| *** tpb has joined #timvideos | 00:00 | |
| *** nueces has joined #timvideos | 00:10 | |
| xfxf | mithro: i'm getting absolutely nowhere with flashing the boards again. i don't know what's different about now than what i did last week. | 02:34 |
|---|---|---|
| xfxf | https://www.irccloud.com/pastebin/GN9yEVjZ/ | 02:34 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 02:34 |
| xfxf | i gig for a couple of hours, have something to do with my son (had intended on letting this record for a few hours while i was out). if i am doing something wrong above, please let me know. | 02:35 |
| mithro | I don't see any mode switch commands there | 02:37 |
| mithro | Did you look at the wip updated instructions? | 02:38 |
| xfxf | yes. the mode switch commands have *never* worked for me | 02:40 |
| xfxf | can you point me to exactly what i should do? | 02:40 |
| xfxf | there are lots of different instructions now | 02:40 |
| mithro | https://github.com/mithro/HDMI2USB-misoc-firmware/blob/docs-scripts-update/scripts/README.md | 02:43 |
| tpb | Title: HDMI2USB-misoc-firmware/README.md at docs-scripts-update · mithro/HDMI2USB-misoc-firmware · GitHub (at github.com) | 02:43 |
| mithro | hdmi2usb-mode-switch --mode=jtag | 02:43 |
| mithro | If you get an ID which starts with 0x18 when programming | 02:44 |
| mithro | Switch to serial mode and then back to JTAG and try again | 02:45 |
| mithro | None of that has changed since LCA | 02:46 |
| *** sb0 has quit IRC | 05:08 | |
| *** danielki has quit IRC | 05:16 | |
| *** ssk1328 has joined #timvideos | 05:34 | |
| *** sb0 has joined #timvideos | 05:46 | |
| *** sb0 has quit IRC | 06:00 | |
| xfxf | mithro: still having issues, but hdmi2usb-mode-switch actually seems to be doing things now. it wasn't at all the other day and i haven't reinstalled it | 07:59 |
| xfxf | i'll try a few other things now | 07:59 |
| xfxf | right now it's complaining about missing files, i'll try building the gateware again | 08:00 |
| mithro | xfxf: When you say things like that, I really worry about what you are doing | 08:00 |
| xfxf | this is why i have pasted you exactly what i have done, so you can verify that i'm doing the right thing | 08:00 |
| xfxf | in all fairness, when the official instructions are full of 'FIXME's', it's hard to know what i am meant to do :) | 08:00 |
| xfxf | but i'm repeating all steps now, but ensuring the opsis is in jtag mode before flashing | 08:01 |
| mithro | xfxf: It will leave JTAG mode after flashing | 08:01 |
| xfxf | yeah i realise, then i use the tool to stick it back into another mode | 08:01 |
| xfxf | like i said the other day, with two boards, that tool did absolutely nothing | 08:01 |
| xfxf | i have no idea why it is now | 08:01 |
| xfxf | i'll go back and install another laptop from scratch later tonight because 'i don't know why this didn't work' concerns me | 08:02 |
| xfxf | hmm, that README mentions jumpers on the opsis, but that they're set properly from factory | 08:21 |
| xfxf | i'm just wondering, maybe some of these opsis's jumpers were messed with @ LCA? | 08:21 |
| xfxf | do you know what the default positions should be? | 08:21 |
| xfxf | the issue i was having the other day is all of them were booting up in jtag mode by default, no serial port appeared, and the mode switch tool wasn't working | 08:22 |
| xfxf | i assumed it was a firmware/gateware thing but maybe it's a jumper? | 08:22 |
| xfxf | and it'd be good to know what they should be anyway - even if they're meant to be set properly from factory, i assume the manafacturer still had a human fit the jumpers | 08:23 |
| xfxf | mithro: i'm getting nowhere | 08:35 |
| mithro | xfxf: That doesn't tell me anything that I can help you with | 08:37 |
| xfxf | https://www.irccloud.com/pastebin/o1yIZguJ/ | 08:40 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 08:40 |
| mithro | xfxf: run hdmi2usb-mode-switch with -v | 08:41 |
| mithro | xfxf: You should also setup hdmi2usb-mode-switch so you don't need to run it as root | 08:42 |
| mithro | xfxf: also, type "which openocd" | 08:43 |
| mithro | xfxf: Also - check that your git tree is actually clean | 08:44 |
| *** ssk1328 has quit IRC | 08:45 | |
| xfxf | https://www.irccloud.com/pastebin/3taxvzz0/ | 09:15 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 09:15 |
| xfxf | git is clean, this is a completely new clone | 09:16 |
| xfxf | (H2U B=opsis) xf@yogacore:~/HDMI2USB-misoc-firmware$ sudo hdmi2usb-mode-switch -v | 09:17 |
| xfxf | WARNING:root:unbind-helper not found, will have to run as root! | 09:17 |
| xfxf | Traceback (most recent call last): | 09:17 |
| xfxf | File "/usr/local/bin/hdmi2usb-mode-switch", line 9, in <module> | 09:17 |
| xfxf | load_entry_point('hdmi2usb.modeswitch==0.0.0+93.gd0eb917', 'console_scripts', 'hdmi2usb-mode-switch')() | 09:17 |
| xfxf | File "/usr/local/lib/python3.5/dist-packages/hdmi2usb.modeswitch-0.0.0+93.gd0eb917-py3.5.egg/hdmi2usb/modeswitch/cli.py", line 183, in main | 09:17 |
| xfxf | assert len(found_boards) == 1, found_boards | 09:17 |
| xfxf | AssertionError: [] | 09:17 |
| xfxf | ugh, sorry, that should have been a pastebin | 09:17 |
| xfxf | that's after the commands pasted above. resetting the board, it then doesn't error + tells me it's in jtag mode | 09:18 |
| xfxf | have no idea what i am doing wrong | 09:18 |
| xfxf | wait, nm, now i flashed it and i'm on on the console | 09:21 |
| xfxf | i literally did the exact same things again on a fresh reset | 09:21 |
| xfxf | hrm | 09:21 |
| *** sb0 has joined #timvideos | 09:21 | |
| mithro | xfxf: I mean use "-v" on the commands it will tell you want the tool is trying to do | 09:34 |
| xfxf | yeah, something super funky is going on. the opsis just hard locked on me again | 09:34 |
| mithro | xfxf: Explain "hard locked" | 09:35 |
| xfxf | and i can't connect to its serial port anymore | 09:35 |
| xfxf | the video on my HDMI monitor has frozen | 09:35 |
| xfxf | sticking on the last frame | 09:35 |
| mithro | xfxf: What is showing on lsub and the LEDs? | 09:35 |
| xfxf | single red led | 09:35 |
| xfxf | Bus 002 Device 050: ID 1d50:60b7 OpenMoko, Inc. | 09:36 |
| xfxf | no errors in dmesg | 09:36 |
| xfxf | this was after 5 minutes of the opsis being on | 09:36 |
| mithro | xfxf: and dmesg | 09:36 |
| xfxf | https://www.irccloud.com/pastebin/OirmZgtu/ | 09:37 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 09:37 |
| mithro | xfxf: So, if you try and connect to the lm32 it doesn't work? | 09:37 |
| xfxf | nope | 09:37 |
| xfxf | just hangs | 09:37 |
| xfxf | [FLTERM] starting | 09:37 |
| xfxf | if i hit enter, nothing | 09:37 |
| mithro | xfxf: okay, Can you switch the board into serial mode and try connecting again | 09:38 |
| xfxf | doesn't work | 09:48 |
| xfxf | https://www.irccloud.com/pastebin/28eHRQgk/ | 09:48 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 09:48 |
| xfxf | i can reset the board, i assume you meant try to do that while it's 'locked up' | 09:48 |
| *** Bertl_zZ is now known as Bertl | 09:53 | |
| mithro | xfxf: for some reason it thinks you have an Atlys board, which might means the wrong fx2 firmware is being loaded | 09:57 |
| xfxf | i don't see how - i've exported the right env variable | 09:57 |
| xfxf | and i have literally reset the device, re-ran the commands, and they worked fine | 09:57 |
| mithro | xfxf: Did you build the fx2 firmware without the env set and then set it? | 09:58 |
| xfxf | the error may be misleading, will it say that if the device is detected but not responding? | 09:58 |
| xfxf | no, rebuilt everything with the env set | 09:58 |
| mithro | I'm just verifying now | 09:58 |
| xfxf | in any case, i'm using the fx2 firmware right now fine after i reset it and re-flashed | 09:59 |
| xfxf | actually, is the fx2 the serial, the UVC, or both? | 10:00 |
| xfxf | or are one of those the lm32? | 10:00 |
| mithro | xfxf: The fact the FX2 is coming up at 1D50 60B7 means it is confused - an Opsis should come up as a 2A19 5442 | 10:00 |
| xfxf | hmm okay | 10:00 |
| mithro | xfxf: The fx2 is UVC+Serial | 10:00 |
| mithro | xfxf: The lm32 is not a real device, it is inside the FPGA | 10:00 |
| xfxf | ah, right, makes sense | 10:01 |
| xfxf | okay, i'll make clean and rebuild everything | 10:01 |
| xfxf | oh wait | 10:01 |
| xfxf | hdmi2usb-mode-switch is trying to flash its own fx2 firmware when changing the mode to serial | 10:01 |
| xfxf | maybe that's where it's going wrong? | 10:01 |
| mithro | "flash" and "load" are two different things | 10:02 |
| xfxf | right. sorry, i mean load | 10:02 |
| mithro | "flash" means to write permanently onto the board | 10:02 |
| xfxf | ack | 10:02 |
| xfxf | how does that tool know which fx2 firmware to flash? the existing usb id's? | 10:02 |
| xfxf | it appears to flash its own seperate to the one i built | 10:02 |
| xfxf | i'm also going to make clean the built misoc stuff & build again but ensure that env variable is 100% set | 10:04 |
| mithro | xfxf: Yes | 10:04 |
| xfxf | BOARD=opsis make all | 10:06 |
| xfxf | that should work, yes? | 10:06 |
| mithro | no | 10:06 |
| xfxf | so i should do what? | 10:07 |
| mithro | xfxf: you need to "export BOARD=opsis" *before* you source setup-env.sh | 10:07 |
| xfxf | ah | 10:07 |
| mithro | Well, actually you don't if you are careful, but at the moment we want to be exact and that is the most exact way | 10:07 |
| xfxf | okay, sure | 10:07 |
| xfxf | new shell, export BOARD=opsis, source'd setup-env.sh, then did make clean && make all | 10:08 |
| xfxf | re the hdmi2usb-mode-switch tool, i assume if i want to force it to use a fx2 firmware i've built i just feed it as the appropriate parameter? | 10:08 |
| mithro | xfxf: While it is doing that check what the contents of your "firmware/fx2/date.inc" | 10:08 |
| mithro | xfxf: Yes | 10:09 |
| xfxf | https://www.irccloud.com/pastebin/qKw5W311/ | 10:09 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 10:09 |
| mithro | xfxf: Great, that is what you want | 10:09 |
| xfxf | okay, will let it build and then will try again | 10:09 |
| xfxf | so to recap, the steps after i have a firmware built, and a brand new reset opsis, is the following: | 10:10 |
| xfxf | hdmi2usb-mode-switch --mode=jtag (if not already in jtag mode), make load-gateware, make load-fx2, make load-lm32, make connect-lm32? | 10:11 |
| xfxf | with a possible hdmi2usb-mode-switch --mode=serial in there if loading the gateware doesn't switch to that mode (but it should by itself IIRC?) | 10:12 |
| xfxf | i seem to keep doing it in the order where i have serial but no UVC, but i think the above is right | 10:12 |
| mithro | You don't need the "make load-lm32" | 10:14 |
| mithro | no | 10:14 |
| mithro | If the "make load-gateware" fails, then do a "hdmi2usb-mode-switch --mode=serial; sleep 5; hdmi2usb-mode-switch --mode=jtag" and try the "make load-gateware" again | 10:15 |
| xfxf | ok | 10:15 |
| mithro | So | 10:15 |
| mithro | Actually - on the Opsis you shouldn't need the "make load-fx2" either | 10:16 |
| mithro | After the "make load-gateware" the FPGA should reset the FX2 and make it read the FX2 firmware out of the FPGA | 10:17 |
| xfxf | that step has never gotten me to a working console (iirc it tries to connect and fails) or with a /dev/videoX device appearing, hence why i do the others | 10:17 |
| mithro | "make flash-gateware" will burn stuff into the FPGA flash permanently (it'll survive a power cycle) | 10:17 |
| xfxf | ack, i'll do that after i've confirmed a firmware we want to use | 10:17 |
| xfxf | so from your POV what is required after 'make load-gateware' to get a fully working device? or that's it? | 10:19 |
| mithro | xfxf: On an Opsis - nothing | 10:19 |
| mithro | xfxf: On an Atlys - "make load-fx2" | 10:19 |
| mithro | xfxf: the "make connect-lm32" is just so you can fiddle with settings and stuff | 10:20 |
| xfxf | right. i haven't had that happen yet on the opsis. i'll paste you the output after this builds + i run it if it does the same thing | 10:20 |
| mithro | xfxf: It'll take about 10 seconds for the Opsis to come up after doing a "make load-gateware" | 10:24 |
| xfxf | okay, maybe i'm not being patient enough. will try | 10:24 |
| mithro | xfxf: Also, you probably want to make sure you don't have ixo-usb-jtag installed anymore | 10:26 |
| mithro | xfxf: As that will install udev rules which makes your opsis go into jtag mode straight away rather then working normally | 10:27 |
| xfxf | https://www.irccloud.com/pastebin/twSxncnE/ | 10:28 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 10:28 |
| xfxf | mithro: see above, load-gateware definitely doesn't give me serial or video on its own | 10:28 |
| mithro | xfxf: see my comment about ixo-usb-jtag above - also, I need dmesg before and after make load-gateware - I have no idea when those messages came in compared to the commands you run | 10:29 |
| xfxf | ah, right, i installed the udev rules that came with the hdmi2usb-mode-switch tool. what should i be using? | 10:29 |
| mithro | xfxf: "sudo apt-get remove ixo-usb-jtag; sudo dpkg --purge ixo-usb-jtag" | 10:30 |
| mithro | https://www.irccloud.com/pastebin/3m69h4SC/ | 10:30 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 10:30 |
| mithro | Those two lines are the first stage of the FX2 on an Opsis renumerating after the FPGA told it to reboot | 10:31 |
| xfxf | okay. yes that package was installed | 10:32 |
| mithro | A couple of sends later the 2a19 5440 device should disappear and a 2a19 5442 | 10:32 |
| mithro | xfxf: It's needed for the Atlys, but we should disable it for the Opsis now the mode-switch tool works okay | 10:33 |
| mithro | actually we could probably remove it now that mode-switch tool works okay.... | 10:33 |
| mithro | but you know, not enough time | 10:33 |
| mithro | I'm going to go get some dinner | 10:33 |
| xfxf | ooohhh right | 10:33 |
| mithro | be back at a computer in 30 minutes but should be on my phone for questions I can answer that way | 10:34 |
| xfxf | okay, now the board doesn't boot up in jtag mode automatically now | 10:35 |
| xfxf | right, this confused the crap out of me, is that package installed during the mode-switch-tool stuff? | 10:35 |
| xfxf | because i installed ubuntu on a brand new laptop and then just the mode-switch tool, and every board i plugged in was booting up in jtag by default | 10:36 |
| xfxf | and i never recalled the opsis doing that by default so i wasn't sure what was happening | 10:36 |
| xfxf | ah, yes, wonderful - now setting board to jtag, make load-gateware, then make connect-lm32 works out of the box | 10:37 |
| mithro | I'm assuming you didn't set "export BOARD=opsis" before get-env-root.sh and hence it assumed you wanted it | 10:39 |
| mithro | CarlFK and tumbleweed probably haven't seen it becuase they have mostly been using the pre-built stuff | 10:40 |
| mithro | Btw you can flash/load an opsis/atlys *just* using the hdmi2usb-mode-switch tool | 10:41 |
| mithro | Oh, CarlFK mainly uses Atlys boards too... | 10:41 |
| xfxf | ack at both of above | 10:43 |
| xfxf | okay, the issue i'm having now is i'm getting nothing from /dev/video1 | 10:43 |
| xfxf | encoder: 1280x720 @ 0fps (0Mbps) from input0 (q: 85) | 10:43 |
| xfxf | encoder is on | 10:43 |
| mithro | There is a bug that you get nothing on the video output if you have the serial console connected | 10:44 |
| mithro | But only the first time | 10:44 |
| xfxf | ah, maybe that's it, i had it connected, but i did try connecting multiple times | 10:44 |
| xfxf | unplugged usb, put back in again, now getting video | 10:44 |
| mithro | That is not what I expected you to do | 10:45 |
| mithro | You should just close the "make connect-lm32" and then try | 10:45 |
| mithro | Once video is going you can connect again | 10:46 |
| *** ssk1328 has joined #timvideos | 10:47 | |
| xfxf | okay. i'm capturing video now so i'm going to leave it for a bit and see what happens | 10:47 |
| xfxf | the 'crash' i seemed to trigger before by trying to connect to the serial port while i was trying to view video | 10:47 |
| xfxf | i'll not do that, but then try that in a hour or two and see what happens | 10:47 |
| xfxf | also may have been coincidence | 10:47 |
| xfxf | either way, thankyou, i think i'm far enough now that i can properly start testing | 10:47 |
| xfxf | bah, only if i knew that package was installed / did that | 10:48 |
| xfxf | that was 90% of what was confusing me | 10:48 |
| xfxf | now the board is behaving as i expect | 10:48 |
| mithro | back now | 11:04 |
| xfxf | okay, capturing fine still, opsis hasn't locked up | 11:40 |
| xfxf | connected/disconnected several times to serial port, is fine | 11:40 |
| xfxf | is it possible that package being installed was causing problems? | 11:41 |
| xfxf | a redmere cable connected to another HDMI cable (in this case, the little HDMI box for my Vive) causes a ton of noise, i assume as expected due to longer cable length? | 11:50 |
| xfxf | actually, ick, i get exactly the same lines of noise from my XA20 using a 0.6m redmere cable :/ | 11:51 |
| xfxf | i'll record a vid and upload | 11:58 |
| xfxf | okay, mithro, this is concerning | 12:09 |
| xfxf | that noise i mentioned | 12:09 |
| xfxf | is not there at all | 12:09 |
| xfxf | until i connect something to the encoder/video device | 12:09 |
| xfxf | and then it appears | 12:09 |
| xfxf | and then it goes away when i disconnect it again | 12:09 |
| xfxf | (comparing to output0, that is) | 12:09 |
| mithro | xfxf: I don't understand what you are explaining - please tell me exactly what is connected and the settings you are using and what "noise" means - preferrable with screenshots | 12:11 |
| xfxf | mithro: http://catstick.com/hdmi2usb-noise.mkv | 12:19 |
| xfxf | https://www.irccloud.com/pastebin/eLG9GlqX/ | 12:20 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 12:20 |
| xfxf | this noise only starts occurring as soon as something is connected to the encoder. | 12:20 |
| xfxf | it's almost like the opsis is running out of grunt | 12:20 |
| xfxf | i took a video from my phone, uploading to YT now, will be up soon | 12:20 |
| mithro | xfxf: Does that appear on the video output too? | 12:20 |
| mithro | xfxf: Or just the encoder output? | 12:21 |
| xfxf | it shows on both output and the encoder | 12:21 |
| xfxf | video i'm uploading now will demonstrate | 12:21 |
| mithro | xfxf: okay | 12:22 |
| mithro | xfxf: If you turn off one output, does the problem go away? | 12:23 |
| xfxf | expect dodgy mobile phone quality but it should be sufficient to demonstrate | 12:23 |
| xfxf | will test that | 12:23 |
| xfxf | i recall something like this with the atlys but thought the opsis was more powerful | 12:23 |
| xfxf | https://youtu.be/2rLiqDXRVII | 12:23 |
| mithro | xfxf: The Opsis is - and its complicated | 12:24 |
| mithro | xfxf: I'm pretty sure those are underflows caused by too much memory contention | 12:25 |
| mithro | xfxf: So.... | 12:26 |
| xfxf | it's still there with output1 turned off, but only about half as bad | 12:27 |
| mithro | xfxf: And if you turn off an input? | 12:27 |
| xfxf | less again but still there | 12:28 |
| xfxf | biggest difference is when i don't have anything connected to encoder or turn it off | 12:28 |
| mithro | xfxf: What is status showing? | 12:28 |
| xfxf | https://www.irccloud.com/pastebin/zM9HMtJY/ | 12:28 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 12:28 |
| mithro | xfxf: Can you restart with just one input, one output and the encoder? | 12:29 |
| xfxf | sure, by restart, you mean power cycle the device? | 12:29 |
| mithro | Yes | 12:30 |
| xfxf | btw this is with two super short redmere cables | 12:30 |
| xfxf | they're like 0.4m or something, shortest i could buy | 12:30 |
| xfxf | yup, still there | 12:34 |
| xfxf | although with just input0/output0/encoder and everything else off it's only slight | 12:35 |
| xfxf | but still definitely noticable, just not the snowfest in that video | 12:35 |
| mithro | xfxf: Are you sure you aren't seeing two different things? | 12:35 |
| xfxf | definitely not, they're the exact same black lines | 12:35 |
| xfxf | if i disconnect the encoder those lines immediately dissapear | 12:35 |
| mithro | xfxf: Okay | 12:36 |
| mithro | xfxf: Well if you have a reproducible test case you can go try and figure out when it started | 12:36 |
| xfxf | okay, any firmware revisions that would be likely to step back to? | 12:36 |
| mithro | xfxf: Do a bisect | 12:37 |
| xfxf | sure, but it's not like i know the internals of this thing well :P | 12:37 |
| xfxf | i'd only be guessing based off variable names / code comments / commit comments | 12:37 |
| mithro | xfxf: A bisect means you choose the oldest version and see if it works -- If it does then you know it was introduced somewhere between those two | 12:38 |
| mithro | xfxf: Then you pick the middle version | 12:38 |
| mithro | xfxf: If the problem is there, then it happened before this version, if the problem isn't, then it started after this version | 12:38 |
| xfxf | ack, i get how it's used, it's more what versions i should strategically pick given rebuilding/testing isn't rapid | 12:39 |
| xfxf | but i'll just step back in ~1 month increments i think | 12:39 |
| mithro | xfxf: you can use the prebuilt firmware | 12:39 |
| xfxf | oh, right, of course | 12:39 |
| mithro | xfxf: Doing a bisect is quicker then stepping randomly, it removes half the possibilities every test | 12:39 |
| mithro | xfxf: assuming that the bug is something which always occurs / always doesn't occur | 12:42 |
| xfxf | well, it's definitely reproducible right now | 12:46 |
| xfxf | i'm going to try this on the firmware we used at LCA | 12:46 |
| xfxf | i think the board i'm using has that flashed on it already | 12:46 |
| xfxf | yup, it's on there too. | 12:50 |
| xfxf | we just didn't notice because it's only slight when only input0/output0/encoder, and the encoder on this firmware seems to be set to 75, not 85 | 12:51 |
| xfxf | hmm - and according to my gstreamer pipeline which shows fps, with the encoder set to 25, i was only getting ~22fps from input0 until i unplugged input1 | 12:52 |
| xfxf | https://www.irccloud.com/pastebin/f9EYhzBm/ | 12:53 |
| tpb | Title: Pastebin | IRCCloud (at www.irccloud.com) | 12:53 |
| xfxf | so this looks like the bug has always been there :/ | 12:53 |
| xfxf | then it went back to ~25fps | 12:53 |
| xfxf | er | 12:54 |
| xfxf | irccloud is being screwy, refreshing page | 12:54 |
| xfxf | okay, 'then it went back to ~25fps` should appear below my 'hmm - and' line. i don't know why that got re-ordered. | 12:54 |
| *** xfxf2 has joined #timvideos | 12:55 | |
| xfxf2 | mithro: sorry, using a local client, irccloud seems broken | 12:55 |
| xfxf2 | did you get the above or was it confusing (it seems to be re-ordering what i'm writing) | 12:55 |
| xfxf2 | this seems like a genuine bug that's been there for at least 6-7 months | 12:56 |
| xfxf2 | i vaguely recall you pulling back the opsis clock speed or something at the LCA hackfest? | 12:57 |
| mithro | We were not getting video artifacts on the Opsis at LCA | 12:58 |
| xfxf2 | i actually suspect we were, they're just subtle | 13:01 |
| xfxf2 | let me go look at a video | 13:01 |
| xfxf2 | herm, what room did we use the opsis in again? | 13:01 |
| mithro | xfxf2: There are white underflow errors in the encoding of the video | 13:02 |
| mithro | xfxf2: but we never got them in the video output | 13:02 |
| xfxf2 | okay, because i'm staring at them on my monitor right now | 13:05 |
| xfxf2 | it may be that they're not as noticable on a projector? | 13:05 |
| xfxf2 | that video i uploaded, it's perhaps only 20% as bad as it is in that | 13:05 |
| xfxf2 | but still there | 13:05 |
| xfxf2 | realistically it sounds like i have to put up with this, although it's suboptimal, and i was intending on using two input's on one of them but i'll can that | 13:06 |
| xfxf2 | how can we fix this though, it seems like a genuine bug? | 13:07 |
| xfxf2 | is there anything i can test? again, vaguely recall you changing timing or something during the hackfest | 13:07 |
| mithro | xfxf2: you can try increasing the fifo depth https://github.com/timvideos/HDMI2USB-misoc-firmware/blob/master/targets/opsis_video.py#L27 | 13:09 |
| tpb | Title: HDMI2USB-misoc-firmware/opsis_video.py at master · timvideos/HDMI2USB-misoc-firmware · GitHub (at github.com) | 13:09 |
| xfxf2 | also, should 'flash-gateware' work? because i get a 'Unknown flash device (ID 0x00818181)' | 13:10 |
| mithro | xfxf2: That is the one you need to mode into serial mode then back to jtag mode | 13:10 |
| xfxf2 | okay | 13:10 |
| xfxf2 | also, fifo depth, try 1024 for both inputs? | 13:10 |
| mithro | xfxf2: Also check if you can see them on the Atlys would be a useful test too | 13:11 |
| xfxf2 | my atlys is mostly garbage, won't attempt | 13:11 |
| xfxf2 | the HDMI ports got damaged during LCA | 13:13 |
| xfxf2 | will see if i can get them repaired at some point | 13:13 |
| *** Bertl is now known as Bertl_oO | 13:21 | |
| *** xfxf2 has quit IRC | 13:32 | |
| xfxf | mithro: nope, with fifo increased to 1024, same issue | 13:32 |
| *** xfxf2 has joined #timvideos | 13:32 | |
| xfxf | also i'm having serious issues reliably getting UVC video | 13:32 |
| xfxf | until i unplug and plug back in the USB | 13:33 |
| xfxf | or repeatedly do connect-lm32 and ctrl-c and re-do it again | 13:33 |
| xfxf | and yes, more than once, usually 4-5 times | 13:33 |
| xfxf | unplugging USB and plugging back in works first time tho | 13:33 |
| mithro | xfxf: My guess is that you have modem manager opening the device | 13:41 |
| xfxf2 | this is UVC, not the serial port | 13:42 |
| xfxf2 | serial connects reliably | 13:42 |
| mithro | xfxf: The video won't start when anything is connected to the serial | 13:42 |
| xfxf2 | yes, this is with me ctrl-c'ing the flterm process that opens | 13:43 |
| xfxf2 | unless that still keeps open a handle or something | 13:43 |
| mithro | xfxf2: Why is the flterm process opening at all? | 13:43 |
| *** danielki has joined #timvideos | 13:52 | |
| *** danielki has quit IRC | 13:57 | |
| tumbleweed | mithro: do we know how to flash atlys boards? | 15:02 |
| *** xfxf2 has quit IRC | 15:02 | |
| mithro | tumbleweed: Kinda | 15:03 |
| mithro | tumbleweed: we can flash the FPGA | 15:03 |
| tumbleweed | that'll do | 15:03 |
| mithro | tumbleweed: but flashing the FX2 is a bit dangerous | 15:03 |
| tumbleweed | how do I do that? | 15:03 |
| tumbleweed | the FPGA bit | 15:03 |
| mithro | tumbleweed: make flash-gateware or use the hdmi2usb-mode-switch --flash-gateware | 15:04 |
| tumbleweed | that sounds easy | 15:04 |
| * tumbleweed tries | 15:04 | |
| tumbleweed | (carlfk's odroids don't have a working openocd, so pre-flashed atlys boards will make our lives easier) | 15:04 |
| *** ssk1328 has quit IRC | 15:05 | |
| tumbleweed | mithro: bin or bit? | 15:07 |
| mithro | tumbleweed: flash .bin, load .bit | 15:07 |
| tumbleweed | OK | 15:08 |
| mithro | tumbleweed: The hdmi2usb-mode-switch tool should stop you doing the wrong thing now | 15:08 |
| mithro | tumbleweed: Have you seen the things that xfxf reported? | 15:10 |
| xfxf | CarlFK: i assume you're using your ingest script for capture? | 15:11 |
| tumbleweed | mithro: what things? | 15:12 |
| xfxf | because i recall you talking about a 1 second delay the other day, i think that's in your gstreamer pipeline if you're muxing alsa + hdmi2usb, i'm seeing the same here | 15:12 |
| tumbleweed | not those reliability problems, no | 15:12 |
| xfxf | CarlFK: i also remember thaytan strongly recommending to use pulsesrc, as alsa is just talking to pulse pretending to be alsa | 15:14 |
| xfxf | er, s/as alsa is/as using alsasrc on ubuntu is/ | 15:14 |
| mithro | well bed time for me | 15:15 |
| xfxf | tumbleweed: are you using atlys's or opsis's though? i assume atlys's | 15:15 |
| xfxf | mithro: night, thanks for your help. i'm leaving a machine capturing from input0 overnight | 15:15 |
| xfxf | mithro: i'd like to solve the noise issue if possible before pycon, but with just input0/output0/encoder enabled it's not overly bad. i strongly suspect that was actually present at LCA, we just didn't look hard enough | 15:16 |
| mithro | Your not going to solve it with 2 weeks out | 15:18 |
| mithro | Unless you have fpga programmers hidden somewhere | 15:19 |
| mithro | It's probably solved with _florent_ new dram controller | 15:20 |
| mithro | But that has a week's more work needed before it's even ready for testing | 15:21 |
| mithro | You should give the heartbeat a go | 15:25 |
| mithro | Anyway, I'm home now, so actually going to sleep | 15:26 |
| xfxf | if we lived with it without noticing for LCA, we can live with it for now | 15:29 |
| xfxf | i guess my only concern is seeing the framerate drop when enabling both inputs - that perhaps might explain some of the sync issues we saw | 15:29 |
| xfxf | i imagine i can work around that with the appropriate gstreamer pipeline, though | 15:29 |
| xfxf | maybe dropped frames from the hdmi2usb was causing the alsa/pulse muxing to get all whacky | 15:30 |
| tumbleweed | xfxf: atlys | 15:32 |
| xfxf | tumbleweed: ah, figures | 15:32 |
| tumbleweed | and yes CarlFK reported something about 1s lag | 15:32 |
| xfxf | is this carlfk's kit or debconf bought a bunch? | 15:32 |
| xfxf | CarlFK: ping me when you're around, i might refactor the ingest script a bit for my use | 15:33 |
| *** rohitksingh has joined #timvideos | 17:32 | |
| tumbleweed | mithro: so, one of the atlys boards won't flash | 17:49 |
| tumbleweed | it has Info : JTAG tap: xc6s.tap tap/device found: 0x34008093 (mfg: 0x049 (Xilinx), part: 0x4008, ver: 0x3) | 17:49 |
| tumbleweed | where a successful one has Info : JTAG tap: xc6s.tap tap/device found: 0x44008093 (mfg: 0x049 (Xilinx), part: 0x4008, ver: 0x4) | 17:50 |
| tumbleweed | the next line on the failed one is Error: Unknown flash device (ID 0x00ffffff) | 17:50 |
| tumbleweed | (when you wake up...) | 17:50 |
| *** rohitksingh has quit IRC | 18:09 | |
| *** froztbyte has joined #timvideos | 19:00 | |
| froztbyte | hai | 19:03 |
| *** nueces has quit IRC | 19:41 | |
| *** CarlFK has quit IRC | 23:10 | |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!