https://bugs.kde.org/show_bug.cgi?id=368907
Szymon Stefanek <pra...@kvirc.net> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |pra...@kvirc.net
Status|UNCONFIRMED |CONFIRMED
Ever confirmed|0 |1
--- Comment #27 from Szymon Stefanek <pra...@kvirc.net> ---
I have the same issue with version 2.0.10. Viewing the same DVB channel
sometimes uses low CPU and sometimes stays almost fixed at 100% (saturating one
core, I guess). Version 2.0.10 installed from
https://launchpad.net/~mtron/+archive/ubuntu/kaffeine-stable seems to be
affected just like the 2.0.5 from the official ubuntu repositories. In addition
2.0.10 also tends to spit out a lot of warnings like the following:
08-07-17 03:55:54.236 [Debug ] kaffeine.epg: channel invalid
08-07-17 03:55:54.263 [Debug ] kaffeine.epg: channel invalid
08-07-17 03:55:54.390 [Debug ] kaffeine.epg: table 0x4f, extension 0x2189,
session 1/1, size 383
08-07-17 03:55:54.392 [Warning ] "2 instead of 1 arguments to message {Country
%1: not rate...} supplied before conversion."
but these might be unrelated as they seem to appear also when CPU usage is low.
Another message I see often (in both versions) is:
[000000002d6da6e8] core input error: ES_OUT_SET_(GROUP_)PCR is called too late
(pts_delay increased to 488 ms)
[000000002d6da6e8] core input error: ES_OUT_RESET_PCR called
Hard to say if it's related.
Profiling shows similar results to the ones obtained by Jan Palus.
Most instruction fetches seem to happen inside src_process in libsamplerate.so.
An interesting fact is that when profiling via valgrind the CPU usage tends to
be lower than when running normally. 100% is very rare. So it's even possible
that the bug is not triggered when running under valgrind.
--
You are receiving this mail because:
You are watching all bug changes.
