When the Partial Parity Log is enabled, circular buffer is used to store
PPL data. Each write to RAID device causes overwrite of data in this buffer
so some write_hint can be set to those request to help drives handle
garbage collection. This patch adds new sysfs attribute which can be used
to specify which write_hint should be assigned to PPL.
Signed-off-by: Mariusz Dabrowski <mariusz.dabrowski@xxxxxxxxx>
---
Documentation/admin-guide/md.rst | 3 +++
drivers/md/raid5-ppl.c | 3 +++
drivers/md/raid5.c | 50 ++++++++++++++++++++++++++++++++++++++++
drivers/md/raid5.h | 1 +
4 files changed, 57 insertions(+)
diff --git a/Documentation/admin-guide/md.rst b/Documentation/admin-guide/md.rst
index 84de718f24a4..3c51084ffd37 100644
--- a/Documentation/admin-guide/md.rst
+++ b/Documentation/admin-guide/md.rst
@@ -756,3 +756,6 @@ These currently include:
The cache mode for raid5. raid5 could include an extra disk for
caching. The mode can be "write-throuth" and "write-back". The
default is "write-through".
+
+ ppl_write_hint
+ NVMe stream ID to be set for each PPL write request.
diff --git a/drivers/md/raid5-ppl.c b/drivers/md/raid5-ppl.c
index 3a7c36326589..46af40960ad2 100644
--- a/drivers/md/raid5-ppl.c
+++ b/drivers/md/raid5-ppl.c
@@ -440,6 +440,7 @@ static void ppl_submit_iounit(struct ppl_io_unit *io)
struct ppl_header *pplhdr = page_address(io->header_page);
struct bio *bio = &io->bio;
struct stripe_head *sh;
+ struct r5conf *r5_conf = ppl_conf->mddev->private;
int i;
bio->bi_private = io;
@@ -476,6 +477,7 @@ static void ppl_submit_iounit(struct ppl_io_unit *io)
bio_set_dev(bio, log->rdev->bdev);
bio->bi_iter.bi_sector = log->next_io_sector;
bio_add_page(bio, io->header_page, PAGE_SIZE, 0);
+ bio->bi_write_hint = r5_conf->ppl_write_hint;
pr_debug("%s: log->current_io_sector: %llu\n", __func__,
(unsigned long long)log->next_io_sector);
@@ -505,6 +507,7 @@ static void ppl_submit_iounit(struct ppl_io_unit *io)
bio = bio_alloc_bioset(GFP_NOIO, BIO_MAX_PAGES,
&ppl_conf->bs);
bio->bi_opf = prev->bi_opf;
+ bio->bi_write_hint = prev->bi_write_hint;
bio_copy_dev(bio, prev);
bio->bi_iter.bi_sector = bio_end_sector(prev);
bio_add_page(bio, sh->ppl_page, PAGE_SIZE, 0);
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index cecea901ab8c..2b2039b89445 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -6652,6 +6652,53 @@ raid5_group_thread_cnt = __ATTR(group_thread_cnt, S_IRUGO | S_IWUSR,
raid5_show_group_thread_cnt,
raid5_store_group_thread_cnt);
+static ssize_t
+raid5_show_ppl_write_hint(struct mddev *mddev, char *buf)
+{
+ size_t ret = 0;
+ struct r5conf *conf;
+
+ spin_lock(&mddev->lock);
+ conf = mddev->private;
+ if (conf)
+ ret = sprintf(buf, "%d\n", conf->ppl_write_hint);
+ spin_unlock(&mddev->lock);
+
+ return ret;
+}
+
+static ssize_t
+raid5_store_ppl_write_hint(struct mddev *mddev, const char *page, size_t len)
+{
+ struct r5conf *conf;
+ int err = 0;
+ unsigned short new;
+
+ if (len >= PAGE_SIZE)
+ return -EINVAL;
+ if (kstrtou16(page, 10, &new))
+ return -EINVAL;
+
+ err = mddev_lock(mddev);
+ if (err)
+ return err;
+
+ conf = mddev->private;
+ if (!conf)
+ err = -ENODEV;
+ else
+ conf->ppl_write_hint = new;
+
+ mddev_unlock(mddev);
+
+ return err ?: len;
+}
+
+static struct md_sysfs_entry
+raid5_ppl_write_hint = __ATTR(ppl_write_hint, S_IRUGO | S_IWUSR,
+ raid5_show_ppl_write_hint,
+ raid5_store_ppl_write_hint);
+
static struct attribute *raid5_attrs[] = {
&raid5_stripecache_size.attr,
&raid5_stripecache_active.attr,
@@ -6660,6 +6707,7 @@ static struct attribute *raid5_attrs[] = {
&raid5_skip_copy.attr,
&raid5_rmw_level.attr,
&r5c_journal_mode.attr,
+ &raid5_ppl_write_hint.attr,
NULL,
};
static struct attribute_group raid5_attrs_group = {
@@ -7104,6 +7152,8 @@ static struct r5conf *setup_conf(struct mddev *mddev)
goto abort;
}
+ conf->ppl_write_hint = RWF_WRITE_LIFE_NOT_SET;
+
return conf;
abort:
diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h
index 8474c224127b..8ff1c06cf5d4 100644
--- a/drivers/md/raid5.h
+++ b/drivers/md/raid5.h
@@ -577,6 +577,7 @@ struct r5conf {
int raid_disks;
int max_nr_stripes;
int min_nr_stripes;
+ unsigned short ppl_write_hint;
/* reshape_progress is the leading edge of a 'reshape'
* It has value MaxSector when no reshape is happening
--
2.16.4
