buzzmarshall has quit [Quit: Konversation terminated!]
leah has quit [Ping timeout: 245 seconds]
leah has joined #linux-amlogic
hexdump0815 has quit [Ping timeout: 255 seconds]
hexdump0815 has joined #linux-amlogic
naoki has joined #linux-amlogic
chewitt has joined #linux-amlogic
chewitt has quit [Quit: Zzz..]
chewitt has joined #linux-amlogic
<wens>
lvrp16: teammate previously told us that the ddr training parameters need to be continously re-trained for higher frequencies, as the hardware will drift with time and temperature
naoki has quit [Quit: naoki]
<lvrp16>
narmstrong: you can't be serious...
<lvrp16>
wens: I knew that before since it is needed for some auto parts as part of testing. I got real world lesson on that with LPDDR4 > 3200MHz.
jacobk has quit [Ping timeout: 240 seconds]
<narmstrong>
lvrp16: yes I am... you can't imagine... those HW tools costs millions and requires very specific (old) linux releases to run and people write plenty of custom scripts around to automate stuff, it'a another world
<lvrp16>
narmstrong: I guess every cycle, they just pile on more crap until the engineer responsible retires and then nobody knows how to do it anymore :D
<narmstrong>
exact :-p
rockosov has quit [Ping timeout: 246 seconds]
rockosov has joined #linux-amlogic
camus has quit [Read error: Connection reset by peer]
camus has joined #linux-amlogic
luka177 has quit [Ping timeout: 255 seconds]
f_[xmpp] has joined #linux-amlogic
f_ has joined #linux-amlogic
<f_>
lvrp16, narmstrong: That's peak software/hardware engineering you got there!
Daanct12 has quit [Ping timeout: 246 seconds]
Daanct12 has joined #linux-amlogic
<f_>
Let's now see how spectacularly my DRAM init code b0rks
<f_>
(at least it feels nicer to look at)
<f_>
By the way that server slowdown issue yesterday was a VPS issue
luka177 has joined #linux-amlogic
<f_>
Guess what?
<f_>
Writing to 0x1000000 seems to work
<narmstrong>
Noice
<f_>
I'll have to test further just to be sure
<f_>
But relocating the stack b0rks
<f_>
Probably a wrong address then
<f_>
I'm proud of my DRAM init code :^)
<f_>
Not that good, at least as far as upstream coding style goes, but I can polish it later
<f_>
It's better than Amlogic code
<f_>
and of course, I'm abusing #if
<f_>
I think I spoke too soon.
luka177 has quit [Ping timeout: 250 seconds]
<f_>
Writing seems to work, but reading b0rks
<chewitt>
half-way done then :)
<f_>
lol
<narmstrong>
Just avoid reading then :-p
<f_>
Yeah run linux on SRAM
<f_>
Just avoid initialising DRAM
<f_>
interesting
<f_>
newer BL2 run memTestDataBus() on 0x0
<f_>
This looks suspicious to me:
<f_>
_DAT_c1300000 = 1;
<f_>
aka.
<f_>
writel(1, 0xc1300000);
<f_>
done before testing the dram
<f_>
and after the tests:
<f_>
writel(0, 0xc1300000);
BlueMatt has quit [Server closed connection]
BlueMatt has joined #linux-amlogic
<f_>
// REMAP THE ADDRESS SPACE BY WRITING TO NIC400 REMAP REGISTER
<f_>
//wr_reg(0xC1300000, 0x00000001);
<f_>
//__asm__ volatile("ISB");
<f_>
//__asm__ volatile("DMB SY");
<f_>
apparently remaps the address space?
<narmstrong>
Looks like something to do
<f_>
it doesn't change anything, at all.
<f_>
BUT I HAVE NO IDEA WHY SOME COMMENTS ARE WRITTEN IN ALL CAPS
<narmstrong>
Perhaps this remaps to 0 so it can be tested from bl2 then it’s remapped to 0x1000000
<f_>
lvrp16: and of course, libre computer is listed at the top of the sponsors page :P
luka177 has joined #linux-amlogic
<f_>
I wonder who had that great idea :P
* f_
sees that sweetpotato is now for sale.
<f_>
next: cottonwood :)
<f_>
Stays true to the librecomputer design fashion: all white with coloured GPIO!
<lvrp16>
5 cent marketing lol
<f_>
Now, with USB-C!
<f_>
I wonder if role-switching works, on that USB-C port
luka177 has quit [Ping timeout: 258 seconds]
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 240 seconds]
luka177 has joined #linux-amlogic
chewitt has quit [Quit: Zzz..]
<lvrp16>
f_: Sweet Potato has no USB on the Type-C, there's not enough channels
<lvrp16>
it's a dummy power port with Rd
<f_>
Sure
<f_>
Still nice to see USB-C
<f_>
I think RPi4 can't do role-switching either, have to check when I'm bored
<lvrp16>
Yeah good riddance to MicroUSB. We have to update La Frite to use Type-C and PoE as well
<f_>
Make it the same colour as lepotato/sweetpotato
<lvrp16>
the IP block requires the ID pin
luka177 has quit [Ping timeout: 258 seconds]
<f_>
lafrite looks slightly blue
<f_>
No idea why ¯\_(ツ)_/¯
<lvrp16>
it's the oils used to make the PCB color
<lvrp16>
white is very difficult to do for PCB
<f_>
and lafrite is supposed to be cheaper than lepotato
<lvrp16>
yeah between now and then end of the year, we're introducing a lot of new software features.
<f_>
Also, what's the difference between AML-S905X-CC-V1.0 and AML-S905X-CC-V1.0-A?
<lvrp16>
clean up the years of work everyone put in to make something usable
<lvrp16>
1.0 is 1.0A
<f_>
In the pictures the GPIO is all black, but what I got had coloured headers
<f_>
So I thought 1.0A was different
<lvrp16>
x.yZ, x is product version, y is revision, Z has no purpose or arbitrary purpose
<lvrp16>
It would be V1.1A V1.2A etc
<lvrp16>
kernel style V1.110A
<lvrp16>
all the revisions are software compatible, versions are different products.
<f_>
But then was that colourful GPIO header added recently?
<lvrp16>
yeah, we are switching everything to them. after counting pins for the 1000th time, I had enough.
<lvrp16>
if it was like 8 pins, ok, but 20 pins...
<f_>
Have to say they look nice
<f_>
And they indeed help. I wish I had that with raspberrypies too
<f_>
Could've saved me from checking pinout.xyz for the 2000th time.
<f_>
...just to hook up UART
<f_>
(I ran some libre/open source boot firmware and only composite video works..no HDMI
<f_>
(that was someone else's work. I did help a little bit back then)
luka177 has joined #linux-amlogic
<f_>
(Pi1-Pi4 are 'supported', it was recommended to use the Pi2 for the time being as it had the most things working)
<f_>
(IIRC Pi4 can't even boot anything)
<f_>
(same for Pi1. Pi2 can boot linux on one core)
<f_>
lvrp16: Anyway I like that change :)
<lvrp16>
hahaha I use pinout.xyz
<lvrp16>
so much
<lvrp16>
Pi 2 was probably the most OK RPi board
luka177 has quit [Ping timeout: 246 seconds]
<f_>
IIRC the reason why RPi don't give full schematics is because of Broadcom
<f_>
I think that's also the reason why their boot firmware is proprietary
<f_>
I'm not helping cleverca22 (the one who's working on that firmware) anymore, but am willing to start working helping him again at some point.
<f_>
s/working //
<f_>
Eventually I'd like to move away from github
<f_>
If all github projects had their home somewhere else I'd consider nuking my github account
<f_>
same for gitlab, which I used for hosting my repos before
<f_>
(I use my own infra for all git repos nowadays..hosting a git server isn't as complicated as one might think)
<f_>
hey, you could run a git server on lepotato and it would run fine
<lvrp16>
le potato runs all of our manufacturing lines
luka177 has joined #linux-amlogic
<f_>
may as well create a "powered by librecomputer lepotato" sticker lol
<f_>
But yeah, hosting your own git server doesn't really require much resources..so I wonder why people don't do it more often instead of relying on centralised proprietary hosting owned by some not very trustworthy company.
<f_>
(or even just relying on one of these small libre hosting)
luka177 has quit [Ping timeout: 245 seconds]
<f_>
I did miss some stuff in dmc init
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 258 seconds]
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 246 seconds]
luka177 has joined #linux-amlogic
hexdump0815 has quit [Quit: WeeChat 1.9.1]
hexdump0815 has joined #linux-amlogic
luka177 has quit [Ping timeout: 255 seconds]
<f_>
Really weird..
<f_>
Amlogic code is true magic
luka177 has joined #linux-amlogic
jacobk has joined #linux-amlogic
luka177 has quit [Ping timeout: 258 seconds]
jacobk has quit [Ping timeout: 248 seconds]
luka177 has joined #linux-amlogic
luka177 has quit [Ping timeout: 255 seconds]
<narmstrong>
f_: g12/sm1 can do role switching very well if routed on an micro-usb, with the proper CC management chip and driver it can also do role switching, everything is in place
luka177 has joined #linux-amlogic
<f_>
Nice
<f_>
minute: ^ could be interesting then
<f_>
Now that I think about it.
<narmstrong>
on gxl, it's another story... it was their first design with DWC3 and rose switching requires a full unbind/rebind of the host and gadget drivers to work cleanly