Friday, 2017-08-04

*** tpb has joined #timvideos00:00
*** Choco31415 has quit IRC00:01
xfxfCarlFK[m]/mithro: i just looked, ohiolinux is 6 streams00:02
xfxfthat guy needs a team of experienced peeps00:02
xfxfencouraging somebody to learn the ropes to do a 6 stream event on their own is a guaranteed disaster imo :/00:03
xfxfhell i wouldn't do it on my own00:03
paddatrappermithro: ok cool00:27
mithropaddatrapper: I'm sure I've seen this issue before, but I can't remember the solution01:10
paddatrappermithro: I wonder if a new commit will fix it01:11
jeamithro: how do i open mplayer or vlc or something like that to show the output of the connected opsis?01:13
tpbTitle: HDMI2USB-litex-firmware/ at master · timvideos/HDMI2USB-litex-firmware · GitHub (at
jeathanks :)01:13
tpbTitle: voctomix-outcasts/ at master · CarlFK/voctomix-outcasts · GitHub (at
mithropaddatrapper: I can't find the commit 75bebb54743bae46e16e6e7e460f30270429500b anywhere in repos...01:13
paddatrappermithro: that is weird...01:14
mithropaddatrapper: It might be worth dumping all the travis values... IE TRAVIS_PULL_REQUEST_SHA and stuff01:15
paddatrappermithro: Yeah, the commit it is supposed to be building 50615fc869f350033dccd3937895135be0b482601:15
CarlFKjea: don't forget to run that before you connect to the serial console01:16
paddatrapperI wonder if it broke thanks to it being a rebase01:16
paddatrappermithro: a new commit fixed it01:26
tpbTitle: Support resetting an video input without resetting the output · Issue #80 · timvideos/HDMI2USB-litex-firmware · GitHub (at
tpbTitle: Add auto-switch mode · Issue #208 · timvideos/HDMI2USB-litex-firmware · GitHub (at
mithropaddatrapper: Oh, is there a weekly report for me to review?01:35
paddatrappermithro: yes there is01:36
*** paradisaeidae has joined #timvideos01:37
tpbTitle: If pixel clock detected (or no input detected) reset the input every X seconds · Issue #337 · timvideos/HDMI2USB-litex-firmware · GitHub (at
mithropaddatrapper: And you try removing those commits and pushing again?01:40
paddatrappermithro: just waiting for it to build01:41
mithropaddatrapper: okay01:41
mithrojea: would be good to fix too :-)01:42
tpbTitle: Label pixel clock in status output to prevent confusion between it and the refresh rate. · Issue #304 · timvideos/HDMI2USB-litex-firmware · GitHub (at
paddatrappermithro: though it seems to not want to build again, uses the cached result (fail)01:42
mithropaddatrapper: There should be two builds, one for your repo and one for the pull request...01:42
* paddatrapper enables builds for my repo01:43
paddatrapperI just find the emails annoying01:43
mithropaddatrapper: Oh, you haven't enabled build for your repo?01:43
mithropaddatrapper: You can filter them pretty easily?01:43
paddatrappermithro: I do when working on the travis stuff, for the rest I let it build on PRs01:43
mithroI've run out of battery power01:44
tumbleweedmithro: will do01:50
tumbleweedinteresting data point, none of our VGA->HDMI convertors were working01:50
tumbleweedchanged the primary mode on the opsis to 60fps, and now they (almost) all work (except the ones that xfxf told us to buy :P )01:50
CarlFKtumbleweed: what video_mode number?01:51
tumbleweedCarlFK: 901:51
mithrotumbleweed: are you using the duel mode option?01:52
tumbleweedmithro: yeah01:52
tumbleweedso, default boot mode, with that commit01:52
CarlFKWhat is duel mode?01:52
CarlFKwhatever mode... Just saw it when I was setting up last week01:53
mithrotumbleweed: I'm not sure that acrobat actually will do anything?01:53
mithroS/acrobat/ /01:53
mithroStupid mobile phone01:53
mithrojea and shenki are our EDID experts (IE the last people to touch the code :-P)01:54
tumbleweedmithro: it seemed to do the trick01:55
mithrotumbleweed: that confuses me :-P01:55
jeawe now display two modes on the EDID. same resolution, but one 50p and the other 60p01:56
CarlFKdisplay ?01:56
jeabut the mode on the opsis is still set to one specifically01:56
jeaCarlFK: the computer will see that two frequencies for 720 resolution01:57
CarlFKif the opsis is only going to use one, ... why?01:58
tumbleweedwe were getting strobing when a camera caught the projector in frame, so we wanted 60fps on the projector01:58
tumbleweedso we made that change01:58
tumbleweedthen suddenly the VGA adaptors we were fighting with played ball01:58
jeawe advertise both because it allows a majority of laptops to work out of the box02:01
CarlFKjea: and it is ok if the laptop picks the one the opsis isn't using?02:02
jeawe thought it should be, because the frequencies are similar enough that it would work with either02:03
jeabut maybe there is an issue?02:04
mithrotumbleweed: We should be using wprintf rather than printf everywhere02:06
tumbleweedw wraps it?02:07
tumbleweedyou should also be doing cr-lf everywhere, because that's the norm for serial terminals02:07
tumbleweedI did a bit of a hack there, because the issue is in litex too...02:08
tumbleweedone of the aims of the hdmi2usb is to never force a mode switch on the projector02:09
tumbleweedbut we're noticing a projector attached to this opsis seems to "reconnect" (for lack of a better word) quite frequently02:10
tumbleweedand doesn't if we bypass the opsis02:10
mithrotumbleweed: Can you explain more?02:10
mithrotumbleweed: Are you getting memory overflows?02:10
tumbleweedhow would I tell?02:10
mithrotumbleweed: The status output02:11
tpbTitle: HDMI2USB-litex-firmware-old/stdio_wrap.c at dhcp · mithro/HDMI2USB-litex-firmware-old · GitHub (at
mithrotumbleweed: Oh yeah02:12
mithrotumbleweed: How many inputs / outputs are you using?02:13
tumbleweed1 in 1 out02:13
tumbleweed(currently vga, because it's proving especially dodgy)02:13
mithrotumbleweed: and the encoder?02:14
tumbleweedyep underflows02:14
tumbleweedinput0:  0x0 (@1.66 MHz) (disabled)02:14
tumbleweedinput1:  1280x720 (@61.87 MHz) (capturing)02:14
tumbleweedoutput0: [email protected] from input1 (underflows: 5059)02:14
tumbleweedoutput1: [email protected] from input1 (underflows: 5050)02:14
mithrotumbleweed: Disable the second output02:14
tumbleweedencoder doesn't seem affected02:14
mithroThe encoder will just stall when it runs out of bandwidth, it's not timing sensitive02:15
mithrotumbleweed: If the HDMI output stalls, then the projector will possibly lose connection02:16
mithroxfxf: ping? Where you testing this setup?02:16
mithrotumbleweed: What version of the firmwarE?02:16
tumbleweedoutput1 off seems to help02:17
jeamithro: 105 and 106 both have dual output at the moment from the opsis02:18
mithrotumbleweed: I wonder if 60Hz is just enough to start getting overflows (compared to 50Hz)02:18
mithrotumbleweed: BTW that pixel clock looks a bit strange, it should be ~75MHz02:19
tumbleweedoops, that's not the modeline I meant to use02:21
tumbleweedinput1:  1280x720 (@74.25 MHz) (capturing)02:21
tumbleweednow I'm feeding this modeline in:
tpbTitle: HDMI2USB-litex-firmware/processor.c at master · timvideos/HDMI2USB-litex-firmware · GitHub (at
mithroWe really should get the detected HSync / VSync refresh rate into that output02:22
tpbTitle: Have frequency (and hsync/vsync) information on the inputs in the status message · Issue #94 · timvideos/HDMI2USB-litex-firmware · GitHub (at
mithrocr1901: So, how is that SPI coming along? :-P02:28
tumbleweedmithro: so, 2 inputs on seems to work02:42
tumbleweedhrm, now we're noticing artifacts with that firmware build, on the encoder02:43
tumbleweedbut not on v0.0.3-732-gccfa97c-dirty02:43
mithrotumbleweed: Hrm?02:44
mithrotumbleweed: you mean 740 is getting overflows, while 732 is not?02:45
mithrotumbleweed: 2 inputs or 2 outputs?02:45
*** http_GK1wmSU has joined #timvideos02:47
tumbleweednot necessarily overflows, but some weird artifacts02:47
tumbleweedI'll show you a stream02:47
*** http_GK1wmSU has left #timvideos02:47
tpbTitle: hls.js demo (at
mithrotumbleweed: sorry, I went to lunch03:32
mithrotumbleweed: So what am I looking at in that video?03:33
tumbleweedmithro: the artifacts03:37
tumbleweedgo fullscreen03:37
mithrotumbleweed: sure, but how are you getting those and under what conditions?03:37
tumbleweedanything we send into this opsis03:38
*** rohitksingh_work has joined #timvideos03:38
tumbleweed(running my branch)03:38
tumbleweedtried from two separate machines03:39
mithrotumbleweed: What is status showing?03:39
tumbleweedinput0:  0x0 (@1.60 MHz) (disabled)03:40
tumbleweedinput1:  1280x720 (@74.25 MHz) (capturing)03:40
tumbleweedoutput0: [email protected] from input1 (underflows: 0)03:40
tumbleweedoutput1: off03:40
tumbleweedEDID primary mode:   [email protected]03:40
tumbleweedEDID secondary mode: [email protected]03:40
tumbleweedencoder: 1280x720 @ 23fps from input1 (q: 85)03:40
tumbleweedlooks fine on HDMI output03:41
mithrotumbleweed: and I assume you haven't loaded any hdmi2usb.hex manually or anything?03:41
tumbleweedwith a different opsis (running a different firmware) we don't see this03:42
mithrotumbleweed: and in a version on master?03:42
tumbleweedabout to do some bisection on firmware to see03:42
tumbleweedbut I was waiting for you to see the issue03:43
mithrotumbleweed: okay03:43
mithrotumbleweed: and you tried restarting the opsis?03:43
mithrotumbleweed: Oh can you get the version output before you do03:43
tumbleweed  git describe: v0.0.3-740-g6a8ced6-dirty03:43
tumbleweed    git commit: 6a8ced607223b36296f20857efd9f085646c7f8a03:43
mithrotumbleweed: The full stanza in pastebin would be useful03:44
tpbTitle: debian Pastezone (at
tumbleweedI'm trying to get an ISE install up, so I can iterate a bit faster on hacks...03:44
tumbleweedok, back to flashing03:44
mithrotumbleweed: You shouldn't need ISE if you just want to hack the firmware level stuff03:45
mithrotumbleweed: You can download gateware from prebuilt and put it in the build directory03:45
tpbTitle: Fix the pre-built download script · Issue #322 · timvideos/HDMI2USB-litex-firmware · GitHub (at
mithrotumbleweed: It would be really nice to get to the stage where people can hack on the firmware without needing the gateware03:57
tumbleweedwhich means packaging the lm32 toolchain03:59
mithrotumbleweed: Well, I mean still the HDMI2USB-litex-firmware repo, just not Xilinx ISE04:05
*** SamSagaZ has joined #timvideos04:12
mithrotumbleweed: I'm actually fixing that right now...04:15
tumbleweedyeah, I get that04:16
tumbleweedI'm just struggling to find my way around the build environment :P04:16
tumbleweedmake firmware-load-opsis just hangs on an flterm prompt04:17
mithrotumbleweed: You need to hit the reset button on the opsis04:22
mithrotumbleweed: firmware-load uses serial to load the firmware, you need to get the fpga into the firmware load mode via either hitting the reset button or typing "reboot" if the console is working04:25
tumbleweedyeah, I can do that04:26
tumbleweedit never seems to be running the new build, though04:27
mithrotumbleweed: Paste the output?04:27
mithrotumbleweed: It should load right after the bios prompt appears04:27
tpbTitle: debian Pastezone (at
tumbleweedat that point it didn't wait for me to type reboot or press a button04:29
tumbleweedbut it still exited 004:29
mithrotumbleweed: What happened at the flterm command?04:34
tumbleweednothing, it exited immediately04:34
* tumbleweed hard reset the board now04:35
tumbleweedif I manually use flterm --kernel=firmware.bin and then reboot, that should load the firmware ,right?04:35
tumbleweedit disconnects me after loading the firmware04:35
tumbleweedand when I reconnect, it isn't running the new firmware04:35
mithrotumbleweed: Can you paste the whole output with the loading of the firmware and the disconnect?04:36
tumbleweedthat was the whole output (for your make04:37
jeaCarlFK: what is the thing to generate the preview videos?04:38
tpbTitle: debian Pastezone (at
mithrotumbleweed: The issue is the flterm errors..04:38
mithro[FLTERM] Uploading kernel (85396 bytes)...04:38
mithro[FLTERM] Got unknown reply 'T' from the device, aborting.04:38
mithro[FLTERM] Booting the device.04:38
mithro[FLTERM] Got unknown reply 'i' from the device, aborting.04:38
tumbleweedwhat does that mean?04:39
jeais the next one an m?04:39
tumbleweed"Timeout" I think04:39
mithrotumbleweed: Yeah, something has gone wrong when it was sending the firmware04:40
jeatimeout starts with tim, so i still get what i was hoping for :P04:41
mithrotumbleweed: You can do a "make firmware-flash" in theory04:41
tumbleweeddoesn't boot now04:43
mithrotumbleweed: Can you paste the output of "make firmware-flash" ?04:44
tpbTitle: debian Pastezone (at
mithroYeah, it's flashing the firmware to the wrong location04:46
mithroread 85404 bytes from file /home/videoteam/HDMI2USB-litex-firmware/build/opsis_video_lm32/software/firmware/firmware.fbi and flash bank 0 at offset 0x00200000 in 3.011761s (27.692 KiB/s)04:46
mithroThat is the old location before we put the BIOS in ROM04:46
mithros/ROM/SPI flash/04:47
CarlFK[m]jea: python  --raw --low04:48
tpbTitle: Fix the download-prebuilt script. by mithro · Pull Request #340 · timvideos/HDMI2USB-litex-firmware · GitHub (at
jeaCarlFK[m]: thanks. so i can run that to generate the previews, then run later to do the actual encoding?04:49
mithrotumbleweed: This is the issue you hit there ->
tpbTitle: Fix flashing firmware on opsis/atlys · Issue #320 · timvideos/HDMI2USB-litex-firmware · GitHub (at
jeai don't remember anything else that needs to be done, until youtube uploads04:49
jeaCarlFK[m]: is there a specific way to login to youtube to get the oauth-<thing>.json file?04:49
CarlFK[m]jea: use to try and post a file (even a dummy one) it will give you a URL to browse to04:51
jeaCarlFK[m]: right, was hoping it would be something like that :)04:51
CarlFK[m]browse the URL, get back a string, cut/paste the string into the prompt, it will create the json file04:52
mithrotumbleweed: Fixing now04:52
tumbleweedmithro: so, how do I build a flash.bin, whith what I have?04:52
mithrotumbleweed: make image04:52
tumbleweedgateware not found04:52
tumbleweedbut I put it there04:52
tumbleweedah, it wanted video_lm32 not hdmi2usb_lm3204:54
tumbleweedno idea what "video" is04:54
mithrotumbleweed: export TARGET=hdmi2usb04:55
jeaCarlFK[m]: the sync_rax script exited after a few seconds. I presume it should take longer than that, so anything that I should be looking at to fix it?04:55
CarlFK[m]jea; add a -v, then stare at the spew ... and paste bin it and I'll stare at it04:56
* tumbleweed must go to bed soon04:57
tumbleweedI stopped being able to think a few hours ago04:57
tpbTitle: Ubuntu Pastebin (at
CarlFK[m]jea:  if this is you, you haven't run adddv yet05:00
tpbTitle: veyepar - Show: PyCon Australia Pycon_au_2017 (at
*** SamSagaZ has quit IRC05:00
jeaCarlFK[m]: no, i haven't done that. let me run that05:01
jeaCarlFK[m]: any flags needed?05:01
tumbleweedmithro: didn't you also have a TFTP method for faster development?05:02
jeaexcellent. i like simple!05:02
mithrotumbleweed: Yes05:02
tumbleweedwell good news after all of that is that I can't reproduce the artifacts on this opsis05:04
tumbleweedback to the other opsis :/05:04
tpbTitle: Fix the location flashing firmware too. by mithro · Pull Request #61 · timvideos/HDMI2USB-mode-switch · GitHub (at
tumbleweedI'll have to take your word on that05:07
tumbleweedlots of magic numbers05:07
tumbleweedafter a power cycle, the artifacts are gone05:09
tumbleweedmithro: I'll play with that branch tomorrow05:12
* tumbleweed -> bed05:12
mithrotumbleweed: woose, sleep is for the week :-P05:14
*** paradisaeidae has quit IRC05:31
*** paradisaeidae_ has joined #timvideos05:46
*** rohitksingh_wor1 has joined #timvideos05:52
*** rohitksingh_work has quit IRC05:53
*** paradisaeidae_ has quit IRC06:03
*** CarlFK has quit IRC06:15
*** cr1901 has quit IRC06:17
*** paradisaeidae has joined #timvideos06:18
*** cr1901 has joined #timvideos06:27
jeamithro: I tried to flash the firmware (after applying your patch to the mode switch tool within my build area), but it doesn't seem to have worked06:51
xfxfjea: you're right, you should have asked him that during his talk06:58
mithroPastebin the output?07:33
tpbTitle: Ubuntu Pastebin (at
mithrojea: that looks like the output of firmware-load not firmware-flash?08:00
jeamithro: do I have to flash to test?08:00
jeai thought i was doing load at hobart08:00
jeaflashing now...08:01
mithrojea: no, but at that prompt you need to reset the Opsis via either typing 'reboot' or hitting the reset button....08:01
mithro(when loading)08:02
jeapressing the reset button did not seem to bring up the device with my edid change08:02
jeaflashing failed, with the 0x1818181818 error08:02
mithroYou should see the bios prompt and the flterm should load the firmware08:02
mithrojea: make opsis-reset then try again08:03
jeadid reset-opsis, then firmware-flash, same 0x18181818 thing08:04
*** Choco31415 has joined #timvideos10:34
Choco31415xfxf: I'm not streaming all 6+ rooms. That's suicide! I'll only stream 1 room, maybe 2 if I get my roommate involved (he has bash/streaming experience as well).10:38
xfxfChoco31415: you want to have plenty of smaller event (i.e. meetup) experience before attempting a conference10:39
xfxfthere's more complication than you likely realise at the moment10:39
xfxfeven if only 1 room10:39
Choco31415xfxf: Carl suggested LUG meetings for some practice.10:44
xfxfChoco31415: that's a v good idea.  get a few of those under your belt and you'll be in an infinitely better position10:44
Choco31415xfxf: I doubt game livestreams count?10:45
xfxfChoco31415: streaming games using OBD has almost nothing in common with recording conferences10:45
xfxfapart from perhaps some of the streaming concepts which makes up about 10% of what you need to be familiar with10:46
xfxfdifferent problem set10:46
Choco31415xfxf: Or better yet, I could stream some of the presentations for clubs at my college.10:59
*** SamSagaZ has joined #timvideos12:10
tumbleweedmithro: :P12:20
*** Choco31415 has quit IRC12:28
*** Choco31415 has joined #timvideos12:30
*** rohitksingh_wor1 has quit IRC12:56
*** CarlFK has joined #timvideos13:28
*** ChanServ sets mode: +v CarlFK13:28
*** SamSagaZ has quit IRC13:49
CarlFK[m]   Return of the Slots14:16
tpbTitle: veyepar - Show: PyCon Australia Pycon_au_2017 (at
paddatrappermithro: Good news - the audio firmware (that just generates a tone) works on the atlys too15:28
*** SamSagaZ has joined #timvideos15:52
*** Choco31415 has quit IRC15:53
*** Choco31415 has joined #timvideos16:33
paddatrappermithro: though annoyingly set interface fails on the opsis17:03
*** SamSagaZ has quit IRC17:27
*** SamSagaZ has joined #timvideos17:34
CarlFK[m]tumbleweed (IRC):  I did "Rebuild the copied sources" from Yakkety to Artful and they all passed.  mind if I copy them into the timvideo PPA?17:42
*** Choco31415 has quit IRC17:43
tumbleweedCarlFK[m]: you won't be able to because you've rebuilt them17:46
CarlFKtumbleweed: um... what can I do?17:46
tumbleweeddon't rebuild when you're copying17:47
tumbleweedit's almost never what you want17:47
CarlFKthere are Yackety binaries in timvideo and Artful binaries in my PPA - I want Artful binaries in timvideos PPA.17:49
tumbleweedthen we copy-forward within the timvideos PPA, without the rebuild17:49
CarlFKcan I do that now?17:50
tumbleweeddo it17:51
CarlFKdone.  yay.  one less source of error/warning/distraction as I track down a gst memory leak17:54
*** SamSagaZ has quit IRC18:32
*** SamSagaZ has joined #timvideos18:45
*** SamSagaZ has quit IRC19:59
*** cr1901_modern has quit IRC22:10
mithrotumbleweed: morning23:23
mithrotumbleweed: I was weak too, actually got 7 hours of sleep :-P23:24
tumbleweedmithro: o/23:29
tumbleweedmithro: would you consider merging the CR-LF hacks?23:30
tumbleweedI think we're pretty happy with the state of the debconf17 branch23:30
mithrotumbleweed: Yes, can you send it as a pull request?23:45
jeaCarlFK: slots ftw23:49
micoloushey tumbleweed, I was trying to make a display that had oscilloscope + vu meter on the side of the video output from vocto, but the pipeline just stalled and failed23:49
micolousDo you know enough gstreamer-fu to fix my code?23:49
tpbTitle: video-scripts/ at master · xfxf/video-scripts · GitHub (at

Generated by 2.13.1 by Marius Gedminas - find it at!