From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([208.118.235.92]:50179) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1TGfMD-00077Y-FZ for qemu-devel@nongnu.org; Tue, 25 Sep 2012 20:20:06 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1TGfMB-0002Fm-QB for qemu-devel@nongnu.org; Tue, 25 Sep 2012 20:20:05 -0400 Received: from lgemrelse6q.lge.com ([156.147.1.121]:42650) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1TGfMB-00029f-Dm for qemu-devel@nongnu.org; Tue, 25 Sep 2012 20:20:03 -0400 Date: Wed, 26 Sep 2012 09:23:06 +0900 From: Minchan Kim Message-ID: <20120926002306.GB7759@bbox> References: <1348224383-1499-1-git-send-email-mgorman@suse.de> <1348224383-1499-6-git-send-email-mgorman@suse.de> <20120925070517.GK13234@bbox> <20120925075105.GC11266@suse.de> <20120925081327.GA7759@bbox> <20120925143931.f404ca22.akpm@linux-foundation.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20120925143931.f404ca22.akpm@linux-foundation.org> Subject: Re: [Qemu-devel] [PATCH 5/9] mm: compaction: Acquire the zone->lru_lock as late as possible List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Andrew Morton Cc: Richard Davies , KVM , QEMU-devel , LKML , Linux-MM , Mel Gorman , Shaohua Li , Avi Kivity On Tue, Sep 25, 2012 at 02:39:31PM -0700, Andrew Morton wrote: > On Tue, 25 Sep 2012 17:13:27 +0900 > Minchan Kim wrote: > > > I see. To me, your saying is better than current comment. > > I hope comment could be more explicit. > > > > diff --git a/mm/compaction.c b/mm/compaction.c > > index df01b4e..f1d2cc7 100644 > > --- a/mm/compaction.c > > +++ b/mm/compaction.c > > @@ -542,8 +542,9 @@ isolate_migratepages_range(struct zone *zone, struct compact_control *cc, > > * splitting and collapsing (collapsing has already happened > > * if PageLRU is set) but the lock is not necessarily taken > > * here and it is wasteful to take it just to check transhuge. > > - * Check transhuge without lock and skip if it's either a > > - * transhuge or hugetlbfs page. > > + * Check transhuge without lock and *skip* if it's either a > > + * transhuge or hugetlbfs page because it's not safe to call > > + * compound_order. > > */ > > if (PageTransHuge(page)) { > > if (!locked) > > Going a bit further: > > --- a/mm/compaction.c~mm-compaction-acquire-the-zone-lru_lock-as-late-as-possible-fix > +++ a/mm/compaction.c > @@ -415,7 +415,8 @@ isolate_migratepages_range(struct zone * > * if PageLRU is set) but the lock is not necessarily taken > * here and it is wasteful to take it just to check transhuge. > * Check transhuge without lock and skip if it's either a > - * transhuge or hugetlbfs page. > + * transhuge or hugetlbfs page because calling compound_order() > + * requires lru_lock to exclude isolation and splitting. > */ > if (PageTransHuge(page)) { > if (!locked) > _ > > > but... the requirement to hold lru_lock for compound_order() is news > to me. It doesn't seem to be written down or explained anywhere, and > one wonders why the cheerily undocumented compound_lock() doesn't have > this effect. What's going on here?? First of all, I don't know why we should mention hugetlbfs in comment. I don't know hugetlbfs well so I had a time to look through code but can't find a place setting PG_lru so I'm not sure hugetlbfs page can reach on this code. Please correct me if I was wrong. On THP, I think compound_lock you mentioned is okay, too but I think it's sort of optimization because we don't need both lru_lock and compound_lock. If we hold lru_lock, we can't prevent race with __split_huge_page_refcount so that the page couldn't be freed. Namely, it's safe to call compound_order. > > -- > 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: email@kvack.org -- Kind regards, Minchan Kim