From: Niklas Cassel <Niklas.Cassel@wdc.com>
To: Shinichiro Kawasaki <shinichiro.kawasaki@wdc.com>
Cc: "fio@vger.kernel.org" <fio@vger.kernel.org>,
Jens Axboe <axboe@kernel.dk>, Vincent Fu <vincentfu@gmail.com>,
Damien Le Moal <damien.lemoal@opensource.wdc.com>,
Dmitry Fomichev <Dmitry.Fomichev@wdc.com>
Subject: Re: [PATCH v2 7/8] zbd: initialize valid data bytes accounting at file set up
Date: Tue, 7 Feb 2023 14:06:32 +0000 [thread overview]
Message-ID: <Y+Ja5jyZq/snxRpI@x1-carbon> (raw)
In-Reply-To: <20230207063739.1661191-8-shinichiro.kawasaki@wdc.com>
On Tue, Feb 07, 2023 at 03:37:38PM +0900, Shin'ichiro Kawasaki wrote:
> The valid data bytes accounting field is initialized at file reset,
> after each job started. Each job locks zones to check write pointer
> positions of its write target zones. This can cause zone lock contention
> with write by other jobs.
>
> To avoid the zone lock contention, move the initialization from file
> reset to file set up before job start. It allows to access the write
> pointers and the accounting field without locks. Remove the lock and
> unlock codes which are no longer required. Ensure the locks are not
> required by checking run-state in the struct thread_data.
>
> Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
> ---
> zbd.c | 93 ++++++++++++++++++++++++++++-------------------------------
> 1 file changed, 44 insertions(+), 49 deletions(-)
>
> diff --git a/zbd.c b/zbd.c
> index ca6816b9..a464d6df 100644
> --- a/zbd.c
> +++ b/zbd.c
> @@ -1074,6 +1074,44 @@ void zbd_recalc_options_with_zone_granularity(struct thread_data *td)
> }
> }
>
> +static uint64_t zbd_set_vdb(struct thread_data *td, const struct fio_file *f)
> +{
> + struct fio_zone_info *zb, *ze, *z;
> + uint64_t wp_vdb = 0;
> + struct zoned_block_device_info *zbdi = f->zbd_info;
> +
> + assert(td->runstate < TD_RUNNING);
> + assert(zbdi);
> +
> + if (!accounting_vdb(td, f))
> + return 0;
zbd_set_vdb() is also called from zbd_file_reset().
The code from here:
> + if (zbdi->wp_write_min_zone != zbdi->wp_write_max_zone) {
> + if (zbdi->wp_write_min_zone != f->min_zone ||
> + zbdi->wp_write_max_zone != f->max_zone) {
> + td_verror(td, EINVAL,
> + "multi-jobs with different write ranges are "
> + "not supported with zone_reset_threshold");
> + log_err("multi-jobs with different write ranges are "
> + "not supported with zone_reset_threshold\n");
> + }
> + return 0;
> + }
> +
> + zbdi->wp_write_min_zone = f->min_zone;
> + zbdi->wp_write_max_zone = f->max_zone;
Up until here. (Basically the code added in patch 6/8.)
Should probably only be called by zbd_setup_files().
Or do we need to perform the verification for every single file reset?
> +
> + zb = zbd_get_zone(f, f->min_zone);
> + ze = zbd_get_zone(f, f->max_zone);
> + for (z = zb; z < ze; z++)
> + if (z->has_wp)
> + wp_vdb += z->wp - z->start;
> +
> + zbdi->wp_valid_data_bytes = wp_vdb;
> +
> + return wp_vdb;
> +}
> +
> int zbd_setup_files(struct thread_data *td)
> {
> struct fio_file *f;
> @@ -1099,6 +1137,7 @@ int zbd_setup_files(struct thread_data *td)
> struct zoned_block_device_info *zbd = f->zbd_info;
> struct fio_zone_info *z;
> int zi;
> + uint64_t vdb;
>
> assert(zbd);
>
> @@ -1106,6 +1145,11 @@ int zbd_setup_files(struct thread_data *td)
> f->max_zone =
> zbd_offset_to_zone_idx(f, f->file_offset + f->io_size);
>
> + vdb = zbd_set_vdb(td, f);
> +
> + dprint(FD_ZBD, "%s(%s): valid data bytes = %" PRIu64 "\n",
> + __func__, f->file_name, vdb);
> +
> /*
> * When all zones in the I/O range are conventional, io_size
> * can be smaller than zone size, making min_zone the same
> @@ -1197,54 +1241,9 @@ static bool zbd_dec_and_reset_write_cnt(const struct thread_data *td,
> return write_cnt == 0;
> }
>
> -static uint64_t zbd_set_vdb(struct thread_data *td, const struct fio_file *f)
> -{
> - struct fio_zone_info *zb, *ze, *z;
> - uint64_t wp_vdb = 0;
> - struct zoned_block_device_info *zbdi = f->zbd_info;
> -
> - if (!accounting_vdb(td, f))
> - return 0;
> -
> - if (zbdi->wp_write_min_zone != zbdi->wp_write_max_zone) {
> - if (zbdi->wp_write_min_zone != f->min_zone ||
> - zbdi->wp_write_max_zone != f->max_zone) {
> - td_verror(td, EINVAL,
> - "multi-jobs with different write ranges are "
> - "not supported with zone_reset_threshold");
> - log_err("multi-jobs with different write ranges are "
> - "not supported with zone_reset_threshold\n");
> - }
> - return 0;
> - }
> -
> - zbdi->wp_write_min_zone = f->min_zone;
> - zbdi->wp_write_max_zone = f->max_zone;
> -
> - zb = zbd_get_zone(f, f->min_zone);
> - ze = zbd_get_zone(f, f->max_zone);
> - for (z = zb; z < ze; z++) {
> - if (z->has_wp) {
> - zone_lock(td, f, z);
> - wp_vdb += z->wp - z->start;
> - }
> - }
> -
> - pthread_mutex_lock(&zbdi->mutex);
> - zbdi->wp_valid_data_bytes = wp_vdb;
> - pthread_mutex_unlock(&zbdi->mutex);
> -
> - for (z = zb; z < ze; z++)
> - if (z->has_wp)
> - zone_unlock(z);
> -
> - return wp_vdb;
> -}
> -
> void zbd_file_reset(struct thread_data *td, struct fio_file *f)
> {
> struct fio_zone_info *zb, *ze;
> - uint64_t vdb;
> bool verify_data_left = false;
>
> if (!f->zbd_info || !td_write(td))
> @@ -1252,10 +1251,6 @@ void zbd_file_reset(struct thread_data *td, struct fio_file *f)
>
> zb = zbd_get_zone(f, f->min_zone);
> ze = zbd_get_zone(f, f->max_zone);
> - vdb = zbd_set_vdb(td, f);
> -
> - dprint(FD_ZBD, "%s(%s): valid data bytes = %" PRIu64 "\n",
> - __func__, f->file_name, vdb);
>
> /*
> * If data verification is enabled reset the affected zones before
> --
> 2.38.1
>
next prev parent reply other threads:[~2023-02-07 14:06 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-02-07 6:37 [PATCH v2 0/8] zbd: fix 'sectors with data' and zone_reset_threshold accounting issues Shin'ichiro Kawasaki
2023-02-07 6:37 ` [PATCH v2 1/8] zbd: refer file->last_start[] instead of sectors with data accounting Shin'ichiro Kawasaki
2023-02-07 14:05 ` Niklas Cassel
2023-02-07 6:37 ` [PATCH v2 2/8] zbd: remove CHECK_SWD feature Shin'ichiro Kawasaki
2023-02-07 14:05 ` Niklas Cassel
2023-02-07 6:37 ` [PATCH v2 3/8] zbd: rename the accounting 'sectors with data' to 'valid data bytes' Shin'ichiro Kawasaki
2023-02-07 14:06 ` Niklas Cassel
2023-02-07 6:37 ` [PATCH v2 4/8] doc: fix unit of zone_reset_threshold and relation to other option Shin'ichiro Kawasaki
2023-02-07 14:06 ` Niklas Cassel
2023-02-07 6:37 ` [PATCH v2 5/8] zbd: account valid data bytes only for zone_reset_threshold option Shin'ichiro Kawasaki
2023-02-07 14:06 ` Niklas Cassel
2023-02-07 6:37 ` [PATCH v2 6/8] zbd: check write ranges " Shin'ichiro Kawasaki
2023-02-07 14:06 ` Niklas Cassel
2023-02-08 4:59 ` Shinichiro Kawasaki
2023-02-07 6:37 ` [PATCH v2 7/8] zbd: initialize valid data bytes accounting at file set up Shin'ichiro Kawasaki
2023-02-07 14:06 ` Niklas Cassel [this message]
2023-02-08 5:03 ` Shinichiro Kawasaki
2023-02-08 8:30 ` Niklas Cassel
2023-02-07 6:37 ` [PATCH v2 8/8] t/zbd: add test cases for zone_reset_threshold option Shin'ichiro Kawasaki
2023-02-07 14:06 ` Niklas Cassel
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=Y+Ja5jyZq/snxRpI@x1-carbon \
--to=niklas.cassel@wdc.com \
--cc=Dmitry.Fomichev@wdc.com \
--cc=axboe@kernel.dk \
--cc=damien.lemoal@opensource.wdc.com \
--cc=fio@vger.kernel.org \
--cc=shinichiro.kawasaki@wdc.com \
--cc=vincentfu@gmail.com \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).