block/diskstats: more accurate approximation of io_ticks for slow disks

Currently io_ticks is approximated by adding one at each start and end of requests if jiffies counter has changed. This works perfectly for requests shorter than a jiffy or if one of requests starts/ends at each jiffy. If disk executes just one request at a time and they are longer than two jiffies then only first and last jiffies will be accounted. Fix is simple: at the end of request add up into io_ticks jiffies passed since last update rather than just one jiffy. Example: common HDD executes random read 4k requests around 12ms. fio --name=test --filename=/dev/sdb --rw=randread --direct=1 --runtime=30 & iostat -x 10 sdb Note changes of iostat's "%util" 8,43% -> 99,99% before/after patch: Before: Device: rrqm/s wrqm/s r/s w/s rkB/s wkB/s avgrq-sz avgqu-sz await r_await w_await svctm %util sdb 0,00 0,00 82,60 0,00 330,40 0,00 8,00 0,96 12,09 12,09 0,00 1,02 8,43 After: Device: rrqm/s wrqm/s r/s w/s rkB/s wkB/s avgrq-sz avgqu-sz await r_await w_await svctm %util sdb 0,00 0,00 82,50 0,00 330,00 0,00 8,00 1,00 12,10 12,10 0,00 12,12 99,99 Now io_ticks does not loose time between start and end of requests, but for queue-depth > 1 some I/O time between adjacent starts might be lost. For load estimation "%util" is not as useful as average queue length, but it clearly shows how often disk queue is completely empty.

Konstantin Khlebnikov committed 6 years ago

Commit 2b8bd421 parent 387048b4 files · +11 −8Browse files

Unified · Split

4 files

Documentation/admin-guide/iostats.rst +4 −1
block/bio.c +4 −4
block/blk-core.c +2 −2
include/linux/genhd.h +1 −1

Documentation/admin-guide/iostats.rst+4−1

     Since 5.0 this field counts jiffies when at least one request was
     started or completed. If request runs more than 2 jiffies then some
-    I/O time will not be accounted unless there are other requests.
+    I/O time might be not accounted in case of concurrent requests.
 Field 11 -- weighted # of milliseconds spent doing I/Os (unsigned int)
     This field is incremented at each I/O start, I/O completion, I/O
 summed to) and the result given to the user.  There is no convenient
 user interface for accessing the per-CPU counters themselves.
+Since 4.19 request times are measured with nanoseconds precision and
+truncated to milliseconds before showing in this interface.
 Disks vs Partitions
 -------------------

block/bio.c+4−4

@@ -1768,14 +1768,14 @@ defer: · 1767 unmodified lines
1768	schedule_work(&bio_dirty_work);	1768	schedule_work(&bio_dirty_work);
1769	}	1769	}
1770		1770
1771	void update_io_ticks(struct hd_struct *part, unsigned long now)	1771	void update_io_ticks(struct hd_struct *part, unsigned long now, bool end)
1772	{	1772	{
1773	unsigned long stamp;	1773	unsigned long stamp;
1774	again:	1774	again:
1775	stamp = READ_ONCE(part->stamp);	1775	stamp = READ_ONCE(part->stamp);
1776	if (unlikely(stamp != now)) {	1776	if (unlikely(stamp != now)) {
1777	if (likely(cmpxchg(&part->stamp, stamp, now) == stamp)) {	1777	if (likely(cmpxchg(&part->stamp, stamp, now) == stamp)) {
1778	__part_stat_add(part, io_ticks, 1);	1778	__part_stat_add(part, io_ticks, end ? now - stamp : 1);
1779	}	1779	}
1780	}	1780	}
1781	if (part->partno) {	1781	if (part->partno) {
@@ -1791,7 +1791,7 @@ void generic_start_io_acct(struct request_queue *q, int op, · 9 unmodified lines
1791		1791
1792	part_stat_lock();	1792	part_stat_lock();
1793		1793
1794	update_io_ticks(part, jiffies);	1794	update_io_ticks(part, jiffies, false);
1795	part_stat_inc(part, ios[sgrp]);	1795	part_stat_inc(part, ios[sgrp]);
1796	part_stat_add(part, sectors[sgrp], sectors);	1796	part_stat_add(part, sectors[sgrp], sectors);
1797	part_inc_in_flight(q, part, op_is_write(op));	1797	part_inc_in_flight(q, part, op_is_write(op));
@@ -1809,7 +1809,7 @@ void generic_end_io_acct(struct request_queue *q, int req_op, · 11 unmodified lines
1809		1809
1810	part_stat_lock();	1810	part_stat_lock();
1811		1811
1812	update_io_ticks(part, now);	1812	update_io_ticks(part, now, true);
1813	part_stat_add(part, nsecs[sgrp], jiffies_to_nsecs(duration));	1813	part_stat_add(part, nsecs[sgrp], jiffies_to_nsecs(duration));
1814	part_stat_add(part, time_in_queue, duration);	1814	part_stat_add(part, time_in_queue, duration);
1815	part_dec_in_flight(q, part, op_is_write(req_op));	1815	part_dec_in_flight(q, part, op_is_write(req_op));

block/blk-core.c+2−2

@@ -1337,7 +1337,7 @@ void blk_account_io_done(struct request *req, u64 now) · 1336 unmodified lines
1337	part_stat_lock();	1337	part_stat_lock();
1338	part = req->part;	1338	part = req->part;
1339		1339
1340	update_io_ticks(part, jiffies);	1340	update_io_ticks(part, jiffies, true);
1341	part_stat_inc(part, ios[sgrp]);	1341	part_stat_inc(part, ios[sgrp]);
1342	part_stat_add(part, nsecs[sgrp], now - req->start_time_ns);	1342	part_stat_add(part, nsecs[sgrp], now - req->start_time_ns);
1343	part_stat_add(part, time_in_queue, nsecs_to_jiffies64(now - req->start_time_ns));	1343	part_stat_add(part, time_in_queue, nsecs_to_jiffies64(now - req->start_time_ns));
@@ -1379,7 +1379,7 @@ void blk_account_io_start(struct request *rq, bool new_io) · 35 unmodified lines
1379	rq->part = part;	1379	rq->part = part;
1380	}	1380	}
1381		1381
1382	update_io_ticks(part, jiffies);	1382	update_io_ticks(part, jiffies, false);
1383		1383
1384	part_stat_unlock();	1384	part_stat_unlock();
1385	}	1385	}

include/linux/genhd.h+1−1

@@ -422,7 +422,7 @@ void part_dec_in_flight(struct request_queue q, struct hd_struct part, · 421 unmodified lines
422	void part_inc_in_flight(struct request_queue q, struct hd_struct part,	422	void part_inc_in_flight(struct request_queue q, struct hd_struct part,
423	int rw);	423	int rw);
424		424
425	void update_io_ticks(struct hd_struct *part, unsigned long now);	425	void update_io_ticks(struct hd_struct *part, unsigned long now, bool end);
426		426
427	/* block/genhd.c */	427	/* block/genhd.c */
428	extern void device_add_disk(struct device parent, struct gendisk disk,	428	extern void device_add_disk(struct device parent, struct gendisk disk,