summaryrefslogtreecommitdiff
path: root/ppsthread.c
diff options
context:
space:
mode:
authorGary E. Miller <gem@rellim.com>2013-11-26 17:09:12 -0800
committerGary E. Miller <gem@rellim.com>2013-11-26 17:09:12 -0800
commit64f447f4780cbf58e259dd538b313fc63ced8be9 (patch)
tree6fbcb866c6ee88ef9474743c546b947bbb92361f /ppsthread.c
parent34dff6b9f7c8cc878131bda3e178603e86c7263f (diff)
downloadgpsd-64f447f4780cbf58e259dd538b313fc63ced8be9.tar.gz
Improve ppsthread comments.
Diffstat (limited to 'ppsthread.c')
-rw-r--r--ppsthread.c8
1 files changed, 6 insertions, 2 deletions
diff --git a/ppsthread.c b/ppsthread.c
index 8d9ef068..d8265c51 100644
--- a/ppsthread.c
+++ b/ppsthread.c
@@ -322,12 +322,16 @@ static /*@null@*/ void *gpsd_ppsmonitor(void *arg)
* The timestamp has already been captured in the kernel, and we
* are merely fetching it here.
*/
- /* on a quad core 2.4GHz Xeon this removes about 20uS of
- * latency, and about +/-5uS of jitter over the other method */
+ /* on a quad core 2.4GHz Xeon using KPPS timestamp instead of plain
+ * PPS timestamp removes about 20uS of latency, and about +/-5uS
+ * of jitter
+ */
memset( (void *)&kernelpps_tv, 0, sizeof(kernelpps_tv));
#else /* not TIOMCIWAIT */
/*
* RFC2783 specifies that a NULL timeval means to wait.
+ *
+ * FIXME, this will fail on 2Hz 'PPS', should wait 3 Sec.
*/
kernelpps_tv.tv_sec = 1;
kernelpps_tv.tv_nsec = 0;