All of lore.kernel.org
 help / color / mirror / Atom feed
From: Guoqing Jiang <jgq516@gmail.com>
To: Artur Paszkiewicz <artur.paszkiewicz@intel.com>, song@kernel.org
Cc: linux-raid@vger.kernel.org, Guoqing Jiang <jiangguoqing@kylinos.cn>
Subject: Re: [PATCH 2/5] md: the latest try for improve io stats accounting
Date: Wed, 19 May 2021 16:09:04 +0800	[thread overview]
Message-ID: <c63b1793-e83e-f729-cd97-2f34b43166dd@gmail.com> (raw)
In-Reply-To: <d788bc1f-3fd9-a293-2f2a-6047fdd45625@intel.com>



On 5/19/21 3:26 PM, Artur Paszkiewicz wrote:
> On 19.05.2021 03:30, Guoqing Jiang wrote:
>> Hmm, raid0 allocates split bio from mddev->bio_set, but raid5 is
>> different, it splits bio from r5conf->bio_split. So either let raid5 also
>> splits bio from mddev->bio_set, or add an additional checking for
>> raid5. Thoughts?
> It looks like raid5 has a different bio set for that because it uses
> mddev->bio_set for something else - allocating a bio for rdev. So I
> think it can be changed to split from mddev->bio_set and have a private
> bio set for the rdev bio allocation.


Instead of add a flag, I tried this way:

diff --git a/drivers/md/raid0.c b/drivers/md/raid0.c
index e5d7411cba9b..d309b639b5d9 100644
--- a/drivers/md/raid0.c
+++ b/drivers/md/raid0.c
@@ -748,6 +748,19 @@ static void raid0_quiesce(struct mddev *mddev, int 
quiesce)
  {
  }

+/*
+ * Don't account the bio if it was split from mddev->bio_set.
+ */
+static bool raid0_accounting_bio(struct mddev *mddev, struct bio *bio)
+{
+       bool ret = true;
+
+       if (bio->bi_pool == &mddev->bio_set)
+               ret = false;
+
+       return ret;
+}
+
  static struct md_personality raid0_personality=
  {
         .name           = "raid0",
@@ -760,6 +773,7 @@ static struct md_personality raid0_personality=
         .size           = raid0_size,
         .takeover       = raid0_takeover,
         .quiesce        = raid0_quiesce,
+       .accounting_bio = raid0_accounting_bio,
  };

diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 841e1c1aa5e6..070ccf55c534 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -8596,6 +8596,20 @@ static void *raid6_takeover(struct mddev *mddev)
         return setup_conf(mddev);
  }

+/*
+ * Don't account the bio if it was split from r5conf->bio_split.
+ */
+static bool raid5_accounting_bio(struct mddev *mddev, struct bio *bio)
+{
+       bool ret = true;
+       struct r5conf *conf = mddev->private;
+
+       if (bio->bi_pool == &conf->bio_split)
+               ret = false;
+
+       return ret;
+}
+
  static int raid5_change_consistency_policy(struct mddev *mddev, const 
char *buf)
  {
         struct r5conf *conf;
@@ -8687,6 +8701,7 @@ static struct md_personality raid6_personality =
         .finish_reshape = raid5_finish_reshape,
         .quiesce        = raid5_quiesce,
         .takeover       = raid6_takeover,
+       .accounting_bio = raid5_accounting_bio,

Thanks,
Guoqing

  parent reply	other threads:[~2021-05-19  8:09 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-05-18  5:32 [PATCH 0/5] md: io stats accounting Guoqing Jiang
2021-05-18  5:32 ` [PATCH 1/5] md: revert " Guoqing Jiang
2021-05-18  5:32 ` [PATCH 2/5] md: the latest try for improve " Guoqing Jiang
2021-05-18 10:12   ` Artur Paszkiewicz
2021-05-19  1:30     ` Guoqing Jiang
2021-05-19  7:26       ` Artur Paszkiewicz
2021-05-19  7:46         ` Christoph Hellwig
2021-05-19  8:09         ` Guoqing Jiang [this message]
2021-05-20  0:45           ` Song Liu
2021-05-18 13:35   ` Christoph Hellwig
2021-05-19  1:40     ` Guoqing Jiang
2021-05-18  5:32 ` [PATCH 3/5] md-multipath: enable io accounting Guoqing Jiang
2021-05-18 13:37   ` Christoph Hellwig
2021-05-19  1:46     ` Guoqing Jiang
2021-05-18  5:32 ` [PATCH 4/5] md/raid1: " Guoqing Jiang
2021-05-18 13:40   ` Christoph Hellwig
2021-05-19  1:47     ` Guoqing Jiang
2021-05-18  5:32 ` [PATCH 5/5] md/raid10: " Guoqing Jiang
2021-05-19  0:14 ` [PATCH 0/5] md: io stats accounting Song Liu
2021-05-19  1:50   ` 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=c63b1793-e83e-f729-cd97-2f34b43166dd@gmail.com \
    --to=jgq516@gmail.com \
    --cc=artur.paszkiewicz@intel.com \
    --cc=jiangguoqing@kylinos.cn \
    --cc=linux-raid@vger.kernel.org \
    --cc=song@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.