The impact is not so significant for raid0 compared to raid5, however
it's still more appropriate to issue IOs evenly to underlying disks.

Signed-off-by: Yu Kuai <yukuai@fnnas.com>
---
 drivers/md/raid0.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/md/raid0.c b/drivers/md/raid0.c
index 47aee1b1d4d1..332f413bcf51 100644
--- a/drivers/md/raid0.c
+++ b/drivers/md/raid0.c
@@ -388,6 +388,8 @@ static int raid0_set_limits(struct mddev *mddev)
 	err = mddev_stack_rdev_limits(mddev, &lim, MDDEV_STACK_INTEGRITY);
 	if (err)
 		return err;
+
+	md_config_align_limits(mddev, &lim);
 	return queue_limits_set(mddev->gendisk->queue, &lim);
 }
 
-- 
2.51.0

Dear Kuai,


Thank you for your patches.

Am 21.11.25 um 06:14 schrieb Yu Kuai:
> The impact is not so significant for raid0 compared to raid5, however
> it's still more appropriate to issue IOs evenly to underlying disks.

Could you please elaborate, what the problem is, and also share your 
benchmark results and test setups?

> Signed-off-by: Yu Kuai <yukuai@fnnas.com>
> ---
>   drivers/md/raid0.c | 2 ++
>   1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/md/raid0.c b/drivers/md/raid0.c
> index 47aee1b1d4d1..332f413bcf51 100644
> --- a/drivers/md/raid0.c
> +++ b/drivers/md/raid0.c
> @@ -388,6 +388,8 @@ static int raid0_set_limits(struct mddev *mddev)
>   	err = mddev_stack_rdev_limits(mddev, &lim, MDDEV_STACK_INTEGRITY);
>   	if (err)
>   		return err;
> +
> +	md_config_align_limits(mddev, &lim);
>   	return queue_limits_set(mddev->gendisk->queue, &lim);
>   }
>   


Kind regards,

Paul