Monday, 2017-02-20

*** tpb has joined #timvideos00:00
tumbleweedCarlFK: hit the reset button please (not a power cycle)00:02
CarlFKtumbleweed: whoops, in trying to see about that.. I unplugged it00:03
CarlFKand plugged it back in.. about 5 sec later.  hmm...00:03
CarlFKtumbleweed: still want me to find the reset button?00:29
CarlFKI think the button on the case might be plugged into a header that likely was deemed "reset"00:30
tumbleweedCarlFK: naah, we're good00:30
tumbleweedalmost done here00:30
tumbleweedCarlFK: now please00:33
tumbleweed(reset button, that is)00:34
tumbleweedloaded the lm32 firmware now...00:34
CarlFKcan you put pattern on out 0 so I can see if00:34
CarlFKhitting the button does anything ?00:34
tumbleweedwe don't have it booted up enough to talk to it00:35
tumbleweedso no, I can't00:35
tumbleweedyou could also power cycle if finding the reset button is too hard00:36
CarlFKI hit th button, a blue led came on00:37
CarlFKnow I have garbage on my lcd00:37
tumbleweedlooks booted00:37
tumbleweedyay garbage?00:37
CarlFKoh, the garabe is moving!00:37
* tumbleweed sends pattern out00:37
tumbleweedok, you have an LCA firmware on there now00:38
CarlFKalthough the glitch art was ... art!00:38
tumbleweedwas the glitch art static?00:38
CarlFKum.. static..00:40
tumbleweedwe're thinking it probably just doesn't initialize the buffer ram on boot, then00:40
CarlFKit ws random bytes and every few seconds it would change00:40
tumbleweedit was booting up on passthrough mode00:40
CarlFKrandom bytes not changing00:40
tumbleweedI mean, 1 -> {0, 1, e}00:41
CarlFKyou realize static has 2 very different meanings in this context00:41
tumbleweedyep :P00:41
tumbleweedbut I've debugged the thing I cuold only debug with an opsis, so thanks00:41
tumbleweednow to see if I can squeeze this into the next debian release (we're already frozen)00:42
CarlFKwould now be a good time to ask for my atlys to be flashed ?00:42
CarlFKI don't need them until July or something00:43
mithroCarlFK: If you want an Atlys that doesn't work, sure!00:43
tumbleweedmithro says he hasn't been home since this farmware was written, so he hasn' ttested on atlys00:43
CarlFKsounds like I should wait00:43
tumbleweedit does00:43
CarlFKno prob - just checking00:43
CarlFKmithro: did you hear I found a local FPGA guy that is interested in helping?00:44
mithroCarlFK: I'll believe it when they actually do something....00:44
CarlFKsure.. but still.. it is exciting ....00:45
CarlFKhe said stuff like "it is easy to find fpga ip. the tricky part if finding stuff that works.  bla bla bla the first thing I will do is figure out how to test it."00:45
CarlFKdamm.  I had an Atlys in my car at the time.  I would have loaned it to him if I would have thought about it00:46
mithroCarlFK: "find FPGA ip" -- that is not what we do really00:47
CarlFKeven for like the serial io?00:47
CarlFKthat was my example of stuff that works but could work better00:47
mithroCarlFK: The serial IO stuff is all on the FX2 side of things, not the FPGA00:48
CarlFKoh.  whoops.00:48
tpbTitle: Check that version is working properly in the debian packages · Issue #50 · timvideos/HDMI2USB-mode-switch · GitHub (at
*** IRCFrEAK has joined #timvideos01:37
*** IRCFrEAK has left #timvideos01:37
tpbTitle: Include udev rules in the sdist by stefanor · Pull Request #52 · timvideos/HDMI2USB-mode-switch · GitHub (at
*** puck`` is now known as puck03:32
tpbTitle: Add --reset-gateware by mithro · Pull Request #54 · timvideos/HDMI2USB-mode-switch · GitHub (at
mithrotumbleweed: Need to test that and then we can merge03:39
xfxfthaytan: you mentioned @ LCA there was probably a missing sync option in our gstreamer ingests which might explain the desync issues04:14
xfxf(copy pasta from a week ago, not sure if you saw - i'm doing another conf in a fortnight so keen to fix)04:15
CarlFKxfxf: are you sure it was ingest and not core or gui?04:39
*** madgoat has joined #timvideos04:44
*** madgoat has left #timvideos04:44
*** SamSagaZ_ has quit IRC05:14
thaytanxfxf, who can remember things from so long ago?05:26
xfxfCarlFK: i have no idea05:26
xfxfthaytan: lol, do you have any idea what you might have meant?05:27
xfxfyou mentioned this on the last day of the conf05:27
CarlFKpretty sure it had to to with the gui elements getting out of sync05:27
xfxfi believe you said we were missing some option on our ingests that actually tells the sinks to stay in sync with the clock master05:27
CarlFKlike the preview lagging behind the master05:27
xfxflike, we have the option of using a clock master, but we haven't actually enabled everything to sync to it, or something05:27
tpbTitle: voctomix-outcasts/ at master · CarlFK/voctomix-outcasts · GitHub (at
xfxfis what we're using05:28
CarlFKim gussing it is the "or soemthing" part of that  ;)05:28
thaytanI don't remember what I meant, but I'll have a think about it05:28
xfxfyeah, i had zero headspace to do anything about it at the time, but i am super keen now to actually fix something if there's a glaring problem05:28
xfxfthaytan: thx05:28
thaytanxfxf, I said a thing about how you don't set the base_time anywhere, so while the capture devices will all share a clock with a common rate and won't drift over time, they won't all have the same initial offset05:42
thaytanso they may start with some desync and then hold onto it05:42
xfxfah, right05:42
thaytanand I think there was also a thing about the gui using an audiosink with sync=false?05:42
xfxfthis probably explains why i have to do the alsa audio delay thing05:42
xfxfis base_time an option i'll see using gst-inspect?05:42
thaytanno, it's a pipeline configuration thing05:43
thaytanthe basetime on a pipeline sets the common 0 reference on the clock for all elements05:43
thaytanto capture on different processes / machines and be able to exactly correlate captured data, you need to distribute the basetime across them all05:43
thaytanor keep it simple and use a basetime of 0 everywhere05:44
thaytanthen the timestamps on frames are actual wall clock times05:44
thaytanxfxf, we also talked about me setting up a test system here  sometime05:52
thaytanmaybe this weekend05:52
*** CarlFK has quit IRC06:07
*** CarlFK has joined #timvideos06:30
*** ChanServ sets mode: +v CarlFK06:30
paddatrappermithro: thanks06:36
paddatrappermithro: I'll try add some stuff about the FX2LP to the docs this evening06:40
*** puck is now known as puck``09:51
xfxfthaytan: humm, actually, are you around on Saturday evening and do you have a spare bed/couch?11:03
xfxfi'm heading to Winton on Sunday and was about to look for accom in shepparton... you're the same distance in the other direction11:03
xfxfcould come early arvo on Sat and drag up an image11:03
xfxfif not, the whole thing is ansible'd now, so i should be able to tell you how to get an image going locally11:03
xfxfthe whole thing _should_ be up on various github repo's but I don't know how and go it is (it assumes a local PXE debian installer)11:04
thaytanxfxf, the same distance in the other direction, except then you're twice as far away from your destination?11:04
xfxfthaytan: driving an extra hour on Sat doesn't phase me11:05
xfxfi used to do melb<>adel trips every two weeks or so11:05
xfxfs/how and go/how run and go/, rather11:06
xfxfshepparton is kinda a wierd diversion in any case, but closest place that has hotels with accom11:06
thaytanxfxf, you're welcome to come stay11:06
xfxfapparently some tennis thing on in benalla this weekend11:06
xfxfonly if you want a hand getting a test install running :)11:07
xfxfyou got a spare laptop?11:07
xfxfi have some spare core 2 duo's around too (which will do 1x 720p stream but no more - i.e. single source into voctocore only) - i use those as spare machines to test hdmi2usb things on11:07
thaytanxfxf, there's spare machines11:09
thaytanone 6 core i7 for the server11:10
thaytanand core 2 duo and/or raspberry pis for inputs11:10
xfxfdo you have a hdmi2usb?11:10
xfxfah, neat, that'll be plenty11:10
thaytanI have an Opsis11:11
thaytanit's never been turned on11:11
xfxflet me know if you want me to give you a hand setting stuff up sat late arvo/evening, more than happy to head up with a copy of the imaging stuff13:04
xfxfif you've got everything there should only take a couple of hours to get you the same setup we used at LCA13:05
xfxfincluding a working build environment for the hdmi2usb firmware13:05
xfxfzero doubts you can do it of course, but i assume this will be exponentially quicker given the documentation for everything is a bit scattered13:05
*** seaLne has quit IRC15:38
*** seaLne has joined #timvideos15:40
tumbleweedCarlFK: thanks :)
tpbTitle: #855555 - unblock: hdmi2usb-fx2-firmware/0.0.0~git20151225-1 - Debian Bug report logs (at
tumbleweedalso, how's that bug number :P16:37
CarlFKtumbleweed: morning - yatzi!16:45
*** cyrozap has quit IRC17:17
*** cyrozap-ZNC has joined #timvideos17:17
CarlFKtumbleweed: is it ok if I image that negk machine you were using ?17:59
mithrotumbleweed: Want to test that reset patch?18:06
tumbleweedgah, a key on my laptop keyboard just broke :(18:07
tumbleweedCarlFK: can I use it a bit more later, to check that I got modeswitch in the PPA, into a good state?18:22
CarlFKtumbleweed: no problem. it can stay of for days or weeks really18:24
tumbleweednaah, hours should be fine :P18:24
CarlFKI just agreed to do video March 25. ;)18:24
tumbleweedjust having breakfast right now18:24
CarlFKjea or anyone know where the thing is that hooks vocto core to youtube?18:25
tumbleweedCarlFK: lca 2017 repository18:26
tumbleweedor the debconf video one. Can't remember18:26
CarlFKone of the ansibles?18:26
CarlFKtumbleweed: I found this for icecast
tpbTitle: debconf-video/ansible - Ansible playbooks for the DebConf Video Team's machines (at
CarlFKI don't see any youtube around there ... or anywhere18:39
mithroCarlFK: jea definitely committed it somewhere18:42
mithropaddatrapper: I'm going to move all the firmware we use on the FX2 in the HDMI2USB project into the repo (rather then part of the misoc/litex firmware repos).19:12
tpbTitle: GitHub - timvideos/HDMI2USB-fx2-firmware: HDMI2USB firmware for the Cypress FX2 found on the Digilent Atlys and Numato Opsis boards. (at
*** SamSagaZ has joined #timvideos19:33
paddatrappermithro: good idea19:47
paddatrapperFractured code doesn't help anyone19:47
jea[m]CarlFK: this one is what you want
tpbTitle: debconf-video/ansible - Ansible playbooks for the DebConf Video Team's machines (at
jea[m]Just need to point that at a YouTube endpoint after you have created it20:11
CarlFKjea[m]: thanks.  how do I not have that on my local repo?20:20
CarlFKI have a feeling I didn't git pull when I should have..20:20
jea[m]Hmm, not sure on that one. I think it went in around Wednesday of LCA, so should be there20:22
CarlFKI guess I get to do a git pull and wade though the merge.. right?20:38
tumbleweedCarlFK: probably. Although a merged branch is usually harder to review than a rebased one20:43
CarlFKtumbleweed: um.. k.  can you walk me though what I should do?   I have 2 sets of changes: 1 veyepar install 2-pxe stuff (minimal changes that I think you will be fine with, a base for me to hack later into what I want)20:52
tumbleweedCarlFK: all entirely uncommited?21:26
CarlFKtumbleweed: yes.  sec.. Ill post a git diff - I think that will easy to take in21:27
tumbleweedCarlFK: ok, git stash21:27
tumbleweedthen you can pull21:27
mithrotumbleweed: so... that testing? :-P21:27
tumbleweedthen git stash pop21:27
tumbleweedmithro: yep, sitting down to do that21:28
*** techman83|clone has quit IRC21:32
*** techman83 has joined #timvideos21:33
*** ChanServ sets mode: +v techman8321:33
CarlFKtumbleweed:  --21:37
tpbTitle: Ubuntu Pastebin (at
CarlFKcan you skim that over before I stash pull etc?21:37
CarlFKalso, how strong do you feel about not installing a web server?  I shoved it in to get things testable. I think I can shift to tftp if you care21:38
tumbleweedCarlFK: we ended up with a webserver at LCA, I think it's something that'll usually be useful21:40
tumbleweedit just hasn't been necessary for anything in the ansible repo,yet21:40
tumbleweedthere are things in that dif that'c can't possibly be right21:40
tumbleweed+Acquire::http::Proxy "http://{{ inventory_hostname }}:8000/";21:40
tumbleweedthat says the apt proxy is always on localhost, right?21:41
CarlFKum. no.21:41
tumbleweedI'm also skeptical of ignore_errors: yes21:41
CarlFKthat was to work around the "reboot" thing21:41
tumbleweedI don't think that's something you want to work-around21:42
tumbleweedyou either want to handle it properly, or require a reboot21:42
tumbleweedhandling it properly means taking down the interfaces before changing the ifupdown config, so ifupdown doesn't get confused21:42
tumbleweedreally, I guess what I want is commits explaining the need for most of this :)21:43
tumbleweedI don't know what problems you are trying to solve21:44
CarlFKmostly making it easier to test21:44
tumbleweedcan you elaborate?21:45
CarlFKI have been createing a pxe server, then using it to build a 2nd pxe server, then the first gets installed from the 2nd. back and forth 20 times or so now21:45
tumbleweedwhy on earth did you need to do that?21:46
CarlFKhaving to track down all the paces the IP of the server is used and change it from .3 to .4 or whatver21:46
tumbleweedyeah, that explains some of this21:46
tumbleweedhonestly, I wouldn't bother with PXE for tetsing21:46
CarlFKto move the host of late_command from github to the local anything (kinda my laptop, ansible copies it to the pxe server)21:46
tumbleweedbut you could have done that by just changing late_command_url, surely?21:48
CarlFKI need to change it to something the client machine can get21:49
tumbleweedand it can't get it from github / your laptop / something?21:49
CarlFKlike the pxe box.  and so it needs to be put on the box.21:49
CarlFKthe client can't get to my laptop all the time21:50
tumbleweedCarlFK: that still doesn't answer my question21:51
tumbleweedyou can give pxe1 the url tftp://pxe1/, and pxe2 the url tftp://pxe2/...21:52
tumbleweedbut I suppose, yes, it makes sense to generate it from hostanme, if we're putting it in place from ansible21:52
tumbleweedI wish this stuff was on github, so we could easily do code review :(21:52
xfxftumbleweed: hey, how's things?  did you see my email?21:53
CarlFKalso that means needs to be moved onto the server.  previously it was moved to github via an undocumented step (kinda)21:53
tumbleweedCarlFK: yeah, because there was no sane default for debian21:53
tumbleweedxfxf: no email, but I do recall a hangouts ping21:54
xfxfoh, maybe it was that - i'm putting a bid together for us pycon, wanted to see if you were interested21:54
xfxfeven tentative interest is good21:54
* tumbleweed is planning to go to pycon21:54
tumbleweedI mean, i've got tickets, hotel etc21:54
xfxfoh, right, you got the rego?21:54
xfxfif you want to go as a normal delegate that's totally understandable21:54
tumbleweedyeah, a couple of months ago21:55
tumbleweedso, I'll be there. And I'm hapy to help21:55
tumbleweedbut I can't do more than a day or two of setup, and don't raelly want to get stuck in a back room21:56
xfxfrad.  no idea if we'll get the bid, but good to know you're happy to be on the team21:56
xfxfthat's fine21:56
tumbleweed:) good luck with that21:56
xfxfi'm likely going to come to pycon anyway as a delegate even if this doesn't go anywhere21:57
xfxfyou still doing the world tour of conferences btw?22:04
* tumbleweed is home22:04
tumbleweedsaw mithro here, yesterday22:04
CarlFKtumbleweed: would it help if I changed my local checkout to github and pushed it all up there so we could use it's history browser soemthing?22:04
xfxfoh lol, mithro is still travelling? i thought he was back in sydney22:06
tumbleweedhe leaves tonight22:06
xfxfthat explains why he's not been around much :P22:06
tumbleweedCarlFK: push current master as master, push your stuff into a separate branch/branches, and then we can do PRs with review22:07
xfxfand why he was having trouble with his Telstra sim card22:07
xfxfmithro: i had Telstra royally f*ck up my roaming too a few trips ago :/  kept insisting it was enabled, it wasn't22:07
CarlFKtumbleweed: push to github, right ?22:09
tumbleweedCarlFK: yeah22:09
mithroxfxf: I'll give the uni a call shortly22:17
xfxfmithro: np, sorry, I didn't realise you were still travelling, this explains why it's complicated22:17
mithroLooking forward to getting home22:20
tumbleweedmithro: found a bug22:34
CarlFKis that why my lcd is black?22:34
tumbleweedmithro: also, reset-gateware doesn't seem to actually do anything22:38
tumbleweedoh, it only works with -v22:39
*** SamSagaZ has quit IRC22:39
mithroHow are you testing?22:40
mithroJust getting lunch be back in 3022:40
tumbleweedmithro: atlys22:40
CarlFKoh right, it went black because I unplugged it22:41
* tumbleweed was going to test there, next22:42
CarlFKjust the lcd - negk and opsis are still up22:46
mithrotumbleweed: I meant, more what commands are you running and how are you "verifying" the reset behaviour?23:03
tumbleweedmithro: hdmi2usb-mode-switch --reset-gateware23:04
tumbleweedverifying with an LCD, and serial terminal23:04
mithrotumbleweed: on the atlys you should be able to stay connected with the serial port while the FPGA reboots and see the bios come up?23:05
tumbleweedyeah, sometimes23:05
tumbleweednot every time23:05
tumbleweedand that happens with -v, but absolutely nothing happens without it23:05
mithrotumbleweed: No errors?23:06
tumbleweedif I hack in real debugging without verbose, I get:23:06
tumbleweedError: Trying to use configured scan chain anyway...23:06
tumbleweedError: xc6s.tap: IR capture error; saw 0x03 not 0x0123:06
tumbleweedWarn : Bypassing JTAG setup events due to errors23:06
tumbleweedFound 1 boards.23:06
tumbleweedwhich is pretty much identical to when it works23:07
tumbleweedexcept for it working :/23:07
mithrohrm... You shouldn't see any IR capture errors... :/23:08
mithrotumbleweed: Can you paste the complete output somewhere?23:08
mithroBTW Does it reboot automatically after loading now?23:08
tpbTitle: debian Pastezone (at
tumbleweeddifference seems to be lines 57 vs 11923:10
mithrotumbleweed: what about loading/flashing?23:10
tumbleweedloading did a reset, without -v23:11
mithrotumbleweed: Can you paste the output of both too?23:12
tumbleweedflashing fails with the 818181 thing. How do we work around taht on atlys?23:12
tumbleweedmithro: with my hax that show you verbose with out -v?23:13
mithrotumbleweed: switch to the production firmware maybe?23:15
* tumbleweed is not sure what that means23:15
CarlFKtumbleweed:  look at all those branches and stuff :p23:19
tpbTitle: GitHub - CarlFK/video-stack-deploy (at
CarlFKincluding somethng that makes me think parsnip23:19
tumbleweedCarlFK: can you file PRs in there?23:20
tumbleweedoh, they all depend on each other23:20
tumbleweedone at a time, then :P23:20
tumbleweed+1 for noroot23:21
tumbleweedI mean, that looks good. I want to discuss the next bits23:22
CarlFK3 prs up23:24
mithropaddatrapper: If you want a simple FX2 related task to look into, do you want to see if you can port the FX2 firmware loading stuff to Python+LibUSB?23:27
mithrotumbleweed: Can you give me remote access to your Atlys?23:27
mithropaddatrapper: The first thing to do would be figure out the FX2 reset sequence23:28
mithropaddatrapper: And add a "--reset-fx2" into HDMI2USB-mode-switch23:28
tumbleweedmithro: yeah, give me 10 mins23:28
CarlFKwant me to add an atlys to negk?23:28
tumbleweedmithro: if you want access to CarlFK's opsis, that's even easier :P23:28
mithrotumbleweed: Probably both23:29
CarlFKk - give me 10 to find/plug in23:29
CarlFKmithro: do you want the atlys on the same host at the opsis?23:29
mithroCarlFK: not a the moment, would want it on a different computer23:29
CarlFKk - I have plenty23:29
CarlFKoh yeah, you have seen my living room.  not much has changed23:30
*** SamSagaZ has joined #timvideos23:57
CarlFKmithro: atlys powered and plugged in to usb - do you need hdmi in/out?23:57
tumbleweeddoubt he does23:58

Generated by 2.13.1 by Marius Gedminas - find it at!