Re: [PATCH 22/24] HWPOISON: add memory cgroup filter

[Wu Fengguang]

On Thu, Dec 03, 2009 at 09:52:29AM +0800, KAMEZAWA Hiroyuki wrote:
> On Wed, 2 Dec 2009 20:58:42 +0800
> Wu Fengguang <fengguang.wu@xxxxxxxxx> wrote:
> 
> > On Wed, Dec 02, 2009 at 08:44:46PM +0800, Andi Kleen wrote:
> > > >  
> > > > +static int hwpoison_filter_task(struct page *p)
> > > > +{
> > > 
> > > Can we make that ifdef instead of depends on ?
> > 
> > Sure. Here is the updated patch.
> > 
> > ---
> > HWPOISON: add memory cgroup filter
> > 
> > The hwpoison test suite need to inject hwpoison to a collection of
> > selected task pages, and must not touch pages not owned by them and
> > thus kill important system processes such as init. (But it's OK to
> > mis-hwpoison free/unowned pages as well as shared clean pages.
> > Mis-hwpoison of shared dirty pages will kill all tasks, so the test
> > suite will target all or non of such tasks in the first place.)
> > 
> > The memory cgroup serves this purpose well. We can put the target
> > processes under the control of a memory cgroup, and tell the hwpoison
> > injection code to only kill pages associated with some active memory
> > cgroup.
> > 
> > The prerequisite for doing hwpoison stress tests with mem_cgroup is,
> > the mem_cgroup code tracks task pages _accurately_ (unless page is
> > locked).  Which we believe is/should be true.
> > 
> > The benifits are simplification of hwpoison injector code. Also the
> > mem_cgroup code will automatically be tested by hwpoison test cases.
> > 
> > The alternative interfaces pin-pfn/unpin-pfn can also delegate the
> > (process and page flags) filtering functions reliably to user space.
> > However prototype implementation shows that this scheme adds more
> > complexity than we wanted.
> > 
> > CC: KOSAKI Motohiro <kosaki.motohiro@xxxxxxxxxxxxxx>
> > CC: Hugh Dickins <hugh.dickins@xxxxxxxxxxxxx>
> > CC: Daisuke Nishimura <nishimura@xxxxxxxxxxxxxxxxx>
> > CC: Balbir Singh <balbir@xxxxxxxxxxxxxxxxxx>
> > CC: KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx>
> > CC: Li Zefan <lizf@xxxxxxxxxxxxxx>
> > CC: Paul Menage <menage@xxxxxxxxxx>
> > CC: Nick Piggin <npiggin@xxxxxxx> 
> > CC: Andi Kleen <andi@xxxxxxxxxxxxxx> 
> > Signed-off-by: Wu Fengguang <fengguang.wu@xxxxxxxxx>
> > ---
> >  mm/Kconfig           |    2 +-
> >  mm/hwpoison-inject.c |    7 +++++++
> >  mm/internal.h        |    1 +
> >  mm/memory-failure.c  |   28 ++++++++++++++++++++++++++++
> >  4 files changed, 37 insertions(+), 1 deletion(-)
> > 
> > --- linux-mm.orig/mm/memory-failure.c	2009-12-01 09:56:06.000000000 +0800
> > +++ linux-mm/mm/memory-failure.c	2009-12-02 20:56:55.000000000 +0800
> > @@ -96,6 +96,31 @@ static int hwpoison_filter_flags(struct 
> >  		return -EINVAL;
> >  }
> >  
> > +#ifdef	CONFIG_CGROUP_MEM_RES_CTLR_SWAP
> > +u32 hwpoison_filter_memcg;
> > +static int hwpoison_filter_task(struct page *p)
> > +{
> > +	struct mem_cgroup *mem;
> > +	struct cgroup_subsys_state *css;
> > +
> > +	if (!hwpoison_filter_memcg)
> > +		return 0;
> > +
> > +	mem = try_get_mem_cgroup_from_page(p);
> > +	if (!mem)
> > +		return -EINVAL;
> > +
> > +	css = mem_cgroup_css(mem);
> > +	if (!css)
> > +		return -EINVAL;
> 
> > +
> > +	css_put(css);
> > +	return 0;
> > +}
> 
> 
> Hmm..can you adds comment ? What does this function is for ?

Good idea. How about this one?

/*
 * This allows stress tests to limit test scope to a collection of tasks
 * by putting them under some memcg. This prevents killing unrelated/important
 * processes such as /sbin/init. Note that the target task may share clean
 * pages with init (eg. libc text), which is harmless. If the target task
 * share _dirty_ pages with another task B, the test scheme must make sure B
 * is also included in the memcg. At last, due to race conditions this filter
 * can only guarantee that the page either belongs to the memcg tasks, or is
 * a freed page.
 */

> Is this more meaningful than PageLRU(page) etc..?

It's mainly for stress testing (randomly killing pages of many tasks
until all of them get killed, and see if it impacts the health of the
whole system).

It could be used in combination with the page flags filter to do more
oriented tests.

A task may map some non-LRU page - eg. the vdso page.

Thanks,
Fengguang
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/