io_uring: split work handling part of SQPOLL into helper

This is done in preparation for handling more than one ctx, but it also
cleans up the code a bit since io_sq_thread() was a bit too unwieldy to
get a get overview on.

__io_sq_thread() is now the main handler, and it returns an enum sq_ret
that tells io_sq_thread() what it ended up doing. The parent then makes
a decision on idle, spinning, or work handling based on that.
Signed-off-by: Jens Axboe <axboe@kernel.dk>

fs/io_uring.c

@@ -6642,32 +6642,26 @@ static int io_sq_wake_function(struct wait_queue_entry *wqe, unsigned mode,
 	return ret;
 }
 
-static int io_sq_thread(void *data)
-{
-	struct io_ring_ctx *ctx = data;
-	const struct cred *old_cred;
-	unsigned long timeout;
-	int ret = 0;
-
-	init_wait(&ctx->sqo_wait_entry);
-	ctx->sqo_wait_entry.func = io_sq_wake_function;
-
-	complete(&ctx->sq_thread_comp);
-
-	old_cred = override_creds(ctx->creds);
+enum sq_ret {
+	SQT_IDLE	= 1,
+	SQT_SPIN	= 2,
+	SQT_DID_WORK	= 4,
+};
 
-	timeout = jiffies + ctx->sq_thread_idle;
-	while (!kthread_should_park()) {
+static enum sq_ret __io_sq_thread(struct io_ring_ctx *ctx,
+				  unsigned long start_jiffies)
+{
+	unsigned long timeout = start_jiffies + ctx->sq_thread_idle;
 	unsigned int to_submit;
+	int ret = 0;
 
+again:
 	if (!list_empty(&ctx->iopoll_list)) {
 		unsigned nr_events = 0;
 
 		mutex_lock(&ctx->uring_lock);
 		if (!list_empty(&ctx->iopoll_list) && !need_resched())
 			io_do_iopoll(ctx, &nr_events, 0);
-			else
-				timeout = jiffies + ctx->sq_thread_idle;
 		mutex_unlock(&ctx->uring_lock);
 	}
 
@@ -6695,11 +6689,8 @@ static int io_sq_thread(void *data)
 		 */
 		if (!list_empty(&ctx->iopoll_list) || need_resched() ||
 		    (!time_after(jiffies, timeout) && ret != -EBUSY &&
-			    !percpu_ref_is_dying(&ctx->refs))) {
-				io_run_task_work();
-				cond_resched();
-				continue;
-			}
+		    !percpu_ref_is_dying(&ctx->refs)))
+			return SQT_SPIN;
 
 		prepare_to_wait(ctx->sqo_wait, &ctx->sqo_wait_entry,
 					TASK_INTERRUPTIBLE);
@@ -6714,38 +6705,56 @@ static int io_sq_thread(void *data)
 		if ((ctx->flags & IORING_SETUP_IOPOLL) &&
 		    !list_empty_careful(&ctx->iopoll_list)) {
 			finish_wait(ctx->sqo_wait, &ctx->sqo_wait_entry);
-				continue;
+			goto again;
 		}
 
 		io_ring_set_wakeup_flag(ctx);
 
 		to_submit = io_sqring_entries(ctx);
-			if (!to_submit || ret == -EBUSY) {
-				if (kthread_should_park()) {
-					finish_wait(ctx->sqo_wait, &ctx->sqo_wait_entry);
-					break;
+		if (!to_submit || ret == -EBUSY)
+			return SQT_IDLE;
 	}
-				if (io_run_task_work()) {
-					finish_wait(ctx->sqo_wait, &ctx->sqo_wait_entry);
-					io_ring_clear_wakeup_flag(ctx);
-					continue;
-				}
-				schedule();
-				finish_wait(ctx->sqo_wait, &ctx->sqo_wait_entry);
 
-				ret = 0;
-				continue;
-			}
 	finish_wait(ctx->sqo_wait, &ctx->sqo_wait_entry);
-
 	io_ring_clear_wakeup_flag(ctx);
-		}
 
 	mutex_lock(&ctx->uring_lock);
 	if (likely(!percpu_ref_is_dying(&ctx->refs)))
 		ret = io_submit_sqes(ctx, to_submit);
 	mutex_unlock(&ctx->uring_lock);
-		timeout = jiffies + ctx->sq_thread_idle;
+	return SQT_DID_WORK;
+}
+
+static int io_sq_thread(void *data)
+{
+	struct io_ring_ctx *ctx = data;
+	const struct cred *old_cred;
+	unsigned long start_jiffies;
+
+	init_wait(&ctx->sqo_wait_entry);
+	ctx->sqo_wait_entry.func = io_sq_wake_function;
+
+	complete(&ctx->sq_thread_comp);
+
+	old_cred = override_creds(ctx->creds);
+
+	start_jiffies = jiffies;
+	while (!kthread_should_park()) {
+		enum sq_ret ret;
+
+		ret = __io_sq_thread(ctx, start_jiffies);
+		switch (ret) {
+		case SQT_IDLE:
+			schedule();
+			start_jiffies = jiffies;
+			continue;
+		case SQT_SPIN:
+			io_run_task_work();
+			cond_resched();
+			fallthrough;
+		case SQT_DID_WORK:
+			continue;
+		}
 	}
 
 	io_run_task_work();