On Mon, Jan 6, 2020 at 1:42 AM Guoqing Jiang
<guoqing.jiang@xxxxxxxxxxxxxxx> wrote:
>
>
>
> On 1/3/20 11:58 PM, Song Liu wrote:
> > On Mon, Dec 23, 2019 at 1:49 AM <jgq516@xxxxxxxxx> wrote:
> >> From: Guoqing Jiang <guoqing.jiang@xxxxxxxxxxxxxxx>
> >>
> >> So far, IO serialization is used for two scenarios:
> >>
> >> 1. raid1 which enables write-behind mode, and there is rdev in the array
> >> which is multi-queue device and flaged with writemostly.
> >> 2. IO serialization is enabled or disabled by change serialize_policy.
> >>
> >> So introduce rdev_need_serial to check the first scenario. And for 1, IO
> >> serialization is enabled automatically while 2 is controlled manually.
> >>
> >> And it is possible that both scenarios are true, so for create serial pool,
> >> rdev/rdevs_init_serial should be separate from check if the pool existed or
> >> not. Then for destroy pool, we need to check if the pool is needed by other
> >> rdevs due to the first scenario.
> >>
> >> Signed-off-by: Guoqing Jiang <guoqing.jiang@xxxxxxxxxxxxxxx>
> >> ---
> >> drivers/md/md.c | 71 +++++++++++++++++++++++++++++--------------------
> >> 1 file changed, 42 insertions(+), 29 deletions(-)
> >>
> >> diff --git a/drivers/md/md.c b/drivers/md/md.c
> >> index 796cf70e1c9f..788559f42d43 100644
> >> --- a/drivers/md/md.c
> >> +++ b/drivers/md/md.c
> >> @@ -147,28 +147,40 @@ static void rdevs_init_serial(struct mddev *mddev)
> >> }
> >>
> >> /*
> >> - * Create serial_info_pool for raid1 under conditions:
> >> - * 1. rdev is the first multi-queue device flaged with writemostly,
> >> - * also write-behind mode is enabled.
> >> - * 2. rdev is NULL, means want to enable serialization for all rdevs.
> >> + * rdev needs to enable serial stuffs if it meets the conditions:
> >> + * 1. it is multi-queue device flaged with writemostly.
> >> + * 2. the write-behind mode is enabled.
> >> + */
> >> +static int rdev_need_serial(struct md_rdev *rdev)
> >> +{
> >> + return (rdev && rdev->mddev->bitmap_info.max_write_behind > 0 &&
> > I guess there is not need to check rdev here?
>
> No, the rdev could be passed from mddev_create_serial_pool, and
> caller could set it to NULL, pls see patch 0003.
I see. Thanks for the clarification.
Song
