2023-11-28 08:01:43

by Uladzislau Rezki

[permalink] [raw]
Subject: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

From: Neeraj Upadhyay <[email protected]>

Currently, processing of the next batch of rcu_synchronize nodes
for the new grace period, requires doing a llist reversal operation
to find the tail element of the list. This can be a very costly
operation (high number of cache misses) for a long list.

To address this, this patch introduces a "dummy-wait-node" entity.
At every grace period init, a new wait node is added to the llist.
This wait node is used as wait tail for this new grace period.

This allows lockless additions of new rcu_synchronize nodes in the
rcu_sr_normal_add_req(), while the cleanup work executes and does
the progress. The dummy nodes are removed on next round of cleanup
work execution.

Signed-off-by: Uladzislau Rezki (Sony) <[email protected]>
Signed-off-by: Neeraj Upadhyay <[email protected]>
---
kernel/rcu/tree.c | 270 +++++++++++++++++++++++++++++++++++++++-------
1 file changed, 233 insertions(+), 37 deletions(-)

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 975621ef40e3..d7b48996825f 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -1384,25 +1384,173 @@ static void rcu_poll_gp_seq_end_unlocked(unsigned long *snap)
raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
}

+#define SR_NORMAL_GP_WAIT_HEAD_MAX 5
+
+struct sr_wait_node {
+ atomic_t inuse;
+ struct llist_node node;
+};
+
/*
- * There are three lists for handling synchronize_rcu() users.
- * A first list corresponds to new coming users, second for users
- * which wait for a grace period and third is for which a grace
- * period is passed.
+ * There is a single llist, which is used for handling
+ * synchronize_rcu() users' enqueued rcu_synchronize nodes.
+ * Within this llist, there are two tail pointers:
+ *
+ * wait tail: Tracks the set of nodes, which need to
+ * wait for the current GP to complete.
+ * done tail: Tracks the set of nodes, for which grace
+ * period has elapsed. These nodes processing
+ * will be done as part of the cleanup work
+ * execution by a kworker.
+ *
+ * At every grace period init, a new wait node is added
+ * to the llist. This wait node is used as wait tail
+ * for this new grace period. Given that there are a fixed
+ * number of wait nodes, if all wait nodes are in use
+ * (which can happen when kworker callback processing
+ * is delayed) and additional grace period is requested.
+ * This means, a system is slow in processing callbacks.
+ *
+ * TODO: If a slow processing is detected, a first node
+ * in the llist should be used as a wait-tail for this
+ * grace period, therefore users which should wait due
+ * to a slow process are handled by _this_ grace period
+ * and not next.
+ *
+ * Below is an illustration of how the done and wait
+ * tail pointers move from one set of rcu_synchronize nodes
+ * to the other, as grace periods start and finish and
+ * nodes are processed by kworker.
+ *
+ *
+ * a. Initial llist callbacks list:
+ *
+ * +----------+ +--------+ +-------+
+ * | | | | | |
+ * | head |---------> | cb2 |--------->| cb1 |
+ * | | | | | |
+ * +----------+ +--------+ +-------+
+ *
+ *
+ *
+ * b. New GP1 Start:
+ *
+ * WAIT TAIL
+ * |
+ * |
+ * v
+ * +----------+ +--------+ +--------+ +-------+
+ * | | | | | | | |
+ * | head ------> wait |------> cb2 |------> | cb1 |
+ * | | | head1 | | | | |
+ * +----------+ +--------+ +--------+ +-------+
+ *
+ *
+ *
+ * c. GP completion:
+ *
+ * WAIT_TAIL == DONE_TAIL
+ *
+ * DONE TAIL
+ * |
+ * |
+ * v
+ * +----------+ +--------+ +--------+ +-------+
+ * | | | | | | | |
+ * | head ------> wait |------> cb2 |------> | cb1 |
+ * | | | head1 | | | | |
+ * +----------+ +--------+ +--------+ +-------+
+ *
+ *
+ *
+ * d. New callbacks and GP2 start:
+ *
+ * WAIT TAIL DONE TAIL
+ * | |
+ * | |
+ * v v
+ * +----------+ +------+ +------+ +------+ +-----+ +-----+ +-----+
+ * | | | | | | | | | | | | | |
+ * | head ------> wait |--->| cb4 |--->| cb3 |--->|wait |--->| cb2 |--->| cb1 |
+ * | | | head2| | | | | |head1| | | | |
+ * +----------+ +------+ +------+ +------+ +-----+ +-----+ +-----+
+ *
+ *
+ *
+ * e. GP2 completion:
+ *
+ * WAIT_TAIL == DONE_TAIL
+ * DONE TAIL
+ * |
+ * |
+ * v
+ * +----------+ +------+ +------+ +------+ +-----+ +-----+ +-----+
+ * | | | | | | | | | | | | | |
+ * | head ------> wait |--->| cb4 |--->| cb3 |--->|wait |--->| cb2 |--->| cb1 |
+ * | | | head2| | | | | |head1| | | | |
+ * +----------+ +------+ +------+ +------+ +-----+ +-----+ +-----+
+ *
+ *
+ * While the llist state transitions from d to e, a kworker
+ * can start executing rcu_sr_normal_gp_cleanup_work() and
+ * can observe either the old done tail (@c) or the new
+ * done tail (@e). So, done tail updates and reads need
+ * to use the rel-acq semantics. If the concurrent kworker
+ * observes the old done tail, the newly queued work
+ * execution will process the updated done tail. If the
+ * concurrent kworker observes the new done tail, then
+ * the newly queued work will skip processing the done
+ * tail, as workqueue semantics guarantees that the new
+ * work is executed only after the previous one completes.
+ *
+ * f. kworker callbacks processing complete:
+ *
+ *
+ * DONE TAIL
+ * |
+ * |
+ * v
+ * +----------+ +--------+
+ * | | | |
+ * | head ------> wait |
+ * | | | head2 |
+ * +----------+ +--------+
+ *
*/
static struct sr_normal_state {
struct llist_head srs_next; /* request a GP users. */
- struct llist_head srs_wait; /* wait for GP users. */
- struct llist_head srs_done; /* ready for GP users. */
-
- /*
- * In order to add a batch of nodes to already
- * existing srs-done-list, a tail of srs-wait-list
- * is maintained.
- */
- struct llist_node *srs_wait_tail;
+ struct llist_node *srs_wait_tail; /* wait for GP users. */
+ struct llist_node *srs_done_tail; /* ready for GP users. */
+ struct sr_wait_node srs_wait_nodes[SR_NORMAL_GP_WAIT_HEAD_MAX];
} sr;

+static bool rcu_sr_is_wait_head(struct llist_node *node)
+{
+ return &(sr.srs_wait_nodes)[0].node <= node &&
+ node <= &(sr.srs_wait_nodes)[SR_NORMAL_GP_WAIT_HEAD_MAX - 1].node;
+}
+
+static struct llist_node *rcu_sr_get_wait_head(void)
+{
+ struct sr_wait_node *sr_wn;
+ int i;
+
+ for (i = 0; i < SR_NORMAL_GP_WAIT_HEAD_MAX; i++) {
+ sr_wn = &(sr.srs_wait_nodes)[i];
+
+ if (!atomic_cmpxchg_acquire(&sr_wn->inuse, 0, 1))
+ return &sr_wn->node;
+ }
+
+ return NULL;
+}
+
+static void rcu_sr_put_wait_head(struct llist_node *node)
+{
+ struct sr_wait_node *sr_wn = container_of(node, struct sr_wait_node, node);
+ atomic_set_release(&sr_wn->inuse, 0);
+}
+
/* Disabled by default. */
static int rcu_normal_wake_from_gp;
module_param(rcu_normal_wake_from_gp, int, 0644);
@@ -1423,14 +1571,44 @@ static void rcu_sr_normal_complete(struct llist_node *node)

static void rcu_sr_normal_gp_cleanup_work(struct work_struct *work)
{
- struct llist_node *done, *rcu, *next;
+ struct llist_node *done, *rcu, *next, *head;

- done = llist_del_all(&sr.srs_done);
+ /*
+ * This work execution can potentially execute
+ * while a new done tail is being updated by
+ * grace period kthread in rcu_sr_normal_gp_cleanup().
+ * So, read and updates of done tail need to
+ * follow acq-rel semantics.
+ *
+ * Given that wq semantics guarantees that a single work
+ * cannot execute concurrently by multiple kworkers,
+ * the done tail list manipulations are protected here.
+ */
+ done = smp_load_acquire(&sr.srs_done_tail);
if (!done)
return;

- llist_for_each_safe(rcu, next, done)
- rcu_sr_normal_complete(rcu);
+ WARN_ON_ONCE(!rcu_sr_is_wait_head(done));
+ head = done->next;
+ done->next = NULL;
+
+ /*
+ * The dummy node, which is pointed to by the
+ * done tail which is acq-read above is not removed
+ * here. This allows lockless additions of new
+ * rcu_synchronize nodes in rcu_sr_normal_add_req(),
+ * while the cleanup work executes. The dummy
+ * nodes is removed, in next round of cleanup
+ * work execution.
+ */
+ llist_for_each_safe(rcu, next, head) {
+ if (!rcu_sr_is_wait_head(rcu)) {
+ rcu_sr_normal_complete(rcu);
+ continue;
+ }
+
+ rcu_sr_put_wait_head(rcu);
+ }
}
static DECLARE_WORK(sr_normal_gp_cleanup, rcu_sr_normal_gp_cleanup_work);

@@ -1439,43 +1617,56 @@ static DECLARE_WORK(sr_normal_gp_cleanup, rcu_sr_normal_gp_cleanup_work);
*/
static void rcu_sr_normal_gp_cleanup(void)
{
- struct llist_node *head, *tail;
+ struct llist_node *wait_tail;

- if (llist_empty(&sr.srs_wait))
+ wait_tail = sr.srs_wait_tail;
+ if (wait_tail == NULL)
return;

- tail = READ_ONCE(sr.srs_wait_tail);
- head = __llist_del_all(&sr.srs_wait);
+ sr.srs_wait_tail = NULL;
+ ASSERT_EXCLUSIVE_WRITER(sr.srs_wait_tail);

- if (head) {
- /* Can be not empty. */
- llist_add_batch(head, tail, &sr.srs_done);
+ // concurrent sr_normal_gp_cleanup work might observe this update.
+ smp_store_release(&sr.srs_done_tail, wait_tail);
+ ASSERT_EXCLUSIVE_WRITER(sr.srs_done_tail);
+
+ if (wait_tail)
queue_work(system_highpri_wq, &sr_normal_gp_cleanup);
- }
}

/*
* Helper function for rcu_gp_init().
*/
-static void rcu_sr_normal_gp_init(void)
+static bool rcu_sr_normal_gp_init(void)
{
- struct llist_node *head, *tail;
+ struct llist_node *first;
+ struct llist_node *wait_head;
+ bool start_new_poll = false;

- if (llist_empty(&sr.srs_next))
- return;
+ first = READ_ONCE(sr.srs_next.first);
+ if (!first || rcu_sr_is_wait_head(first))
+ return start_new_poll;
+
+ wait_head = rcu_sr_get_wait_head();
+ if (!wait_head) {
+ // Kick another GP to retry.
+ start_new_poll = true;
+ return start_new_poll;
+ }

- tail = llist_del_all(&sr.srs_next);
- head = llist_reverse_order(tail);
+ /* Inject a wait-dummy-node. */
+ llist_add(wait_head, &sr.srs_next);

/*
- * A waiting list of GP should be empty on this step,
- * since a GP-kthread, rcu_gp_init() -> gp_cleanup(),
+ * A waiting list of rcu_synchronize nodes should be empty on
+ * this step, since a GP-kthread, rcu_gp_init() -> gp_cleanup(),
* rolls it over. If not, it is a BUG, warn a user.
*/
- WARN_ON_ONCE(!llist_empty(&sr.srs_wait));
+ WARN_ON_ONCE(sr.srs_wait_tail != NULL);
+ sr.srs_wait_tail = wait_head;
+ ASSERT_EXCLUSIVE_WRITER(sr.srs_wait_tail);

- WRITE_ONCE(sr.srs_wait_tail, tail);
- __llist_add_batch(head, tail, &sr.srs_wait);
+ return start_new_poll;
}

static void rcu_sr_normal_add_req(struct rcu_synchronize *rs)
@@ -1493,6 +1684,7 @@ static noinline_for_stack bool rcu_gp_init(void)
unsigned long mask;
struct rcu_data *rdp;
struct rcu_node *rnp = rcu_get_root();
+ bool start_new_poll;

WRITE_ONCE(rcu_state.gp_activity, jiffies);
raw_spin_lock_irq_rcu_node(rnp);
@@ -1517,11 +1709,15 @@ static noinline_for_stack bool rcu_gp_init(void)
/* Record GP times before starting GP, hence rcu_seq_start(). */
rcu_seq_start(&rcu_state.gp_seq);
ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
- rcu_sr_normal_gp_init();
+ start_new_poll = rcu_sr_normal_gp_init();
trace_rcu_grace_period(rcu_state.name, rcu_state.gp_seq, TPS("start"));
rcu_poll_gp_seq_start(&rcu_state.gp_seq_polled_snap);
raw_spin_unlock_irq_rcu_node(rnp);

+ // New poll request after rnp unlock
+ if (start_new_poll)
+ (void) start_poll_synchronize_rcu();
+
/*
* Apply per-leaf buffered online and offline operations to
* the rcu_node tree. Note that this new grace period need not
--
2.39.2


2023-12-20 01:38:07

by Paul E. McKenney

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Tue, Nov 28, 2023 at 09:00:30AM +0100, Uladzislau Rezki (Sony) wrote:
> From: Neeraj Upadhyay <[email protected]>
>
> Currently, processing of the next batch of rcu_synchronize nodes
> for the new grace period, requires doing a llist reversal operation
> to find the tail element of the list. This can be a very costly
> operation (high number of cache misses) for a long list.
>
> To address this, this patch introduces a "dummy-wait-node" entity.
> At every grace period init, a new wait node is added to the llist.
> This wait node is used as wait tail for this new grace period.
>
> This allows lockless additions of new rcu_synchronize nodes in the
> rcu_sr_normal_add_req(), while the cleanup work executes and does
> the progress. The dummy nodes are removed on next round of cleanup
> work execution.
>
> Signed-off-by: Uladzislau Rezki (Sony) <[email protected]>
> Signed-off-by: Neeraj Upadhyay <[email protected]>

This says that Uladzislau created the patch and that Neeraj
acted as maintainer. I am guessing that you both worked on it,
in which case is should have the Co-developed-by tags as shown in
Documentation/process/submitting-patches.rst. Could you please update
these to reflect the actual origin?

One question below toward the end. There are probably others that I
should be asking, but I have to start somewhere. ;-)

Thanx, Paul

> ---
> kernel/rcu/tree.c | 270 +++++++++++++++++++++++++++++++++++++++-------
> 1 file changed, 233 insertions(+), 37 deletions(-)
>
> diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> index 975621ef40e3..d7b48996825f 100644
> --- a/kernel/rcu/tree.c
> +++ b/kernel/rcu/tree.c
> @@ -1384,25 +1384,173 @@ static void rcu_poll_gp_seq_end_unlocked(unsigned long *snap)
> raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
> }
>
> +#define SR_NORMAL_GP_WAIT_HEAD_MAX 5
> +
> +struct sr_wait_node {
> + atomic_t inuse;
> + struct llist_node node;
> +};
> +
> /*
> - * There are three lists for handling synchronize_rcu() users.
> - * A first list corresponds to new coming users, second for users
> - * which wait for a grace period and third is for which a grace
> - * period is passed.
> + * There is a single llist, which is used for handling
> + * synchronize_rcu() users' enqueued rcu_synchronize nodes.
> + * Within this llist, there are two tail pointers:
> + *
> + * wait tail: Tracks the set of nodes, which need to
> + * wait for the current GP to complete.
> + * done tail: Tracks the set of nodes, for which grace
> + * period has elapsed. These nodes processing
> + * will be done as part of the cleanup work
> + * execution by a kworker.
> + *
> + * At every grace period init, a new wait node is added
> + * to the llist. This wait node is used as wait tail
> + * for this new grace period. Given that there are a fixed
> + * number of wait nodes, if all wait nodes are in use
> + * (which can happen when kworker callback processing
> + * is delayed) and additional grace period is requested.
> + * This means, a system is slow in processing callbacks.
> + *
> + * TODO: If a slow processing is detected, a first node
> + * in the llist should be used as a wait-tail for this
> + * grace period, therefore users which should wait due
> + * to a slow process are handled by _this_ grace period
> + * and not next.
> + *
> + * Below is an illustration of how the done and wait
> + * tail pointers move from one set of rcu_synchronize nodes
> + * to the other, as grace periods start and finish and
> + * nodes are processed by kworker.
> + *
> + *
> + * a. Initial llist callbacks list:
> + *
> + * +----------+ +--------+ +-------+
> + * | | | | | |
> + * | head |---------> | cb2 |--------->| cb1 |
> + * | | | | | |
> + * +----------+ +--------+ +-------+
> + *
> + *
> + *
> + * b. New GP1 Start:
> + *
> + * WAIT TAIL
> + * |
> + * |
> + * v
> + * +----------+ +--------+ +--------+ +-------+
> + * | | | | | | | |
> + * | head ------> wait |------> cb2 |------> | cb1 |
> + * | | | head1 | | | | |
> + * +----------+ +--------+ +--------+ +-------+
> + *
> + *
> + *
> + * c. GP completion:
> + *
> + * WAIT_TAIL == DONE_TAIL
> + *
> + * DONE TAIL
> + * |
> + * |
> + * v
> + * +----------+ +--------+ +--------+ +-------+
> + * | | | | | | | |
> + * | head ------> wait |------> cb2 |------> | cb1 |
> + * | | | head1 | | | | |
> + * +----------+ +--------+ +--------+ +-------+
> + *
> + *
> + *
> + * d. New callbacks and GP2 start:
> + *
> + * WAIT TAIL DONE TAIL
> + * | |
> + * | |
> + * v v
> + * +----------+ +------+ +------+ +------+ +-----+ +-----+ +-----+
> + * | | | | | | | | | | | | | |
> + * | head ------> wait |--->| cb4 |--->| cb3 |--->|wait |--->| cb2 |--->| cb1 |
> + * | | | head2| | | | | |head1| | | | |
> + * +----------+ +------+ +------+ +------+ +-----+ +-----+ +-----+
> + *
> + *
> + *
> + * e. GP2 completion:
> + *
> + * WAIT_TAIL == DONE_TAIL
> + * DONE TAIL
> + * |
> + * |
> + * v
> + * +----------+ +------+ +------+ +------+ +-----+ +-----+ +-----+
> + * | | | | | | | | | | | | | |
> + * | head ------> wait |--->| cb4 |--->| cb3 |--->|wait |--->| cb2 |--->| cb1 |
> + * | | | head2| | | | | |head1| | | | |
> + * +----------+ +------+ +------+ +------+ +-----+ +-----+ +-----+
> + *
> + *
> + * While the llist state transitions from d to e, a kworker
> + * can start executing rcu_sr_normal_gp_cleanup_work() and
> + * can observe either the old done tail (@c) or the new
> + * done tail (@e). So, done tail updates and reads need
> + * to use the rel-acq semantics. If the concurrent kworker
> + * observes the old done tail, the newly queued work
> + * execution will process the updated done tail. If the
> + * concurrent kworker observes the new done tail, then
> + * the newly queued work will skip processing the done
> + * tail, as workqueue semantics guarantees that the new
> + * work is executed only after the previous one completes.
> + *
> + * f. kworker callbacks processing complete:
> + *
> + *
> + * DONE TAIL
> + * |
> + * |
> + * v
> + * +----------+ +--------+
> + * | | | |
> + * | head ------> wait |
> + * | | | head2 |
> + * +----------+ +--------+
> + *
> */
> static struct sr_normal_state {
> struct llist_head srs_next; /* request a GP users. */
> - struct llist_head srs_wait; /* wait for GP users. */
> - struct llist_head srs_done; /* ready for GP users. */
> -
> - /*
> - * In order to add a batch of nodes to already
> - * existing srs-done-list, a tail of srs-wait-list
> - * is maintained.
> - */
> - struct llist_node *srs_wait_tail;
> + struct llist_node *srs_wait_tail; /* wait for GP users. */
> + struct llist_node *srs_done_tail; /* ready for GP users. */
> + struct sr_wait_node srs_wait_nodes[SR_NORMAL_GP_WAIT_HEAD_MAX];
> } sr;
>
> +static bool rcu_sr_is_wait_head(struct llist_node *node)
> +{
> + return &(sr.srs_wait_nodes)[0].node <= node &&
> + node <= &(sr.srs_wait_nodes)[SR_NORMAL_GP_WAIT_HEAD_MAX - 1].node;
> +}
> +
> +static struct llist_node *rcu_sr_get_wait_head(void)
> +{
> + struct sr_wait_node *sr_wn;
> + int i;
> +
> + for (i = 0; i < SR_NORMAL_GP_WAIT_HEAD_MAX; i++) {
> + sr_wn = &(sr.srs_wait_nodes)[i];
> +
> + if (!atomic_cmpxchg_acquire(&sr_wn->inuse, 0, 1))
> + return &sr_wn->node;
> + }
> +
> + return NULL;
> +}
> +
> +static void rcu_sr_put_wait_head(struct llist_node *node)
> +{
> + struct sr_wait_node *sr_wn = container_of(node, struct sr_wait_node, node);
> + atomic_set_release(&sr_wn->inuse, 0);
> +}
> +
> /* Disabled by default. */
> static int rcu_normal_wake_from_gp;
> module_param(rcu_normal_wake_from_gp, int, 0644);
> @@ -1423,14 +1571,44 @@ static void rcu_sr_normal_complete(struct llist_node *node)
>
> static void rcu_sr_normal_gp_cleanup_work(struct work_struct *work)
> {
> - struct llist_node *done, *rcu, *next;
> + struct llist_node *done, *rcu, *next, *head;
>
> - done = llist_del_all(&sr.srs_done);
> + /*
> + * This work execution can potentially execute
> + * while a new done tail is being updated by
> + * grace period kthread in rcu_sr_normal_gp_cleanup().
> + * So, read and updates of done tail need to
> + * follow acq-rel semantics.
> + *
> + * Given that wq semantics guarantees that a single work
> + * cannot execute concurrently by multiple kworkers,
> + * the done tail list manipulations are protected here.
> + */
> + done = smp_load_acquire(&sr.srs_done_tail);
> if (!done)
> return;
>
> - llist_for_each_safe(rcu, next, done)
> - rcu_sr_normal_complete(rcu);
> + WARN_ON_ONCE(!rcu_sr_is_wait_head(done));
> + head = done->next;
> + done->next = NULL;
> +
> + /*
> + * The dummy node, which is pointed to by the
> + * done tail which is acq-read above is not removed
> + * here. This allows lockless additions of new
> + * rcu_synchronize nodes in rcu_sr_normal_add_req(),
> + * while the cleanup work executes. The dummy
> + * nodes is removed, in next round of cleanup
> + * work execution.
> + */
> + llist_for_each_safe(rcu, next, head) {
> + if (!rcu_sr_is_wait_head(rcu)) {
> + rcu_sr_normal_complete(rcu);
> + continue;
> + }
> +
> + rcu_sr_put_wait_head(rcu);
> + }
> }
> static DECLARE_WORK(sr_normal_gp_cleanup, rcu_sr_normal_gp_cleanup_work);
>
> @@ -1439,43 +1617,56 @@ static DECLARE_WORK(sr_normal_gp_cleanup, rcu_sr_normal_gp_cleanup_work);
> */
> static void rcu_sr_normal_gp_cleanup(void)
> {
> - struct llist_node *head, *tail;
> + struct llist_node *wait_tail;
>
> - if (llist_empty(&sr.srs_wait))
> + wait_tail = sr.srs_wait_tail;
> + if (wait_tail == NULL)
> return;
>
> - tail = READ_ONCE(sr.srs_wait_tail);
> - head = __llist_del_all(&sr.srs_wait);
> + sr.srs_wait_tail = NULL;
> + ASSERT_EXCLUSIVE_WRITER(sr.srs_wait_tail);
>
> - if (head) {
> - /* Can be not empty. */
> - llist_add_batch(head, tail, &sr.srs_done);
> + // concurrent sr_normal_gp_cleanup work might observe this update.
> + smp_store_release(&sr.srs_done_tail, wait_tail);
> + ASSERT_EXCLUSIVE_WRITER(sr.srs_done_tail);
> +
> + if (wait_tail)
> queue_work(system_highpri_wq, &sr_normal_gp_cleanup);
> - }
> }
>
> /*
> * Helper function for rcu_gp_init().
> */
> -static void rcu_sr_normal_gp_init(void)
> +static bool rcu_sr_normal_gp_init(void)
> {
> - struct llist_node *head, *tail;
> + struct llist_node *first;
> + struct llist_node *wait_head;
> + bool start_new_poll = false;
>
> - if (llist_empty(&sr.srs_next))
> - return;
> + first = READ_ONCE(sr.srs_next.first);
> + if (!first || rcu_sr_is_wait_head(first))
> + return start_new_poll;
> +
> + wait_head = rcu_sr_get_wait_head();
> + if (!wait_head) {
> + // Kick another GP to retry.
> + start_new_poll = true;
> + return start_new_poll;
> + }
>
> - tail = llist_del_all(&sr.srs_next);
> - head = llist_reverse_order(tail);
> + /* Inject a wait-dummy-node. */
> + llist_add(wait_head, &sr.srs_next);
>
> /*
> - * A waiting list of GP should be empty on this step,
> - * since a GP-kthread, rcu_gp_init() -> gp_cleanup(),
> + * A waiting list of rcu_synchronize nodes should be empty on
> + * this step, since a GP-kthread, rcu_gp_init() -> gp_cleanup(),
> * rolls it over. If not, it is a BUG, warn a user.
> */
> - WARN_ON_ONCE(!llist_empty(&sr.srs_wait));
> + WARN_ON_ONCE(sr.srs_wait_tail != NULL);
> + sr.srs_wait_tail = wait_head;
> + ASSERT_EXCLUSIVE_WRITER(sr.srs_wait_tail);
>
> - WRITE_ONCE(sr.srs_wait_tail, tail);
> - __llist_add_batch(head, tail, &sr.srs_wait);
> + return start_new_poll;
> }
>
> static void rcu_sr_normal_add_req(struct rcu_synchronize *rs)
> @@ -1493,6 +1684,7 @@ static noinline_for_stack bool rcu_gp_init(void)
> unsigned long mask;
> struct rcu_data *rdp;
> struct rcu_node *rnp = rcu_get_root();
> + bool start_new_poll;
>
> WRITE_ONCE(rcu_state.gp_activity, jiffies);
> raw_spin_lock_irq_rcu_node(rnp);
> @@ -1517,11 +1709,15 @@ static noinline_for_stack bool rcu_gp_init(void)
> /* Record GP times before starting GP, hence rcu_seq_start(). */
> rcu_seq_start(&rcu_state.gp_seq);
> ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> - rcu_sr_normal_gp_init();
> + start_new_poll = rcu_sr_normal_gp_init();
> trace_rcu_grace_period(rcu_state.name, rcu_state.gp_seq, TPS("start"));
> rcu_poll_gp_seq_start(&rcu_state.gp_seq_polled_snap);
> raw_spin_unlock_irq_rcu_node(rnp);
>
> + // New poll request after rnp unlock
> + if (start_new_poll)
> + (void) start_poll_synchronize_rcu();

You lost me on this one. Anything that got moved to the wait list
should be handled by the current grace period, right? Or is the
problem that rcu_sr_normal_gp_init() is being invoked after the call
to rcu_seq_start()? If that is the case, could it be moved ahead so
that we don't need the extra grace period?

Or am I missing something subtle here?

> +
> /*
> * Apply per-leaf buffered online and offline operations to
> * the rcu_node tree. Note that this new grace period need not
> --
> 2.39.2
>

2023-12-21 10:53:01

by Uladzislau Rezki

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Tue, Dec 19, 2023 at 05:37:56PM -0800, Paul E. McKenney wrote:
> On Tue, Nov 28, 2023 at 09:00:30AM +0100, Uladzislau Rezki (Sony) wrote:
> > From: Neeraj Upadhyay <[email protected]>
> >
> > Currently, processing of the next batch of rcu_synchronize nodes
> > for the new grace period, requires doing a llist reversal operation
> > to find the tail element of the list. This can be a very costly
> > operation (high number of cache misses) for a long list.
> >
> > To address this, this patch introduces a "dummy-wait-node" entity.
> > At every grace period init, a new wait node is added to the llist.
> > This wait node is used as wait tail for this new grace period.
> >
> > This allows lockless additions of new rcu_synchronize nodes in the
> > rcu_sr_normal_add_req(), while the cleanup work executes and does
> > the progress. The dummy nodes are removed on next round of cleanup
> > work execution.
> >
> > Signed-off-by: Uladzislau Rezki (Sony) <[email protected]>
> > Signed-off-by: Neeraj Upadhyay <[email protected]>
>
> This says that Uladzislau created the patch and that Neeraj
> acted as maintainer. I am guessing that you both worked on it,
> in which case is should have the Co-developed-by tags as shown in
> Documentation/process/submitting-patches.rst. Could you please update
> these to reflect the actual origin?
>
Right. We both worked on it. Neeraj is an author whereas i should mark
myself as a Co-developed-by. This is a correct way. Thank you for
pointing on it!

>
> One question below toward the end. There are probably others that I
> should be asking, but I have to start somewhere. ;-)
>
Good :)

> >
> > /*
> > * Helper function for rcu_gp_init().
> > */
> > -static void rcu_sr_normal_gp_init(void)
> > +static bool rcu_sr_normal_gp_init(void)
> > {
> > - struct llist_node *head, *tail;
> > + struct llist_node *first;
> > + struct llist_node *wait_head;
> > + bool start_new_poll = false;
> >
> > - if (llist_empty(&sr.srs_next))
> > - return;
> > + first = READ_ONCE(sr.srs_next.first);
> > + if (!first || rcu_sr_is_wait_head(first))
> > + return start_new_poll;
> > +
> > + wait_head = rcu_sr_get_wait_head();
> > + if (!wait_head) {
> > + // Kick another GP to retry.
> > + start_new_poll = true;
> > + return start_new_poll;
> > + }
> >
> > - tail = llist_del_all(&sr.srs_next);
> > - head = llist_reverse_order(tail);
> > + /* Inject a wait-dummy-node. */
> > + llist_add(wait_head, &sr.srs_next);
> >
> > /*
> > - * A waiting list of GP should be empty on this step,
> > - * since a GP-kthread, rcu_gp_init() -> gp_cleanup(),
> > + * A waiting list of rcu_synchronize nodes should be empty on
> > + * this step, since a GP-kthread, rcu_gp_init() -> gp_cleanup(),
> > * rolls it over. If not, it is a BUG, warn a user.
> > */
> > - WARN_ON_ONCE(!llist_empty(&sr.srs_wait));
> > + WARN_ON_ONCE(sr.srs_wait_tail != NULL);
> > + sr.srs_wait_tail = wait_head;
> > + ASSERT_EXCLUSIVE_WRITER(sr.srs_wait_tail);
> >
> > - WRITE_ONCE(sr.srs_wait_tail, tail);
> > - __llist_add_batch(head, tail, &sr.srs_wait);
> > + return start_new_poll;
> > }
> >
> > static void rcu_sr_normal_add_req(struct rcu_synchronize *rs)
> > @@ -1493,6 +1684,7 @@ static noinline_for_stack bool rcu_gp_init(void)
> > unsigned long mask;
> > struct rcu_data *rdp;
> > struct rcu_node *rnp = rcu_get_root();
> > + bool start_new_poll;
> >
> > WRITE_ONCE(rcu_state.gp_activity, jiffies);
> > raw_spin_lock_irq_rcu_node(rnp);
> > @@ -1517,11 +1709,15 @@ static noinline_for_stack bool rcu_gp_init(void)
> > /* Record GP times before starting GP, hence rcu_seq_start(). */
> > rcu_seq_start(&rcu_state.gp_seq);
> > ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> > - rcu_sr_normal_gp_init();
> > + start_new_poll = rcu_sr_normal_gp_init();
> > trace_rcu_grace_period(rcu_state.name, rcu_state.gp_seq, TPS("start"));
> > rcu_poll_gp_seq_start(&rcu_state.gp_seq_polled_snap);
> > raw_spin_unlock_irq_rcu_node(rnp);
> >
> > + // New poll request after rnp unlock
> > + if (start_new_poll)
> > + (void) start_poll_synchronize_rcu();
>
> You lost me on this one. Anything that got moved to the wait list
> should be handled by the current grace period, right? Or is the
> problem that rcu_sr_normal_gp_init() is being invoked after the call
> to rcu_seq_start()? If that is the case, could it be moved ahead so
> that we don't need the extra grace period?
>
> Or am I missing something subtle here?
>
The problem is that, we are limited in number of "wait-heads" which we
add as a marker node for this/current grace period. If there are more clients
and there is no a wait-head available it means that a system, the deferred
kworker, is slow in processing callbacks, thus all wait-nodes are in use.

That is why we need an extra grace period. Basically to repeat our try one
more time, i.e. it might be that a current grace period is not able to handle
users due to the fact that a system is doing really slow, but this is rather
a corner case and is not a problem.

--
Uladzislau Rezki

2023-12-21 18:40:32

by Paul E. McKenney

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Thu, Dec 21, 2023 at 11:52:33AM +0100, Uladzislau Rezki wrote:
> On Tue, Dec 19, 2023 at 05:37:56PM -0800, Paul E. McKenney wrote:
> > On Tue, Nov 28, 2023 at 09:00:30AM +0100, Uladzislau Rezki (Sony) wrote:
> > > From: Neeraj Upadhyay <[email protected]>
> > >
> > > Currently, processing of the next batch of rcu_synchronize nodes
> > > for the new grace period, requires doing a llist reversal operation
> > > to find the tail element of the list. This can be a very costly
> > > operation (high number of cache misses) for a long list.
> > >
> > > To address this, this patch introduces a "dummy-wait-node" entity.
> > > At every grace period init, a new wait node is added to the llist.
> > > This wait node is used as wait tail for this new grace period.
> > >
> > > This allows lockless additions of new rcu_synchronize nodes in the
> > > rcu_sr_normal_add_req(), while the cleanup work executes and does
> > > the progress. The dummy nodes are removed on next round of cleanup
> > > work execution.
> > >
> > > Signed-off-by: Uladzislau Rezki (Sony) <[email protected]>
> > > Signed-off-by: Neeraj Upadhyay <[email protected]>
> >
> > This says that Uladzislau created the patch and that Neeraj
> > acted as maintainer. I am guessing that you both worked on it,
> > in which case is should have the Co-developed-by tags as shown in
> > Documentation/process/submitting-patches.rst. Could you please update
> > these to reflect the actual origin?
> >
> Right. We both worked on it. Neeraj is an author whereas i should mark
> myself as a Co-developed-by. This is a correct way. Thank you for
> pointing on it!

Sounds good, thank you!

> > One question below toward the end. There are probably others that I
> > should be asking, but I have to start somewhere. ;-)
> >
> Good :)
>
> > >
> > > /*
> > > * Helper function for rcu_gp_init().
> > > */
> > > -static void rcu_sr_normal_gp_init(void)
> > > +static bool rcu_sr_normal_gp_init(void)
> > > {
> > > - struct llist_node *head, *tail;
> > > + struct llist_node *first;
> > > + struct llist_node *wait_head;
> > > + bool start_new_poll = false;
> > >
> > > - if (llist_empty(&sr.srs_next))
> > > - return;
> > > + first = READ_ONCE(sr.srs_next.first);
> > > + if (!first || rcu_sr_is_wait_head(first))
> > > + return start_new_poll;
> > > +
> > > + wait_head = rcu_sr_get_wait_head();
> > > + if (!wait_head) {
> > > + // Kick another GP to retry.
> > > + start_new_poll = true;
> > > + return start_new_poll;
> > > + }
> > >
> > > - tail = llist_del_all(&sr.srs_next);
> > > - head = llist_reverse_order(tail);
> > > + /* Inject a wait-dummy-node. */
> > > + llist_add(wait_head, &sr.srs_next);
> > >
> > > /*
> > > - * A waiting list of GP should be empty on this step,
> > > - * since a GP-kthread, rcu_gp_init() -> gp_cleanup(),
> > > + * A waiting list of rcu_synchronize nodes should be empty on
> > > + * this step, since a GP-kthread, rcu_gp_init() -> gp_cleanup(),
> > > * rolls it over. If not, it is a BUG, warn a user.
> > > */
> > > - WARN_ON_ONCE(!llist_empty(&sr.srs_wait));
> > > + WARN_ON_ONCE(sr.srs_wait_tail != NULL);
> > > + sr.srs_wait_tail = wait_head;
> > > + ASSERT_EXCLUSIVE_WRITER(sr.srs_wait_tail);
> > >
> > > - WRITE_ONCE(sr.srs_wait_tail, tail);
> > > - __llist_add_batch(head, tail, &sr.srs_wait);
> > > + return start_new_poll;
> > > }
> > >
> > > static void rcu_sr_normal_add_req(struct rcu_synchronize *rs)
> > > @@ -1493,6 +1684,7 @@ static noinline_for_stack bool rcu_gp_init(void)
> > > unsigned long mask;
> > > struct rcu_data *rdp;
> > > struct rcu_node *rnp = rcu_get_root();
> > > + bool start_new_poll;
> > >
> > > WRITE_ONCE(rcu_state.gp_activity, jiffies);
> > > raw_spin_lock_irq_rcu_node(rnp);
> > > @@ -1517,11 +1709,15 @@ static noinline_for_stack bool rcu_gp_init(void)
> > > /* Record GP times before starting GP, hence rcu_seq_start(). */
> > > rcu_seq_start(&rcu_state.gp_seq);
> > > ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> > > - rcu_sr_normal_gp_init();
> > > + start_new_poll = rcu_sr_normal_gp_init();
> > > trace_rcu_grace_period(rcu_state.name, rcu_state.gp_seq, TPS("start"));
> > > rcu_poll_gp_seq_start(&rcu_state.gp_seq_polled_snap);
> > > raw_spin_unlock_irq_rcu_node(rnp);
> > >
> > > + // New poll request after rnp unlock
> > > + if (start_new_poll)
> > > + (void) start_poll_synchronize_rcu();
> >
> > You lost me on this one. Anything that got moved to the wait list
> > should be handled by the current grace period, right? Or is the
> > problem that rcu_sr_normal_gp_init() is being invoked after the call
> > to rcu_seq_start()? If that is the case, could it be moved ahead so
> > that we don't need the extra grace period?
> >
> > Or am I missing something subtle here?
> >
> The problem is that, we are limited in number of "wait-heads" which we
> add as a marker node for this/current grace period. If there are more clients
> and there is no a wait-head available it means that a system, the deferred
> kworker, is slow in processing callbacks, thus all wait-nodes are in use.
>
> That is why we need an extra grace period. Basically to repeat our try one
> more time, i.e. it might be that a current grace period is not able to handle
> users due to the fact that a system is doing really slow, but this is rather
> a corner case and is not a problem.

But in that case, the real issue is not the need for an extra grace
period, but rather the need for the wakeup processing to happen, correct?
Or am I missing something subtle here?

Thanx, Paul

2023-12-22 09:27:58

by Uladzislau Rezki

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Thu, Dec 21, 2023 at 10:40:21AM -0800, Paul E. McKenney wrote:
> On Thu, Dec 21, 2023 at 11:52:33AM +0100, Uladzislau Rezki wrote:
> > On Tue, Dec 19, 2023 at 05:37:56PM -0800, Paul E. McKenney wrote:
> > > On Tue, Nov 28, 2023 at 09:00:30AM +0100, Uladzislau Rezki (Sony) wrote:
> > > > From: Neeraj Upadhyay <[email protected]>
> > > >
> > > > Currently, processing of the next batch of rcu_synchronize nodes
> > > > for the new grace period, requires doing a llist reversal operation
> > > > to find the tail element of the list. This can be a very costly
> > > > operation (high number of cache misses) for a long list.
> > > >
> > > > To address this, this patch introduces a "dummy-wait-node" entity.
> > > > At every grace period init, a new wait node is added to the llist.
> > > > This wait node is used as wait tail for this new grace period.
> > > >
> > > > This allows lockless additions of new rcu_synchronize nodes in the
> > > > rcu_sr_normal_add_req(), while the cleanup work executes and does
> > > > the progress. The dummy nodes are removed on next round of cleanup
> > > > work execution.
> > > >
> > > > Signed-off-by: Uladzislau Rezki (Sony) <[email protected]>
> > > > Signed-off-by: Neeraj Upadhyay <[email protected]>
> > >
> > > This says that Uladzislau created the patch and that Neeraj
> > > acted as maintainer. I am guessing that you both worked on it,
> > > in which case is should have the Co-developed-by tags as shown in
> > > Documentation/process/submitting-patches.rst. Could you please update
> > > these to reflect the actual origin?
> > >
> > Right. We both worked on it. Neeraj is an author whereas i should mark
> > myself as a Co-developed-by. This is a correct way. Thank you for
> > pointing on it!
>
> Sounds good, thank you!
>
> > > One question below toward the end. There are probably others that I
> > > should be asking, but I have to start somewhere. ;-)
> > >
> > Good :)
> >
> > > >
> > > > /*
> > > > * Helper function for rcu_gp_init().
> > > > */
> > > > -static void rcu_sr_normal_gp_init(void)
> > > > +static bool rcu_sr_normal_gp_init(void)
> > > > {
> > > > - struct llist_node *head, *tail;
> > > > + struct llist_node *first;
> > > > + struct llist_node *wait_head;
> > > > + bool start_new_poll = false;
> > > >
> > > > - if (llist_empty(&sr.srs_next))
> > > > - return;
> > > > + first = READ_ONCE(sr.srs_next.first);
> > > > + if (!first || rcu_sr_is_wait_head(first))
> > > > + return start_new_poll;
> > > > +
> > > > + wait_head = rcu_sr_get_wait_head();
> > > > + if (!wait_head) {
> > > > + // Kick another GP to retry.
> > > > + start_new_poll = true;
> > > > + return start_new_poll;
> > > > + }
> > > >
> > > > - tail = llist_del_all(&sr.srs_next);
> > > > - head = llist_reverse_order(tail);
> > > > + /* Inject a wait-dummy-node. */
> > > > + llist_add(wait_head, &sr.srs_next);
> > > >
> > > > /*
> > > > - * A waiting list of GP should be empty on this step,
> > > > - * since a GP-kthread, rcu_gp_init() -> gp_cleanup(),
> > > > + * A waiting list of rcu_synchronize nodes should be empty on
> > > > + * this step, since a GP-kthread, rcu_gp_init() -> gp_cleanup(),
> > > > * rolls it over. If not, it is a BUG, warn a user.
> > > > */
> > > > - WARN_ON_ONCE(!llist_empty(&sr.srs_wait));
> > > > + WARN_ON_ONCE(sr.srs_wait_tail != NULL);
> > > > + sr.srs_wait_tail = wait_head;
> > > > + ASSERT_EXCLUSIVE_WRITER(sr.srs_wait_tail);
> > > >
> > > > - WRITE_ONCE(sr.srs_wait_tail, tail);
> > > > - __llist_add_batch(head, tail, &sr.srs_wait);
> > > > + return start_new_poll;
> > > > }
> > > >
> > > > static void rcu_sr_normal_add_req(struct rcu_synchronize *rs)
> > > > @@ -1493,6 +1684,7 @@ static noinline_for_stack bool rcu_gp_init(void)
> > > > unsigned long mask;
> > > > struct rcu_data *rdp;
> > > > struct rcu_node *rnp = rcu_get_root();
> > > > + bool start_new_poll;
> > > >
> > > > WRITE_ONCE(rcu_state.gp_activity, jiffies);
> > > > raw_spin_lock_irq_rcu_node(rnp);
> > > > @@ -1517,11 +1709,15 @@ static noinline_for_stack bool rcu_gp_init(void)
> > > > /* Record GP times before starting GP, hence rcu_seq_start(). */
> > > > rcu_seq_start(&rcu_state.gp_seq);
> > > > ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> > > > - rcu_sr_normal_gp_init();
> > > > + start_new_poll = rcu_sr_normal_gp_init();
> > > > trace_rcu_grace_period(rcu_state.name, rcu_state.gp_seq, TPS("start"));
> > > > rcu_poll_gp_seq_start(&rcu_state.gp_seq_polled_snap);
> > > > raw_spin_unlock_irq_rcu_node(rnp);
> > > >
> > > > + // New poll request after rnp unlock
> > > > + if (start_new_poll)
> > > > + (void) start_poll_synchronize_rcu();
> > >
> > > You lost me on this one. Anything that got moved to the wait list
> > > should be handled by the current grace period, right? Or is the
> > > problem that rcu_sr_normal_gp_init() is being invoked after the call
> > > to rcu_seq_start()? If that is the case, could it be moved ahead so
> > > that we don't need the extra grace period?
> > >
> > > Or am I missing something subtle here?
> > >
> > The problem is that, we are limited in number of "wait-heads" which we
> > add as a marker node for this/current grace period. If there are more clients
> > and there is no a wait-head available it means that a system, the deferred
> > kworker, is slow in processing callbacks, thus all wait-nodes are in use.
> >
> > That is why we need an extra grace period. Basically to repeat our try one
> > more time, i.e. it might be that a current grace period is not able to handle
> > users due to the fact that a system is doing really slow, but this is rather
> > a corner case and is not a problem.
>
> But in that case, the real issue is not the need for an extra grace
> period, but rather the need for the wakeup processing to happen, correct?
> Or am I missing something subtle here?
>
Basically, yes. If we had a spare dummy-node we could process the users
by the current GP(no need in extra). Why we may not have it - it is because
like you pointed:

- wake-up issue, i.e. wake-up time + when we are on_cpu;
- slow list process. For example priority. The kworker is not
given enough CPU time to do the progress, thus "dummy-nodes"
are not released in time for reuse.

Therefore, en extra GP is requested if there is a high flow of
synchronize_rcu() users and kworker is not able to do a progress
in time.

For example 60K+ parallel synchronize_rcu() users will trigger it.

--
Uladzislau Rezki

2023-12-22 18:58:43

by Paul E. McKenney

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Fri, Dec 22, 2023 at 10:27:41AM +0100, Uladzislau Rezki wrote:
> On Thu, Dec 21, 2023 at 10:40:21AM -0800, Paul E. McKenney wrote:
> > On Thu, Dec 21, 2023 at 11:52:33AM +0100, Uladzislau Rezki wrote:
> > > On Tue, Dec 19, 2023 at 05:37:56PM -0800, Paul E. McKenney wrote:
> > > > On Tue, Nov 28, 2023 at 09:00:30AM +0100, Uladzislau Rezki (Sony) wrote:
> > > > > From: Neeraj Upadhyay <[email protected]>
> > > > >
> > > > > Currently, processing of the next batch of rcu_synchronize nodes
> > > > > for the new grace period, requires doing a llist reversal operation
> > > > > to find the tail element of the list. This can be a very costly
> > > > > operation (high number of cache misses) for a long list.
> > > > >
> > > > > To address this, this patch introduces a "dummy-wait-node" entity.
> > > > > At every grace period init, a new wait node is added to the llist.
> > > > > This wait node is used as wait tail for this new grace period.
> > > > >
> > > > > This allows lockless additions of new rcu_synchronize nodes in the
> > > > > rcu_sr_normal_add_req(), while the cleanup work executes and does
> > > > > the progress. The dummy nodes are removed on next round of cleanup
> > > > > work execution.
> > > > >
> > > > > Signed-off-by: Uladzislau Rezki (Sony) <[email protected]>
> > > > > Signed-off-by: Neeraj Upadhyay <[email protected]>
> > > >
> > > > This says that Uladzislau created the patch and that Neeraj
> > > > acted as maintainer. I am guessing that you both worked on it,
> > > > in which case is should have the Co-developed-by tags as shown in
> > > > Documentation/process/submitting-patches.rst. Could you please update
> > > > these to reflect the actual origin?
> > > >
> > > Right. We both worked on it. Neeraj is an author whereas i should mark
> > > myself as a Co-developed-by. This is a correct way. Thank you for
> > > pointing on it!
> >
> > Sounds good, thank you!
> >
> > > > One question below toward the end. There are probably others that I
> > > > should be asking, but I have to start somewhere. ;-)
> > > >
> > > Good :)
> > >
> > > > >
> > > > > /*
> > > > > * Helper function for rcu_gp_init().
> > > > > */
> > > > > -static void rcu_sr_normal_gp_init(void)
> > > > > +static bool rcu_sr_normal_gp_init(void)
> > > > > {
> > > > > - struct llist_node *head, *tail;
> > > > > + struct llist_node *first;
> > > > > + struct llist_node *wait_head;
> > > > > + bool start_new_poll = false;
> > > > >
> > > > > - if (llist_empty(&sr.srs_next))
> > > > > - return;
> > > > > + first = READ_ONCE(sr.srs_next.first);
> > > > > + if (!first || rcu_sr_is_wait_head(first))
> > > > > + return start_new_poll;
> > > > > +
> > > > > + wait_head = rcu_sr_get_wait_head();
> > > > > + if (!wait_head) {
> > > > > + // Kick another GP to retry.
> > > > > + start_new_poll = true;
> > > > > + return start_new_poll;
> > > > > + }
> > > > >
> > > > > - tail = llist_del_all(&sr.srs_next);
> > > > > - head = llist_reverse_order(tail);
> > > > > + /* Inject a wait-dummy-node. */
> > > > > + llist_add(wait_head, &sr.srs_next);
> > > > >
> > > > > /*
> > > > > - * A waiting list of GP should be empty on this step,
> > > > > - * since a GP-kthread, rcu_gp_init() -> gp_cleanup(),
> > > > > + * A waiting list of rcu_synchronize nodes should be empty on
> > > > > + * this step, since a GP-kthread, rcu_gp_init() -> gp_cleanup(),
> > > > > * rolls it over. If not, it is a BUG, warn a user.
> > > > > */
> > > > > - WARN_ON_ONCE(!llist_empty(&sr.srs_wait));
> > > > > + WARN_ON_ONCE(sr.srs_wait_tail != NULL);
> > > > > + sr.srs_wait_tail = wait_head;
> > > > > + ASSERT_EXCLUSIVE_WRITER(sr.srs_wait_tail);
> > > > >
> > > > > - WRITE_ONCE(sr.srs_wait_tail, tail);
> > > > > - __llist_add_batch(head, tail, &sr.srs_wait);
> > > > > + return start_new_poll;
> > > > > }
> > > > >
> > > > > static void rcu_sr_normal_add_req(struct rcu_synchronize *rs)
> > > > > @@ -1493,6 +1684,7 @@ static noinline_for_stack bool rcu_gp_init(void)
> > > > > unsigned long mask;
> > > > > struct rcu_data *rdp;
> > > > > struct rcu_node *rnp = rcu_get_root();
> > > > > + bool start_new_poll;
> > > > >
> > > > > WRITE_ONCE(rcu_state.gp_activity, jiffies);
> > > > > raw_spin_lock_irq_rcu_node(rnp);
> > > > > @@ -1517,11 +1709,15 @@ static noinline_for_stack bool rcu_gp_init(void)
> > > > > /* Record GP times before starting GP, hence rcu_seq_start(). */
> > > > > rcu_seq_start(&rcu_state.gp_seq);
> > > > > ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> > > > > - rcu_sr_normal_gp_init();
> > > > > + start_new_poll = rcu_sr_normal_gp_init();
> > > > > trace_rcu_grace_period(rcu_state.name, rcu_state.gp_seq, TPS("start"));
> > > > > rcu_poll_gp_seq_start(&rcu_state.gp_seq_polled_snap);
> > > > > raw_spin_unlock_irq_rcu_node(rnp);
> > > > >
> > > > > + // New poll request after rnp unlock
> > > > > + if (start_new_poll)
> > > > > + (void) start_poll_synchronize_rcu();
> > > >
> > > > You lost me on this one. Anything that got moved to the wait list
> > > > should be handled by the current grace period, right? Or is the
> > > > problem that rcu_sr_normal_gp_init() is being invoked after the call
> > > > to rcu_seq_start()? If that is the case, could it be moved ahead so
> > > > that we don't need the extra grace period?
> > > >
> > > > Or am I missing something subtle here?
> > > >
> > > The problem is that, we are limited in number of "wait-heads" which we
> > > add as a marker node for this/current grace period. If there are more clients
> > > and there is no a wait-head available it means that a system, the deferred
> > > kworker, is slow in processing callbacks, thus all wait-nodes are in use.
> > >
> > > That is why we need an extra grace period. Basically to repeat our try one
> > > more time, i.e. it might be that a current grace period is not able to handle
> > > users due to the fact that a system is doing really slow, but this is rather
> > > a corner case and is not a problem.
> >
> > But in that case, the real issue is not the need for an extra grace
> > period, but rather the need for the wakeup processing to happen, correct?
> > Or am I missing something subtle here?
> >
> Basically, yes. If we had a spare dummy-node we could process the users
> by the current GP(no need in extra). Why we may not have it - it is because
> like you pointed:
>
> - wake-up issue, i.e. wake-up time + when we are on_cpu;
> - slow list process. For example priority. The kworker is not
> given enough CPU time to do the progress, thus "dummy-nodes"
> are not released in time for reuse.
>
> Therefore, en extra GP is requested if there is a high flow of
> synchronize_rcu() users and kworker is not able to do a progress
> in time.
>
> For example 60K+ parallel synchronize_rcu() users will trigger it.

OK, but what bad thing would happen if that was moved to precede the
rcu_seq_start(&rcu_state.gp_seq)? That way, the requested grace period
would be the same as the one that is just now starting.

Something like this?

start_new_poll = rcu_sr_normal_gp_init();

/* Record GP times before starting GP, hence rcu_seq_start(). */
rcu_seq_start(&rcu_state.gp_seq);
ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);

trace_rcu_grace_period(rcu_state.name, rcu_state.gp_seq, TPS("start"));
rcu_poll_gp_seq_start(&rcu_state.gp_seq_polled_snap);
raw_spin_unlock_irq_rcu_node(rnp);

// New poll request after rnp unlock
if (start_new_poll)
(void) start_poll_synchronize_rcu();

Yes, rcu_sr_normal_gp_init() might need some adjustment given that it
is seeing the pre-GP value of rcu_state.gp_seq.

But unless I am missing something, what you have now can result in
extra grace periods, which incur overhead on what would otherwise be an
idle system.

Thanx, Paul

2024-01-02 12:52:51

by Uladzislau Rezki

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

Hello, Paul!

Sorry for late answer, it is because of holidays :)

> > > > The problem is that, we are limited in number of "wait-heads" which we
> > > > add as a marker node for this/current grace period. If there are more clients
> > > > and there is no a wait-head available it means that a system, the deferred
> > > > kworker, is slow in processing callbacks, thus all wait-nodes are in use.
> > > >
> > > > That is why we need an extra grace period. Basically to repeat our try one
> > > > more time, i.e. it might be that a current grace period is not able to handle
> > > > users due to the fact that a system is doing really slow, but this is rather
> > > > a corner case and is not a problem.
> > >
> > > But in that case, the real issue is not the need for an extra grace
> > > period, but rather the need for the wakeup processing to happen, correct?
> > > Or am I missing something subtle here?
> > >
> > Basically, yes. If we had a spare dummy-node we could process the users
> > by the current GP(no need in extra). Why we may not have it - it is because
> > like you pointed:
> >
> > - wake-up issue, i.e. wake-up time + when we are on_cpu;
> > - slow list process. For example priority. The kworker is not
> > given enough CPU time to do the progress, thus "dummy-nodes"
> > are not released in time for reuse.
> >
> > Therefore, en extra GP is requested if there is a high flow of
> > synchronize_rcu() users and kworker is not able to do a progress
> > in time.
> >
> > For example 60K+ parallel synchronize_rcu() users will trigger it.
>
> OK, but what bad thing would happen if that was moved to precede the
> rcu_seq_start(&rcu_state.gp_seq)? That way, the requested grace period
> would be the same as the one that is just now starting.
>
> Something like this?
>
> start_new_poll = rcu_sr_normal_gp_init();
>
> /* Record GP times before starting GP, hence rcu_seq_start(). */
> rcu_seq_start(&rcu_state.gp_seq);
> ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
>
I had a concern about the case when rcu_sr_normal_gp_init() handles what
we currently have, in terms of requests. Right after that there is/are
extra sync requests which invoke the start_poll_synchronize_rcu() but
since a GP has been requested before it will not request an extra one. So
"last" incoming users might not be processed.

That is why i have placed the rcu_sr_normal_gp_init() after a gp_seq is
updated.

I can miss something, so please comment. Apart of that we can move it
as you proposed.

--
Uladzislau Rezki

2024-01-02 19:25:22

by Paul E. McKenney

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Tue, Jan 02, 2024 at 01:52:26PM +0100, Uladzislau Rezki wrote:
> Hello, Paul!
>
> Sorry for late answer, it is because of holidays :)
>
> > > > > The problem is that, we are limited in number of "wait-heads" which we
> > > > > add as a marker node for this/current grace period. If there are more clients
> > > > > and there is no a wait-head available it means that a system, the deferred
> > > > > kworker, is slow in processing callbacks, thus all wait-nodes are in use.
> > > > >
> > > > > That is why we need an extra grace period. Basically to repeat our try one
> > > > > more time, i.e. it might be that a current grace period is not able to handle
> > > > > users due to the fact that a system is doing really slow, but this is rather
> > > > > a corner case and is not a problem.
> > > >
> > > > But in that case, the real issue is not the need for an extra grace
> > > > period, but rather the need for the wakeup processing to happen, correct?
> > > > Or am I missing something subtle here?
> > > >
> > > Basically, yes. If we had a spare dummy-node we could process the users
> > > by the current GP(no need in extra). Why we may not have it - it is because
> > > like you pointed:
> > >
> > > - wake-up issue, i.e. wake-up time + when we are on_cpu;
> > > - slow list process. For example priority. The kworker is not
> > > given enough CPU time to do the progress, thus "dummy-nodes"
> > > are not released in time for reuse.
> > >
> > > Therefore, en extra GP is requested if there is a high flow of
> > > synchronize_rcu() users and kworker is not able to do a progress
> > > in time.
> > >
> > > For example 60K+ parallel synchronize_rcu() users will trigger it.
> >
> > OK, but what bad thing would happen if that was moved to precede the
> > rcu_seq_start(&rcu_state.gp_seq)? That way, the requested grace period
> > would be the same as the one that is just now starting.
> >
> > Something like this?
> >
> > start_new_poll = rcu_sr_normal_gp_init();
> >
> > /* Record GP times before starting GP, hence rcu_seq_start(). */
> > rcu_seq_start(&rcu_state.gp_seq);
> > ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> >
> I had a concern about the case when rcu_sr_normal_gp_init() handles what
> we currently have, in terms of requests. Right after that there is/are
> extra sync requests which invoke the start_poll_synchronize_rcu() but
> since a GP has been requested before it will not request an extra one. So
> "last" incoming users might not be processed.
>
> That is why i have placed the rcu_sr_normal_gp_init() after a gp_seq is
> updated.
>
> I can miss something, so please comment. Apart of that we can move it
> as you proposed.

Couldn't that possibility be handled by a check in rcu_gp_cleanup()?

Thanx, Paul

2024-01-03 13:16:24

by Uladzislau Rezki

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Tue, Jan 02, 2024 at 11:25:13AM -0800, Paul E. McKenney wrote:
> On Tue, Jan 02, 2024 at 01:52:26PM +0100, Uladzislau Rezki wrote:
> > Hello, Paul!
> >
> > Sorry for late answer, it is because of holidays :)
> >
> > > > > > The problem is that, we are limited in number of "wait-heads" which we
> > > > > > add as a marker node for this/current grace period. If there are more clients
> > > > > > and there is no a wait-head available it means that a system, the deferred
> > > > > > kworker, is slow in processing callbacks, thus all wait-nodes are in use.
> > > > > >
> > > > > > That is why we need an extra grace period. Basically to repeat our try one
> > > > > > more time, i.e. it might be that a current grace period is not able to handle
> > > > > > users due to the fact that a system is doing really slow, but this is rather
> > > > > > a corner case and is not a problem.
> > > > >
> > > > > But in that case, the real issue is not the need for an extra grace
> > > > > period, but rather the need for the wakeup processing to happen, correct?
> > > > > Or am I missing something subtle here?
> > > > >
> > > > Basically, yes. If we had a spare dummy-node we could process the users
> > > > by the current GP(no need in extra). Why we may not have it - it is because
> > > > like you pointed:
> > > >
> > > > - wake-up issue, i.e. wake-up time + when we are on_cpu;
> > > > - slow list process. For example priority. The kworker is not
> > > > given enough CPU time to do the progress, thus "dummy-nodes"
> > > > are not released in time for reuse.
> > > >
> > > > Therefore, en extra GP is requested if there is a high flow of
> > > > synchronize_rcu() users and kworker is not able to do a progress
> > > > in time.
> > > >
> > > > For example 60K+ parallel synchronize_rcu() users will trigger it.
> > >
> > > OK, but what bad thing would happen if that was moved to precede the
> > > rcu_seq_start(&rcu_state.gp_seq)? That way, the requested grace period
> > > would be the same as the one that is just now starting.
> > >
> > > Something like this?
> > >
> > > start_new_poll = rcu_sr_normal_gp_init();
> > >
> > > /* Record GP times before starting GP, hence rcu_seq_start(). */
> > > rcu_seq_start(&rcu_state.gp_seq);
> > > ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> > >
> > I had a concern about the case when rcu_sr_normal_gp_init() handles what
> > we currently have, in terms of requests. Right after that there is/are
> > extra sync requests which invoke the start_poll_synchronize_rcu() but
> > since a GP has been requested before it will not request an extra one. So
> > "last" incoming users might not be processed.
> >
> > That is why i have placed the rcu_sr_normal_gp_init() after a gp_seq is
> > updated.
> >
> > I can miss something, so please comment. Apart of that we can move it
> > as you proposed.
>
> Couldn't that possibility be handled by a check in rcu_gp_cleanup()?
>
It is controlled by the caller anyway, i.e. if a new GP is needed.

I am not 100% sure it is as straightforward as it could look like to
handle it in the rcu_sr_normal_gp_cleaup() function. At least i see
that we need to access to the first element of llist and find out if
it is a wait-dummy-head or not. If not we know there are extra incoming
calls.

So that way requires extra calling of start_poll_synchronize_rcu().

I can add a comment about your concern and we can find the best approach
later, if it is OK with you!

--
Uladzislau Rezki

2024-01-03 14:47:38

by Paul E. McKenney

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Wed, Jan 03, 2024 at 02:16:00PM +0100, Uladzislau Rezki wrote:
> On Tue, Jan 02, 2024 at 11:25:13AM -0800, Paul E. McKenney wrote:
> > On Tue, Jan 02, 2024 at 01:52:26PM +0100, Uladzislau Rezki wrote:
> > > Hello, Paul!
> > >
> > > Sorry for late answer, it is because of holidays :)
> > >
> > > > > > > The problem is that, we are limited in number of "wait-heads" which we
> > > > > > > add as a marker node for this/current grace period. If there are more clients
> > > > > > > and there is no a wait-head available it means that a system, the deferred
> > > > > > > kworker, is slow in processing callbacks, thus all wait-nodes are in use.
> > > > > > >
> > > > > > > That is why we need an extra grace period. Basically to repeat our try one
> > > > > > > more time, i.e. it might be that a current grace period is not able to handle
> > > > > > > users due to the fact that a system is doing really slow, but this is rather
> > > > > > > a corner case and is not a problem.
> > > > > >
> > > > > > But in that case, the real issue is not the need for an extra grace
> > > > > > period, but rather the need for the wakeup processing to happen, correct?
> > > > > > Or am I missing something subtle here?
> > > > > >
> > > > > Basically, yes. If we had a spare dummy-node we could process the users
> > > > > by the current GP(no need in extra). Why we may not have it - it is because
> > > > > like you pointed:
> > > > >
> > > > > - wake-up issue, i.e. wake-up time + when we are on_cpu;
> > > > > - slow list process. For example priority. The kworker is not
> > > > > given enough CPU time to do the progress, thus "dummy-nodes"
> > > > > are not released in time for reuse.
> > > > >
> > > > > Therefore, en extra GP is requested if there is a high flow of
> > > > > synchronize_rcu() users and kworker is not able to do a progress
> > > > > in time.
> > > > >
> > > > > For example 60K+ parallel synchronize_rcu() users will trigger it.
> > > >
> > > > OK, but what bad thing would happen if that was moved to precede the
> > > > rcu_seq_start(&rcu_state.gp_seq)? That way, the requested grace period
> > > > would be the same as the one that is just now starting.
> > > >
> > > > Something like this?
> > > >
> > > > start_new_poll = rcu_sr_normal_gp_init();
> > > >
> > > > /* Record GP times before starting GP, hence rcu_seq_start(). */
> > > > rcu_seq_start(&rcu_state.gp_seq);
> > > > ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> > > >
> > > I had a concern about the case when rcu_sr_normal_gp_init() handles what
> > > we currently have, in terms of requests. Right after that there is/are
> > > extra sync requests which invoke the start_poll_synchronize_rcu() but
> > > since a GP has been requested before it will not request an extra one. So
> > > "last" incoming users might not be processed.
> > >
> > > That is why i have placed the rcu_sr_normal_gp_init() after a gp_seq is
> > > updated.
> > >
> > > I can miss something, so please comment. Apart of that we can move it
> > > as you proposed.
> >
> > Couldn't that possibility be handled by a check in rcu_gp_cleanup()?
> >
> It is controlled by the caller anyway, i.e. if a new GP is needed.
>
> I am not 100% sure it is as straightforward as it could look like to
> handle it in the rcu_sr_normal_gp_cleaup() function. At least i see
> that we need to access to the first element of llist and find out if
> it is a wait-dummy-head or not. If not we know there are extra incoming
> calls.
>
> So that way requires extra calling of start_poll_synchronize_rcu().

If this is invoked early enough in rcu_gp_cleanup(), all that needs to
happen is to set the need_gp flag. Plus you can count the number of
requests, and snapshot that number at rcu_gp_init() time and check to
see if it changed at rcu_gp_cleanup() time. Later on, this could be
used to reduce the number of wakeups, correct?

> I can add a comment about your concern and we can find the best approach
> later, if it is OK with you!

I agree that this should be added via a later patch, though I have not
yet given up on the possibility that this patch might be simple enough
to be later in this same series.

Thanx, Paul

2024-01-03 17:36:55

by Uladzislau Rezki

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Wed, Jan 03, 2024 at 06:47:30AM -0800, Paul E. McKenney wrote:
> On Wed, Jan 03, 2024 at 02:16:00PM +0100, Uladzislau Rezki wrote:
> > On Tue, Jan 02, 2024 at 11:25:13AM -0800, Paul E. McKenney wrote:
> > > On Tue, Jan 02, 2024 at 01:52:26PM +0100, Uladzislau Rezki wrote:
> > > > Hello, Paul!
> > > >
> > > > Sorry for late answer, it is because of holidays :)
> > > >
> > > > > > > > The problem is that, we are limited in number of "wait-heads" which we
> > > > > > > > add as a marker node for this/current grace period. If there are more clients
> > > > > > > > and there is no a wait-head available it means that a system, the deferred
> > > > > > > > kworker, is slow in processing callbacks, thus all wait-nodes are in use.
> > > > > > > >
> > > > > > > > That is why we need an extra grace period. Basically to repeat our try one
> > > > > > > > more time, i.e. it might be that a current grace period is not able to handle
> > > > > > > > users due to the fact that a system is doing really slow, but this is rather
> > > > > > > > a corner case and is not a problem.
> > > > > > >
> > > > > > > But in that case, the real issue is not the need for an extra grace
> > > > > > > period, but rather the need for the wakeup processing to happen, correct?
> > > > > > > Or am I missing something subtle here?
> > > > > > >
> > > > > > Basically, yes. If we had a spare dummy-node we could process the users
> > > > > > by the current GP(no need in extra). Why we may not have it - it is because
> > > > > > like you pointed:
> > > > > >
> > > > > > - wake-up issue, i.e. wake-up time + when we are on_cpu;
> > > > > > - slow list process. For example priority. The kworker is not
> > > > > > given enough CPU time to do the progress, thus "dummy-nodes"
> > > > > > are not released in time for reuse.
> > > > > >
> > > > > > Therefore, en extra GP is requested if there is a high flow of
> > > > > > synchronize_rcu() users and kworker is not able to do a progress
> > > > > > in time.
> > > > > >
> > > > > > For example 60K+ parallel synchronize_rcu() users will trigger it.
> > > > >
> > > > > OK, but what bad thing would happen if that was moved to precede the
> > > > > rcu_seq_start(&rcu_state.gp_seq)? That way, the requested grace period
> > > > > would be the same as the one that is just now starting.
> > > > >
> > > > > Something like this?
> > > > >
> > > > > start_new_poll = rcu_sr_normal_gp_init();
> > > > >
> > > > > /* Record GP times before starting GP, hence rcu_seq_start(). */
> > > > > rcu_seq_start(&rcu_state.gp_seq);
> > > > > ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> > > > >
> > > > I had a concern about the case when rcu_sr_normal_gp_init() handles what
> > > > we currently have, in terms of requests. Right after that there is/are
> > > > extra sync requests which invoke the start_poll_synchronize_rcu() but
> > > > since a GP has been requested before it will not request an extra one. So
> > > > "last" incoming users might not be processed.
> > > >
> > > > That is why i have placed the rcu_sr_normal_gp_init() after a gp_seq is
> > > > updated.
> > > >
> > > > I can miss something, so please comment. Apart of that we can move it
> > > > as you proposed.
> > >
> > > Couldn't that possibility be handled by a check in rcu_gp_cleanup()?
> > >
> > It is controlled by the caller anyway, i.e. if a new GP is needed.
> >
> > I am not 100% sure it is as straightforward as it could look like to
> > handle it in the rcu_sr_normal_gp_cleaup() function. At least i see
> > that we need to access to the first element of llist and find out if
> > it is a wait-dummy-head or not. If not we know there are extra incoming
> > calls.
> >
> > So that way requires extra calling of start_poll_synchronize_rcu().
>
> If this is invoked early enough in rcu_gp_cleanup(), all that needs to
> happen is to set the need_gp flag. Plus you can count the number of
> requests, and snapshot that number at rcu_gp_init() time and check to
> see if it changed at rcu_gp_cleanup() time. Later on, this could be
> used to reduce the number of wakeups, correct?
>
You mean instead of waking-up a gp-kthread just continue processing of
new users if they are exist? If so, i think, we can implement it as separate
patches.

> > I can add a comment about your concern and we can find the best approach
> > later, if it is OK with you!
>
> I agree that this should be added via a later patch, though I have not
> yet given up on the possibility that this patch might be simple enough
> to be later in this same series.
>
Maybe there is a small misunderstanding. Please note, the rcu_sr_normal_gp_init()
function does not request any new gp, i.e. our approach does not do any extra GP
requests. It happens only if there are no any dummy-wait-head available as we
discussed it earlier.

--
Uladzislau Rezki

2024-01-03 17:56:50

by Paul E. McKenney

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Wed, Jan 03, 2024 at 06:35:20PM +0100, Uladzislau Rezki wrote:
> On Wed, Jan 03, 2024 at 06:47:30AM -0800, Paul E. McKenney wrote:
> > On Wed, Jan 03, 2024 at 02:16:00PM +0100, Uladzislau Rezki wrote:
> > > On Tue, Jan 02, 2024 at 11:25:13AM -0800, Paul E. McKenney wrote:
> > > > On Tue, Jan 02, 2024 at 01:52:26PM +0100, Uladzislau Rezki wrote:
> > > > > Hello, Paul!
> > > > >
> > > > > Sorry for late answer, it is because of holidays :)
> > > > >
> > > > > > > > > The problem is that, we are limited in number of "wait-heads" which we
> > > > > > > > > add as a marker node for this/current grace period. If there are more clients
> > > > > > > > > and there is no a wait-head available it means that a system, the deferred
> > > > > > > > > kworker, is slow in processing callbacks, thus all wait-nodes are in use.
> > > > > > > > >
> > > > > > > > > That is why we need an extra grace period. Basically to repeat our try one
> > > > > > > > > more time, i.e. it might be that a current grace period is not able to handle
> > > > > > > > > users due to the fact that a system is doing really slow, but this is rather
> > > > > > > > > a corner case and is not a problem.
> > > > > > > >
> > > > > > > > But in that case, the real issue is not the need for an extra grace
> > > > > > > > period, but rather the need for the wakeup processing to happen, correct?
> > > > > > > > Or am I missing something subtle here?
> > > > > > > >
> > > > > > > Basically, yes. If we had a spare dummy-node we could process the users
> > > > > > > by the current GP(no need in extra). Why we may not have it - it is because
> > > > > > > like you pointed:
> > > > > > >
> > > > > > > - wake-up issue, i.e. wake-up time + when we are on_cpu;
> > > > > > > - slow list process. For example priority. The kworker is not
> > > > > > > given enough CPU time to do the progress, thus "dummy-nodes"
> > > > > > > are not released in time for reuse.
> > > > > > >
> > > > > > > Therefore, en extra GP is requested if there is a high flow of
> > > > > > > synchronize_rcu() users and kworker is not able to do a progress
> > > > > > > in time.
> > > > > > >
> > > > > > > For example 60K+ parallel synchronize_rcu() users will trigger it.
> > > > > >
> > > > > > OK, but what bad thing would happen if that was moved to precede the
> > > > > > rcu_seq_start(&rcu_state.gp_seq)? That way, the requested grace period
> > > > > > would be the same as the one that is just now starting.
> > > > > >
> > > > > > Something like this?
> > > > > >
> > > > > > start_new_poll = rcu_sr_normal_gp_init();
> > > > > >
> > > > > > /* Record GP times before starting GP, hence rcu_seq_start(). */
> > > > > > rcu_seq_start(&rcu_state.gp_seq);
> > > > > > ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> > > > > >
> > > > > I had a concern about the case when rcu_sr_normal_gp_init() handles what
> > > > > we currently have, in terms of requests. Right after that there is/are
> > > > > extra sync requests which invoke the start_poll_synchronize_rcu() but
> > > > > since a GP has been requested before it will not request an extra one. So
> > > > > "last" incoming users might not be processed.
> > > > >
> > > > > That is why i have placed the rcu_sr_normal_gp_init() after a gp_seq is
> > > > > updated.
> > > > >
> > > > > I can miss something, so please comment. Apart of that we can move it
> > > > > as you proposed.
> > > >
> > > > Couldn't that possibility be handled by a check in rcu_gp_cleanup()?
> > > >
> > > It is controlled by the caller anyway, i.e. if a new GP is needed.
> > >
> > > I am not 100% sure it is as straightforward as it could look like to
> > > handle it in the rcu_sr_normal_gp_cleaup() function. At least i see
> > > that we need to access to the first element of llist and find out if
> > > it is a wait-dummy-head or not. If not we know there are extra incoming
> > > calls.
> > >
> > > So that way requires extra calling of start_poll_synchronize_rcu().
> >
> > If this is invoked early enough in rcu_gp_cleanup(), all that needs to
> > happen is to set the need_gp flag. Plus you can count the number of
> > requests, and snapshot that number at rcu_gp_init() time and check to
> > see if it changed at rcu_gp_cleanup() time. Later on, this could be
> > used to reduce the number of wakeups, correct?
> >
> You mean instead of waking-up a gp-kthread just continue processing of
> new users if they are exist? If so, i think, we can implement it as separate
> patches.

Agreed, this is an optimization, and thus should be a separate patch.

> > > I can add a comment about your concern and we can find the best approach
> > > later, if it is OK with you!
> >
> > I agree that this should be added via a later patch, though I have not
> > yet given up on the possibility that this patch might be simple enough
> > to be later in this same series.
> >
> Maybe there is a small misunderstanding. Please note, the rcu_sr_normal_gp_init()
> function does not request any new gp, i.e. our approach does not do any extra GP
> requests. It happens only if there are no any dummy-wait-head available as we
> discussed it earlier.

The start_poll_synchronize_rcu() added by your patch 4/7 will request
an additional grace period because it is invoked after rcu_seq_start()
is called, correct? Or am I missing something subtle here?

Thanx, Paul

2024-01-03 19:02:16

by Uladzislau Rezki

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Wed, Jan 03, 2024 at 09:56:42AM -0800, Paul E. McKenney wrote:
> On Wed, Jan 03, 2024 at 06:35:20PM +0100, Uladzislau Rezki wrote:
> > On Wed, Jan 03, 2024 at 06:47:30AM -0800, Paul E. McKenney wrote:
> > > On Wed, Jan 03, 2024 at 02:16:00PM +0100, Uladzislau Rezki wrote:
> > > > On Tue, Jan 02, 2024 at 11:25:13AM -0800, Paul E. McKenney wrote:
> > > > > On Tue, Jan 02, 2024 at 01:52:26PM +0100, Uladzislau Rezki wrote:
> > > > > > Hello, Paul!
> > > > > >
> > > > > > Sorry for late answer, it is because of holidays :)
> > > > > >
> > > > > > > > > > The problem is that, we are limited in number of "wait-heads" which we
> > > > > > > > > > add as a marker node for this/current grace period. If there are more clients
> > > > > > > > > > and there is no a wait-head available it means that a system, the deferred
> > > > > > > > > > kworker, is slow in processing callbacks, thus all wait-nodes are in use.
> > > > > > > > > >
> > > > > > > > > > That is why we need an extra grace period. Basically to repeat our try one
> > > > > > > > > > more time, i.e. it might be that a current grace period is not able to handle
> > > > > > > > > > users due to the fact that a system is doing really slow, but this is rather
> > > > > > > > > > a corner case and is not a problem.
> > > > > > > > >
> > > > > > > > > But in that case, the real issue is not the need for an extra grace
> > > > > > > > > period, but rather the need for the wakeup processing to happen, correct?
> > > > > > > > > Or am I missing something subtle here?
> > > > > > > > >
> > > > > > > > Basically, yes. If we had a spare dummy-node we could process the users
> > > > > > > > by the current GP(no need in extra). Why we may not have it - it is because
> > > > > > > > like you pointed:
> > > > > > > >
> > > > > > > > - wake-up issue, i.e. wake-up time + when we are on_cpu;
> > > > > > > > - slow list process. For example priority. The kworker is not
> > > > > > > > given enough CPU time to do the progress, thus "dummy-nodes"
> > > > > > > > are not released in time for reuse.
> > > > > > > >
> > > > > > > > Therefore, en extra GP is requested if there is a high flow of
> > > > > > > > synchronize_rcu() users and kworker is not able to do a progress
> > > > > > > > in time.
> > > > > > > >
> > > > > > > > For example 60K+ parallel synchronize_rcu() users will trigger it.
> > > > > > >
> > > > > > > OK, but what bad thing would happen if that was moved to precede the
> > > > > > > rcu_seq_start(&rcu_state.gp_seq)? That way, the requested grace period
> > > > > > > would be the same as the one that is just now starting.
> > > > > > >
> > > > > > > Something like this?
> > > > > > >
> > > > > > > start_new_poll = rcu_sr_normal_gp_init();
> > > > > > >
> > > > > > > /* Record GP times before starting GP, hence rcu_seq_start(). */
> > > > > > > rcu_seq_start(&rcu_state.gp_seq);
> > > > > > > ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> > > > > > >
> > > > > > I had a concern about the case when rcu_sr_normal_gp_init() handles what
> > > > > > we currently have, in terms of requests. Right after that there is/are
> > > > > > extra sync requests which invoke the start_poll_synchronize_rcu() but
> > > > > > since a GP has been requested before it will not request an extra one. So
> > > > > > "last" incoming users might not be processed.
> > > > > >
> > > > > > That is why i have placed the rcu_sr_normal_gp_init() after a gp_seq is
> > > > > > updated.
> > > > > >
> > > > > > I can miss something, so please comment. Apart of that we can move it
> > > > > > as you proposed.
> > > > >
> > > > > Couldn't that possibility be handled by a check in rcu_gp_cleanup()?
> > > > >
> > > > It is controlled by the caller anyway, i.e. if a new GP is needed.
> > > >
> > > > I am not 100% sure it is as straightforward as it could look like to
> > > > handle it in the rcu_sr_normal_gp_cleaup() function. At least i see
> > > > that we need to access to the first element of llist and find out if
> > > > it is a wait-dummy-head or not. If not we know there are extra incoming
> > > > calls.
> > > >
> > > > So that way requires extra calling of start_poll_synchronize_rcu().
> > >
> > > If this is invoked early enough in rcu_gp_cleanup(), all that needs to
> > > happen is to set the need_gp flag. Plus you can count the number of
> > > requests, and snapshot that number at rcu_gp_init() time and check to
> > > see if it changed at rcu_gp_cleanup() time. Later on, this could be
> > > used to reduce the number of wakeups, correct?
> > >
> > You mean instead of waking-up a gp-kthread just continue processing of
> > new users if they are exist? If so, i think, we can implement it as separate
> > patches.
>
> Agreed, this is an optimization, and thus should be a separate patch.
>
> > > > I can add a comment about your concern and we can find the best approach
> > > > later, if it is OK with you!
> > >
> > > I agree that this should be added via a later patch, though I have not
> > > yet given up on the possibility that this patch might be simple enough
> > > to be later in this same series.
> > >
> > Maybe there is a small misunderstanding. Please note, the rcu_sr_normal_gp_init()
> > function does not request any new gp, i.e. our approach does not do any extra GP
> > requests. It happens only if there are no any dummy-wait-head available as we
> > discussed it earlier.
>
> The start_poll_synchronize_rcu() added by your patch 4/7 will request
> an additional grace period because it is invoked after rcu_seq_start()
> is called, correct? Or am I missing something subtle here?
>
<snip>
+ // New poll request after rnp unlock
+ if (start_new_poll)
+ (void) start_poll_synchronize_rcu();
+
<snip>

The "start_new_poll" is set to "true" only when _this_ GP is not able
to handle anything and there are outstanding users. It happens when the
rcu_sr_normal_gp_init() function was not able to insert a dummy separator
to the llist, because there were no left dummy-nodes(fixed number of them)
due to the fact that all of them are "in-use". The reason why there are no
dummy-nodes is because of slow progress because it is done by dedicated
kworker.

I can trigger it, i mean when we need an addition GP, start_new_pool is 1,
only when i run 20 000 processes concurrently in a tight loop:

<snip>
while (1)
synchronize_rcu();
<snip>

in that scenario we start to ask for an addition GP because we are not up
to speed, i.e. a system is slow in processing callbacks and we need some
time until wait-node/nodes is/are released for reuse.

We need a next GP to move it forward, i.e. to repeat a try of attaching
a dummy-node.

--
Uladzislau Rezki

2024-01-03 19:03:26

by Uladzislau Rezki

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Wed, Jan 03, 2024 at 08:02:00PM +0100, Uladzislau Rezki wrote:
> On Wed, Jan 03, 2024 at 09:56:42AM -0800, Paul E. McKenney wrote:
> > On Wed, Jan 03, 2024 at 06:35:20PM +0100, Uladzislau Rezki wrote:
> > > On Wed, Jan 03, 2024 at 06:47:30AM -0800, Paul E. McKenney wrote:
> > > > On Wed, Jan 03, 2024 at 02:16:00PM +0100, Uladzislau Rezki wrote:
> > > > > On Tue, Jan 02, 2024 at 11:25:13AM -0800, Paul E. McKenney wrote:
> > > > > > On Tue, Jan 02, 2024 at 01:52:26PM +0100, Uladzislau Rezki wrote:
> > > > > > > Hello, Paul!
> > > > > > >
> > > > > > > Sorry for late answer, it is because of holidays :)
> > > > > > >
> > > > > > > > > > > The problem is that, we are limited in number of "wait-heads" which we
> > > > > > > > > > > add as a marker node for this/current grace period. If there are more clients
> > > > > > > > > > > and there is no a wait-head available it means that a system, the deferred
> > > > > > > > > > > kworker, is slow in processing callbacks, thus all wait-nodes are in use.
> > > > > > > > > > >
> > > > > > > > > > > That is why we need an extra grace period. Basically to repeat our try one
> > > > > > > > > > > more time, i.e. it might be that a current grace period is not able to handle
> > > > > > > > > > > users due to the fact that a system is doing really slow, but this is rather
> > > > > > > > > > > a corner case and is not a problem.
> > > > > > > > > >
> > > > > > > > > > But in that case, the real issue is not the need for an extra grace
> > > > > > > > > > period, but rather the need for the wakeup processing to happen, correct?
> > > > > > > > > > Or am I missing something subtle here?
> > > > > > > > > >
> > > > > > > > > Basically, yes. If we had a spare dummy-node we could process the users
> > > > > > > > > by the current GP(no need in extra). Why we may not have it - it is because
> > > > > > > > > like you pointed:
> > > > > > > > >
> > > > > > > > > - wake-up issue, i.e. wake-up time + when we are on_cpu;
> > > > > > > > > - slow list process. For example priority. The kworker is not
> > > > > > > > > given enough CPU time to do the progress, thus "dummy-nodes"
> > > > > > > > > are not released in time for reuse.
> > > > > > > > >
> > > > > > > > > Therefore, en extra GP is requested if there is a high flow of
> > > > > > > > > synchronize_rcu() users and kworker is not able to do a progress
> > > > > > > > > in time.
> > > > > > > > >
> > > > > > > > > For example 60K+ parallel synchronize_rcu() users will trigger it.
> > > > > > > >
> > > > > > > > OK, but what bad thing would happen if that was moved to precede the
> > > > > > > > rcu_seq_start(&rcu_state.gp_seq)? That way, the requested grace period
> > > > > > > > would be the same as the one that is just now starting.
> > > > > > > >
> > > > > > > > Something like this?
> > > > > > > >
> > > > > > > > start_new_poll = rcu_sr_normal_gp_init();
> > > > > > > >
> > > > > > > > /* Record GP times before starting GP, hence rcu_seq_start(). */
> > > > > > > > rcu_seq_start(&rcu_state.gp_seq);
> > > > > > > > ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> > > > > > > >
> > > > > > > I had a concern about the case when rcu_sr_normal_gp_init() handles what
> > > > > > > we currently have, in terms of requests. Right after that there is/are
> > > > > > > extra sync requests which invoke the start_poll_synchronize_rcu() but
> > > > > > > since a GP has been requested before it will not request an extra one. So
> > > > > > > "last" incoming users might not be processed.
> > > > > > >
> > > > > > > That is why i have placed the rcu_sr_normal_gp_init() after a gp_seq is
> > > > > > > updated.
> > > > > > >
> > > > > > > I can miss something, so please comment. Apart of that we can move it
> > > > > > > as you proposed.
> > > > > >
> > > > > > Couldn't that possibility be handled by a check in rcu_gp_cleanup()?
> > > > > >
> > > > > It is controlled by the caller anyway, i.e. if a new GP is needed.
> > > > >
> > > > > I am not 100% sure it is as straightforward as it could look like to
> > > > > handle it in the rcu_sr_normal_gp_cleaup() function. At least i see
> > > > > that we need to access to the first element of llist and find out if
> > > > > it is a wait-dummy-head or not. If not we know there are extra incoming
> > > > > calls.
> > > > >
> > > > > So that way requires extra calling of start_poll_synchronize_rcu().
> > > >
> > > > If this is invoked early enough in rcu_gp_cleanup(), all that needs to
> > > > happen is to set the need_gp flag. Plus you can count the number of
> > > > requests, and snapshot that number at rcu_gp_init() time and check to
> > > > see if it changed at rcu_gp_cleanup() time. Later on, this could be
> > > > used to reduce the number of wakeups, correct?
> > > >
> > > You mean instead of waking-up a gp-kthread just continue processing of
> > > new users if they are exist? If so, i think, we can implement it as separate
> > > patches.
> >
> > Agreed, this is an optimization, and thus should be a separate patch.
> >
> > > > > I can add a comment about your concern and we can find the best approach
> > > > > later, if it is OK with you!
> > > >
> > > > I agree that this should be added via a later patch, though I have not
> > > > yet given up on the possibility that this patch might be simple enough
> > > > to be later in this same series.
> > > >
> > > Maybe there is a small misunderstanding. Please note, the rcu_sr_normal_gp_init()
> > > function does not request any new gp, i.e. our approach does not do any extra GP
> > > requests. It happens only if there are no any dummy-wait-head available as we
> > > discussed it earlier.
> >
> > The start_poll_synchronize_rcu() added by your patch 4/7 will request
> > an additional grace period because it is invoked after rcu_seq_start()
> > is called, correct? Or am I missing something subtle here?
> >
> <snip>
> + // New poll request after rnp unlock
> + if (start_new_poll)
> + (void) start_poll_synchronize_rcu();
> +
> <snip>
>
> The "start_new_poll" is set to "true" only when _this_ GP is not able
> to handle anything and there are outstanding users. It happens when the
> rcu_sr_normal_gp_init() function was not able to insert a dummy separator
> to the llist, because there were no left dummy-nodes(fixed number of them)
> due to the fact that all of them are "in-use". The reason why there are no
> dummy-nodes is because of slow progress because it is done by dedicated
> kworker.
>
> I can trigger it, i mean when we need an addition GP, start_new_pool is 1,
> only when i run 20 000 processes concurrently in a tight loop:
>
> <snip>
> while (1)
> synchronize_rcu();
> <snip>
>
> in that scenario we start to ask for an addition GP because we are not up
> to speed, i.e. a system is slow in processing callbacks and we need some
> time until wait-node/nodes is/are released for reuse.
>
> We need a next GP to move it forward, i.e. to repeat a try of attaching
> a dummy-node.
>
Probably i should add a comment about it :)

--
Uladzislau Rezki

2024-01-03 19:33:12

by Paul E. McKenney

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Wed, Jan 03, 2024 at 08:03:10PM +0100, Uladzislau Rezki wrote:
> On Wed, Jan 03, 2024 at 08:02:00PM +0100, Uladzislau Rezki wrote:
> > On Wed, Jan 03, 2024 at 09:56:42AM -0800, Paul E. McKenney wrote:
> > > On Wed, Jan 03, 2024 at 06:35:20PM +0100, Uladzislau Rezki wrote:
> > > > On Wed, Jan 03, 2024 at 06:47:30AM -0800, Paul E. McKenney wrote:
> > > > > On Wed, Jan 03, 2024 at 02:16:00PM +0100, Uladzislau Rezki wrote:
> > > > > > On Tue, Jan 02, 2024 at 11:25:13AM -0800, Paul E. McKenney wrote:
> > > > > > > On Tue, Jan 02, 2024 at 01:52:26PM +0100, Uladzislau Rezki wrote:
> > > > > > > > Hello, Paul!
> > > > > > > >
> > > > > > > > Sorry for late answer, it is because of holidays :)
> > > > > > > >
> > > > > > > > > > > > The problem is that, we are limited in number of "wait-heads" which we
> > > > > > > > > > > > add as a marker node for this/current grace period. If there are more clients
> > > > > > > > > > > > and there is no a wait-head available it means that a system, the deferred
> > > > > > > > > > > > kworker, is slow in processing callbacks, thus all wait-nodes are in use.
> > > > > > > > > > > >
> > > > > > > > > > > > That is why we need an extra grace period. Basically to repeat our try one
> > > > > > > > > > > > more time, i.e. it might be that a current grace period is not able to handle
> > > > > > > > > > > > users due to the fact that a system is doing really slow, but this is rather
> > > > > > > > > > > > a corner case and is not a problem.
> > > > > > > > > > >
> > > > > > > > > > > But in that case, the real issue is not the need for an extra grace
> > > > > > > > > > > period, but rather the need for the wakeup processing to happen, correct?
> > > > > > > > > > > Or am I missing something subtle here?
> > > > > > > > > > >
> > > > > > > > > > Basically, yes. If we had a spare dummy-node we could process the users
> > > > > > > > > > by the current GP(no need in extra). Why we may not have it - it is because
> > > > > > > > > > like you pointed:
> > > > > > > > > >
> > > > > > > > > > - wake-up issue, i.e. wake-up time + when we are on_cpu;
> > > > > > > > > > - slow list process. For example priority. The kworker is not
> > > > > > > > > > given enough CPU time to do the progress, thus "dummy-nodes"
> > > > > > > > > > are not released in time for reuse.
> > > > > > > > > >
> > > > > > > > > > Therefore, en extra GP is requested if there is a high flow of
> > > > > > > > > > synchronize_rcu() users and kworker is not able to do a progress
> > > > > > > > > > in time.
> > > > > > > > > >
> > > > > > > > > > For example 60K+ parallel synchronize_rcu() users will trigger it.
> > > > > > > > >
> > > > > > > > > OK, but what bad thing would happen if that was moved to precede the
> > > > > > > > > rcu_seq_start(&rcu_state.gp_seq)? That way, the requested grace period
> > > > > > > > > would be the same as the one that is just now starting.
> > > > > > > > >
> > > > > > > > > Something like this?
> > > > > > > > >
> > > > > > > > > start_new_poll = rcu_sr_normal_gp_init();
> > > > > > > > >
> > > > > > > > > /* Record GP times before starting GP, hence rcu_seq_start(). */
> > > > > > > > > rcu_seq_start(&rcu_state.gp_seq);
> > > > > > > > > ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> > > > > > > > >
> > > > > > > > I had a concern about the case when rcu_sr_normal_gp_init() handles what
> > > > > > > > we currently have, in terms of requests. Right after that there is/are
> > > > > > > > extra sync requests which invoke the start_poll_synchronize_rcu() but
> > > > > > > > since a GP has been requested before it will not request an extra one. So
> > > > > > > > "last" incoming users might not be processed.
> > > > > > > >
> > > > > > > > That is why i have placed the rcu_sr_normal_gp_init() after a gp_seq is
> > > > > > > > updated.
> > > > > > > >
> > > > > > > > I can miss something, so please comment. Apart of that we can move it
> > > > > > > > as you proposed.
> > > > > > >
> > > > > > > Couldn't that possibility be handled by a check in rcu_gp_cleanup()?
> > > > > > >
> > > > > > It is controlled by the caller anyway, i.e. if a new GP is needed.
> > > > > >
> > > > > > I am not 100% sure it is as straightforward as it could look like to
> > > > > > handle it in the rcu_sr_normal_gp_cleaup() function. At least i see
> > > > > > that we need to access to the first element of llist and find out if
> > > > > > it is a wait-dummy-head or not. If not we know there are extra incoming
> > > > > > calls.
> > > > > >
> > > > > > So that way requires extra calling of start_poll_synchronize_rcu().
> > > > >
> > > > > If this is invoked early enough in rcu_gp_cleanup(), all that needs to
> > > > > happen is to set the need_gp flag. Plus you can count the number of
> > > > > requests, and snapshot that number at rcu_gp_init() time and check to
> > > > > see if it changed at rcu_gp_cleanup() time. Later on, this could be
> > > > > used to reduce the number of wakeups, correct?
> > > > >
> > > > You mean instead of waking-up a gp-kthread just continue processing of
> > > > new users if they are exist? If so, i think, we can implement it as separate
> > > > patches.
> > >
> > > Agreed, this is an optimization, and thus should be a separate patch.
> > >
> > > > > > I can add a comment about your concern and we can find the best approach
> > > > > > later, if it is OK with you!
> > > > >
> > > > > I agree that this should be added via a later patch, though I have not
> > > > > yet given up on the possibility that this patch might be simple enough
> > > > > to be later in this same series.
> > > > >
> > > > Maybe there is a small misunderstanding. Please note, the rcu_sr_normal_gp_init()
> > > > function does not request any new gp, i.e. our approach does not do any extra GP
> > > > requests. It happens only if there are no any dummy-wait-head available as we
> > > > discussed it earlier.
> > >
> > > The start_poll_synchronize_rcu() added by your patch 4/7 will request
> > > an additional grace period because it is invoked after rcu_seq_start()
> > > is called, correct? Or am I missing something subtle here?
> > >
> > <snip>
> > + // New poll request after rnp unlock
> > + if (start_new_poll)
> > + (void) start_poll_synchronize_rcu();
> > +
> > <snip>
> >
> > The "start_new_poll" is set to "true" only when _this_ GP is not able
> > to handle anything and there are outstanding users. It happens when the
> > rcu_sr_normal_gp_init() function was not able to insert a dummy separator
> > to the llist, because there were no left dummy-nodes(fixed number of them)
> > due to the fact that all of them are "in-use". The reason why there are no
> > dummy-nodes is because of slow progress because it is done by dedicated
> > kworker.
> >
> > I can trigger it, i mean when we need an addition GP, start_new_pool is 1,
> > only when i run 20 000 processes concurrently in a tight loop:
> >
> > <snip>
> > while (1)
> > synchronize_rcu();
> > <snip>
> >
> > in that scenario we start to ask for an addition GP because we are not up
> > to speed, i.e. a system is slow in processing callbacks and we need some
> > time until wait-node/nodes is/are released for reuse.
> >
> > We need a next GP to move it forward, i.e. to repeat a try of attaching
> > a dummy-node.
> >
> Probably i should add a comment about it :)

Sounds good, and thank you for bearing with me!

Thanx, Paul

2024-01-04 11:18:36

by Uladzislau Rezki

[permalink] [raw]
Subject: Re: [PATCH v3 4/7] rcu: Improve handling of synchronize_rcu() users

On Wed, Jan 03, 2024 at 11:33:01AM -0800, Paul E. McKenney wrote:
> On Wed, Jan 03, 2024 at 08:03:10PM +0100, Uladzislau Rezki wrote:
> > On Wed, Jan 03, 2024 at 08:02:00PM +0100, Uladzislau Rezki wrote:
> > > On Wed, Jan 03, 2024 at 09:56:42AM -0800, Paul E. McKenney wrote:
> > > > On Wed, Jan 03, 2024 at 06:35:20PM +0100, Uladzislau Rezki wrote:
> > > > > On Wed, Jan 03, 2024 at 06:47:30AM -0800, Paul E. McKenney wrote:
> > > > > > On Wed, Jan 03, 2024 at 02:16:00PM +0100, Uladzislau Rezki wrote:
> > > > > > > On Tue, Jan 02, 2024 at 11:25:13AM -0800, Paul E. McKenney wrote:
> > > > > > > > On Tue, Jan 02, 2024 at 01:52:26PM +0100, Uladzislau Rezki wrote:
> > > > > > > > > Hello, Paul!
> > > > > > > > >
> > > > > > > > > Sorry for late answer, it is because of holidays :)
> > > > > > > > >
> > > > > > > > > > > > > The problem is that, we are limited in number of "wait-heads" which we
> > > > > > > > > > > > > add as a marker node for this/current grace period. If there are more clients
> > > > > > > > > > > > > and there is no a wait-head available it means that a system, the deferred
> > > > > > > > > > > > > kworker, is slow in processing callbacks, thus all wait-nodes are in use.
> > > > > > > > > > > > >
> > > > > > > > > > > > > That is why we need an extra grace period. Basically to repeat our try one
> > > > > > > > > > > > > more time, i.e. it might be that a current grace period is not able to handle
> > > > > > > > > > > > > users due to the fact that a system is doing really slow, but this is rather
> > > > > > > > > > > > > a corner case and is not a problem.
> > > > > > > > > > > >
> > > > > > > > > > > > But in that case, the real issue is not the need for an extra grace
> > > > > > > > > > > > period, but rather the need for the wakeup processing to happen, correct?
> > > > > > > > > > > > Or am I missing something subtle here?
> > > > > > > > > > > >
> > > > > > > > > > > Basically, yes. If we had a spare dummy-node we could process the users
> > > > > > > > > > > by the current GP(no need in extra). Why we may not have it - it is because
> > > > > > > > > > > like you pointed:
> > > > > > > > > > >
> > > > > > > > > > > - wake-up issue, i.e. wake-up time + when we are on_cpu;
> > > > > > > > > > > - slow list process. For example priority. The kworker is not
> > > > > > > > > > > given enough CPU time to do the progress, thus "dummy-nodes"
> > > > > > > > > > > are not released in time for reuse.
> > > > > > > > > > >
> > > > > > > > > > > Therefore, en extra GP is requested if there is a high flow of
> > > > > > > > > > > synchronize_rcu() users and kworker is not able to do a progress
> > > > > > > > > > > in time.
> > > > > > > > > > >
> > > > > > > > > > > For example 60K+ parallel synchronize_rcu() users will trigger it.
> > > > > > > > > >
> > > > > > > > > > OK, but what bad thing would happen if that was moved to precede the
> > > > > > > > > > rcu_seq_start(&rcu_state.gp_seq)? That way, the requested grace period
> > > > > > > > > > would be the same as the one that is just now starting.
> > > > > > > > > >
> > > > > > > > > > Something like this?
> > > > > > > > > >
> > > > > > > > > > start_new_poll = rcu_sr_normal_gp_init();
> > > > > > > > > >
> > > > > > > > > > /* Record GP times before starting GP, hence rcu_seq_start(). */
> > > > > > > > > > rcu_seq_start(&rcu_state.gp_seq);
> > > > > > > > > > ASSERT_EXCLUSIVE_WRITER(rcu_state.gp_seq);
> > > > > > > > > >
> > > > > > > > > I had a concern about the case when rcu_sr_normal_gp_init() handles what
> > > > > > > > > we currently have, in terms of requests. Right after that there is/are
> > > > > > > > > extra sync requests which invoke the start_poll_synchronize_rcu() but
> > > > > > > > > since a GP has been requested before it will not request an extra one. So
> > > > > > > > > "last" incoming users might not be processed.
> > > > > > > > >
> > > > > > > > > That is why i have placed the rcu_sr_normal_gp_init() after a gp_seq is
> > > > > > > > > updated.
> > > > > > > > >
> > > > > > > > > I can miss something, so please comment. Apart of that we can move it
> > > > > > > > > as you proposed.
> > > > > > > >
> > > > > > > > Couldn't that possibility be handled by a check in rcu_gp_cleanup()?
> > > > > > > >
> > > > > > > It is controlled by the caller anyway, i.e. if a new GP is needed.
> > > > > > >
> > > > > > > I am not 100% sure it is as straightforward as it could look like to
> > > > > > > handle it in the rcu_sr_normal_gp_cleaup() function. At least i see
> > > > > > > that we need to access to the first element of llist and find out if
> > > > > > > it is a wait-dummy-head or not. If not we know there are extra incoming
> > > > > > > calls.
> > > > > > >
> > > > > > > So that way requires extra calling of start_poll_synchronize_rcu().
> > > > > >
> > > > > > If this is invoked early enough in rcu_gp_cleanup(), all that needs to
> > > > > > happen is to set the need_gp flag. Plus you can count the number of
> > > > > > requests, and snapshot that number at rcu_gp_init() time and check to
> > > > > > see if it changed at rcu_gp_cleanup() time. Later on, this could be
> > > > > > used to reduce the number of wakeups, correct?
> > > > > >
> > > > > You mean instead of waking-up a gp-kthread just continue processing of
> > > > > new users if they are exist? If so, i think, we can implement it as separate
> > > > > patches.
> > > >
> > > > Agreed, this is an optimization, and thus should be a separate patch.
> > > >
> > > > > > > I can add a comment about your concern and we can find the best approach
> > > > > > > later, if it is OK with you!
> > > > > >
> > > > > > I agree that this should be added via a later patch, though I have not
> > > > > > yet given up on the possibility that this patch might be simple enough
> > > > > > to be later in this same series.
> > > > > >
> > > > > Maybe there is a small misunderstanding. Please note, the rcu_sr_normal_gp_init()
> > > > > function does not request any new gp, i.e. our approach does not do any extra GP
> > > > > requests. It happens only if there are no any dummy-wait-head available as we
> > > > > discussed it earlier.
> > > >
> > > > The start_poll_synchronize_rcu() added by your patch 4/7 will request
> > > > an additional grace period because it is invoked after rcu_seq_start()
> > > > is called, correct? Or am I missing something subtle here?
> > > >
> > > <snip>
> > > + // New poll request after rnp unlock
> > > + if (start_new_poll)
> > > + (void) start_poll_synchronize_rcu();
> > > +
> > > <snip>
> > >
> > > The "start_new_poll" is set to "true" only when _this_ GP is not able
> > > to handle anything and there are outstanding users. It happens when the
> > > rcu_sr_normal_gp_init() function was not able to insert a dummy separator
> > > to the llist, because there were no left dummy-nodes(fixed number of them)
> > > due to the fact that all of them are "in-use". The reason why there are no
> > > dummy-nodes is because of slow progress because it is done by dedicated
> > > kworker.
> > >
> > > I can trigger it, i mean when we need an addition GP, start_new_pool is 1,
> > > only when i run 20 000 processes concurrently in a tight loop:
> > >
> > > <snip>
> > > while (1)
> > > synchronize_rcu();
> > > <snip>
> > >
> > > in that scenario we start to ask for an addition GP because we are not up
> > > to speed, i.e. a system is slow in processing callbacks and we need some
> > > time until wait-node/nodes is/are released for reuse.
> > >
> > > We need a next GP to move it forward, i.e. to repeat a try of attaching
> > > a dummy-node.
> > >
> > Probably i should add a comment about it :)
>
> Sounds good, and thank you for bearing with me!
>
Thanks to you :)

--
Uladzislau Rezki