From: Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx> == tl;dr == Automatic demotion opted for a simple, lazy approach to handling hotplug events. This noticeably slows down memory hotplug[1]. Optimize away updates to the demotion order when memory hotplug events should have no effect. This has no effect on CPU hotplug. There is no known problem on the CPU side and any work there will be in a separate series. == Background == Automatic demotion is a memory migration strategy to ensure that new allocations have room in faster memory tiers on tiered memory systems. The kernel maintains an array (node_demotion[]) to drive these migrations. The node_demotion[] path is calculated by starting at nodes with CPUs and then "walking" to nodes with memory. Only hotplug events which online or offline a node with memory (N_ONLINE) or CPUs (N_CPU) will actually affect the migration order. == Problem == However, the current code is lazy. It completely regenerates the migration order on *any* CPU or memory hotplug event. The logic was that these events are extremely rare and that the overhead from indiscriminate order regeneration is minimal. Part of the update logic involves a synchronize_rcu(), which is a pretty big hammer. Its overhead was large enough to be detected by some 0day tests that watch memory hotplug performance[1]. == Solution == Add a new helper (node_demotion_topo_changed()) which can differentiate between superfluous and impactful hotplug events. Skip the expensive update operation for superfluous events. == Aside: Locking == It took me a few moments to declare the locking to be safe enough for node_demotion_topo_changed() to work. It all hinges on the memory hotplug lock: During memory hotplug events, 'mem_hotplug_lock' is held for write. This ensures that two memory hotplug events can not be called simultaneously. CPU hotplug has a similar lock (cpuhp_state_mutex) which also provides mutual exclusion between CPU hotplug events. In addition, the demotion code acquire and hold the mem_hotplug_lock for read during its CPU hotplug handlers. This provides mutual exclusion between the demotion memory hotplug callbacks and the CPU hotplug callbacks. This effectively allows treating the migration target generation code to act as if it is single-threaded. 1. https://lore.kernel.org/all/20210905135932.GE15026@xsang-OptiPlex-9020/ Fixes: 884a6e5d1f93 ("mm/migrate: update node demotion order on hotplug events") Reported-by: kernel test robot <oliver.sang@xxxxxxxxx> Signed-off-by: Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx> Cc: "Huang, Ying" <ying.huang@xxxxxxxxx> Cc: Michal Hocko <mhocko@xxxxxxxx> Cc: Wei Xu <weixugc@xxxxxxxxxx> Cc: Oscar Salvador <osalvador@xxxxxxx> Cc: David Rientjes <rientjes@xxxxxxxxxx> Cc: Dan Williams <dan.j.williams@xxxxxxxxx> Cc: David Hildenbrand <david@xxxxxxxxxx> Cc: Greg Thelen <gthelen@xxxxxxxxxx> Cc: Yang Shi <yang.shi@xxxxxxxxxxxxxxxxx> Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- b/mm/migrate.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) diff -puN mm/migrate.c~faster-node-order mm/migrate.c --- a/mm/migrate.c~faster-node-order 2021-09-24 09:12:30.988377798 -0700 +++ b/mm/migrate.c 2021-09-24 09:12:30.988377798 -0700 @@ -3239,8 +3239,18 @@ static int migration_offline_cpu(unsigne * set_migration_target_nodes(). */ static int __meminit migrate_on_reclaim_callback(struct notifier_block *self, - unsigned long action, void *arg) + unsigned long action, void *_arg) { + struct memory_notify *arg = _arg; + + /* + * Only update the node migration order when a node is + * changing status, like online->offline. This avoids + * the overhead of synchronize_rcu() in most cases. + */ + if (arg->status_change_nid < 0) + return notifier_from_errno(0); + switch (action) { case MEM_GOING_OFFLINE: /* _
