1 files changed, 62 insertions, 95 deletions

diff --git a/block/blk-timeout.c b/block/blk-timeout.c
index 65f103563969..1b8de0417fc1 100644
--- a/block/blk-timeout.c
+++ b/block/blk-timeout.c
@@ -1,3 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Functions related to generic timeout handling of requests.
  */
@@ -7,6 +8,7 @@
 #include <linux/fault-inject.h>
 
 #include "blk.h"
+#include "blk-mq.h"
 
 #ifdef CONFIG_FAIL_IO_TIMEOUT
 
@@ -18,20 +20,18 @@ static int __init setup_fail_io_timeout(char *str)
 }
 __setup("fail_io_timeout=", setup_fail_io_timeout);
 
-int blk_should_fake_timeout(struct request_queue *q)
+bool __blk_should_fake_timeout(struct request_queue *q)
 {
-	if (!test_bit(QUEUE_FLAG_FAIL_IO, &q->queue_flags))
-		return 0;
-
 	return should_fail(&fail_io_timeout, 1);
 }
+EXPORT_SYMBOL_GPL(__blk_should_fake_timeout);
 
 static int __init fail_io_timeout_debugfs(void)
 {
 	struct dentry *dir = fault_create_debugfs_attr("fail_io_timeout",
 						NULL, &fail_io_timeout);
 
-	return IS_ERR(dir) ? PTR_ERR(dir) : 0;
+	return PTR_ERR_OR_ZERO(dir);
 }
 
 late_initcall(fail_io_timeout_debugfs);
@@ -56,12 +56,10 @@ ssize_t part_timeout_store(struct device *dev, struct device_attribute *attr,
 		char *p = (char *) buf;
 
 		val = simple_strtoul(p, &p, 10);
-		spin_lock_irq(q->queue_lock);
 		if (val)
-			queue_flag_set(QUEUE_FLAG_FAIL_IO, q);
+			blk_queue_flag_set(QUEUE_FLAG_FAIL_IO, q);
 		else
-			queue_flag_clear(QUEUE_FLAG_FAIL_IO, q);
-		spin_unlock_irq(q->queue_lock);
+			blk_queue_flag_clear(QUEUE_FLAG_FAIL_IO, q);
 	}
 
 	return count;
@@ -69,93 +67,55 @@ ssize_t part_timeout_store(struct device *dev, struct device_attribute *attr,
 
 #endif /* CONFIG_FAIL_IO_TIMEOUT */
 
-/*
- * blk_delete_timer - Delete/cancel timer for a given function.
- * @req:	request that we are canceling timer for
+/**
+ * blk_abort_request - Request recovery for the specified command
+ * @req:	pointer to the request of interest
  *
+ * This function requests that the block layer start recovery for the
+ * request by deleting the timer and calling the q's timeout function.
+ * LLDDs who implement their own error recovery MAY ignore the timeout
+ * event if they generated blk_abort_request.
  */
-void blk_delete_timer(struct request *req)
+void blk_abort_request(struct request *req)
 {
-	list_del_init(&req->timeout_list);
+	/*
+	 * All we need to ensure is that timeout scan takes place
+	 * immediately and that scan sees the new timeout value.
+	 * No need for fancy synchronizations.
+	 */
+	WRITE_ONCE(req->deadline, jiffies);
+	kblockd_schedule_work(&req->q->timeout_work);
 }
+EXPORT_SYMBOL_GPL(blk_abort_request);
 
-static void blk_rq_timed_out(struct request *req)
-{
-	struct request_queue *q = req->q;
-	enum blk_eh_timer_return ret = BLK_EH_RESET_TIMER;
-
-	if (q->rq_timed_out_fn)
-		ret = q->rq_timed_out_fn(req);
-	switch (ret) {
-	case BLK_EH_HANDLED:
-		__blk_complete_request(req);
-		break;
-	case BLK_EH_RESET_TIMER:
-		blk_clear_rq_complete(req);
-		blk_add_timer(req);
-		break;
-	case BLK_EH_NOT_HANDLED:
-		/*
-		 * LLD handles this for now but in the future
-		 * we can send a request msg to abort the command
-		 * and we can move more of the generic scsi eh code to
-		 * the blk layer.
-		 */
-		break;
-	default:
-		printk(KERN_ERR "block: bad eh return: %d\n", ret);
-		break;
-	}
-}
+static unsigned long blk_timeout_mask __read_mostly;
 
-void blk_rq_timed_out_timer(unsigned long data)
+static int __init blk_timeout_init(void)
 {
-	struct request_queue *q = (struct request_queue *) data;
-	unsigned long flags, next = 0;
-	struct request *rq, *tmp;
-	int next_set = 0;
-
-	spin_lock_irqsave(q->queue_lock, flags);
-
-	list_for_each_entry_safe(rq, tmp, &q->timeout_list, timeout_list) {
-		if (time_after_eq(jiffies, rq->deadline)) {
-			list_del_init(&rq->timeout_list);
-
-			/*
-			 * Check if we raced with end io completion
-			 */
-			if (blk_mark_rq_complete(rq))
-				continue;
-			blk_rq_timed_out(rq);
-		} else if (!next_set || time_after(next, rq->deadline)) {
-			next = rq->deadline;
-			next_set = 1;
-		}
-	}
+	blk_timeout_mask = roundup_pow_of_two(HZ) - 1;
+	return 0;
+}
 
-	if (next_set)
-		mod_timer(&q->timeout, round_jiffies_up(next));
+late_initcall(blk_timeout_init);
 
-	spin_unlock_irqrestore(q->queue_lock, flags);
+/*
+ * Just a rough estimate, we don't care about specific values for timeouts.
+ */
+static inline unsigned long blk_round_jiffies(unsigned long j)
+{
+	return (j + blk_timeout_mask) + 1;
 }
 
-/**
- * blk_abort_request -- Request request recovery for the specified command
- * @req:	pointer to the request of interest
- *
- * This function requests that the block layer start recovery for the
- * request by deleting the timer and calling the q's timeout function.
- * LLDDs who implement their own error recovery MAY ignore the timeout
- * event if they generated blk_abort_req. Must hold queue lock.
- */
-void blk_abort_request(struct request *req)
+unsigned long blk_rq_timeout(unsigned long timeout)
 {
-	if (blk_mark_rq_complete(req))
-		return;
-	blk_delete_timer(req);
-	blk_rq_timed_out(req);
+	unsigned long maxt;
+
+	maxt = blk_round_jiffies(jiffies + BLK_MAX_TIMEOUT);
+	if (time_after(timeout, maxt))
+		timeout = maxt;
+
+	return timeout;
 }
-EXPORT_SYMBOL_GPL(blk_abort_request);
 
 /**
  * blk_add_timer - Start timeout timer for a single request
@@ -170,12 +130,6 @@ void blk_add_timer(struct request *req)
 	struct request_queue *q = req->q;
 	unsigned long expiry;
 
-	if (!q->rq_timed_out_fn)
-		return;
-
-	BUG_ON(!list_empty(&req->timeout_list));
-	BUG_ON(test_bit(REQ_ATOM_COMPLETE, &req->atomic_flags));
-
 	/*
 	 * Some LLDs, like scsi, peek at the timeout to prevent a
 	 * command from being retried forever.
@@ -183,18 +137,31 @@ void blk_add_timer(struct request *req)
 	if (!req->timeout)
 		req->timeout = q->rq_timeout;
 
-	req->deadline = jiffies + req->timeout;
-	list_add_tail(&req->timeout_list, &q->timeout_list);
+	req->rq_flags &= ~RQF_TIMED_OUT;
+
+	expiry = jiffies + req->timeout;
+	WRITE_ONCE(req->deadline, expiry);
 
 	/*
 	 * If the timer isn't already pending or this timeout is earlier
 	 * than an existing one, modify the timer. Round up to next nearest
 	 * second.
 	 */
-	expiry = round_jiffies_up(req->deadline);
+	expiry = blk_rq_timeout(blk_round_jiffies(expiry));
 
 	if (!timer_pending(&q->timeout) ||
-	    time_before(expiry, q->timeout.expires))
-		mod_timer(&q->timeout, expiry);
-}
+	    time_before(expiry, q->timeout.expires)) {
+		unsigned long diff = q->timeout.expires - expiry;
 
+		/*
+		 * Due to added timer slack to group timers, the timer
+		 * will often be a little in front of what we asked for.
+		 * So apply some tolerance here too, otherwise we keep
+		 * modifying the timer because expires for value X
+		 * will be X + something.
+		 */
+		if (!timer_pending(&q->timeout) || (diff >= HZ / 2))
+			mod_timer(&q->timeout, expiry);
+	}
+
+}