dm cache policy mq: tweak algorithm that decides when to promote a block

Rather than maintaining a separate promote_threshold variable that we
periodically update we now use the hit count of the oldest clean
block.  Also add a fudge factor to discourage demoting dirty blocks.

With some tests this has a sizeable difference, because the old code
was too eager to demote blocks.  For example, device-mapper-test-suite's
git_extract_cache_quick test goes from taking 190 seconds, to 142
(linear on spindle takes 250).

Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>

diff --git a/Documentation/device-mapper/cache-policies.txt b/Documentation/device-mapper/cache-policies.txt
index 66c2774c0c64c0e8bf3a9aadc2a6f1b8102b5a12..7746e5dbfd40192ab86f00140a35a9e85b9c8ac0 100644 (file)
--- a/Documentation/device-mapper/cache-policies.txt
+++ b/Documentation/device-mapper/cache-policies.txt
@@ -58,9 +58,9 @@ since spindles tend to have good bandwidth. The io_tracker counts
 contiguous I/Os to try to spot when the io is in one of these sequential
 modes.
 
-Internally the mq policy maintains a promotion threshold variable.  If
-the hit count of a block not in the cache goes above this threshold it
-gets promoted to the cache.  The read, write and discard promote adjustment
+Internally the mq policy determines a promotion threshold.  If the hit
+count of a block not in the cache goes above this threshold it gets
+promoted to the cache.  The read, write and discard promote adjustment
 tunables allow you to tweak the promotion threshold by adding a small
 value based on the io type.  They default to 4, 8 and 1 respectively.
 If you're trying to quickly warm a new cache device you may wish to

diff --git a/drivers/md/dm-cache-policy-mq.c b/drivers/md/dm-cache-policy-mq.c
index 0e385e40909e74fcde4da4ee5d785149101ad9cb..334d098d720da90ec2ed3d31c2ecb43fdb8c4629 100644 (file)
--- a/drivers/md/dm-cache-policy-mq.c
+++ b/drivers/md/dm-cache-policy-mq.c
@@ -181,24 +181,30 @@ static void queue_shift_down(struct queue *q)
  * Gives us the oldest entry of the lowest popoulated level.  If the first
  * level is emptied then we shift down one level.
  */
-static struct list_head *queue_pop(struct queue *q)
+static struct list_head *queue_peek(struct queue *q)
 {
        unsigned level;
-       struct list_head *r;
 
        for (level = 0; level < NR_QUEUE_LEVELS; level++)
-               if (!list_empty(q->qs + level)) {
-                       r = q->qs[level].next;
-                       list_del(r);
+               if (!list_empty(q->qs + level))
+                       return q->qs[level].next;
 
-                       /* have we just emptied the bottom level? */
-                       if (level == 0 && list_empty(q->qs))
-                               queue_shift_down(q);
+       return NULL;
+}
 
-                       return r;
-               }
+static struct list_head *queue_pop(struct queue *q)
+{
+       struct list_head *r = queue_peek(q);
 
-       return NULL;
+       if (r) {
+               list_del(r);
+
+               /* have we just emptied the bottom level? */
+               if (list_empty(q->qs))
+                       queue_shift_down(q);
+       }
+
+       return r;
 }
 
 static struct list_head *list_pop(struct list_head *lh)
@@ -383,13 +389,6 @@ struct mq_policy {
        unsigned generation;
        unsigned generation_period; /* in lookups (will probably change) */
 
-       /*
-        * Entries in the pre_cache whose hit count passes the promotion
-        * threshold move to the cache proper.  Working out the correct
-        * value for the promotion_threshold is crucial to this policy.
-        */
-       unsigned promote_threshold;
-
        unsigned discard_promote_adjustment;
        unsigned read_promote_adjustment;
        unsigned write_promote_adjustment;
@@ -406,6 +405,7 @@ struct mq_policy {
 #define DEFAULT_DISCARD_PROMOTE_ADJUSTMENT 1
 #define DEFAULT_READ_PROMOTE_ADJUSTMENT 4
 #define DEFAULT_WRITE_PROMOTE_ADJUSTMENT 8
+#define DISCOURAGE_DEMOTING_DIRTY_THRESHOLD 128
 
 /*----------------------------------------------------------------*/
 
@@ -518,6 +518,12 @@ static struct entry *pop(struct mq_policy *mq, struct queue *q)
        return e;
 }
 
+static struct entry *peek(struct queue *q)
+{
+       struct list_head *h = queue_peek(q);
+       return h ? container_of(h, struct entry, list) : NULL;
+}
+
 /*
  * Has this entry already been updated?
  */
@@ -570,10 +576,6 @@ static void check_generation(struct mq_policy *mq)
                                        break;
                        }
                }
-
-               mq->promote_threshold = nr ? total / nr : 1;
-               if (mq->promote_threshold * nr < total)
-                       mq->promote_threshold++;
        }
 }
 
@@ -640,6 +642,30 @@ static int demote_cblock(struct mq_policy *mq, dm_oblock_t *oblock)
        return 0;
 }
 
+/*
+ * Entries in the pre_cache whose hit count passes the promotion
+ * threshold move to the cache proper.  Working out the correct
+ * value for the promotion_threshold is crucial to this policy.
+ */
+static unsigned promote_threshold(struct mq_policy *mq)
+{
+       struct entry *e;
+
+       if (any_free_cblocks(mq))
+               return 0;
+
+       e = peek(&mq->cache_clean);
+       if (e)
+               return e->hit_count;
+
+       e = peek(&mq->cache_dirty);
+       if (e)
+               return e->hit_count + DISCOURAGE_DEMOTING_DIRTY_THRESHOLD;
+
+       /* This should never happen */
+       return 0;
+}
+
 /*
  * We modify the basic promotion_threshold depending on the specific io.
  *
@@ -653,7 +679,7 @@ static unsigned adjusted_promote_threshold(struct mq_policy *mq,
                                           bool discarded_oblock, int data_dir)
 {
        if (data_dir == READ)
-               return mq->promote_threshold + mq->read_promote_adjustment;
+               return promote_threshold(mq) + mq->read_promote_adjustment;
 
        if (discarded_oblock && (any_free_cblocks(mq) || any_clean_cblocks(mq))) {
                /*
@@ -663,7 +689,7 @@ static unsigned adjusted_promote_threshold(struct mq_policy *mq,
                return mq->discard_promote_adjustment;
        }
 
-       return mq->promote_threshold + mq->write_promote_adjustment;
+       return promote_threshold(mq) + mq->write_promote_adjustment;
 }
 
 static bool should_promote(struct mq_policy *mq, struct entry *e,
@@ -1230,7 +1256,6 @@ static struct dm_cache_policy *mq_create(dm_cblock_t cache_size,
        mq->tick = 0;
        mq->hit_count = 0;
        mq->generation = 0;
-       mq->promote_threshold = 0;
        mq->discard_promote_adjustment = DEFAULT_DISCARD_PROMOTE_ADJUSTMENT;
        mq->read_promote_adjustment = DEFAULT_READ_PROMOTE_ADJUSTMENT;
        mq->write_promote_adjustment = DEFAULT_WRITE_PROMOTE_ADJUSTMENT;