mm: multi-gen LRU: minimal implementation

To avoid confusion, the terms "promotion" and "demotion" will be applied
to the multi-gen LRU, as a new convention; the terms "activation" and
"deactivation" will be applied to the active/inactive LRU, as usual.

The aging produces young generations.  Given an lruvec, it increments
max_seq when max_seq-min_seq+1 approaches MIN_NR_GENS.  The aging promotes
hot pages to the youngest generation when it finds them accessed through
page tables; the demotion of cold pages happens consequently when it
increments max_seq.  Promotion in the aging path does not involve any LRU
list operations, only the updates of the gen counter and
lrugen->nr_pages[]; demotion, unless as the result of the increment of
max_seq, requires LRU list operations, e.g., lru_deactivate_fn().  The
aging has the complexity O(nr_hot_pages), since it is only interested in
hot pages.

The eviction consumes old generations.  Given an lruvec, it increments
min_seq when lrugen->lists[] indexed by min_seq%MAX_NR_GENS becomes empty.
A feedback loop modeled after the PID controller monitors refaults over
anon and file types and decides which type to evict when both types are
available from the same generation.

The protection of pages accessed multiple times through file descriptors
takes place in the eviction path.  Each generation is divided into
multiple tiers.  A page accessed N times through file descriptors is in
tier order_base_2(N).  Tiers do not have dedicated lrugen->lists[], only
bits in folio->flags.  The aforementioned feedback loop also monitors
refaults over all tiers and decides when to protect pages in which tiers
(N>1), using the first tier (N=0,1) as a baseline.  The first tier
contains single-use unmapped clean pages, which are most likely the best
choices.  In contrast to promotion in the aging path, the protection of a
page in the eviction path is achieved by moving this page to the next
generation, i.e., min_seq+1, if the feedback loop decides so.  This
approach has the following advantages:

1. It removes the cost of activation in the buffered access path by
   inferring whether pages accessed multiple times through file
   descriptors are statistically hot and thus worth protecting in the
   eviction path.
2. It takes pages accessed through page tables into account and avoids
   overprotecting pages accessed multiple times through file
   descriptors. (Pages accessed through page tables are in the first
   tier, since N=0.)
3. More tiers provide better protection for pages accessed more than
   twice through file descriptors, when under heavy buffered I/O
   workloads.

Server benchmark results:
  Single workload:
    fio (buffered I/O): +[30, 32]%
                IOPS         BW
      5.19-rc1: 2673k        10.2GiB/s
      patch1-6: 3491k        13.3GiB/s

  Single workload:
    memcached (anon): -[4, 6]%
                Ops/sec      KB/sec
      5.19-rc1: 1161501.04   45177.25
      patch1-6: 1106168.46   43025.04

  Configurations:
    CPU: two Xeon 6154
    Mem: total 256G

    Node 1 was only used as a ram disk to reduce the variance in the
    results.

    patch drivers/block/brd.c <<EOF
    99,100c99,100
    < 	gfp_flags = GFP_NOIO | __GFP_ZERO | __GFP_HIGHMEM;
    < 	page = alloc_page(gfp_flags);
    ---
    > 	gfp_flags = GFP_NOIO | __GFP_ZERO | __GFP_HIGHMEM | __GFP_THISNODE;
    > 	page = alloc_pages_node(1, gfp_flags, 0);
    EOF

    cat >>/etc/systemd/system.conf <<EOF
    CPUAffinity=numa
    NUMAPolicy=bind
    NUMAMask=0
    EOF

    cat >>/etc/memcached.conf <<EOF
    -m 184320
    -s /var/run/memcached/memcached.sock
    -a 0766
    -t 36
    -B binary
    EOF

    cat fio.sh
    modprobe brd rd_nr=1 rd_size=113246208
    swapoff -a
    mkfs.ext4 /dev/ram0
    mount -t ext4 /dev/ram0 /mnt

    mkdir /sys/fs/cgroup/user.slice/test
    echo 38654705664 >/sys/fs/cgroup/user.slice/test/memory.max
    echo $$ >/sys/fs/cgroup/user.slice/test/cgroup.procs
    fio -name=mglru --numjobs=72 --directory=/mnt --size=1408m \
      --buffered=1 --ioengine=io_uring --iodepth=128 \
      --iodepth_batch_submit=32 --iodepth_batch_complete=32 \
      --rw=randread --random_distribution=random --norandommap \
      --time_based --ramp_time=10m --runtime=5m --group_reporting

    cat memcached.sh
    modprobe brd rd_nr=1 rd_size=113246208
    swapoff -a
    mkswap /dev/ram0
    swapon /dev/ram0

    memtier_benchmark -S /var/run/memcached/memcached.sock \
      -P memcache_binary -n allkeys --key-minimum=1 \
      --key-maximum=65000000 --key-pattern=P:P -c 1 -t 36 \
      --ratio 1:0 --pipeline 8 -d 2000

    memtier_benchmark -S /var/run/memcached/memcached.sock \
      -P memcache_binary -n allkeys --key-minimum=1 \
      --key-maximum=65000000 --key-pattern=R:R -c 1 -t 36 \
      --ratio 0:1 --pipeline 8 --randomize --distinct-client-seed

Client benchmark results:
  kswapd profiles:
    5.19-rc1
      40.33%  page_vma_mapped_walk (overhead)
      21.80%  lzo1x_1_do_compress (real work)
       7.53%  do_raw_spin_lock
       3.95%  _raw_spin_unlock_irq
       2.52%  vma_interval_tree_iter_next
       2.37%  folio_referenced_one
       2.28%  vma_interval_tree_subtree_search
       1.97%  anon_vma_interval_tree_iter_first
       1.60%  ptep_clear_flush
       1.06%  __zram_bvec_write

    patch1-6
      39.03%  lzo1x_1_do_compress (real work)
      18.47%  page_vma_mapped_walk (overhead)
       6.74%  _raw_spin_unlock_irq
       3.97%  do_raw_spin_lock
       2.49%  ptep_clear_flush
       2.48%  anon_vma_interval_tree_iter_first
       1.92%  folio_referenced_one
       1.88%  __zram_bvec_write
       1.48%  memmove
       1.31%  vma_interval_tree_iter_next

  Configurations:
    CPU: single Snapdragon 7c
    Mem: total 4G

    ChromeOS MemoryPressure [1]

[1] https://chromium.googlesource.com/chromiumos/platform/tast-tests/

Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Yu Zhao <[email protected]>
Acked-by: Brian Geffon <[email protected]>
Acked-by: Jan Alexander Steffens (heftig) <[email protected]>
Acked-by: Oleksandr Natalenko <[email protected]>
Acked-by: Steven Barrett <[email protected]>
Acked-by: Suleiman Souhlal <[email protected]>
Tested-by: Daniel Byrne <[email protected]>
Tested-by: Donald Carr <[email protected]>
Tested-by: Holger Hoffstätte <[email protected]>
Tested-by: Konstantin Kharlamov <[email protected]>
Tested-by: Shuang Zhai <[email protected]>
Tested-by: Sofia Trinh <[email protected]>
Tested-by: Vaibhav Jain <[email protected]>
Cc: Andi Kleen <[email protected]>
Cc: Aneesh Kumar K.V <[email protected]>
Cc: Barry Song <[email protected]>
Cc: Catalin Marinas <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Hillf Danton <[email protected]>
Cc: Jens Axboe <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: Jonathan Corbet <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: Matthew Wilcox <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Miaohe Lin <[email protected]>
Cc: Michael Larabel <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Mike Rapoport <[email protected]>
Cc: Mike Rapoport <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Qi Zheng <[email protected]>
Cc: Tejun Heo <[email protected]>
Cc: Vlastimil Babka <[email protected]>
Cc: Will Deacon <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

include/linux/mm_inline.h

@@ -121,6 +121,33 @@ static inline int lru_gen_from_seq(unsigned long seq)
 	return seq % MAX_NR_GENS;
 }
 
+static inline int lru_hist_from_seq(unsigned long seq)
+{
+	return seq % NR_HIST_GENS;
+}
+
+static inline int lru_tier_from_refs(int refs)
+{
+	VM_WARN_ON_ONCE(refs > BIT(LRU_REFS_WIDTH));
+
+	/* see the comment in folio_lru_refs() */
+	return order_base_2(refs + 1);
+}
+
+static inline int folio_lru_refs(struct folio *folio)
+{
+	unsigned long flags = READ_ONCE(folio->flags);
+	bool workingset = flags & BIT(PG_workingset);
+
+	/*
+	 * Return the number of accesses beyond PG_referenced, i.e., N-1 if the
+	 * total number of accesses is N>1, since N=0,1 both map to the first
+	 * tier. lru_tier_from_refs() will account for this off-by-one. Also see
+	 * the comment on MAX_NR_TIERS.
+	 */
+	return ((flags & LRU_REFS_MASK) >> LRU_REFS_PGOFF) + workingset;
+}
+
 static inline int folio_lru_gen(struct folio *folio)
 {
 	unsigned long flags = READ_ONCE(folio->flags);
@@ -173,6 +200,15 @@ static inline void lru_gen_update_size(struct lruvec *lruvec, struct folio *foli
 		__update_lru_size(lruvec, lru, zone, -delta);
 		return;
 	}
+
+	/* promotion */
+	if (!lru_gen_is_active(lruvec, old_gen) && lru_gen_is_active(lruvec, new_gen)) {
+		__update_lru_size(lruvec, lru, zone, -delta);
+		__update_lru_size(lruvec, lru + LRU_ACTIVE, zone, delta);
+	}
+
+	/* demotion requires isolation, e.g., lru_deactivate_fn() */
+	VM_WARN_ON_ONCE(lru_gen_is_active(lruvec, old_gen) && !lru_gen_is_active(lruvec, new_gen));
 }
 
 static inline bool lru_gen_add_folio(struct lruvec *lruvec, struct folio *folio, bool reclaiming)

include/linux/mmzone.h

@@ -350,6 +350,28 @@ enum lruvec_flags {
 #define MIN_NR_GENS		2U
 #define MAX_NR_GENS		4U
 
+/*
+ * Each generation is divided into multiple tiers. A page accessed N times
+ * through file descriptors is in tier order_base_2(N). A page in the first tier
+ * (N=0,1) is marked by PG_referenced unless it was faulted in through page
+ * tables or read ahead. A page in any other tier (N>1) is marked by
+ * PG_referenced and PG_workingset. This implies a minimum of two tiers is
+ * supported without using additional bits in folio->flags.
+ *
+ * In contrast to moving across generations which requires the LRU lock, moving
+ * across tiers only involves atomic operations on folio->flags and therefore
+ * has a negligible cost in the buffered access path. In the eviction path,
+ * comparisons of refaulted/(evicted+protected) from the first tier and the
+ * rest infer whether pages accessed multiple times through file descriptors
+ * are statistically hot and thus worth protecting.
+ *
+ * MAX_NR_TIERS is set to 4 so that the multi-gen LRU can support twice the
+ * number of categories of the active/inactive LRU when keeping track of
+ * accesses through file descriptors. This uses MAX_NR_TIERS-2 spare bits in
+ * folio->flags.
+ */
+#define MAX_NR_TIERS		4U
+
 #ifndef __GENERATING_BOUNDS_H
 
 struct lruvec;
@@ -364,6 +386,16 @@ enum {
 	LRU_GEN_FILE,
 };
 
+#define MIN_LRU_BATCH		BITS_PER_LONG
+#define MAX_LRU_BATCH		(MIN_LRU_BATCH * 64)
+
+/* whether to keep historical stats from evicted generations */
+#ifdef CONFIG_LRU_GEN_STATS
+#define NR_HIST_GENS		MAX_NR_GENS
+#else
+#define NR_HIST_GENS		1U
+#endif
+
 /*
  * The youngest generation number is stored in max_seq for both anon and file
  * types as they are aged on an equal footing. The oldest generation numbers are
@@ -386,6 +418,15 @@ struct lru_gen_struct {
 	struct list_head lists[MAX_NR_GENS][ANON_AND_FILE][MAX_NR_ZONES];
 	/* the multi-gen LRU sizes, eventually consistent */
 	long nr_pages[MAX_NR_GENS][ANON_AND_FILE][MAX_NR_ZONES];
+	/* the exponential moving average of refaulted */
+	unsigned long avg_refaulted[ANON_AND_FILE][MAX_NR_TIERS];
+	/* the exponential moving average of evicted+protected */
+	unsigned long avg_total[ANON_AND_FILE][MAX_NR_TIERS];
+	/* the first tier doesn't need protection, hence the minus one */
+	unsigned long protected[NR_HIST_GENS][ANON_AND_FILE][MAX_NR_TIERS - 1];
+	/* can be modified without holding the LRU lock */
+	atomic_long_t evicted[NR_HIST_GENS][ANON_AND_FILE][MAX_NR_TIERS];
+	atomic_long_t refaulted[NR_HIST_GENS][ANON_AND_FILE][MAX_NR_TIERS];
 };
 
 void lru_gen_init_lruvec(struct lruvec *lruvec);

include/linux/page-flags-layout.h

@@ -106,7 +106,10 @@
 #error "Not enough bits in page flags"
 #endif
 
-#define LRU_REFS_WIDTH	0
+/* see the comment on MAX_NR_TIERS */
+#define LRU_REFS_WIDTH	min(__LRU_REFS_WIDTH, BITS_PER_LONG - NR_PAGEFLAGS - \
+			    ZONES_WIDTH - LRU_GEN_WIDTH - SECTIONS_WIDTH - \
+			    NODES_WIDTH - KASAN_TAG_WIDTH - LAST_CPUPID_WIDTH)
 
 #endif
 #endif /* _LINUX_PAGE_FLAGS_LAYOUT */

kernel/bounds.c

@@ -24,8 +24,10 @@ int main(void)
 	DEFINE(SPINLOCK_SIZE, sizeof(spinlock_t));
 #ifdef CONFIG_LRU_GEN
 	DEFINE(LRU_GEN_WIDTH, order_base_2(MAX_NR_GENS + 1));
+	DEFINE(__LRU_REFS_WIDTH, MAX_NR_TIERS - 2);
 #else
 	DEFINE(LRU_GEN_WIDTH, 0);
+	DEFINE(__LRU_REFS_WIDTH, 0);
 #endif
 	/* End of constants */
 

mm/Kconfig

@@ -1118,6 +1118,7 @@ config PTE_MARKER_UFFD_WP
 	  purposes.  It is required to enable userfaultfd write protection on
 	  file-backed memory types like shmem and hugetlbfs.
 
+# multi-gen LRU {
 config LRU_GEN
 	bool "Multi-Gen LRU"
 	depends on MMU
@@ -1126,6 +1127,16 @@ config LRU_GEN
 	help
 	  A high performance LRU implementation to overcommit memory.
 
+config LRU_GEN_STATS
+	bool "Full stats for debugging"
+	depends on LRU_GEN
+	help
+	  Do not enable this option unless you plan to look at historical stats
+	  from evicted generations for debugging purpose.
+
+	  This option has a per-memcg and per-node memory overhead.
+# }
+
 source "mm/damon/Kconfig"
 
 endmenu

mm/swap.c

@@ -428,6 +428,40 @@ static void __lru_cache_activate_folio(struct folio *folio)
 	local_unlock(&cpu_fbatches.lock);
 }
 
+#ifdef CONFIG_LRU_GEN
+static void folio_inc_refs(struct folio *folio)
+{
+	unsigned long new_flags, old_flags = READ_ONCE(folio->flags);
+
+	if (folio_test_unevictable(folio))
+		return;
+
+	if (!folio_test_referenced(folio)) {
+		folio_set_referenced(folio);
+		return;
+	}
+
+	if (!folio_test_workingset(folio)) {
+		folio_set_workingset(folio);
+		return;
+	}
+
+	/* see the comment on MAX_NR_TIERS */
+	do {
+		new_flags = old_flags & LRU_REFS_MASK;
+		if (new_flags == LRU_REFS_MASK)
+			break;
+
+		new_flags += BIT(LRU_REFS_PGOFF);
+		new_flags |= old_flags & ~LRU_REFS_MASK;
+	} while (!try_cmpxchg(&folio->flags, &old_flags, new_flags));
+}
+#else
+static void folio_inc_refs(struct folio *folio)
+{
+}
+#endif /* CONFIG_LRU_GEN */
+
 /*
  * Mark a page as having seen activity.
  *
@@ -440,6 +474,11 @@ static void __lru_cache_activate_folio(struct folio *folio)
  */
 void folio_mark_accessed(struct folio *folio)
 {
+	if (lru_gen_enabled()) {
+		folio_inc_refs(folio);
+		return;
+	}
+
 	if (!folio_test_referenced(folio)) {
 		folio_set_referenced(folio);
 	} else if (folio_test_unevictable(folio)) {