[Bug 2306] Keep pps hack for Win32 even if user-mode/loopback PPS API is activated...

bk: 50aa7cb32nlG230T4EKaLxIwO4rH8w

diff --git a/ChangeLog b/ChangeLog
index a2dde231e919e76bcb87f98baef0056db64bbf1a..edf346f106e9ea783db71f8f89ac94ce314afc30 100644 (file)
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,5 @@
+* [Bug 2306] keep pps hack for Win32 even if user-mode/loopback
+  PPS API is activated on a serial line.
 (4.2.7p324) 2012/11/19 Released by Harlan Stenn <stenn@ntp.org>
 * Reinstate doc fix to authentic.html from Mike T.
 * [Bug 1223] cleanup for rlimit changes.

diff --git a/ports/winnt/ntpd/ntp_iocompletionport.c b/ports/winnt/ntpd/ntp_iocompletionport.c
index 031c34c4b997f133db42a467329f06345158a388..ad35d5f52eb2436351001562b1bc642fb3585265 100644 (file)
--- a/ports/winnt/ntpd/ntp_iocompletionport.c
+++ b/ports/winnt/ntpd/ntp_iocompletionport.c
@@ -796,7 +796,16 @@ OnSerialWaitComplete(
                        ppsbuf->data = dev->pps_data;
                        InterlockedExchange((PLONG)&dev->cov_count, covc);
                        /* Note: InterlockedExchange() is a full barrier. */
-               } else {
+               }
+               /* perlinger@ntp.org, 2012-11-19
+               It can be argued that once you have the PPS API active, you can
+               disable the old pps hack. This would give a behaviour that's much
+               more like the behaviour under a UN*Xish OS. On the other hand, it
+               will give a nasty surprise for people which have until now happily
+               taken the pps hack for granted, and after the first complaint, I have
+               decided to keep the old implementation unconditionally. So here it is:
+               /*else*/ 
+               {
                        /* backward compat: 'usermode-pps-hack' */
                        if (MS_RLSD_ON & modem_status) {
                                lpo->DCDSTime = lpo->RecvTime;