pepijndevos changed the topic of #apicula to: Project Apicula: bitstream documentation and tooling for Gowin FPGAs https://github.com/YosysHQ/apicula -- logs https://freenode.irclog.whitequark.org/apicula
_whitelogger has joined #apicula
FabM has joined #apicula
<pepijndevos> oh no...
<pepijndevos> updating the unpacker
<pepijndevos> unpacking vendor bitstream the tap is on R7C14_SPINE27 but the pin drives R7C14_SPINE24
<daveshah> Huh
<daveshah> Permutation or irregularities somewhere maybe?
<pepijndevos> Dunno... got some debugging to do it seems...
<pepijndevos> During lunch I think I realised the problem... AND I MANAGED TO FINISH MY LUNCH. Such self control. Please clap.
<Lofty> Congrats pepijndevos
<omnitechnomancer> good work
<pepijndevos> yay, it unpacks
<pepijndevos> welp, I kinda forgot the format for the nextpnr ROUTING attribute
<pepijndevos> My regex is kinda broken now that I added a bunch of new things
<pepijndevos> seems dest;pip;1;
<pepijndevos> working???!!!!
<pepijndevos> Tang Nano is running a globally clocked blinky
<daveshah> neat!
<pepijndevos> trabucayre, if I have two boards plugged in, how do I tell openFPGALoader which one to use?
<pepijndevos> ah -d /dev/ttyUSB2
<trabucayre> yep
<trabucayre> or --ftdi-serial
<pepijndevos> fffffff
<pepijndevos> k
<pepijndevos> bitstream doesn't work on TEC0117
<trabucayre> snifff
<trabucayre> pastbin ?
<pepijndevos> of what?
<pepijndevos> I don't think it's an openFPGALoader problem this time
<trabucayre> you saif doesn't work -> my first idea is "openFPGALoader is again broken" :)
<pepijndevos> just my own fuckery
<trabucayre> I'm happy because it's not my fault ;-)
<trabucayre> pff Need to work with ISE ;-(
<pepijndevos> trabucayre, maybe you DO know the solution to the problem though. I recall you looked a bit into the checksum stuff in the header.
<pepijndevos> In a 1.9.6 vendor generated bitstream it starts with 1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111
<pepijndevos> while my bitstream starts with 1011011010100001
<pepijndevos> that's.... not even the same length...
<pepijndevos> the document daveshah wrote also mentions 20 0xff bytes, which... I never saw?? https://github.com/YosysHQ/apicula/blob/master/doc/commandstructure.md#preamble
<trabucayre> yep
<trabucayre> my note said a data checksum ...
<trabucayre> I build a design to see this
<trabucayre> ok with 1.9.6 all my scripts are broken :-/
<pepijndevos> I'm still using 1.9.3 for fuzzing, but the 1.9.3 programmer is even more broken, so I can only use openFPGALoader and the 1.9.6 programmer
<trabucayre> good mix :)
<pepijndevos> and... I can still load a vendor bitstream generated with 1.9.3 sooo...
<pepijndevos> aha, but what does not work is loading a 1.9.3 bitstream with a 1.9.6 programmers. wtfffff
<trabucayre> it's funny... or not
<pepijndevos> ah no, that's just the security bit
<pepijndevos> alright, false alarm...
<pepijndevos> so it's just my Apicula bitstream that's somehow fucked up
<trabucayre> pfff set_option seems different
<trabucayre> don't want -device or -pn ...
<pepijndevos> oh yea, the tcl is completely different
<pepijndevos> one sec
<trabucayre> ok
<pepijndevos> that's using the new tcl api to build stuff
<trabucayre> thanks
<pepijndevos> specifically _build_script
<trabucayre> ok
<pepijndevos> But yea, updating to 1.9.6 will probably completely break Apicula
<pepijndevos> wait wat... the line before the flash address is just.... not there??
<pepijndevos> no wait, it's not there on the vendor bitstream
<pepijndevos> wtfffff
<trabucayre> secur bit add one line in the header
<trabucayre> have you a run.tcl already filled ?
<trabucayre> run_all is unkown ...
<trabucayre> I'm stupid "run all" not "run_all"
<trabucayre> fuck my device.cfg seems not taken
<pepijndevos> okay so... old code with new DB still works, means fuzzing and chipdb are ok
<pepijndevos> new gowin_pack with old nextpnr json also works, meaning the old parts of the packer are at least fine
<pepijndevos> so either I broke the nextpnr script, or the new bits are wrong...
<pepijndevos> ah, actually in both cases the checksum fails, due to secure bit I assume, so it's not a programmer problem... something is just wrong with my clock routing
<pepijndevos> yea okay, there is just something legit wrong with clock routing on GW1N-9
<trabucayre> secur bit add a new line in header (after line 6)
<trabucayre> (in my note)
<pepijndevos> I bet there is just some enable bit that I'm missing that causes everything to parse just fine, but not actually turn on the clock
<pepijndevos> Like... I can parse a vendor bitstream, I can pack/unpack my own, but my own is just dead.
<pepijndevos> And only on GW1N-9, on GW1N-1 everything is fine.
FabM has quit [Quit: Leaving]