[PATCH] Move page_assign_page_cgroup to VM_BUG_ON in free_hot_cold_page

[PATCH] Move page_assign_page_cgroup to VM_BUG_ON in free_hot_cold_page

[Balbir Singh]

Based on the recommendation and observations of Hugh Dickins,
page_cgroup_assign_cgroup() is not required. This patch replaces it with
a VM_BUG_ON, so that we can catch them in free_hot_cold_page()

Signed-off-by: Balbir Singh <balbir@linux.vnet.ibm.com>
---

 mm/page_alloc.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff -puN mm/page_alloc.c~memory-controller-move-to-bug-on-in-free_hot_cold_page mm/page_alloc.c
--- linux-2.6.24-rc5/mm/page_alloc.c~memory-controller-move-to-bug-on-in-free_hot_cold_page	2007-12-19 11:31:46.000000000 +0530
+++ linux-2.6.24-rc5-balbir/mm/page_alloc.c	2007-12-19 11:33:45.000000000 +0530
@@ -995,7 +995,7 @@ static void fastcall free_hot_cold_page(
 
 	if (!PageHighMem(page))
 		debug_check_no_locks_freed(page_address(page), PAGE_SIZE);
-	page_assign_page_cgroup(page, NULL);
+	VM_BUG_ON(page_get_page_cgroup(page));
 	arch_free_page(page, 0);
 	kernel_map_pages(page, 1, 0);
 
_

-- 
	Warm Regards,
	Balbir Singh
	Linux Technology Center
	IBM, ISTL

Re: [PATCH] Move page_assign_page_cgroup to VM_BUG_ON in free_hot_cold_page

[Dave Hansen]

On Wed, 2007-12-19 at 11:48 +0530, Balbir Singh wrote:
> diff -puN mm/page_alloc.c~memory-controller-move-to-bug-on-in-free_hot_cold_page mm/page_alloc.c
> --- linux-2.6.24-rc5/mm/page_alloc.c~memory-controller-move-to-bug-on-in-free_hot_cold_page     2007-12-19 11:31:46.000000000 +0530
> +++ linux-2.6.24-rc5-balbir/mm/page_alloc.c     2007-12-19 11:33:45.000000000 +0530
> @@ -995,7 +995,7 @@ static void fastcall free_hot_cold_page(
>  
>         if (!PageHighMem(page))
>                 debug_check_no_locks_freed(page_address(page), PAGE_SIZE);
> -       page_assign_page_cgroup(page, NULL);
> +       VM_BUG_ON(page_get_page_cgroup(page));
>         arch_free_page(page, 0);
>         kernel_map_pages(page, 1, 0); 

Hi Balbir,

You generally want to do these like:

	foo = page_assign_page_cgroup(page, NULL);
	VM_BUG_ON(foo);

Some embedded people have been known to optimize kernel size like this:

	#define VM_BUG_ON(x) do{}while(0)

-- Dave

Re: [PATCH] Move page_assign_page_cgroup to VM_BUG_ON in free_hot_cold_page

[Hugh Dickins]

On Wed, 19 Dec 2007, Dave Hansen wrote:
> > --- linux-2.6.24-rc5/mm/page_alloc.c~memory-controller-move-to-bug-on-in-free_hot_cold_page     2007-12-19 11:31:46.000000000 +0530
> > +++ linux-2.6.24-rc5-balbir/mm/page_alloc.c     2007-12-19 11:33:45.000000000 +0530
> > @@ -995,7 +995,7 @@ static void fastcall free_hot_cold_page(
> >  
> >         if (!PageHighMem(page))
> >                 debug_check_no_locks_freed(page_address(page), PAGE_SIZE);
> > -       page_assign_page_cgroup(page, NULL);
> > +       VM_BUG_ON(page_get_page_cgroup(page));
> >         arch_free_page(page, 0);
> >         kernel_map_pages(page, 1, 0); 
> 
> Hi Balbir,
> 
> You generally want to do these like:
> 
> 	foo = page_assign_page_cgroup(page, NULL);
> 	VM_BUG_ON(foo);
> 
> Some embedded people have been known to optimize kernel size like this:
> 
> 	#define VM_BUG_ON(x) do{}while(0)

Balbir's patch looks fine to me: I don't get your point there, Dave.

Hugh

Re: [PATCH] Move page_assign_page_cgroup to VM_BUG_ON in free_hot_cold_page

[Peter Zijlstra]

On Thu, 2007-12-20 at 13:14 +0000, Hugh Dickins wrote:
> On Wed, 19 Dec 2007, Dave Hansen wrote:
> > > --- linux-2.6.24-rc5/mm/page_alloc.c~memory-controller-move-to-bug-on-in-free_hot_cold_page     2007-12-19 11:31:46.000000000 +0530
> > > +++ linux-2.6.24-rc5-balbir/mm/page_alloc.c     2007-12-19 11:33:45.000000000 +0530
> > > @@ -995,7 +995,7 @@ static void fastcall free_hot_cold_page(
> > >  
> > >         if (!PageHighMem(page))
> > >                 debug_check_no_locks_freed(page_address(page), PAGE_SIZE);
> > > -       page_assign_page_cgroup(page, NULL);
> > > +       VM_BUG_ON(page_get_page_cgroup(page));
> > >         arch_free_page(page, 0);
> > >         kernel_map_pages(page, 1, 0); 
> > 
> > Hi Balbir,
> > 
> > You generally want to do these like:
> > 
> > 	foo = page_assign_page_cgroup(page, NULL);
> > 	VM_BUG_ON(foo);
> > 
> > Some embedded people have been known to optimize kernel size like this:
> > 
> > 	#define VM_BUG_ON(x) do{}while(0)
> 
> Balbir's patch looks fine to me: I don't get your point there, Dave.

There was a lengthy discussion here:
  http://lkml.org/lkml/2007/12/14/131

on the merit of debug statements with side effects. But looking at our
definition:

#ifdef CONFIG_DEBUG_VM
#define VM_BUG_ON(cond) BUG_ON(cond)
#else
#define VM_BUG_ON(condition) do { } while(0)
#endif

disabling CONFIG_DEBUG_VM breaks the code as proposed by Balbir in that
it will no longer acquire the reference.

Re: [PATCH] Move page_assign_page_cgroup to VM_BUG_ON in free_hot_cold_page

[Hugh Dickins]

On Thu, 20 Dec 2007, Peter Zijlstra wrote:
> On Thu, 2007-12-20 at 13:14 +0000, Hugh Dickins wrote:
> > On Wed, 19 Dec 2007, Dave Hansen wrote:
> > > > -       page_assign_page_cgroup(page, NULL);
> > > > +       VM_BUG_ON(page_get_page_cgroup(page));
> > > 
> > > Hi Balbir,
> > > 
> > > You generally want to do these like:
> > > 
> > > 	foo = page_assign_page_cgroup(page, NULL);
> > > 	VM_BUG_ON(foo);
> > > 
> > > Some embedded people have been known to optimize kernel size like this:
> > > 
> > > 	#define VM_BUG_ON(x) do{}while(0)
> > 
> > Balbir's patch looks fine to me: I don't get your point there, Dave.
> 
> There was a lengthy discussion here:
>   http://lkml.org/lkml/2007/12/14/131
> 
> on the merit of debug statements with side effects.

Of course, but what's the relevance?

> But looking at our definition:
> 
> #ifdef CONFIG_DEBUG_VM
> #define VM_BUG_ON(cond) BUG_ON(cond)
> #else
> #define VM_BUG_ON(condition) do { } while(0)
> #endif
> 
> disabling CONFIG_DEBUG_VM breaks the code as proposed by Balbir in that
> it will no longer acquire the reference.

But what reference?

struct page_cgroup *page_get_page_cgroup(struct page *page)
{
	return (struct page_cgroup *)
		(page->page_cgroup & ~PAGE_CGROUP_LOCK);
}

I guess the issue is that often a "get" function has a complementary
"put" function, but this isn't one of them.  Would page_page_cgroup
be a better name, perhaps?  I don't know.

Hugh

Re: [PATCH] Move page_assign_page_cgroup to VM_BUG_ON in free_hot_cold_page

[Peter Zijlstra]

On Thu, 2007-12-20 at 14:16 +0000, Hugh Dickins wrote:
> On Thu, 20 Dec 2007, Peter Zijlstra wrote:
> > On Thu, 2007-12-20 at 13:14 +0000, Hugh Dickins wrote:
> > > On Wed, 19 Dec 2007, Dave Hansen wrote:
> > > > > -       page_assign_page_cgroup(page, NULL);
> > > > > +       VM_BUG_ON(page_get_page_cgroup(page));
> > > > 
> > > > Hi Balbir,
> > > > 
> > > > You generally want to do these like:
> > > > 
> > > > 	foo = page_assign_page_cgroup(page, NULL);
> > > > 	VM_BUG_ON(foo);
> > > > 
> > > > Some embedded people have been known to optimize kernel size like this:
> > > > 
> > > > 	#define VM_BUG_ON(x) do{}while(0)
> > > 
> > > Balbir's patch looks fine to me: I don't get your point there, Dave.
> > 
> > There was a lengthy discussion here:
> >   http://lkml.org/lkml/2007/12/14/131
> > 
> > on the merit of debug statements with side effects.
> 
> Of course, but what's the relevance?
> 
> > But looking at our definition:
> > 
> > #ifdef CONFIG_DEBUG_VM
> > #define VM_BUG_ON(cond) BUG_ON(cond)
> > #else
> > #define VM_BUG_ON(condition) do { } while(0)
> > #endif
> > 
> > disabling CONFIG_DEBUG_VM breaks the code as proposed by Balbir in that
> > it will no longer acquire the reference.
> 
> But what reference?
> 
> struct page_cgroup *page_get_page_cgroup(struct page *page)
> {
> 	return (struct page_cgroup *)
> 		(page->page_cgroup & ~PAGE_CGROUP_LOCK);
> }
> 
> I guess the issue is that often a "get" function has a complementary
> "put" function, but this isn't one of them.  Would page_page_cgroup
> be a better name, perhaps?  I don't know.

Ah, yes, I mistakenly assumed it was a reference get. In that case I
stand corrected and do not have any objections.

Re: [PATCH] Move page_assign_page_cgroup to VM_BUG_ON in free_hot_cold_page

[Balbir Singh]

Peter Zijlstra wrote:
> On Thu, 2007-12-20 at 14:16 +0000, Hugh Dickins wrote:
>> On Thu, 20 Dec 2007, Peter Zijlstra wrote:
>>> On Thu, 2007-12-20 at 13:14 +0000, Hugh Dickins wrote:
>>>> On Wed, 19 Dec 2007, Dave Hansen wrote:
>>>>>> -       page_assign_page_cgroup(page, NULL);
>>>>>> +       VM_BUG_ON(page_get_page_cgroup(page));
>>>>> Hi Balbir,
>>>>>
>>>>> You generally want to do these like:
>>>>>
>>>>> 	foo = page_assign_page_cgroup(page, NULL);
>>>>> 	VM_BUG_ON(foo);
>>>>>
>>>>> Some embedded people have been known to optimize kernel size like this:
>>>>>
>>>>> 	#define VM_BUG_ON(x) do{}while(0)
>>>> Balbir's patch looks fine to me: I don't get your point there, Dave.
>>> There was a lengthy discussion here:
>>>   http://lkml.org/lkml/2007/12/14/131
>>>
>>> on the merit of debug statements with side effects.
>> Of course, but what's the relevance?
>>
>>> But looking at our definition:
>>>
>>> #ifdef CONFIG_DEBUG_VM
>>> #define VM_BUG_ON(cond) BUG_ON(cond)
>>> #else
>>> #define VM_BUG_ON(condition) do { } while(0)
>>> #endif
>>>
>>> disabling CONFIG_DEBUG_VM breaks the code as proposed by Balbir in that
>>> it will no longer acquire the reference.
>> But what reference?
>>
>> struct page_cgroup *page_get_page_cgroup(struct page *page)
>> {
>> 	return (struct page_cgroup *)
>> 		(page->page_cgroup & ~PAGE_CGROUP_LOCK);
>> }
>>
>> I guess the issue is that often a "get" function has a complementary
>> "put" function, but this isn't one of them.  Would page_page_cgroup
>> be a better name, perhaps?  I don't know.
> 
> Ah, yes, I mistakenly assumed it was a reference get. In that case I
> stand corrected and do not have any objections.
> 

I was going to say the same thing, page_get_page_cgroup() does not hold
any references. May be _get_ in the name is confusing.


-- 
	Warm Regards,
	Balbir Singh
	Linux Technology Center
	IBM, ISTL

Re: [PATCH] Move page_assign_page_cgroup to VM_BUG_ON in free_hot_cold_page

[Dave Hansen]

On Thu, 2007-12-20 at 21:54 +0530, Balbir Singh wrote:
> 
> I was going to say the same thing, page_get_page_cgroup() does not hold
> any references. May be _get_ in the name is confusing. 

OK, you three had the entire conversation outing me before I even fot to
respond! :)

Yeah, I thought it was a refcount acquisition.  Balbir, it's fine the
way it stands (except for the slightly confusing name).

-- Dave

Re: [PATCH] Move page_assign_page_cgroup to VM_BUG_ON in free_hot_cold_page

[Andrew Morton]

On Thu, 20 Dec 2007 21:54:15 +0530 Balbir Singh <balbir@linux.vnet.ibm.com> wrote:

> >> struct page_cgroup *page_get_page_cgroup(struct page *page)
> >> {
> >> 	return (struct page_cgroup *)
> >> 		(page->page_cgroup & ~PAGE_CGROUP_LOCK);
> >> }
> >>
> >> I guess the issue is that often a "get" function has a complementary
> >> "put" function, but this isn't one of them.  Would page_page_cgroup
> >> be a better name, perhaps?  I don't know.
> > 
> > Ah, yes, I mistakenly assumed it was a reference get. In that case I
> > stand corrected and do not have any objections.
> > 
> 
> I was going to say the same thing, page_get_page_cgroup() does not hold
> any references. May be _get_ in the name is confusing.

It is a bit unconventional.  page_cgroup()?

Re: [PATCH] Move page_assign_page_cgroup to VM_BUG_ON in free_hot_cold_page

[Hugh Dickins]

On Thu, 20 Dec 2007, Andrew Morton wrote:
> On Thu, 20 Dec 2007 21:54:15 +0530 Balbir Singh <balbir@linux.vnet.ibm.com> wrote:
> > I was going to say the same thing, page_get_page_cgroup() does not hold
> > any references. May be _get_ in the name is confusing.
> 
> It is a bit unconventional.  page_cgroup()?

Seems ideal to me
(though there is some argument for page_page_cgroup(), weird as it is).

Hugh

Re: [PATCH] Move page_assign_page_cgroup to VM_BUG_ON in free_hot_cold_page

[Balbir Singh]

Hugh Dickins wrote:
> On Thu, 20 Dec 2007, Andrew Morton wrote:
>> On Thu, 20 Dec 2007 21:54:15 +0530 Balbir Singh <balbir@linux.vnet.ibm.com> wrote:
>>> I was going to say the same thing, page_get_page_cgroup() does not hold
>>> any references. May be _get_ in the name is confusing.
>> It is a bit unconventional.  page_cgroup()?
> 
> Seems ideal to me
> (though there is some argument for page_page_cgroup(), weird as it is).
> 
> Hugh

Sounds good to me as well, except that it sounds like a constructor :-)
I'll try and make the changes and submit a patch. I am in the middle of
losing control_type now (based on Hugh's comments)

-- 
	Warm Regards,
	Balbir Singh
	Linux Technology Center
	IBM, ISTL