lkml.org 
[lkml]   [2023]   [May]   [22]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[PATCH v4 3/3] md/raid10: optimize check_decay_read_errors()
Date
From: Li Nan <linan122@huawei.com>

check_decay_read_errors() is used to handle rdev->read_errors. But
read_errors is inc and read after check_decay_read_errors() is invoked
in fix_read_error().

Put all operations of read_errors into check_decay_read_errors() and
clean up unnecessary atomic_read of read_errors.

Suggested-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Li Nan <linan122@huawei.com>
---
drivers/md/raid10.c | 41 ++++++++++++++++++++++++-----------------
1 file changed, 24 insertions(+), 17 deletions(-)

diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index 4fcfcb350d2b..d31eed17f186 100644
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -2655,23 +2655,24 @@ static void recovery_request_write(struct mddev *mddev, struct r10bio *r10_bio)
}

/*
- * Used by fix_read_error() to decay the per rdev read_errors.
+ * Used by fix_read_error() to decay the per rdev read_errors and check if
+ * read_error > max_read_errors.
* We halve the read error count for every hour that has elapsed
* since the last recorded read error.
*
*/
-static void check_decay_read_errors(struct mddev *mddev, struct md_rdev *rdev)
+static bool check_decay_read_errors(struct mddev *mddev, struct md_rdev *rdev)
{
- long cur_time_mon;
+ time64_t cur_time_mon = ktime_get_seconds();
unsigned long hours_since_last;
- unsigned int read_errors = atomic_read(&rdev->read_errors);
-
- cur_time_mon = ktime_get_seconds();
+ unsigned int read_errors;
+ unsigned int max_read_errors =
+ atomic_read(&mddev->max_corr_read_errors);

if (rdev->last_read_error == 0) {
/* first time we've seen a read error */
rdev->last_read_error = cur_time_mon;
- return;
+ goto increase;
}

hours_since_last = (long)(cur_time_mon -
@@ -2684,10 +2685,25 @@ static void check_decay_read_errors(struct mddev *mddev, struct md_rdev *rdev)
* just set read errors to 0. We do this to avoid
* overflowing the shift of read_errors by hours_since_last.
*/
+ read_errors = atomic_read(&rdev->read_errors);
if (hours_since_last >= 8 * sizeof(read_errors))
atomic_set(&rdev->read_errors, 0);
else
atomic_set(&rdev->read_errors, read_errors >> hours_since_last);
+
+increase:
+ read_errors = atomic_inc_return(&rdev->read_errors);
+ if (read_errors > max_read_errors) {
+ pr_notice("md/raid10:%s: %pg: Raid device exceeded read_error threshold [cur %d:max %d]\n",
+ mdname(mddev), rdev->bdev,
+ read_errors, max_read_errors);
+ pr_notice("md/raid10:%s: %pg: Failing raid device\n",
+ mdname(mddev), rdev->bdev);
+ md_error(mddev, rdev);
+ return false;
+ }
+
+ return true;
}

static int r10_sync_page_io(struct md_rdev *rdev, sector_t sector,
@@ -2727,7 +2743,6 @@ static void fix_read_error(struct r10conf *conf, struct mddev *mddev, struct r10
int sect = 0; /* Offset from r10_bio->sector */
int sectors = r10_bio->sectors;
struct md_rdev *rdev;
- int max_read_errors = atomic_read(&mddev->max_corr_read_errors);
int d = r10_bio->devs[r10_bio->read_slot].devnum;

/* still own a reference to this rdev, so it cannot
@@ -2740,15 +2755,7 @@ static void fix_read_error(struct r10conf *conf, struct mddev *mddev, struct r10
more fix_read_error() attempts */
return;

- check_decay_read_errors(mddev, rdev);
- atomic_inc(&rdev->read_errors);
- if (atomic_read(&rdev->read_errors) > max_read_errors) {
- pr_notice("md/raid10:%s: %pg: Raid device exceeded read_error threshold [cur %d:max %d]\n",
- mdname(mddev), rdev->bdev,
- atomic_read(&rdev->read_errors), max_read_errors);
- pr_notice("md/raid10:%s: %pg: Failing raid device\n",
- mdname(mddev), rdev->bdev);
- md_error(mddev, rdev);
+ if (!check_decay_read_errors(mddev, rdev)) {
r10_bio->devs[r10_bio->read_slot].bio = IO_BLOCKED;
return;
}
--
2.31.1
\
 
 \ /
  Last update: 2023-05-22 09:29    [W:0.074 / U:0.412 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site