Skip to contentSkip to conversation

mm: fix huge page reallocated in soft_offline_page

PATCH In review Xishi Qiu 1 participant 2 messages Dec 16, 2013

Subsystems (2)
Memory management HWPoison memory failure handling
lore
Share & tools
lore

Xishi Qiu

author
#0 lore raw

PATCH 1/1 mm: fix huge page reallocated in soft_offline_page

The huge page may be reallocated in soft_offline_page, because MIGRATE_ISOLATE can not keep the page until after setting PG_hwpoison. alloc_huge_page()
dequeue_huge_page_vma()
dequeue_huge_page_node()
If the huge page was reallocated, we need to try offline it again.

Signed-off-by Xishi Qiu <[email protected]>
---
 mm/memory-failure.c |   21 ++++++++++++++++++---
 1 files changed, 18 insertions(+), 3 deletions(-)
mm/memory-failure.c +18 -3
index b7c1716..f384249 100644
@@ -1505,8 +1505,11 @@ static int soft_offline_huge_page(struct page *page, int flags)
 		if (ret > 0)
 			ret = -EIO;
 	} else {
+		ret = dequeue_hwpoisoned_huge_page(hpage);
+		/* If the page was reallocated, we need to try again. */
+		if (ret)
+			return -EAGAIN;
 		set_page_hwpoison_huge_page(hpage);
-		dequeue_hwpoisoned_huge_page(hpage);
 		atomic_long_add(1 << compound_order(hpage),
 				&num_poisoned_pages);
 	}

@@ -1624,10 +1627,11 @@ static int __soft_offline_page(struct page *page, int flags)
  */
 int soft_offline_page(struct page *page, int flags)
 {
-	int ret;
+	int ret, retry_max = 3;
 	unsigned long pfn = page_to_pfn(page);
 	struct page *hpage = compound_trans_head(page);
 
+retry:
 	if (PageHWPoison(page)) {
 		pr_info("soft offline: %#lx page already poisoned\n", pfn);
 		return -EBUSY;

@@ -1663,8 +1667,15 @@ int soft_offline_page(struct page *page, int flags)
 			ret = __soft_offline_page(page, flags);
 	} else if (ret == 0) { /* for free pages */
 		if (PageHuge(page)) {
+			ret = dequeue_hwpoisoned_huge_page(hpage);
+			/* If the page was reallocated, we need to try again. */
+			if (ret) {
+				unset_migratetype_isolate(page,
+						MIGRATE_MOVABLE);
+				if (retry_max-- > 0)
+					goto retry;
+			}
 			set_page_hwpoison_huge_page(hpage);
-			dequeue_hwpoisoned_huge_page(hpage);
 			atomic_long_add(1 << compound_order(hpage),
 					&num_poisoned_pages);
 		} else {

@@ -1673,5 +1684,9 @@ int soft_offline_page(struct page *page, int flags)
 		}
 	}
 	unset_migratetype_isolate(page, MIGRATE_MOVABLE);
+
+	if (ret == -EAGAIN && retry_max-- > 0)
+		goto retry;
+
 	return ret;
 }
Signature · 8 lines
1.7.1


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
author
#1 lore raw
Quoted 7 lines · Xishi Qiu

The huge page may be reallocated in soft_offline_page, because MIGRATE_ISOLATE can not keep the page until after setting PG_hwpoison. alloc_huge_page()
dequeue_huge_page_vma()
dequeue_huge_page_node()
If the huge page was reallocated, we need to try offline it again.

Sorry, I made a mistake.
This bug was fixed in
commit c8721bbbdd36382de51cd6b7a56322e0acca2414

Thanks,
Xishi Qiu

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>