lkml.org 
[lkml]   [2010]   [Sep]   [7]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [BUGFIX][PATCH 1/3] memory hotplug: fix next block calculation in is_removable
On Tue, Sep 07, 2010 at 02:15:01AM +0900, Hiroyuki Kamezawa wrote:
> 2010/9/6 Mel Gorman <mel@csn.ul.ie>:
> > On Mon, Sep 06, 2010 at 02:42:28PM +0900, KAMEZAWA Hiroyuki wrote:
> >>
> >> From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> >>
> >> next_active_pageblock() is for finding next _used_ freeblock. It skips
> >> several blocks when it finds there are a chunk of free pages lager than
> >> pageblock. But it has 2 bugs.
> >>
> >>   1. We have no lock. page_order(page) - pageblock_order can be minus.
> >>   2. pageblocks_stride += is wrong. it should skip page_order(p) of pages.
> >>
> >> CC: stable@kernel.org
> >> Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> >> ---
> >>  mm/memory_hotplug.c |   16 ++++++++--------
> >>  1 file changed, 8 insertions(+), 8 deletions(-)
> >>
> >> Index: kametest/mm/memory_hotplug.c
> >> ===================================================================
> >> --- kametest.orig/mm/memory_hotplug.c
> >> +++ kametest/mm/memory_hotplug.c
> >> @@ -584,19 +584,19 @@ static inline int pageblock_free(struct
> >>  /* Return the start of the next active pageblock after a given page */
> >>  static struct page *next_active_pageblock(struct page *page)
> >>  {
> >> -     int pageblocks_stride;
> >> -
> >>       /* Ensure the starting page is pageblock-aligned */
> >>       BUG_ON(page_to_pfn(page) & (pageblock_nr_pages - 1));
> >>
> >> -     /* Move forward by at least 1 * pageblock_nr_pages */
> >> -     pageblocks_stride = 1;
> >> -
> >>       /* If the entire pageblock is free, move to the end of free page */
> >> -     if (pageblock_free(page))
> >> -             pageblocks_stride += page_order(page) - pageblock_order;
> >> +     if (pageblock_free(page)) {
> >> +             int order;
> >> +             /* be careful. we don't have locks, page_order can be changed.*/
> >> +             order = page_order(page);
> >> +             if (order > pageblock_order)
> >> +                     return page + (1 << order);
> >> +     }
> >
> > As you note in your changelog, page_order() is unsafe because we do not have
> > the zone lock but you don't check if order is somewhere between pageblock_order
> > and MAX_ORDER_NR_PAGES. How is this safer?
> >
> Ah, I missed that.
>
> if ((pageblock_order <= order) && (order < MAX_ORDER))
> return page + (1 << order);
> ok ?
>

Seems ok. There will still be some false usage of order but it should be
harmless.

--
Mel Gorman
Part-time Phd Student Linux Technology Center
University of Limerick IBM Dublin Software Lab
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2010-09-07 11:33    [W:0.048 / U:2.456 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site