feat: P2-3 repl_wait_ack() with pthread_cond_timedwait + ACK deadlock prevention
Implements sync replication wait: primary blocks until at least one non-catching-up standby ACKs the target LSN, or times out. Uses pthread_cond_timedwait with CLOCK_REALTIME deadline. Inner loop rechecks all standbys after each wakeup to guard against spurious wakeups. ACK deadlock prevention: repl_on_ack() is called exclusively from the io_worker read callback (not vtaskmgr), so it only does atomic_store + pthread_cond_signal — safe to call while vcpu threads block in repl_wait_ack(). 4 new TDD tests (all green, 9 total in test_repl_unit): - test_wait_ack_no_standbys_returns_enodev - test_wait_ack_already_acked - test_wait_ack_timeout (verifies >=80ms elapsed) - test_wait_ack_woken_by_background_ack (pthread thread signals after 30ms)
This commit is contained in:
parent
abefc8a128
commit
d326ee1b2e
2 changed files with 144 additions and 5 deletions
|
|
@ -201,12 +201,61 @@ int repl_broadcast(replicator_t *r, const wal_entry_hdr_t *hdr,
|
|||
|
||||
int repl_wait_ack(replicator_t *r, uint64_t target_lsn, uint32_t timeout_ms)
|
||||
{
|
||||
(void)r;
|
||||
(void)target_lsn;
|
||||
(void)timeout_ms;
|
||||
if (!r) return -EINVAL;
|
||||
|
||||
/* Full implementation in P2-3 */
|
||||
return -ENODEV;
|
||||
/* Check whether any non-catching-up standby exists */
|
||||
mylock_lock(&r->lock);
|
||||
int has_standby = 0;
|
||||
for (int i = 0; i < MAX_STANDBYS; i++) {
|
||||
if (r->standbys[i].ep != NULL &&
|
||||
!atomic_load_explicit(&r->standbys[i].in_catch_up,
|
||||
memory_order_acquire)) {
|
||||
has_standby = 1;
|
||||
break;
|
||||
}
|
||||
}
|
||||
mylock_unlock(&r->lock);
|
||||
|
||||
if (!has_standby) return -ENODEV;
|
||||
|
||||
/* Compute absolute deadline for pthread_cond_timedwait (CLOCK_REALTIME) */
|
||||
struct timespec deadline;
|
||||
clock_gettime(CLOCK_REALTIME, &deadline);
|
||||
deadline.tv_sec += (long)(timeout_ms / 1000);
|
||||
deadline.tv_nsec += (long)(timeout_ms % 1000) * 1000000L;
|
||||
if (deadline.tv_nsec >= 1000000000L) {
|
||||
deadline.tv_sec++;
|
||||
deadline.tv_nsec -= 1000000000L;
|
||||
}
|
||||
|
||||
pthread_mutex_lock(&r->ack_lock);
|
||||
int rc = -ETIMEDOUT;
|
||||
while (1) {
|
||||
/* Recheck every standby after each wakeup (spurious-wakeup guard) */
|
||||
mylock_lock(&r->lock);
|
||||
for (int i = 0; i < MAX_STANDBYS; i++) {
|
||||
standby_conn_t *s = &r->standbys[i];
|
||||
if (s->ep == NULL) continue;
|
||||
if (atomic_load_explicit(&s->in_catch_up, memory_order_acquire))
|
||||
continue;
|
||||
if (atomic_load_explicit(&s->acked_lsn, memory_order_acquire)
|
||||
>= target_lsn) {
|
||||
rc = 0;
|
||||
break;
|
||||
}
|
||||
}
|
||||
mylock_unlock(&r->lock);
|
||||
if (rc == 0) break;
|
||||
|
||||
int wait_rc = pthread_cond_timedwait(&r->ack_cond, &r->ack_lock,
|
||||
&deadline);
|
||||
if (wait_rc == ETIMEDOUT) {
|
||||
rc = -ETIMEDOUT;
|
||||
break;
|
||||
}
|
||||
}
|
||||
pthread_mutex_unlock(&r->ack_lock);
|
||||
return rc;
|
||||
}
|
||||
|
||||
/* ============================================================
|
||||
|
|
|
|||
|
|
@ -1,7 +1,10 @@
|
|||
#include <assert.h>
|
||||
#include <errno.h>
|
||||
#include <stdint.h>
|
||||
#include <string.h>
|
||||
#include <stdio.h>
|
||||
#include <pthread.h>
|
||||
#include <time.h>
|
||||
/* Use the same Unity include pattern as other controller tests */
|
||||
#include <unity/unity.h>
|
||||
#include "repl.h"
|
||||
|
|
@ -74,6 +77,89 @@ static void test_repl_trim_lsn_disconnected_slot_expiry(void) {
|
|||
printf("test_repl_trim_lsn_disconnected_slot_expiry: PASS\n");
|
||||
}
|
||||
|
||||
/* ============================================================
|
||||
* P2-3: repl_wait_ack() tests
|
||||
* ============================================================ */
|
||||
|
||||
/* Test 1: repl_wait_ack returns -ENODEV when no standbys connected */
|
||||
static void test_wait_ack_no_standbys_returns_enodev(void) {
|
||||
replicator_t r;
|
||||
repl_init(&r);
|
||||
int rc = repl_wait_ack(&r, /*target_lsn=*/10, /*timeout_ms=*/50);
|
||||
TEST_ASSERT_EQUAL_INT(-ENODEV, rc);
|
||||
repl_fini(&r);
|
||||
printf("test_wait_ack_no_standbys_returns_enodev: PASS\n");
|
||||
}
|
||||
|
||||
/* Test 2: repl_wait_ack returns 0 when standby already ACKed target LSN */
|
||||
static void test_wait_ack_already_acked(void) {
|
||||
replicator_t r;
|
||||
int dummy_ep = 0;
|
||||
repl_init(&r);
|
||||
repl_add_standby(&r, (void*)&dummy_ep, "sb", 7001);
|
||||
/* Pre-set acked_lsn to 20 (already past target 10) */
|
||||
repl_on_ack(&r, (void*)&dummy_ep, 20);
|
||||
int rc = repl_wait_ack(&r, 10, 50);
|
||||
TEST_ASSERT_EQUAL_INT(0, rc);
|
||||
repl_fini(&r);
|
||||
printf("test_wait_ack_already_acked: PASS\n");
|
||||
}
|
||||
|
||||
/* Test 3: repl_wait_ack times out when standby never ACKs */
|
||||
static void test_wait_ack_timeout(void) {
|
||||
replicator_t r;
|
||||
int dummy_ep = 0;
|
||||
repl_init(&r);
|
||||
repl_add_standby(&r, (void*)&dummy_ep, "sb", 7001);
|
||||
/* acked_lsn stays at 0 — never reaches target 100 */
|
||||
struct timespec start, end;
|
||||
clock_gettime(CLOCK_MONOTONIC, &start);
|
||||
int rc = repl_wait_ack(&r, 100, 100 /* 100ms timeout */);
|
||||
clock_gettime(CLOCK_MONOTONIC, &end);
|
||||
TEST_ASSERT_EQUAL_INT(-ETIMEDOUT, rc);
|
||||
/* Verify we actually waited ~100ms (not returned immediately) */
|
||||
long elapsed_ms = (end.tv_sec - start.tv_sec) * 1000
|
||||
+ (end.tv_nsec - start.tv_nsec) / 1000000;
|
||||
TEST_ASSERT_TRUE(elapsed_ms >= 80); /* at least 80ms */
|
||||
repl_fini(&r);
|
||||
printf("test_wait_ack_timeout: PASS\n");
|
||||
}
|
||||
|
||||
/* Test 4: ACK from background thread wakes repl_wait_ack */
|
||||
struct ack_args {
|
||||
replicator_t *r;
|
||||
void *ep;
|
||||
uint64_t lsn;
|
||||
int delay_ms;
|
||||
};
|
||||
|
||||
static void* ack_sender_thread(void* arg) {
|
||||
struct ack_args* a = arg;
|
||||
struct timespec ts = { 0, (long)a->delay_ms * 1000000L };
|
||||
nanosleep(&ts, NULL);
|
||||
repl_on_ack(a->r, a->ep, a->lsn);
|
||||
return NULL;
|
||||
}
|
||||
|
||||
static void test_wait_ack_woken_by_background_ack(void) {
|
||||
replicator_t r;
|
||||
int dummy_ep = 0;
|
||||
repl_init(&r);
|
||||
repl_add_standby(&r, (void*)&dummy_ep, "sb", 7001);
|
||||
|
||||
struct ack_args args = { &r, (void*)&dummy_ep, 15, 30 /* 30ms delay */ };
|
||||
pthread_t th;
|
||||
pthread_create(&th, NULL, ack_sender_thread, &args);
|
||||
|
||||
/* Wait for LSN 15 with 200ms timeout — ACK arrives after 30ms */
|
||||
int rc = repl_wait_ack(&r, 15, 200);
|
||||
pthread_join(th, NULL);
|
||||
|
||||
TEST_ASSERT_EQUAL_INT(0, rc);
|
||||
repl_fini(&r);
|
||||
printf("test_wait_ack_woken_by_background_ack: PASS\n");
|
||||
}
|
||||
|
||||
void test_repl_unit(void) {
|
||||
UNITY_BEGIN();
|
||||
RUN_TEST(test_repl_init_zeroes_nstandbys);
|
||||
|
|
@ -81,5 +167,9 @@ void test_repl_unit(void) {
|
|||
RUN_TEST(test_repl_add_standby_max);
|
||||
RUN_TEST(test_repl_trim_lsn_no_standbys);
|
||||
RUN_TEST(test_repl_trim_lsn_disconnected_slot_expiry);
|
||||
RUN_TEST(test_wait_ack_no_standbys_returns_enodev);
|
||||
RUN_TEST(test_wait_ack_already_acked);
|
||||
RUN_TEST(test_wait_ack_timeout);
|
||||
RUN_TEST(test_wait_ack_woken_by_background_ack);
|
||||
UNITY_END();
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue