block: add a switch for precise iostat accounting

hulk inclusion
category: bugfix
bugzilla: 39265, https://gitee.com/openeuler/kernel/issues/I4WC06
CVE: NA

-----------------------------------------------

When the inflight IOs are slow and no new IOs are issued, we expect
iostat could manifest the IO hang problem. However after
commit 9c6dea45 ("block: delete part_round_stats and switch to less
precise counting"), io_tick and time_in_queue will not be updated until
the end of IO, and the avgqu-sz and %util columns of iostat will be zero.

To fix it, we could fallback to the implementation before commit
9c6dea45, but it may cause performance regression on NVMe device
or bio-based device (due to overhead of inflight calculation),
so add a switch to control whether or not to use precise iostat
accounting. It can be enabled by adding "precise_iostat=1" in kernel
boot cmdline. When precise accouting is enabled, io_tick and time_in_queue
will be updated when accessing /proc/diskstats and
/sys/block/sdX/sdXN/stat.

Fixes: 9c6dea45 ("block: delete part_round_stats and switch to less precise counting")
Signed-off-by: NZhang Wensheng <zhangwensheng5@huawei.com>
Reviewed-by: NJason Yan <yanaijie@huawei.com>
Signed-off-by: NYang Yingliang <yangyingliang@huawei.com>

block/bio.c

@@ -1706,9 +1706,13 @@ void generic_end_io_acct(struct request_queue *q, int req_op,
 	const int sgrp = op_stat_group(req_op);
 	int cpu = part_stat_lock();
 
-	update_io_ticks(cpu, part, now);
+	if (precise_iostat) {
+		part_round_stats(q, cpu, part);
+	} else {
+		update_io_ticks(cpu, part, now);
+		part_stat_add(cpu, part, time_in_queue, duration);
+	}
 	part_stat_add(cpu, part, nsecs[sgrp], jiffies_to_nsecs(duration));
-	part_stat_add(cpu, part, time_in_queue, duration);
 	part_dec_in_flight(q, part, op_is_write(req_op));
 
 	part_stat_unlock();

block/blk-core.c

@@ -56,6 +56,20 @@ EXPORT_TRACEPOINT_SYMBOL_GPL(block_unplug);
 
 DEFINE_IDA(blk_queue_ida);
 
+bool precise_iostat;
+static int __init precise_iostat_setup(char *str)
+{
+	bool precise;
+
+	if (!strtobool(str, &precise)) {
+		precise_iostat = precise;
+		pr_info("precise iostat %d\n", precise_iostat);
+	}
+
+	return 1;
+}
+__setup("precise_iostat=", precise_iostat_setup);
+
 /*
  * For the allocated request tables
  */
@@ -1700,8 +1714,13 @@ static void part_round_stats_single(struct request_queue *q, int cpu,
 				    struct hd_struct *part, unsigned long now,
 				    unsigned int inflight)
 {
-	if (inflight)
+	if (inflight) {
+		if (precise_iostat) {
+			__part_stat_add(cpu, part, time_in_queue,
+					inflight * (now - part->stamp));
+		}
 		__part_stat_add(cpu, part, io_ticks, (now - part->stamp));
+	}
 	part->stamp = now;
 }
 
@@ -2771,10 +2790,15 @@ void blk_account_io_done(struct request *req, u64 now)
 		cpu = part_stat_lock();
 		part = req->part;
 
-		update_io_ticks(cpu, part, jiffies);
+		if (!precise_iostat) {
+			update_io_ticks(cpu, part, jiffies);
+			part_stat_add(cpu, part, time_in_queue,
+				nsecs_to_jiffies64(now - req->start_time_ns));
+		} else {
+			part_round_stats(req->q, cpu, part);
+		}
 		part_stat_inc(cpu, part, ios[sgrp]);
 		part_stat_add(cpu, part, nsecs[sgrp], now - req->start_time_ns);
-		part_stat_add(cpu, part, time_in_queue, nsecs_to_jiffies64(now - req->start_time_ns));
 		part_dec_in_flight(req->q, part, rq_data_dir(req));
 
 		hd_struct_put(part);

block/blk-merge.c

@@ -669,6 +669,8 @@ static void blk_account_io_merge(struct request *req)
 		cpu = part_stat_lock();
 		part = req->part;
 
+		if (precise_iostat)
+			part_round_stats(req->q, cpu, part);
 		part_dec_in_flight(req->q, part, rq_data_dir(req));
 
 		hd_struct_put(part);

block/genhd.c

@@ -1352,6 +1352,7 @@ static int diskstats_show(struct seq_file *seqf, void *v)
 	struct hd_struct *hd;
 	char buf[BDEVNAME_SIZE];
 	unsigned int inflight[2];
+	int cpu;
 
 	/*
 	if (&disk_to_dev(gp)->kobj.entry == block_class.devices.next)
@@ -1363,6 +1364,12 @@ static int diskstats_show(struct seq_file *seqf, void *v)
 
 	disk_part_iter_init(&piter, gp, DISK_PITER_INCL_EMPTY_PART0);
 	while ((hd = disk_part_iter_next(&piter))) {
+		if (precise_iostat) {
+			cpu = part_stat_lock();
+			part_round_stats(gp->queue, cpu, hd);
+			part_stat_unlock();
+		}
+
 		part_in_flight(gp->queue, hd, inflight);
 		seq_printf(seqf, "%4d %7d %s "
 			   "%lu %lu %lu %u "

block/partition-generic.c

@@ -18,6 +18,7 @@
 #include <linux/ctype.h>
 #include <linux/genhd.h>
 #include <linux/blktrace_api.h>
+#include <linux/blkdev.h>
 
 #include "partitions/check.h"
 
@@ -121,6 +122,13 @@ ssize_t part_stat_show(struct device *dev,
 	struct hd_struct *p = dev_to_part(dev);
 	struct request_queue *q = part_to_disk(p)->queue;
 	unsigned int inflight[2];
+	int cpu;
+
+	if (precise_iostat) {
+		cpu = part_stat_lock();
+		part_round_stats(q, cpu, p);
+		part_stat_unlock();
+	}
 
 	part_in_flight(q, p, inflight);
 	return sprintf(buf,

include/linux/blkdev.h

@@ -28,6 +28,7 @@
 #include <linux/scatterlist.h>
 #include <linux/blkzoned.h>
 
+extern bool precise_iostat;
 struct module;
 struct scsi_ioctl_command;