Monday, 2018-11-19

*** tpb has joined #timvideos00:00
tannewtmithro: they can be up to 64 bytes. On the control endpoint the setup packet will include a max length00:05
mithrotannewt: But it seems the first data0 following a setup pid token is always 8 bytes00:06
tannewtYup. The setup packet is always 8 bytes00:06
mithrocoffee time and then I think I can get these data in / out transfers working00:07
tannewtThe nrf has registers for those values because you can receive a setup packet at any time00:07
tannewtNice! Do you have a setup packet interrupt as well?00:08
*** TheAssassin has joined #timvideos00:20
mithrotannewt: I'll need some code to generate the packet data soon00:27
tannewtI may have some time tonight for it. You can see the osal in tiny usb for what needs to be implemented at the lowest level.00:29
*** sc00bz1 has joined #timvideos00:47
*** sc00bz has quit IRC00:50
xobsmithro: yes, U10 is the RGB LED.  And U7 is the crystal I forgot to give them.  And U4 is the footprint that apparently doesn't fit the parts I gave them :-/01:03
*** puck_ has quit IRC01:14
*** futarisIRCcloud has joined #timvideos02:41
futarisIRCcloudmithro: What device(s) were you trying to get tinyusb running on over the weekend?02:46
mithrotannewt / xobs: The crc5 is done over the addr+endp -- so 11 bits?04:08
tannewtmithro, not sure I never have to do that part04:12
futarisIRCcloudaddress & endp.
xobsmithro: I believe so. Though in practice I never had to calculate it in Grainuum.04:22
mithrotannewt: I figure out what that strange request is before the SET_ADDRESS04:30
tpbTitle: Snippet | IRCCloud (at
*** rohitksingh_work has joined #timvideos04:35
xobsYeah, it's completely valid to only request part of a descriptor!04:41
futarisIRCcloudYep, part of the spec.04:44
futarisIRCcloudRelevant - - "please find the people who wrote the USB spec, and fire every single one of them for gross incompetence"04:45
tannewtthat's easy to say after the fact04:45
futarisIRCcloudIt was designed in 1996. Bluetooth specs aren't much better.04:47
*** rohitksingh_wor1 has joined #timvideos04:57
*** rohitksingh_work has quit IRC04:58
xobsThe descriptor system, with various sizes that may (or may not) be accurate, is a large source of host implementation bugs.05:17
*** rohitksingh_work has joined #timvideos05:26
*** rohitksingh_wor1 has quit IRC05:29
xobsWhen compiling nextpnr, what does "Could NOT find Boost" mean?05:58
xobsThis is following "Found the following boost libraries: filesystem, thread, program_options, system, chrono, date_time, atomic".05:58
mithroxobs: If I recall, it is complaining about not finding python-py27....06:17
xobsmithro: I see.  It tries to find Boost one method, and then when it fails it falls back to a different method.  But the error still gets printed out.06:18
*** rohitksingh_wor1 has joined #timvideos06:26
*** rohitksingh_work has quit IRC06:28
*** futarisIRCcloud has quit IRC07:00
CarlFKmithro: why is this booting h2u firmware and not linux?
tpbTitle: Ubuntu Pastebin (at
mithroCarlFK: your script is bad07:37
mithroLook in the middle07:37
CarlFKmithro: i looked.  now what? ;)07:40
mithroYour script is broken, fix your script07:41
CarlFKmithro: I don't know how07:45
mithroCarlFK: go compare your script to the tutorial07:46
mithroCarlFK: Also make your script stop on errors07:46
CarlFKmithro: make flterm not error :p07:46
mithroCarlFK: you have the source code07:48
CarlFKmithro: this just got too hard.07:48
mithroCarlFK: of just do a "flterm || true" ?07:57
CarlFKmithro: i can't  just do that - the script clones your repo, your repo has the flterm commands in arty/ - if you'll do it your repo, then I'll test it and see what happens08:12
mithroIsn't it the last command anyway?08:13
CarlFKmithro: 4 - firmware,linux  x qemu, arty08:15
mithroCarlFK: I have no idea what you are saying08:18
CarlFKmithro: flterm isn't the last command08:18
mithroWell, I'm calling it a night08:32
mithroxobs: You can see the latest at
tpbTitle: GitHub - mithro/valentyusb: USB Full-Speed core written in migen/LiteX (at
mithroxobs: Still probably won't work08:50
mithroxobs: Currently working on this bit ->
tpbTitle: valentyusb/ at master · mithro/valentyusb · GitHub (at
xobsIck.  Turns out the footprint of the crystal is mirrored on the datasheet.  As a result my PCBs have no working 48 MHz crystal.11:00
tpbTitle: Imgur: The magic of the Internet (at
daveshahjoin the club11:02
xobsdaveshah: that... makes me feel a lot better, actually.  Nice job.11:03
daveshahI had made the same mistake I think11:03
daveshahthen I ordered a second batch of boards to be assembled by elecrow11:03
daveshahwith a correct footprint11:03
xobsWell, this is why we make prototypes.11:03
daveshahbut they soldered it the wrong way round so I had to bodge those too!11:03
xobsI also made the SPI flash footprint a bit small, but that can be fixed with not-too-terrible hand-soldering.  The oscillator footprint thing is annoying.11:04
daveshaheveryone gets SPI flash footprints too small lol11:04
xobsIt's an FPGA, so I'm thinking maybe I can get away with using a different input for clocks.  Unfortunately, the pad I picked for that is I2C, which has a pullup on it.  That'll make the clock line not so good.11:05
daveshahI doubt a pullup is too big an issue tbh11:05
xobsWell, I'll try feeding it in.  I need a 48 MHz clock, and happily the Raspberry Pi has a 19.2 MHz clock that's two pins away, which the PLL can eat nicely.11:06
daveshahah, that's nice11:06
xobsAnd this PCB is designed to work with a Raspberry Pi, so it's not a catastrophic failure.  Just a big one.11:06
xobsAt least I got the LED polarity right.  And the voltage regulators seem happy.  Tomorrow I'll see if I can talk to the ICE40.11:07
* felix_ also wonders why some crystal manufacturers have the footprint seen from the bottom and not from the top like everone else does that11:11
*** rohitksingh_wor1 has quit IRC11:54
*** diamondman has quit IRC11:57
*** diamondman has joined #timvideos11:57
*** rohitksingh_work has joined #timvideos11:59
*** rohitksingh_wor1 has joined #timvideos12:03
*** rohitksingh_work has quit IRC12:03
*** rohitksingh_wor1 has quit IRC13:02
*** Kripton has quit IRC13:15
*** Kripton has joined #timvideos13:28
*** rohitksingh has joined #timvideos13:47
*** rohitksingh has quit IRC14:55
*** rohitksingh has joined #timvideos15:27
*** rohitksingh has quit IRC16:50
*** rohitksingh has joined #timvideos17:22
*** rohitksingh has quit IRC17:31
*** rohitksingh has joined #timvideos17:39
mithroxobs: I'm very interested in seeing if I can get the USB stuff working from the internal oscillator18:52
mithroxobs: tinyfpga doesn't think it's stable enough18:52
mithroxobs: He's probably right, but I'm stubborn :-P18:53
mithro_florent_: ping?18:53
tannewt@mithro can you sync to the internal osc?20:34
mithrotannewt: you mean internal oscillator feeding the PLL?20:37
tannewtI think samd syncs the internal osc to the start of frame20:37
tannewtI don't know much about clocks20:38
CarlFKmithro: ./scripts/ ... buildroot login: <- is that flterm or qemu's --nographic console?20:47
CarlFKI don't see flterm runing.  which explains whey I can ^c out of it and flterm doesn't error20:48
*** rohitksingh has quit IRC21:09
mithroxobs: Also see
tpbTitle: litex-buildenv/ at tinyusb-work · mithro/litex-buildenv · GitHub (at
CarlFKmithro: this looks new: or1k-elf-newlib-gcc: error trying to exec 'cc1': execvp: No such file or directory21:24
CarlFKmithro: any guess what that is about, or should I dig?21:24
daveshahCarlFK: sounds like it might be a path issue21:25
daveshahPerhaps the gcc binary is in not in its proper folder so it can't find cc121:26
CarlFKlitex-buildenv/build/conda/bin/or1k-elf-newlib-gcc  litex-buildenv/build/conda/libexec/gcc/or1k-elf/8.2.0/cc121:35
CarlFKno idea what proper is21:35
tannewtmithro, thanks. my bxes should come today22:03
mithrotannewt: Great!22:03
mithroCarlFK: you got the full output?22:04
CarlFKmithro:  (I think thats enough?22:05
tpbTitle: Ubuntu Pastebin (at
*** puck_ has joined #timvideos22:08
mithroCarlFK: dunno, will have to look at it later -- need to concentrate on this USB stuff22:19
CarlFKmithro: no prob - was thinking it might have been something you did in the last few hours22:22
*** tsglove2 has joined #timvideos22:40
*** cr1901_modern1 has joined #timvideos22:42
*** springermac_ has joined #timvideos22:45
*** techman83 has quit IRC22:45
*** cyrozap has quit IRC22:45
*** cr1901_modern has quit IRC22:46
*** tsglove has quit IRC22:46
*** paddatrapper has quit IRC22:46
*** springermac has quit IRC22:46
*** springermac_ is now known as springermac22:46
*** paddatrapper has joined #timvideos22:47
*** cyrozap has joined #timvideos22:47
*** techman83 has joined #timvideos22:52
*** ChanServ sets mode: +v techman8322:52
*** sc00bz1 is now known as Sc00bz23:32
*** cr1901_modern1 has quit IRC23:39
*** cr1901_modern has joined #timvideos23:39

Generated by 2.13.1 by Marius Gedminas - find it at!