <tglx@...utronix.de>
Cc:     Wanpeng Li <kernellwp@...il.com>, Mike Galbraith <efault@....de>,
        Rik van Riel <riel@...hat.com>,
        Luiz Capitulino <lcapitulino@...hat.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        Peter Zijlstra <peterz@...radead.org>
Subject: Re: [BUG nohz]: wrong user and system time accounting

On Tue, Apr 11, 2017 at 04:22:48PM +0200, Thomas Gleixner wrote:
> On Thu, 30 Mar 2017, Wanpeng Li wrote:
> > diff --git a/kernel/sched/cputime.c b/kernel/sched/cputime.c
> > index f3778e2b..f1ee393 100644
> > --- a/kernel/sched/cputime.c
> > +++ b/kernel/sched/cputime.c
> > @@ -676,18 +676,21 @@ void thread_group_cputime_adjusted(struct
> > task_struct *p, u64 *ut, u64 *st)
> >  #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
> >  static u64 vtime_delta(struct task_struct *tsk)
> >  {
> > -    unsigned long now = READ_ONCE(jiffies);
> > +    u64 now = local_clock();
> > +    u64 delta;
> > +
> > +    delta = now - tsk->vtime_snap;
> > 
> > -    if (time_before(now, (unsigned long)tsk->vtime_snap))
> > +    if (delta < TICK_NSEC)
> >          return 0;
> > 
> > -    return jiffies_to_nsecs(now - tsk->vtime_snap);
> > +    return jiffies_to_nsecs(delta / TICK_NSEC);
> 
> So you replaced a jiffies based approach with a jiffies based approach.
> 
> >  }
> > 
> >  static u64 get_vtime_delta(struct task_struct *tsk)
> >  {
> > -    unsigned long now = READ_ONCE(jiffies);
> > -    u64 delta, other;
> > +    u64 delta = vtime_delta(tsk);
> > +    u64 other;
> > 
> >      /*
> >       * Unlike tick based timing, vtime based timing never has lost
> > @@ -696,10 +699,9 @@ static u64 get_vtime_delta(struct task_struct *tsk)
> >       * elapsed time. Limit account_other_time to prevent rounding
> >       * errors from causing elapsed vtime to go negative.
> >       */
> > -    delta = jiffies_to_nsecs(now - tsk->vtime_snap);
> >      other = account_other_time(delta);
> >      WARN_ON_ONCE(tsk->vtime_snap_whence == VTIME_INACTIVE);
> > -    tsk->vtime_snap = now;
> > +    tsk->vtime_snap += delta;
> 
> Here is how it works^Wfails
> 
> 	       For simplicity tsk->vtime_snap starts at 0
> 	       HZ = 1000
> 
> CPU0 	       CPU1
> 	       sysexit()
> 		 account_system()
> 		   now == 0
> 		   delta = vtime_delta()	<- 0ns
> 	           tsk->vtime_snap += delta;	== 0ns
> 
> 	       busy_loop(995us)
> 
> 	       sysenter()
> 		 now == 996us
> 		 account_user()
> 		   delta = vtime_delta()	<- 0ns
> 		   tsk->vtime_snap += delta	== 0ns
> 		   
> 	       sysexit()
> 		 account_system()
> 		   now == 1001us
> 		   delta = vtime_delta()	<- 10000000ns
> 
> 		   ^^^^ Gets accounted to system
> 
> 	           tsk->vtime_snap += delta;	== 10000000ns
> 
> It's not different from the current jiffies based stuff at all. Same
> failure mode.

Yes you're right, I got confused again. So to fix this we could do our snapshots
at a frequency lower than HZ but still high enough to avoid overhead.

Something like TICK_NSEC / 2 ?