All of lore.kernel.org
 help / color / mirror / Atom feed
From: Guoqing Jiang <guoqing.jiang@cloud.ionos.com>
To: axboe@kernel.dk
Cc: linux-block@vger.kernel.org,
	Guoqing Jiang <guoqing.jiang@cloud.ionos.com>,
	Florian-Ewald Mueller <florian-ewald.mueller@cloud.ionos.com>
Subject: [PATCH RFC V2 1/4] block: add a statistic table for io latency
Date: Mon, 13 Jul 2020 23:13:18 +0200	[thread overview]
Message-ID: <20200713211321.21123-2-guoqing.jiang@cloud.ionos.com> (raw)
In-Reply-To: <20200713211321.21123-1-guoqing.jiang@cloud.ionos.com>

Usually, we get the status of block device by cat stat file,
but we can only know the total time with that file. And we
would like to know more accurate statistic, such as each
latency range, which helps people to diagnose if there is
issue about the hardware.

Also a new config option is introduced to control if people
want to know the additional statistics or not, and we use
the option for io sector in next patch.

Signed-off-by: Florian-Ewald Mueller <florian-ewald.mueller@cloud.ionos.com>
Signed-off-by: Guoqing Jiang <guoqing.jiang@cloud.ionos.com>
---
 block/Kconfig             |  8 ++++++++
 block/blk-core.c          | 34 ++++++++++++++++++++++++++++++++++
 block/genhd.c             | 26 ++++++++++++++++++++++++++
 include/linux/part_stat.h |  7 +++++++
 4 files changed, 75 insertions(+)

diff --git a/block/Kconfig b/block/Kconfig
index bbad5e8bbffe..360f63111e2d 100644
--- a/block/Kconfig
+++ b/block/Kconfig
@@ -176,6 +176,14 @@ config BLK_DEBUG_FS
 	Unless you are building a kernel for a tiny system, you should
 	say Y here.
 
+config BLK_ADDITIONAL_DISKSTAT
+	bool "Block layer additional diskstat"
+	default n
+	help
+	Enabling this option adds io latency statistics for each block device.
+
+	If unsure, say N.
+
 config BLK_DEBUG_FS_ZONED
        bool
        default BLK_DEBUG_FS && BLK_DEV_ZONED
diff --git a/block/blk-core.c b/block/blk-core.c
index d9d632639bd1..036eb04782de 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1411,6 +1411,34 @@ static void update_io_ticks(struct hd_struct *part, unsigned long now, bool end)
 	}
 }
 
+#ifdef CONFIG_BLK_ADDITIONAL_DISKSTAT
+/*
+ * Either account additional stat for request if req is not NULL or account for bio.
+ */
+static void blk_additional_latency(struct hd_struct *part, const int sgrp,
+				   struct request *req, unsigned long start_jiffies)
+{
+	unsigned int idx;
+	unsigned long duration, now = READ_ONCE(jiffies);
+
+	if (req)
+		duration = jiffies_to_nsecs(now) - req->start_time_ns;
+	else
+		duration = jiffies_to_nsecs(now - start_jiffies);
+
+	duration /= NSEC_PER_MSEC;
+	duration /= HZ_TO_MSEC_NUM;
+	if (likely(duration > 0)) {
+		idx = ilog2(duration);
+		if (idx > ADD_STAT_NUM - 1)
+			idx = ADD_STAT_NUM - 1;
+	} else
+		idx = 0;
+	part_stat_inc(part, latency_table[idx][sgrp]);
+
+}
+#endif
+
 static void blk_account_io_completion(struct request *req, unsigned int bytes)
 {
 	if (req->part && blk_do_io_stat(req)) {
@@ -1440,6 +1468,9 @@ void blk_account_io_done(struct request *req, u64 now)
 		part = req->part;
 
 		update_io_ticks(part, jiffies, true);
+#ifdef CONFIG_BLK_ADDITIONAL_DISKSTAT
+		blk_additional_latency(part, sgrp, req, 0);
+#endif
 		part_stat_inc(part, ios[sgrp]);
 		part_stat_add(part, nsecs[sgrp], now - req->start_time_ns);
 		part_stat_unlock();
@@ -1488,6 +1519,9 @@ void disk_end_io_acct(struct gendisk *disk, unsigned int op,
 
 	part_stat_lock();
 	update_io_ticks(part, now, true);
+#ifdef CONFIG_BLK_ADDITIONAL_DISKSTAT
+	blk_additional_latency(part, sgrp, NULL, start_time);
+#endif
 	part_stat_add(part, nsecs[sgrp], jiffies_to_nsecs(duration));
 	part_stat_local_dec(part, in_flight[op_is_write(op)]);
 	part_stat_unlock();
diff --git a/block/genhd.c b/block/genhd.c
index c42a49f2f537..f5d2f110fb34 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -1420,6 +1420,29 @@ static struct device_attribute dev_attr_fail_timeout =
 	__ATTR(io-timeout-fail, 0644, part_timeout_show, part_timeout_store);
 #endif
 
+#ifdef CONFIG_BLK_ADDITIONAL_DISKSTAT
+static ssize_t io_latency_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	struct hd_struct *p = dev_to_part(dev);
+	size_t count = 0;
+	int i, sgrp;
+
+	for (i = 0; i < ADD_STAT_NUM; i++) {
+		count += scnprintf(buf + count, PAGE_SIZE - count, "%5d ms: ",
+				   (1 << i) * HZ_TO_MSEC_NUM);
+		for (sgrp = 0; sgrp < NR_STAT_GROUPS; sgrp++)
+			count += scnprintf(buf + count, PAGE_SIZE - count, "%lu ",
+					   part_stat_read(p, latency_table[i][sgrp]));
+		count += scnprintf(buf + count, PAGE_SIZE - count, "\n");
+	}
+
+	return count;
+}
+
+static struct device_attribute dev_attr_io_latency =
+	__ATTR(io_latency, 0444, io_latency_show, NULL);
+#endif
+
 static struct attribute *disk_attrs[] = {
 	&dev_attr_range.attr,
 	&dev_attr_ext_range.attr,
@@ -1438,6 +1461,9 @@ static struct attribute *disk_attrs[] = {
 #endif
 #ifdef CONFIG_FAIL_IO_TIMEOUT
 	&dev_attr_fail_timeout.attr,
+#endif
+#ifdef CONFIG_BLK_ADDITIONAL_DISKSTAT
+	&dev_attr_io_latency.attr,
 #endif
 	NULL
 };
diff --git a/include/linux/part_stat.h b/include/linux/part_stat.h
index 24125778ef3e..fe3def8c69d7 100644
--- a/include/linux/part_stat.h
+++ b/include/linux/part_stat.h
@@ -9,6 +9,13 @@ struct disk_stats {
 	unsigned long sectors[NR_STAT_GROUPS];
 	unsigned long ios[NR_STAT_GROUPS];
 	unsigned long merges[NR_STAT_GROUPS];
+#ifdef CONFIG_BLK_ADDITIONAL_DISKSTAT
+/*
+ * We measure latency (ms) for 1, 2, ..., 1024 and >=1024.
+ */
+#define ADD_STAT_NUM	12
+	unsigned long latency_table[ADD_STAT_NUM][NR_STAT_GROUPS];
+#endif
 	unsigned long io_ticks;
 	local_t in_flight[2];
 };
-- 
2.17.1


  reply	other threads:[~2020-07-13 21:13 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-07-13 21:13 [PATCH RFC V2 0/4] block: add two statistic tables Guoqing Jiang
2020-07-13 21:13 ` Guoqing Jiang [this message]
2020-07-14  7:43   ` [PATCH RFC V2 1/4] block: add a statistic table for io latency Johannes Thumshirn
2020-07-14  8:25     ` Guoqing Jiang
2020-08-11  7:22   ` Danil Kipnis
2020-07-13 21:13 ` [PATCH RFC V2 2/4] block: add a statistic table for io sector Guoqing Jiang
2020-08-11 15:04   ` Aleksei Marov
2020-08-11 15:48     ` Guoqing Jiang
2020-07-13 21:13 ` [PATCH RFC V2 3/4] block: add io_extra_stats node Guoqing Jiang
2020-07-13 21:13 ` [PATCH RFC V2 4/4] block: call blk_additional_{latency,sector} only when io_extra_stats is true Guoqing Jiang

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200713211321.21123-2-guoqing.jiang@cloud.ionos.com \
    --to=guoqing.jiang@cloud.ionos.com \
    --cc=axboe@kernel.dk \
    --cc=florian-ewald.mueller@cloud.ionos.com \
    --cc=linux-block@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.