Saturday, 2017-02-25

*** tpb has joined #timvideos00:00
thaytanxfxf, sorry - I missed your entire reply00:58
thaytanI meant that I have a core 2 ubuntu machine that's our webserver that could get some extra things on it, and my main dev machine that could take a chroot if needed, plus misc raspberry pi's and things that could generate test feeds00:59
thaytanxfxf, let me know when you're on your way00:59
*** gk_-1wm_- has joined #timvideos01:39
*** gk_-1wm_- has left #timvideos01:39
CarlFKthaytan: I am surprised you dont have some i3 sort of class machines for testing hardware things01:43
xfxfthaytan: I'm omw now, running a little late sorry. Google maps says 4 hrs02:40
*** puck is now known as puck``03:08
*** [d__d] has quit IRC05:25
*** [d__d] has joined #timvideos05:27
*** CarlFK has quit IRC06:42
*** CarlFK has joined #timvideos06:52
*** ChanServ sets mode: +v CarlFK06:52
xfxfmithro: you around? having issues building gateware on latest nextgen branch09:31
xfxf+ rolling back doesn't work as submodules have changed, repo's have gone etc09:37
mithroxfxf: I noticed that the CI was broken but haven't had a chance to look at it09:47
xfxfright, thaytan and I are trying to get a build env running on his system (Fedora)09:47
xfxfWe think it's good but want to actually build a firmware09:48
mithroxfxf: Well, at the moment all I know is you are having issues :-P some error outputs might be helpful09:59
xfxfwe're progressing as i managed to tarball up a version i had running at lca w/ submodules already installed, but i'll paste the current error10:01
thaytanmithro, Sherlock Holmes would have worked it out by now :(10:01
thaytanValueError: Cannot extract CSR name from code, need to specify.10:01
thaytan  File "/home/jan/devel/HDMI2USB/HDMI2USB-litex-firmware/third_party/litex/litex/soc/interconnect/", line 37, in __init__10:01
xfxf^ that10:01
tpbTitle: 2>&1 | tee -a /home/jan/devel/HDMI2USB/HDMI2USB-litex-firmware/build/opsis_video - (at
thaytanand that I'm running on fedora, where there is no tempfile command :)10:02
xfxfdoes same thing on my machine which i repeatedly built gateware/firmware on at lca (ubuntu LTS)10:02
mithroThat is what the CI is failing with10:29
mithrooh... I might know what the problem is10:30
mithroxfxf / thaytan: Can you run "python --verison"10:31
thaytan[[email protected] ~]$ python --version10:32
thaytanPython 2.7.1310:32
thaytan[[email protected] ~]$ python3 --version10:32
thaytanPython 3.5.210:32
thaytanmithro, might just need to change a #! somewhere?10:34
xfxf(H2U P=opsis R=nextgen) [email protected]:~/dev/video/HDMI2USB-litex-firmware$ python --version10:34
xfxfPython 3.6.0 :: Continuum Analytics, Inc.10:34
xfxf(H2U P=opsis R=nextgen) [email protected]:~/dev/video/HDMI2USB-litex-firmware$ python3 --version10:34
xfxfPython 3.6.0 :: Continuum Analytics, Inc.10:34
xfxfit's the same for thaytan, he hadn't source'd setup-env.sh10:35
thaytanpython --version10:35
thaytanPython 3.6.0 :: Continuum Analytics, Inc.10:35
mithroI'll puth a tempory fix shortly10:36
mithroTwo options it seems10:51
mithro or make conda use python 3.5 for now10:51
tpbTitle: gen/fhdl/tracer: port to python 3.6 · povauboin/[email protected] · GitHub (at
mithroxfxf / thaytan: BTW There is an open issue where you build gateware and it results in an output which fails memtest11:02
thaytanmithro, yep, that commit fixes it11:02
thaytanI'm building 2 gateware now - xfxf's tree from a month or so ago and nextgen with that fix cherry-picked11:03
mithroSadly the current fix is, "just try again"11:03
thaytanoh great11:03
mithrothaytan: I'm testing a different "work around" ->
tpbTitle: Force conda version to Python 3.5 by mithro · Pull Request #44 · mithro/HDMI2USB-litex-firmware · GitHub (at
thaytannon-deterministic build11:03
thaytanmithro, 3.5 is what ryan's "old tree" has11:04
thaytanI wish the build could multithread11:04
mithrothaytan: Yeah, the setup script is pointing at "conda latest" which moved to 3.6 like 2 weeks ago it seems11:04
thaytanthere's 11 cores sitting idle11:04
mithrothaytan: In what way?11:04
thaytanxilinx is single thread it seems11:05
mithrothaytan: Yes correct11:08
mithroISE is mostly single threaded11:08
mithrothaytan: so, why are you and xfxf having a party without me? :-P11:09
xfxfoh humm - i'm trying to do 'make gateware-load' but the usb device goes down after loading and doesn't come back up again11:09
*** thaytan has quit IRC11:09
xfxfi vaguely recall this is related to the etheernet thing11:09
*** thaytan has joined #timvideos11:14
*** ChanServ sets mode: +v thaytan11:15
mithroxfxf: What version are you building?11:17
mithroxfxf: I was sure that I made it so the FX2 would still come up by default if no firmware is loaded11:18
xfxfoh it's an old version from like a month back, the recent one doesn't build for me11:19
xfxfactually i'm a bit tired, d/w, me trying to report bugs to you for an old version is stupid11:19
tpbTitle: Make the reset signals be Tristates · mithro/[email protected] · GitHub (at
tpbTitle: debconf-video/ansible - Ansible playbooks for the DebConf Video Team's machines (at
mithroxfxf / thaytan: That change has been merged11:27
xfxfta, i'll try building agan11:28
mithroxfxf: You'll need to re-run ./scripts/ (running ./scripts/ should complain)11:29
xfxfmithro: ack, noticed those files changed, doing that11:30
thaytanI'm building with the python 3.6 change instead11:32
xfxfappears to be building now \o/11:34
mithrothaytan: Yeah - that fix needs to be cleaned up to allow both Python 3.6 and Python 3.511:35
thaytanI built a gateware11:41
thaytanI built a firmware11:41
thaytanI loaded a gateware11:41
thaytan(it's a Koan)11:41
xfxfmithro: it's make gateware-load, firmware-load, then flterm and 'serialboot', right?11:43
xfxf(i may be remembering wrong, docco is super out of date now)11:43
xfxfnm, sorted11:44
mithrofirmware-load will run flterm11:46
mithroand serialboot is correct11:46
thaytanit didn't because my udev alias didn't appear11:46
xfxfyeah, there was a udev issue, so it wasn't running, thaytan figured it out11:46
thaytanbut it's OK, I ran it manually11:46
thaytanmithro, I'm doing all this on Fedora 25, btw11:47
mithroHowever, I highly recommend setting up tftp loading of firmware11:47
thaytanso some differences but it's working11:47
thaytanfor now, I just want a working board to do GStreamer / vocto stuff against11:47
mithrothaytan: If anything needs to change in the scripts, happy to see Fedora tested11:47
thaytanno need to load firmware too often11:47
thaytanmithro, the only thing that was actually wrong is Fedora has a newer libftdi that has and the included openocd wants (different lib name and .so version)11:48
thaytanbut the API seems compat, because a symlink worked11:48
mithroyou might want to install the udev rules from
tpbTitle: HDMI2USB-mode-switch/udev at master · timvideos/HDMI2USB-mode-switch · GitHub (at
mithroBTW You have roughly until I finish my dinner before I head to bed11:49
thaytanmithro, udev rules are installed, but I only got /dev/hdmi2usb/by-num/opsis0/usbdev instead of /dev/hdmi2usb/by-num/opsis0/tty11:50
xfxfmithro: yeah, already pointed him to that11:50
thaytanI have a firmware flashed, but it seems to be booting in JTAG mode until I switch it11:51
mithrothaytan: Hrm... the bot should *never* boot in JTAG mode - have you installed the old ixo-usb-jtag rules which do auto-fxload stuff?11:52
thaytanmithro, I have the rules from master udev/ subdir11:52
tpbTitle: GitHub - timvideos/HDMI2USB-mode-switch: Tool for switching boards supported by HDMI2USB firmware between multiple different modes (programming, webcam, etc). (at
thaytanI don't see any automatic fxload there11:53
mithrothaytan: Those are the right ones11:53
mithroYou *dont* want this one ->
tpbTitle: ixo-usb-jtag/ixo-usb-jtag.rules at master · mithro/ixo-usb-jtag · GitHub (at
thaytanoh, I do have that11:55
thaytansomething installed it, must be from the HDMI2USB-litex-fimrware repo11:55
mithrothaytan: I'm guessing its from the old firmware11:55
mithrothaytan: were was that file installed?11:56
mithroBTW the gateware we used at LCA is here ->
tpbTitle: Index of /~tim (at
thaytanmithro, has 98-hdmi2usbaux-ixo-usb-jtag.rules right there11:58
tpbTitle: HDMI2USB-mode-switch/udev at master · timvideos/HDMI2USB-mode-switch · GitHub (at
mithrothaytan: That is different to the ixo-usb-jtag.rules above11:58
mithroCheck for a /lib/udev/rules.d/85-ixo-usb-jtag.rules11:59
mithroshould have "fxload" in it11:59
thaytannothing in /lib/udev/rules.d no12:00
thaytanmithro, when I power the board12:03
tpbTitle: pastebin - Miscellany - post number 3772126 (at
thaytandoes the board put itself in serial mode, or should udev?12:03
mithrothaytan: Oh, 5440 is not JTAG mode, that is the "failsafe mode"12:04
mithroThe board will always some up with that at first and then disconnect and come up with the right one12:04
mithrothaytan: HDMI2USB-mode-switch can be used to put it into serial mode12:04
mithrothaytan: See targets/opsis/Makefile.mk12:05
thaytanmithro, OK - I can switch it to serial OK12:06
thaytanit's coming up with the new firmware12:06
thaytanbut it's not switching on its own at boot12:06
thaytanI get WARNING:root:unbind-helper not found, will have to run as root! too12:06
thaytannot sure what that means12:07
mithrothaytan: It'll only switch after firmware is loaded12:07
tpbTitle: pastebin - Something - post number 3772129 (at
thaytanthat's after I switch to serial12:07
mithroso unless you flash the firmware or load the firmware via network it won't come up on it's own until after the serialboot command12:07
thaytanthat's the firmware in flash12:08
thaytan(I ran make flash-opsis)12:08
mithrothaytan: and you power cycled after flashing?12:08
thaytanand if I run 'reboot' in the flterm console, it'll come back up in JTAG mode12:08
mithrothaytan: 5440 != JTAG mode12:09
mithrothaytan: reboot also != power cycle12:09
mithroDoes xfxf have the Opsis with the LowSpeedIO board + ftdi adapter there?12:11
thaytanmithro, I have also pulled the cable since uploading the firmware to flash12:12
mithrothaytan: So, after disconnecting the power and connecting the power again the board should come up as 5440 for a couple of seconds and then disappear and come back as 544212:13
thaytanand OK - powersafe != JTAG (except that's what it says in the syslog :-P)12:13
thaytanpower cycle12:14
thaytansorry, getting late for me12:14
tpbTitle: pastebin - Mine - post number 3772132 (at
thaytanthat's a clean power cycle (actually the same as the one before)12:14
mithroOpsis (JTAG Mode) is idVendor=16c0, idProduct=06ad12:14
thaytanit goes 5440 then disconnects and comes back as 16c0, idProduct=06ad12:14
mithroWhich target did you build the gateware for?12:15
thaytanthat's the same as the earlier paste - I think you missed it before12:15
thaytanmithro, opsis, nextgen branch12:15
thaytan(H2U P=opsis)12:15
mithroexport TARGET=hdmi2usb12:16
mithroLooks like we set the default target to video, which doesn't include the JPEG encoder + usb interface12:16
thaytanOK, cool12:17
thaytancompiling gatware12:17
thaytanwhat's the difference between gateware and firmware?12:17
mithroWhich is why its coming up in JTAG mode12:17
thaytanFPGA bitstream vs CPU code?12:17
mithrogateware == stuff which configures the FPGA (IE the hardware description in FPGA gate terms)12:18
mithrofirmware == stuff which runs on the softcore embedded inside the FPGA gateware12:18
thaytanOK, so reloading the gateware should be sufficient, or firmware too?12:18
mithroFirmware is gateware specific12:19
mithrobecause it depends on location of registers and stuff12:19
mithrowhich will move about if features are enabled/disabled12:19
thaytanbuilt gateware in 8m20.725s12:30
thaytanunlike ryan's "old tree" which seemed to get stuck12:30
xfxfI take no responsibility for that tree :P12:33
xfxfthat said, built for me okay in 15 mins or so12:33
thaytantime python -u ./ --platform=opsis --target=video --cpu-type=lm32 --iprange=192.168.100    \12:42
thaytaneven though I have (H2U P=opsis T=hdmi2usb R=nextgen)12:42
mithrothaytan: The gateware should take ~30 minutes12:43
mithrothaytan: did you do a "TARGET=hdmi2usb" or an "export TARGET=hdmi2usb" ?12:46
mithrothaytan: Try "make help" and see what that says too...12:46
thaytanerr derr12:46
mithrothaytan: Do you know if there is a way to "detect" the export value on a variable?12:47
thaytanyes, that's building it correctly now12:47
thaytanmithro, export | grep $VAR ?12:47
mithrothaytan: I was thinking "hdmi2usb_prompt() {" function which updates the prompt could check that export is set on the TARGET/PLATFORM/PROG stuff...12:49
thaytanexport with no args lists all exported vars12:59
thaytanso that'd work12:59
thaytantook considerably longer that time13:05
thaytanmithro, there's no line hooked to a hardware reset that JTAG mode could exercise?13:11
thaytan(is the physical power-cycle really needed?)13:11
thaytannm - I'll check it in the morning when I'm back next to the board13:14
mithrothaytan: We just added something like that to mode-switch13:15
mithrothaytan: but I haven't tested on an Opsis yet13:15
thaytanmithro, whereabouts?13:16
mithrohdmi2usb-mode-switch --reset-gateware13:16
mithromight not be merged yet13:16
tpbTitle: Add --reset-gateware by mithro · Pull Request #54 · timvideos/HDMI2USB-mode-switch · GitHub (at
thaytanyeah, not merged13:21
thaytanI think I won't play with that tonight, I'll just wait until morning and walk back to the shed and pull the cable13:22
*** springermac has quit IRC15:38
*** springermac has joined #timvideos15:39
*** rohitksingh has joined #timvideos16:35
*** SamSagaZ__ has joined #timvideos17:26
*** SamSagaZ_ has quit IRC17:29
*** CarlFK has quit IRC20:53
*** rohitksingh has quit IRC21:36
*** rohitksingh has joined #timvideos21:36
*** rohitksingh has quit IRC22:24

Generated by 2.13.1 by Marius Gedminas - find it at!