linux-kernel - Re: [PATCH 3/3] mm/memcg: apply add/del

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <alpine.LSU.2.00.1205311544400.4561@eggly.anvils>
Date:	Thu, 31 May 2012 15:58:59 -0700 (PDT)
From:	Hugh Dickins <hughd@...gle.com>
To:	baozich <baozich@...il.com>
cc:	Andrew Morton <akpm@...ux-foundation.org>,
	Konstantin Khlebnikov <khlebnikov@...nvz.org>,
	KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>,
	Johannes Weiner <hannes@...xchg.org>,
	Michal Hocko <mhocko@...e.cz>, linux-mm@...ck.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH 3/3] mm/memcg: apply add/del_page to lruvec

On Thu, 31 May 2012, baozich wrote:
> > 
> > In their place, mem_cgroup_page_lruvec() to decide the lruvec,
> > previously a side-effect of add, and mem_cgroup_update_lru_size()
> > to maintain the lru_size stats.
> I have a stupid question. I'm not sure whether there is reduplication
> to put both "page" and "zone" parameter in mem_cgroup_page_lruvec(),
> for I noticed that the "struct zone *zone" parameter are usually from 
> page_zone(page) in most cases. I think that the semantics of this function
> is to grab the lruvec the page belongs to. So will it be ok if we pass
> only "page" as the parameter, which I think would be cleaner? Please
> fix me if I missed something.

I share your dislike for passing down an "unnecessary" argument,
but I do think it's justified here.

If the zone pointer were available simply by page->zone, then yes,
I'd agree with you that it's probably silly to pass zone separately.

But page_zone(page) is never as trivial as that, and on some memory
layouts it can be a lookup that you'd really prefer to avoid repeating.

In every(?) case where we're using mem_cgroup_page_lruvec(), the zone
is already known: it's just been used for spin_lock_irq(&zone->lru_lock).

And when CONFIG_CGROUP_MEM_RES_CTLR is not set, the inline function
mem_cgroup_page_lruvec() uses only zone, not page at all: I wouldn't
want to be slowing down that case with another page_zone(page) lookup.

Also it's somewhat academic (though not for v3.5), in that this function
goes away in the patches I build upon it; and I expect it also to go away
in the patches Konstantin would build upon it - mem_cgroup_page_lruvec()
is a staging point, before we combine memcg/zone lookup with the locking.

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