| *** tpb has joined #timvideos | 00:00 | |
| mithro | _florent_: You wouldn't happen to still be up? | 00:01 |
|---|---|---|
| tumbleweed | mithro: how do you use modeswitch in development? it seems all the entry points are created by setup.py | 00:21 |
| tumbleweed | I ended up doing | 00:21 |
| tumbleweed | for cmd in hdmi2usb-find-board hdmi2usb-mode-switch hdmi2usb-manage-firmware opsis-find-board opsis-mode-switch opsis-manage-firmware atlys-find-board atlys-mode-switch atlys-manage-firmware; do echo -e '#!/usr/bin/python3\nimport pathlib, sys\nsys.path.insert(0, str(pathlib.Path(__file__).parent.parent.resolve()))\nimport hdmi2usb.modeswitch.cli\nhdmi2usb.modeswitch.cli.main()' > $cmd; chmod +x $cmd; done | 00:21 |
| tumbleweed | which is rather hacky | 00:21 |
| tumbleweed | err that probably clipped | 00:21 |
| mithro | It didn't I think | 00:21 |
| tumbleweed | https://paste.debian.net/894479/ | 00:22 |
| tpb | Title: debian Pastezone (at paste.debian.net) | 00:22 |
| tumbleweed | ah | 00:22 |
| mithro | tumbleweed: for development I run "make setup" | 00:22 |
| tumbleweed | I avoided that makefile, too much conda | 00:23 |
| mithro | tumbleweed: Well, you can just do a "python3 setup.py develop" or "python3 setup.py install" | 00:23 |
| tumbleweed | anyway, as you can see I just switched from that opsis-prod branch to master | 00:24 |
| mithro | you should be using master from https://github.com/timvideos/HDMI2USB-mode-switch | 00:24 |
| tpb | Title: GitHub - timvideos/HDMI2USB-mode-switch: Tool for switching boards supported by HDMI2USB firmware between multiple different modes (programming, webcam, etc). (at github.com) | 00:24 |
| tumbleweed | yes, exactly | 00:24 |
| mithro | Great! | 00:24 |
| tumbleweed | sure, but usually it's useful to be able to run code from the checkout | 00:24 |
| mithro | tumbleweed: Hrm? | 00:25 |
| tumbleweed | I mean, if you're hacking on it | 00:25 |
| mithro | tumbleweed: This is pretty much how any normal python module works | 00:25 |
| tumbleweed | usually there are entry points in the module that you can run, without any setup | 00:25 |
| tumbleweed | in this case the entry point isn't executable, because it's depending on setuptools wrapper entryopints, and sys.argv[0] | 00:26 |
| mithro | tumbleweed: python setup.py will generate those wrappers | 00:26 |
| tumbleweed | yes | 00:26 |
| tumbleweed | but usually you don't have to touch setup.py when hacking | 00:26 |
| mithro | tumbleweed: So you could use virtualenv if you don't like conda | 00:27 |
| tumbleweed | no ve necessary | 00:27 |
| tumbleweed | it has no crazy dependencies | 00:27 |
| tumbleweed | anyway, you see my query about ixo-usb-jtag.hex ? | 00:27 |
| mithro | tumbleweed: Yes, and I need a bunch more information | 00:27 |
| mithro | tumbleweed: When ixo-usb-jtag package is installed, things won't work properly | 00:27 |
| tumbleweed | so | 00:28 |
| tumbleweed | firstly, the 0.0.2 firmwares don't seem to work | 00:29 |
| tumbleweed | and neither do recent 0.0.0 ones, on this board | 00:29 |
| mithro | tumbleweed: what is the 0.0.2 firmwares? | 00:29 |
| tumbleweed | e.g. v0.0.2-81-gd0d3aea | 00:29 |
| tumbleweed | the most recent build | 00:29 |
| mithro | tumbleweed: How are you loading them? | 00:29 |
| tumbleweed | with modeswitch | 00:29 |
| tumbleweed | so, after loading one of those, I had a "bricked" board | 00:30 |
| tumbleweed | because the fx2 wouldn't come up | 00:30 |
| tumbleweed | until I jumpered it to failsafe | 00:30 |
| tumbleweed | but then modeswitch would loop when trying to load the jtag firmware | 00:30 |
| mithro | tumbleweed: There is a lot of things going on here - please can you give me a set of instructions to reproduce the problem and I'll give it a go | 00:30 |
| tumbleweed | not right now, becaus I'm in bed, and the opsis is on a desk at ARM | 00:31 |
| tumbleweed | tomorrow morning I can | 00:31 |
| mithro | tumbleweed: Okay, that would be helpful | 00:31 |
| tumbleweed | but basically: flash the latest firmware | 00:31 |
| tumbleweed | it'll be bricked | 00:31 |
| tumbleweed | then try to unbrick it | 00:31 |
| mithro | tumbleweed: I also need to know what you have installed | 00:31 |
| mithro | tumbleweed: and how you have the jumpers configured | 00:32 |
| tumbleweed | as in? | 00:32 |
| tumbleweed | default | 00:32 |
| mithro | tumbleweed: linux, any udev rules, ixo-usb-jtag, openocd, any other packages which interact with them | 00:32 |
| tumbleweed | surely if it flashed, nothing else matters that much? | 00:32 |
| mithro | tumbleweed: You aren't having problems with flashing, you are having problems with it coming up after flashing, right? | 00:33 |
| tumbleweed | all my ixo-usb-jtag udev rules are commented out, except for the permissions one | 00:33 |
| tumbleweed | correct | 00:33 |
| mithro | tumbleweed: I've been successfully using the firmware on this production opsis board here | 00:34 |
| tumbleweed | and then, once it won't come up, re-flashing it with something else | 00:34 |
| tumbleweed | the latest firmware? | 00:34 |
| tumbleweed | this board is the one that wouldn't run every firmware, we can try a wider range tomorrow | 00:34 |
| tumbleweed | wider range of boards | 00:34 |
| mithro | tumbleweed: If you have a set of steps which works on one board and not another, that is useful info | 00:35 |
| tumbleweed | yeah, I only had one board today | 00:35 |
| tumbleweed | hopefully rattus will bring another one tomorrow | 00:35 |
| mithro | tumbleweed: but a set of steps which I can try and reproduce exactly here would be useful | 00:35 |
| mithro | tumbleweed: but generally, you either use ixo-usb-jtag package *or* HDMI2USB-mode-switch | 00:36 |
| tumbleweed | modeswitch --mode jtag on a board in falesafe mode (P18 disconnected, and K3 on 1-2) | 00:36 |
| tumbleweed | that's where I ran into that problem | 00:36 |
| tumbleweed | it did nothing | 00:37 |
| tumbleweed | surely it's not either/or if you hobble the udev rules | 00:37 |
| tumbleweed | because that packade doesn't do anything without them | 00:37 |
| tumbleweed | it's just some udev rules, and some firmware blobs | 00:37 |
| mithro | tumbleweed: Assuming you reloaded the udev rules and a bunch of other things which might be going wrong | 00:38 |
| mithro | tumbleweed: If you add a -v you can see what the mode-switch is doing / waiting for | 00:38 |
| tumbleweed | I had -v, obviously | 00:38 |
| mithro | tumbleweed: Then the log would be useful | 00:38 |
| mithro | tumbleweed: Your fxload version information would be useful too | 00:39 |
| tumbleweed | it won't be *that* useful, because it hangs | 00:39 |
| tumbleweed | but yeah, we'll get theret morrow | 00:39 |
| mithro | tumbleweed: where does it hang? | 00:39 |
| tumbleweed | and we don't get the fxload verbose output, unless you run it by hand | 00:39 |
| tumbleweed | after fxload | 00:39 |
| tumbleweed | and the krnel does a lot of this: | 00:40 |
| tumbleweed | [1662542.512044] usb 1-2: new high-speed USB device number 99 using xhci_hcd | 00:40 |
| tumbleweed | [1662547.672073] usb 1-2: device descriptor read/64, error -110 | 00:40 |
| tumbleweed | [1662563.280364] usb 1-2: device descriptor read/64, error -110 | 00:40 |
| tumbleweed | you can see, by running fxload by hand, that it is loading the firmware, and rebooting the fx2 | 00:40 |
| mithro | tumbleweed: So I'm guessing it is hanging when doing a lsusb | 00:40 |
| tumbleweed | but into some broken state | 00:40 |
| tumbleweed | changing the firmware for the (suprisingly different) noe that ixo-usb-jtag had, fixed that issue for me | 00:41 |
| tumbleweed | my fxload is 0.0.20081013-1ubuntu2 | 00:41 |
| mithro | tumbleweed: I'm afraid there is too much unknown here for me to figure out the problem | 00:42 |
| tumbleweed | sure | 00:42 |
| tumbleweed | and I spent a few hours being confused by a USB problem that was on the host-side (but caused by doing things on the opsis), so I hope that didn't lead me into a weird debugging corner | 00:43 |
| tumbleweed | but I think that was resolved before most of this | 00:44 |
| * tumbleweed -> bed | 00:45 | |
| mithro | tumbleweed: Hopefully we can put more logic into modeswitch to help figure out problems like this | 00:46 |
| *** rohitksingh_work has joined #timvideos | 04:03 | |
| *** Bertl is now known as Bertl_zZ | 06:39 | |
| *** rohitksingh_wor1 has joined #timvideos | 08:05 | |
| *** rohitksingh_work has quit IRC | 08:06 | |
| *** rohitksingh_work has joined #timvideos | 08:14 | |
| *** rohitksingh_wor1 has quit IRC | 08:16 | |
| tumbleweed | mithro: right, I'm in front of an opsis | 10:31 |
| tumbleweed | are you still around? | 10:31 |
| tumbleweed | mithro: ok, it seems I need a different fx2 firmware, depending on whether I'm re-flashing a happy opsis or a failsafe opsis | 10:39 |
| tumbleweed | mithro: so, when the board is in failsafe, this happens if I try to use modeswitch to flash | 11:35 |
| tumbleweed | https://paste.debian.net/894614/ | 11:35 |
| tpb | Title: debian Pastezone (at paste.debian.net) | 11:35 |
| tumbleweed | latest firmware that works is v0.0.0-654-gbca243b | 11:50 |
| tumbleweed | except the encoder is fucked | 11:50 |
| *** Bertl_zZ is now known as Bertl | 12:19 | |
| *** rohitksingh_work has quit IRC | 13:06 | |
| tumbleweed | the last firmware version that works (with encoder) is v0.0.0-598-g07f8476 | 13:06 |
| tumbleweed | latest firmware (v0.0.2-81-gd0d3aea) works on another (less faulty) opsis | 13:22 |
| tumbleweed | but streaming is borked | 13:22 |
| tumbleweed | also | 13:22 |
| tumbleweed | input0: 0x0 (@ 42773 kHz) | 13:23 |
| tumbleweed | input1: 1280x720 (@ 74249 kHz) | 13:23 |
| tumbleweed | output0: 1280x720@60Hz from pattern | 13:23 |
| tumbleweed | output1: 1280x720@60Hz from input1 | 13:23 |
| tumbleweed | encoder: 1280x720 @ 0fps (0Mbps) from input1 (q: 50) | 13:23 |
| tumbleweed | ddr: read: 1760Mbps write: 884Mbps all: 2644Mbps | 13:23 |
| tumbleweed | all the kHz | 13:23 |
| tumbleweed | and v0.0.2-72-g70cb993 has working encoder | 13:40 |
| tumbleweed | or did, for a bit... | 13:52 |
| * tumbleweed works his way back again | 13:53 | |
| *** rohitksingh has joined #timvideos | 14:00 | |
| tumbleweed | o_O no, that was my laptop | 14:02 |
| *** puck`` has quit IRC | 15:31 | |
| *** puck`` has joined #timvideos | 15:34 | |
| *** rohitksingh has quit IRC | 16:16 | |
| *** rohitksingh has joined #timvideos | 16:18 | |
| *** Bertl is now known as Bertl_zZ | 18:57 | |
| *** rohitksingh has quit IRC | 21:02 | |
| *** Bertl_zZ is now known as Bertl | 23:31 | |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!