dm thin: sort the per thin deferred bios using an rb_tree
A thin-pool will allocate blocks using FIFO order for all thin devices
which share the thin-pool. Because of this simplistic allocation the
thin-pool's space can become fragmented quite easily; especially when
multiple threads are requesting blocks in parallel.
Sort each thin device's deferred_bio_list based on logical sector to
help reduce fragmentation of the thin-pool's ondisk layout.
The following tables illustrate the realized gains/potential offered by
sorting each thin device's deferred_bio_list. An "io size"-sized random
read of the device would result in "seeks/io" fragments being read, with
an average "distance/seek" between each fragment.
Data was written to a single thin device using multiple threads via
iozone (8 threads, 64K for both the block_size and io_size).
unsorted:
io size seeks/io distance/seek
--------------------------------------
4k 0.000 0b
16k 0.013 11m
64k 0.065 11m
256k 0.274 10m
1m 1.109 10m
4m 4.411 10m
16m 17.097 11m
64m 60.055 13m
256m 148.798 25m
1g 809.929 21m
sorted:
io size seeks/io distance/seek
--------------------------------------
4k 0.000 0b
16k 0.000 1g
64k 0.001 1g
256k 0.003 1g
1m 0.011 1g
4m 0.045 1g
16m 0.181 1g
64m 0.747 1011m
256m 3.299 1g
1g 14.373 1g
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Acked-by: Joe Thornber <ejt@redhat.com>
diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c
index 08e62aef..53728be 100644
--- a/drivers/md/dm-thin.c
+++ b/drivers/md/dm-thin.c
@@ -16,6 +16,7 @@
#include <linux/init.h>
#include <linux/module.h>
#include <linux/slab.h>
+#include <linux/rbtree.h>
#define DM_MSG_PREFIX "thin"
@@ -230,6 +231,7 @@
spinlock_t lock;
struct bio_list deferred_bio_list;
struct bio_list retry_on_resume_list;
+ struct rb_root sort_bio_list; /* sorted list of deferred bios */
};
/*----------------------------------------------------------------*/
@@ -371,6 +373,7 @@
struct dm_deferred_entry *shared_read_entry;
struct dm_deferred_entry *all_io_entry;
struct dm_thin_new_mapping *overwrite_mapping;
+ struct rb_node rb_node;
};
static void requeue_bio_list(struct thin_c *tc, struct bio_list *master)
@@ -1367,12 +1370,77 @@
jiffies > pool->last_commit_jiffies + COMMIT_PERIOD;
}
+#define thin_pbd(node) rb_entry((node), struct dm_thin_endio_hook, rb_node)
+#define thin_bio(pbd) dm_bio_from_per_bio_data((pbd), sizeof(struct dm_thin_endio_hook))
+
+static void __thin_bio_rb_add(struct thin_c *tc, struct bio *bio)
+{
+ struct rb_node **rbp, *parent;
+ struct dm_thin_endio_hook *pbd;
+ sector_t bi_sector = bio->bi_iter.bi_sector;
+
+ rbp = &tc->sort_bio_list.rb_node;
+ parent = NULL;
+ while (*rbp) {
+ parent = *rbp;
+ pbd = thin_pbd(parent);
+
+ if (bi_sector < thin_bio(pbd)->bi_iter.bi_sector)
+ rbp = &(*rbp)->rb_left;
+ else
+ rbp = &(*rbp)->rb_right;
+ }
+
+ pbd = dm_per_bio_data(bio, sizeof(struct dm_thin_endio_hook));
+ rb_link_node(&pbd->rb_node, parent, rbp);
+ rb_insert_color(&pbd->rb_node, &tc->sort_bio_list);
+}
+
+static void __extract_sorted_bios(struct thin_c *tc)
+{
+ struct rb_node *node;
+ struct dm_thin_endio_hook *pbd;
+ struct bio *bio;
+
+ for (node = rb_first(&tc->sort_bio_list); node; node = rb_next(node)) {
+ pbd = thin_pbd(node);
+ bio = thin_bio(pbd);
+
+ bio_list_add(&tc->deferred_bio_list, bio);
+ rb_erase(&pbd->rb_node, &tc->sort_bio_list);
+ }
+
+ WARN_ON(!RB_EMPTY_ROOT(&tc->sort_bio_list));
+}
+
+static void __sort_thin_deferred_bios(struct thin_c *tc)
+{
+ struct bio *bio;
+ struct bio_list bios;
+
+ bio_list_init(&bios);
+ bio_list_merge(&bios, &tc->deferred_bio_list);
+ bio_list_init(&tc->deferred_bio_list);
+
+ /* Sort deferred_bio_list using rb-tree */
+ while ((bio = bio_list_pop(&bios)))
+ __thin_bio_rb_add(tc, bio);
+
+ /*
+ * Transfer the sorted bios in sort_bio_list back to
+ * deferred_bio_list to allow lockless submission of
+ * all bios.
+ */
+ __extract_sorted_bios(tc);
+}
+
static void process_thin_deferred_bios(struct thin_c *tc)
{
struct pool *pool = tc->pool;
unsigned long flags;
struct bio *bio;
struct bio_list bios;
+ struct blk_plug plug;
if (tc->requeue_mode) {
requeue_bio_list(tc, &tc->deferred_bio_list);
@@ -1382,10 +1450,20 @@
bio_list_init(&bios);
spin_lock_irqsave(&tc->lock, flags);
+
+ if (bio_list_empty(&tc->deferred_bio_list)) {
+ spin_unlock_irqrestore(&tc->lock, flags);
+ return;
+ }
+
+ __sort_thin_deferred_bios(tc);
+
bio_list_merge(&bios, &tc->deferred_bio_list);
bio_list_init(&tc->deferred_bio_list);
+
spin_unlock_irqrestore(&tc->lock, flags);
+ blk_start_plug(&plug);
while ((bio = bio_list_pop(&bios))) {
/*
* If we've got no free new_mapping structs, and processing
@@ -1405,6 +1483,7 @@
else
pool->process_bio(tc, bio);
}
+ blk_finish_plug(&plug);
}
static void process_deferred_bios(struct pool *pool)
@@ -2964,7 +3043,7 @@
.name = "thin-pool",
.features = DM_TARGET_SINGLETON | DM_TARGET_ALWAYS_WRITEABLE |
DM_TARGET_IMMUTABLE,
- .version = {1, 11, 0},
+ .version = {1, 12, 0},
.module = THIS_MODULE,
.ctr = pool_ctr,
.dtr = pool_dtr,
@@ -3040,6 +3119,7 @@
spin_lock_init(&tc->lock);
bio_list_init(&tc->deferred_bio_list);
bio_list_init(&tc->retry_on_resume_list);
+ tc->sort_bio_list = RB_ROOT;
if (argc == 3) {
r = dm_get_device(ti, argv[2], FMODE_READ, &origin_dev);
@@ -3287,7 +3367,7 @@
static struct target_type thin_target = {
.name = "thin",
- .version = {1, 11, 0},
+ .version = {1, 12, 0},
.module = THIS_MODULE,
.ctr = thin_ctr,
.dtr = thin_dtr,
