summaryrefslogtreecommitdiffstats
path: root/drivers/md
diff options
context:
space:
mode:
authorShaohua Li <shli@kernel.org>2013-08-29 15:40:32 +0800
committerNeilBrown <neilb@suse.de>2013-09-02 10:31:29 +1000
commitbfc90cb0936f5b972706625f38f72c7cb726c20a (patch)
tree131194fdf489b6dd9eb3e165e6cd1c68bb4c629e /drivers/md
parent4d77e3ba88d085836f1e8e475e3131844dd89d04 (diff)
downloadop-kernel-dev-bfc90cb0936f5b972706625f38f72c7cb726c20a.zip
op-kernel-dev-bfc90cb0936f5b972706625f38f72c7cb726c20a.tar.gz
raid5: only wakeup necessary threads
If there are not enough stripes to handle, we'd better not always queue all available work_structs. If one worker can only handle small or even none stripes, it will impact request merge and create lock contention. With this patch, the number of work_struct running will depend on pending stripes number. Note: some statistics info used in the patch are accessed without locking protection. This should doesn't matter, we just try best to avoid queue unnecessary work_struct. Signed-off-by: Shaohua Li <shli@fusionio.com> Signed-off-by: NeilBrown <neilb@suse.de>
Diffstat (limited to 'drivers/md')
-rw-r--r--drivers/md/raid5.c41
-rw-r--r--drivers/md/raid5.h3
2 files changed, 38 insertions, 6 deletions
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 663a8e5..7ff4f25 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -77,6 +77,7 @@ static struct workqueue_struct *raid5_wq;
#define BYPASS_THRESHOLD 1
#define NR_HASH (PAGE_SIZE / sizeof(struct hlist_head))
#define HASH_MASK (NR_HASH - 1)
+#define MAX_STRIPE_BATCH 8
static inline struct hlist_head *stripe_hash(struct r5conf *conf, sector_t sect)
{
@@ -209,6 +210,7 @@ static void raid5_wakeup_stripe_thread(struct stripe_head *sh)
{
struct r5conf *conf = sh->raid_conf;
struct r5worker_group *group;
+ int thread_cnt;
int i, cpu = sh->cpu;
if (!cpu_online(cpu)) {
@@ -220,6 +222,8 @@ static void raid5_wakeup_stripe_thread(struct stripe_head *sh)
struct r5worker_group *group;
group = conf->worker_groups + cpu_to_group(cpu);
list_add_tail(&sh->lru, &group->handle_list);
+ group->stripes_cnt++;
+ sh->group = group;
}
if (conf->worker_cnt_per_group == 0) {
@@ -229,8 +233,20 @@ static void raid5_wakeup_stripe_thread(struct stripe_head *sh)
group = conf->worker_groups + cpu_to_group(sh->cpu);
- for (i = 0; i < conf->worker_cnt_per_group; i++)
- queue_work_on(sh->cpu, raid5_wq, &group->workers[i].work);
+ group->workers[0].working = true;
+ /* at least one worker should run to avoid race */
+ queue_work_on(sh->cpu, raid5_wq, &group->workers[0].work);
+
+ thread_cnt = group->stripes_cnt / MAX_STRIPE_BATCH - 1;
+ /* wakeup more workers */
+ for (i = 1; i < conf->worker_cnt_per_group && thread_cnt > 0; i++) {
+ if (group->workers[i].working == false) {
+ group->workers[i].working = true;
+ queue_work_on(sh->cpu, raid5_wq,
+ &group->workers[i].work);
+ thread_cnt--;
+ }
+ }
}
static void do_release_stripe(struct r5conf *conf, struct stripe_head *sh)
@@ -589,6 +605,10 @@ get_active_stripe(struct r5conf *conf, sector_t sector,
!test_bit(STRIPE_EXPANDING, &sh->state))
BUG();
list_del_init(&sh->lru);
+ if (sh->group) {
+ sh->group->stripes_cnt--;
+ sh->group = NULL;
+ }
}
}
} while (sh == NULL);
@@ -4153,15 +4173,18 @@ static struct stripe_head *__get_priority_stripe(struct r5conf *conf, int group)
{
struct stripe_head *sh = NULL, *tmp;
struct list_head *handle_list = NULL;
+ struct r5worker_group *wg = NULL;
if (conf->worker_cnt_per_group == 0) {
handle_list = &conf->handle_list;
} else if (group != ANY_GROUP) {
handle_list = &conf->worker_groups[group].handle_list;
+ wg = &conf->worker_groups[group];
} else {
int i;
for (i = 0; i < conf->group_cnt; i++) {
handle_list = &conf->worker_groups[i].handle_list;
+ wg = &conf->worker_groups[i];
if (!list_empty(handle_list))
break;
}
@@ -4208,11 +4231,16 @@ static struct stripe_head *__get_priority_stripe(struct r5conf *conf, int group)
if (conf->bypass_count < 0)
conf->bypass_count = 0;
}
+ wg = NULL;
}
if (!sh)
return NULL;
+ if (wg) {
+ wg->stripes_cnt--;
+ sh->group = NULL;
+ }
list_del_init(&sh->lru);
atomic_inc(&sh->count);
BUG_ON(atomic_read(&sh->count) != 1);
@@ -4919,8 +4947,8 @@ static int retry_aligned_read(struct r5conf *conf, struct bio *raid_bio)
return handled;
}
-#define MAX_STRIPE_BATCH 8
-static int handle_active_stripes(struct r5conf *conf, int group)
+static int handle_active_stripes(struct r5conf *conf, int group,
+ struct r5worker *worker)
{
struct stripe_head *batch[MAX_STRIPE_BATCH], *sh;
int i, batch_size = 0;
@@ -4963,7 +4991,8 @@ static void raid5_do_work(struct work_struct *work)
released = release_stripe_list(conf);
- batch_size = handle_active_stripes(conf, group_id);
+ batch_size = handle_active_stripes(conf, group_id, worker);
+ worker->working = false;
if (!batch_size && !released)
break;
handled += batch_size;
@@ -5025,7 +5054,7 @@ static void raid5d(struct md_thread *thread)
handled++;
}
- batch_size = handle_active_stripes(conf, ANY_GROUP);
+ batch_size = handle_active_stripes(conf, ANY_GROUP, NULL);
if (!batch_size && !released)
break;
handled += batch_size;
diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h
index 435b12d..2113ffa 100644
--- a/drivers/md/raid5.h
+++ b/drivers/md/raid5.h
@@ -213,6 +213,7 @@ struct stripe_head {
enum reconstruct_states reconstruct_state;
spinlock_t stripe_lock;
int cpu;
+ struct r5worker_group *group;
/**
* struct stripe_operations
* @target - STRIPE_OP_COMPUTE_BLK target
@@ -369,12 +370,14 @@ struct disk_info {
struct r5worker {
struct work_struct work;
struct r5worker_group *group;
+ bool working;
};
struct r5worker_group {
struct list_head handle_list;
struct r5conf *conf;
struct r5worker *workers;
+ int stripes_cnt;
};
struct r5conf {
OpenPOWER on IntegriCloud