[PATCH RFC] mm: increase page waitqueue hash size

[Nicholas Piggin <npiggin@gmail.com>]

The page waitqueue hash is a bit small (256 entries) on very big systems. A
16 socket 1536 thread POWER9 system was found to encounter hash collisions
and excessive time in waitqueue locking at times. This was intermittent and
hard to reproduce easily with the setup we had (very little real IO
capacity). The thought is some important pages happened to collide in the
hash, slowing down page locking, causing the problem to snowball.

An small test case was made where threads would write and fsync different
pages, generating just a small amount of contention across many pages.

Increasing page waitqueue hash size to 262144 entries increased throughput
by 182% while also reducing standard deviation 3x. perf before the increase:

  36.23%  [k] _raw_spin_lock_irqsave                -      -
              |
              |--34.60%--wake_up_page_bit
              |          0
              |          iomap_write_end.isra.38
              |          iomap_write_actor
              |          iomap_apply
              |          iomap_file_buffered_write
              |          xfs_file_buffered_aio_write
              |          new_sync_write

  17.93%  [k] native_queued_spin_lock_slowpath      -      -
              |
              |--16.74%--_raw_spin_lock_irqsave
              |          |
              |           --16.44%--wake_up_page_bit
              |                     iomap_write_end.isra.38
              |                     iomap_write_actor
              |                     iomap_apply
              |                     iomap_file_buffered_write
              |                     xfs_file_buffered_aio_write

This patch uses alloc_large_system_hash to allocate a bigger system hash
that scales somewhat with memory size.

This hash could be made per-node, which should help reduce remote accesses
on well localised workloads, but that adds some complexity with hotplug,
so until we get a less artificial workload to test with, let's keep it
simple.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---

 mm/filemap.c | 24 +++++++++++++++++++++---
 1 file changed, 21 insertions(+), 3 deletions(-)

diff --git a/mm/filemap.c b/mm/filemap.c
index 1aaea26556cc..d3cd158f0c3f 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -34,6 +34,7 @@
 #include <linux/security.h>
 #include <linux/cpuset.h>
 #include <linux/hugetlb.h>
+#include <linux/memblock.h>
 #include <linux/memcontrol.h>
 #include <linux/cleancache.h>
 #include <linux/shmem_fs.h>
@@ -969,19 +970,36 @@ EXPORT_SYMBOL(__page_cache_alloc);
  * at a cost of "thundering herd" phenomena during rare hash
  * collisions.
  */
-#define PAGE_WAIT_TABLE_BITS 8
-#define PAGE_WAIT_TABLE_SIZE (1 << PAGE_WAIT_TABLE_BITS)
+#define PAGE_WAIT_TABLE_SIZE (1 << page_wait_table_bits)
+#if CONFIG_BASE_SMALL
+static const unsigned int page_wait_table_bits = 4;
 static wait_queue_head_t page_wait_table[PAGE_WAIT_TABLE_SIZE] __cacheline_aligned;
+#else
+static unsigned int page_wait_table_bits __ro_after_init;
+static wait_queue_head_t *page_wait_table __ro_after_init;
+#endif
 
 static wait_queue_head_t *page_waitqueue(struct page *page)
 {
-	return &page_wait_table[hash_ptr(page, PAGE_WAIT_TABLE_BITS)];
+	return &page_wait_table[hash_ptr(page, page_wait_table_bits)];
 }
 
 void __init pagecache_init(void)
 {
 	int i;
 
+	if (!CONFIG_BASE_SMALL) {
+		page_wait_table = alloc_large_system_hash("Page waitqueue hash",
+							sizeof(wait_queue_head_t),
+							0,
+							21,
+							0,
+							&page_wait_table_bits,
+							NULL,
+							0,
+							0);
+	}
+
 	for (i = 0; i < PAGE_WAIT_TABLE_SIZE; i++)
 		init_waitqueue_head(&page_wait_table[i]);
 
-- 
2.23.0