Hi PƔdraig,
> Hopefully the attached avoids this.
Your patch avoids the first 3 "fail=1" occurrences, but not the fourth one:
+ timeout 2.34e+5d sleep 0
+ test 124 = 0
+ fail=1
But is this really the desirable workaround? Do people want to have
"timeout 2.34e+5d sleep 0" really do different things on Linux/hppa
than on Linux/x86?
Why not add a workaround to the 'timeout' program, rather than to the tests?
Your guess about an overflow seems correct:
$ ./timeout 821760555 sleep 0 ; echo $?
t = 1325723090, duration_ceil = 821760555
0
$ ./timeout 821760557 sleep 0 ; echo $?
t = 1325723093, duration_ceil = 821760557
124
The limit appears to be INT_MAX.
The following patch fixes it for me.
2012-01-04 Bruno Haible <address@hidden>
timeout: Fix behaviour for end times beyond 2038 on Linux/hppa.
* src/timeout.c (settimeout) [Linux/hppa]: Cap the duration, to avoid
an overflow.
--- src/timeout.c.bak 2012-01-01 10:04:06.000000000 +0100
+++ src/timeout.c 2012-01-05 01:36:38.405115464 +0100
@@ -111,6 +111,13 @@
deprecated by POSIX. Instead we fallback to single second
resolution provided by alarm(). */
+#if defined __linux__ && defined __hppa__
+ /* Work around an overflow that happens when the end date is beyond 2038.
*/
+ time_t t = time (NULL);
+ if (t >= 0 && t <= INT_MAX && duration > (double) (INT_MAX - t))
+ duration = (double) (INT_MAX - t);
+#endif
+
#if HAVE_TIMER_SETTIME
struct timespec ts = dtotimespec (duration);
struct itimerspec its = { {0, 0}, ts };