<MoeIcenowy>
_florent_: I cannot imagine the adjust is so simple
<MoeIcenowy>
My own lead to get the read latency -= 1 change (because there is RBURST (BURSTDET called by Lattice)), however for the write latency -= 1 I didn't find it (because I have no knowledge about the internal of LiteDRAM)
<MoeIcenowy>
nickoe: not possible w/o proper primitives from Gowin
<MoeIcenowy>
a primitive called GW_JTAG is known to exist in prim_syn.v in Gowin toolchain
<MoeIcenowy>
but not documented at all
<MoeIcenowy>
and GAO is surely encrypted
Degi has quit [Ping timeout: 246 seconds]
Degi has joined #litex
nickoe has quit [Quit: Client closed]
nickoe has joined #litex
nickoe has quit [Quit: Client closed]
nickoe has joined #litex
somlo has quit [Remote host closed the connection]
somlo has joined #litex
nickoe has quit [Quit: Client closed]
nickoe has joined #litex
<sensille>
Hoernchen: the patch is only for the eth_clock. 30mhz is awfully low, where is the critical path?
FabM has joined #litex
FabM has quit [Changing host]
FabM has joined #litex
<Hoernchen>
sensille, https://dpaste.org/BrVRK - apparently to a dpram? this is the unmodified "colorlite"
<sensille>
diamond can get to 50, but fails in the eth path. my patch might help there
<sensille>
but 40 should be enough?
<Hoernchen>
i guess so
<Hoernchen>
i'm just wondering if there ever was a way to hit those default frequencies
<sensille>
no luck with abc9/flow3 either
<Hoernchen>
yeah i'm currently trying some of those magic ravenslofty cookbook incantations
<sensille>
i don't feel well running a design with failed timings either
nickoe has quit [Quit: Client closed]
nickoe has joined #litex
nickoe67 has joined #litex
nickoe has quit [Ping timeout: 246 seconds]
cr1901__ has joined #litex
cr1901_ has quit [Ping timeout: 246 seconds]
nickoe29 has joined #litex
nickoe67 has quit [Ping timeout: 246 seconds]
cr1901_ has joined #litex
cr1901__ has quit [Ping timeout: 244 seconds]
nickoe57 has joined #litex
nickoe29 has quit [Ping timeout: 246 seconds]
nickoe51 has joined #litex
nickoe57 has quit [Ping timeout: 246 seconds]
cr1901__ has joined #litex
cr1901_ has quit [Ping timeout: 245 seconds]
SpaceCoaster has quit [Ping timeout: 246 seconds]
fts-tmassey has joined #litex
<fts-tmassey>
Hello! I'm having trouble with linux-on-litex-vexriscv and a Terasic de0nano. I've downloaded to the board, I get the side-to-side LED's, but I can't communicate with it.
<fts-tmassey>
I think it's because I need to connect to a UART via the GPIO header pins, but I don't know what pins to use! I'm happy to answer any questions I can, of course.
<fts-tmassey>
I'm running on Debian Bookworm, with a Terasic de0-nano. Quartus 22.1free. I'm happy to provide parameters, logs, timings, etc. but you would have to help me to find these... I basically followed the bouncing ball from the readme.
jevinskie[m] has joined #litex
<jevinskie[m]>
<MoeIcenowy> "and GAO is surely encrypted" <- I’d bet we can get around that little road bump if that’s what is necessary for figuring out the primitive/enabling litejtag. ;)
<fts-tmassey>
I have also successfully used Quartus to download a quick procedure from the Terasic manual. Seeing as I see the LED's moving like I've seen in other tutorials, it seems I've successfully downloaded the FPGA code. The problem is, I can't connect to the serial interface to download the Linux kernel, etc.
<josuah>
whitequark[cis]: maybe they had bug when turning encryption off :D
nickoe51 has quit [Quit: Client closed]
nickoe51 has joined #litex
<whitequark[cis]>
no it's just the implementation is like that
<whitequark[cis]>
they're in the gwsyn.so library
<josuah>
one would have to be in a country where reverse engineering allows as far as decrypting IPs to understand how the hardware works...
<josuah>
even if the IP content is not included directly
<josuah>
not sure how this all works to be honest.
<gurki>
whitequark[cis]: thats hilarious, thanks for mentioning this :D
<whitequark[cis]>
you just decrypt the IP and get done with it
<whitequark[cis]>
don't tell me you've never pirated a movie either
<whitequark[cis]>
it's no one's business what you do in the privacy of your own computer
<josuah>
yes, but I did not write a novel out of the scenario of a pirated movie either
<josuah>
if something needs to be integrated with open-source project for instance
<gurki>
whitequark[cis]: im laughing about the plaintext key, not you doing whatever you do with it
<whitequark[cis]>
yeah i was responding to josuah
<gurki>
ah :)
<josuah>
the question is about how to face a lawyer if that had to happen
<josuah>
oh, nevermind
<josuah>
if someone of the size of i.e. intel, nvidia wants me legally down, they probably do not need me to be actually stealing anything ^_^'
<whitequark[cis]>
power does as power wants
<whitequark[cis]>
though, i have once got a c&d from xilinx, and i'm still alive
<josuah>
\o/ kudos on being alive, I mean it
<whitequark[cis]>
you don't actually need to tell anyone where you know how to use GW_JTAG from
<Hoernchen>
someone decrypts it and describes it, someone else implements it according to the docs
<whitequark[cis]>
or you can always do a parallel construction, exactly the same thing as the state does when it doesn't want to give up sources of evidence
<whitequark[cis]>
get it from RE, and then pretend you got it from very inspired experimentation
<whitequark[cis]>
Hoernchen: I think this might not work in this particular case
<whitequark[cis]>
since the person decrypting is still liable
<Hoernchen>
in this particular case the "thing" is... not particularly unique or exciting, it should be reasonably easy
<whitequark[cis]>
I mean
<josuah>
I just understood all the 4chan leaks... someone needs it, someone "else" publishes it, it's out, grab it :S
<whitequark[cis]>
I looked at GW_JTAG, I could describe it without any RE work
<Hoernchen>
ah yes but: who is that person? imagine docs showing up somewhere...
<gurki>
josuah: its a very, very (...) very complex topic with a lot of strong oppinions
<whitequark[cis]>
I can trivially break Gowin encryption too but it's not enough of a challenge to bother
<josuah>
gurki: sorry for opening the lid
<gurki>
for me personally its a minefield im not touching with a long stick
<Hoernchen>
the way it should not be done is the ham guys and their "reversed" ambe codec which is pretty much the dsp code decompiled and translated into c after running in the emulator..
<Hoernchen>
that is just tainted beyond repair
<whitequark[cis]>
as someone who thinks almost all IP should not exist I can only admire the degree of commitment there
<Hoernchen>
ambe is a major issue, it prevents open source DMR because you're supposed to buy a physical chip to use it
<whitequark[cis]>
yeah, I know about ambe
<Hoernchen>
and i'm still trying to forget all the things i never saw while not looking at the nice lattice diamond text f.. uh.. binary blobs
<whitequark[cis]>
i thought diamond ships ip as netlists?
<whitequark[cis]>
and whatever the hell is that cursed simulation format
<Hoernchen>
oh not the ip
<Hoernchen>
the part and bit descriptions
<Hoernchen>
long, long chains of boolean equations that tell you which bit does what
<Hoernchen>
the lattice ip that is used by sinplify was rsa bsafe protected at the time, so.. slightly more difficult than grepping for the key
<Hoernchen>
i.e. the tri speed ethernet mac
<whitequark[cis]>
ahhhh
<Hoernchen>
just found one of those pretty ascii tile maps
<Hoernchen>
alld of that was such a long time ago.. 2012 or something like that