<loki666>
probably unrelated to HDMI but could explain why I lost wifi
Guest6901 has quit [Ping timeout: 480 seconds]
<tokyovigilante>
ah right, sorry, not sure how that's got in, has been working for me. I'll fix the IOMMU commit. The emac node is for the ethernet pins on the H616 which would be connected directly to a socket. Not sure how that 0 slipped in but nothing to do with wifi, I'd check you have the rtw88 module built and the correct firmware first
<tokyovigilante>
the wifi is an entirely separate chip and uses SDIO for communication
<loki666>
rtw88 module is there with firmware, but I get an error about external_clock
<tokyovigilante>
That emac0 line is correct it seems
<loki666>
ok
<tokyovigilante>
have you had wifi working before?
<loki666>
yes with 6.9 + RG35XX dts
<tokyovigilante>
you probably just need to check your kconfig again, and make sure you have all the power patches in particular, I've never seen that pwrseq message but that's related to the power supply to the wifi chip, there's a pwrseq node in the -Plus DT which you can have a look at
<tokyovigilante>
Have you updated the DT since then? that's a pretty old WIP
<loki666>
I'll check not on my computer right now
<tokyovigilante>
you should have all the bits there, certainly wifi and hdmi can coexist, good work getting a buildroot config going, I'm just using a fedora userspace
<tokyovigilante>
might be worth looking into FEL-boot and TFTP for the kernel if you are interested in developing
<ItsKaitlyn03>
i'll likely go through everything, i can guarantee this code is not stable and stuff is kinda janky
<ItsKaitlyn03>
I need to fix stuff before I feel comfortable just putting this code out in the wild
<ItsKaitlyn03>
it also only supports LPDDR4
<jernej>
ItsKaitlyn03: did you compare register dump with vendor and your driver?
<jernej>
suppoting only LPDDR4 is of course fine, you have to start somewhere :)
<ItsKaitlyn03>
i dumped my dram params and based my stuff off of the h616 dram and just looked at the libdram a133 decomp and my own ida decomp, found any differences and patched them
<ItsKaitlyn03>
also removed things I couldn't test
<ItsKaitlyn03>
or things that don't exist on a133, that otherwise do on h616
loki666 has quit [Remote host closed the connection]
loki666 has joined #linux-sunxi
<ItsKaitlyn03>
So now, should I try loading a kernel?
<ItsKaitlyn03>
I doubt it'll work
<jernej>
ItsKaitlyn03: you can try with memtest command in U-Boot, but you have to enable it in config first
<ItsKaitlyn03>
whats the option?
<jernej>
CONFIG_CMD_MEMTEST
<ItsKaitlyn03>
okay i think u-boot is slightly messed up
<ItsKaitlyn03>
i think it thinks the base SDRAM addr is 0
<ItsKaitlyn03>
"Mem error @ 0x0000000000000510: found E3C0A003E28D00B7, expected 00000000000000A2"
<ItsKaitlyn03>
oops
<jernej>
ah, you have to define CONFIG_SYS_MEMTEST_START and CONFIG_SYS_MEMTEST_END
machinehum has joined #linux-sunxi
<jernej>
start is 0x40000000
<ItsKaitlyn03>
end would be 0x13FFFFFFF
<ItsKaitlyn03>
yep?
<jernej>
that's if you have 4 GB RAM
<ItsKaitlyn03>
yeah i have 4gb of ram, but due to some weirdness i had with u-boot, it reports as 2GiB
<jernej>
please check and/or test and then I'll send patches to ML
<ItsKaitlyn03>
pure H616 dram init doesn't work on A133, I tried. There is a bit of differences between H616 DRAM and A133 DRAM that causes H616 DRAM init to just not work, even though they're quite similar in some ways :P
<ItsKaitlyn03>
I already have dual-phy stuff :P
<jernej>
I pinged you so you can evaluate differences further :)
<jernej>
in any case, H6 and H616 seem to have completely the same controller, so some functions can be shared
<jernej>
I'm sure that's the case with A133 too, right?
<ItsKaitlyn03>
nope, a lot of functions are handled differently in slightly different ways
<ItsKaitlyn03>
prcm for ohms/rescal does not exist on a133
<ItsKaitlyn03>
they do it in SYS_CFG
<jernej>
yeah, but that's not controller part
<jernej>
I have sunxi_mctl_ctl_reg structure in mind
<ItsKaitlyn03>
they share that in common afaik, but dram_para is different (addition of some tpr parameters) and some of the hacks h616 dram init does crashes on a133
<ItsKaitlyn03>
i did try adding a bunch of "is_enabled" and try to conditionally handle a133, but i figured it would've made h616 dram init messy
<ItsKaitlyn03>
i think I'll try to build mainline linux and see if I can just write it into memory somewhere and load it
<ItsKaitlyn03>
afaik, A133 is supported in mainline somewhat(?)
<jernej>
well, first you need something working and then we can discuss where to put it. but if PHY is not the same, then there is not much sense of having it in the same file
<ItsKaitlyn03>
yeah, I want to go over DRAM a bit more
<jernej>
do A100 and A133 use same die?
<ItsKaitlyn03>
afaik, not really sure what A100 is. allwinner makes no mention of it, other than that A133 is based on A100. A100 might just be either an earlier variant of A100, or a slower A133 due to manufacturing constraints. hard to say really
<jernej>
well, there are a few A100 drivers, so I guess A133 can use them
<ItsKaitlyn03>
yeah a133 is based on a100. for a133, they use the a100 dts
<ItsKaitlyn03>
in fact
<ItsKaitlyn03>
they share the same chipid afaik
<jernej>
other than that, I don't remember A133 specific support
<ItsKaitlyn03>
both R818/A100/A133 are both 0x1855 chipid afaik
<ItsKaitlyn03>
not both, all of those 3 are 0x1855
<ItsKaitlyn03>
i think a100 was the first to release, then a133, and then r818
<ItsKaitlyn03>
wonder what even happened to a100, its documented nowhere
JohnDoe_71Rus has quit [Quit: KVIrc KVIrc Quasar 5.2.2, revision: 5.2.0+git-7584-200e7bcec, build type: debug, sources date: 20160102, built on: 2024-05-13 20:04:01 UTC 5.2.0+git-7584-]
KREYREN_ has joined #linux-sunxi
KREYREN_oftc has quit [Ping timeout: 480 seconds]
<tokyovigilante>
jernej: nice, thanks. will give them a test
<tokyovigilante>
today.
<Jookia>
i woke up in a cold sweat realizing i may have gotten punked by the t113 pwm datasheet
<Jookia>
ItsKaitlyn03: congrats on the dram init!
<ItsKaitlyn03>
currently going insane as to why linux wont init correctly with my own ramfs image
<Jookia>
what's the error?
<ItsKaitlyn03>
it just hangs at "Run <xyz> as init process"
<ItsKaitlyn03>
nothing else
<Jookia>
hmm
<ItsKaitlyn03>
ive been trying various other ramfs images, i tried using alpine
<ItsKaitlyn03>
nothing
<ItsKaitlyn03>
it just doesnt work
<ItsKaitlyn03>
and i dont understand why
<Jookia>
interesting. maybe you could use kdb?
<Jookia>
finding out if the kernel is hanging would be a good start
<jernej>
ItsKaitlyn03: do you have TF-A?
<Jookia>
yeah i think i got tricked by the datasheet
<Jookia>
when it said the total cycles is (entire_period + 1) i assumed that meant it added an extra cycle HOWEVER, it could literally just be that its refering to the entire_period register being the (actual value - 1)...
<ItsKaitlyn03>
i did stub out TF-A which in my defense I did get a bit hasty with it
<ItsKaitlyn03>
i guess ill go ahead and actually make a proper BL31 impl for A133
<jernej>
that would be a good step :)
<Jookia>
kernel may be waiting for some atf thing to happen before running userspace
<ItsKaitlyn03>
good point
<Jookia>
that said i don't have atf on my t113 only u-boot and sadness
<ItsKaitlyn03>
I kinda janked everything together and I'll need to unjank before I put anything online
KREYREN_ has quit []
KREYREN_oftc has joined #linux-sunxi
<Jookia>
:D
<ItsKaitlyn03>
but just having this work is a really good sign :D
<KREYREN_oftc>
How do you format the SPI chip for AllWinner A64 with u-boot so that it's able to boot? (the built binary can't be flashed as is bcs it's not the size of the chip and i am unable to get it past https://dpaste.org/6HHO1/raw)
* KREYREN_oftc
assumes that the u-boot is corrupted on the SPI
<Jookia>
ItsKaitlyn03: yeah, it's a lot of progress. thanks for the work :)
<ItsKaitlyn03>
still need to figure out whats up with mmc0
<ItsKaitlyn03>
but im also now booted into alpine linux
<ItsKaitlyn03>
yeah it only detects one core, it might be due to my dtb though
<ItsKaitlyn03>
im using the same one i am for u-boot
<ItsKaitlyn03>
not optimal but i mean, it works
<Jookia>
possibly yeah
JuniorJPDJ has quit []
<jernej>
ItsKaitlyn03: bl31 is responsible for core management. You have to add code for enabling/disabling cores there
<ItsKaitlyn03>
makes sense
t4h4[m] has quit []
obbardc has quit [Quit: Client limit exceeded: 20000]
cperon has quit []
ftg has quit [Read error: Connection reset by peer]
<tokyovigilante>
ItsKaitlyn03: nice! first boot always very satisfying, and always after remembering something key you forgot, after you fixed the actual problem hours before
<ItsKaitlyn03>
second goal tonight is to clean up dram initialization
<ItsKaitlyn03>
i only tested it on my dram parameters, and it only supports LPDDR4 as that's all i have