Re: [PATCH] Make is_mem_section_removable more conformable with offlining code

[Michal Hocko <mhocko@suse.cz>]

On Thu 02-09-10 23:19:18, Hiroyuki Kamezawa wrote:
[...]
> > diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> > index 6e6e626..0bd941b 100644
> > --- a/include/linux/mmzone.h
> > +++ b/include/linux/mmzone.h
> > @@ -669,6 +669,30 @@ unsigned long __init node_memmap_size_bytes(int, unsigned long, unsigned long);
> > ?*/
> > ?#define zone_idx(zone) ? ? ? ? ((zone) - (zone)->zone_pgdat->node_zones)
> >
> > +#ifdef CONFIG_MEMORY_HOTREMOVE
> > +/*
> > + * A free or LRU pages block are removable
> > + * Do not use MIGRATE_MOVABLE because it can be insufficient and
> > + * other MIGRATE types are tricky.
> > + */
> > +static inline bool is_page_removable(struct page *page)
> > +{
> > + ? ? ? int page_block = 1 << pageblock_order;
> > + ? ? ? for (page_block > 0) {
> 
> for ?

Bahh. The old and half backed patch. See the up-to-date one bellow.

> > + ? ? ? ? ? ? ? if (PageBuddy(page)) {
> > + ? ? ? ? ? ? ? ? ? ? ? page_block -= page_order(page);
> > + ? ? ? ? ? ? ? }else if (PageLRU(page))
> > + ? ? ? ? ? ? ? ? ? ? ? page_block--;
> > + ? ? ? ? ? ? ? else
> > + ? ? ? ? ? ? ? ? ? ? ? return false;
> > + ? ? ? }
> > +
> > + ? ? ? return true;
> > +}
> 
> Hmm. above for is intending to check all pages in the block ?

Yes, by their orders.

> I'll look into details, tomorrow.

Thanks!

---
>From 0a8650594280446ecfa2ae8b609f25877801c55d Mon Sep 17 00:00:00 2001
From: Michal Hocko <mhocko@suse.cz>
Date: Fri, 20 Aug 2010 15:39:16 +0200
Subject: [PATCH] Make is_mem_section_removable more conformable with offlining code

Currently is_mem_section_removable checks whether each pageblock from
the given pfn range is of MIGRATE_MOVABLE type or if it is free. If both
are false then the range is considered non removable.

On the other hand, offlining code (more specifically
set_migratetype_isolate) doesn't care whether a page is free and instead
it just checks the migrate type of the page and whether the page's zone
is movable.

This can lead into a situation when we can mark a node as not removable
just because a pageblock is MIGRATE_RESERVE and it is not free but still
movable.

Let's make a common helper is_page_removable which unifies both tests
at one place.

Do not rely on any of MIGRATE_* types as all others but MIGRATE_MOVABLE
may be tricky. MIGRATE_RESERVE can be anything that just happened to
fallback to that allocation. MIGRATE_RECLAIMABLE can be unmovable
because slab (or what ever) has this page currently in use and cannot
release it.  If we tried to remove those pages and the isolation failed
then those blocks would get into the MIRAGTE_MOVABLE list
unconditionally and we will end up having unmovable pages in the movable
list.

Let's, instead, check just whether a pageblock contains only free or LRU
pages.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
---
 include/linux/mmzone.h |   24 ++++++++++++++++++++++++
 mm/memory_hotplug.c    |   19 +------------------
 mm/page_alloc.c        |    5 +----
 3 files changed, 26 insertions(+), 22 deletions(-)

diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index 6e6e626..f64297e 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -669,6 +669,30 @@ unsigned long __init node_memmap_size_bytes(int, unsigned long, unsigned long);
  */
 #define zone_idx(zone)		((zone) - (zone)->zone_pgdat->node_zones)
 
+#ifdef CONFIG_MEMORY_HOTREMOVE
+/*
+ * A free or LRU pages block are removable
+ * Do not use MIGRATE_MOVABLE because it can be insufficient and
+ * other MIGRATE types are tricky.
+ */
+static inline bool is_page_removable(struct page *page)
+{
+	int page_block = 1 << pageblock_order;
+	while (page_block > 0) {
+		if (PageBuddy(page)) {
+			page_block -= page_order(page);
+		}else if (PageLRU(page))
+			page_block--;
+		else 
+			return false;
+	}
+
+	return true;
+}
+#else
+#define is_page_removable(p) 0
+#endif
+
 static inline int populated_zone(struct zone *zone)
 {
 	return (!!zone->present_pages);
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index a4cfcdc..66195b8 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -569,17 +569,6 @@ out:
 EXPORT_SYMBOL_GPL(add_memory);
 
 #ifdef CONFIG_MEMORY_HOTREMOVE
-/*
- * A free page on the buddy free lists (not the per-cpu lists) has PageBuddy
- * set and the size of the free page is given by page_order(). Using this,
- * the function determines if the pageblock contains only free pages.
- * Due to buddy contraints, a free page at least the size of a pageblock will
- * be located at the start of the pageblock
- */
-static inline int pageblock_free(struct page *page)
-{
-	return PageBuddy(page) && page_order(page) >= pageblock_order;
-}
 
 /* Return the start of the next active pageblock after a given page */
 static struct page *next_active_pageblock(struct page *page)
@@ -608,13 +597,7 @@ int is_mem_section_removable(unsigned long start_pfn, unsigned long nr_pages)
 
 	/* Check the starting page of each pageblock within the range */
 	for (; page < end_page; page = next_active_pageblock(page)) {
-		type = get_pageblock_migratetype(page);
-
-		/*
-		 * A pageblock containing MOVABLE or free pages is considered
-		 * removable
-		 */
-		if (type != MIGRATE_MOVABLE && !pageblock_free(page))
+		if (!is_page_removable(page))
 			return 0;
 
 		/*
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index a9649f4..c2e2576 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -5277,14 +5277,11 @@ int set_migratetype_isolate(struct page *page)
 	struct memory_isolate_notify arg;
 	int notifier_ret;
 	int ret = -EBUSY;
-	int zone_idx;
 
 	zone = page_zone(page);
-	zone_idx = zone_idx(zone);
 
 	spin_lock_irqsave(&zone->lock, flags);
-	if (get_pageblock_migratetype(page) == MIGRATE_MOVABLE ||
-	    zone_idx == ZONE_MOVABLE) {
+	if (is_page_removable(page)) {
 		ret = 0;
 		goto out;
 	}
-- 
1.7.1


-- 
Michal Hocko
L3 team 
SUSE LINUX s.r.o.
Lihovarska 1060/12
190 00 Praha 9    
Czech Republic