Linux-mm Archive on lore.kernel.org
 help / color / Atom feed
From: Dave Hansen <dave.hansen@linux.intel.com>
To: linux-kernel@vger.kernel.org
Cc: linux-mm@kvack.org,Dave Hansen
	<dave.hansen@linux.intel.com>,yang.shi@linux.alibaba.com,rientjes@google.com,ying.huang@intel.com,dan.j.williams@intel.com
Subject: [RFC][PATCH 5/8] mm/numa: automatically generate node migration order
Date: Mon, 29 Jun 2020 16:45:12 -0700
Message-ID: <20200629234512.F34EDC44@viggo.jf.intel.com> (raw)
In-Reply-To: <20200629234503.749E5340@viggo.jf.intel.com>


From: Dave Hansen <dave.hansen@linux.intel.com>

When memory fills up on a node, memory contents can be
automatically migrated to another node.  The biggest problems are
knowing when to migrate and to where the migration should be
targeted.

The most straightforward way to generate the "to where" list
would be to follow the page allocator fallback lists.  Those
lists already tell us if memory is full where to look next.  It
would also be logical to move memory in that order.

But, the allocator fallback lists have a fatal flaw: most nodes
appear in all the lists.  This would potentially lead to
migration cycles (A->B, B->A, A->B, ...).

Instead of using the allocator fallback lists directly, keep a
separate node migration ordering.  But, reuse the same data used
to generate page allocator fallback in the first place:
find_next_best_node().

This means that the firmware data used to populate node distances
essentially dictates the ordering for now.  It should also be
architecture-neutral since all NUMA architectures have a working
find_next_best_node().

The protocol for node_demotion[] access and writing is not
standard.  It has no specific locking and is intended to be read
locklessly.  Readers must take care to avoid observing changes
that appear incoherent.  This was done so that node_demotion[]
locking has no chance of becoming a bottleneck on large systems
with lots of CPUs in direct reclaim.

This code is unused for now.  It will be called later in the
series.

Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Yang Shi <yang.shi@linux.alibaba.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Huang Ying <ying.huang@intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
---

 b/mm/internal.h   |    1 
 b/mm/migrate.c    |  130 +++++++++++++++++++++++++++++++++++++++++++++++++++++-
 b/mm/page_alloc.c |    2 
 3 files changed, 131 insertions(+), 2 deletions(-)

diff -puN mm/internal.h~auto-setup-default-migration-path-from-firmware mm/internal.h
--- a/mm/internal.h~auto-setup-default-migration-path-from-firmware	2020-06-29 16:34:41.629312597 -0700
+++ b/mm/internal.h	2020-06-29 16:34:41.638312597 -0700
@@ -192,6 +192,7 @@ extern int user_min_free_kbytes;
 
 extern void zone_pcp_update(struct zone *zone);
 extern void zone_pcp_reset(struct zone *zone);
+extern int find_next_best_node(int node, nodemask_t *used_node_mask);
 
 #if defined CONFIG_COMPACTION || defined CONFIG_CMA
 
diff -puN mm/migrate.c~auto-setup-default-migration-path-from-firmware mm/migrate.c
--- a/mm/migrate.c~auto-setup-default-migration-path-from-firmware	2020-06-29 16:34:41.631312597 -0700
+++ b/mm/migrate.c	2020-06-29 16:34:41.639312597 -0700
@@ -1128,6 +1128,10 @@ out:
 	return rc;
 }
 
+/*
+ * Writes to this array occur without locking.  READ_ONCE()
+ * is recommended for readers.
+ */
 static int node_demotion[MAX_NUMNODES] = {[0 ...  MAX_NUMNODES - 1] = NUMA_NO_NODE};
 
 /**
@@ -1141,7 +1145,13 @@ int next_demotion_node(int node)
 {
 	get_online_mems();
 	while (true) {
-		node = node_demotion[node];
+		/*
+		 * node_demotion[] is updated without excluding
+		 * this function from running.  READ_ONCE() avoids
+		 * 'node' checks reading different values from
+		 * node_demotion[].
+		 */
+		node = READ_ONCE(node_demotion[node]);
 		if (node == NUMA_NO_NODE)
 			break;
 		if (node_online(node))
@@ -3086,3 +3096,121 @@ void migrate_vma_finalize(struct migrate
 }
 EXPORT_SYMBOL(migrate_vma_finalize);
 #endif /* CONFIG_DEVICE_PRIVATE */
+
+/* Disable reclaim-based migration. */
+static void disable_all_migrate_targets(void)
+{
+	int node;
+
+	for_each_online_node(node)
+		node_demotion[node] = NUMA_NO_NODE;
+}
+
+/*
+ * Find an automatic demotion target for 'node'.
+ * Failing here is OK.  It might just indicate
+ * being at the end of a chain.
+ */
+static int establish_migrate_target(int node, nodemask_t *used)
+{
+	int migration_target;
+
+	/*
+	 * Can not set a migration target on a
+	 * node with it already set.
+	 *
+	 * No need for READ_ONCE() here since this
+	 * in the write path for node_demotion[].
+	 * This should be the only thread writing.
+	 */
+	if (node_demotion[node] != NUMA_NO_NODE)
+		return NUMA_NO_NODE;
+
+	migration_target = find_next_best_node(node, used);
+	if (migration_target == NUMA_NO_NODE)
+		return NUMA_NO_NODE;
+
+	node_demotion[node] = migration_target;
+
+	return migration_target;
+}
+
+/*
+ * When memory fills up on a node, memory contents can be
+ * automatically migrated to another node instead of
+ * discarded at reclaim.
+ *
+ * Establish a "migration path" which will start at nodes
+ * with CPUs and will follow the priorities used to build the
+ * page allocator zonelists.
+ *
+ * The difference here is that cycles must be avoided.  If
+ * node0 migrates to node1, then neither node1, nor anything
+ * node1 migrates to can migrate to node0.
+ *
+ * This function can run simultaneously with readers of
+ * node_demotion[].  However, it can not run simultaneously
+ * with itself.  Exclusion is provided by memory hotplug events
+ * being single-threaded.
+ */
+void set_migration_target_nodes(void)
+{
+	nodemask_t next_pass = NODE_MASK_NONE;
+	nodemask_t this_pass = NODE_MASK_NONE;
+	nodemask_t used_targets = NODE_MASK_NONE;
+	int node;
+
+	get_online_mems();
+	/*
+	 * Avoid any oddities like cycles that could occur
+	 * from changes in the topology.  This will leave
+	 * a momentary gap when migration is disabled.
+	 */
+	disable_all_migrate_targets();
+
+	/*
+	 * Ensure that the "disable" is visible across the system.
+	 * Readers will see either a combination of before+disable
+	 * state or disable+after.  They will never see before and
+	 * after state together.
+	 *
+	 * The before+after state together might have cycles and
+	 * could cause readers to do things like loop until this
+	 * function finishes.  This ensures they can only see a
+	 * single "bad" read and would, for instance, only loop
+	 * once.
+	 */
+	smp_wmb();
+
+	/*
+	 * Allocations go close to CPUs, first.  Assume that
+	 * the migration path starts at the nodes with CPUs.
+	 */
+	next_pass = node_states[N_CPU];
+again:
+	this_pass = next_pass;
+	next_pass = NODE_MASK_NONE;
+	/*
+	 * To avoid cycles in the migration "graph", ensure
+	 * that migration sources are not future targets by
+	 * setting them in 'used_targets'.
+	 *
+	 * But, do this only once per pass so that multiple
+	 * source nodes can share a target node.
+	 */
+	nodes_or(used_targets, used_targets, this_pass);
+	for_each_node_mask(node, this_pass) {
+		int target_node = establish_migrate_target(node, &used_targets);
+
+		if (target_node == NUMA_NO_NODE)
+			continue;
+
+		/* Visit targets from this pass in the next pass: */
+		node_set(target_node, next_pass);
+	}
+	/* Is another pass necessary? */
+	if (!nodes_empty(next_pass))
+		goto again;
+
+	put_online_mems();
+}
diff -puN mm/page_alloc.c~auto-setup-default-migration-path-from-firmware mm/page_alloc.c
--- a/mm/page_alloc.c~auto-setup-default-migration-path-from-firmware	2020-06-29 16:34:41.634312597 -0700
+++ b/mm/page_alloc.c	2020-06-29 16:34:41.641312597 -0700
@@ -5591,7 +5591,7 @@ static int node_load[MAX_NUMNODES];
  *
  * Return: node id of the found node or %NUMA_NO_NODE if no node is found.
  */
-static int find_next_best_node(int node, nodemask_t *used_node_mask)
+int find_next_best_node(int node, nodemask_t *used_node_mask)
 {
 	int n, val;
 	int min_val = INT_MAX;
_


  parent reply index

Thread overview: 43+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-06-29 23:45 [RFC][PATCH 0/8] Migrate Pages in lieu of discard Dave Hansen
2020-06-29 23:45 ` [RFC][PATCH 1/8] mm/numa: node demotion data structure and lookup Dave Hansen
2020-06-29 23:45 ` [RFC][PATCH 2/8] mm/migrate: Defer allocating new page until needed Dave Hansen
2020-07-01  8:47   ` Greg Thelen
2020-07-01 14:46     ` Dave Hansen
2020-07-01 18:32       ` Yang Shi
2020-06-29 23:45 ` [RFC][PATCH 3/8] mm/vmscan: Attempt to migrate page in lieu of discard Dave Hansen
2020-07-01  0:47   ` David Rientjes
2020-07-01  1:29     ` Yang Shi
2020-07-01  5:41       ` David Rientjes
2020-07-01  8:54         ` Huang, Ying
2020-07-01 18:20           ` Dave Hansen
2020-07-01 19:50             ` David Rientjes
2020-07-02  1:50               ` Huang, Ying
2020-07-01 15:15         ` Dave Hansen
2020-07-01 17:21         ` Yang Shi
2020-07-01 19:45           ` David Rientjes
2020-07-02 10:02             ` Jonathan Cameron
2020-07-01  1:40     ` Huang, Ying
2020-07-01 16:48     ` Dave Hansen
2020-07-01 19:25       ` David Rientjes
2020-07-02  5:02         ` Huang, Ying
2020-06-29 23:45 ` [RFC][PATCH 4/8] mm/vmscan: add page demotion counter Dave Hansen
2020-06-29 23:45 ` Dave Hansen [this message]
2020-06-30  8:22   ` [RFC][PATCH 5/8] mm/numa: automatically generate node migration order Huang, Ying
2020-07-01 18:23     ` Dave Hansen
2020-07-02  1:20       ` Huang, Ying
2020-06-29 23:45 ` [RFC][PATCH 6/8] mm/vmscan: Consider anonymous pages without swap Dave Hansen
2020-06-29 23:45 ` [RFC][PATCH 7/8] mm/vmscan: never demote for memcg reclaim Dave Hansen
2020-06-29 23:45 ` [RFC][PATCH 8/8] mm/numa: new reclaim mode to enable reclaim-based migration Dave Hansen
2020-06-30  7:23   ` Huang, Ying
2020-06-30 17:50     ` Yang Shi
2020-07-01  0:48       ` Huang, Ying
2020-07-01  1:12         ` Yang Shi
2020-07-01  1:28           ` Huang, Ying
2020-07-01 16:02       ` Dave Hansen
2020-07-03  9:30   ` Huang, Ying
2020-06-30 18:36 ` [RFC][PATCH 0/8] Migrate Pages in lieu of discard Shakeel Butt
2020-06-30 18:51   ` Dave Hansen
2020-06-30 19:25     ` Shakeel Butt
2020-06-30 19:31       ` Dave Hansen
2020-07-01 14:24         ` [RFC] [PATCH " Zi Yan
2020-07-01 14:32           ` Dave Hansen

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200629234512.F34EDC44@viggo.jf.intel.com \
    --to=dave.hansen@linux.intel.com \
    --cc=dan.j.williams@intel.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=rientjes@google.com \
    --cc=yang.shi@linux.alibaba.com \
    --cc=ying.huang@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

Linux-mm Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-mm/0 linux-mm/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-mm linux-mm/ https://lore.kernel.org/linux-mm \
		linux-mm@kvack.org
	public-inbox-index linux-mm

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kvack.linux-mm


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git