Friday, 2016-04-29

*** tpb has joined #timvideos00:00
*** panther has joined #timvideos02:00
*** panther is now known as sab_12302:01
*** sb0 has joined #timvideos02:13
mithrotumbleweed: capture doesn't work at 1080p02:45
mithrotumbleweed: it would also half your frame rate02:53
*** CarlFK has joined #timvideos03:16
*** ChanServ sets mode: +v CarlFK03:16
sab_123mithro, to get started for the UART, I would write from USB into an end point after setting the correct vid and pid and use the hardware UART after setting up the timer . FOr the rx side, I would first set the isr interrupt after receiving the data arm the endpoint and wait correct?03:27
sab_123mithro, I remember that the last time we had this conversation you had some doubts if I understtod correctly. Just want to make sure this is what you need03:28
*** CarlFK has quit IRC03:44
*** CarlFK has joined #timvideos03:56
*** ChanServ sets mode: +v CarlFK03:56
mithrosab_123: no, what we need is a bunch of well documented, tested, composable parts which can be connected together to create what you described04:32
*** CarlFK has quit IRC04:51
sab_123mithro, UART needs to be bitbanged too right?04:52
sab_123mithro, because like you were saying , not all have serial ports04:52
mithrosab_123: yes, there are multiple different possibilities for the UART side - hardware uart, bit banged uart, fifo04:58
sab_123ok mithro , got it04:59
sab_123mithro, for the auto vectoring , on everyline I see BYTE dummy=INT2JT; which doesnt seem to be used at all. Do you want me to remove this??05:11
mithrosab_123: I have NFI what you are talking about - please be more descriptive05:12
tpbTitle: Fx2lib: include/usbjt.h Source File (at
sab_123line number05:12
mithrosab_123: I don't know why you are looking at the very old stuff on sourceforge?05:16
sab_123Its the same code on your fx2lib too05:16
mithrosab_123: Can you also explain why it exists?05:16
sab_123This one doesnt have it though05:18
tpbTitle: fx2lib/autovector.h at aaa470e377ca0bf861244b10cdf4bdc4615d8bc1 · djmuhlestein/fx2lib · GitHub (at
sab_123I have to look at the makefile05:19
sab_123see which git is being pulled in05:19
*** Bertl_zZ is now known as Bertl05:26
sab_123strange, its pulling in an older version05:27
sab_123 doesnt have it05:27
sab_123I just did a git clone and it pulls in one which has that line05:28
tpbTitle: fx2lib/autovector.h at master · timvideos/fx2lib · GitHub (at
mithrosab_123: do you know how to use gitk?05:33
sab_123nope, only git. If I do git clone
tpbTitle: GitHub - timvideos/fx2lib: Library routines for creating firmware for the Cypress FX2 (CY7C68013 and variants) with SDCC (at
sab_123I get a new one05:34
sab_123but on the HDMI2MISOC when I do make and it pulls in third party libs , I get an older version05:34
mithrosab_123: You should learn how to use gitk05:35
mithrosab_123: you also understand that there are different branches?05:36
sab_123master is default05:36
sab_123I havent changed it05:36
sab_123its pulling in detached branch by default05:38
sab_123that was the problem05:38
mithrosab_123: that is not a problem, it is how submodules work05:39
mithrosab_123: git submodules always point to a specific commit so that they are always consistent05:43
sab_123mithro, ah submodules , I have to read up on that, currently I will use the latest branch by checking out master06:04
sab_123I dont think there will be conflicts, the current commit is 201506:04
sab_123for the detached submodule which is being pulled in06:05
*** sb0 has quit IRC06:17
*** sb0 has joined #timvideos06:49
*** Bertl is now known as Bertl_oO06:51
*** aps_ is now known as aps07:23
tumbleweedmithro: ah. Half frame rate would probably be fine...07:59
mithrotumbleweed: I'm be around a bit later if you have people who want to hack on Opsis  / HDMI2USB stuff08:02
olasd(yes, I'm frustrated from yesterday :P)08:03
mithroolasd: I'm around now08:53
mithroolasd: what are you trying to do?08:53
olasdso, I have one of the Opsis boards DebConf bought in front of me; I'm trying to flash a more recent firmware to try and get some more video modes available08:54
mithroolasd: okay - step 1 is to get the FX2 on the board into "JTAG programming mode", there are a number of different ways to do that08:55
olasd(we're using one of the three other boards in 720p mode; the last one was DOA / the USB doesn't enumerate)08:55
olasdright, thus far I'm following :)08:55
mithroolasd: The "firmware" for running the FX2 in JTAG programming mode comes from my ixo-usb-jtag repository08:56
tpbTitle: GitHub - mithro/ixo-usb-jtag: usb-jtag - Altera USB Blaster Emulation with a FX2 (at
mithroolasd: or the ixo-usb-jtag deb08:57
mithroIt can be loaded using the fxload command08:57
mithroolasd: I have a tool called "HDMI2USB-mode-switch" which is designed to make it easy to do this08:58
mithroolasd: but it needs a bunch of love to get into a good state08:59
*** sab_123 has quit IRC08:59
mithroolasd: and
tpbTitle: GitHub - mithro/HDMI2USB-mode-switch at opsis-prod (at
mithroolasd: if you know people who know Python, they could definitely help with finishing that stuff09:03
mithroolasd: it would be good to get it packaged up for debian too09:04
mithrothen you could just go09:04
mithrohdmi2usb-mode-switch --mode=jtag09:04
olasdthat worked09:06
mithroolasd: which works? :P09:11
olasdthe jtag mode switch09:12
mithroolasd: what output did you get?09:15
olasdmithro: I'm now in a state where the board registers as idVendor=16c0, idProduct=06ad (Product: Opsis (JTAG Mode)).09:15
olasdthe output was just "found 1 board"09:15
mithro olasd run it with --verbose :)09:15
mithroolasd: you'll see the fxload and other commands it runs09:16
olasd"Board already in required mode (jtag)" :P09:16
tpbTitle: HDMI2USB-mode-switch/ at opsis-prod · mithro/HDMI2USB-mode-switch · GitHub (at
mithroProbably want to remove the eeprom programming steps09:17
olasdshould I replace the .bin with something newer?09:18
mithroolasd: if you are building / hacking on the firmware then you'll want to use "make flash" in
tpbTitle: GitHub - timvideos/HDMI2USB-misoc-firmware: A version of the HDMI2USB firmware based around the misoc+migen tools produced by m-labs. (at
olasdright, okay09:21
mithroolasd: but you could use prebuilt ones from
tpbTitle: GitHub - timvideos/HDMI2USB-misoc-firmware: A version of the HDMI2USB firmware based around the misoc+migen tools produced by m-labs. (at
mithroolasd: you should hack on
tpbTitle: Allow user to add custom EDID mode · Issue #63 · timvideos/HDMI2USB-misoc-firmware · GitHub (at
olasdokay; the flashing seems to be unhappy09:28
olasdlet me paste the full output09:29
tpbTitle: debian Pastezone (at
mithroolasd: if you get a device DNA that looks like this09:32
mithroDNA = 110000001100000011000000110000001100000010011110101000111 (0x181818181813d47)09:32
mithroit means the ixo-usb-jtag is confused09:32
olasdI do09:32
olasdwhich means it is09:32
tpbTitle: Pastebin | IRCCloud (at
mithroSwitch to the serial mode and then back to the jtag mode09:32
olasddone, and DNA is still the same09:33
mithroolasd: power cycle and try again09:34
mithrowe need to improve the FX2 firmware09:34
olasdDNA = 110000001100000011000000110000001100000011000000110000001 (0x181818181818181)09:35
olasdso; output was different and still failed; another serial -> jtag cycle and apparently the board is flashing09:37
olasdwrote 1507328 bytes from file spiflash/opsis_hdmi2usb-hdmi2usbsoc-opsis.bin in 56.280357s (26.155 KiB/s)09:38
mithroyeah, 110000001100000011000000110000001100000011000000110000001 means that its just looping back the JTAG rather than getting to the FPGA properly09:41
mithroolasd: Possibly because it isn't resetting the FPGA properly09:52
olasdI succeeded in flashing the HEAD revision of the HDMI2USB-misoc-firmware, that I built09:53
olasdmithro: is the output from make flash in the misoc-firmware repo09:56
tpbTitle: debian Pastezone (at
mithroIt's kind of hard for me to read french - but it looks like the flashing part worked09:56
olasdah, sorry09:57
mithroI think it says09:57
mithromake: *** No rule to make target "flash-fx2" necessary to "flash". Stop.09:57
olasdapparently there is no flash-fx2 target09:57
olasdthe rest of the flashing worked09:57
mithroYeah, you don't need to flash-fx2 on the Opsis and I haven't written the code to flash the fx on an atlys :P09:58
mithroolasd: if you power cycle after flashing09:58
mithroolasd: hopefully it will come up with new firmware09:58
olasdI did power cycle09:58
mithroolasd: did it come back? :P09:58
olasdand it just doesn't come back ;_;09:58
mithroolasd: it can take about 30 seconds?09:59
mithroolasd: what happens on the USB?09:59
olasdyeah; I see [72815.373462] usb 1-2: New USB device found, idVendor=2a19, idProduct=5440 appear09:59
olasdthen disappear09:59
olasdand then nothing09:59
mithroolasd: sounds like the FPGA is resetting the FX2 but it isn't successfully loading the new firmware10:01
mithro2a19 5440 is the "FX2 has power but hasn't loaded new firmware"10:02
olasdyeah, I gathered that from the usb ids page10:02
mithroolasd: try disconnecting the FX2 RST jumper10:02
mithroolasd: and loading the serial mode10:02
mithroolasd: and see what the firmware is saying10:02
mithroolasd: tumbleweed wasn't smart enough to get a LowSpeedIO board right?10:02
tumbleweedit didn't seem like something that would be useful to us10:03
mithroyou guys are in Vienna right?10:03
* olasd nods10:03
mithrotumbleweed: the LowSpeedIO has a UART on it, you can use that to connect to the serial console even when the FPGA is resetting the FX2 :P10:04
mithromakes it easy to debug what is going on with the FX210:04
olasdI'm sure we can find a ftdi somewhere10:04
olasdlots of people from the metalab around10:04
tumbleweedwe found one :)10:05
mithrotumbleweed: but were you going to connect it :P10:05
mithroI guess you could try shoving some wires into the TOFE connector10:05
olasdah, yeah :-)10:06
olasdso anyway10:06
mithroolasd: btw - you have the SDA jumper connected right?10:06
olasdI'm in serial mode; nothing happening on the serial port10:07
mithroolasd: hit enter?10:07
olasdI've done that :)10:07
mithroyou might need to be quick to catch what ever its outputting?10:08
mithroI was sure that I asked for the FX2 serial lines to be put on a header, but it never seem to end up in the final design10:08
olasdyeah, don't look like it10:09
mithroolasd: otherwise you could have connected a FTDI device there10:10
mithroolasd: _florent_ uses an adapter connected to the microSD pins10:11
tpbTitle: HDMI2USB-misoc-firmware/ at master · timvideos/HDMI2USB-misoc-firmware · GitHub (at
olasdokay; I've flashed the firmware from the mode-switch repo and the board is back10:13
mithroolasd: great10:13
mithroolasd: I haven't been testing with a production opsis board recently10:13
mithroolasd: so maybe something else is broken10:13
mithroNow since it is 8:30pm on a Friday evening and SLUG is over, I'm going to head home10:25
olasdwell, I'm in a situation where i'm able to recover a working firmware, so I should be fine for a while10:31
olasdat some point we might try to do something about the board that doesn't enumerate usb, if someone brings an external jtag board10:34
mithroolasd: can you generate gateware?10:51
olasdyes, I'm pretty sure make gateware succeeded earlier10:52
mithroIE "make gateware"10:52
olasdI have been multitasking as it takes a while, however10:53
mithroBtw you did an "export BOARD=opsis" right?10:53
olasdyeah :)10:53
mithroHa great, I was thinking maybe you had flashed Atlys gateware on an Opsis which doesn't work10:54
mithroolasd: if your any good at packaging, it work be good to get lm32 and or1k cross compilers into Debian10:55
olasdI won't touch cross compiler packaging with a ten feet pole, sorry :)10:55
olasdI tried that a few years back, it was a horrible horrible experience10:56
mithroThat's why we currently get you a working cross compiler with conda10:56
olasdoh, yeah, I understand that; can't say I love it, but I've seen the other side10:57
mithroLm32 stuff is all upstream10:57
mithroDunno if they helps10:57
olasd(getting a bare metal arm cortex m3 toolchain properly packaged; I think I started in 2008, got some packages working, but spent years explaining that no, the linux toolchains couldn't build properly for a bare metal target)10:57
olasd(then I threw up my hands in the air and got back to doing more productive stuff)10:58
mithroolasd: you a C programmer?10:59
olasdit's been a long while since I've done embedded stuff; I maybe last did some real C programming in 2009 or so, and my last fray with vhdl was in 2010. and I don't have nearly enough free time to get back into it unfortunately11:00
olasd(sorry :))11:01
mithroolasd: that is what you think ;)11:06
*** panther has joined #timvideos11:06
*** panther is now known as sab_12311:06
olasdno, that is what I know11:07
olasdI'm going to spend the summer herding 29 interns and their mentors for GSoC/Outreachy in Debian11:07
olasdand stop trying to sideband-volunteer me through tumbleweed :D11:07
*** panther_ has joined #timvideos11:32
*** sab_123 has quit IRC11:32
*** panther_ has quit IRC13:28
*** CarlFK has joined #timvideos14:22
*** ChanServ sets mode: +v CarlFK14:22
*** springermac_ has joined #timvideos14:29
*** springermac has quit IRC14:29
*** springermac_ is now known as springermac14:29
*** cr1901_modern1 has quit IRC14:35
*** kaalikahn has quit IRC14:35
*** _florent_ has quit IRC14:35
*** cr1901_modern1 has joined #timvideos14:39
*** kaalikahn has joined #timvideos14:40
*** _florent_ has joined #timvideos14:45
*** sb0 has quit IRC17:12
CarlFKmithro:  "Purchase a board from CrowdSupply or directly from"17:37
CarlFKI don't see how to purchase in either place17:37
*** CarlFK has quit IRC18:28
*** CarlFK has joined #timvideos18:37
*** ChanServ sets mode: +v CarlFK18:37
*** CarlFK has quit IRC18:59
*** CarlFK has joined #timvideos19:21
*** ChanServ sets mode: +v CarlFK19:21
*** CarlFK has quit IRC20:43
*** CarlFK has joined #timvideos20:58
*** ChanServ sets mode: +v CarlFK20:58
*** Bertl_oO is now known as Bertl_zZ21:34
*** CarlFK has quit IRC22:13

Generated by 2.13.1 by Marius Gedminas - find it at!