memcg: migration clean up

Now memcg in unmap_and_move checks BUG_ON of charge. But mem_cgroup_prepare_migration returns either 0 or -ENOMEM. If it returns -ENOMEM, it jumps out unlock without the check. If it returns 0, it can pass BUG_ON. So it's meaningless. Let's remove it.

Cc Johannes Weiner <[email protected]>

Cc Balbir Singh <[email protected]>

Acked-by Daisuke Nishimura <[email protected]>

Acked-by KAMEZAWA Hiroyuki <[email protected]>

Signed-off-by Minchan Kim <[email protected]>

---
 mm/migrate.c |    1 -
 1 files changed, 0 insertions(+), 1 deletions(-)

mm/migrate.c +0 -0

index eb083a6..2abc9c9 100644

Quoted 20 lines · Minchan Kim

Now memcg in unmap_and_move checks BUG_ON of charge. But mem_cgroup_prepare_migration returns either 0 or -ENOMEM. If it returns -ENOMEM, it jumps out unlock without the check. If it returns 0, it can pass BUG_ON. So it's meaningless. Let's remove it.

Cc Johannes Weiner <[email protected]>

Cc Balbir Singh <[email protected]>

Acked-by Daisuke Nishimura <[email protected]>

Acked-by KAMEZAWA Hiroyuki <[email protected]>

Signed-off-by Minchan Kim <[email protected]>

---
 mm/migrate.c |    1 -

 1 files changed, 0 insertions(+), 1 deletions(-)

mm/migrate.c +0 -0

index eb083a6..2abc9c9 100644

mm/migrate.c static int unmap_and_move(new_page_t get_new_page, unsigned long private, · line 686

Quoted 4 lines

mm/migrate.c +0 -1

@@ -683,7 +683,6 @@ static int unmap_and_move(new_page_t get_new_page, unsigned long private,
               rc = -ENOMEM;
               goto unlock;
       }
-     BUG_ON(charge);

You remove this assertion of the mem_cgroup_prepare_migration() return value but only add a comment about the expectations in the next patch.

Could you write a full-blown kerneldoc on mem_cgroup_prepare_migration and remove this BUG_ON() in the same patch?

memcg charge/uncharge could be handled by mem_cgroup_[prepare/end] migration itself so charge local variable in unmap_and_move lost the role since we introduced 01b1ae63c2.

In addition, the variable name is not good like below.

int unmap_and_move()

{
	charge = mem_cgroup_prepare_migration(xxx);
	..
		BUG_ON(charge); <-- BUG if it is charged?
		..
uncharge:
		if (!charge)    <-- why do we have to uncharge !charge?
			mem_group_end_migration(xxx);
	..
}

So let's remove unnecessary and confusing variable.

Cc Johannes Weiner <[email protected]>

Cc Balbir Singh <[email protected]>

Suggested-by Daisuke Nishimura <[email protected]>

Acked-by Daisuke Nishimura <[email protected]>

Acked-by KAMEZAWA Hiroyuki <[email protected]>

Signed-off-by Minchan Kim <[email protected]>

---
 mm/memcontrol.c |    1 +
 mm/migrate.c    |    9 +++------
 2 files changed, 4 insertions(+), 6 deletions(-)

mm/memcontrol.c +1 -0

index 8a97571..3c91d5c 100644
@@ -2873,6 +2873,7 @@ static inline int mem_cgroup_move_swap_account(swp_entry_t entry,
 /*
  * Before starting migration, account PAGE_SIZE to mem_cgroup that the old
  * page belongs to.
+ * Note: Should not return -EAGAIN. unmap_and_move depens on it.
  */
 int mem_cgroup_prepare_migration(struct page *page,
 	struct page *newpage, struct mem_cgroup **ptr, gfp_t gfp_mask)

mm/migrate.c +0 -1

index 2abc9c9..37055d0 100644
@@ -622,7 +622,6 @@ static int unmap_and_move(new_page_t get_new_page, unsigned long private,
 	int *result = NULL;
 	struct page *newpage = get_new_page(page, private, &result);
 	int remap_swapcache = 1;
-	int charge = 0;
 	struct mem_cgroup *mem;
 	struct anon_vma *anon_vma = NULL;