libglusterfs: Add monotonic clocking counter for timer thread

gettimeofday() returns the current wall clock time and timezone.
Using these functions in order to measure the passage of time
(how long an operation took) therefore seems like a no-brainer.

This time suffer's from some limitations:

a. They have a low resolution: “High-performance” timing by
definition, requires clock resolutions into the microseconds
or better.

b. They can jump forwards and backwards in time: Computer
clocks all tick at slightly different rates, which causes
the time to drift. Most systems have NTP enabled which
periodically adjusts the system clock to keep them in sync
with “actual” time. The adjustment can cause the clock to
suddenly jump forward (artificially inflating your timing
numbers) or jump backwards (causing your timing calculations
to go negative or hugely positive). In such cases timer
thread could go into an infinite loop.

From 'man gettimeofday':
----------
..
..
The time returned by gettimeofday() is affected by discontinuous
jumps in the system time (e.g., if the system administrator manually
changes the system time).  If you need a monotonically increasing
clock, see clock_gettime(2).
..
..
----------

Rationale:

For calculating interval timing for Timer thread, all that’s
needed should be clock as a simple counter that increments
at a stable rate.

This is necessary to avoid the jumps which are caused by using
"wall time", this counter must be monotonic that can never
“tick” backwards, ever.

Change-Id: I701d31e71a85a73d21a6c5cd15583e7a5a645eeb
BUG: 1017993
Signed-off-by: Harshavardhana <harsha@harshavardhana.net>
Reviewed-on: http://review.gluster.org/6070
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Anand Avati <avati@redhat.com>

5 files changed, 107 insertions, 18 deletions

diff --git a/libglusterfs/src/Makefile.am b/libglusterfs/src/Makefile.am
index 65ff2654827..2f2ab97697b 100644
--- a/libglusterfs/src/Makefile.am
+++ b/libglusterfs/src/Makefile.am
@@ -15,7 +15,7 @@ CONTRIB_BUILDDIR = $(top_builddir)/contrib
 libglusterfs_la_SOURCES = dict.c xlator.c logging.c \
 	hashfn.c defaults.c common-utils.c timer.c inode.c call-stub.c \
 	compat.c fd.c compat-errno.c event.c mem-pool.c gf-dirent.c syscall.c \
-	iobuf.c globals.c statedump.c stack.c checksum.c daemon.c \
+	iobuf.c globals.c statedump.c stack.c checksum.c daemon.c timespec.c \
 	$(CONTRIBDIR)/rbtree/rb.c rbthash.c store.c latency.c \
 	graph.c $(CONTRIBDIR)/uuid/clear.c $(CONTRIBDIR)/uuid/copy.c \
 	$(CONTRIBDIR)/uuid/gen_uuid.c $(CONTRIBDIR)/uuid/pack.c \
@@ -33,7 +33,7 @@ nodist_libglusterfs_la_SOURCES = y.tab.c graph.lex.c gf-error-codes.h
 
 BUILT_SOURCES = graph.lex.c
 
-noinst_HEADERS = common-utils.h defaults.h dict.h glusterfs.h hashfn.h \
+noinst_HEADERS = common-utils.h defaults.h dict.h glusterfs.h hashfn.h timespec.h \
 	logging.h xlator.h stack.h timer.h list.h inode.h call-stub.h compat.h \
 	fd.h revision.h compat-errno.h event.h mem-pool.h byte-order.h \
 	gf-dirent.h locking.h syscall.h iobuf.h globals.h statedump.h \
diff --git a/libglusterfs/src/timer.c b/libglusterfs/src/timer.c
index 97a746b91d7..0ad6ea59fc2 100644
--- a/libglusterfs/src/timer.c
+++ b/libglusterfs/src/timer.c
@@ -17,19 +17,18 @@
 #include "logging.h"
 #include "common-utils.h"
 #include "globals.h"
-
-#define TS(tv) ((((unsigned long long) tv.tv_sec) * 1000000) + (tv.tv_usec))
+#include "timespec.h"
 
 gf_timer_t *
 gf_timer_call_after (glusterfs_ctx_t *ctx,
-                     struct timeval delta,
+                     struct timespec delta,
                      gf_timer_cbk_t callbk,
                      void *data)
 {
         gf_timer_registry_t *reg = NULL;
         gf_timer_t *event = NULL;
         gf_timer_t *trav = NULL;
-        unsigned long long at = 0L;
+        uint64_t at = 0;
 
         if (ctx == NULL)
         {
@@ -48,10 +47,8 @@ gf_timer_call_after (glusterfs_ctx_t *ctx,
         if (!event) {
                 return NULL;
         }
-        gettimeofday (&event->at, NULL);
-        event->at.tv_usec = ((event->at.tv_usec + delta.tv_usec) % 1000000);
-        event->at.tv_sec += ((event->at.tv_usec + delta.tv_usec) / 1000000);
-        event->at.tv_sec += delta.tv_sec;
+        timespec_now (&event->at);
+        timespec_adjust_delta (event->at, delta);
         at = TS (event->at);
         event->callbk = callbk;
         event->data = data;
@@ -127,7 +124,7 @@ void *
 gf_timer_proc (void *ctx)
 {
         gf_timer_registry_t *reg = NULL;
-	const struct timespec sleepts = {.tv_sec = 1, .tv_nsec = 0, };
+        const struct timespec sleepts = {.tv_sec = 1, .tv_nsec = 0, };
 
         if (ctx == NULL)
         {
@@ -142,14 +139,14 @@ gf_timer_proc (void *ctx)
         }
 
         while (!reg->fin) {
-                unsigned long long now;
-                struct timeval now_tv;
+                uint64_t now;
+                struct timespec now_ts;
                 gf_timer_t *event = NULL;
 
-                gettimeofday (&now_tv, NULL);
-                now = TS (now_tv);
+                timespec_now (&now_ts);
+                now = TS (now_ts);
                 while (1) {
-                        unsigned long long at;
+                        uint64_t at;
                         char need_cbk = 0;
 
                         pthread_mutex_lock (&reg->lock);
diff --git a/libglusterfs/src/timer.h b/libglusterfs/src/timer.h
index 2954f6aff5d..2f963adbf55 100644
--- a/libglusterfs/src/timer.h
+++ b/libglusterfs/src/timer.h
@@ -25,7 +25,7 @@ typedef void (*gf_timer_cbk_t) (void *);
 
 struct _gf_timer {
         struct _gf_timer *next, *prev;
-        struct timeval    at;
+        struct timespec    at;
         gf_timer_cbk_t    callbk;
         void             *data;
         xlator_t         *xl;
@@ -44,7 +44,7 @@ typedef struct _gf_timer_registry gf_timer_registry_t;
 
 gf_timer_t *
 gf_timer_call_after (glusterfs_ctx_t *ctx,
-                     struct timeval delta,
+                     struct timespec delta,
                      gf_timer_cbk_t cbk,
                      void *data);
 
diff --git a/libglusterfs/src/timespec.c b/libglusterfs/src/timespec.c
new file mode 100644
index 00000000000..0df74e61c20
--- /dev/null
+++ b/libglusterfs/src/timespec.c
@@ -0,0 +1,68 @@
+/*
+  Copyright (c) 2013 Red Hat, Inc. <http://www.redhat.com>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#include <stdio.h>
+#include <inttypes.h>
+#if defined GF_LINUX_HOST_OS || defined GF_SOLARIS_HOST_OS || defined GF_BSD_HOST_OS
+#include <time.h>
+#include <sys/time.h>
+#endif
+
+#if defined GF_DARWIN_HOST_OS
+#include <mach/mach_time.h>
+#endif
+
+#include "logging.h"
+#include "time.h"
+
+
+void tv2ts (struct timeval tv, struct timespec *ts)
+{
+        ts->tv_sec  = tv.tv_sec;
+        ts->tv_nsec = tv.tv_usec * 1000;
+}
+
+void timespec_now (struct timespec *ts)
+{
+#if defined GF_LINUX_HOST_OS || defined GF_SOLARIS_HOST_OS || defined GF_BSD_HOST_OS
+
+        if (0 == clock_gettime(CLOCK_MONOTONIC, ts))
+                return;
+        else {
+                struct timeval tv;
+                if (0 == gettimeofday(&tv, NULL))
+                        tv2ts(tv, ts);
+        }
+#elif defined GF_DARWIN_HOST_OS
+        mach_timebase_info_data_t tb = { 0 };
+        static double timebase = 0.0;
+        uint64_t time = 0;
+        mach_timebase_info (&tb);
+
+        timebase *= info.numer;
+        timebase /= info.denom;
+
+        time = mach_absolute_time();
+        time *= timebase;
+
+        ts->tv_sec = (time * NANO);
+        ts->tv_nsec = (time - (ts.tv_sec * GIGA));
+
+#endif /* Platform verification */
+        gf_log_callingfn ("timer", GF_LOG_DEBUG, "%"PRIu64".%09"PRIu64,
+                          ts->tv_sec, ts->tv_nsec);
+}
+
+void timespec_adjust_delta (struct timespec ts, struct timespec delta)
+{
+        ts.tv_nsec = ((ts.tv_nsec + delta.tv_nsec) % 1000000000);
+        ts.tv_sec += ((ts.tv_nsec + delta.tv_nsec) / 1000000000);
+        ts.tv_sec += delta.tv_sec;
+}
diff --git a/libglusterfs/src/timespec.h b/libglusterfs/src/timespec.h
new file mode 100644
index 00000000000..c584c82c092
--- /dev/null
+++ b/libglusterfs/src/timespec.h
@@ -0,0 +1,24 @@
+/*
+  Copyright (c) 2013 Red Hat, Inc. <http://www.redhat.com>
+  This file is part of GlusterFS.
+
+  This file is licensed to you under your choice of the GNU Lesser
+  General Public License, version 3 or any later version (LGPLv3 or
+  later), or the GNU General Public License, version 2 (GPLv2), in all
+  cases as published by the Free Software Foundation.
+*/
+
+#ifndef __INCLUDE_TIMESPEC_H__
+#define __INCLUDE_TIMESPEC_H__
+
+#include <stdint.h>
+
+#define TS(ts)  ((ts.tv_sec * 1000000000LL) + ts.tv_nsec)
+#define NANO (+1.0E-9)
+#define GIGA UINT64_C(1000000000)
+
+void tv2ts (struct timeval tv, struct timespec *ts);
+void timespec_now (struct timespec *ts);
+void timespec_adjust_delta (struct timespec ts, struct timespec delta);
+
+#endif /*  __INCLUDE_TIMESPEC_H__ */