urcu-qsbr: batch concurrent synchronize_rcu()

Here are benchmarks on batching of synchronize_rcu(), and it leads to
very interesting scalability improvement and speedups, e.g., on a
24-core AMD, with a write-heavy scenario (4 readers threads, 20 updater
threads, each updater using synchronize_rcu()):

* Serialized grace periods :

./test_urcu_qsbr 4 20 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers   4
rdur      0 wdur      0 nr_writers  20 wdelay      0
nr_reads  20251412728 nr_writes      1826331 nr_ops  20253239059

* Batched grace periods :

./test_urcu_qsbr 4 20 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers   4
rdur      0 wdur      0 nr_writers  20 wdelay      0
nr_reads  15141994746 nr_writes      9382515 nr_ops  15151377261

For a 9382515/1826331 = 5.13 speedup for 20 updaters.

Of course, we can see that readers have slowed down, probably due to
increased update traffic, given there is no change to the read-side code
whatsoever.

Now let's see the penality of managing the stack for single-updater.
With 4 readers, single updater:

* Serialized grace periods :

./test_urcu_qsbr 4 1 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers   4
rdur      0 wdur      0 nr_writers   1 wdelay      0
nr_reads  19240784755 nr_writes      2130839 nr_ops  19242915594

* Batched grace periods :

./test_urcu_qsbr 4 1 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers   4
rdur      0 wdur      0 nr_writers   1 wdelay      0
nr_reads  19160162768 nr_writes      2253068 nr_ops  1916241583

2253068 vs 2137036 -> a couple of runs show that this difference lost in
the noise for single updater.

More benchmark results:

* Serialized synchronize_rcu() -- test_urcu_qsbr

./test_urcu_qsbr 4 1 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers   4 rdur      0 wdur      0 nr_writers   1 wdelay      0 nr_reads  18841016559 nr_writes      1857130 nr_ops  18842873689
./test_urcu_qsbr 4 20 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers   4 rdur      0 wdur      0 nr_writers  20 wdelay      0 nr_reads  20272811733 nr_writes      1837027 nr_ops  20274648760
./test_urcu_qsbr 12 12 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers  12 rdur      0 wdur      0 nr_writers  12 wdelay      0 nr_reads  60343516643 nr_writes      2353685 nr_ops  60345870328
./test_urcu_qsbr 16 8 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers  16 rdur      0 wdur      0 nr_writers   8 wdelay      0 nr_reads  78202711840 nr_writes      2326331 nr_ops  78205038171
./test_urcu_qsbr 20 4 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers  20 rdur      0 wdur      0 nr_writers   4 wdelay      0 nr_reads  94553396003 nr_writes      2238396 nr_ops  94555634399
./test_urcu_qsbr 20 3 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers  20 rdur      0 wdur      0 nr_writers   3 wdelay      0 nr_reads  95004708661 nr_writes      2165966 nr_ops  95006874627
./test_urcu_qsbr 20 2 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers  20 rdur      0 wdur      0 nr_writers   2 wdelay      0 nr_reads  95386506198 nr_writes      2194352 nr_ops  95388700550
./test_urcu_qsbr 20 1 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers  20 rdur      0 wdur      0 nr_writers   1 wdelay      0 nr_reads  84705972017 nr_writes      2609595 nr_ops  84708581612

* Batched synchronize_rcu() -- test_urcu_qsbr

./test_urcu_qsbr 4 1 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers   4 rdur      0 wdur      0 nr_writers   1 wdelay      0 nr_reads  19154850714 nr_writes      2238834 nr_ops  19157089548
./test_urcu_qsbr 4 20 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers   4 rdur      0 wdur      0 nr_writers  20 wdelay      0 nr_reads  15114131760 nr_writes      9370255 nr_ops  15123502015
./test_urcu_qsbr 12 12 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers  12 rdur      0 wdur      0 nr_writers  12 wdelay      0 nr_reads  45541854970 nr_writes      5786496 nr_ops  45547641466
./test_urcu_qsbr 16 8 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers  16 rdur      0 wdur      0 nr_writers   8 wdelay      0 nr_reads  66217337547 nr_writes      4257427 nr_ops  66221594974
./test_urcu_qsbr 20 4 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers  20 rdur      0 wdur      0 nr_writers   4 wdelay      0 nr_reads  95048642908 nr_writes      2416266 nr_ops  95051059174
./test_urcu_qsbr 20 3 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers  20 rdur      0 wdur      0 nr_writers   3 wdelay      0 nr_reads  96679609928 nr_writes      2211168 nr_ops  96681821096
./test_urcu_qsbr 20 2 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers  20 rdur      0 wdur      0 nr_writers   2 wdelay      0 nr_reads  92166219811 nr_writes      1968725 nr_ops  92168188536
./test_urcu_qsbr 20 1 20
SUMMARY ./test_urcu_qsbr          testdur   20 nr_readers  20 rdur      0 wdur      0 nr_writers   1 wdelay      0 nr_reads  87986181951 nr_writes      3278737 nr_ops  87989460688

CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: Lai Jiangshan <laijs@cn.fujitsu.com>
CC: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

diff --git a/urcu-qsbr.c b/urcu-qsbr.c
index 5b341b5cfd6d97034b08ca5ea6df913a0c8a7736..7f747ed5f9ce8ef133df3c2d026b8b76154e2e9e 100644 (file)
--- a/urcu-qsbr.c
+++ b/urcu-qsbr.c
@@ -36,6 +36,7 @@
 #include <poll.h>
 
 #include "urcu/wfcqueue.h"
+#include "urcu/wfstack.h"
 #include "urcu/map/urcu-qsbr.h"
 #define BUILD_QSBR_LIB
 #include "urcu/static/urcu-qsbr.h"
@@ -78,6 +79,35 @@ DEFINE_URCU_TLS(unsigned int, rcu_rand_yield);
 
 static CDS_LIST_HEAD(registry);
 
+/*
+ * Number of busy-loop attempts before waiting on futex for grace period
+ * batching.
+ */
+#define RCU_AWAKE_ATTEMPTS 1000
+
+enum adapt_wakeup_state {
+       /* AWAKE_WAITING is compared directly (futex compares it). */
+       AWAKE_WAITING =         0,
+       /* non-zero are used as masks. */
+       AWAKE_WAKEUP =          (1 << 0),
+       AWAKE_AWAKENED =        (1 << 1),
+       AWAKE_TEARDOWN =        (1 << 2),
+};
+
+struct gp_waiters_thread {
+       struct cds_wfs_node node;
+       int32_t wait_futex;
+};
+
+/*
+ * Stack keeping threads awaiting to wait for a grace period. Contains
+ * struct gp_waiters_thread objects.
+ */
+static struct cds_wfs_stack gp_waiters = {
+       .head = CDS_WFS_END,
+       .lock = PTHREAD_MUTEX_INITIALIZER,
+};
+
 static void mutex_lock(pthread_mutex_t *mutex)
 {
        int ret;
@@ -116,6 +146,58 @@ static void wait_gp(void)
                      NULL, NULL, 0);
 }
 
+/*
+ * Note: urcu_adaptative_wake_up needs "value" to stay allocated
+ * throughout its execution. In this scheme, the waiter owns the futex
+ * memory, and we only allow it to free this memory when it receives the
+ * AWAKE_TEARDOWN flag.
+ */
+static void urcu_adaptative_wake_up(int32_t *value)
+{
+       cmm_smp_mb();
+       assert(uatomic_read(value) == AWAKE_WAITING);
+       uatomic_set(value, AWAKE_WAKEUP);
+       if (!(uatomic_read(value) & AWAKE_AWAKENED))
+               futex_noasync(value, FUTEX_WAKE, 1, NULL, NULL, 0);
+       /* Allow teardown of "value" memory. */
+       uatomic_or(value, AWAKE_TEARDOWN);
+}
+
+/*
+ * Caller must initialize "value" to AWAKE_WAITING before passing its
+ * memory to waker thread.
+ */
+static void urcu_adaptative_busy_wait(int32_t *value)
+{
+       unsigned int i;
+
+       /* Load and test condition before read futex */
+       cmm_smp_rmb();
+       for (i = 0; i < RCU_AWAKE_ATTEMPTS; i++) {
+               if (uatomic_read(value) != AWAKE_WAITING)
+                       goto skip_futex_wait;
+               caa_cpu_relax();
+       }
+       futex_noasync(value, FUTEX_WAIT, AWAKE_WAITING, NULL, NULL, 0);
+skip_futex_wait:
+
+       /* Tell waker thread than we are awakened. */
+       uatomic_or(value, AWAKE_AWAKENED);
+
+       /*
+        * Wait until waker thread lets us know it's ok to tear down
+        * memory allocated for value.
+        */
+       for (i = 0; i < RCU_AWAKE_ATTEMPTS; i++) {
+               if (uatomic_read(value) & AWAKE_TEARDOWN)
+                       break;
+               caa_cpu_relax();
+       }
+       while (!(uatomic_read(value) & AWAKE_TEARDOWN))
+               poll(NULL, 0, 10);
+       assert(uatomic_read(value) & AWAKE_TEARDOWN);
+}
+
 static void wait_for_readers(struct cds_list_head *input_readers,
                        struct cds_list_head *cur_snap_readers,
                        struct cds_list_head *qsreaders)
@@ -198,6 +280,9 @@ void synchronize_rcu(void)
        CDS_LIST_HEAD(cur_snap_readers);
        CDS_LIST_HEAD(qsreaders);
        unsigned long was_online;
+       struct gp_waiters_thread gp_waiters_thread;
+       struct cds_wfs_head *gp_waiters_head;
+       struct cds_wfs_node *waiters_iter, *waiters_iter_n;
 
        was_online = URCU_TLS(rcu_reader).ctr;
 
@@ -214,8 +299,26 @@ void synchronize_rcu(void)
        else
                cmm_smp_mb();
 
+       /*
+        * Add ourself to gp_waiters stack of threads awaiting to wait
+        * for a grace period. Proceed to perform the grace period only
+        * if we are the first thread added into the stack.
+        */
+       cds_wfs_node_init(&gp_waiters_thread.node);
+       gp_waiters_thread.wait_futex = AWAKE_WAITING;
+       if (cds_wfs_push(&gp_waiters, &gp_waiters_node) != 0) {
+               /* Not first in stack: will be awakened by another thread. */
+               urcu_adaptative_busy_wait(&gp_waiters_thread.wait_futex);
+               goto gp_end;
+       }
+
        mutex_lock(&rcu_gp_lock);
 
+       /*
+        * Pop all waiters into our local stack head.
+        */
+       gp_waiters_head = __cds_wfs_pop_all(&gp_waiters);
+
        if (cds_list_empty(&registry))
                goto out;
 
@@ -272,6 +375,19 @@ void synchronize_rcu(void)
 out:
        mutex_unlock(&rcu_gp_lock);
 
+       /* Wake all waiters in our stack head, excluding ourself. */
+       cds_wfs_for_each_blocking_safe(gp_waiters_head, waiters_iter,
+                               waiters_iter_n) {
+               struct gp_waiters_thread *wt;
+
+               wt = caa_container_of(waiters_iter,
+                               struct gp_waiters_thread, node);
+               if (wt == &gp_waiters_thread)
+                       continue;
+               urcu_adaptative_wake_up(&wt->wait_futex);
+       }
+
+gp_end:
        /*
         * Finish waiting for reader threads before letting the old ptr being
         * freed.
@@ -286,6 +402,9 @@ void synchronize_rcu(void)
 {
        CDS_LIST_HEAD(qsreaders);
        unsigned long was_online;
+       struct gp_waiters_thread gp_waiters_thread;
+       struct cds_wfs_head *gp_waiters_head;
+       struct cds_wfs_node *waiters_iter, *waiters_iter_n;
 
        was_online = URCU_TLS(rcu_reader).ctr;
 
@@ -299,7 +418,26 @@ void synchronize_rcu(void)
        else
                cmm_smp_mb();
 
+       /*
+        * Add ourself to gp_waiters stack of threads awaiting to wait
+        * for a grace period. Proceed to perform the grace period only
+        * if we are the first thread added into the stack.
+        */
+       cds_wfs_node_init(&gp_waiters_thread.node);
+       gp_waiters_thread.wait_futex = AWAKE_WAITING;
+       if (cds_wfs_push(&gp_waiters, &gp_waiters_thread.node) != 0) {
+               /* Not first in stack: will be awakened by another thread. */
+               urcu_adaptative_busy_wait(&gp_waiters_thread.wait_futex);
+               goto gp_end;
+       }
+
        mutex_lock(&rcu_gp_lock);
+
+       /*
+        * Pop all waiters into our local stack head.
+        */
+       gp_waiters_head = __cds_wfs_pop_all(&gp_waiters);
+
        if (cds_list_empty(&registry))
                goto out;
 
@@ -334,6 +472,19 @@ void synchronize_rcu(void)
 out:
        mutex_unlock(&rcu_gp_lock);
 
+       /* Wake all waiters in our stack head, excluding ourself. */
+       cds_wfs_for_each_blocking_safe(gp_waiters_head, waiters_iter,
+                               waiters_iter_n) {
+               struct gp_waiters_thread *wt;
+
+               wt = caa_container_of(waiters_iter,
+                               struct gp_waiters_thread, node);
+               if (wt == &gp_waiters_thread)
+                       continue;
+               urcu_adaptative_wake_up(&wt->wait_futex);
+       }
+
+gp_end:
        if (was_online)
                rcu_thread_online();
        else