Skip to contentSkip to conversation

decouple work_list protection from the big wqe->lock

PATCH Applied Hao Xu 2 participants 7 messages Feb 6, 2022 – Feb 14, 2022

Subsystems (2)
lore
Share & tools
lore

Hao Xu

author
#0 lore raw

cover decouple work_list protection from the big wqe->lock

wqe->lock is abused, it now protects acct->work_list, hash stuff, nr_workers, wqe->free_list and so on. Lets first get the work_list out of the wqe-lock mess by introduce a specific lock for work list. This is the first step to solve the huge contension between work insertion and work consumption.
good thing:
- split locking for bound and unbound work list
- reduce contension between work_list visit and (worker's)free_list.

For the hash stuff, since there won't be a work with same file in both bound and unbound work list, thus they won't visit same hash entry. it works well to use the new lock to protect hash stuff.

Results:
set max_unbound_worker = 4, test with echo-server:
nice -n -15 ./io_uring_echo_server -p 8081 -f -n 1000 -l 16 (-n connection, -l workload)
before this patch:
Samples: 2M of event 'cycles:ppp', Event count (approx.): 1239982111074

Overhead  Command          Shared Object         Symbol
  28.59%  iou-wrk-10021    [kernel.vmlinux]      [k] native_queued_spin_lock_slowpath
   8.89%  io_uring_echo_s  [kernel.vmlinux]      [k] native_queued_spin_lock_slowpath
   6.20%  iou-wrk-10021    [kernel.vmlinux]      [k] _raw_spin_lock
   2.45%  io_uring_echo_s  [kernel.vmlinux]      [k] io_prep_async_work
   2.36%  iou-wrk-10021    [kernel.vmlinux]      [k] _raw_spin_lock_irqsave
   2.29%  iou-wrk-10021    [kernel.vmlinux]      [k] io_worker_handle_work
   1.29%  io_uring_echo_s  [kernel.vmlinux]      [k] io_wqe_enqueue
   1.06%  iou-wrk-10021    [kernel.vmlinux]      [k] io_wqe_worker
   1.06%  io_uring_echo_s  [kernel.vmlinux]      [k] _raw_spin_lock
   1.03%  iou-wrk-10021    [kernel.vmlinux]      [k] __schedule
   0.99%  iou-wrk-10021    [kernel.vmlinux]      [k] tcp_sendmsg_locked

with this patch:
Samples: 1M of event 'cycles:ppp', Event count (approx.): 708446691943

Overhead  Command          Shared Object         Symbol
  16.86%  iou-wrk-10893    [kernel.vmlinux]      [k] native_queued_spin_lock_slowpat
   9.10%  iou-wrk-10893    [kernel.vmlinux]      [k] _raw_spin_lock
   4.53%  io_uring_echo_s  [kernel.vmlinux]      [k] native_queued_spin_lock_slowpat
   2.87%  iou-wrk-10893    [kernel.vmlinux]      [k] io_worker_handle_work
   2.57%  iou-wrk-10893    [kernel.vmlinux]      [k] _raw_spin_lock_irqsave
   2.56%  io_uring_echo_s  [kernel.vmlinux]      [k] io_prep_async_work
   1.82%  io_uring_echo_s  [kernel.vmlinux]      [k] _raw_spin_lock
   1.33%  iou-wrk-10893    [kernel.vmlinux]      [k] io_wqe_worker
   1.26%  io_uring_echo_s  [kernel.vmlinux]      [k] try_to_wake_up

spin_lock failure from 25.59% + 8.89% = 34.48% to 16.86% + 4.53% = 21.39% TPS is similar, while cpu usage is from almost 400% to 350%

Hao Xu (3):
io-wq: decouple work_list protection from the big wqe->lock
io-wq: reduce acct->lock crossing functions lock/unlock
io-wq: use IO_WQ_ACCT_NR rather than hardcoded number

 fs/io-wq.c | 114 ++++++++++++++++++++++++++---------------------------
 1 file changed, 57 insertions(+), 57 deletions(-)

base-commit: f6133fbd373811066c8441737e65f384c8f31974

Signature · 1 line
2.25.1
author
#1 lore raw

PATCH 1/3 io-wq: decouple work_list protection from the big wqe->lock

wqe->lock is abused, it now protects acct->work_list, hash stuff, nr_workers, wqe->free_list and so on. Lets first get the work_list out of the wqe-lock mess by introduce a specific lock for work list. This is the first step to solve the huge contension between work insertion and work consumption.
good thing:
- split locking for bound and unbound work list
- reduce contension between work_list visit and (worker's)free_list.

For the hash stuff, since there won't be a work with same file in both bound and unbound work list, thus they won't visit same hash entry. it works well to use the new lock to protect hash stuff.

Results:
set max_unbound_worker = 4, test with echo-server:
nice -n -15 ./io_uring_echo_server -p 8081 -f -n 1000 -l 16 (-n connection, -l workload)
before this patch:
Samples: 2M of event 'cycles:ppp', Event count (approx.): 1239982111074

Overhead  Command          Shared Object         Symbol
  28.59%  iou-wrk-10021    [kernel.vmlinux]      [k] native_queued_spin_lock_slowpath
   8.89%  io_uring_echo_s  [kernel.vmlinux]      [k] native_queued_spin_lock_slowpath
   6.20%  iou-wrk-10021    [kernel.vmlinux]      [k] _raw_spin_lock
   2.45%  io_uring_echo_s  [kernel.vmlinux]      [k] io_prep_async_work
   2.36%  iou-wrk-10021    [kernel.vmlinux]      [k] _raw_spin_lock_irqsave
   2.29%  iou-wrk-10021    [kernel.vmlinux]      [k] io_worker_handle_work
   1.29%  io_uring_echo_s  [kernel.vmlinux]      [k] io_wqe_enqueue
   1.06%  iou-wrk-10021    [kernel.vmlinux]      [k] io_wqe_worker
   1.06%  io_uring_echo_s  [kernel.vmlinux]      [k] _raw_spin_lock
   1.03%  iou-wrk-10021    [kernel.vmlinux]      [k] __schedule
   0.99%  iou-wrk-10021    [kernel.vmlinux]      [k] tcp_sendmsg_locked

with this patch:
Samples: 1M of event 'cycles:ppp', Event count (approx.): 708446691943

Overhead  Command          Shared Object         Symbol
  16.86%  iou-wrk-10893    [kernel.vmlinux]      [k] native_queued_spin_lock_slowpat
   9.10%  iou-wrk-10893    [kernel.vmlinux]      [k] _raw_spin_lock
   4.53%  io_uring_echo_s  [kernel.vmlinux]      [k] native_queued_spin_lock_slowpat
   2.87%  iou-wrk-10893    [kernel.vmlinux]      [k] io_worker_handle_work
   2.57%  iou-wrk-10893    [kernel.vmlinux]      [k] _raw_spin_lock_irqsave
   2.56%  io_uring_echo_s  [kernel.vmlinux]      [k] io_prep_async_work
   1.82%  io_uring_echo_s  [kernel.vmlinux]      [k] _raw_spin_lock
   1.33%  iou-wrk-10893    [kernel.vmlinux]      [k] io_wqe_worker
   1.26%  io_uring_echo_s  [kernel.vmlinux]      [k] try_to_wake_up

spin_lock failure from 25.59% + 8.89% = 34.48% to 16.86% + 4.53% = 21.39% TPS is similar, while cpu usage is from almost 400% to 350%

---
 fs/io-wq.c | 96 +++++++++++++++++++++++++++++-------------------------
 1 file changed, 52 insertions(+), 44 deletions(-)
fs/io-wq.c +52 -44
index bb7f161bb19c..9595616ccaa3 100644
@@ -76,6 +76,7 @@ struct io_wqe_acct {
 	unsigned max_workers;
 	int index;
 	atomic_t nr_running;
+	raw_spinlock_t lock;
 	struct io_wq_work_list work_list;
 	unsigned long flags;
 };
@@ -224,12 +225,12 @@ static void io_worker_exit(struct io_worker *worker)
 	if (worker->flags & IO_WORKER_F_FREE)
 		hlist_nulls_del_rcu(&worker->nulls_node);
 	list_del_rcu(&worker->all_list);
-	preempt_disable();
+	raw_spin_unlock(&wqe->lock);
 	io_wqe_dec_running(worker);
 	worker->flags = 0;
+	preempt_disable();
 	current->flags &= ~PF_IO_WORKER;
 	preempt_enable();
-	raw_spin_unlock(&wqe->lock);
 
 	kfree_rcu(worker, rcu);
 	io_worker_ref_put(wqe->wq);
@@ -385,7 +386,6 @@ static bool io_queue_worker_create(struct io_worker *worker,
 }
 
 static void io_wqe_dec_running(struct io_worker *worker)
-	__must_hold(wqe->lock)
 {
 	struct io_wqe_acct *acct = io_wqe_get_acct(worker);
 	struct io_wqe *wqe = worker->wqe;
@@ -393,13 +393,18 @@ static void io_wqe_dec_running(struct io_worker *worker)
 	if (!(worker->flags & IO_WORKER_F_UP))
 		return;
 
-	if (atomic_dec_and_test(&acct->nr_running) && io_acct_run_queue(acct)) {
-		atomic_inc(&acct->nr_running);
-		atomic_inc(&wqe->wq->worker_refs);
+	if (!atomic_dec_and_test(&acct->nr_running))
+		return;
+	raw_spin_lock(&wqe->lock);
+	if (!io_acct_run_queue(acct)) {
 		raw_spin_unlock(&wqe->lock);
-		io_queue_worker_create(worker, acct, create_worker_cb);
-		raw_spin_lock(&wqe->lock);
+		return;
 	}
+
+	raw_spin_unlock(&wqe->lock);
+	atomic_inc(&acct->nr_running);
+	atomic_inc(&wqe->wq->worker_refs);
+	io_queue_worker_create(worker, acct, create_worker_cb);
 }
 
 /*
@@ -407,11 +412,12 @@ static void io_wqe_dec_running(struct io_worker *worker)
  * it's currently on the freelist
  */
 static void __io_worker_busy(struct io_wqe *wqe, struct io_worker *worker)
-	__must_hold(wqe->lock)
 {
 	if (worker->flags & IO_WORKER_F_FREE) {
 		worker->flags &= ~IO_WORKER_F_FREE;
+		raw_spin_lock(&wqe->lock);
 		hlist_nulls_del_init_rcu(&worker->nulls_node);
+		raw_spin_unlock(&wqe->lock);
 	}
 }
 
@@ -456,7 +462,7 @@ static bool io_wait_on_hash(struct io_wqe *wqe, unsigned int hash)
 
 static struct io_wq_work *io_get_next_work(struct io_wqe_acct *acct,
 					   struct io_worker *worker)
-	__must_hold(wqe->lock)
+	__must_hold(acct->lock)
 {
 	struct io_wq_work_node *node, *prev;
 	struct io_wq_work *work, *tail;
@@ -498,9 +504,9 @@ static struct io_wq_work *io_get_next_work(struct io_wqe_acct *acct,
 		 * work being added and clearing the stalled bit.
 		 */
 		set_bit(IO_ACCT_STALLED_BIT, &acct->flags);
-		raw_spin_unlock(&wqe->lock);
+		raw_spin_unlock(&acct->lock);
 		unstalled = io_wait_on_hash(wqe, stall_hash);
-		raw_spin_lock(&wqe->lock);
+		raw_spin_lock(&acct->lock);
 		if (unstalled) {
 			clear_bit(IO_ACCT_STALLED_BIT, &acct->flags);
 			if (wq_has_sleeper(&wqe->wq->hash->wait))
@@ -538,7 +544,7 @@ static void io_assign_current_work(struct io_worker *worker,
 static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work);
 
 static void io_worker_handle_work(struct io_worker *worker)
-	__releases(wqe->lock)
+	__releases(acct->lock)
 {
 	struct io_wqe_acct *acct = io_wqe_get_acct(worker);
 	struct io_wqe *wqe = worker->wqe;
@@ -556,6 +562,7 @@ static void io_worker_handle_work(struct io_worker *worker)
 		 * clear the stalled flag.
 		 */
 		work = io_get_next_work(acct, worker);
+		raw_spin_unlock(&acct->lock);
 		if (work) {
 			__io_worker_busy(wqe, worker);
 
@@ -569,10 +576,9 @@ static void io_worker_handle_work(struct io_worker *worker)
 			raw_spin_lock(&worker->lock);
 			worker->next_work = work;
 			raw_spin_unlock(&worker->lock);
-		}
-		raw_spin_unlock(&wqe->lock);
-		if (!work)
+		} else {
 			break;
+		}
 		io_assign_current_work(worker, work);
 		__set_current_state(TASK_RUNNING);
 
@@ -609,7 +615,7 @@ static void io_worker_handle_work(struct io_worker *worker)
 			}
 		} while (work);
 
-		raw_spin_lock(&wqe->lock);
+		raw_spin_lock(&acct->lock);
 	} while (1);
 }
 
@@ -634,11 +640,14 @@ static int io_wqe_worker(void *data)
 
 		set_current_state(TASK_INTERRUPTIBLE);
 loop:
-		raw_spin_lock(&wqe->lock);
+		raw_spin_lock(&acct->lock);
 		if (io_acct_run_queue(acct)) {
 			io_worker_handle_work(worker);
 			goto loop;
+		} else {
+			raw_spin_unlock(&acct->lock);
 		}
+		raw_spin_lock(&wqe->lock);
 		/* timed out, exit unless we're the last worker */
 		if (last_timeout && acct->nr_workers > 1) {
 			acct->nr_workers--;
@@ -663,7 +672,7 @@ static int io_wqe_worker(void *data)
 	}
 
 	if (test_bit(IO_WQ_BIT_EXIT, &wq->state)) {
-		raw_spin_lock(&wqe->lock);
+		raw_spin_lock(&acct->lock);
 		io_worker_handle_work(worker);
 	}
 
@@ -705,10 +714,7 @@ void io_wq_worker_sleeping(struct task_struct *tsk)
 		return;
 
 	worker->flags &= ~IO_WORKER_F_RUNNING;
-
-	raw_spin_lock(&worker->wqe->lock);
 	io_wqe_dec_running(worker);
-	raw_spin_unlock(&worker->wqe->lock);
 }
 
 static void io_init_new_worker(struct io_wqe *wqe, struct io_worker *worker,
@@ -778,10 +784,12 @@ static void create_worker_cont(struct callback_head *cb)
 				.cancel_all	= true,
 			};
 
+			raw_spin_unlock(&wqe->lock);
 			while (io_acct_cancel_pending_work(wqe, acct, &match))
-				raw_spin_lock(&wqe->lock);
+				;
+		} else {
+			raw_spin_unlock(&wqe->lock);
 		}
-		raw_spin_unlock(&wqe->lock);
 		io_worker_ref_put(wqe->wq);
 		kfree(worker);
 		return;
@@ -914,6 +922,7 @@ static bool io_wq_work_match_item(struct io_wq_work *work, void *data)
 static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work)
 {
 	struct io_wqe_acct *acct = io_work_get_acct(wqe, work);
+	struct io_cb_cancel_data match;
 	unsigned work_flags = work->flags;
 	bool do_create;
 
@@ -927,10 +936,12 @@ static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work)
 		return;
 	}
 
-	raw_spin_lock(&wqe->lock);
+	raw_spin_lock(&acct->lock);
 	io_wqe_insert_work(wqe, work);
 	clear_bit(IO_ACCT_STALLED_BIT, &acct->flags);
+	raw_spin_unlock(&acct->lock);
 
+	raw_spin_lock(&wqe->lock);
 	rcu_read_lock();
 	do_create = !io_wqe_activate_free_worker(wqe, acct);
 	rcu_read_unlock();
@@ -946,18 +957,18 @@ static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work)
 			return;
 
 		raw_spin_lock(&wqe->lock);
-		/* fatal condition, failed to create the first worker */
-		if (!acct->nr_workers) {
-			struct io_cb_cancel_data match = {
-				.fn		= io_wq_work_match_item,
-				.data		= work,
-				.cancel_all	= false,
-			};
-
-			if (io_acct_cancel_pending_work(wqe, acct, &match))
-				raw_spin_lock(&wqe->lock);
+		if (acct->nr_workers) {
+			raw_spin_unlock(&wqe->lock);
+			return;
 		}
 		raw_spin_unlock(&wqe->lock);
+
+		/* fatal condition, failed to create the first worker */
+		match.fn		= io_wq_work_match_item,
+		match.data		= work,
+		match.cancel_all	= false,
+
+		io_acct_cancel_pending_work(wqe, acct, &match);
 	}
 }
 
@@ -1032,22 +1043,23 @@ static inline void io_wqe_remove_pending(struct io_wqe *wqe,
 static bool io_acct_cancel_pending_work(struct io_wqe *wqe,
 					struct io_wqe_acct *acct,
 					struct io_cb_cancel_data *match)
-	__releases(wqe->lock)
 {
 	struct io_wq_work_node *node, *prev;
 	struct io_wq_work *work;
 
+	raw_spin_lock(&acct->lock);
 	wq_list_for_each(node, prev, &acct->work_list) {
 		work = container_of(node, struct io_wq_work, list);
 		if (!match->fn(work, match->data))
 			continue;
 		io_wqe_remove_pending(wqe, work, prev);
-		raw_spin_unlock(&wqe->lock);
+		raw_spin_unlock(&acct->lock);
 		io_run_cancel(work, wqe);
 		match->nr_pending++;
 		/* not safe to continue after unlock */
 		return true;
 	}
+	raw_spin_unlock(&acct->lock);
 
 	return false;
 }
@@ -1061,7 +1073,6 @@ static void io_wqe_cancel_pending_work(struct io_wqe *wqe,
 		struct io_wqe_acct *acct = io_get_acct(wqe, i == 0);
 
 		if (io_acct_cancel_pending_work(wqe, acct, match)) {
-			raw_spin_lock(&wqe->lock);
 			if (match->cancel_all)
 				goto retry;
 			break;
@@ -1103,13 +1114,11 @@ enum io_wq_cancel io_wq_cancel_cb(struct io_wq *wq, work_cancel_fn *cancel,
 	for_each_node(node) {
 		struct io_wqe *wqe = wq->wqes[node];
 
-		raw_spin_lock(&wqe->lock);
 		io_wqe_cancel_pending_work(wqe, &match);
-		if (match.nr_pending && !match.cancel_all) {
-			raw_spin_unlock(&wqe->lock);
+		if (match.nr_pending && !match.cancel_all)
 			return IO_WQ_CANCEL_OK;
-		}
 
+		raw_spin_lock(&wqe->lock);
 		io_wqe_cancel_running_work(wqe, &match);
 		raw_spin_unlock(&wqe->lock);
 		if (match.nr_running && !match.cancel_all)
@@ -1190,6 +1199,7 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
 			acct->index = i;
 			atomic_set(&acct->nr_running, 0);
 			INIT_WQ_LIST(&acct->work_list);
+			raw_spin_lock_init(&acct->lock);
 		}
 		wqe->wq = wq;
 		raw_spin_lock_init(&wqe->lock);
@@ -1282,9 +1292,7 @@ static void io_wq_destroy(struct io_wq *wq)
 			.fn		= io_wq_work_match_all,
 			.cancel_all	= true,
 		};
-		raw_spin_lock(&wqe->lock);
 		io_wqe_cancel_pending_work(wqe, &match);
-		raw_spin_unlock(&wqe->lock);
 		free_cpumask_var(wqe->cpu_mask);
 		kfree(wqe);
 	}
Signature · 1 line
2.25.1
author
#2 lore raw

PATCH 2/3 io-wq: reduce acct->lock crossing functions lock/unlock

In reply to Hao Xu

reduce acct->lock lock and unlock in different functions to make the code clearer.

---
 fs/io-wq.c | 32 ++++++++++++--------------------
 1 file changed, 12 insertions(+), 20 deletions(-)
fs/io-wq.c +12 -20
index 9595616ccaa3..f7b7fa396faf 100644
@@ -239,10 +239,15 @@ static void io_worker_exit(struct io_worker *worker)
 
 static inline bool io_acct_run_queue(struct io_wqe_acct *acct)
 {
+	bool ret = false;
+
+	raw_spin_lock(&acct->lock);
 	if (!wq_list_empty(&acct->work_list) &&
 	    !test_bit(IO_ACCT_STALLED_BIT, &acct->flags))
-		return true;
-	return false;
+		ret = true;
+	raw_spin_unlock(&acct->lock);
+
+	return ret;
 }
 
 /*
@@ -395,13 +400,9 @@ static void io_wqe_dec_running(struct io_worker *worker)
 
 	if (!atomic_dec_and_test(&acct->nr_running))
 		return;
-	raw_spin_lock(&wqe->lock);
-	if (!io_acct_run_queue(acct)) {
-		raw_spin_unlock(&wqe->lock);
+	if (!io_acct_run_queue(acct))
 		return;
-	}
 
-	raw_spin_unlock(&wqe->lock);
 	atomic_inc(&acct->nr_running);
 	atomic_inc(&wqe->wq->worker_refs);
 	io_queue_worker_create(worker, acct, create_worker_cb);
@@ -544,7 +545,6 @@ static void io_assign_current_work(struct io_worker *worker,
 static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work);
 
 static void io_worker_handle_work(struct io_worker *worker)
-	__releases(acct->lock)
 {
 	struct io_wqe_acct *acct = io_wqe_get_acct(worker);
 	struct io_wqe *wqe = worker->wqe;
@@ -561,6 +561,7 @@ static void io_worker_handle_work(struct io_worker *worker)
 		 * can't make progress, any work completion or insertion will
 		 * clear the stalled flag.
 		 */
+		raw_spin_lock(&acct->lock);
 		work = io_get_next_work(acct, worker);
 		raw_spin_unlock(&acct->lock);
 		if (work) {
@@ -614,8 +615,6 @@ static void io_worker_handle_work(struct io_worker *worker)
 					wake_up(&wq->hash->wait);
 			}
 		} while (work);
-
-		raw_spin_lock(&acct->lock);
 	} while (1);
 }
 
@@ -639,14 +638,9 @@ static int io_wqe_worker(void *data)
 		long ret;
 
 		set_current_state(TASK_INTERRUPTIBLE);
-loop:
-		raw_spin_lock(&acct->lock);
-		if (io_acct_run_queue(acct)) {
+		while (io_acct_run_queue(acct))
 			io_worker_handle_work(worker);
-			goto loop;
-		} else {
-			raw_spin_unlock(&acct->lock);
-		}
+
 		raw_spin_lock(&wqe->lock);
 		/* timed out, exit unless we're the last worker */
 		if (last_timeout && acct->nr_workers > 1) {
@@ -671,10 +665,8 @@ static int io_wqe_worker(void *data)
 		last_timeout = !ret;
 	}
 
-	if (test_bit(IO_WQ_BIT_EXIT, &wq->state)) {
-		raw_spin_lock(&acct->lock);
+	if (test_bit(IO_WQ_BIT_EXIT, &wq->state))
 		io_worker_handle_work(worker);
-	}
 
 	audit_free(current);
 	io_worker_exit(worker);
Signature · 1 line
2.25.1
author
#3 lore raw

PATCH 3/3 io-wq: use IO_WQ_ACCT_NR rather than hardcoded number

In reply to Hao Xu

It's better to use the defined enum stuff not the hardcoded number to define array.

---
 fs/io-wq.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
fs/io-wq.c +2 -2
index f7b7fa396faf..5b93fa67d346 100644
@@ -92,7 +92,7 @@ enum {
  */
 struct io_wqe {
 	raw_spinlock_t lock;
-	struct io_wqe_acct acct[2];
+	struct io_wqe_acct acct[IO_WQ_ACCT_NR];
 
 	int node;
 
@@ -1376,7 +1376,7 @@ int io_wq_max_workers(struct io_wq *wq, int *new_count)
 	BUILD_BUG_ON((int) IO_WQ_ACCT_UNBOUND != (int) IO_WQ_UNBOUND);
 	BUILD_BUG_ON((int) IO_WQ_ACCT_NR      != 2);
 
-	for (i = 0; i < 2; i++) {
+	for (i = 0; i < IO_WQ_ACCT_NR; i++) {
 		if (new_count[i] > task_rlimit(current, RLIMIT_NPROC))
 			new_count[i] = task_rlimit(current, RLIMIT_NPROC);
 	}
Signature · 1 line
2.25.1
maintainer
#4 lore raw
In reply to Hao Xu
Quoted 45 lines · Hao Xu

wqe->lock is abused, it now protects acct->work_list, hash stuff, nr_workers, wqe->free_list and so on. Lets first get the work_list out of the wqe-lock mess by introduce a specific lock for work list. This is the first step to solve the huge contension between work insertion and work consumption.
good thing:
- split locking for bound and unbound work list
- reduce contension between work_list visit and (worker's)free_list.

For the hash stuff, since there won't be a work with same file in both bound and unbound work list, thus they won't visit same hash entry. it works well to use the new lock to protect hash stuff.

Results:
set max_unbound_worker = 4, test with echo-server:
nice -n -15 ./io_uring_echo_server -p 8081 -f -n 1000 -l 16 (-n connection, -l workload)
before this patch:
Samples: 2M of event 'cycles:ppp', Event count (approx.): 1239982111074

Overhead  Command          Shared Object         Symbol
  28.59%  iou-wrk-10021    [kernel.vmlinux]      [k] native_queued_spin_lock_slowpath
   8.89%  io_uring_echo_s  [kernel.vmlinux]      [k] native_queued_spin_lock_slowpath
   6.20%  iou-wrk-10021    [kernel.vmlinux]      [k] _raw_spin_lock
   2.45%  io_uring_echo_s  [kernel.vmlinux]      [k] io_prep_async_work
   2.36%  iou-wrk-10021    [kernel.vmlinux]      [k] _raw_spin_lock_irqsave
   2.29%  iou-wrk-10021    [kernel.vmlinux]      [k] io_worker_handle_work
   1.29%  io_uring_echo_s  [kernel.vmlinux]      [k] io_wqe_enqueue
   1.06%  iou-wrk-10021    [kernel.vmlinux]      [k] io_wqe_worker
   1.06%  io_uring_echo_s  [kernel.vmlinux]      [k] _raw_spin_lock
   1.03%  iou-wrk-10021    [kernel.vmlinux]      [k] __schedule
   0.99%  iou-wrk-10021    [kernel.vmlinux]      [k] tcp_sendmsg_locked

with this patch:
Samples: 1M of event 'cycles:ppp', Event count (approx.): 708446691943

Overhead  Command          Shared Object         Symbol
  16.86%  iou-wrk-10893    [kernel.vmlinux]      [k] native_queued_spin_lock_slowpat
   9.10%  iou-wrk-10893    [kernel.vmlinux]      [k] _raw_spin_lock
   4.53%  io_uring_echo_s  [kernel.vmlinux]      [k] native_queued_spin_lock_slowpat
   2.87%  iou-wrk-10893    [kernel.vmlinux]      [k] io_worker_handle_work
   2.57%  iou-wrk-10893    [kernel.vmlinux]      [k] _raw_spin_lock_irqsave
   2.56%  io_uring_echo_s  [kernel.vmlinux]      [k] io_prep_async_work
   1.82%  io_uring_echo_s  [kernel.vmlinux]      [k] _raw_spin_lock
   1.33%  iou-wrk-10893    [kernel.vmlinux]      [k] io_wqe_worker
   1.26%  io_uring_echo_s  [kernel.vmlinux]      [k] try_to_wake_up
Quoted 2 lines

spin_lock failure from 25.59% + 8.89% = 34.48% to 16.86% + 4.53% = 21.39% TPS is similar, while cpu usage is from almost 400% to 350%

I think this looks like a good start to improving the io-wq locking. I didnt spot anything immediately wrong with the series, my only worker was worker->flags protection, but I _think_ that looks OK to in terms of the worker itself doing the manipulations.

Let's queue this up for 5.18 testing, thanks!

Signature · 1 line
Jens Axboe
author
#5 lore raw
Quoted 46 lines · Jens Axboe
Quoted 45 lines · Hao Xu

wqe->lock is abused, it now protects acct->work_list, hash stuff, nr_workers, wqe->free_list and so on. Lets first get the work_list out of the wqe-lock mess by introduce a specific lock for work list. This is the first step to solve the huge contension between work insertion and work consumption.
good thing:
- split locking for bound and unbound work list
- reduce contension between work_list visit and (worker's)free_list.

For the hash stuff, since there won't be a work with same file in both bound and unbound work list, thus they won't visit same hash entry. it works well to use the new lock to protect hash stuff.

Results:
set max_unbound_worker = 4, test with echo-server:
nice -n -15 ./io_uring_echo_server -p 8081 -f -n 1000 -l 16 (-n connection, -l workload)
before this patch:
Samples: 2M of event 'cycles:ppp', Event count (approx.): 1239982111074
Overhead Command Shared Object Symbol

   28.59%  iou-wrk-10021    [kernel.vmlinux]      [k] native_queued_spin_lock_slowpath
    8.89%  io_uring_echo_s  [kernel.vmlinux]      [k] native_queued_spin_lock_slowpath
    6.20%  iou-wrk-10021    [kernel.vmlinux]      [k] _raw_spin_lock
    2.45%  io_uring_echo_s  [kernel.vmlinux]      [k] io_prep_async_work
    2.36%  iou-wrk-10021    [kernel.vmlinux]      [k] _raw_spin_lock_irqsave
    2.29%  iou-wrk-10021    [kernel.vmlinux]      [k] io_worker_handle_work
    1.29%  io_uring_echo_s  [kernel.vmlinux]      [k] io_wqe_enqueue
    1.06%  iou-wrk-10021    [kernel.vmlinux]      [k] io_wqe_worker
    1.06%  io_uring_echo_s  [kernel.vmlinux]      [k] _raw_spin_lock
    1.03%  iou-wrk-10021    [kernel.vmlinux]      [k] __schedule
    0.99%  iou-wrk-10021    [kernel.vmlinux]      [k] tcp_sendmsg_locked

with this patch:
Samples: 1M of event 'cycles:ppp', Event count (approx.): 708446691943
Overhead Command Shared Object Symbol

   16.86%  iou-wrk-10893    [kernel.vmlinux]      [k] native_queued_spin_lock_slowpat
    9.10%  iou-wrk-10893    [kernel.vmlinux]      [k] _raw_spin_lock
    4.53%  io_uring_echo_s  [kernel.vmlinux]      [k] native_queued_spin_lock_slowpat
    2.87%  iou-wrk-10893    [kernel.vmlinux]      [k] io_worker_handle_work
    2.57%  iou-wrk-10893    [kernel.vmlinux]      [k] _raw_spin_lock_irqsave
    2.56%  io_uring_echo_s  [kernel.vmlinux]      [k] io_prep_async_work
    1.82%  io_uring_echo_s  [kernel.vmlinux]      [k] _raw_spin_lock
    1.33%  iou-wrk-10893    [kernel.vmlinux]      [k] io_wqe_worker
    1.26%  io_uring_echo_s  [kernel.vmlinux]      [k] try_to_wake_up
Quoted 6 lines
Quoted 2 lines

spin_lock failure from 25.59% + 8.89% = 34.48% to 16.86% + 4.53% = 21.39% TPS is similar, while cpu usage is from almost 400% to 350%

I think this looks like a good start to improving the io-wq locking. I didnt spot anything immediately wrong with the series, my only worker was worker->flags protection, but I _think_ that looks OK to in terms of the worker itself doing the manipulations.

Yes, I went over the code  to make sure worker->flags is manipulated by the worker

itself when doing coding.

Regards,

Hao

Quoted 3 lines

Let's queue this up for 5.18 testing, thanks!

maintainer Applied
#6 lore raw
In reply to Hao Xu
Quoted 10 lines · Hao Xu

wqe->lock is abused, it now protects acct->work_list, hash stuff, nr_workers, wqe->free_list and so on. Lets first get the work_list out of the wqe-lock mess by introduce a specific lock for work list. This is the first step to solve the huge contension between work insertion and work consumption.
good thing:
- split locking for bound and unbound work list
- reduce contension between work_list visit and (worker's)free_list.

[...]

Applied, thanks!

[1/3] io-wq: decouple work_list protection from the big wqe->lock
(no commit info)
[2/3] io-wq: reduce acct->lock crossing functions lock/unlock
(no commit info)
[3/3] io-wq: use IO_WQ_ACCT_NR rather than hardcoded number
(no commit info)

Best regards,

Signature · 1 line
Jens Axboe