[mythtv-users] mythtv v30 'Could not open decoder' error

[Mike Bibbings]

On 31/01/2019 14:47, Tim Pletcher wrote:
> On Tue, Jan 29, 2019 at 2:11 PM Tim Pletcher <pletchtd at gmail.com 
> <mailto:pletchtd at gmail.com>> wrote:
>
>     On Tue, Jan 29, 2019 at 10:28 AM John <jksjdevelop at gmail.com
>     <mailto:jksjdevelop at gmail.com>> wrote:
>
>         I also get this error on a remote frontend with version 30.
>         Some timeout must need a tweak. Note the issue has been there
>         for several weeks it is not due to a very recent change. I
>         don't use live tv enough to make it a major issue for me.
>
>
>
>     Good to know that the problem is not unique to me.  I only updated
>     to v30 on Sunday the day prior so it is likely that I simply
>     didn't stumble upon it in the limited use of the system since then.
>
>     According to this in the log:
>
>     Jan 28 19:28:13 k-mint mythfrontend.real: mythfrontend[1273]: I CoreContext avformatdecoder.cpp:2772 (OpenAVCodec) AFD: Opened codec 0x55a532b95180, id(AC3) type(Audio)
>     Jan 28 19:28:13 k-mint mythfrontend.real: mythfrontend[1273]: E CoreContext avformatdecoder.cpp:2635 (ScanStreams) AFD: Unknown video codec - defaulting to MPEG2
>
>     It appears that this is related to fallback to software based
>     decoding rather than GPU Hardware based decoding. Since this is
>     occurring intermittently, you are probably right that it is a
>     timing issue.
>
>     Looking at avformatdecoder.cpp and totally speculating because i
>     have no experience with C++, I wonder if is related to
>     SEQ_PKT_ERR_MAX10 (line 108) which is a criterion used to apply
>     logic to fall back to software decoding? Is 10 packets perhaps
>     sometimes too sensitive when changing channels via LiveTV?
>
> I pulled down the source, increased the SEQ_PKT_ERR_MAX value to 15 in 
> avformatdecoder.cpp, and recompiled the frontend.  Since doing that, I 
> have not experienced the 'Could not open decoder' error for the last 
> three days now when using LiveTV.  I have not noticed any other 
> deleterious effects though I will continue to monitor for any.
>
> On a separate topic, when I recompile the frontend from source, the 
> mythfrontend binary I generate is an order of magnitude larger than 
> the one in the mythbuntu repo and I can't seem to figure out why.  
> Looking and comparing the ./configure options from the mythbuntu build 
> package with the ones I am using, the only tangible difference i see 
> is that my system is using NASM rather than YASM for the compilation.  
> Can anyone educate me on what is causing this difference?
>
> -Tim
>
>

Tim,

I tried your change to  SEQ_PKT_ERR_MAX on current mythtv master and it 
did not resolve the problem. I also tried other values of 50 and 1000, 
again no resolution.

For reference my system is Xubuntu 18.04 with Hauppauge QuadHD DVB-T/T2  
using current Hauppauge kernel from the ppa.

My workround is attached, it disables error checking in 
avformatdecoder.cpp, and only affects LiveTV.


Mike

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mythtv.org/pipermail/mythtv-users/attachments/20190131/3caff0a3/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: disable_avformatdecoder_return_FAIL_DEBUGONLY.patch
Type: text/x-patch
Size: 613 bytes
Desc: not available
URL: <http://lists.mythtv.org/pipermail/mythtv-users/attachments/20190131/3caff0a3/attachment.bin>