A few weeks back, I reported timekeeping problems with the new ivtv version running. Here's a recap and workaround.
My main system is an AMD64 3000+ on an ECS NForce4-A motherboard with a PVR 500, running FC5 and updated with ATrpms. Running with the newest ivtv, I got the following in /var/log/messages when mythbackend was recording a program: Mar 3 21:27:45 nero kernel: warning: many lost ticks. Mar 3 21:27:45 nero kernel: Your time source seems to be instable or some driver is hogging interupts Mar 3 21:27:45 nero kernel: rip __do_softirq+0x53/0xd5 And the system time lost about 2.5 seconds every minute: nero# sh -c 'while ntpdate -b 192.168.1.3; do sleep 10; done' 3 Mar 21:27:43 ntpdate[3422]: step time server 192.168.1.3 offset 1.218678 sec 3 Mar 21:27:54 ntpdate[3426]: step time server 192.168.1.3 offset 0.406938 sec 3 Mar 21:28:04 ntpdate[3440]: step time server 192.168.1.3 offset 0.394011 sec 3 Mar 21:28:14 ntpdate[3443]: step time server 192.168.1.3 offset 0.411024 sec For a while, I went back to old ivtv, but the current mythtv and video4linux seem to need the new version, otherwise codec or playback params seem confused, resulting in high-pitched audio with lots of dropouts. So I tried the new stuff again tonight and noticed another kernel message: Mar 3 22:36:25 nero kernel: ivtv1 warning: CX2341X_ENC_SET_VBI_LINE took 411 jiffies (1000 per HZ) I also noticed that time slippage didn't happen if I ran "mplayer /dev/video1" before mythbackend recorded anything on that device. So I went to mythtvsetup and removed closed captioning from the VBI setup. Voila, no more time slippage... _______________________________________________ ivtv-devel mailing list ivtv-devel@ivtvdriver.org http://ivtvdriver.org/mailman/listinfo/ivtv-devel
