[PATCH] Allocate memory cgroup structures in local nodes v2

[PATCH] Allocate memory cgroup structures in local nodes v2

[Andi Kleen]

From: Andi Kleen <ak@linux.intel.com>

[Andrew: since this is a regression and a very simple fix
could you still consider it for .39? Thanks]

dde79e005a769 added a regression that the memory cgroup data structures
all end up in node 0 because the first attempt at allocating them
would not pass in a node hint. Since the initialization runs on CPU #0
it would all end up node 0. This is a problem on large memory systems,
where node 0 would lose a lot of memory.

Change the alloc_pages_exact to alloc_pages_exact_node. This will
still fall back to other nodes if not enough memory is available.

[RED-PEN: right now it would fall back first before trying
vmalloc_node. Probably not the best strategy ... But I left it like
that for now.]

v2: Fix argument order. Thanks David Rientjes.
Reported-by: Doug Nelson
Cc: rientjes@google.com
CC: Michal Hocko <mhocko@suse.cz>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
 mm/page_cgroup.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/mm/page_cgroup.c b/mm/page_cgroup.c
index 9905501..a362215 100644
--- a/mm/page_cgroup.c
+++ b/mm/page_cgroup.c
@@ -134,7 +134,7 @@ static void *__init_refok alloc_page_cgroup(size_t size, int nid)
 {
 	void *addr = NULL;
 
-	addr = alloc_pages_exact(size, GFP_KERNEL | __GFP_NOWARN);
+	addr = alloc_pages_exact_node(nid, GFP_KERNEL | __GFP_NOWARN, size);
 	if (addr)
 		return addr;
 
-- 
1.7.4.4

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH] Allocate memory cgroup structures in local nodes v2

[Johannes Weiner]

On Wed, May 04, 2011 at 01:26:23PM -0700, Andi Kleen wrote:
> diff --git a/mm/page_cgroup.c b/mm/page_cgroup.c
> index 9905501..a362215 100644
> --- a/mm/page_cgroup.c
> +++ b/mm/page_cgroup.c
> @@ -134,7 +134,7 @@ static void *__init_refok alloc_page_cgroup(size_t size, int nid)
>  {
>  	void *addr = NULL;
>  
> -	addr = alloc_pages_exact(size, GFP_KERNEL | __GFP_NOWARN);
> +	addr = alloc_pages_exact_node(nid, GFP_KERNEL | __GFP_NOWARN, size);

alloc_pages_exact_node is not the 'specify node as well'-version of
alloc_pages_exact, it refers to 'exact node'.  Thus the
free_pages_exact call is no longer the right counter-part.

alloc_pages_exact_node takes an order, not a size argument.

alloc_pages_exact_node returns a pointer to the struct page, not to
the allocated memory, like all other alloc_pages* functions with the
exception of alloc_pages_exact.

I don't think any of those mistakes even triggers a compiler warning.
Wow.  This API is so thoroughly fscked beyond belief that I think the
only way to top this is to have one of the functions invert the bits
of its return value depending on the parity of the uptime counter.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH] Allocate memory cgroup structures in local nodes v2

[Andi Kleen]

> I don't think any of those mistakes even triggers a compiler warning.
> Wow.  This API is so thoroughly fscked beyond belief that I think the
> only way to top this is to have one of the functions invert the bits
> of its return value depending on the parity of the uptime counter.

Yes I must agree. Oops. Ok I'm retracting the patch for now
and do more testing (i think it just hit the fallback)

-Andi

-- 
ak@linux.intel.com -- Speaking for myself only.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH] Allocate memory cgroup structures in local nodes v2 II

[Andi Kleen]

> alloc_pages_exact_node takes an order, not a size argument.
> 
> alloc_pages_exact_node returns a pointer to the struct page, not to
> the allocated memory, like all other alloc_pages* functions with the
> exception of alloc_pages_exact.

In addition to all of this it's also not exact, but just a normal
order of two allocation.

-Andi

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH] Allocate memory cgroup structures in local nodes v2

[Michal Hocko]

On Wed 04-05-11 23:38:50, Johannes Weiner wrote:
> On Wed, May 04, 2011 at 01:26:23PM -0700, Andi Kleen wrote:
> > diff --git a/mm/page_cgroup.c b/mm/page_cgroup.c
> > index 9905501..a362215 100644
> > --- a/mm/page_cgroup.c
> > +++ b/mm/page_cgroup.c
> > @@ -134,7 +134,7 @@ static void *__init_refok alloc_page_cgroup(size_t size, int nid)
> >  {
> >  	void *addr = NULL;
> >  
> > -	addr = alloc_pages_exact(size, GFP_KERNEL | __GFP_NOWARN);
> > +	addr = alloc_pages_exact_node(nid, GFP_KERNEL | __GFP_NOWARN, size);
> 
> alloc_pages_exact_node is not the 'specify node as well'-version of
> alloc_pages_exact, it refers to 'exact node'.  Thus the
> free_pages_exact call is no longer the right counter-part.
> 
> alloc_pages_exact_node takes an order, not a size argument.
> 
> alloc_pages_exact_node returns a pointer to the struct page, not to
> the allocated memory, like all other alloc_pages* functions with the
> exception of alloc_pages_exact.
> 
> I don't think any of those mistakes even triggers a compiler warning.
> Wow.  This API is so thoroughly fscked beyond belief that I think the
> only way to top this is to have one of the functions invert the bits
> of its return value depending on the parity of the uptime counter.

I think Dave Hansen is doing a cleanup in that area
(https://lkml.org/lkml/2011/4/11/337).

-- 
Michal Hocko
SUSE Labs
SUSE LINUX s.r.o.
Lihovarska 1060/12
190 00 Praha 9    
Czech Republic

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>