Saturday, 2019-08-31

*** tpb has joined #tomu00:00
*** emeb has quit IRC00:18
*** rohitksingh has quit IRC02:59
*** rohitksingh has joined #tomu03:46
*** ademski has joined #tomu04:29
*** ademski has quit IRC05:29
*** cedric has quit IRC06:04
*** cedric has joined #tomu06:08
*** cedric has joined #tomu06:08
*** rohitksingh has quit IRC08:25
*** spacekookie has quit IRC09:04
*** spacekookie has joined #tomu09:04
xobstnt: gtkwave doesn't have the best documentation, or the best API, but I finally got it to auto-add multiple sigrok lines on init:
tpbTitle: spibone/gtkwave.init at master · xobs/spibone · GitHub (at
tntxobs: Oh, that's useful, thanks ! I had no idea that was doable ... or called Insert_Comment :)11:00
xobstnt: you can do Insert_Blank too (basically you can do any menu function), and it looks like you can specify arguments if it pops up a dialog box.  Not documented as far as I can see.11:01
xobsI do Insert_Comment and then use Hilight_Regexp to only pick out the signals I want.11:01
xobsAlso, it seems as though you can add the blank lines above the target. Which was a key piece of information I discovered.11:01
xobsAlso, Sigrok doesn't decode Wishbone. Which kinda makes sense since it's internal-only.11:04
MadHackerArgh, my CDC config descriptor is > wMaxPacketSize. Suddenly everything is hard. :/11:20
xobsMadHacker: :(11:21
MadHackerJust because there's no sequencing in dummyusb, so I can't easily send multiple response packets.11:22
MadHackerGetting there otherwise.11:22
xobsI wonder if it's possible to trim it down to fit in 64 bytes.11:26
xobs63 bytes, rather.11:27
MadHackerYeah, it's at 67, so it's frustratingly close.11:27
MadHackerLooking about though, that seems to be the normal length for a CDC one.11:27
tntYou can probably cheat ...11:27
tntsince a lot of device have wrong cdc descriptors with non union descriptors and everything in a single interface, most OS have quirks to deal with those ...11:28
MadHackerThat'd be small enough.11:28
MadHackerOr maybe send some of the descriptors undersized that end in all zeroes? Might get away with that?11:28
MadHackerSo, OS X seems to allow it if you just miss out the union.11:31
MadHackerGood enough for me for now. :D11:32
xobsI wonder if Windows will allow it.  It's pretty strict.11:43
MadHackerI'll try it shortly. I want to get some data flowing first; if the only problem is the descriptor stuff then I can go back to it and work out some way to trigger the 2nd packet of the descriptor send.11:44
MadHackerI reworked dummyusb's GetDescriptor stuff a lot, so there's no len(blah)+len(blah)+... etc stuff going on.11:45
MadHackerBit tidier, but I want to get some basic data in and out of the CDC data endpoints before I'd share it.11:45
MadHackerThen again, - works for sending descriptors more prettily, doesn't do anything useful yet. Will do that next.11:59
tpbTitle: 6502-on-fomu/ at master · jamesacraig/6502-on-fomu · GitHub (at
tntMadHacker: how big is dummyusb + cdc btw ?12:00
MadHackerDunno, I haven't built the CDC bits yet. :D12:00
xobstnt: I know dummyusb by itself is ~680 LCs.12:00
xobsWell, just something that enumerates and does nothing interesting.12:00
MadHackerInfo:          ICESTORM_LC:  217112:01
MadHacker 5280    41%12:01
MadHackerThat's my 6502 + dummyusb stuff together tho.12:01
MadHackerSo not very helpful for you. :(12:01
MadHackerStill, that's enough to blink a LED from software, and to enumerate as a USB device.12:01
MadHackerSo, y'know, getting there. Next step, make the LED colour reflect data from USB.12:02
xobsMadHacker: If it's helpful at all, you can use the wishbone-over-SPI module I just created to debug it.12:02
xobsAssuming peeking and poking memory is useful to you.12:02
MadHackerFor the 6502 end? The USB side is OK, since I can use wireshark to see what it's sending/receiving reliably enough. The 6502's got a wishbone bridge but doesn't use wishbone for its own RAM/ROM.12:04
MadHackerSo I could talk to it that way, but not debug with it.12:04
MadHacker(The 6502's only got 64KB of address space, and litex seemed to want lots for CSRs; also, the cycles really don't fit well with the 6502's "the bus is always busy" approach)12:05
xobsWell, you could always expose various signals onto Wishbone and poke at them taht way.12:05
MadHackerI did put together a bridge, so in theory I could dump a little shared memory between the two and have your wishbone-over-SPI thing read/write from it.12:05
xobsMadHacker: I do like how you've reorganized the dummyusb stuff.12:30
MadHackerThanks. Still working away on that. I want to get it neater again, but it's a start. Turning it into a declaration at the top and then a Case generator seems much easier to maintain, if nothing else.12:54
*** im-tomu has quit IRC12:59
*** im-tomu has joined #tomu13:11
MadHackerOK, I'm restarting (somewhat), albeit with a lot of cut and paste. epfifo is probably closer to what I want - minus the wishbone bits - but a pseudo-FIFO that implements the setup packet stuff cut-and-paste from dummyusb seems a better way to go.14:12
MadHackerThat way I get FIFOs already handled for the CDC bits, and I get the descriptors and setup stuff managed in gateware.14:12
xobsMadHacker: I'm in the middle of a rewrite (hence the SPI bridge) that has three FIFOs at
tpbTitle: GitHub - xobs/valentyusb at tri-fifo (at
xobsOne for the SETUP endpoint, one for IN, and one for OUT.  I still don't have it working, but I'm at the point where I'd like to be able to debug it on-device.14:16
MadHackerHm. OK. I'm liking epfifo's initialiser side for the "just tell me what EPs you want" list approach. I just want a setup endpoint to be able to live in the gateware.14:18
MadHackerThe assumption of wishbone is a bit of an obstacle for me, but I feel that epfifo's maybe a cleaner model than dummyusb to extend.14:19
*** NoGodDamnIdea has joined #tomu15:52
xobsMadHacker: You sure you don't want to get dummyusb working?16:28
xobsFor input, you ought to be able to just add into a FIFO if usbcore.endp == 2 and == PID.IN and fifo.empty == 1.16:30
*** rohitksingh has joined #tomu16:39
*** emeb has joined #tomu17:07
*** rohitksingh has quit IRC18:19
*** rohitksingh has joined #tomu18:21
*** rohitksingh has quit IRC19:00
*** rohitksingh has joined #tomu19:00
*** ademski has joined #tomu20:02
* Xark finds that FOMU development has some unique challenges. This is the first time he has had to crawl around on the floor under his desk to find an FPGA dev board. :D20:06
XarkIt is cool that the whole FOMU seems toolchain works under ChromeOS Linux prompt - including device programming with dfu-util (however still need to use Chrome browser for serial port terminal access).20:09
Xarker, whole FOMU toolchain seems to work...20:09
*** emeb has quit IRC20:30
*** emeb has joined #tomu20:33
*** js`xmpp has joined #tomu20:59
*** NoGodDamnIdea has quit IRC21:16
*** rohitksingh has quit IRC21:24
*** xkapastel has joined #tomu21:29
*** rohitksingh has joined #tomu22:10
*** emeb has quit IRC22:46
*** rohitksingh has quit IRC23:40

Generated by 2.13.1 by Marius Gedminas - find it at!