------- You are receiving this mail because: -------
You are on the CC list for the bug.
http://bugs.exim.org/show_bug.cgi?id=1426
--- Comment #6 from Git Commit <git@???> 2014-08-11 19:17:15 ---
Git commit:
http://git.exim.org/exim.git/commitdiff/0f8ba377cb458ec0c5de18a450189e701f710a1e
commit 0f8ba377cb458ec0c5de18a450189e701f710a1e
Author: Jeremy Harris <jgh146exb@???>
AuthorDate: Tue May 27 21:50:41 2014 +0100
Commit: Jeremy Harris <jgh146exb@???>
CommitDate: Tue Jul 22 23:06:15 2014 +0100
Do not sleep for tiny periods, or hang trying to sleep for zero. Bug 1426
---
src/src/exim.c | 10 ++++++++--
1 files changed, 8 insertions(+), 2 deletions(-)
diff --git a/src/src/exim.c b/src/src/exim.c
index 6a23364..51daa55 100644
--- a/src/src/exim.c
+++ b/src/src/exim.c
@@ -267,6 +267,10 @@ will wait for ever, so we panic in this instance. (There
was a case of this
when a bug in a function that calls milliwait() caused it to pass invalid
data.
That's when I added the check. :-)
+We assume it to be not worth sleeping for under 100us; this value will
+require revisiting as hardware advances. This avoids the issue of
+a zero-valued timer setting meaning "never fire".
+
Argument: an itimerval structure containing the interval
Returns: nothing
*/
@@ -276,6 +280,9 @@ milliwait(struct itimerval *itval)
{
sigset_t sigmask;
sigset_t old_sigmask;
+
+if (itval->it_value.tv_usec < 100 && itval->it_value.tv_sec == 0)
+ return;
(void)sigemptyset(&sigmask); /* Empty mask */
(void)sigaddset(&sigmask, SIGALRM); /* Add SIGALRM */
(void)sigprocmask(SIG_BLOCK, &sigmask, &old_sigmask); /* Block SIGALRM */
@@ -310,8 +317,7 @@ struct itimerval itval;
itval.it_interval.tv_sec = 0;
itval.it_interval.tv_usec = 0;
itval.it_value.tv_sec = msec/1000;
-if ((itval.it_value.tv_usec = (msec % 1000) * 1000) == 0)
- itval.it_value.tv_usec = 1;
+itval.it_value.tv_usec = (msec % 1000) * 1000;
milliwait(&itval);
}
--
Configure bugmail:
http://bugs.exim.org/userprefs.cgi?tab=email
