lkml.org 
[lkml]   [2016]   [Jan]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
Subject[PATCH 1/3] mm: workingset: eviction buckets for bigmem/lowbit machines
space will get tight once we need to identify the memcg. add this to
stretch out the necessary distance by sacrificing granularity.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
---
mm/workingset.c | 40 +++++++++++++++++++++++++++++++++++-----
1 file changed, 35 insertions(+), 5 deletions(-)

diff --git a/mm/workingset.c b/mm/workingset.c
index 61ead9e5549d..6f3ba184ffb2 100644
--- a/mm/workingset.c
+++ b/mm/workingset.c
@@ -152,8 +152,23 @@
* refault distance will immediately activate the refaulting page.
*/

+#define EVICTION_SHIFT (NODES_SHIFT + ZONES_SHIFT + \
+ RADIX_TREE_EXCEPTIONAL_SHIFT)
+#define EVICTION_MASK (~0UL >> EVICTION_SHIFT)
+
+/*
+ * Eviction timestamps need to be able to cover the full range of
+ * actionable refaults. However, bits are tight in the radix tree
+ * entry, and after storing the identifier for the lruvec there might
+ * not be enough left to represent every single actionable refault. In
+ * that case, we have to sacrifice granularity for distance, and group
+ * evictions into coarser buckets by shaving off lower timestamp bits.
+ */
+static unsigned int bucket_order;
+
static void *pack_shadow(unsigned long eviction, struct zone *zone)
{
+ eviction >>= bucket_order;
eviction = (eviction << NODES_SHIFT) | zone_to_nid(zone);
eviction = (eviction << ZONES_SHIFT) | zone_idx(zone);
eviction = (eviction << RADIX_TREE_EXCEPTIONAL_SHIFT);
@@ -168,7 +183,6 @@ static void unpack_shadow(void *shadow,
unsigned long entry = (unsigned long)shadow;
unsigned long eviction;
unsigned long refault;
- unsigned long mask;
int zid, nid;

entry >>= RADIX_TREE_EXCEPTIONAL_SHIFT;
@@ -176,13 +190,12 @@ static void unpack_shadow(void *shadow,
entry >>= ZONES_SHIFT;
nid = entry & ((1UL << NODES_SHIFT) - 1);
entry >>= NODES_SHIFT;
- eviction = entry;
+ eviction = entry << bucket_order;

*zone = NODE_DATA(nid)->node_zones + zid;

refault = atomic_long_read(&(*zone)->inactive_age);
- mask = ~0UL >> (NODES_SHIFT + ZONES_SHIFT +
- RADIX_TREE_EXCEPTIONAL_SHIFT);
+
/*
* The unsigned subtraction here gives an accurate distance
* across inactive_age overflows in most cases.
@@ -199,7 +212,7 @@ static void unpack_shadow(void *shadow,
* inappropriate activation leading to pressure on the active
* list is not a problem.
*/
- *distance = (refault - eviction) & mask;
+ *distance = (refault - eviction) & EVICTION_MASK;
}

/**
@@ -398,8 +411,25 @@ static struct lock_class_key shadow_nodes_key;

static int __init workingset_init(void)
{
+ unsigned int timestamp_bits;
+ unsigned int max_order;
int ret;

+ BUILD_BUG_ON(BITS_PER_LONG < EVICTION_SHIFT);
+ /*
+ * Calculate the eviction bucket size to cover the longest
+ * actionable refault distance, which is currently half of
+ * memory (totalram_pages/2). However, memory hotplug may add
+ * some more pages at runtime, so keep working with up to
+ * double the initial memory by using totalram_pages as-is.
+ */
+ timestamp_bits = BITS_PER_LONG - EVICTION_SHIFT;
+ max_order = fls_long(totalram_pages - 1);
+ if (max_order > timestamp_bits)
+ bucket_order = max_order - timestamp_bits;
+ printk("workingset: timestamp_bits=%d max_order=%d bucket_order=%u\n",
+ timestamp_bits, max_order, bucket_order);
+
ret = list_lru_init_key(&workingset_shadow_nodes, &shadow_nodes_key);
if (ret)
goto err;
--
2.7.0
\
 
 \ /
  Last update: 2016-01-25 18:01    [W:0.051 / U:0.208 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site