pg_time_now(). This uses INSTR_TIME_SET_CURRENT in it, but this macro
can call clock_gettime(CLOCK_MONOTONIC[_RAW], ) instead of gettimeofday
or clock_gettime(CLOCK_REALTIME, ). When CLOCK_MONOTONIC[_RAW] is used,
clock_gettime doesn't return epoch time. Therefore, we can use
INSTR_TIME_SET_CURRENT aiming to calculate a duration, but we should
not have used this to get the current timestamp.
I think we can fix this issue by using gettimeofday for logging as before
this was changed. I attached the patch.
I cannot say that I'm thrilled by having multiple tv stuff back in several
place. I can be okay with one, though. What about the attached? Does it
make sense?
--
Fabien.
diff --git a/src/bin/pgbench/pgbench.c b/src/bin/pgbench/pgbench.c
index d7479925cb..5fe2a9286f 100644
--- a/src/bin/pgbench/pgbench.c
+++ b/src/bin/pgbench/pgbench.c
@@ -672,11 +672,11 @@ static const PsqlScanCallbacks pgbench_callbacks = {
static inline pg_time_usec_t
pg_time_now(void)
{
- instr_time now;
+ struct timeval tv;
- INSTR_TIME_SET_CURRENT(now);
+ gettimeofday(&tv, NULL);
- return (pg_time_usec_t) INSTR_TIME_GET_MICROSEC(now);
+ return tv.tv_sec * 1000000 + tv.tv_usec;
}
static inline void
