Friday, 2019-05-10

*** tpb has joined #tomu00:00
xobsWeird.  I plugged Fomu into my PC's keyboard, and according to the Beagle sometimes the PC isn't "ACK"ing debug packets.01:12
xobsWhich, granted, it's totally fine to do that.  But it's not something I'd seen before.  Or handled.01:13
*** emeb has quit IRC01:20
xobsOkay, /that's/ interesting.  If the value "0xe0007537" is READ from the device, Windows doesn't ever send an ACK.  That value can be written, and many other values can be written too...01:46
*** futarisIRCcloud has quit IRC02:03
*** futarisIRCcloud has joined #tomu02:10
*** xkapastel has quit IRC03:16
xobsIt appears to be a problem where USB isn't properly stuffing the last bit, as it should.  Well, I'm glad I caught this now!03:17
xobsAnd not in two weeks when we go to production.03:18
*** futarisIRCcloud has quit IRC04:13
*** awe00 has quit IRC04:41
*** rohitksingh_work has joined #tomu05:17
tntxobs: lol. Look what I fixed in my core a couple weeks ago : https://github.com/smunaut/ice40-playground/commit/a8ca29653ce8e373360269a74e4fe73c3bb9325905:26
tpbTitle: cores/usb: Fix bitstuffing in TX path for border line case · smunaut/ice40-playground@a8ca296 · GitHub (at github.com)05:26
mithrotnt: Did I ever get you some (more?) Fomu hacker boards?05:28
mithrotnt: FYI - I'll be back in Europe for WOSH (https://fossi-foundation.org/wosh/) in June...05:29
tpbTitle: WOSH: Week of Open Source Hardware (at fossi-foundation.org)05:29
xobstnt: it's one of those weird edgecases that I always forget about.05:30
xobsIt's becoming more of an issue now that I'm sending random bytes across the wire.05:30
tntmithro: I don't think so (at least I never received any).  But yeah, if you can take a few with you and post them then, that'd be nice :)05:30
xobsWhich, incidentally, appears to be a good way to shake out bugs.05:30
mithroI still have plenty05:30
tntxobs: yeah, I only found out about the issue when I started to send random 4 Mits isochronous traffic and notice sometime I would miss some packets ...05:31
mithroanyway, heading home now - have a gnight!05:31
tntmithro: gn !05:31
xobsmithro: night!05:31
xobs...and after spending all day on it, it turns out the problem was probably caused by the timeout feature being weird.09:14
*** AmosSam has left #tomu09:15
xobsI removed the timeout logic and now it works well.09:16
*** AmosSam has joined #tomu09:24
*** AmosSam has left #tomu09:30
*** AmosSam has joined #tomu09:30
*** futarisIRCcloud has joined #tomu09:34
*** AmosSam has left #tomu09:37
*** AmosSam has joined #tomu09:41
tntxobs: timeout ?10:05
xobstnt: I put a timeout on the debug bus, because otherwise if you try to access a part of the Wishbone that is unmapped, nothing will ACK it and you'll wait forever.10:07
xobsSomehow that timeout was doing the wrong thing.10:07
*** johnhmay has quit IRC10:57
*** johnhmay has joined #tomu10:59
*** alexhw has quit IRC10:59
*** alexhw has joined #tomu11:01
*** spacekookie has quit IRC11:04
*** spacekookie has joined #tomu11:05
*** xkapastel has joined #tomu11:20
*** futarisIRCcloud has quit IRC11:43
_florent__xobs: just curious, it is the litex wishbone timeout or something else?12:23
xobs_florent__: it was mostly a copy of the timeout in the UART bridge.12:24
*** ToxicFrog has joined #tomu13:01
ToxicFrogI'm having a weird issue with the quickstart repo. The prebuilt DFUs all work fine, but when I build the same ones from source, they don't -- main() runs fine but the sys_tick_handler() is never called.13:05
xobsToxicFrog: are you on Ubuntu 18.04 by any chance?13:05
ToxicFrogThe only change I've made to the source or the build process is using arm-none-eabi-gcc instead of arm-none-eabi-g++ to link, since I don't have the latter, but it's all written in plain C anyways, so that shouldn't be an issue...and if it were I'd expect an actual linker error.13:06
ToxicFrogxobs: OpenSUSE.13:06
ToxicFrogWhy, is this a known issue on Ubuntu 18.04? Perhaps it also affects SUSE.13:06
ToxicFrog(this is not a showstopper for me, since everything I want to do, at least for now, I can do with loops and cycle counting in main(), but I would like to get to the bottom of it)13:07
xobsThe issue on Ubuntu was that it wasn't linking the correct library.13:08
xobsE. G. https://bugs.launchpad.net/ubuntu/+source/newlib/+bug/176812513:08
tpbTitle: Bug #1768125 “libnewlib-arm-none-eabi generates wrong code for c...” : Bugs : newlib package : Ubuntu (at bugs.launchpad.net)13:08
*** im-tomu has quit IRC13:12
*** im-tomu has joined #tomu13:12
*** awe00 has joined #tomu13:26
ToxicFrogAah. I'll look into that on SUSE, thanks.13:30
*** rohitksingh_work has quit IRC13:32
*** AmosSam has left #tomu13:46
*** AmosSam has joined #tomu13:46
*** rohitksingh has joined #tomu14:34
*** emeb has joined #tomu15:18
*** emeb has quit IRC15:43
xobsalexhw: looks like the theory I had on the github issue isn't quite correct.  Still getting missing bytes.16:15
xobsOr extra bytes.16:15
xobsOr both.16:15
xobsI've been staring at this for too long now.  I'm going to give up for the night.16:16
*** xkapastel has quit IRC18:22
*** rohitksingh has quit IRC19:17
*** rohitksingh has joined #tomu19:21
*** emeb has joined #tomu19:23
*** rohitksingh has quit IRC19:46
alexhw[m]xobs: I'll look into it a some more this night or tomorrow21:05
*** futarisIRCcloud has joined #tomu21:55
futarisIRCcloudhttps://twitter.com/_conorpp/status/112666892599812505621:59

Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!