Monday, 2019-08-19

*** tpb has joined #tomu00:00
*** awe00 has quit IRC01:43
*** whatnick has joined #tomu03:15
*** whatnick has quit IRC04:10
*** whatnick has joined #tomu04:11
*** rohitksingh has quit IRC04:20
*** rohitksingh has joined #tomu04:29
*** whatnick_ has joined #tomu04:47
*** whatnick has quit IRC04:49
*** whatnick_ has quit IRC04:56
*** whatnick has joined #tomu04:57
xobsWoo, managed to build the on the in-office Mac!05:03
xobsI had to add libffi from conda, so naturally that means doing another release.  But it's in a good position now, much better than it was last week.05:04
*** rohitksingh has quit IRC05:10
mithroxobs: I feel like you'd have been more productive to get our conda stuff working on Mac / Win06:41
xobsmithro: probably! I'll have to investigate how conda works at some point I'm sure.07:15
mithroxobs: You can even build a single conda installer07:15
mithrowith everything in the same binary07:15
xobsmithro: that sounds complicated and lovely. I still need to figure out how to bundle something like that for nextpnr!07:18
MadHackerxobs: re: git dependency; sorry for being properly grouchy about it. I tend to work from a "download a zip" not a git clone for stuff from github. As a result I had both a dependency I didn't expect on git, *and* it wasn't salvagable because I wasn't working from a git checkout, and therefore the submodules wouldn't work either.08:23
*** futarisIRCcloud has quit IRC08:23
MadHackerI could deal with the git dependency but I'd have preferred the lxbuildenv stuff just explicitly did its own git clone, rather than trying to just use the recursive submodules thing when it isn't necessarily starting out from a git-y environment.08:24
MadHackerAnyway, sorry, I was probably needlessly grumpy about it.08:24
xobsMadHacker: it's a good datapoint. I appreciate it.08:36
xobsI know a lot about being grouchy at unexpected software behavior.08:36
*** awe00 has joined #tomu09:59
*** awe00 has quit IRC10:10
*** awe00 has joined #tomu10:29
MadHackerxobs: So, working from a clean checkout of foboot, on OS X, I get: ERROR: Top-level input 'I0' also driven by I0.O.10:31
MadHackerwhen I try to build for --revision pvt10:31
MadHackerIs this likely to be an odd versioning things on yosys or something on my machine?10:32
MadHackerActually, yes, it will be, since I see it's just doing SB_LUT4 at this point.10:32
xobsMadHacker: I'm not really sure.  It's a distinct possibility.10:33
MadHackerYep, think it's my own fault on this one. Yosys versioning.10:36
*** cangaroo_ has joined #tomu11:17
*** awe00 has quit IRC11:49
*** mifune has joined #tomu11:57
*** awe00 has joined #tomu12:05
*** whatnick1 has joined #tomu12:31
*** jas4711 has quit IRC12:45
mithroxobs: FYI -
tpbTitle: LiteX BuildEnv Support - Google Sheets (at
mithroxobs: Where are you working on the workshop stuff?12:50
xobsmithro: It's at specifically ``12:51
tpbTitle: GitHub - im-tomu/fomu-workshop: Support files for participating in a Fomu workshop (at
mithroxobs: For  all commands you should show example output when successful12:52
mithroShall I log some github issues?12:52
xobsmithro: I generally do.  What's missing?12:52
xobsSure, that's a good idea.12:53
tpbTitle: GitHub - im-tomu/fomu-workshop: Support files for participating in a Fomu workshop (at
xobsAh, that README.  Okay.12:53
mithroJust getting to the airport now13:06
mithrowill be back in a bit13:06
*** awe00 has quit IRC13:09
whatnick1Hey xobs do you a tutorial for VexRiscv on Fomu ?13:21
xobswhatnick1: what are you looking to do?13:21
whatnick1Just VexRiscv + Micropython ...13:22
whatnick1Or help out some more on Fomu ..13:22
whatnick1I tried it on the TinyFPGA BX , did not take, trying to see how the Fomu is different13:23
xobsI don't, unfortunately.  It's early days yet.13:26
xobsBut I'm more than happy to help people get set up!13:27
xobsI haven't even moved micropython into the im-fomu repository yet.  It's currently at
tpbTitle: GitHub - xobs/micropython at fomu (at
*** whatnick1 has left #tomu13:28
whatnickhey just closed a client13:29
xobsAh, okay!13:29
xobsThe basic build process is to go into and run "make".  Then load `bin/micropython.bin` onto Fomu.  NOT `bin/micropython.dfu`.13:29
tpbTitle: micropython/ports/fomu at fomu · xobs/micropython · GitHub (at
whatnicklet me do some work then will have fun with the fomu at home.13:30
whatnickmake will grab the toolchain ?13:30
xobsYou can also run `wishbone-tool -s gdb`, then in another window run `riscv64-unknown-elf-gdb -ex "target remote localhost:1234" build/firmware.elf` and debug it over gdb.13:31
whatnickxobs we have met in a pub in Sydney ... and ICC...13:31
whatnickokay .. will keep that in the logs.. looking forward to tutorial material.13:31
xobswhatnick: oh. Hello!13:31
whatnickin Kenya now :)13:32
whatnickspreading fomus around like viruses13:32
xobsWelcome home13:32
*** cangaroo_ has quit IRC13:38
*** awe00 has joined #tomu13:45
MadHackerGoodness but nextpnr is slow on my macbook on the normal foboot hw.14:03
MadHackerI'll have to remember to run that on something beefier in future.14:03
xobsIt does take a while.  It /is/ about 99% full.14:04
xobsDid you run it with `--placer heap`?14:04
MadHackerI'm just running your build script at this point (
MadHackerSeems to have completed at least.14:05
xobsTry adding `--placer heap`.  I don't have that set as default, since the HeAP placer is relatively new.  But now that we ship nextpnr, it's pretty likely to be enabled.14:06
*** awe00 has quit IRC14:06
MadHackerWill do.14:06
*** emeb has joined #tomu14:27
*** whatnick has quit IRC14:47
mithroxobs: We should get zephyr working into the workshop somehow...14:49
xobsmithro: good suggestion.14:54
xobsThough I did just try printing the workshop document, and it's 27 pages.14:54
mithroxobs: Poke the antmicro people about it?14:54
mithroxobs: I suggest you just print a bunch of business cards with links to the webpage?14:54
*** whatnick has joined #tomu14:54
ovfthere's probably not enough space for usb mass storage in foboot? it'd be nice for fomu to be self-documenting15:03
xobsovf: not in foboot, but there's a provision for transitioning to a secondary bootloader.15:04
xobsThe problem right now is I've been needing to get this toolchain stuff done, and I need to chase the factory to figure out why they're behind.15:04
xobsAnd I'm in the middle of a refactor of the USB stack to enable more than just the three endpoints, which will be crucial to getting circuitpython support.15:05
xobsBut there's a plan for it!15:06
*** tmbinc has joined #tomu15:08
ovfcool. :-) the nice thing is that with fomu you can always ask people later to update their hardware.15:08
tmbincJust received my crowdsupply fomus. It fails to enumerate on windows, is that a known bug? Works in Linux and I have OpenVizsla logs. Happy to dive into it, but if it's a known fixed issue, I can save the time :)15:09
tmbincI think it fails in 80 06 00 01 00 00 40 00 (i.e. bmRequestType=D2H|Standard|Device, bRequest=06 READ DESCRIPTOR, wValue=0001, wIndex=0000)15:12
xobstmbinc: there's a known issue where it doesn't properly respect the USB address value.  Can you try it with a hub to make sure it's behind a transaction translator?15:14
*** whatnick has quit IRC15:15
tmbincit's on a USB3 hub15:15
*** whatnick has joined #tomu15:15
tmbincI think it's still on address 0?15:16
tmbincah, hm15:17
tmbincmeh if openvizsla would have a proper decoder this would be more easy15:17
tnttmbinc: ViewSB15:17
tpbTitle: GitHub - usb-tools/ViewSB: open-source USB analyzer toolkit with support for a variety of capture hardware (at
*** xkapastel has joined #tomu15:27
xobsmithro: Hmm... You're much better at Twitter than I am.  Twitter doesn't seem very reliable about sending me communication.15:29
xobs"If you enable notifications for Retweets, replies or mentions, likes, or follows, you may not receive a notification for every Retweet, reply or mention, like, or follow. "15:29
xobstmbinc: it's not a known issue, but based on what you're seeing it's trying to query the configuration descriptor but not getting a response.15:30
mithroxobs: I'm using tweetdeck and have columns for important keywords15:30
tmbinctnt: oh, that's nicer than expected. Triggers a table_name=None bug somehow  but i can see the request, trying to figure out if my pcap is wrong or what ViewSB displays is wrong though15:30
xobsThe only known issue is that it ignores the "SET_ADDRESS" command, and therefore will respond when other devices on the bus respond.15:30
xobsmithro: but do you have to constantly check it, or does it send you email notifications?15:31
*** awe00 has joined #tomu15:31
mithroxobs: I check it when procrastinating doing more important things :-P15:31
tmbincxobs:it may be related actually15:31
tmbinci think ViewSB incorrectly shows the checksum as the device address?15:33
tmbincoh i think what _actually_ happens is that there's a port reset (for whatever reason) but the FOMU doesn't reset to address 015:34
tmbincso it's actually trying to obtain the device descriptor with a "2D 00 10" (=SETUP address=0, endpoint=0), but there's no response15:35
xobsFomu currently ignores the "ADDR" field of the token.15:35
tmbincok, fair point, as i do see a NAK15:36
xobsSo for whatever reason, Fomu isn't able to understand the packet.15:36
tmbincbut why would it NAK the device descriptor request?15:36
xobsYou see a NAK?  That's interesting.15:36
xobsIs the LED fading in and out?15:36
tmbincanyway there seem to be two issues: a.) the thing that causes a reset, and b.) that the re-enumeration doesn't work15:37
tmbincxobs: fyi with a nightly wireshark. Lots of crap before the actual USB traffic, please ignore.15:37
tmbinc(did I mention OpenVizlsa needs work? ;)15:37
tmbincso around packet 2200 there's the first SETUP asking for the device descriptor, which is NAK'ed a few times but then works15:38
xobstmbinc: If the RGB LED is fading in and out, that means the crystal is working and the CPU is functioning, which is a good sign.15:38
tmbincit is blinking15:38
tmbincand i can see the initial descriptors going by15:39
*** awe00 has quit IRC15:40
xobsMy copy of wireshark refuses to open that file :(15:40
tmbinc3.1.1rc0 works here15:41
xobsBut can you actually see responses coming back from the device?15:41
tmbincI can see the string descriptor, "Fomu DFU Bootlaoder v1.8.7" etc.15:41
tmbincthere's something weird with the addressig though15:41
tmbincit sends the initial device descriptor request to address 015:42
tmbincthen sends the next request already to a non-zero address15:42
tmbincshouldn't the PC first need to set the address and move it away from 0?15:42
xobsHmm... Sounds like I really, really need to get that working.15:42
tmbinchm i think ViewSB just fails decoding this correctly unfortunately15:43
xobsI'm sorry it's not working, tmbinc .  I'm going to cccamp tomorrow, so I need to prepare for that.15:44
tmbincyeah the second SETUP is a set_address15:44
tmbincxobs:oh, no worry.15:44
tmbincxobs: I'll be there as well so worst case we can debug it there15:44
xobsOh yay.  I'll bring the beagle (busted though it may be).  And if I have time on the plane maybe I'll work on it.15:45
xobsI need to head to bed now.  Goodnight!15:45
*** awe00 has joined #tomu16:41
*** awe00 has quit IRC16:52
*** whatnick_ has joined #tomu16:52
*** whatnick has quit IRC16:54
*** whatnick_ has quit IRC16:57
*** whatnick has joined #tomu16:58
*** NoGodDamnIdea has joined #tomu17:06
*** awe00 has joined #tomu17:15
*** awe00 has quit IRC17:55
*** awe00 has joined #tomu19:02
shaprrandom feedback: the fomu case lets the USB pins side of the PCB easily flip up, sometimes it gets stuck when I try to insert it into a USB port. I haven't had that problem with any of ~25 tomu19:29
shapris that a known problem? maybe I just got a weird one?19:29
shaprI just assembled the second one I ordered, it's perfectly fine, as good as the tomus I've gotten. I think I just have a weird one.19:31
shaprThe tomu getting started repo is wonderful, I gave a class and everybody loved it. Is there any of that for the fomu?19:32
shaprshould I start with the foboot or fomu-toolchain?19:33
*** rohitksingh has joined #tomu19:59
*** NoGodDamnIdea has quit IRC20:09
*** rohitksingh has quit IRC20:20
*** rohitksingh has joined #tomu20:22
*** rohitksingh has quit IRC20:33

Generated by 2.13.1 by Marius Gedminas - find it at!