<!-- 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>
crope: nox-: pong
nox-: moin crope, have you noticed reception issues with some rtl28xxu tuners recently?
<br> someone it seems the driver is less sensitive now...
<br> (with 0x1d19:0x1101 and nooelec 0x0bda:0x2838 at least, others still seem ok)
<br> <u>working</u>: 4f4274af7009890f0d4724909bf9038193955489, broken: 4f671fe2f9523a1ea206f63fe60a7c7b3a56d5c7
<br> nooelec gets corrupted streams after channel switch, 0x1d19:0x1101 receives nothing anymore at all
<br> s/someone/somehow/
crope: there is no commit 4f671fe2f9523a1ea206f63fe60a7c7b3a56d5c7
<br> nox-: which tree has commit 4f671fe2f9523a1ea206f63fe60a7c7b3a56d5c7 ?
nox-: git://github.com/torvalds/linux.git media_tree apparently
<br> or is that the wrong one?
crope: nox-: which is commit name and date?
nox-: hmm thats what hps wrote
<br> guess ill ask him
crope: I did a lot of changes recently to rtl2832u driver. some potential mistakes could be for example register caching, some volatile register is cached
nox-: i c...
<br> this is supposed to be a 4.0rc tree
<br> i seem to have https://github.com/torvalds/linux/blob/master/drivers/media/dvb-frontends/rtl2832.c
crope: nox-: could you remove register caching and try if it helps
nox-: where is that?
crope: dev->regmap_config.cache_type = REGCACHE_RBTREE,
<br> change it to dev->regmap_config.cache_type = REGCACHE_NONE,
nox-: hmm nope, thats even worse
<br> now even the rtl28xxu tuners that still worked are broken
crope: how many tuners you have?
<br> all rtl2832u devices were not broken, but only one?
nox-: yeah two were broken
<br> i have 4 or 5 or so
<br> <nox-> someone it seems the driver is less sensitive now...
<br> <nox-> (with 0x1d19:0x1101 and nooelec 0x0bda:0x2838 at least, others still seem ok)
crope: those numbers does not say anything
nox-: 0x0ccd:0x00d3 now is broken too
crope: which are RF tuner IC used?
nox-: oh
<br> r820t was one broken one
<br> https://wiki.freebsd.org/WebcamCompat - mine are in ther
<br> e
crope: what is tuners for those others?
nox-: http://rtlsdr.org/hardware-usb
<br> E4000 now is broken too
<br> (after your patch)
<br> that one still worked before
<br> FC2580 might be the other broken one
<br> (somehow i thought it was FC0013, hmm)
<br> 0x1d19 0x1101 FC2580 Dexatek DK DVB-T Dongle (Logilink VG0002A)
FSanches: <u>mchehab</u>: Hello, here's Felipe Sanches. I talked to you at some conference (probably at FISL) a few years ago.
<br> <u>mchehab</u>: I was working on a userspace driver for a Zinwell ISDB-T 1-Seg USB receiver
<br> <u>mchehab</u>: it was like 5 years ago, I guess.
mchehab: hi FSanches
FSanches: Recently I found the device in a box here in my house and got interested in working on it again
<br> I improved the code a bit but still did not get it to work properly
<br> I also started to port it to kernel-space
<br> I'm commiting to a fork of your linux-media tree hosted at my github account here: https://github.com/felipesanches/linux-media/commits/zinwell
<br> I do have access to the datasheet of the tuner (Maxim MAX2163) but I lack documentation on the other 2 chips:
<br> MegaChips MA50159
<br> and Somagic SMI-2020CBE
crope: nox-: for me it seems to work, though I saw some I2C errors when used VLC... I think there could not be big issue as I havent got any bug reports...
FSanches: so, it means my work is based on USB sniffer logs of the device while it was being controlled by the crappy proprietary windows program
nox-: well 4.0rc is still quite new so maybe you'll get the reports later
mchehab: <u>FSanches</u>: ok. usb sniffing should work
<br> especially for 1seg
FSanches: <u>mchehab</u>: Lucas Villa Real is also working with me on this
nox-: crope, and i tested using vdr...
FSanches: maybe you know him...
mchehab: yes, I know him
FSanches: :-)
mchehab: don't hear about him for a while
FSanches: So... it seems to me that the SMI-2020CBE chip is the one responsible for working as an i2c bridge over USB to allow the driver to communicate with the other 2 chips
mchehab: it would be nice if you had the demod datasheet
crope: nox-: I tested with VLC and dvbscan. And I have surely tested it using tzap, dvbv5-zap and dvbv5-scan when I did those changes as those are the tools I am using. usually I will get bug reports in a few days after patches are committed to media/master if there is big issues
mchehab: as isdb-t has lots of parameters to set there
<br> and it should be a little hard to get
<br> yet, thankfully, most demods work in "auto" mode
nox-: crope, lets me test using vlc too
<br> crope, getting corrupted streams with r820t
<br> with vlc
mchehab: getting everything they need from the TMCC carrier
crope: nox-: is your signal very weak?
nox-: it is kinda weak but worked with eaerlier driver versions
FSanches: let me ask you something simple... Is this i2c bridging over usb functionality the thing that characterizes a "dvb_frontend" in the code ?
nox-: crope, as i said it looks like the driver has become less sensitive
<br> still waiting for picture with e4000...
<br> ..on vlc
FSanches: As far as I can tell (by searching the current media codebase) this chip (Somagic SMI2020CBE) is still not supported by Linux so I'll have to add some code for that. Should I start fresh or should I append it to an existing source module? I mean... it is hard to classify this one grouped to other already supported devices. For instance, is there any ISDB-T device already supported?
crope: nox-: e4000 works too. what it reports as a NNR?
<br> CNR
<br> Lock (0x1f) Signal= 93.73% C/N= 27.25dB postBER= 0
nox-: where do i see that with vlc?
crope: vlc does not show any signal statistics
<br> use dvbv5-zap
nox-: ah ok
<br> ill ise vdr-pluging-femon
<br> use
<br> no str reading, snd 00f3
<br> snr
<br> (0%)
crope: Lock (0x1f) Signal= 92.94% C/N= 32.67dB postBER= 0
<br> first CNR was e4000 and second one r820t
nox-: mind you with e4000 this is only broken with your oneliner patch
<br> let me remove that and test again
<br> now vdr works again but snr is still as low
<br> ..and vlc gets no picture, only vdr does
<br> vlc is weird, i dont even see how i switch channels...
<br> (i get picture with r820t w/o your oneliner patch)
crope: you have to make playlist first which contains all the channels. after that could could pick channel from playlist
nox-: hm ok i just entered mux freq...
<br> but anyway, vlc is broken with e4000 here
FSanches: <u>mchehab</u>: I just got my first good isochronous responses from the Zinwell ISDB-T device :-)
<br> I'll finish porting this userspace script to kernelspace and hopefully I'll get this thing working on Linux! I'm very excited about it! Woo-hoo!!!
nox-: crope, and your oneliner patch only makes reception worse here
<br> crope, so i suppose there are still bugs
<br> bbl
***: nox- has quit IRC (Quit: nox-)
lokita23: ola
<br> hola
<br> q tal
***: lokita23 has quit IRC (Quit: lokita23)
mchehab: <u>FSanches</u>: sorry, was busy with other things...
<br> a dvb_frontend is actually a tuner + demod
<br> (and SEC, for DVB-S/S2)
<br> except for max2163, you should write a new set of drivers, as those chips are different than the ones we use
<br> I think we have already a driver for max216x
<br> if so, the best is to try to reuse it