ASSI via devel writes:
> is… intriguing. If you follow the code through the function, ts_min in
> the log should always be ts_prev+sys_fuzz; these two values can't be
> identical or differ by any other value, ts_min must strictly be greater
> than ts_prev by sys_fuzz. I notice that there is always a DNS operation
> logged at the same time when I get these errors, so I am pretty sure
> this is related to the DNS thread. This would also explain why a server
> that has many clients would see many more such errors.
While I have not yet identified the code path that is involved (likely
something in NTS, I think), you might want to try the following monkey
patch:
Subject: [PATCH] Wrap get_systime in critical section to serialize
timestamping among multiple threads
---
libntp/systime.c | 18 ++++++++++++++++++
1 file changed, 18 insertions(+)
diff --git a/libntp/systime.c b/libntp/systime.c
index 8efea6cfe..5e87aee15 100644
--- a/libntp/systime.c
+++ b/libntp/systime.c
@@ -6,6 +6,8 @@
*/
#include "config.h"
+#include <pthread.h>
+
#include "ntp.h"
#include "ntp_syslog.h"
#include "ntp_stdlib.h"
@@ -209,14 +211,30 @@ normalize_time(
/*
* get_systime - return system time in NTP timestamp format.
*/
+static pthread_mutex_t systime_lock = PTHREAD_MUTEX_INITIALIZER;
+
void
get_systime(
l_fp *now /* system time */
)
{
+ struct timespec ts_lock = { 0, 0 };
+ struct timespec ts_delta = { 0, 0 };
struct timespec ts; /* seconds and nanoseconds */
+ if (pthread_mutex_trylock(&systime_lock)) {
+ get_ostime(&ts_lock);
+ pthread_mutex_lock(&systime_lock);
+ }
get_ostime(&ts);
+ ts_delta = ts;
normalize_time(ts, sys_fuzz > 0.0 ? ntp_random() : 0, now);
+ pthread_mutex_unlock(&systime_lock);
+ if (ts_lock.tv_nsec != 0 || ts_lock.tv_sec != 0) {
+ ts_delta = sub_tspec(ts_delta, ts_lock);
+ msyslog(LOG_WARNING,
+ "CLOCK: we were blocked for %ld.%09ld s while waiting for another thread",
+ (long)ts_delta.tv_sec, ts_delta.tv_nsec);
+ }
}
--
2.24.0
Let me know if makes a difference on your system. The error crops up
too sparingly on mine to really tell if it helps or not.
Regards,
Achim.
--
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+
SD adaptations for Waldorf Q V3.00R3 and Q+ V3.54R2:
http://Synth.Stromeko.net/Downloads.html#WaldorfSDada
_______________________________________________
devel mailing list
devel@ntpsec.org
http://lists.ntpsec.org/mailman/listinfo/devel
