patch 9.0.1411: accuracy of profiling is not optimal
Problem: Accuracy of profiling is not optimal.
Solution: Use CLOCK_MONOTONIC if possible. (Ernie Rael, closes #12129)
diff --git a/src/time.c b/src/time.c
index c82a79a..bcc7eef 100644
--- a/src/time.c
+++ b/src/time.c
@@ -163,7 +163,7 @@
tm->LowPart = n2;
# else
tm->tv_sec = n1;
- tm->tv_usec = n2;
+ tm->tv_fsec = n2;
# endif
return error ? FAIL : OK;
}
@@ -222,7 +222,7 @@
n2 = res.LowPart;
# else
n1 = res.tv_sec;
- n2 = res.tv_usec;
+ n2 = res.tv_fsec;
# endif
list_append_number(rettv->vval.v_list, (varnumber_T)n1);
list_append_number(rettv->vval.v_list, (varnumber_T)n2);
@@ -258,6 +258,7 @@
void
f_reltimestr(typval_T *argvars UNUSED, typval_T *rettv)
{
+ static char buf[50];
# ifdef FEAT_RELTIME
proftime_T tm;
# endif
@@ -269,7 +270,15 @@
return;
if (list2proftime(&argvars[0], &tm) == OK)
+ {
+# ifdef MSWIN
rettv->vval.v_string = vim_strsave((char_u *)profile_msg(&tm));
+# else
+ long usec = tm.tv_fsec / (TV_FSEC_SEC / 1000000);
+ sprintf(buf, "%3ld.%06ld", (long)tm.tv_sec, usec);
+ rettv->vval.v_string = vim_strsave((char_u *)buf);
+# endif
+ }
else if (in_vim9script())
emsg(_(e_invalid_argument));
# endif
@@ -392,7 +401,7 @@
static long last_timer_id = 0;
/*
- * Return time left until "due". Negative if past "due".
+ * Return time left, in "msec", until "due". Negative if past "due".
*/
long
proftime_time_left(proftime_T *due, proftime_T *now)
@@ -409,7 +418,7 @@
if (now->tv_sec > due->tv_sec)
return 0;
return (due->tv_sec - now->tv_sec) * 1000
- + (due->tv_usec - now->tv_usec) / 1000;
+ + (due->tv_fsec - now->tv_fsec) / (TV_FSEC_SEC / 1000);
# endif
}