| *** tpb has joined #tomu | 00:00 | |
| xobs | People are getting them, yay! I didn't know they were shipping them already. I just know I sent a big box to the US. | 00:34 |
|---|---|---|
| xobs | As many of you in this channel know, I'm working on packaging up all the required tools. | 01:08 |
| xobs | The synthesizer/hardware tools are at https://github.com/xobs/toolchain-icestorm/releases/latest and the pnr is at https://github.com/xobs/toolchain-nextpnr-ice40/releases | 01:09 |
| tpb | Title: Release v1.15-fomu: travis: only build when tags: true · xobs/toolchain-icestorm · GitHub (at github.com) | 01:09 |
| xobs | Those are just intermediate repos -- I'll have a third repo in im-fomu that packages those up, plus a risc-v toolchain and (on Windows) Python too. | 01:10 |
| pollo | xobs: I'm in Canada , so it even crossed the border | 01:47 |
| *** xkapastel has quit IRC | 03:21 | |
| futarisIRCcloud | “USBSamurai — A Remotely Controlled Malicious USB HID Injecting Cable for less than 10$” by Luca Bongiorni https://link.medium.com/DqVbFNo5bZ | 07:38 |
| MadHacker | Whee, so my first fomu order arrived. | 08:21 |
| MadHacker | Oddly came via mouser, which is interesting. :) | 08:21 |
| xobs | MadHacker: whee! | 08:21 |
| *** Kitlith_ has quit IRC | 08:21 | |
| MadHacker | Toys. :) | 08:21 |
| MadHacker | OK, so plugging that in locked up my USB controller entirely. | 08:25 |
| MadHacker | (which was a bit awkward because my keyboard is on that same controller!) | 08:25 |
| xobs | MadHacker: :( | 08:25 |
| MadHacker | Also, crowdsupply have put their stickers on top of the label with the "find out more at"... on the bag. | 08:26 |
| MadHacker | 'tis a shame because the fomu sticker is much prettier than the CS-FOMU-01 QTY: 1 sticker. :) | 08:26 |
| MadHacker | Right, onto getting this working. So, it shows up as "Fomu DFU Bootloader v 1.8.7" OK, but windows is giving it "This device cannot start (Code 10)". | 08:28 |
| MadHacker | About to have a rummage through the various quick start bits and see if I can get it going sans jig. | 08:29 |
| MadHacker | Nope. Unfortunately it completely locks up the bus on whatever I plug it into most of the time. This is on a generic HP desktop. I have two fomus here; both do the same. :( | 08:37 |
| MadHacker | Occasionally it does indentify. | 08:37 |
| xobs | MadHacker: Great :( | 08:38 |
| MadHacker | identify even. | 08:38 |
| MadHacker | Sorry xobs. | 08:38 |
| xobs | This is on a Mac? | 08:38 |
| MadHacker | This is on a generic HP desktop. | 08:38 |
| MadHacker | My work windows machine. | 08:38 |
| MadHacker | Intel everything, no hubs involved. | 08:39 |
| xobs | So I've seen at least one case where a board only shows up when toggling the C_RESET pin once after everything is powered up. | 08:39 |
| xobs | That will show up about 5% of the time when rapidly inserting and removing it. | 08:39 |
| MadHacker | I'll give them a go on my mac, and see if I can find someone else's machine to try as well, just try and scope out the problem. | 08:40 |
| MadHacker | We can hope it's only slightly older machines like this that care. (This is pre-USB3 and being plugged into USB2 ports; i5-3470 machine) | 08:40 |
| MadHacker | OK. It's fine on my macbook. | 08:43 |
| xobs | That's promising. | 08:44 |
| MadHacker | Yeah, absolutely fine on this. Perfectly reliable. | 08:44 |
| MadHacker | Works OK on the mac on a hub, too. Two of them even. | 08:46 |
| xobs | Thanks for checking. I wonder what the problem could be. Maybe the response timing is off? | 08:46 |
| MadHacker | OK. Experiment #2, plug in fomu, plug in mouse to 2nd port on same bus on my desktop - the mouse gets "the last device plugged in has not been recognised" but the fomu itself didn't present anything at all. | 08:47 |
| MadHacker | Works fine as soon as you pull fomu out. | 08:48 |
| xobs | That probably means Fomu is getting confused seeing EOP and is stuck transmitting. | 08:48 |
| MadHacker | That's my general impression, yep. It's basically drowning out anything else on the bus. | 08:49 |
| MadHacker | Don't have a hub handy for this PC, but I'll see if I can find one and give that a try. Might fix the issue somewhat given the little bit of isolation the hub provides. | 08:50 |
| xobs | That's a very good point. | 08:52 |
| xobs | I should find a USB 1.1 hub without TTs. | 08:52 |
| xobs | That could actually be the issue. | 08:52 |
| xobs | Bugger. I think I know the issue. | 08:54 |
| xobs | I think it's ignoring the "Address" field. Which isn't an issue if your hub has a TT. | 08:54 |
| MadHacker | This is going straight into a non-hubbed port. | 08:58 |
| MadHacker | (Well, root hub or whatever) | 08:58 |
| xobs | No onboard hub? | 09:02 |
| MadHacker | No; it's one of those very cheap machines where the only USB ports are straight from the chipset. | 09:04 |
| MadHacker | Or at least I think it is... | 09:06 |
| MadHacker | OK, life is strange. No. There's an integrated hub providing an SD card reader slot. | 09:07 |
| MadHacker | xobs: FWIW if it's the only device on the bus it enumerates fine. | 09:14 |
| MadHacker | So at least in terms of telling people how to get started, "try it with nothing else plugged in to that hub" is not usually particularly burdensome. | 09:15 |
| MadHacker | And dfu-util is happy like that, too. | 09:17 |
| tnt | btw, this behavior of other device not recognized when I had an experimental core plugged in was when I was not properly filtering packets based on assigned address. (i.e. not checking ADDR field is what I got in SET_ADDRESS). | 10:12 |
| tnt | (oh I should have continued reading :p but yeah, that's exactly the same conclusion you came to) | 10:14 |
| xobs | tnt: Nice to have it corroborated. | 10:15 |
| xobs | Fortunately I have a software updater created that should work. | 10:16 |
| xobs | I wish I'd caught the issue sooner, though. | 10:16 |
| xobs | I found a Mac! And I've determined that Apple isn't very good with forwards-compatibility. It's running OS X 10.11, and nothing I'm building actually runs on it. | 10:30 |
| MadHacker | I'm pretty sure XCode has various options to change what platform release you're targetting. | 10:47 |
| MadHacker | There's presumably a matching option for the CLI. | 10:47 |
| MadHacker | -mmacosx-version-min=<version> | 10:48 |
| MadHacker | Or just export MACOSX_DEPLOYMENT_TARGET=10.6 or whatever. | 10:48 |
| xobs | MadHacker: oh! That's good to know! I wonder if I should do that for my Rust project. | 10:52 |
| *** rohitksingh has joined #tomu | 12:38 | |
| xobs | MadHacker: Could I get you to download and test https://github.com/xobs/wishbone-utils/releases/tag/v0.2.8-test3 to see if you can get it to talk to Fomu? | 13:02 |
| tpb | Title: Release v0.2.8-test3: travis: rust looks like it needs 10.7 at a minimum · xobs/wishbone-utils · GitHub (at github.com) | 13:02 |
| xobs | It should be linked without libusb, but I had trouble running it on the 10.11 machine I was using. | 13:02 |
| *** rohitksingh has quit IRC | 13:34 | |
| *** xkapastel has joined #tomu | 13:40 | |
| *** rohitksingh has joined #tomu | 13:48 | |
| *** rohitksingh has quit IRC | 14:05 | |
| MadHacker | xobs: Will do. | 14:07 |
| *** emeb has joined #tomu | 14:10 | |
| MadHacker | It broadly seems to work. | 14:12 |
| MadHacker | I mean, I can make it crash by using --list, but it's somewhat together and doing its thing. | 14:12 |
| xobs | MadHacker: crash? how so? | 14:13 |
| MadHacker | https://paste.debian.net/1095977/ | 14:13 |
| tpb | Title: debian Pastezone (at paste.debian.net) | 14:13 |
| MadHacker | Stray tar command at the bottom, sorry about that, mispaste. Nonetheless the rest is right. | 14:13 |
| xobs | Thanks, that's promising! | 14:23 |
| xobs | Now I know I can build that, I can work on dfu-util and then nextpnr-ice40. | 14:24 |
| xobs | Python will be the challenge to statically-link | 14:24 |
| MadHacker | There's various build-an-exe tools for python. | 14:24 |
| MadHacker | It's not actually that bad usually. | 14:24 |
| MadHacker | pyinstaller comes to mind. | 14:25 |
| MadHacker | cxfreeze is the other similar thing. | 14:26 |
| xobs | MadHacker: those will provide a libpython3.7.a? | 14:28 |
| MadHacker | Believe they will take care of it "somehow", where somehow is platform-dependent. | 14:31 |
| xobs | The thing I'm trying to do is embed it into nextpnr-ice40. The version I'm shipping assumes it's installed in /usr/local/opt/Python/ which is where homebrew puts it. | 14:34 |
| ovf | interestingly they make a special provision for windows: https://docs.python.org/3.7/using/windows.html#windows-embeddable | 14:36 |
| tpb | Title: 3. Using Python on Windows Python 3.7.4 documentation (at docs.python.org) | 14:36 |
| MadHacker | -DBUILD_PYTHON=OFF ? | 14:44 |
| xobs | We need Python in order to support multiple clock domains. | 14:44 |
| MadHacker | Or are you using nextpnr's python bits specifically? | 14:44 |
| MadHacker | OK. | 14:44 |
| MadHacker | OK. I managed to get the examples stuff for foboot working on windows no fuss. lxbuildenv can die in a fire at this point, though. Depends on git silently (just gives a file not found error) is not cool, but it then sulks if your copy of foboot isn't an actual git checkout etc. | 17:23 |
| MadHacker | Was pip so *very* difficult to deal with? | 17:23 |
| *** xkapastel has quit IRC | 17:25 | |
| * MadHacker grumbles. | 17:30 | |
| MadHacker | I have to say that the usb-cdcacm demo worked well, and ditto the blinky, tho. | 17:30 |
| ovf | xobs: i think you can pretty easily have python local to your application. i've tested with neither macos or nextpnr (sorry), but it should be similar to my test here: http://vpaste.net/hYbI6?raw . does this help at all? it defaults to trying to load python stuff from next to the binary | 17:48 |
| *** xkapastel has joined #tomu | 18:51 | |
| *** NoGodDamnIdea has joined #tomu | 20:16 | |
| *** xkapastel has quit IRC | 21:52 | |
| *** emeb has quit IRC | 23:38 | |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!