It's just a variant of wait_event_cmd, with exclusive flag being set.
For cases like RAID5, which puts many processes to sleep until 1/4
resources are free, a wake_up wakes up all processes to run, but
there is one process being able to get the resource as it's protected
by a spin lock. That ends up introducing heavy lock contentions, and
hurts performance badly.
Here introduce wait_event_cmd_exclusive to relieve the lock contention
naturally by letting wake_up() just wake up one process.
Cc: Ingo Molnar <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Signed-off-by: Yuanhan Liu <[email protected]>
---
include/linux/wait.h | 14 +++++++++++---
1 file changed, 11 insertions(+), 3 deletions(-)
diff --git a/include/linux/wait.h b/include/linux/wait.h
index 2db8334..6c3b4de 100644
--- a/include/linux/wait.h
+++ b/include/linux/wait.h
@@ -358,10 +358,18 @@ do { \
__ret; \
})
-#define __wait_event_cmd(wq, condition, cmd1, cmd2) \
- (void)___wait_event(wq, condition, TASK_UNINTERRUPTIBLE, 0, 0, \
+#define __wait_event_cmd(wq, condition, cmd1, cmd2, exclusive) \
+ (void)___wait_event(wq, condition, TASK_UNINTERRUPTIBLE, exclusive, 0, \
cmd1; schedule(); cmd2)
+
+#define wait_event_cmd_exclusive(wq, condition, cmd1, cmd2) \
+do { \
+ if (condition) \
+ break; \
+ __wait_event_cmd(wq, condition, cmd1, cmd2, 1); \
+} while (0)
+
/**
* wait_event_cmd - sleep until a condition gets true
* @wq: the waitqueue to wait on
@@ -380,7 +388,7 @@ do { \
do { \
if (condition) \
break; \
- __wait_event_cmd(wq, condition, cmd1, cmd2); \
+ __wait_event_cmd(wq, condition, cmd1, cmd2, 0); \
} while (0)
#define __wait_event_interruptible(wq, condition) \
--
1.9.0
I noticed heavy spin lock contention at get_active_stripe(), introduced
at being wake up stage, where a bunch of processes try to re-hold the
spin lock again.
After giving some thoughts on this issue, I found the lock could be
relieved(and even avoided) if we turn the wait_for_stripe to per
waitqueue for each lock hash and make the wake up exclusive: wake up
one process each time, which avoids the lock contention naturally.
Before go hacking with wait_for_stripe, I found it actually has 2
usages: for the array to enter or leave the quiescent state, and also
to wait for an available stripe in each of the hash lists.
So this patch splits the first usage off into a separate wait_queue,
wait_for_quiescent, and the next patch will turn the second usage into
one waitqueue for each hash value, and make it exclusive, to relieve
the lock contention.
v2: wake_up(wait_for_quiescent) when (active_stripes == 0)
Commit log refactor suggestion from Neil.
Signed-off-by: Yuanhan Liu <[email protected]>
---
drivers/md/raid5.c | 15 +++++++++------
drivers/md/raid5.h | 1 +
2 files changed, 10 insertions(+), 6 deletions(-)
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 77dfd72..64d5bea 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -374,6 +374,8 @@ static void release_inactive_stripe_list(struct r5conf *conf,
if (do_wakeup) {
wake_up(&conf->wait_for_stripe);
+ if (atomic_read(&conf->active_stripes) == 0)
+ wake_up(&conf->wait_for_quiescent);
if (conf->retry_read_aligned)
md_wakeup_thread(conf->mddev->thread);
}
@@ -667,7 +669,7 @@ get_active_stripe(struct r5conf *conf, sector_t sector,
spin_lock_irq(conf->hash_locks + hash);
do {
- wait_event_lock_irq(conf->wait_for_stripe,
+ wait_event_lock_irq(conf->wait_for_quiescent,
conf->quiesce == 0 || noquiesce,
*(conf->hash_locks + hash));
sh = __find_stripe(conf, sector, conf->generation - previous);
@@ -4729,7 +4731,7 @@ static void raid5_align_endio(struct bio *bi, int error)
raid_bi, 0);
bio_endio(raid_bi, 0);
if (atomic_dec_and_test(&conf->active_aligned_reads))
- wake_up(&conf->wait_for_stripe);
+ wake_up(&conf->wait_for_quiescent);
return;
}
@@ -4824,7 +4826,7 @@ static int chunk_aligned_read(struct mddev *mddev, struct bio * raid_bio)
align_bi->bi_iter.bi_sector += rdev->data_offset;
spin_lock_irq(&conf->device_lock);
- wait_event_lock_irq(conf->wait_for_stripe,
+ wait_event_lock_irq(conf->wait_for_quiescent,
conf->quiesce == 0,
conf->device_lock);
atomic_inc(&conf->active_aligned_reads);
@@ -5668,7 +5670,7 @@ static int retry_aligned_read(struct r5conf *conf, struct bio *raid_bio)
bio_endio(raid_bio, 0);
}
if (atomic_dec_and_test(&conf->active_aligned_reads))
- wake_up(&conf->wait_for_stripe);
+ wake_up(&conf->wait_for_quiescent);
return handled;
}
@@ -6399,6 +6401,7 @@ static struct r5conf *setup_conf(struct mddev *mddev)
goto abort;
spin_lock_init(&conf->device_lock);
seqcount_init(&conf->gen_lock);
+ init_waitqueue_head(&conf->wait_for_quiescent);
init_waitqueue_head(&conf->wait_for_stripe);
init_waitqueue_head(&conf->wait_for_overlap);
INIT_LIST_HEAD(&conf->handle_list);
@@ -7422,7 +7425,7 @@ static void raid5_quiesce(struct mddev *mddev, int state)
* active stripes can drain
*/
conf->quiesce = 2;
- wait_event_cmd(conf->wait_for_stripe,
+ wait_event_cmd(conf->wait_for_quiescent,
atomic_read(&conf->active_stripes) == 0 &&
atomic_read(&conf->active_aligned_reads) == 0,
unlock_all_device_hash_locks_irq(conf),
@@ -7436,7 +7439,7 @@ static void raid5_quiesce(struct mddev *mddev, int state)
case 0: /* re-enable writes */
lock_all_device_hash_locks_irq(conf);
conf->quiesce = 0;
- wake_up(&conf->wait_for_stripe);
+ wake_up(&conf->wait_for_quiescent);
wake_up(&conf->wait_for_overlap);
unlock_all_device_hash_locks_irq(conf);
break;
diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h
index 7dc0dd8..4cc05ec 100644
--- a/drivers/md/raid5.h
+++ b/drivers/md/raid5.h
@@ -508,6 +508,7 @@ struct r5conf {
struct list_head inactive_list[NR_STRIPE_HASH_LOCKS];
atomic_t empty_inactive_list_nr;
struct llist_head released_stripes;
+ wait_queue_head_t wait_for_quiescent;
wait_queue_head_t wait_for_stripe;
wait_queue_head_t wait_for_overlap;
unsigned long cache_state;
--
1.9.0
I noticed heavy spin lock contention at get_active_stripe() with fsmark
multiple thread write workloads.
Here is how this hot contention comes from. We have limited stripes, and
it's a multiple thread write workload. Hence, those stripes will be taken
soon, which puts later processes to sleep for waiting free stripes. When
enough stripes(>= 1/4 total stripes) are released, all process are woken,
trying to get the lock. But there is one only being able to get this lock
for each hash lock, making other processes spinning out there for acquiring
the lock.
Thus, it's effectiveless to wakeup all processes and let them battle for
a lock that permits one to access only each time. Instead, we could make
it be a exclusive wake up: wake up one process only. That avoids the heavy
spin lock contention naturally.
To do the exclusive wake up, we've to split wait_for_stripe into multiple
wait queues, to make it per hash value, just like the hash lock.
Here are some test results I have got with this patch applied(all test run
3 times):
`fsmark.files_per_sec'
=====================
next-20150317 this patch
------------------------- -------------------------
metric_value ±stddev metric_value ±stddev change testbox/benchmark/testcase-params
------------------------- ------------------------- -------- ------------------------------
25.600 ±0.0 92.700 ±2.5 262.1% ivb44/fsmark/1x-64t-4BRD_12G-RAID5-btrfs-4M-30G-fsyncBeforeClose
25.600 ±0.0 77.800 ±0.6 203.9% ivb44/fsmark/1x-64t-9BRD_6G-RAID5-btrfs-4M-30G-fsyncBeforeClose
32.000 ±0.0 93.800 ±1.7 193.1% ivb44/fsmark/1x-64t-4BRD_12G-RAID5-ext4-4M-30G-fsyncBeforeClose
32.000 ±0.0 81.233 ±1.7 153.9% ivb44/fsmark/1x-64t-9BRD_6G-RAID5-ext4-4M-30G-fsyncBeforeClose
48.800 ±14.5 99.667 ±2.0 104.2% ivb44/fsmark/1x-64t-4BRD_12G-RAID5-xfs-4M-30G-fsyncBeforeClose
6.400 ±0.0 12.800 ±0.0 100.0% ivb44/fsmark/1x-64t-3HDD-RAID5-btrfs-4M-40G-fsyncBeforeClose
63.133 ±8.2 82.800 ±0.7 31.2% ivb44/fsmark/1x-64t-9BRD_6G-RAID5-xfs-4M-30G-fsyncBeforeClose
245.067 ±0.7 306.567 ±7.9 25.1% ivb44/fsmark/1x-64t-4BRD_12G-RAID5-f2fs-4M-30G-fsyncBeforeClose
17.533 ±0.3 21.000 ±0.8 19.8% ivb44/fsmark/1x-1t-3HDD-RAID5-xfs-4M-40G-fsyncBeforeClose
188.167 ±1.9 215.033 ±3.1 14.3% ivb44/fsmark/1x-1t-4BRD_12G-RAID5-btrfs-4M-30G-NoSync
254.500 ±1.8 290.733 ±2.4 14.2% ivb44/fsmark/1x-1t-9BRD_6G-RAID5-btrfs-4M-30G-NoSync
`time.system_time'
=====================
next-20150317 this patch
------------------------- -------------------------
metric_value ±stddev metric_value ±stddev change testbox/benchmark/testcase-params
------------------------- ------------------------- -------- ------------------------------
7235.603 ±1.2 185.163 ±1.9 -97.4% ivb44/fsmark/1x-64t-4BRD_12G-RAID5-btrfs-4M-30G-fsyncBeforeClose
7666.883 ±2.9 202.750 ±1.0 -97.4% ivb44/fsmark/1x-64t-9BRD_6G-RAID5-btrfs-4M-30G-fsyncBeforeClose
14567.893 ±0.7 421.230 ±0.4 -97.1% ivb44/fsmark/1x-64t-3HDD-RAID5-btrfs-4M-40G-fsyncBeforeClose
3697.667 ±14.0 148.190 ±1.7 -96.0% ivb44/fsmark/1x-64t-4BRD_12G-RAID5-xfs-4M-30G-fsyncBeforeClose
5572.867 ±3.8 310.717 ±1.4 -94.4% ivb44/fsmark/1x-64t-9BRD_6G-RAID5-ext4-4M-30G-fsyncBeforeClose
5565.050 ±0.5 313.277 ±1.5 -94.4% ivb44/fsmark/1x-64t-4BRD_12G-RAID5-ext4-4M-30G-fsyncBeforeClose
2420.707 ±17.1 171.043 ±2.7 -92.9% ivb44/fsmark/1x-64t-9BRD_6G-RAID5-xfs-4M-30G-fsyncBeforeClose
3743.300 ±4.6 379.827 ±3.5 -89.9% ivb44/fsmark/1x-64t-3HDD-RAID5-ext4-4M-40G-fsyncBeforeClose
3308.687 ±6.3 363.050 ±2.0 -89.0% ivb44/fsmark/1x-64t-3HDD-RAID5-xfs-4M-40G-fsyncBeforeClose
Where,
1x: where 'x' means iterations or loop, corresponding to the 'L' option of fsmark
1t, 64t: where 't' means thread
4M: means the single file size, corresponding to the '-s' option of fsmark
40G, 30G, 120G: means the total test size
4BRD_12G: BRD is the ramdisk, where '4' means 4 ramdisk, and where '12G' means
the size of one ramdisk. So, it would be 48G in total. And we made a
raid on those ramdisk
As you can see, though there are no much performance gain for hard disk
workload, the system time is dropped heavily, up to 97%. And as expected,
the performance increased a lot, up to 260%, for fast device(ram disk).
v2: use bits instead of array to note down wait queue need to wake up.
Signed-off-by: Yuanhan Liu <[email protected]>
---
drivers/md/raid5.c | 27 +++++++++++++++++++--------
drivers/md/raid5.h | 2 +-
2 files changed, 20 insertions(+), 9 deletions(-)
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 64d5bea..1b11bbf 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -344,7 +344,8 @@ static void release_inactive_stripe_list(struct r5conf *conf,
int hash)
{
int size;
- bool do_wakeup = false;
+ unsigned long do_wakeup = 0;
+ int i = 0;
unsigned long flags;
if (hash == NR_STRIPE_HASH_LOCKS) {
@@ -365,15 +366,19 @@ static void release_inactive_stripe_list(struct r5conf *conf,
!list_empty(list))
atomic_dec(&conf->empty_inactive_list_nr);
list_splice_tail_init(list, conf->inactive_list + hash);
- do_wakeup = true;
+ do_wakeup |= 1 << (size - 1);
spin_unlock_irqrestore(conf->hash_locks + hash, flags);
}
size--;
hash--;
}
+ for (i = 0; i < NR_STRIPE_HASH_LOCKS; i++) {
+ if (do_wakeup & (1 << i))
+ wake_up(&conf->wait_for_stripe[i]);
+ }
+
if (do_wakeup) {
- wake_up(&conf->wait_for_stripe);
if (atomic_read(&conf->active_stripes) == 0)
wake_up(&conf->wait_for_quiescent);
if (conf->retry_read_aligned)
@@ -686,14 +691,15 @@ get_active_stripe(struct r5conf *conf, sector_t sector,
if (!sh) {
set_bit(R5_INACTIVE_BLOCKED,
&conf->cache_state);
- wait_event_lock_irq(
- conf->wait_for_stripe,
+ wait_event_cmd_exclusive(
+ conf->wait_for_stripe[hash],
!list_empty(conf->inactive_list + hash) &&
(atomic_read(&conf->active_stripes)
< (conf->max_nr_stripes * 3 / 4)
|| !test_bit(R5_INACTIVE_BLOCKED,
&conf->cache_state)),
- *(conf->hash_locks + hash));
+ spin_unlock_irq(conf->hash_locks + hash),
+ spin_lock_irq(conf->hash_locks + hash));
clear_bit(R5_INACTIVE_BLOCKED,
&conf->cache_state);
} else {
@@ -718,6 +724,9 @@ get_active_stripe(struct r5conf *conf, sector_t sector,
}
} while (sh == NULL);
+ if (!list_empty(conf->inactive_list + hash))
+ wake_up(&conf->wait_for_stripe[hash]);
+
spin_unlock_irq(conf->hash_locks + hash);
return sh;
}
@@ -2138,7 +2147,7 @@ static int resize_stripes(struct r5conf *conf, int newsize)
cnt = 0;
list_for_each_entry(nsh, &newstripes, lru) {
lock_device_hash_lock(conf, hash);
- wait_event_cmd(conf->wait_for_stripe,
+ wait_event_cmd_exclusive(conf->wait_for_stripe[hash],
!list_empty(conf->inactive_list + hash),
unlock_device_hash_lock(conf, hash),
lock_device_hash_lock(conf, hash));
@@ -6402,7 +6411,9 @@ static struct r5conf *setup_conf(struct mddev *mddev)
spin_lock_init(&conf->device_lock);
seqcount_init(&conf->gen_lock);
init_waitqueue_head(&conf->wait_for_quiescent);
- init_waitqueue_head(&conf->wait_for_stripe);
+ for (i = 0; i < NR_STRIPE_HASH_LOCKS; i++) {
+ init_waitqueue_head(&conf->wait_for_stripe[i]);
+ }
init_waitqueue_head(&conf->wait_for_overlap);
INIT_LIST_HEAD(&conf->handle_list);
INIT_LIST_HEAD(&conf->hold_list);
diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h
index 4cc05ec..6307b90 100644
--- a/drivers/md/raid5.h
+++ b/drivers/md/raid5.h
@@ -509,7 +509,7 @@ struct r5conf {
atomic_t empty_inactive_list_nr;
struct llist_head released_stripes;
wait_queue_head_t wait_for_quiescent;
- wait_queue_head_t wait_for_stripe;
+ wait_queue_head_t wait_for_stripe[NR_STRIPE_HASH_LOCKS];
wait_queue_head_t wait_for_overlap;
unsigned long cache_state;
#define R5_INACTIVE_BLOCKED 1 /* release of inactive stripes blocked,
--
1.9.0
On Mon, Apr 27, 2015 at 12:51:01PM +0800, Yuanhan Liu wrote:
> It's just a variant of wait_event_cmd, with exclusive flag being set.
>
> For cases like RAID5, which puts many processes to sleep until 1/4
> resources are free, a wake_up wakes up all processes to run, but
> there is one process being able to get the resource as it's protected
> by a spin lock. That ends up introducing heavy lock contentions, and
> hurts performance badly.
>
> Here introduce wait_event_cmd_exclusive to relieve the lock contention
> naturally by letting wake_up() just wake up one process.
>
> Cc: Ingo Molnar <[email protected]>
> Cc: Peter Zijlstra <[email protected]>
> Signed-off-by: Yuanhan Liu <[email protected]>
> ---
> include/linux/wait.h | 14 +++++++++++---
> 1 file changed, 11 insertions(+), 3 deletions(-)
>
> diff --git a/include/linux/wait.h b/include/linux/wait.h
> index 2db8334..6c3b4de 100644
> --- a/include/linux/wait.h
> +++ b/include/linux/wait.h
> @@ -358,10 +358,18 @@ do { \
> __ret; \
> })
>
> -#define __wait_event_cmd(wq, condition, cmd1, cmd2) \
> - (void)___wait_event(wq, condition, TASK_UNINTERRUPTIBLE, 0, 0, \
> +#define __wait_event_cmd(wq, condition, cmd1, cmd2, exclusive) \
> + (void)___wait_event(wq, condition, TASK_UNINTERRUPTIBLE, exclusive, 0, \
> cmd1; schedule(); cmd2)
>
> +
> +#define wait_event_cmd_exclusive(wq, condition, cmd1, cmd2) \
> +do { \
> + if (condition) \
> + break; \
> + __wait_event_cmd(wq, condition, cmd1, cmd2, 1); \
> +} while (0)
> +
> /**
> * wait_event_cmd - sleep until a condition gets true
> * @wq: the waitqueue to wait on
> @@ -380,7 +388,7 @@ do { \
> do { \
> if (condition) \
> break; \
> - __wait_event_cmd(wq, condition, cmd1, cmd2); \
> + __wait_event_cmd(wq, condition, cmd1, cmd2, 0); \
> } while (0)
>
No, that's wrong, its assumed that wait*() and __wait*() have the same
arguments.
On Tue, Apr 28, 2015 at 04:13:15PM +0200, Peter Zijlstra wrote:
> On Mon, Apr 27, 2015 at 12:51:01PM +0800, Yuanhan Liu wrote:
> > It's just a variant of wait_event_cmd, with exclusive flag being set.
> >
> > For cases like RAID5, which puts many processes to sleep until 1/4
> > resources are free, a wake_up wakes up all processes to run, but
> > there is one process being able to get the resource as it's protected
> > by a spin lock. That ends up introducing heavy lock contentions, and
> > hurts performance badly.
> >
> > Here introduce wait_event_cmd_exclusive to relieve the lock contention
> > naturally by letting wake_up() just wake up one process.
> >
> > Cc: Ingo Molnar <[email protected]>
> > Cc: Peter Zijlstra <[email protected]>
> > Signed-off-by: Yuanhan Liu <[email protected]>
> > ---
> > include/linux/wait.h | 14 +++++++++++---
> > 1 file changed, 11 insertions(+), 3 deletions(-)
> >
> > diff --git a/include/linux/wait.h b/include/linux/wait.h
> > index 2db8334..6c3b4de 100644
> > --- a/include/linux/wait.h
> > +++ b/include/linux/wait.h
> > @@ -358,10 +358,18 @@ do { \
> > __ret; \
> > })
> >
> > -#define __wait_event_cmd(wq, condition, cmd1, cmd2) \
> > - (void)___wait_event(wq, condition, TASK_UNINTERRUPTIBLE, 0, 0, \
> > +#define __wait_event_cmd(wq, condition, cmd1, cmd2, exclusive) \
> > + (void)___wait_event(wq, condition, TASK_UNINTERRUPTIBLE, exclusive, 0, \
> > cmd1; schedule(); cmd2)
> >
> > +
> > +#define wait_event_cmd_exclusive(wq, condition, cmd1, cmd2) \
> > +do { \
> > + if (condition) \
> > + break; \
> > + __wait_event_cmd(wq, condition, cmd1, cmd2, 1); \
> > +} while (0)
> > +
> > /**
> > * wait_event_cmd - sleep until a condition gets true
> > * @wq: the waitqueue to wait on
> > @@ -380,7 +388,7 @@ do { \
> > do { \
> > if (condition) \
> > break; \
> > - __wait_event_cmd(wq, condition, cmd1, cmd2); \
> > + __wait_event_cmd(wq, condition, cmd1, cmd2, 0); \
> > } while (0)
> >
>
> No, that's wrong, its assumed that wait*() and __wait*() have the same
> arguments.
Thanks. Will send an updated patch soon.
--yliu