init_tlb_ubc() looked unnecessary to me: tlb_ubc is statically initialized
with zeroes in the init_task, and copied from parent to child while it is
quiescent in arch_dup_task_struct(); so I went to delete it.

But inserted temporary debug WARN_ONs in place of init_tlb_ubc() to check
that it was always empty at that point, and found them firing: because
memcg reclaim can recurse into global reclaim (when allocating biosets
for swapout in my case), and arrive back at the init_tlb_ubc() in

Resetting tlb_ubc.flush_required at that point is wrong: if the upper
level needs a deferred TLB flush, but the lower level turns out not to,
we miss a TLB flush.  But fortunately, that's the only part of the
protocol that does not nest: with the initialization removed, cpumask 
collects bits from upper and lower levels, and flushes TLB when needed.

Fixes: 72b252aed506 ("mm: send one IPI per CPU to TLB flush all entries after 
unmapping pages")
Signed-off-by: Hugh Dickins <>
Cc: # 4.3+

 mm/vmscan.c |   19 -------------------
 1 file changed, 19 deletions(-)

--- 4.8-rc7/mm/vmscan.c 2016-09-05 16:42:52.496692429 -0700
+++ linux/mm/vmscan.c   2016-09-22 09:32:37.900894833 -0700
@@ -2303,23 +2303,6 @@ out:
-static void init_tlb_ubc(void)
-       /*
-        * This deliberately does not clear the cpumask as it's expensive
-        * and unnecessary. If there happens to be data in there then the
-        * first SWAP_CLUSTER_MAX pages will send an unnecessary IPI and
-        * then will be cleared.
-        */
-       current->tlb_ubc.flush_required = false;
-static inline void init_tlb_ubc(void)
  * This is a basic per-node page freer.  Used by both kswapd and direct 
@@ -2355,8 +2338,6 @@ static void shrink_node_memcg(struct pgl
        scan_adjusted = (global_reclaim(sc) && !current_is_kswapd() &&
                         sc->priority == DEF_PRIORITY);
-       init_tlb_ubc();
        while (nr[LRU_INACTIVE_ANON] || nr[LRU_ACTIVE_FILE] ||
                                        nr[LRU_INACTIVE_FILE]) {

Reply via email to