[mythtv] [mythtv-commits] mythtv commit: r19341 by jarod

Otto Kolsi otto at kolsi.fi
Sat Dec 13 16:56:01 UTC 2008


Same here (actually ~200% CPU in dualcore environment). System is in 
Finland with DVB-T.

Peter Stokes wrote:
> This commit causes my mythbackend to consume 100% CPU :-(
> 
> My system uses EIT data from DVB-T transmissions in the UK.
> 
> Please let me know if you require any log output.
> 
> Thanks
> 
> Peter
> 
> 
> On 12 Dec 2008, at 03:04, mythtv at cvs.mythtv.org wrote:
> 
>>      Author: jarod
>>        Date: 2008-12-12 03:04:58 +0000 (Fri, 12 Dec 2008)
>> New Revision: 19341
>>   Changeset: http://cvs.mythtv.org/trac/changeset/19341
>>
>> Modified:
>>
>>   trunk/mythtv/libs/libmythtv/mpeg/mpegstreamdata.cpp
>>
>> Log:
>>
>> mpeg: don't try to overwrite any of the incoming packet buffer
>>
>> In the event the mpeg stream parser fails to process a given packet,
>> it was overwriting the current packet's sync byte w/a non-sync-byte
>> value, to signal that we should resync the stream. This works with
>> many buffers, but not those produced by the new(er) Linux FireWire
>> stack, as the incoming buffer is write-only for the controller.
>> Trying to overwrite the buffer results in a segfault. This works
>> around the issue by simply adding a resync flag, and now I've got
>> my cable box working pretty darned reliably (about 20 consecutive
>> livetv start/stops, with multiple channel changes, and I haven't
>> hit the segfault, which used to trigger roughly 60% of the time).



More information about the mythtv-dev mailing list