crope: "dvb_usb_start_feed: adap=0 active_fe=-1 feed_type=0 setting pid [no]: 2000 (8192) at index 0" <- this explains my EINVAL issue from yesterday (active_fe=-1 is rejected by dvb-usb-v2's dvb_usb_start_feed obviously)
got to dig into the VLC source code now to see why :(
any skystar driver expert in here? =)
(right now)
maybe i should write on some ML
crope: I read somewhere that you are using vlc to watch TV - what playlist type are you feeding it?
crope: I have the strange feeling that dvb-usb-v2 devices are not working with vlc
or it may not even be restricted to dvb-usb-v2 devices - it may simple break when you want to use frontendN where N != 0
xdarklight: you cannot set frontend number on vlc
it is always 0
crope: yep, just discovered that - until some version it auto-detected the correct frontend: https://github.com/videolan/vlc/commit/ea4e38b539b4c0a4b36d047292a84ad92b6bed25
22:59 < crope> xdarklight: can you test it using vlc?
22:59 < crope> $ dvbv5-zap "Yle TV1 HD" --adapter=0 --frontend=1 --record
22:59 < crope> $ vlc --dvb-budget-mode dvb://
crope: dvbv5-zap + vlc in budget mode works fine
also tvheadend is able to receive everything
so only that vlc problem is left, but that is a vlc problem...
crope: is there any way to see if merbanan's slave TS PID filter patch is working correctly?
I have it applied now, I can watch DVB-C HD channels without stuttering etc.
is there any other way to tell if it's working as it should?
xdarklight: use dvbv5-zap with monitor mode
monitor should only shows pids that are allowed to pass filter
is your usb bw so limited that you cannot stream whole stream?
it is around 50Mbit/sec at maximum T2 or C modulation
no, I just read somewhere that pid filtering is required for slave TS on rtl2832
no it is not
you already said vlc with dvb-budget-mode was working. dvb-budget-mode == no pid filters
dvb-budget-mode takes whole TS and does software filtering
crope: https://paste.kde.org/pehn9kqds -> https://paste.kde.org/pa61xrico
so pid filter doesn't seem to be working
almost 50Mbit/sec, thats quite much maximum possible stream size
(there are 4 channels at 122MHz, and there are 4 entries with large p/s count)
yep, it's actually 4 HD channels
larger pids are video
medium are audio, those smallest ones are some other data like tables
if you enable hw pid filters there is only those selected pids shown on that monitor listing
crope: "if you enable hw pid filters" <- I thought these are enabled automatically with dvbv5-zap if the device supports it?
no
those are only enabled if driver wants to enable or you set some moduleparameter. application cannot make decision
application can just request some pids from kernel demuxer and by default demuxer does software pid filtering
crope: ah, so that's what force_pid_filter_usage in dvb_usb_v2 is for, right?
yes
those pid filters are pretty much useless unless you are using old usb1.1. as almost all usb ports are nowadays usb2.0 or usb3 pid filters are not used
crope: thanks for that explanation!
crope: one last question: I want to send some linux-media patches, what should I use as base revision (linux-tree.git/master)?
s/linux-tree/media_tree/
yes
media master
crope: ok, any idea then when this is going to be applied: https://patchwork.linuxtv.org/patch/34312/ ?
I have a patch that auto-selects a few drivers for dvb_usb_rtl28xxu, and mn88472 would be one of the drivers
(and I'm not sure if I'm allowed auto auto-select drivers which are still in staging)
xdarklight: it will go to the kernel 4.8. it is too late for 4.7
crope: ok, so I'm assuming my patch will go into 4.8 as well -> I'll add the select
xdarklight: ah yes, both 72 and 73 demods are missin
as well si2168
yeah, I already figured them all out
crope: https://paste.kde.org/pzxnbrcnb <-going to send that one in a minute
xdarklight: many of those tuners are not used even driver probes so you can remove
MT2063 MT2266 MXL5007T
crope: that was last minute - patches are now sent!