> /var/log/mythtv/mythbackend.log contains a cycling of these messages:
> 
> ...
> 2008-02-16 15:59:32.445 MPEGRec(/dev/video0) Error: select timeout - 
> ivtv driver has stopped responding
> 2008-02-16 15:59:38.052 MPEGRec(/dev/video0) Error: select timeout - 
> ivtv driver has stopped responding
> ...


In my Nov 2007 version of MythTV, that timeout is 5 seconds.  Every time
you see that message MythTV has waited for 5 seconds for the cx18 driver
to report it has data ready to read.  There's a driver/kernel problem.
The logic in the driver to support select() by the kernel is implemented
in: cx18-fileops.c: cx18_v4l2_enc_poll()


> dmesg contains a large number of these messages:
> 
> cx18-0 file: Encoder poll started capture
> cx18-0 file: open encoder MPEG
> cx18-0 file: Encoder poll started capture
> cx18-0 file: open encoder MPEG

When MythTV times out on the select(), it closes the fd opened
on /dev/videoN and reopens it on the next attempt to get video data.
These messages reflect that behavior of MythTV and of the cx18 driver
allowing select() to restart a capture in
cx18-fileops.c:cx18_v4l2_enc_poll()



The big question is why does the select() on a cx18 device node when
video data should be there?  I guess I'm going to have to start
debugging the IRQ and DMA to see when/why the data stops flowing from
the encoder chip. 

I still get the occasional select() timeout and freeze, but it has been
usually resolving itself.


> Fedora 8, single Pentium 4, 2.6G, 2G RAM, cx18 data:

I have dual core:

$ grep Core /proc/cpuinfo
model name      : AMD Athlon(tm) 64 X2 Dual Core Processor 4200+
model name      : AMD Athlon(tm) 64 X2 Dual Core Processor 4200+

so that might make some difference in our results.

-Andy


_______________________________________________
ivtv-devel mailing list
[email protected]
http://ivtvdriver.org/mailman/listinfo/ivtv-devel

Reply via email to