Sunday, 2017-07-16

*** tpb has joined #timvideos00:00
jeaxfxf: I am around. currently setting up machines for recording wordcamp next weekend, with ben and kim01:52
xfxfcool, i'm spending most of my spare time on av related stuff, my todo list is stupid01:53
xfxfworking all of today on stuff01:54
jeai can imagine it would be pretty long. always a lot to do before lca/pycon01:54
*** tvCommitBot has joined #timvideos02:57
-tvCommitBot- [av-foss-stack] CarlFK pushed 1 new commit to master:
-tvCommitBot- av-foss-stack/master 73adb6a Carl Karsten: add the Best Buy laptops to the pool.02:57
*** tvCommitBot has left #timvideos02:57
CarlFKjea: xfxf:  moved docs to readme - try soon05:44
tpbTitle: video-stack-deploy/scripts at usb_late · CarlFK/video-stack-deploy · GitHub (at
jeaCarlFK: thanks for that05:59
jeaCarlFK: there is a section that uses desktopvideo stuff for blackmagic. where do I get them from?05:59
CarlFKjea: you fumble around on the bm site ...  hmm, I think I have a good starting URL.. sec...06:01
CarlFKhow do I get stuff back that I git stash -ed ?06:10
jeagit stash pop06:10
CarlFKunstash isn't it :p06:10
CarlFKda, tx06:10
xfxfstash... stack... popping makes sense, right? :P06:12
jeai have another question that I think you might be able to help with CarlFK - how do you flash your atlys?06:12
CarlFKbm - see if this makes sense
tpbTitle: video-stack-deploy/main.yml at bm-urls · CarlFK/video-stack-deploy · GitHub (at
jeayep, that makes sense thanks :)06:16
CarlFKflashing -
tpbTitle: at carlfk-pull-1 · timvideos/ · GitHub (at
tpbTitle: at carlfk-pull-1 · timvideos/ · GitHub (at
CarlFKI've been told that should work.06:21
CarlFKexcept opsis_hdmi2usb-hdmi2usbsoc-opsis.bin  doesn't seem to exist any more06:22
CarlFKer, the simple answer to your Q is: I don't know06:22
jeai guess you haven't flashed them for a while then?06:23
jeaI think we were doing litex firmware at the start of the year. maybe i can compile it and then flash like we did on the opsis06:23
tpbTitle: GitHub - timvideos/HDMI2USB-firmware-prebuilt: Prebuilt firmware for the HDMI2USB devices (such as the Numato Opsis and the Digilent Atlys board) and OS drivers. (at
CarlFKthat should have the blobs you need06:24
CarlFKif you can build something from source that isn't there, something is very broken06:24
CarlFKbut maybe you have to walk though the build process to figure out what file(s) to be using06:25
jeaoh maybe I just wasn't looking in the right place of the prebuilt repo06:25
jealooks like I might want this
tpbTitle: HDMI2USB-firmware-prebuilt/archive/master/v0.0.3-715-gbc835e3/atlys at master · timvideos/HDMI2USB-firmware-prebuilt · GitHub (at
CarlFKit looks like the dir/file names changed a few months ago06:25
CarlFK3 months ago06:26
CarlFKmithro says  use the recent version - but I can't figure out how06:27
tpbTitle: HDMI2USB-firmware-prebuilt/flash.bin at master · timvideos/HDMI2USB-firmware-prebuilt · GitHub (at
jeamithro: so I just need that one file to flash the atlys using make-gateware?06:30
*** vystrela has joined #timvideos06:31
mithroIf you have the dev environment, just `make flash` should work?06:31
jeayes, that is also a thing I could do06:31
jeai'll just switch the targets from opsis to atlys06:32
mithroPlatform you mean?06:32
*** vystrela is now known as vys06:32
vysalso, hello again06:33
tpbTitle: HDMI2USB-litex-firmware/ at master · timvideos/HDMI2USB-litex-firmware · GitHub (at
mithroI'm just heading into the office now06:34
mithrojea: we should get the 'make download' bit working again06:35
mithroxfxf: ping?06:36
jeayou'll have to remind me what it did again, because I've forgotten. but if it is what I think, that sounds reasonable06:36
mithroIt downloads from the pre-built repo into the dev env06:37
mithroSo you don't have to have the Xilinx tools if you just want to do firmware (as appose to gateware) development06:38
mithroI need to review some PCB routing and figure out why the bios isn't working on the mimasv2 with or1k06:39
mithroAnd somehow it is already almost 5pm06:40
CarlFKmithro: should this work:06:40
CarlFKcd HDMI2USB-firmware-prebuilt/archive/master/v0.0.3-698-g7b7714d/atlys/hdmi2usb/lm3206:40
CarlFKhdmi2usb-modeswitch --flash-gateware flash.bin06:40
CarlFKand is v0.0.3-698-g7b7714d what we should be testing ?06:40
mithroxfxf: Are you expecting me to bring Opsis boards to PyCon AU?06:42
mithroCarlFK: possibly, I haven't tested Atlys in like forever06:42
mithroShould probably just have a "--flash" option now06:43
mithroAs the flash.bin has all of the gateware, bios and firmware in one file06:44
CarlFKmithro: can you linke this to whatever you want us to be running:    HDMI2USB-firmware-prebuilt/atlys/firmware/testing06:45
*** sb0 has quit IRC06:46
mithroCan you log an issue about that and assign it to me, I'll try and do it in a moment but will forget if I don't get to it06:49
mithrounstable should be the current "head", testing should be what ever people have actually tested and confirmed working but not used at a conference, stable should be what ever we last used at a conference and was happy with06:50
CarlFKmithro:  doesn't have issues ...06:53
tpbTitle: GitHub - timvideos/HDMI2USB-firmware-prebuilt: Prebuilt firmware for the HDMI2USB devices (such as the Numato Opsis and the Digilent Atlys board) and OS drivers. (at
mithroCarlFK: put it on the HDMI2USB-litex-firmware repo06:53
tpbTitle: HDMI2USB-firmware-prebuilt links to unstable testing, stable · Issue #328 · timvideos/HDMI2USB-litex-firmware · GitHub (at
mithroCarlFK: thank you06:59
mithroCarlFK: are you going to buy some Opsis boards any time soon?07:01
mithrojea: Are you likely to want Opsis boards in Brisbane or do you already have enough?07:05
mithrojea: Actually - the make download script could just be the same as this one I think ->
tpbTitle: micropython/ at master · mithro/micropython · GitHub (at
jeamithro: i have one opsis here. might be good to get more at some point, because then I can actually do things like wordcamp07:08
CarlFKmithro: no.   PyCon is the only show that has enough budget to get nice stuff07:08
jeai probably only need one more though so I have a full room setup07:09
mithrojea: Probably good to have atleast one spare too07:09
mithroI'm working on trying to make sure we have the ability to order individual opsis in the future07:09
jeathat would be good07:09
mithroBut that is basically going to require me to front for a whole batch07:10
CarlFKjea; any chance of you running the script to make a usb stick?  in the next 30 min?07:10
jeamithro: ah, that is annoying07:11
CarlFKbooting a box would be good too, but that can come later07:11
mithroCarlFK: How are you doing recording of presenters? Still twinpacs? Or do you have enough Atlys?07:11
CarlFKmithro: I have 4 Atlys and 1 Opsis07:11
mithroCarlFK: What would you need to get that to 4 Atlys and 4 Opsis?07:12
jeaCarlFK: what would you like to see in the next 30mins?07:15
jeait can be done07:15
jeamithro: are we using master branch or something else for the litex repo?07:16
mithrojea: master07:20
CarlFKjea:  "How to use:"07:20
tpbTitle: video-stack-deploy/scripts at usb_late · CarlFK/video-stack-deploy · GitHub (at
jeamithro: thanks07:21
jeaCarlFK: ok07:21
CarlFKjea: skip steps 2-507:21
mithroCarlFK: What is this USB stick thing?07:23
jeaCarlFK: will it wipe out all partitions on an existing drive, or can i choose where it installs to?07:23
CarlFKjea: as is it will blow away sda07:26
CarlFKjea: (do a git pull, I just removed a #line that I pulled out to speed up testing07:26
CarlFKmithro: Make a usb installer: debian that loads a preseed and runs ansible07:27
CarlFKmithro: like my pxe insller, only boots from usb, then tries to use the same preseed/ansible07:27
CarlFKjea: but if you don't boot it, it won't do anything... so first I want to make sure the doesn't error07:28
mithrobtw Do we still want to do the VGA expansion board for the Opsis?07:28
mithrotumbleweed: ^07:28
CarlFKmithro: i would not want time spent on  the vga board that could be spent else where07:30
CarlFKmithro:  we have done enough shows without it, and it is just going to get less useful07:31
mithroCarlFK: That was my thinking07:31
jeaCarlFK: oh right. i'll run it now07:32
jea(just don't have a machine right now that i want to blow away, but making the usb is fine)07:32
CarlFKjea: the scritp will blow away whatever sdX drive you tell it to hit...07:32
CarlFKthat should be your usb stick07:32
jeaok, that is alright07:33
jeablowing away a usb is fine, just don't want to blow away a machine when installing debian/ubuntu/etc07:33
jeabut as you said, that part is when we do the booting, which can be later07:33
CarlFKwell, don't blindly cut/paste the sdb from the docs07:33
CarlFKjea: I have stopped testing ubuntu because it is annoying to test extra stuff and the debian side works fine07:37
CarlFKjea: so "this" wont work for you now, but if you fee like ... hacking...07:37
CarlFKjea: this is what you want
tpbTitle: video-stack-deploy/preseed_disk.cfg at usb_late · CarlFK/video-stack-deploy · GitHub (at
CarlFKyou # that out and the installer will prompt you for what disk/partition etc.07:38
jeai can use debian or ubuntu, so that is fine07:38
CarlFKI kinda want to do that for debian/stretch but I don't expect it anytime soon07:39
CarlFKthat = split the preseed.cfg up into parts so that we can do what you want .. becuase in the past I have wanted that too07:40
CarlFKjea; you can test the usb stick with qemu:
tpbTitle: video-stack-deploy/ at usb_late · CarlFK/video-stack-deploy · GitHub (at
CarlFKjea:and notice all this stuff is in the usb_late branch07:42
CarlFKif it seems to run and the docs sound sane, I'll push to get it into master in ~10 hours07:43
jeacan i just use your repo as the inventory for now?07:44
CarlFKnone of us are really happy with how the inventory file is managed07:45
CarlFKwe are using github as both our version control and a cloudy file server07:45
jeayeah, i was wondering about that07:46
jeaare you still using the debconf ansible git repo, or is this repo replacing it?07:46
jeaalso, script is now running07:46
CarlFKI am hacking on my repo, tumbleweed is cherry picking  most of my changes into the DC repo07:47
CarlFKso DC repo is stabler - but also untested-er :p07:49
jeamake gateware with T=atlys is taking so far failing, using latest master07:59
mithrojea: PLATFORM=atlys TARGET=hdmi2usb ?08:00
xfxfmithro: i don't care about the VGA expansion board personally.  like, i guess it'd be nice, but i'd much prefer effort was put into other things given finite time08:03
mithroxfxf: See my above questions about PyCon AU and Opsis08:03
jeamithro: no, hadn't set target. trying again08:04
jeawhat does target default to?08:04
xfxfmithro: if you can bring extra boards, that'd be great.  i think i have enough, but spares definitely do not hurt, especially if we have problems08:04
mithroxfxf: Well, I always plan to bring spares08:04
xfxfi think i own 5, LCA owns 708:04
xfxfplus i'll have a handful of blackmagic capture devices too (either pci-e or the web presenters)08:05
xfxfi'm tempted to use those for camera capture actually as they let me pipe audio in via HDMI/SDI and i'm yet to see voctomix get out of sync doing that08:05
xfxfi still am not clear why the seperate audio device gets out of sync with the video, thaytan  looked over our script AFAIK and thought it was OK08:06
xfxfbut i'm up for debate on that.  i have captured a couple of small events over the past few months doing it the way above though (BM for capture, hdmi2usb for podium) and it's worked well08:06
jeawe adjusted some scripts based on his suggestions and it didn't seem to change things08:06
jeaxfxf: which BM card do you have?08:07
CarlFKjea:  (scroll down past the product spam)08:09
tpbTitle: Blackmagic Design: DeckLink Tech Specs (at
jeaok, so both have the same then08:10
jeamithro: failed again. some process being killed line 5: 40450 Killed                  xst -ifn top.xst08:10
jeamaybe my vm is running out of resources?08:10
CarlFKjea: did the mk script end with + python -m SimpleHTTPServer08:11
CarlFKServing HTTP on port 8000 ...08:11
jeait is stil downloading debian08:11
CarlFKk - the good news is it will only do that once08:12
jeafor whatever reason it selected a swedish mirror instead of something in aus, so it is slower than i expected08:12
xfxfmithro: if i'm building and testing new hdmi2usb firmware/gateware, just build out of master on the misoc firmware on the main repo?08:13
jeawith the bm card, is it alright having a camera come in at 1080 and then go into vocto?08:13
mithroxfxf: Yes08:13
jeamisoc-firmware is now litex-firmware, right?08:14
xfxfoh right, yes, litex08:15
CarlFKjea: ingest will down sample the 1080 but that uses a good bit of cpu08:15
xfxfjea: i don't do that, i run everything at 720p natively08:15
CarlFKI am not sure if the bm cad will scale it08:15
xfxfhence using cameras where i can change the HDMI res08:15
xfxfBM card doesn't scale08:15
xfxfi've given up on the BM PCIe cards anyway, the web presenters have a built in scaler08:16
xfxfand present as USB UVC and an audio device08:16
jeathat throws a spanner in the works08:16
jeaoh really? tell me more about these web presenters08:16
xfxfi'll go back to the HDMI2USB tho once USB audio is working, with exception if ingesting SDI devices08:17
tpbTitle: Blackmagic Design: Blackmagic Web Presenter (at
xfxfthey do SDI/HDMI passthrough, and give a 720p downscaled version from whatever input over USB08:17
xfxfthey're not a replacement for lecturn capture (hdmi2usb still wins there by a mile), but they're way better than using a BM PCIe card IMO due to the scaler and the fact it works on a laptop and you don't need wierd drivers08:18
xfxfonly caveat is they're much more expensive08:18
xfxfthey don't come with the front panel, the front panel isn't needed but there are settings you won't be able to change without it08:18
xfxfit's about $800 for the unit + front panel08:18
jeathat all sounds good apart from the price08:19
jeaCarlFK: script worked08:20
CarlFKjea: woot. thanks.08:20
CarlFKjea: im off to bed - look over the config files and stuff - I know I can tweak the preseed file to prompt me for where to install - let me know how comfortable you are doing that - and how bad it will be if you get it wrong08:24
jeaok, will do08:25
jeaspeak to you later08:25
CarlFKyou can install debian however you want  and then ansible later - but that takes fiddeling too08:25
jeayes, that is what we have done so far, but on ubuntu08:25
jeabut i'll give the usb a go08:26
xfxfhey CarlFK, does your usb installer work on minnowboards?08:56
CarlFKxfxf: no idea - do you have one?08:57
xfxfbecause i have two i'm yet to do anything with, and i don't really want to reformat the working LCA image on the laptops because they're a 'known working' image, vocto hasn't had a major release since then, and i'm hesitant to start screwing with it08:57
xfxfbut if they work on those, then i'll give it a go tonight08:57
xfxfaren't you using minnowboards for the lecturn capture machines?08:57
xfxfi bought a pair to see if i can make that side of things neater08:58
CarlFKi am using odroid c208:58
xfxfoh right, i thought you were using minnowboards08:58
xfxfwhich is why i bought a pair :P08:58
CarlFKdo you have ssd's mounted in them?09:02
mithroxfxf: DebConf is using MinnowBoards - so am I09:02
CarlFKDC is not using MinnowBoards09:02
xfxfi swear tumbleweed said that DC weren't but you were CarlFK09:03
CarlFKthey have like 20 i7 desktops09:03
xfxfeither way, i now have two :P09:03
CarlFKI was going to if DC used them. then the came up with desktops, so none of us got any09:03
mithroCarlFK: MinnowBoards connected to the Opsis is listed on their kits?09:03
CarlFKmithro: that was the plan before they discovered a local stash of desktops09:05
CarlFKxfxf: if you are usung the ssd, it may need this funky dev name: disk=/dev/nvme0n109:06
mithroDid jea end up with my other MinnowBoard?09:10
jeai set one up at hobart, but i didn't take it with me09:11
xfxfshipping around desktops has to be $$$ tho09:11
thaytanxfxf, is that what we said? I've forgotten already09:16
thaytanI thought we talked about needing to set a base time as well as distributing the clock09:16
mithrookay, PCB stuff done! Next is repoing Hasjim's issue09:30
xfxfthaytan: i can't remember - if there are issues with the existing pipeline then I need help to fix that09:40
xfxfi remember about the base time thing but i thought one was actually being set09:40
xfxfmemory fuzzy now tho09:40
thaytanyou'll have to come for another visit :)09:51
mithroDoes anyone know where to source one of these? ?09:57
tpbTitle: Tronsmart Pavo M9 4K TV BOX Mstar MSO9180D1R 1GB/8G (at
mithroCarlFK: Any idea how many watts the confidence monitors you use at the front of the room consume?10:12
mithrojea: Looks like you actually want "make image-flash"10:16
mithrocr1901_modern: ping?10:44
mithrocr1901_modern: Do you remember what was going on with puts failing?10:44
mithrocr1901_modern: It was something to do with loops and the sram IIRC?10:45
*** shane_ has joined #timvideos10:47
*** shane_ has left #timvideos10:47
*** sb0 has joined #timvideos11:34
mithropaddatrapper: ping?11:52
paddatrappermithro: pong?11:52
mithropaddatrapper: Just reading your daily updates11:53
mithropaddatrapper: You mentioned "I spent today trying, and failing, to get printf(...) to print to serial. I think this is something I will need to put on hold for a while and concentrate on ISO transfers for the coming week."11:53
paddatrappermithro: yeah, it's not using the getchar, putchar I defined, but there is the work around of using usart_send_string directly11:54
paddatrapperand ISO trafser is more pressing11:54
mithropaddatrapper: What did you do to debug?11:55
paddatrappermithro: monitor serial and write to I/O pins in the putchar function, neither showed output11:56
mithropaddatrapper: Did you look at if your putchar is getting linked into the output?11:56
paddatrappermithro: how do I do that?11:57
mithroLook at the .lst and similar files11:59
paddatrapperOk, what am I looking for, mention of putchar?12:00
paddatrappermithro: I can find mentions of _printf, but they are all lcalls, with no associated assembly12:06
mithropaddatrapper: Where is your example?12:09
paddatrappermithro: I'm using
tpbTitle: GitHub - paddatrapper/HDMI2USB-fx2-firmware at audio-data (at
paddatrapperbuilding from the audio directory12:11
paddatrappermithro: I'm following the video firmware's example, using soft_putchar(..) (softserial.c), and #define putchar soft_putchar in the firmware.c source12:15
mithropaddatrapper: So I think it must be broken in the hdmi2usb firmware12:18
mithropaddatrapper: the #defines have to be *before* you include stdio.h12:19
paddatrappermithro: that is my though too12:19
paddatrappermithro: tried that too, with no change12:19     00000017  _vprintf_PARM_2                    vprintf12:21
mithropaddatrapper: So you also need to make sure you are not linking in fx2lib/lib/serial.c code12:22
paddatrappermithro: unfortunately I have to, need to call sio0_init(57600) from serial.h, otherwise I get a descriptor read error when the device reconnects12:24
mithropaddatrapper: Then there is something in sio0_init which is needed elsewhere....12:26
mithropaddatrapper: I replaced the `soft_` names in soft_serial.{c,h} with nothing12:28
mithroThen I get12:28   00000C4C  _sio0_init                         softserial12:29   00000C4F  _putchar                           softserial12:29   00000C61  _getchar                           softserial12:29                            [  ]12:29
mithroBefore that I get12:29   00000C4C  _soft_sio0_init                    softserial12:29   00000C4F  _soft_putchar                      softserial12:29   00000C61  _soft_getchar                      softserial12:29   00002243  _sio0_init                         serial12:29   000022F9  _getchar                           serial12:29   00002302  __transchar                        serial12:29   0000230C  _putchar                           serial12:29
paddatrappermithro: is that with removing the sio0_init call?12:30
mithroWell, the sio0_init call is going to the softserial version now....12:30
mithroI haven't actually tested :-P12:31
paddatrapperJust tested and got the descriptors as expected...12:31
paddatrapperAnd the serial output12:31
paddatrappermithro: thanks, it now works :)12:32
mithropaddatrapper: You'll probably need to look at the .map stuff reasonable frequently :-P12:34
paddatrappermithro: ok, I shall keep it in mind then12:34
paddatrappermithro: any idea why we always return false to a get descriptor request?12:39
tpbTitle: HDMI2USB-fx2-firmware/audiodata.c at audio-data · paddatrapper/HDMI2USB-fx2-firmware · GitHub (at
mithropaddatrapper: False means we didn't handle the request12:40
mithropaddatrapper: So returning false means that the code in setupdat.c will handle the request12:40
paddatrappermithro: Ah ok cool12:41
tpbTitle: fx2lib/setupdat.c at master · mithro/fx2lib · GitHub (at
tpbTitle: ADV748x HDMI/Analog video receiver [] (at
mithro is kind of interesting too...13:33
tpbTitle: Synopsys Designware HDMI Video Capture Controller + PHY [] (at
mithroWell that is me for the night13:47
mithropaddatrapper: Need anything from me before I go?13:47
paddatrappermithro: sorry, was dragged away, but nope14:19
*** [d__d] has quit IRC15:40
cr1901_modernmithro: Correct; for certain SRAM sizes, once the CPU started looping it would go into the weeds. I never did figure out why.16:16
*** puck has quit IRC17:32
*** puck has joined #timvideos17:36
*** [d__d] has joined #timvideos17:39
*** techman83 has quit IRC22:00
*** techman83 has joined #timvideos22:01
*** ChanServ sets mode: +v techman8322:01
jeaxfxf: ping23:12

Generated by 2.13.1 by Marius Gedminas - find it at!