[mythtv] [patch] Recording HDTV when no signal causes segfault

[John Patrick Poet]

I see the same behaviour as Wendy.  Started with the switch to 2.6.x kernel
drivers -- at least that is when I first noticed it.

I was going to try and investigate more, before reporting the problem.

I have not tried the 1.3 version of the 2.6.6 kernel driver on the pcHDTV
site yet.

Daniel, I don't think my patch for the segfault should hurt anything -- I
don't see any reason to call JOIN if the thread was never started in the
first place.  I am not sure why you did not see the problem before.  Once
again, I did not notice this problem until after switching to a 2.6 kernel.


John
--

On Tue, 25 May 2004, Daniel Thor Kristjansson wrote:

>
> Well I had a cursory look and it doesn't seem to be due to TuneTo()
> being broken in any obvious way. Unfortunately, I won't have the time to
> debug this until at least Thursday night, and then I'm away Friday -
> Tuesday.
>
> I don't get the segfaults John Poet reported so at least this seems to
> be a frontend only problem for me. ATSC channels seem to go off the air
> a lot more frequently than NTSC channels so it's still a significant
> problem, but I wouldn't hold up 0.15 for this HDTV only problem in light
> of the DataDirect migration deadline.
>
> -- Daniel
>
> On Tue, 25 May 2004, Wendy Seltzer wrote:
>
> ]I'm also having pcHDTV problems in the last few days of CVS.  If myth
> ]starts off recording from a channel, it works fine.  If it starts off
> ]by trying to tune live TV (on the same channel), it fails and leaves
> ]the card in a state where dtvsignal can't reset it.
> ]
> ]If there's anything I can help test, please let me know.
> ]
> ]--Wendy
> ]
> ]At 3:52 PM -0400 5/25/04, Daniel Thor Kristjansson wrote:
> ]>It isn't segfaulting on a bad channel here, but it does seem to be
> ]>entering some kind of invalid state. First, the recorder started when
> ]>it shouldn't have. Of course, it didn't record any video, but while
> ]>'mythtv' complained about the lack of video, it didn't return to
> ]>'mythfrontend' which I think it should have after the 15 second timeout.
> ]>I'll look at it tonight.
> ]>
> ]>-- Daniel
> ]>
> ]>On Tue, 25 May 2004, Isaac Richards wrote:
> ]>
> ]>]On Tuesday 25 May 2004 12:03 am, John Patrick Poet wrote:
> ]>]> With HDTV, the signal strength is checked when tunning to a channel.
> ]>]> When recording, if the signal strength is too low, the encoder thread is
> ]>]> never created, but the pthread_join call was still being made.  This
> ]>]> caused a segfault.
> ]>]>
> ]>]> The attached patch prevents the pthread_join call from being made, if
> ]>]> the encoder thread was not created.
> ]>]
> ]>]I've applied this, but I had to change the signal check stuff on recordings
> ]>](double channel changes are bad sometimes) -- could you please test to make
> ]>]sure it's still working on hdtv?
> ]>]
> ]>]Isaac
> ]
> ]
>