linux-kernel - Re: [patch 5/5]thp: split huge page if head page is isolated

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: <20111031082440.GB21440@barrios-laptop.redhat.com>
Date:	Mon, 31 Oct 2011 17:24:40 +0900
From:	Minchan Kim <minchan.kim@...il.com>
To:	Shaohua Li <shaohua.li@...el.com>
Cc:	Andrew Morton <akpm@...ux-foundation.org>,
	"aarcange@...hat.com" <aarcange@...hat.com>,
	Hugh Dickins <hughd@...gle.com>,
	Rik van Riel <riel@...hat.com>, mel <mel@....ul.ie>,
	KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>,
	linux-mm <linux-mm@...ck.org>,
	lkml <linux-kernel@...r.kernel.org>
Subject: Re: [patch 5/5]thp: split huge page if head page is isolated

On Mon, Oct 31, 2011 at 09:10:49AM +0800, Shaohua Li wrote:
> On Fri, 2011-10-28 at 17:50 +0800, Minchan Kim wrote:
> > On Fri, Oct 28, 2011 at 04:25:56PM +0800, Shaohua Li wrote:
> > > On Fri, 2011-10-28 at 15:30 +0800, Minchan Kim wrote:
> > > > On Fri, Oct 28, 2011 at 01:11:55PM +0800, Shaohua Li wrote:
> > > > > On Fri, 2011-10-28 at 07:34 +0800, Minchan Kim wrote:
> > > > > > On Tue, Oct 25, 2011 at 10:59:40AM +0800, Shaohua Li wrote:
> > > > > > > With current logic, if page reclaim finds a huge page, it will just reclaim
> > > > > > > the head page and leave tail pages reclaimed later. Let's take an example,
> > > > > > > lru list has page A and B, page A is huge page:
> > > > > > > 1. page A is isolated
> > > > > > > 2. page B is isolated
> > > > > > > 3. shrink_page_list() adds page A to swap page cache. so page A is split.
> > > > > > > page A+1, page A+2, ... are added to lru list.
> > > > > > > 4. shrink_page_list() adds page B to swap page cache.
> > > > > > > 5. page A and B is written out and reclaimed.
> > > > > > > 6. page A+1, A+2 ... is isolated and reclaimed later.
> > > > > > > So the reclaim order is A, B, ...(maybe other pages), A+1, A+2 ...
> > > > > > 
> > > > > > I don't see your code yet but have a question.
> > > > > > You mitigate this problem by 4/5 which could add subpages into lru tail
> > > > > > so subpages would reclaim next interation of reclaim.
> > > > > > 
> > > > > > What do we need 5/5?
> > > > > > Do I miss something?
> > > > > Both patches are required. without this patch, current page reclaim will
> > > > > only reclaim the first page of a huge page, because the hugepage isn't
> > > > > split yet. The hugepage is split when the first page is being written to
> > > > > swap, which is too later and page reclaim might already isolated a lot
> > > > > of pages.
> > > > 
> > > > When split happens, subpages would be located in tail of LRU by your 4/5.
> > > > (Assume tail of LRU is old age).
> > > yes, but a lot of other pages already isolated. we will reclaim those
> > > pages first. for example, reclaim huge page A, B. current reclaim order
> > > is A, B, A+1, ... B+1, because we will isolated A and B first, all tail
> > > pages are not isolated yet. While with my patch, the order is A, A
> > > +1, ... B, B+1,.... with my patch, we can avoid unnecessary page split
> > > or page isolation. This is exactly why my patch reduces the thp_split
> > > count.
> > 
> > It's possbile but I doubt how it is effective becuase add_to_swap has a unlikely as follows
> > 
> > 	if (unlikely(PageTransHuge(page)))
> > 
> > I don't mean unlikely assumption is absolutely right.
> > But at least, you have to convince us of it's wrong.
> > Personally, I don't want to add more logic and handling THP pages
> > different with normal page unless it's real concern.
> if you actually use THP, you will find it's a problem. The data I posted
> already clearly showed it.
> 

If so, could you fix above thing in next iteration if you don't mind?

Thanks.

-- 
Kind regards,
Minchan Kim
--
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/