<!-- Some styling for better description lists --><style type='text/css'>dt { font-weight: bold;float: left;display:inline;margin-right: 1em} dd { display:block; margin-left: 2em}</style>
***: Voce changes his alias to mchehab
<br> whyz_ changes his alias to whyz
swiftgeek: Is this the right channel for RTL2832u DVB-T regressions?
iive: that's the right channel for dvb.
swiftgeek: but for kernel module regression too? happened on 3.19->4.0 (unrelated bug forced me to stay on lts)
iive: well, there are developers around. but they might not be looking at their irc clients atm.
<br> sometimes you need to be patient.
swiftgeek: http://permalink.gmane.org/gmane.linux.drivers.video-input-infrastructure/88157 it looks in keywords kinda similar
<br> i don't get errors after disabling remote
iive: and yes. there are mostly kernel developers. most of the drivers are in kernel and they maintain the media branch that gets merged in the main.
swiftgeek: i always get surprised by how many times this dongle breaks and nobody notices for few releases, because everybody is on rtlsdr ;D
<br> but this is different completely for sure
<br> automatic bandwidth selection in vlc doesn't work anymore, seemingly selecting bandwidth=0
<br> " rtl2832 7-0010: invalid bandwidth_hz 0"
iive: i think i've seen that too. I thought it is vlc problem.
swiftgeek: don't want to be a dick but "do not break userspace"
<br> :P
<br> i know, rule doesn't apply after so much time passed
<br> huuuuh
<br> this is getting weird
<br> so behaviour has changed
<br> low quality signal gets bandwidth set to 0
iive: this doesn't make sense. bandwidth is something you need in order to lock the demodulator. quality is function of the demodulation.
swiftgeek: i can only tell this one thing
<br> changing kernel makes things works perfectly fine
<br> best test case for immediately spotting difference would be not connecting antenna at all :D
<br> yup
<br> as soon as i have not perfect signal something chooses bandwidth 0
<br> since such bandwidth wouldn't make any sense i will assume that it's a new error code that wasn't there before
<br> k let's try now with IR remote, i thought that i had to disable that part
<br> with remote enabled it completely fails
<br> eg "r820t_read: i2c rd failed=-32 reg=00 len=2: 61 c8"
iive: i2c is bus that could be shared by more than one device.
swiftgeek: the point is that it's a regression anyway
iive: yeh...
<br> you are right again.
swiftgeek: and it's not the first time it is mentioned
<br> but since kernel trap is gone i'm lost
iive: i think you should really talk with some developer
swiftgeek: so better channel than this?
iive: mchehab, crope these i know .
<br> just stay around. they will came out, eventually.
swiftgeek: my experience with mailinglists isn't particularly good (leads to acid atmosphere in the same time as resolved on irc ;>)
iive: yeh...
swiftgeek: eg. radeon issue was fixed and patched upstream in just 3h
<br> (was killing boot)
iive: i could ask you to do a bisect, but i've bisected kernel before... the chance that rc0 contains bugs that hang your system is non trivial.
swiftgeek: if i had machine to bisect i would be doing that now
<br> unless i can compile this outside of tree?
iive: yes, i think the this branch have something like that.
<br> but i haven't used it.
swiftgeek: just to leave it here - if scope will be needed for debugging of this issue i has DS1052E so can do
iive: oscilloscope ?
swiftgeek: yup
iive: let's hope it won't came to that :)
swiftgeek: well it's light and very portable
<br> but it will be pain to setup it
iive: well, it also increase the risk of damaging your dongle...
swiftgeek: nope
<br> they all have SMA now
<br> also all R820T tuners are coming corroded and barely soldered
<br> so one has to desolder, use sandpaper and then solder back on the board ;D
<br> this is why i love QFN packages so much
<br> there is no way you could possibly damage package in any sane way
iive: hehe... ok.
<br> you know what you do :)
swiftgeek: oh
<br> those dongles are used for rtl-sdr
<br> but thanks to DVB-T i could buy more of them
iive: btw, check if the old (working) kernel supports remote at all.
<br> what is rtl-sdr?
swiftgeek: software defined radio driver (userspace)
<br> and old working kernel supports remote fine
<br> no idea actually about the new one
<br> i have three identical dongles, two very accessible at any time
<br> i have a feel that one issue is directly hw/sw related (i^2c/remote/tuner one) and the one with selecting bandwidth is of that do not break userspace kind
<br> but too dumb to do more than guess
iive: it's definitely software related.
swiftgeek: i'm trying to understand device trees now
<br> i mean like bad timing causing hw to glitch
<br> like my recently discovered gate phone / lock that doesn't accept nearly half of iButtons
<br> i should probably go back to device trees :<
<br> reverse eng. working binary -> source
<br> opensource.: source -> working binary