Re: [PATCH -mmotm 7/8] memcg: move charges of anonymous swap

[Balbir Singh <balbir@linux.vnet.ibm.com>]

* KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> [2010-02-04 16:44:41]:

> On Thu, 4 Feb 2010 16:18:40 +0900
> Paul Mundt <lethal@linux-sh.org> wrote:
> 
> > On Thu, Feb 04, 2010 at 02:27:36PM +0900, KAMEZAWA Hiroyuki wrote:
> 
> > > I think memcg should depends on CONIFG_MMU.
> > > 
> > > How do you think ?
> > > 
> > Unless there's a real technical reason to make it depend on CONFIG_MMU,
> > that's just papering over the problem, and means that some nommu person
> > will have to come back and fix it properly at a later point in time.
> > 
> I have no strong opinion this. It's ok to support as much as possible.
> My concern is that there is no !MMU architecture developper around memcg. So,
> error report will be delayed.
>

I don't mind making it depend on CONFIG_MMU, enabling it for
!CONFIG_MMU will require some careful thought and work, so I'd rather
make that obvious by making it depend on CONFIG_MMU
 
> 
> > CONFIG_SWAP itself is configurable even with CONFIG_MMU=y, so having
> > stubbed out helpers for the CONFIG_SWAP=n case would give the compiler a
> > chance to optimize things away in those cases, too. Embedded systems
> > especially will often have MMU=y and BLOCK=n, resulting in SWAP being
> > unset but swap cache encodings still defined.
> > 
> > How about just changing the is_swap_pte() definition to depend on SWAP
> > instead?
> > 
> I think the new feature as "move task charge" itself depends on CONFIG_MMU
> because it walks a process's page table. 
> 
-- 
	Balbir

--
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/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>