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

Groups > linux.kernel > #1687136 > unrolled thread

[PATCH 3/9] mm, page_alloc: do not set_cpu_numa_mem on empty nodes initialization

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

  [PATCH 3/9] mm, page_alloc: do not set_cpu_numa_mem on empty nodes initialization Michal Hocko <mhocko@kernel.org> - 2017-07-14 10:10 +0200
    Re: [PATCH 3/9] mm, page_alloc: do not set_cpu_numa_mem on empty  nodes initialization Mel Gorman <mgorman@suse.de> - 2017-07-14 11:50 +0200
      Re: [PATCH 3/9] mm, page_alloc: do not set_cpu_numa_mem on empty  nodes initialization Michal Hocko <mhocko@kernel.org> - 2017-07-14 13:00 +0200
        Re: [PATCH 3/9] mm, page_alloc: do not set_cpu_numa_mem on empty  nodes initialization Mel Gorman <mgorman@suse.de> - 2017-07-14 14:40 +0200
          Re: [PATCH 3/9] mm, page_alloc: do not set_cpu_numa_mem on empty  nodes initialization Michal Hocko <mhocko@kernel.org> - 2017-07-14 14:40 +0200
            Re: [PATCH 3/9] mm, page_alloc: do not set_cpu_numa_mem on empty  nodes initialization Mel Gorman <mgorman@suse.de> - 2017-07-14 15:00 +0200

#1687136 — [PATCH 3/9] mm, page_alloc: do not set_cpu_numa_mem on empty nodes initialization

From: Michal Hocko <mhocko@suse.com>

__build_all_zonelists reinitializes each online cpu local node for
CONFIG_HAVE_MEMORYLESS_NODES. This makes sense because previously memory
less nodes could gain some memory during memory hotplug and so the local
node should be changed for CPUs close to such a node. It makes less
sense to do that unconditionally for a newly creaded NUMA node which is
still offline and without any memory.

Let's also simplify the cpu loop and use for_each_online_cpu instead of
an explicit cpu_online check for all possible cpus.

Signed-off-by: Michal Hocko <mhocko@suse.com>
---
 mm/page_alloc.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 7746824a425d..ebc3311555b1 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -5096,10 +5096,8 @@ static int __build_all_zonelists(void *data)
 
 			build_zonelists(pgdat);
 		}
-	}
 
 #ifdef CONFIG_HAVE_MEMORYLESS_NODES
-	for_each_possible_cpu(cpu) {
 		/*
 		 * We now know the "local memory node" for each node--
 		 * i.e., the node of the first zone in the generic zonelist.
@@ -5108,10 +5106,10 @@ static int __build_all_zonelists(void *data)
 		 * secondary cpus' numa_mem as they come on-line.  During
 		 * node/memory hotplug, we'll fixup all on-line cpus.
 		 */
-		if (cpu_online(cpu))
+		for_each_online_cpu(cpu)
 			set_cpu_numa_mem(cpu, local_memory_node(cpu_to_node(cpu)));
-	}
 #endif
+	}
 
 	return 0;
 }
-- 
2.11.0

[toc] | [next] | [standalone]

#1687207 — Re: [PATCH 3/9] mm, page_alloc: do not set_cpu_numa_mem on empty nodes initialization

On Fri, Jul 14, 2017 at 10:00:00AM +0200, Michal Hocko wrote:
> From: Michal Hocko <mhocko@suse.com>
> 
> __build_all_zonelists reinitializes each online cpu local node for
> CONFIG_HAVE_MEMORYLESS_NODES. This makes sense because previously memory
> less nodes could gain some memory during memory hotplug and so the local
> node should be changed for CPUs close to such a node. It makes less
> sense to do that unconditionally for a newly creaded NUMA node which is
> still offline and without any memory.
> 
> Let's also simplify the cpu loop and use for_each_online_cpu instead of
> an explicit cpu_online check for all possible cpus.
> 
> Signed-off-by: Michal Hocko <mhocko@suse.com>
> ---
>  mm/page_alloc.c | 6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
> 
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 7746824a425d..ebc3311555b1 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -5096,10 +5096,8 @@ static int __build_all_zonelists(void *data)
>  
>  			build_zonelists(pgdat);
>  		}
> -	}
>  
>  #ifdef CONFIG_HAVE_MEMORYLESS_NODES
> -	for_each_possible_cpu(cpu) {
>  		/*
>  		 * We now know the "local memory node" for each node--
>  		 * i.e., the node of the first zone in the generic zonelist.
> @@ -5108,10 +5106,10 @@ static int __build_all_zonelists(void *data)
>  		 * secondary cpus' numa_mem as they come on-line.  During
>  		 * node/memory hotplug, we'll fixup all on-line cpus.
>  		 */
> -		if (cpu_online(cpu))
> +		for_each_online_cpu(cpu)
>  			set_cpu_numa_mem(cpu, local_memory_node(cpu_to_node(cpu)));
> -	}
>  #endif
> +	}
>  

This is not as clear a benefit. For each online node, we now go through
all online CPUs once per node. There would be some rationale for using
for_each_online_cpu.

-- 
Mel Gorman
SUSE Labs

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

#1687245 — Re: [PATCH 3/9] mm, page_alloc: do not set_cpu_numa_mem on empty nodes initialization

On Fri 14-07-17 10:48:10, Mel Gorman wrote:
> On Fri, Jul 14, 2017 at 10:00:00AM +0200, Michal Hocko wrote:
> > From: Michal Hocko <mhocko@suse.com>
> > 
> > __build_all_zonelists reinitializes each online cpu local node for
> > CONFIG_HAVE_MEMORYLESS_NODES. This makes sense because previously memory
> > less nodes could gain some memory during memory hotplug and so the local
> > node should be changed for CPUs close to such a node. It makes less
> > sense to do that unconditionally for a newly creaded NUMA node which is
> > still offline and without any memory.
> > 
> > Let's also simplify the cpu loop and use for_each_online_cpu instead of
> > an explicit cpu_online check for all possible cpus.
> > 
> > Signed-off-by: Michal Hocko <mhocko@suse.com>
> > ---
> >  mm/page_alloc.c | 6 ++----
> >  1 file changed, 2 insertions(+), 4 deletions(-)
> > 
> > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > index 7746824a425d..ebc3311555b1 100644
> > --- a/mm/page_alloc.c
> > +++ b/mm/page_alloc.c
> > @@ -5096,10 +5096,8 @@ static int __build_all_zonelists(void *data)
> >  
> >  			build_zonelists(pgdat);
> >  		}
> > -	}
> >  
> >  #ifdef CONFIG_HAVE_MEMORYLESS_NODES
> > -	for_each_possible_cpu(cpu) {
> >  		/*
> >  		 * We now know the "local memory node" for each node--
> >  		 * i.e., the node of the first zone in the generic zonelist.
> > @@ -5108,10 +5106,10 @@ static int __build_all_zonelists(void *data)
> >  		 * secondary cpus' numa_mem as they come on-line.  During
> >  		 * node/memory hotplug, we'll fixup all on-line cpus.
> >  		 */
> > -		if (cpu_online(cpu))
> > +		for_each_online_cpu(cpu)
> >  			set_cpu_numa_mem(cpu, local_memory_node(cpu_to_node(cpu)));
> > -	}
> >  #endif
> > +	}
> >  
> 
> This is not as clear a benefit. For each online node, we now go through
> all online CPUs once per node. There would be some rationale for using
> for_each_online_cpu.

I am not sure I understand. I am using for_each_online_cpu...
-- 
Michal Hocko
SUSE Labs

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

#1687341 — Re: [PATCH 3/9] mm, page_alloc: do not set_cpu_numa_mem on empty nodes initialization

On Fri, Jul 14, 2017 at 12:50:04PM +0200, Michal Hocko wrote:
> On Fri 14-07-17 10:48:10, Mel Gorman wrote:
> > On Fri, Jul 14, 2017 at 10:00:00AM +0200, Michal Hocko wrote:
> > > From: Michal Hocko <mhocko@suse.com>
> > > 
> > > __build_all_zonelists reinitializes each online cpu local node for
> > > CONFIG_HAVE_MEMORYLESS_NODES. This makes sense because previously memory
> > > less nodes could gain some memory during memory hotplug and so the local
> > > node should be changed for CPUs close to such a node. It makes less
> > > sense to do that unconditionally for a newly creaded NUMA node which is
> > > still offline and without any memory.
> > > 
> > > Let's also simplify the cpu loop and use for_each_online_cpu instead of
> > > an explicit cpu_online check for all possible cpus.
> > > 
> > > Signed-off-by: Michal Hocko <mhocko@suse.com>
> > > ---
> > >  mm/page_alloc.c | 6 ++----
> > >  1 file changed, 2 insertions(+), 4 deletions(-)
> > > 
> > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > > index 7746824a425d..ebc3311555b1 100644
> > > --- a/mm/page_alloc.c
> > > +++ b/mm/page_alloc.c
> > > @@ -5096,10 +5096,8 @@ static int __build_all_zonelists(void *data)
> > >  
> > >  			build_zonelists(pgdat);
> > >  		}
> > > -	}
> > >  
> > >  #ifdef CONFIG_HAVE_MEMORYLESS_NODES
> > > -	for_each_possible_cpu(cpu) {
> > >  		/*
> > >  		 * We now know the "local memory node" for each node--
> > >  		 * i.e., the node of the first zone in the generic zonelist.
> > > @@ -5108,10 +5106,10 @@ static int __build_all_zonelists(void *data)
> > >  		 * secondary cpus' numa_mem as they come on-line.  During
> > >  		 * node/memory hotplug, we'll fixup all on-line cpus.
> > >  		 */
> > > -		if (cpu_online(cpu))
> > > +		for_each_online_cpu(cpu)
> > >  			set_cpu_numa_mem(cpu, local_memory_node(cpu_to_node(cpu)));
> > > -	}
> > >  #endif
> > > +	}
> > >  
> > 
> > This is not as clear a benefit. For each online node, we now go through
> > all online CPUs once per node. There would be some rationale for using
> > for_each_online_cpu.
> 
> I am not sure I understand. I am using for_each_online_cpu...

Yes, but within a loop that looks like

for_each_online_node(nid)
	...
	for_each_online_cpu(cpu)

Or maybe you aren't because we are looking at different baselines. I had
minor fuzz and conflicts applying the series.

-- 
Mel Gorman
SUSE Labs

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

#1687343 — Re: [PATCH 3/9] mm, page_alloc: do not set_cpu_numa_mem on empty nodes initialization

On Fri 14-07-17 13:32:42, Mel Gorman wrote:
> On Fri, Jul 14, 2017 at 12:50:04PM +0200, Michal Hocko wrote:
> > On Fri 14-07-17 10:48:10, Mel Gorman wrote:
> > > On Fri, Jul 14, 2017 at 10:00:00AM +0200, Michal Hocko wrote:
> > > > From: Michal Hocko <mhocko@suse.com>
> > > > 
> > > > __build_all_zonelists reinitializes each online cpu local node for
> > > > CONFIG_HAVE_MEMORYLESS_NODES. This makes sense because previously memory
> > > > less nodes could gain some memory during memory hotplug and so the local
> > > > node should be changed for CPUs close to such a node. It makes less
> > > > sense to do that unconditionally for a newly creaded NUMA node which is
> > > > still offline and without any memory.
> > > > 
> > > > Let's also simplify the cpu loop and use for_each_online_cpu instead of
> > > > an explicit cpu_online check for all possible cpus.
> > > > 
> > > > Signed-off-by: Michal Hocko <mhocko@suse.com>
> > > > ---
> > > >  mm/page_alloc.c | 6 ++----
> > > >  1 file changed, 2 insertions(+), 4 deletions(-)
> > > > 
> > > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > > > index 7746824a425d..ebc3311555b1 100644
> > > > --- a/mm/page_alloc.c
> > > > +++ b/mm/page_alloc.c
> > > > @@ -5096,10 +5096,8 @@ static int __build_all_zonelists(void *data)
> > > >  
> > > >  			build_zonelists(pgdat);
> > > >  		}
> > > > -	}
> > > >  
> > > >  #ifdef CONFIG_HAVE_MEMORYLESS_NODES
> > > > -	for_each_possible_cpu(cpu) {
> > > >  		/*
> > > >  		 * We now know the "local memory node" for each node--
> > > >  		 * i.e., the node of the first zone in the generic zonelist.
> > > > @@ -5108,10 +5106,10 @@ static int __build_all_zonelists(void *data)
> > > >  		 * secondary cpus' numa_mem as they come on-line.  During
> > > >  		 * node/memory hotplug, we'll fixup all on-line cpus.
> > > >  		 */
> > > > -		if (cpu_online(cpu))
> > > > +		for_each_online_cpu(cpu)
> > > >  			set_cpu_numa_mem(cpu, local_memory_node(cpu_to_node(cpu)));
> > > > -	}
> > > >  #endif
> > > > +	}
> > > >  
> > > 
> > > This is not as clear a benefit. For each online node, we now go through
> > > all online CPUs once per node. There would be some rationale for using
> > > for_each_online_cpu.
> > 
> > I am not sure I understand. I am using for_each_online_cpu...
> 
> Yes, but within a loop that looks like
> 
> for_each_online_node(nid)
> 	...
> 	for_each_online_cpu(cpu)
> 
> Or maybe you aren't because we are looking at different baselines. I had
> minor fuzz and conflicts applying the series.

The current mmotm after this patch looks like this
	if (self && !node_online(self->node_id)) {
		build_zonelists(self);
	} else {
		for_each_online_node(nid) {
			pg_data_t *pgdat = NODE_DATA(nid);

			build_zonelists(pgdat);
		}

#ifdef CONFIG_HAVE_MEMORYLESS_NODES
		/*
		 * We now know the "local memory node" for each node--
		 * i.e., the node of the first zone in the generic zonelist.
		 * Set up numa_mem percpu variable for on-line cpus.  During
		 * boot, only the boot cpu should be on-line;  we'll init the
		 * secondary cpus' numa_mem as they come on-line.  During
		 * node/memory hotplug, we'll fixup all on-line cpus.
		 */
		for_each_online_cpu(cpu)
			set_cpu_numa_mem(cpu, local_memory_node(cpu_to_node(cpu)));
#endif
	}

So for_each_online_cpu is called outside of the for_each_online_node.
Have a look at
git://git.kernel.org/pub/scm/linux/kernel/git/mhocko/mm.git
branch attempts/zonlists-build-simplification
-- 
Michal Hocko
SUSE Labs

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

#1687378 — Re: [PATCH 3/9] mm, page_alloc: do not set_cpu_numa_mem on empty nodes initialization

On Fri, Jul 14, 2017 at 02:39:41PM +0200, Michal Hocko wrote:
> On Fri 14-07-17 13:32:42, Mel Gorman wrote:
> > On Fri, Jul 14, 2017 at 12:50:04PM +0200, Michal Hocko wrote:
> > > On Fri 14-07-17 10:48:10, Mel Gorman wrote:
> > > > On Fri, Jul 14, 2017 at 10:00:00AM +0200, Michal Hocko wrote:
> > > > > From: Michal Hocko <mhocko@suse.com>
> > > > > 
> > > > > __build_all_zonelists reinitializes each online cpu local node for
> > > > > CONFIG_HAVE_MEMORYLESS_NODES. This makes sense because previously memory
> > > > > less nodes could gain some memory during memory hotplug and so the local
> > > > > node should be changed for CPUs close to such a node. It makes less
> > > > > sense to do that unconditionally for a newly creaded NUMA node which is
> > > > > still offline and without any memory.
> > > > > 
> > > > > Let's also simplify the cpu loop and use for_each_online_cpu instead of
> > > > > an explicit cpu_online check for all possible cpus.
> > > > > 
> > > > > Signed-off-by: Michal Hocko <mhocko@suse.com>
> > > > > ---
> > > > >  mm/page_alloc.c | 6 ++----
> > > > >  1 file changed, 2 insertions(+), 4 deletions(-)
> > > > > 
> > > > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > > > > index 7746824a425d..ebc3311555b1 100644
> > > > > --- a/mm/page_alloc.c
> > > > > +++ b/mm/page_alloc.c
> > > > > @@ -5096,10 +5096,8 @@ static int __build_all_zonelists(void *data)
> > > > >  
> > > > >  			build_zonelists(pgdat);
> > > > >  		}
> > > > > -	}
> > > > >  
> > > > >  #ifdef CONFIG_HAVE_MEMORYLESS_NODES
> > > > > -	for_each_possible_cpu(cpu) {
> > > > >  		/*
> > > > >  		 * We now know the "local memory node" for each node--
> > > > >  		 * i.e., the node of the first zone in the generic zonelist.
> > > > > @@ -5108,10 +5106,10 @@ static int __build_all_zonelists(void *data)
> > > > >  		 * secondary cpus' numa_mem as they come on-line.  During
> > > > >  		 * node/memory hotplug, we'll fixup all on-line cpus.
> > > > >  		 */
> > > > > -		if (cpu_online(cpu))
> > > > > +		for_each_online_cpu(cpu)
> > > > >  			set_cpu_numa_mem(cpu, local_memory_node(cpu_to_node(cpu)));
> > > > > -	}
> > > > >  #endif
> > > > > +	}
> > > > >  
> > > > 
> > > > This is not as clear a benefit. For each online node, we now go through
> > > > all online CPUs once per node. There would be some rationale for using
> > > > for_each_online_cpu.
> > > 
> > > I am not sure I understand. I am using for_each_online_cpu...
> > 
> > Yes, but within a loop that looks like
> > 
> > for_each_online_node(nid)
> > 	...
> > 	for_each_online_cpu(cpu)
> > 
> > Or maybe you aren't because we are looking at different baselines. I had
> > minor fuzz and conflicts applying the series.
> 
> The current mmotm after this patch looks like this
> 	if (self && !node_online(self->node_id)) {
> 		build_zonelists(self);
> 	} else {
> 		for_each_online_node(nid) {
> 			pg_data_t *pgdat = NODE_DATA(nid);
> 
> 			build_zonelists(pgdat);
> 		}
> 
> #ifdef CONFIG_HAVE_MEMORYLESS_NODES
> 		/*
> 		 * We now know the "local memory node" for each node--
> 		 * i.e., the node of the first zone in the generic zonelist.
> 		 * Set up numa_mem percpu variable for on-line cpus.  During
> 		 * boot, only the boot cpu should be on-line;  we'll init the
> 		 * secondary cpus' numa_mem as they come on-line.  During
> 		 * node/memory hotplug, we'll fixup all on-line cpus.
> 		 */
> 		for_each_online_cpu(cpu)
> 			set_cpu_numa_mem(cpu, local_memory_node(cpu_to_node(cpu)));
> #endif
> 	}
> 

Ok, that makes more sense.

Acked-by: Mel Gorman <mgorman@suse.de>

-- 
Mel Gorman
SUSE Labs

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web