[commit: ghc] master: Make `timer_create(CLOCK_REALTIME)` autoconf test more reliable (8ad9e74)

[git at git.haskell.org]

Repository : ssh://git@git.haskell.org/ghc

On branch  : master
Link       : http://ghc.haskell.org/trac/ghc/changeset/8ad9e74f1f88d5c86d2e76f9992f9b2d267421d3/ghc

>---------------------------------------------------------------

commit 8ad9e74f1f88d5c86d2e76f9992f9b2d267421d3
Author: Herbert Valerio Riedel <hvr at gnu.org>
Date:   Tue Nov 17 15:30:56 2015 +0100

    Make `timer_create(CLOCK_REALTIME)` autoconf test more reliable
    
    I've noticed that on a platform with a coarse timer/scheduling
    granularity of 10ms this autoconf tests fails to detect a working
    `timer_create(CLOCK_REALTIME)`.
    
    On AIX, this effectively means that intervals/timers are rounded up to
    multiples of 10ms, so a 13ms delay is effectively a 20ms delay.
    
    By using a 100ms timeout we are on the safe side.
    
    Reviewers: austin, bgamari
    
    Reviewed By: bgamari
    
    Subscribers: thomie, erikd
    
    Differential Revision: https://phabricator.haskell.org/D1483


>---------------------------------------------------------------

8ad9e74f1f88d5c86d2e76f9992f9b2d267421d3
 aclocal.m4 | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/aclocal.m4 b/aclocal.m4
index 2bf27be..33f05e5 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1591,17 +1591,18 @@ int main(int argc, char *argv[])
         exit(2);
     }
 
+    tock = 0;
+
     it.it_value.tv_sec = 0;
-    it.it_value.tv_nsec = 1000000;
+    it.it_value.tv_nsec = 1000000; // 1ms
     it.it_interval = it.it_value;
     if (timer_settime(timer, 0, &it, NULL) != 0) {
         fprintf(stderr,"settime problem\n");
         exit(4);
     }
 
-    tock = 0;
-
-    usleep(3000);
+    // some environments have coarse scheduler/timer granularity of ~10ms and worse
+    usleep(100000); // 100ms
 
     if (!tock) {
         fprintf(stderr,"no CLOCK_REALTIME signal\n");