2022-04-22 01:51:14

by Logan Gunthorpe

Subject: [PATCH v2 06/12] md/raid5: Drop the do_prepare flag in raid5_make_request()

prepare_to_wait() can be reasonably called after schedule instead of
setting a flag and preparing in the next loop iteration.

This means that prepare_to_wait() will be called before
read_seqcount_begin(), but there shouldn't be any reason that
the order matters here. On the first iteration of the loop
prepare_to_wait() is already called first.

Signed-off-by: Logan Gunthorpe <[email protected]>
---
drivers/md/raid5.c | 8 ++------
1 file changed, 2 insertions(+), 6 deletions(-)

diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 5a7334ba0997..b9f618356446 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -5915,7 +5915,6 @@ static bool raid5_make_request(struct mddev *mddev, struct bio * bi)
const int rw = bio_data_dir(bi);
enum stripe_result res;
DEFINE_WAIT(w);
- bool do_prepare;

if (unlikely(bi->bi_opf & REQ_PREFLUSH)) {
int ret = log_handle_flush_request(conf, bi);
@@ -5973,12 +5972,8 @@ static bool raid5_make_request(struct mddev *mddev, struct bio * bi)
for (; logical_sector < last_sector; logical_sector += RAID5_STRIPE_SECTORS(conf)) {
int seq;

- do_prepare = false;
retry:
seq = read_seqcount_begin(&conf->gen_lock);
- if (do_prepare)
- prepare_to_wait(&conf->wait_for_overlap, &w,
- TASK_UNINTERRUPTIBLE);

res = make_stripe_request(mddev, conf, &ctx, logical_sector,
bi, seq);
@@ -5988,7 +5983,8 @@ static bool raid5_make_request(struct mddev *mddev, struct bio * bi)
goto retry;
} else if (res == STRIPE_SCHEDULE_AND_RETRY) {
schedule();
- do_prepare = true;
+ prepare_to_wait(&conf->wait_for_overlap, &w,
+ TASK_UNINTERRUPTIBLE);
goto retry;
}
}
--
2.30.2

2022-04-27 10:03:43

by Guoqing Jiang

Subject: Re: [PATCH v2 06/12] md/raid5: Drop the do_prepare flag in raid5_make_request()

On 4/21/22 3:54 AM, Logan Gunthorpe wrote:
> prepare_to_wait() can be reasonably called after schedule instead of
> setting a flag and preparing in the next loop iteration.
>
> This means that prepare_to_wait() will be called before
> read_seqcount_begin(), but there shouldn't be any reason that
> the order matters here. On the first iteration of the loop
> prepare_to_wait() is already called first.
>
> Signed-off-by: Logan Gunthorpe <[email protected]>
> ---
> drivers/md/raid5.c | 8 ++------
> 1 file changed, 2 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
> index 5a7334ba0997..b9f618356446 100644
> --- a/drivers/md/raid5.c
> +++ b/drivers/md/raid5.c
> @@ -5915,7 +5915,6 @@ static bool raid5_make_request(struct mddev *mddev, struct bio * bi)
> const int rw = bio_data_dir(bi);
> enum stripe_result res;
> DEFINE_WAIT(w);
> - bool do_prepare;
>
> if (unlikely(bi->bi_opf & REQ_PREFLUSH)) {
> int ret = log_handle_flush_request(conf, bi);
> @@ -5973,12 +5972,8 @@ static bool raid5_make_request(struct mddev *mddev, struct bio * bi)
> for (; logical_sector < last_sector; logical_sector += RAID5_STRIPE_SECTORS(conf)) {
> int seq;
>
> - do_prepare = false;
> retry:
> seq = read_seqcount_begin(&conf->gen_lock);
> - if (do_prepare)
> - prepare_to_wait(&conf->wait_for_overlap, &w,
> - TASK_UNINTERRUPTIBLE);
>
> res = make_stripe_request(mddev, conf, &ctx, logical_sector,
> bi, seq);
> @@ -5988,7 +5983,8 @@ static bool raid5_make_request(struct mddev *mddev, struct bio * bi)
> goto retry;
> } else if (res == STRIPE_SCHEDULE_AND_RETRY) {
> schedule();
> - do_prepare = true;
> + prepare_to_wait(&conf->wait_for_overlap, &w,
> + TASK_UNINTERRUPTIBLE);
> goto retry;
> }
> }

Reviewed-by: Guoqing Jiang <[email protected]>

Thanks,
Guoqing