mm, page_owner: copy page owner info during migration

The page_owner mechanism stores gfp_flags of an allocation and stack
trace that lead to it.  During page migration, the original information
is practically replaced by the allocation of free page as the migration
target.  Arguably this is less useful and might lead to all the
page_owner info for migratable pages gradually converge towards
compaction or numa balancing migrations.  It has also lead to
inaccuracies such as one fixed by commit e2cfc91120fa ("mm/page_owner:
set correct gfp_mask on page_owner").

This patch thus introduces copying the page_owner info during migration.
However, since the fact that the page has been migrated from its
original place might be useful for debugging, the next patch will
introduce a way to track that information as well.

Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Sasha Levin <sasha.levin@oracle.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/include/linux/page_owner.h b/include/linux/page_owner.h
index 8e2eb15..6440daa 100644 (file)
--- a/include/linux/page_owner.h
+++ b/include/linux/page_owner.h
@@ -11,6 +11,7 @@ extern void __reset_page_owner(struct page *page, unsigned int order);
 extern void __set_page_owner(struct page *page,
                        unsigned int order, gfp_t gfp_mask);
 extern gfp_t __get_page_owner_gfp(struct page *page);
+extern void __copy_page_owner(struct page *oldpage, struct page *newpage);
 
 static inline void reset_page_owner(struct page *page, unsigned int order)
 {
@@ -32,6 +33,11 @@ static inline gfp_t get_page_owner_gfp(struct page *page)
        else
                return 0;
 }
+static inline void copy_page_owner(struct page *oldpage, struct page *newpage)
+{
+       if (static_branch_unlikely(&page_owner_inited))
+               __copy_page_owner(oldpage, newpage);
+}
 #else
 static inline void reset_page_owner(struct page *page, unsigned int order)
 {
@@ -44,6 +50,8 @@ static inline gfp_t get_page_owner_gfp(struct page *page)
 {
        return 0;
 }
-
+static inline void copy_page_owner(struct page *oldpage, struct page *newpage)
+{
+}
 #endif /* CONFIG_PAGE_OWNER */
 #endif /* __LINUX_PAGE_OWNER_H */

diff --git a/mm/migrate.c b/mm/migrate.c
index 3ad0fea..8133805 100644 (file)
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -38,6 +38,7 @@
 #include <linux/balloon_compaction.h>
 #include <linux/mmu_notifier.h>
 #include <linux/page_idle.h>
+#include <linux/page_owner.h>
 
 #include <asm/tlbflush.h>
 
@@ -578,6 +579,8 @@ void migrate_page_copy(struct page *newpage, struct page *page)
         */
        if (PageWriteback(newpage))
                end_page_writeback(newpage);
+
+       copy_page_owner(page, newpage);
 }
 
 /************************************************************

diff --git a/mm/page_owner.c b/mm/page_owner.c
index feaa28b..774b556 100644 (file)
--- a/mm/page_owner.c
+++ b/mm/page_owner.c
@@ -84,6 +84,31 @@ gfp_t __get_page_owner_gfp(struct page *page)
        return page_ext->gfp_mask;
 }
 
+void __copy_page_owner(struct page *oldpage, struct page *newpage)
+{
+       struct page_ext *old_ext = lookup_page_ext(oldpage);
+       struct page_ext *new_ext = lookup_page_ext(newpage);
+       int i;
+
+       new_ext->order = old_ext->order;
+       new_ext->gfp_mask = old_ext->gfp_mask;
+       new_ext->nr_entries = old_ext->nr_entries;
+
+       for (i = 0; i < ARRAY_SIZE(new_ext->trace_entries); i++)
+               new_ext->trace_entries[i] = old_ext->trace_entries[i];
+
+       /*
+        * We don't clear the bit on the oldpage as it's going to be freed
+        * after migration. Until then, the info can be useful in case of
+        * a bug, and the overal stats will be off a bit only temporarily.
+        * Also, migrate_misplaced_transhuge_page() can still fail the
+        * migration and then we want the oldpage to retain the info. But
+        * in that case we also don't need to explicitly clear the info from
+        * the new page, which will be freed.
+        */
+       __set_bit(PAGE_EXT_OWNER, &new_ext->flags);
+}
+
 static ssize_t
 print_page_owner(char __user *buf, size_t count, unsigned long pfn,
                struct page *page, struct page_ext *page_ext)