Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1530778 > unrolled thread

Re: [PATCH 2/2] rcu: Force resched_cpu when jiffies >= rcu_state.jiffies_resched

Started by"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
First post2016-11-26 21:50 +0100
Last post2016-11-28 12:00 +0100
Articles 3 — 2 participants

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.


Contents

  Re: [PATCH 2/2] rcu: Force resched_cpu when jiffies >=  rcu_state.jiffies_resched "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> - 2016-11-26 21:50 +0100
    Re: [PATCH 2/2] rcu: Force resched_cpu when jiffies >=  rcu_state.jiffies_resched Byungchul Park <byungchul.park@lge.com> - 2016-11-28 08:00 +0100
      Re: [PATCH 2/2] rcu: Force resched_cpu when jiffies >=  rcu_state.jiffies_resched "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> - 2016-11-28 12:00 +0100

#1530778 — Re: [PATCH 2/2] rcu: Force resched_cpu when jiffies >= rcu_state.jiffies_resched

From"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Date2016-11-26 21:50 +0100
SubjectRe: [PATCH 2/2] rcu: Force resched_cpu when jiffies >= rcu_state.jiffies_resched
Message-ID<sHS54-32s-3@gated-at.bofh.it>
On Fri, Nov 25, 2016 at 12:34:26AM -0800, Paul E. McKenney wrote:
> On Tue, Nov 22, 2016 at 05:12:20PM +0900, Byungchul Park wrote:
> > On Wed, Nov 09, 2016 at 03:32:15PM +0900, Byungchul Park wrote:
> > > Currently rcu code forces CPU into scheduler when jiffies >=
> > > rcu_state.gp_start + jiffies_till_sched_qs, via resched_cpu().
> > > 
> > > It would be better to force CPU into scheduler when jiffies >=
> > > rcu_state.jiffies_resched, too.
> > 
> > Hello,
> > 
> > I think these two patches are necessary to call resched_cpu() even in
> > case of jiffies >= rcu_state.jiffies_resched, too. Am I wrong?
> > 
> > It would be appriciated if you let me know if I was wrong.
> 
> My current thought is that both the "if" statement and the call to
> resched_cpu() should be removed, but I am still testing and working
> through the timing.
> 
> Either way, I do very much appreciate your having called my attention
> to this code!

And the patch shown below seems to do the trick.  Thoughts?

							Thanx, Paul

------------------------------------------------------------------------

commit d2db185bfee894c573faebed93461e9938bdbb61
Author: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Date:   Fri Nov 25 00:07:23 2016 -0800

    rcu: Remove short-term CPU kicking
    
    Commit 4914950aaa12d ("rcu: Stop treating in-kernel CPU-bound workloads
    as errors") added a (relatively) short-timeout call to resched_cpu().
    This was inspired by as issue that was fixed by b7e7ade34e61 ("sched/core:
    Fix remote wakeups").  But given that this issue was fixed, it is time
    for the current commit to remove this call to resched_cpu().
    
    Reported-by: Byungchul Park <byungchul.park@lge.com>
    Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 5f4f80d07852..91a68e4e6671 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -1345,12 +1345,6 @@ static int rcu_implicit_dynticks_qs(struct rcu_data *rdp,
 		rdp->rsp->jiffies_resched += 5; /* Re-enable beating. */
 	}
 
-	/* And if it has been a really long time, kick the CPU as well. */
-	if (ULONG_CMP_GE(jiffies,
-			 rdp->rsp->gp_start + 2 * jiffies_till_sched_qs) ||
-	    ULONG_CMP_GE(jiffies, rdp->rsp->gp_start + jiffies_till_sched_qs))
-		resched_cpu(rdp->cpu);  /* Force CPU into scheduler. */
-
 	return 0;
 }
 

[toc] | [next] | [standalone]


#1531063

FromByungchul Park <byungchul.park@lge.com>
Date2016-11-28 08:00 +0100
Message-ID<sIo4V-75o-1@gated-at.bofh.it>
In reply to#1530778
On Sat, Nov 26, 2016 at 04:00:46AM -0800, Paul E. McKenney wrote:
> On Fri, Nov 25, 2016 at 12:34:26AM -0800, Paul E. McKenney wrote:
> > On Tue, Nov 22, 2016 at 05:12:20PM +0900, Byungchul Park wrote:
> > > On Wed, Nov 09, 2016 at 03:32:15PM +0900, Byungchul Park wrote:
> > > > Currently rcu code forces CPU into scheduler when jiffies >=
> > > > rcu_state.gp_start + jiffies_till_sched_qs, via resched_cpu().
> > > > 
> > > > It would be better to force CPU into scheduler when jiffies >=
> > > > rcu_state.jiffies_resched, too.
> > > 
> > > Hello,
> > > 
> > > I think these two patches are necessary to call resched_cpu() even in
> > > case of jiffies >= rcu_state.jiffies_resched, too. Am I wrong?
> > > 
> > > It would be appriciated if you let me know if I was wrong.
> > 
> > My current thought is that both the "if" statement and the call to
> > resched_cpu() should be removed, but I am still testing and working
> > through the timing.
> > 
> > Either way, I do very much appreciate your having called my attention
> > to this code!
> 
> And the patch shown below seems to do the trick.  Thoughts?

Hello,

I see.. I didn't check the log 'rcu: Stop treating in-kernel CPU-bound
workloads as errors'.

If you are convinced to remove them, no objection.

Thank you,
Byungchul

> 
> 							Thanx, Paul
> 
> ------------------------------------------------------------------------
> 
> commit d2db185bfee894c573faebed93461e9938bdbb61
> Author: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> Date:   Fri Nov 25 00:07:23 2016 -0800
> 
>     rcu: Remove short-term CPU kicking
>     
>     Commit 4914950aaa12d ("rcu: Stop treating in-kernel CPU-bound workloads
>     as errors") added a (relatively) short-timeout call to resched_cpu().
>     This was inspired by as issue that was fixed by b7e7ade34e61 ("sched/core:
>     Fix remote wakeups").  But given that this issue was fixed, it is time
>     for the current commit to remove this call to resched_cpu().
>     
>     Reported-by: Byungchul Park <byungchul.park@lge.com>
>     Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> 
> diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> index 5f4f80d07852..91a68e4e6671 100644
> --- a/kernel/rcu/tree.c
> +++ b/kernel/rcu/tree.c
> @@ -1345,12 +1345,6 @@ static int rcu_implicit_dynticks_qs(struct rcu_data *rdp,
>  		rdp->rsp->jiffies_resched += 5; /* Re-enable beating. */
>  	}
>  
> -	/* And if it has been a really long time, kick the CPU as well. */
> -	if (ULONG_CMP_GE(jiffies,
> -			 rdp->rsp->gp_start + 2 * jiffies_till_sched_qs) ||
> -	    ULONG_CMP_GE(jiffies, rdp->rsp->gp_start + jiffies_till_sched_qs))
> -		resched_cpu(rdp->cpu);  /* Force CPU into scheduler. */
> -
>  	return 0;
>  }
>  

[toc] | [prev] | [next] | [standalone]


#1531202

From"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Date2016-11-28 12:00 +0100
Message-ID<sIrPb-12d-9@gated-at.bofh.it>
In reply to#1531063
On Mon, Nov 28, 2016 at 03:46:39PM +0900, Byungchul Park wrote:
> On Sat, Nov 26, 2016 at 04:00:46AM -0800, Paul E. McKenney wrote:
> > On Fri, Nov 25, 2016 at 12:34:26AM -0800, Paul E. McKenney wrote:
> > > On Tue, Nov 22, 2016 at 05:12:20PM +0900, Byungchul Park wrote:
> > > > On Wed, Nov 09, 2016 at 03:32:15PM +0900, Byungchul Park wrote:
> > > > > Currently rcu code forces CPU into scheduler when jiffies >=
> > > > > rcu_state.gp_start + jiffies_till_sched_qs, via resched_cpu().
> > > > > 
> > > > > It would be better to force CPU into scheduler when jiffies >=
> > > > > rcu_state.jiffies_resched, too.
> > > > 
> > > > Hello,
> > > > 
> > > > I think these two patches are necessary to call resched_cpu() even in
> > > > case of jiffies >= rcu_state.jiffies_resched, too. Am I wrong?
> > > > 
> > > > It would be appriciated if you let me know if I was wrong.
> > > 
> > > My current thought is that both the "if" statement and the call to
> > > resched_cpu() should be removed, but I am still testing and working
> > > through the timing.
> > > 
> > > Either way, I do very much appreciate your having called my attention
> > > to this code!
> > 
> > And the patch shown below seems to do the trick.  Thoughts?
> 
> Hello,
> 
> I see.. I didn't check the log 'rcu: Stop treating in-kernel CPU-bound
> workloads as errors'.
> 
> If you are convinced to remove them, no objection.

It is quite possible that additional time-related adjustments are needed,
but this change is a step in the right direction.

							Thanx, Paul

> Thank you,
> Byungchul
> 
> > 
> > 							Thanx, Paul
> > 
> > ------------------------------------------------------------------------
> > 
> > commit d2db185bfee894c573faebed93461e9938bdbb61
> > Author: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> > Date:   Fri Nov 25 00:07:23 2016 -0800
> > 
> >     rcu: Remove short-term CPU kicking
> >     
> >     Commit 4914950aaa12d ("rcu: Stop treating in-kernel CPU-bound workloads
> >     as errors") added a (relatively) short-timeout call to resched_cpu().
> >     This was inspired by as issue that was fixed by b7e7ade34e61 ("sched/core:
> >     Fix remote wakeups").  But given that this issue was fixed, it is time
> >     for the current commit to remove this call to resched_cpu().
> >     
> >     Reported-by: Byungchul Park <byungchul.park@lge.com>
> >     Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> > 
> > diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> > index 5f4f80d07852..91a68e4e6671 100644
> > --- a/kernel/rcu/tree.c
> > +++ b/kernel/rcu/tree.c
> > @@ -1345,12 +1345,6 @@ static int rcu_implicit_dynticks_qs(struct rcu_data *rdp,
> >  		rdp->rsp->jiffies_resched += 5; /* Re-enable beating. */
> >  	}
> >  
> > -	/* And if it has been a really long time, kick the CPU as well. */
> > -	if (ULONG_CMP_GE(jiffies,
> > -			 rdp->rsp->gp_start + 2 * jiffies_till_sched_qs) ||
> > -	    ULONG_CMP_GE(jiffies, rdp->rsp->gp_start + jiffies_till_sched_qs))
> > -		resched_cpu(rdp->cpu);  /* Force CPU into scheduler. */
> > -
> >  	return 0;
> >  }
> >  
> 

[toc] | [prev] | [standalone]


Back to top | Article view | linux.kernel


csiph-web