Re: [PATCH -next v2 7/7] md/raid1-10: limit the number of plugged bio

From: Yu Kuai
Date: Mon May 29 2023 - 04:50:45 EST


在 2023/05/29 15:57, Xiao Ni 写道:
On Mon, May 29, 2023 at 11:18 AM Yu Kuai <yukuai1@xxxxxxxxxxxxxxx> wrote:


在 2023/05/29 11:10, Xiao Ni 写道:
On Mon, May 29, 2023 at 10:20 AM Yu Kuai <yukuai1@xxxxxxxxxxxxxxx> wrote:


在 2023/05/29 10:08, Xiao Ni 写道:
Hi Kuai

There is a limitation of the memory in your test. But for most
situations, customers should not set this. Can this change introduce a
performance regression against other situations?

Noted that this limitation is just to triggered writeback as soon as
possible in the test, and it's 100% sure real situations can trigger
dirty pages write back asynchronously and continue to produce new dirty


I'm confused here. If we want to trigger write back quickly, it needs
to set these two values with a smaller number, rather than 0 and 60.

60 is not required, I'll remove this setting.

0 just means write back if there are any dirty pages.

Hi Kuai

Does 0 mean disabling write back? I tried to find the doc that
describes the meaning when setting dirty_background_ratio to 0, but I
didn't find it.
In it
doesn't describe this. But it says something like this

dirty_background_bytes is the counterpart of dirty_background_ratio. Only
one of them may be specified at a time. When one sysctl is written it is
immediately taken into account to evaluate the dirty memory limits and the
other appears as 0 when read.

Maybe you can specify dirty_background_ratio to 1 if you want to
trigger write back ASAP.

The purpose here is to trigger write back ASAP, I'm not an expert here,
but based on test result, 0 obviously doesn't mean disable write back.

Set dirty_background_bytes to a value, dirty_background_ratio will be
set to 0 together, which means dirty_background_ratio is disabled.
However, change dirty_background_ratio from default value to 0, will end
up both dirty_background_ratio and dirty_background_bytes to be 0, and
based on following related code, I think 0 just means write back if
there are any dirty pages.

bg_bytes = dirty_background_bytes -> 0
bg_ratio = (dirty_background_ratio * PAGE_SIZE) / 100 -> 0

if (bg_bytes)
bg_thresh = DIV_ROUND_UP(bg_bytes, PAGE_SIZE);
bg_thresh = (bg_ratio * available_memory) / PAGE_SIZE; -> 0

dtc->bg_thresh = bg_thresh; -> 0

nr_reclaimable = global_node_page_state(NR_FILE_DIRTY);
if (!laptop_mode && nr_reclaimable > gdtc->bg_thresh &&
wb_start_background_writeback(wb); -> writeback ASAP


If a lot of bio is not plugged, then it's the same as before; if a lot
of bio is plugged, noted that before this patchset, these bio will spent
quite a long time in plug, and hence I think performance should be

Hmm, it depends on if it's sequential or not? If it's a big io
request, can it miss the merge opportunity?

The bio will still be merged to underlying disks' rq(if it's rq based),
underlying disk won't flush plug untill the number of request exceed

Thanks for this.





Best Regards

On Wed, Apr 26, 2023 at 4:24 PM Yu Kuai <yukuai1@xxxxxxxxxxxxxxx> wrote:

From: Yu Kuai <yukuai3@xxxxxxxxxx>

bio can be added to plug infinitely, and following writeback test can
trigger huge amount of plugged bio:

Test script:
modprobe brd rd_nr=4 rd_size=10485760
mdadm -CR /dev/md0 -l10 -n4 /dev/ram[0123] --assume-clean
echo 0 > /proc/sys/vm/dirty_background_ratio
echo 60 > /proc/sys/vm/dirty_ratio
fio -filename=/dev/md0 -ioengine=libaio -rw=write -bs=4k -numjobs=1 -iodepth=128 -name=test

Test result:
Monitor /sys/block/md0/inflight will found that inflight keep increasing
until fio finish writing, after running for about 2 minutes:

[root@fedora ~]# cat /sys/block/md0/inflight
0 4474191

Fix the problem by limiting the number of plugged bio based on the number
of copies for original bio.

Signed-off-by: Yu Kuai <yukuai3@xxxxxxxxxx>
drivers/md/raid1-10.c | 9 ++++++++-
drivers/md/raid1.c | 2 +-
drivers/md/raid10.c | 2 +-
3 files changed, 10 insertions(+), 3 deletions(-)

diff --git a/drivers/md/raid1-10.c b/drivers/md/raid1-10.c
index 98d678b7df3f..35fb80aa37aa 100644
--- a/drivers/md/raid1-10.c
+++ b/drivers/md/raid1-10.c
@@ -21,6 +21,7 @@
#define IO_MADE_GOOD ((struct bio *)2)

#define BIO_SPECIAL(bio) ((unsigned long)bio <= 2)
+#define MAX_PLUG_BIO 32

/* for managing resync I/O pages */
struct resync_pages {
@@ -31,6 +32,7 @@ struct resync_pages {
struct raid1_plug_cb {
struct blk_plug_cb cb;
struct bio_list pending;
+ unsigned int count;

static void rbio_pool_free(void *rbio, void *data)
@@ -127,7 +129,7 @@ static inline void md_submit_write(struct bio *bio)

static inline bool md_add_bio_to_plug(struct mddev *mddev, struct bio *bio,
- blk_plug_cb_fn unplug)
+ blk_plug_cb_fn unplug, int copies)
struct raid1_plug_cb *plug = NULL;
struct blk_plug_cb *cb;
@@ -147,6 +149,11 @@ static inline bool md_add_bio_to_plug(struct mddev *mddev, struct bio *bio,

plug = container_of(cb, struct raid1_plug_cb, cb);
bio_list_add(&plug->pending, bio);
+ if (++plug->count / MAX_PLUG_BIO >= copies) {
+ list_del(&cb->list);
+ cb->callback(cb, false);
+ }

return true;
diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
index 639e09cecf01..c6066408a913 100644
--- a/drivers/md/raid1.c
+++ b/drivers/md/raid1.c
@@ -1562,7 +1562,7 @@ static void raid1_write_request(struct mddev *mddev, struct bio *bio,
/* flush_pending_writes() needs access to the rdev so...*/
mbio->bi_bdev = (void *)rdev;
- if (!md_add_bio_to_plug(mddev, mbio, raid1_unplug)) {
+ if (!md_add_bio_to_plug(mddev, mbio, raid1_unplug, disks)) {
spin_lock_irqsave(&conf->device_lock, flags);
bio_list_add(&conf->pending_bio_list, mbio);
spin_unlock_irqrestore(&conf->device_lock, flags);
diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index bd9e655ca408..7135cfaf75db 100644
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -1306,7 +1306,7 @@ static void raid10_write_one_disk(struct mddev *mddev, struct r10bio *r10_bio,


- if (!md_add_bio_to_plug(mddev, mbio, raid10_unplug)) {
+ if (!md_add_bio_to_plug(mddev, mbio, raid10_unplug, conf->copies)) {
spin_lock_irqsave(&conf->device_lock, flags);
bio_list_add(&conf->pending_bio_list, mbio);
spin_unlock_irqrestore(&conf->device_lock, flags);