Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1439226
| From | Mel Gorman <mgorman@techsingularity.net> |
|---|---|
| Newsgroups | linux.kernel |
| Subject | [PATCH 09/34] mm, vmscan: simplify the logic deciding whether kswapd sleeps |
| Date | 2016-07-08 11:40 +0200 |
| Message-ID | <rSAqm-2Tv-39@gated-at.bofh.it> (permalink) |
| References | <rSAqm-2Tv-3@gated-at.bofh.it> |
| Organization | linux.* mail to news gateway |
kswapd goes through some complex steps trying to figure out if it should
stay awake based on the classzone_idx and the requested order. It is
unnecessarily complex and passes in an invalid classzone_idx to
balance_pgdat(). What matters most of all is whether a larger order has
been requsted and whether kswapd successfully reclaimed at the previous
order. This patch irons out the logic to check just that and the end
result is less headache inducing.
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
---
include/linux/mmzone.h | 5 ++-
mm/memory_hotplug.c | 5 ++-
mm/page_alloc.c | 2 +-
mm/vmscan.c | 101 ++++++++++++++++++++++++-------------------------
4 files changed, 57 insertions(+), 56 deletions(-)
diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index edafdaf62e90..4062fa74526f 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -668,8 +668,9 @@ typedef struct pglist_data {
wait_queue_head_t pfmemalloc_wait;
struct task_struct *kswapd; /* Protected by
mem_hotplug_begin/end() */
- int kswapd_max_order;
- enum zone_type classzone_idx;
+ int kswapd_order;
+ enum zone_type kswapd_classzone_idx;
+
#ifdef CONFIG_COMPACTION
int kcompactd_max_order;
enum zone_type kcompactd_classzone_idx;
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index c5278360ca66..065140ecd081 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -1209,9 +1209,10 @@ static pg_data_t __ref *hotadd_new_pgdat(int nid, u64 start)
arch_refresh_nodedata(nid, pgdat);
} else {
- /* Reset the nr_zones and classzone_idx to 0 before reuse */
+ /* Reset the nr_zones, order and classzone_idx before reuse */
pgdat->nr_zones = 0;
- pgdat->classzone_idx = 0;
+ pgdat->kswapd_order = 0;
+ pgdat->kswapd_classzone_idx = 0;
}
/* we can use NODE_DATA(nid) from here */
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index b84b85ae54ff..d25dc24f65f2 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -6084,7 +6084,7 @@ void __paginginit free_area_init_node(int nid, unsigned long *zones_size,
unsigned long end_pfn = 0;
/* pg_data_t should be reset to zero when it's allocated */
- WARN_ON(pgdat->nr_zones || pgdat->classzone_idx);
+ WARN_ON(pgdat->nr_zones || pgdat->kswapd_classzone_idx);
reset_deferred_meminit(pgdat);
pgdat->node_id = nid;
diff --git a/mm/vmscan.c b/mm/vmscan.c
index a52167eabc96..905c60473126 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -2762,7 +2762,7 @@ static bool pfmemalloc_watermark_ok(pg_data_t *pgdat)
/* kswapd must be awake if processes are being throttled */
if (!wmark_ok && waitqueue_active(&pgdat->kswapd_wait)) {
- pgdat->classzone_idx = min(pgdat->classzone_idx,
+ pgdat->kswapd_classzone_idx = min(pgdat->kswapd_classzone_idx,
(enum zone_type)ZONE_NORMAL);
wake_up_interruptible(&pgdat->kswapd_wait);
}
@@ -3042,11 +3042,11 @@ static bool prepare_kswapd_sleep(pg_data_t *pgdat, int order, long remaining,
if (!populated_zone(zone))
continue;
- if (zone_balanced(zone, order, classzone_idx))
- return true;
+ if (!zone_balanced(zone, order, classzone_idx))
+ return false;
}
- return false;
+ return true;
}
/*
@@ -3238,8 +3238,8 @@ static int balance_pgdat(pg_data_t *pgdat, int order, int classzone_idx)
return sc.order;
}
-static void kswapd_try_to_sleep(pg_data_t *pgdat, int order,
- int classzone_idx, int balanced_classzone_idx)
+static void kswapd_try_to_sleep(pg_data_t *pgdat, int alloc_order, int reclaim_order,
+ unsigned int classzone_idx)
{
long remaining = 0;
DEFINE_WAIT(wait);
@@ -3250,8 +3250,7 @@ static void kswapd_try_to_sleep(pg_data_t *pgdat, int order,
prepare_to_wait(&pgdat->kswapd_wait, &wait, TASK_INTERRUPTIBLE);
/* Try to sleep for a short interval */
- if (prepare_kswapd_sleep(pgdat, order, remaining,
- balanced_classzone_idx)) {
+ if (prepare_kswapd_sleep(pgdat, reclaim_order, remaining, classzone_idx)) {
/*
* Compaction records what page blocks it recently failed to
* isolate pages from and skips them in the future scanning.
@@ -3264,9 +3263,20 @@ static void kswapd_try_to_sleep(pg_data_t *pgdat, int order,
* We have freed the memory, now we should compact it to make
* allocation of the requested order possible.
*/
- wakeup_kcompactd(pgdat, order, classzone_idx);
+ wakeup_kcompactd(pgdat, alloc_order, classzone_idx);
remaining = schedule_timeout(HZ/10);
+
+ /*
+ * If woken prematurely then reset kswapd_classzone_idx and
+ * order. The values will either be from a wakeup request or
+ * the previous request that slept prematurely.
+ */
+ if (remaining) {
+ pgdat->kswapd_classzone_idx = max(pgdat->kswapd_classzone_idx, classzone_idx);
+ pgdat->kswapd_order = max(pgdat->kswapd_order, reclaim_order);
+ }
+
finish_wait(&pgdat->kswapd_wait, &wait);
prepare_to_wait(&pgdat->kswapd_wait, &wait, TASK_INTERRUPTIBLE);
}
@@ -3275,8 +3285,7 @@ static void kswapd_try_to_sleep(pg_data_t *pgdat, int order,
* After a short sleep, check if it was a premature sleep. If not, then
* go fully to sleep until explicitly woken up.
*/
- if (prepare_kswapd_sleep(pgdat, order, remaining,
- balanced_classzone_idx)) {
+ if (prepare_kswapd_sleep(pgdat, reclaim_order, remaining, classzone_idx)) {
trace_mm_vmscan_kswapd_sleep(pgdat->node_id);
/*
@@ -3317,9 +3326,7 @@ static void kswapd_try_to_sleep(pg_data_t *pgdat, int order,
*/
static int kswapd(void *p)
{
- unsigned long order, new_order;
- int classzone_idx, new_classzone_idx;
- int balanced_classzone_idx;
+ unsigned int alloc_order, reclaim_order, classzone_idx;
pg_data_t *pgdat = (pg_data_t*)p;
struct task_struct *tsk = current;
@@ -3349,38 +3356,20 @@ static int kswapd(void *p)
tsk->flags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD;
set_freezable();
- order = new_order = 0;
- classzone_idx = new_classzone_idx = pgdat->nr_zones - 1;
- balanced_classzone_idx = classzone_idx;
+ pgdat->kswapd_order = alloc_order = reclaim_order = 0;
+ pgdat->kswapd_classzone_idx = classzone_idx = 0;
for ( ; ; ) {
bool ret;
- /*
- * While we were reclaiming, there might have been another
- * wakeup, so check the values.
- */
- new_order = pgdat->kswapd_max_order;
- new_classzone_idx = pgdat->classzone_idx;
- pgdat->kswapd_max_order = 0;
- pgdat->classzone_idx = pgdat->nr_zones - 1;
+kswapd_try_sleep:
+ kswapd_try_to_sleep(pgdat, alloc_order, reclaim_order,
+ classzone_idx);
- if (order < new_order || classzone_idx > new_classzone_idx) {
- /*
- * Don't sleep if someone wants a larger 'order'
- * allocation or has tigher zone constraints
- */
- order = new_order;
- classzone_idx = new_classzone_idx;
- } else {
- kswapd_try_to_sleep(pgdat, order, classzone_idx,
- balanced_classzone_idx);
- order = pgdat->kswapd_max_order;
- classzone_idx = pgdat->classzone_idx;
- new_order = order;
- new_classzone_idx = classzone_idx;
- pgdat->kswapd_max_order = 0;
- pgdat->classzone_idx = pgdat->nr_zones - 1;
- }
+ /* Read the new order and classzone_idx */
+ alloc_order = reclaim_order = pgdat->kswapd_order;
+ classzone_idx = pgdat->kswapd_classzone_idx;
+ pgdat->kswapd_order = 0;
+ pgdat->kswapd_classzone_idx = 0;
ret = try_to_freeze();
if (kthread_should_stop())
@@ -3390,12 +3379,24 @@ static int kswapd(void *p)
* We can speed up thawing tasks if we don't call balance_pgdat
* after returning from the refrigerator
*/
- if (!ret) {
- trace_mm_vmscan_kswapd_wake(pgdat->node_id, order);
+ if (ret)
+ continue;
+
+ /*
+ * Reclaim begins at the requested order but if a high-order
+ * reclaim fails then kswapd falls back to reclaiming for
+ * order-0. If that happens, kswapd will consider sleeping
+ * for the order it finished reclaiming at (reclaim_order)
+ * but kcompactd is woken to compact for the original
+ * request (alloc_order).
+ */
+ trace_mm_vmscan_kswapd_wake(pgdat->node_id, alloc_order);
+ reclaim_order = balance_pgdat(pgdat, alloc_order, classzone_idx);
+ if (reclaim_order < alloc_order)
+ goto kswapd_try_sleep;
- /* return value ignored until next patch */
- balance_pgdat(pgdat, order, classzone_idx);
- }
+ alloc_order = reclaim_order = pgdat->kswapd_order;
+ classzone_idx = pgdat->kswapd_classzone_idx;
}
tsk->flags &= ~(PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD);
@@ -3418,10 +3419,8 @@ void wakeup_kswapd(struct zone *zone, int order, enum zone_type classzone_idx)
if (!cpuset_zone_allowed(zone, GFP_KERNEL | __GFP_HARDWALL))
return;
pgdat = zone->zone_pgdat;
- if (pgdat->kswapd_max_order < order) {
- pgdat->kswapd_max_order = order;
- pgdat->classzone_idx = min(pgdat->classzone_idx, classzone_idx);
- }
+ pgdat->kswapd_classzone_idx = max(pgdat->kswapd_classzone_idx, classzone_idx);
+ pgdat->kswapd_order = max(pgdat->kswapd_order, order);
if (!waitqueue_active(&pgdat->kswapd_wait))
return;
if (zone_balanced(zone, order, 0))
--
2.6.4
Back to linux.kernel | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread
[PATCH 00/34] Move LRU page reclaim from zones to nodes v9 Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:40 +0200 [PATCH 09/34] mm, vmscan: simplify the logic deciding whether kswapd sleeps Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:40 +0200 [PATCH 01/34] mm, vmstat: add infrastructure for per-node vmstats Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:40 +0200 [PATCH 13/34] mm, vmscan: make shrink_node decisions more node-centric Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:40 +0200 [PATCH 16/34] mm, page_alloc: consider dirtyable memory in terms of nodes Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:40 +0200 [PATCH 20/34] mm: move vmscan writes and file write accounting to the node Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:40 +0200 [PATCH 12/34] mm: vmscan: do not reclaim from kswapd if there is any eligible zone Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:40 +0200 [PATCH 24/34] mm, vmscan: avoid passing in classzone_idx unnecessarily to shrink_node Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:40 +0200 [PATCH 14/34] mm, memcg: move memcg limit enforcement from zones to nodes Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:40 +0200 [PATCH 04/34] mm, mmzone: clarify the usage of zone padding Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:40 +0200 [PATCH 02/34] mm, vmscan: move lru_lock to the node Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:40 +0200 [PATCH 15/34] mm, workingset: make working set detection node-aware Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:40 +0200 [PATCH 23/34] mm: convert zone_reclaim to node_reclaim Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:40 +0200 [PATCH 07/34] mm, vmscan: make kswapd reclaim in terms of nodes Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200 [PATCH 30/34] mm: page_alloc: cache the last node whose dirty limit is reached Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200 [PATCH 31/34] mm: vmstat: replace __count_zone_vm_events with a zone id equivalent Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200 [PATCH 34/34] mm, vmstat: remove zone and node double accounting by approximating retries Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200 [PATCH 10/34] mm, vmscan: by default have direct reclaim only shrink once per node Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200 [PATCH 32/34] mm: vmstat: account per-zone stalls and pages skipped during reclaim Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200 [PATCH 25/34] mm, vmscan: avoid passing in classzone_idx unnecessarily to compaction_ready Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200 [PATCH 28/34] mm, vmscan: add classzone information to tracepoints Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200 [PATCH 27/34] mm, vmscan: Have kswapd reclaim from all zones if reclaiming and buffer_heads_over_limit Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200 [PATCH 26/34] mm, vmscan: avoid passing in remaining unnecessarily to prepare_kswapd_sleep Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200 [PATCH 29/34] mm, page_alloc: remove fair zone allocation policy Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200 [PATCH 22/34] mm, page_alloc: wake kswapd based on the highest eligible zone Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200 [PATCH 33/34] mm, vmstat: print node-based stats in zoneinfo file Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200 [PATCH 19/34] mm: move most file-based accounting to the node Mel Gorman <mgorman@techsingularity.net> - 2016-07-08 11:50 +0200
csiph-web