Monday, 2015-11-02

*** tpb has joined #timvideos00:00
tumbleweedapt had solved a dependency problem by using sid's openocd00:00
tumbleweed(which is newer)00:00
CarlFKthe problem is you are using my packaging :p00:01
tumbleweednow I get:00:03
tumbleweedError: unable to open ftdi device: device not found00:03
CarlFKpower cycle the atlys00:06
CarlFKa udev rule should auto load something00:06
CarlFKthat makes lusb show:00:06
CarlFKBus 002 Device 015: ID 16c0:06ad Van Ooijen Technische Informatica00:06
CarlFKBus 005 Device 004: ID 04e2:1410 Exar Corp.00:06
tumbleweedwhere do you get that udev rule?00:07
CarlFKum.. one of my apt-gets from notes.txt00:07
tumbleweedI have the Exar Corp device, and a Digilent Development board JTAG device00:07
tumbleweedyeah, I just figured that out :)00:08
mithrotumbleweed: don't use Carl's openocd package :-)00:10
tumbleweedmuch better :)00:10
tumbleweedmithro: what should I use?00:10
mithrotumbleweed: the firmware will download openocd from conda00:10
tumbleweed"the firmware" ?00:11
mithroThe get-env scripts00:11
tumbleweedthose sound like things I want to run, a lot less than carl's packages00:11
tumbleweedthe packages may be messy, but at least I know what they're doing00:12
tumbleweedand it looks like the messyness will resolve (one way or another)00:12
mithroCarlFK's instructions are not useful if you want to actually help with development or do anything apart from basic testing00:13
tumbleweedthat's fine. I don't want to be installing proprietary crap on my laptop00:13
tumbleweedgetting it working at all is the most useful thing for me to do now00:13
tpbTitle: HDMI2USB-misoc-firmware/ at master · timvideos/HDMI2USB-misoc-firmware · GitHub (at
mithroLove some help making those work on Debian00:14
mithroI'll be back in about 30 minutes00:15
tumbleweedI think I'll scare Debian Video people off a lot more with "install a proprietary development environment" than: "here are some messy debs that'll make it work"00:15
mithroJust getting up and heading to work00:15
tumbleweedI'm pretty sure we can make the production setup of this on Debian nice and simple00:15
tumbleweeddev is probably a lot harder (due to the Xilinx ISE)00:16
mithrotumbleweed: the production setup should be just install HDMI2USB-mode-switch00:16
mithrotumbleweed: but that needs packaging00:17
tumbleweedwell, the production setup should be "turn it on" :P (eventually)00:17
mithroYes, true00:17
tumbleweedanyway, it works00:17
mithroMode-switch will be needed if you want to upgrade the firmware00:17
mithroIf you wanted to be really fancy, you could have the gateware binary and the lm32 headers packaged, then people could still work on the lm32 firmware (which is just C) without needing the proprietary toolchain00:30
tumbleweedthat sounds useful00:31
mithrotumbleweed: do you know udev at all?00:34
mithroI'd like to setup /dev/hdmi/<serial number>/{video,tty} symlinks00:35
tumbleweedonly very vaguely00:43
tumbleweedCarlFK: did you find any VGA->HDMI convertors that work, yet?00:43
tumbleweedI got a couple to try, but they don't provide the right modes by EDID00:44
CarlFKtumbleweed: other than the one I got a year ago, I havent tried00:44
tumbleweedthis does 1080p, but not 720p
CarlFKmy current plan for Node: if your laptop doesn't do hdmi, you get the twinpact.00:46
CarlFKthe two rooms we are in only have vga to the projector00:47
CarlFKand I am not inclined to rewire them00:47
tumbleweedthat means you need working HDMI->VGA00:48
tumbleweeddo we have that?00:48
CarlFKI think so.. I need to make sure I have 200:53
CarlFKor just get 2 more.  they are cheep.. $15 ish00:53
*** jamesh has quit IRC01:05
*** jamesh has joined #timvideos01:05
mithroOkay, I'm around now01:39
mithroCarlFK: ping?01:58
CarlFKmithro: pong01:58
mithroCarlFK: on that fresh setup, can you do a "conda update openocd" and then try "make load-gateware" step again?01:59
CarlFKmithro: sure.01:59
CarlFKand I confirmed that it didn't work before the "conda update openocd"02:11
CarlFKmithro: ^^^02:12
mithrookay, great - can you try from scratch again?02:12
mithroCarlFK: A while back I asked if you wanted to manage the stable/testing links in the prebuilt repo02:12
CarlFKmithro: sure - before I reimage the box, next step is erroring02:19
CarlFK[email protected]:~/HDMI2USB-misoc-firmware$ make load-fx2-firmware02:19
CarlFKmake: *** No rule to make target `load-fx2-firmware'.  Stop.02:19
mithroIt should just be "make load-fx2"02:20
CarlFKwant me to update:
tpbTitle: HDMI2USB-misoc-firmware/ at master · mithro/HDMI2USB-misoc-firmware · GitHub (at
CarlFKremove "PROG=fpgalink"  and change to  "make load-fx2"02:21
mithroCarlFK: gah - why did you sent that pull request to my repo and not the timvideos repo?02:32
mithroCarlFK: can you resend that to the timvideos repo?02:32
CarlFKbecause thats what I tested02:32
CarlFKI have no idea how to send that somewhere else02:33
mithroWhen you create the pull request, it will say where you are sending it too02:34
mithroWhen you click the green "pull request" button, the first drop down is where you send the pull request02:35
jameshmithro: have you had a chance to look at the test suite fixes pull request for gst-switch?02:44
mithrojamesh: sorry, not really02:45
mithrojamesh: haven't really had any time to think about gst-switch stuff at the moment02:45
mithrojamesh: have less then a month to get everything ready for shipping the Opsis boards :-/02:57
*** travis-ci has joined #timvideos03:19
travis-ci[mithro/HDMI2USB-misoc-firmware/master#40] (d7497b6): The build passed. (
*** travis-ci has left #timvideos03:19
mithroBertl_zZ: ping? My OHS-Park order of your boards arrived!04:52
mithroBertl_zZ: the joins / berrs on the PCB board are a bit nasty04:55
*** travis-ci has joined #timvideos05:38
travis-ci[mithro/HDMI2USB-misoc-firmware/master#41] (ff33b4f): The build passed. (
*** travis-ci has left #timvideos05:38
mithroCarlFK: Okay, I merged your changes upstream05:43
mithroCarlFK: and fixed a couple of small things too05:43
*** CarlFK has quit IRC05:46
*** CarlFK has joined #timvideos05:52
*** ChanServ sets mode: +v CarlFK05:52
CarlFKmithro: thanks05:54
CarlFKmithro: about managing testing/stable - I am not sure how to evaluate stability05:56
CarlFKmithro: I am happy to report my level of comfort and for the next two weeks it would be handy to me if thats all that mattered05:57
mithroCarlFK: at the moment you are the biggest user, so "stable for you" is a good proxy for th emoment?05:57
CarlFKfor now05:58
*** travis-ci has joined #timvideos05:59
travis-ci[mithro/HDMI2USB-misoc-firmware/master#42] (74ad203): The build passed. (
*** travis-ci has left #timvideos05:59
mithroCarlFK: basically, the idea would be you test the unstable branch and if you are happy it works, promote to testing06:00
CarlFKI am happy enough to say promote it now06:01
mithroCarlFK: which version?06:01
CarlFKfirmware revision: 8e860e3a, built Nov  1 2015 09:32:3106:03
*** travis-ci has joined #timvideos06:19
travis-ci[timvideos/HDMI2USB-misoc-firmware/master#275] (6a57c02): The build passed. (
*** travis-ci has left #timvideos06:19
*** Bertl_zZ is now known as Bertl06:50
*** CarlFK has quit IRC07:37
*** se6astian|away is now known as se6astian07:53
*** CarlFK has joined #timvideos08:28
*** ChanServ sets mode: +v CarlFK08:28
CarlFKmithro: it has been on this for over 30 min:08:30
CarlFKUsing target part "6slx45csg324-3".08:30
CarlFKMapping design into LUTs...08:30
CarlFK3am.. bed time :p09:03
CarlFKoh hell...09:04
CarlFKWrite failed: Broken pipe09:04
CarlFKtmux... make gateware  ... lets see what happens when I wake up09:06
*** Bertl is now known as Bertl_oO09:44
mithro_florent_: you about? I now have a pretty good understanding of how the USB needs to work (IE What needs to happen on the FX2 and signals the gateware needs to send) but I'm unsure on the correct way to implement it10:33
mithro_florent_: basic theory is that the LM32 needs to generate a header, set a couple of bits and then DMA out the header+encoder data following it (and stalling on a FIFO full bit)10:43
*** travis-ci has joined #timvideos11:18
travis-ci[mithro/HDMI2USB-misoc-firmware/master#43] (36b6341): The build passed. (
*** travis-ci has left #timvideos11:18
*** travis-ci has joined #timvideos11:19
travis-ci[mithro/HDMI2USB-misoc-firmware/fx2-refactor#45] (e177d3a): The build has errored. (
*** travis-ci has left #timvideos11:19
_florent_hi mithro, sorry I don't understand what you are doing exactly: why do we need to change the way the USB works?11:42
mithro_florent_: because I'd like to decouple the USB and the JPEG parts and move the USB stuff into migen/misoc11:44
_florent_ah OK,11:44
*** travis-ci has joined #timvideos11:45
travis-ci[mithro/HDMI2USB-misoc-firmware/opsis-fixes#44] (48477f3): The build passed. (
*** travis-ci has left #timvideos11:45
_florent_so moving code from the FX2 to the gateware?11:45
mithro_florent_: Well, the gateware / lm32 firmware needs to eventually send configuration information over to the FX2 (IE the resolutions)11:47
mithroso, I've been rewriting almost all of the FX2 firmware to be more understandable11:49
tpbTitle: Google Docs - create and edit documents online, for free. (at
_florent_so we should change the usb streamer and add a ram to it that will contain the header and for each JPEG frame, send the header first followed by the frame11:49
mithro_florent_: yes, the FX2 will act pretty much like a FIFO11:50
mithro_florent_: you just keep sending bytes until the "FIFO full flag" is raised11:50
mithro_florent_: then you pause until it is deasserted11:50
_florent_ok, that seems easier indeed11:51
mithro_florent_: when you finish sending a full frame, you then toggle the "packet end" signal11:51
_florent_yes that seems fine11:52
*** travis-ci has joined #timvideos11:52
travis-ci[mithro/HDMI2USB-misoc-firmware/fx2-refactor#46] (6a8fbed): The build failed. (
*** travis-ci has left #timvideos11:52
mithro_florent_: I was really hoping to figure out a way to put the other end points under the LM32 control as the FX2 is super sucky - but it isn't really possible11:53
mithro_florent_: Getting the CDC port working again is pretty simple too.11:53
mithro_florent_: there are two endpoints, one you write data to11:54
mithro_florent_: the other you read data from11:54
mithro_florent_: the FIFOADDR[1.0] pins control which endpoint the fifo interface is connected too11:54
_florent_ok, for now you are emulating an UART with the FX2?11:55
mithro_florent_: yeah - but that only works on the opsis11:56
mithro_florent_: Sadly, the hardware UARTs arn't exposed on the FX2 onboard the Atlys11:57
mithro_florent_: so, the process for sending a UART byte, is to interrupt the pending video data being sent, setting the FIFO address pins, writing the UART bytes, toggling pktend, restoring the FIFO address pins and allowing the video the continue where it left off11:59
mithro_florent_: but for now we can ignore that12:01
*** travis-ci has joined #timvideos12:02
travis-ci[mithro/HDMI2USB-misoc-firmware/fx2-refactor#47] (f45dede): The build failed. (
*** travis-ci has left #timvideos12:02
mithro_florent_: I might just do software bit-banging at 9600 board or something on the spare pins12:03
mithro_florent_: What I also wanted to mention is that I've been working on some FX2 firmware which makes it emulate a FTDI 232H chip as a better replacement for ixo-usb-jtag - I figured that might mean it would work pretty well with your liteUSB stuff12:12
_florent_ah ok, then we will be able to share code between liteUSB and the HDMI2USB gateware12:15
mithro_florent_: yeah - this is more to make the Opsis more compatible with xc3sprogs and a bunch of proprietary JTAG tools under Windows / Mac12:17
mithro_florent_: heading out now12:58
shenkimithro: im in singapore!13:02
shenkimithro: guess what the suburb im in is called? Novena!13:03
*** rohitksingh has joined #timvideos13:09
mithroshenki: ha!13:40
mithroIs it open source13:40
CarlFKmithro:  all works:
tpbTitle: HDMI2USB-misoc-firmware/scripts at master · timvideos/HDMI2USB-misoc-firmware · GitHub (at
*** jamesh has quit IRC15:37
*** jamesh has joined #timvideos15:46
*** se6astian is now known as se6astian|away16:00
*** rohitksingh1 has joined #timvideos16:07
*** rohitksingh has quit IRC16:08
*** Bertl_oO is now known as Bertl19:22
*** rohitksingh1 has quit IRC19:43
*** rohitksingh has joined #timvideos19:49
*** se6astian|away is now known as se6astian21:25
*** se6astian is now known as se6astian|away22:17
mithrohey Bertl22:46
CarlFKmithro: the GChris pipeline does not transcode, it just dumps jpeg into a matroska container.  and Shotcut video editor gets hartburn (eats up memory, cpu etc.. crashes)22:53
mithroCarlFK: poor old shotcut22:54
CarlFKdo you have a suggestion for what gstreamer can transocde to?22:54
mithroCarlFK: btw you probably want to delete your HDMI2USB-misoc-firmware that you forked from my repo and refork from the timvideos repo22:54
mithro_florent_: heyo, I see the "encoder: new buffer almost working" patch in your repo - That stuff looks close enough that I think I'm understanding the right way to do this23:01
mithro_florent_: sorry that should have been the "replace buf_fifo with an optimized reader/buffer" patch23:02

Generated by 2.13.1 by Marius Gedminas - find it at!