Merge branch 'lukas/erts/cpu_time_thread/OTP-15090'

* lukas/erts/cpu_time_thread/OTP-15090:
  erts: Make cpu_timestamp use per thread on Linux

6 files changed, 21 insertions, 27 deletions

diff --git a/erts/aclocal.m4 b/erts/aclocal.m4
index a4d09810bd..99b96eb5bc 100644
--- a/erts/aclocal.m4
+++ b/erts/aclocal.m4
@@ -2645,7 +2645,7 @@ case $erl_gethrvtime in
 	dnl Check if clock_gettime (linux) is working
 	dnl
 
-	AC_MSG_CHECKING([if clock_gettime can be used to get process CPU time])
+	AC_MSG_CHECKING([if clock_gettime can be used to get thread CPU time])
 	save_libs=$LIBS
 	LIBS="-lrt"
 	AC_TRY_RUN([
@@ -2659,11 +2659,11 @@ case $erl_gethrvtime in
 	    int i;
 	    struct timespec tp;
 
-	    if (clock_gettime(CLOCK_PROCESS_CPUTIME_ID, &tp) < 0)
+	    if (clock_gettime(CLOCK_THREAD_CPUTIME_ID, &tp) < 0)
 	      exit(1);
 	    start = ((long long)tp.tv_sec * 1000000000LL) + (long long)tp.tv_nsec;
 	    for (i = 0; i < 100; i++)
-	      clock_gettime(CLOCK_PROCESS_CPUTIME_ID, &tp);
+	      clock_gettime(CLOCK_THREAD_CPUTIME_ID, &tp);
 	    stop = ((long long)tp.tv_sec * 1000000000LL) + (long long)tp.tv_nsec;
 	    if (start == 0)
 	      exit(4);
@@ -2686,7 +2686,7 @@ case $erl_gethrvtime in
 	case $erl_clock_gettime_cpu_time in
 		yes)
 			AC_DEFINE(HAVE_CLOCK_GETTIME_CPU_TIME,[],
-				  [define if clock_gettime() works for getting process time])
+				  [define if clock_gettime() works for getting thread time])
 			LIBRT=-lrt
 			;;
 		cross)
diff --git a/erts/emulator/beam/erl_bif_trace.c b/erts/emulator/beam/erl_bif_trace.c
index f9d351e69e..9861483bf0 100644
--- a/erts/emulator/beam/erl_bif_trace.c
+++ b/erts/emulator/beam/erl_bif_trace.c
@@ -643,12 +643,12 @@ Eterm erts_internal_trace_3(BIF_ALIST_3)
 			    SysTimespec tp;
 			    int i;
 			    
-			    if (sys_get_proc_cputime(start, tp) < 0)
+			    if (sys_get_cputime(start, tp) < 0)
 				goto error;
 			    start = ((SysCpuTime)tp.tv_sec * 1000000000LL) + 
 				    (SysCpuTime)tp.tv_nsec;
 			    for (i = 0; i < 100; i++)
-				sys_get_proc_cputime(stop, tp);
+				sys_get_cputime(stop, tp);
 			    stop = ((SysCpuTime)tp.tv_sec * 1000000000LL) + 
 				   (SysCpuTime)tp.tv_nsec;
 			    if (start == 0) goto error;
diff --git a/erts/emulator/beam/erl_time_sup.c b/erts/emulator/beam/erl_time_sup.c
index 4f91d9ad07..29c698e34f 100644
--- a/erts/emulator/beam/erl_time_sup.c
+++ b/erts/emulator/beam/erl_time_sup.c
@@ -1860,7 +1860,7 @@ void erts_get_now_cpu(Uint* megasec, Uint* sec, Uint* microsec) {
   SysCpuTime t;
   SysTimespec tp;
 
-  sys_get_proc_cputime(t, tp);
+  sys_get_cputime(t, tp);
   *microsec = (Uint)(tp.tv_nsec / 1000);
   t = (tp.tv_sec / 1000000);
   *megasec = (Uint)(t % 1000000);
diff --git a/erts/emulator/beam/erl_trace.c b/erts/emulator/beam/erl_trace.c
index f074dd8bdb..f4161b14f2 100644
--- a/erts/emulator/beam/erl_trace.c
+++ b/erts/emulator/beam/erl_trace.c
@@ -625,20 +625,6 @@ erts_get_system_profile(void) {
     return profile;
 }
 
-
-#ifdef HAVE_ERTS_NOW_CPU
-#  define GET_NOW(m, s, u) \
-do { \
-    if (erts_cpu_timestamp) \
-	erts_get_now_cpu(m, s, u); \
-    else \
-	get_now(m, s, u); \
-} while (0)
-#else
-#  define GET_NOW(m, s, u) do {get_now(m, s, u);} while (0)
-#endif
-
-
 static void
 write_sys_msg_to_port(Eterm unused_to,
 		      Port* trace_port,
diff --git a/erts/emulator/sys/unix/erl_unix_sys.h b/erts/emulator/sys/unix/erl_unix_sys.h
index 10adf80875..5bfe5a8e2d 100644
--- a/erts/emulator/sys/unix/erl_unix_sys.h
+++ b/erts/emulator/sys/unix/erl_unix_sys.h
@@ -292,6 +292,8 @@ erts_sys_perf_counter()
 
 /*
  * Functions for measuring CPU time
+ *
+ * Note that gethrvtime is time per process and clock_gettime is per thread.
  */
 
 #if (defined(HAVE_GETHRVTIME) || defined(HAVE_CLOCK_GETTIME_CPU_TIME))
@@ -300,15 +302,15 @@ typedef struct timespec SysTimespec;
 
 #if defined(HAVE_GETHRVTIME)
 #define sys_gethrvtime() gethrvtime()
-#define sys_get_proc_cputime(t,tp) (t) = sys_gethrvtime(), \
-                                   (tp).tv_sec = (time_t)((t)/1000000000LL), \
-                                   (tp).tv_nsec = (long)((t)%1000000000LL)
+#define sys_get_cputime(t,tp) (t) = sys_gethrvtime(), \
+        (tp).tv_sec = (time_t)((t)/1000000000LL),     \
+        (tp).tv_nsec = (long)((t)%1000000000LL)
 int sys_start_hrvtime(void);
 int sys_stop_hrvtime(void);
 
 #elif defined(HAVE_CLOCK_GETTIME_CPU_TIME)
 #define sys_clock_gettime(cid,tp) clock_gettime((cid),&(tp))
-#define sys_get_proc_cputime(t,tp) sys_clock_gettime(CLOCK_PROCESS_CPUTIME_ID,(tp))
+#define sys_get_cputime(t,tp) sys_clock_gettime(CLOCK_THREAD_CPUTIME_ID,(tp))
 
 #endif
 #endif
diff --git a/lib/tools/doc/src/fprof.xml b/lib/tools/doc/src/fprof.xml
index 4c9e48045e..72624bd33b 100644
--- a/lib/tools/doc/src/fprof.xml
+++ b/lib/tools/doc/src/fprof.xml
@@ -328,10 +328,16 @@
            purposes. This option is only
            allowed with the <c>start</c> option.</item>
           <tag><c>cpu_time</c>| <c>{cpu_time, bool()}</c></tag>
-          <item>The options <c>cpu_time</c> or <c>{cpu_time, true></c>
+          <item>The options <c>cpu_time</c> or <c>{cpu_time, true}</c>
            makes the timestamps in the trace be in CPU time instead
            of wallclock time which is the default. This option is
-           only allowed with the <c>start</c> option.</item>
+           only allowed with the <c>start</c> option.
+           <warning><p>Getting correct values out of cpu_time can be difficult.
+           The best way to get correct values is to run using a single
+           scheduler and bind that scheduler to a specific CPU,
+           i.e. <c>erl +S 1 +sbt db</c>.</p>
+           </warning>
+          </item>
           <tag><c>{procs, PidSpec}</c>| <c>{procs, [PidSpec]}</c></tag>
           <item>Specifies which processes that shall be traced. If
            this option is not given, the calling process is