The patch titled
memcg: fix race at move_parent around compound_order()
has been added to the -mm tree. Its filename is
memcg-fix-race-at-move_parent-around-compound_order.patch
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/SubmitChecklist when testing your code ***
See http://userweb.kernel.org/~akpm/stuff/added-to-mm.txt to find
out what to do about this
The current -mm tree may be found at http://userweb.kernel.org/~akpm/mmotm/
------------------------------------------------------
Subject: memcg: fix race at move_parent around compound_order()
From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx>
A fix up mem_cgroup_move_parent() which use compound_order() in
asynchrnous manner. This compound_order() may return unknown value
because we don't take lock. Use PageTransHuge() and HPAGE_SIZE instead of
it.
Also clean up for mem_cgroup_move_parent().
- remove unnecessary initialization of local variable.
- rename charge_size -> page_size
- remove unnecessary (wrong) comment.
- added a comment about THP.
Note:
Current design take compound_page_lock() in caller of move_account().
This should be revisited when we implement direct move_task of hugepage
without splitting.
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx>
Reviewed-by: Johannes Weiner <hannes@xxxxxxxxxxx>
Acked-by: Daisuke Nishimura <nishimura@xxxxxxxxxxxxxxxxx>
Cc: Balbir Singh <balbir@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
mm/memcontrol.c | 25 ++++++++++++++++---------
1 file changed, 16 insertions(+), 9 deletions(-)
diff -puN mm/memcontrol.c~memcg-fix-race-at-move_parent-around-compound_order mm/memcontrol.c
--- a/mm/memcontrol.c~memcg-fix-race-at-move_parent-around-compound_order
+++ a/mm/memcontrol.c
@@ -2236,7 +2236,12 @@ static int mem_cgroup_move_account(struc
{
int ret = -EINVAL;
unsigned long flags;
-
+ /*
+ * The page is isolated from LRU. So, collapse function
+ * will not handle this page. But page splitting can happen.
+ * Do this check under compound_page_lock(). The caller should
+ * hold it.
+ */
if ((charge_size > PAGE_SIZE) && !PageTransHuge(pc->page))
return -EBUSY;
@@ -2268,7 +2273,7 @@ static int mem_cgroup_move_parent(struct
struct cgroup *cg = child->css.cgroup;
struct cgroup *pcg = cg->parent;
struct mem_cgroup *parent;
- int charge = PAGE_SIZE;
+ int page_size = PAGE_SIZE;
unsigned long flags;
int ret;
@@ -2281,22 +2286,24 @@ static int mem_cgroup_move_parent(struct
goto out;
if (isolate_lru_page(page))
goto put;
- /* The page is isolated from LRU and we have no race with splitting */
- charge = PAGE_SIZE << compound_order(page);
+
+ if (PageTransHuge(page))
+ page_size = HPAGE_SIZE;
parent = mem_cgroup_from_cont(pcg);
- ret = __mem_cgroup_try_charge(NULL, gfp_mask, &parent, false, charge);
+ ret = __mem_cgroup_try_charge(NULL, gfp_mask,
+ &parent, false, page_size);
if (ret || !parent)
goto put_back;
- if (charge > PAGE_SIZE)
+ if (page_size > PAGE_SIZE)
flags = compound_lock_irqsave(page);
- ret = mem_cgroup_move_account(pc, child, parent, true, charge);
+ ret = mem_cgroup_move_account(pc, child, parent, true, page_size);
if (ret)
- mem_cgroup_cancel_charge(parent, charge);
+ mem_cgroup_cancel_charge(parent, page_size);
- if (charge > PAGE_SIZE)
+ if (page_size > PAGE_SIZE)
compound_unlock_irqrestore(page, flags);
put_back:
putback_lru_page(page);
_
Patches currently in -mm which might be from kamezawa.hiroyu@xxxxxxxxxxxxxx are
mm-fix-deferred-congestion-timeout-if-preferred-zone-is-not-allowed.patch
mm-clear-pages_scanned-only-if-draining-a-pcp-adds-pages-to-the-buddy-allocator.patch
mm-memcontrolc-fix-uninitialized-variable-use-in-mem_cgroup_move_parent.patch
memcg-fix-account-leak-at-failure-of-memsw-acconting.patch
memcg-bugfix-check-mem_cgroup_disabled-at-split-fixup.patch
memcg-fix-race-at-move_parent-around-compound_order.patch
oom-suppress-nodes-that-are-not-allowed-from-meminfo-on-oom-kill.patch
oom-suppress-show_mem-for-many-nodes-in-irq-context-on-page-alloc-failure.patch
oom-suppress-nodes-that-are-not-allowed-from-meminfo-on-page-alloc-failure.patch
mm-add-replace_page_cache_page-function.patch
--
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
