<ballerburg9005>
I have been sieving through the drivers but I don't even really understand how the color space is set
<ballerburg9005>
in the datasheet it lists "VPU Registers" but not the base address
<ballerburg9005>
I suppose it would be there
<ballerburg9005>
I have been accessing the registers with mmap through /dev/mem, from the drivers' code it only seems nebulous to me which exact addresses are accessed
kaspter has joined #linux-amlogic
<ballerburg9005>
I tried ff900000-ff9fffff : ff900000.vpu vpu but it has no effects
zkrx has quit [Ping timeout: 240 seconds]
<ballerburg9005>
maybe there was something wrong with my code, have to sleep now
<ballerburg9005>
I just made it blue instead of pink and it crashed the box twice
buzzmarshall has quit [Remote host closed the connection]
sputnik_ has joined #linux-amlogic
zkrx has joined #linux-amlogic
sputnik_ has quit [Remote host closed the connection]
sputnik_ has joined #linux-amlogic
Barada has joined #linux-amlogic
Barada has quit [Client Quit]
random_yanek has quit [Quit: random_yanek]
random_yanek has joined #linux-amlogic
camus has joined #linux-amlogic
kaspter has quit [Ping timeout: 265 seconds]
camus is now known as kaspter
camus has joined #linux-amlogic
kaspter has quit [Ping timeout: 260 seconds]
camus is now known as kaspter
adema has quit [Ping timeout: 258 seconds]
adema has joined #linux-amlogic
ldevulder_ is now known as ldevulder
yann has joined #linux-amlogic
chewitt has quit [Read error: Connection reset by peer]
chewitt has joined #linux-amlogic
kaspter has quit [Ping timeout: 268 seconds]
kaspter has joined #linux-amlogic
<narmstrong>
ballerburg9005: yes it's ff900000, but the reg offsets must be *4 to be correct for HW memory access
<narmstrong>
ballerburg9005: for the distorted color, I only spent a few minutes on it, but AFAIK the new Amlogic u-boot set the HDR post-processing block, but we don't use them in the kernel, so the color is bad
chewitt_ has joined #linux-amlogic
chewitt has quit [Ping timeout: 268 seconds]
drieschel has joined #linux-amlogic
Barada has joined #linux-amlogic
<ballerburg9005>
narmstrong: so basically just flip this block back to normal?
<ballerburg9005>
narmstrong: it was my first guess as well that this is due to some new "higher quality" graphics setting set from uboot
<ballerburg9005>
though, it looks extremely like 8bit RBG set to 10bit BGR or something like that
<ballerburg9005>
with the pink and green and the washed down black
<narmstrong>
ballerburg9005: if you find how, yes !
drieschel has quit [Quit: drieschel]
sputnik_ has quit [Ping timeout: 240 seconds]
<chewitt_>
ballerburg9005 the long-running workaround is to chainload mainline u-boot from vendor, but this complicates support for lots of devices
chewitt_ is now known as chewitt
<chewitt>
finding the proper solution would be awesome
jelly is now known as alot
alot is now known as jelly
<chewitt>
recent vendor u-boot allows you to set "use_rgb_to_yuv=0" in boot.ini which "fixes" the problem
<chewitt>
(not much use on devices that don't use boot.ini)
<ballerburg9005>
narmstrong: it was the HDR, like you said
<ballerburg9005>
thanks!
<narmstrong>
Nice thanks !
<ballerburg9005>
narmstrong: should I write a patch?
<narmstrong>
ballerburg9005: yes please !
kaspter has quit [Ping timeout: 246 seconds]
kaspter has joined #linux-amlogic
Barada has quit [Quit: Barada]
<ballerburg9005>
yeah it is a bug, the datasheet says "reg_only_mat : // unsigned , default = 0 ,only use input matrix ,work when hdr
<ballerburg9005>
oh, it says disable
<ballerburg9005>
then its not
camus has joined #linux-amlogic
kaspter has quit [Ping timeout: 240 seconds]
camus is now known as kaspter
drieschel has joined #linux-amlogic
drieschel has quit [Quit: drieschel]
drieschel has joined #linux-amlogic
drieschel has quit [Client Quit]
<ballerburg9005>
narmstrong: I sent the patch to linux-amlogic@lists.infradead.org but I haven't received a mail back from the list
<narmstrong>
ballerburg9005: I think its normal, anyway I got it
<ballerburg9005>
you never know these days with all the anti-spam
drieschel has joined #linux-amlogic
drieschel has quit [Client Quit]
<ballerburg9005>
damn I didn't realize how bad the tearing was with the pink colors
<ballerburg9005>
I suppose there is no double-buffering at the moment?
<ballerburg9005>
also, no sound?
<ballerburg9005>
it says audio in the wiki
<narmstrong>
ballerburg9005: tearing ? where ? with what ?
<narmstrong>
ballerburg9005: audio depends on the DT board you use, it may not have sound
cmeerw has joined #linux-amlogic
<narmstrong>
is there someone having an Odroid HC4 here ? It would be interesting to check if the U-Boot PCIe support I just finished to push upstream works with the HC4 SATA controller
<narmstrong>
AHCI_PCI would certainly do the trick
<ballerburg9005>
narmstrong: it seems to be the typical video tearing with Xorg. Without double or triple buffering the video rendering is never sync to vblank, hence the buffer empties half way through the picture, producing very annoying tears. #tinymce.body
<chewitt>
ballerburg9005 audio also works, but needs mixer settings and probably an alsa conf
B1oHazard has joined #linux-amlogic
camus has joined #linux-amlogic
<B1oHazard>
Hello, can someone help me? I tried to build mainline u-boot for my G12B board, but in newer versions HDMI output didn't work. It boot linux kernel successfully, but without screen. v2020.10 works fine, so i think its regression.
kaspter has quit [Ping timeout: 240 seconds]
camus is now known as kaspter
<B1oHazard>
I tried chewitt prebuilt u-boot, and it also work, but without display.
<narmstrong>
B1oHazard: hello, can you try multiple version since 2020.10 and report which ones are working ?
<narmstrong>
Like 2021.01 and 2020.04 ?
<narmstrong>
And what defconfig are you using for your g12b board ?
<B1oHazard>
narmstrong hello, yes, i can try. I used odroid-n2 defconfig for old version u-boot, and beelink-gtking and same odroid-n2 for master branch.
<B1oHazard>
few minutes, I will check few versions
<B1oHazard>
narmstrong v2021.01-rc1 is affected, v2020.10 works. odroid-n2 defconfig used for both.
<B1oHazard>
v2021.01 and v2021.04 didn't work too. I think every version >=v2021.01-rc1 has broken display. And it broken not for u-boot only, it boot kernel, but without display functions.
<B1oHazard>
narmstrong on LibreELEC forums one guy have same problem on dreambox, and same versions works for him. But he use hack on u-boot master branch and display work for him after that, but not for me.
<B1oHazard>
He say after this display works on DreamBox. But for my GT-King this commit didn't help. Maybe this information will be useful for fix bug.
<narmstrong>
B1oHazard: it doesn’t help much
<narmstrong>
B1oHazard: do you know how to use git bisect ?
<narmstrong>
B1oHazard: you give the last know working commit and the last know broken commit, it will make you test multiple revisions and will give you the offending change
<narmstrong>
B1oHazard: I’ll have a look on my side on the next few days
<B1oHazard>
narmstrong As i know bisect can help me to find commit, but I already found it.
<narmstrong>
This merge commit is not precise enough
<B1oHazard>
Oh, ok. I will try now to use bisect, v2020.10 will be good commit, master head or v2021.01 will be bad. Any suggestions?
<B1oHazard>
narmstrong 139e4a1cbe60de96d4febbc6db5882929801ca46 is the first bad commit
<B1oHazard>
My bisect result
GNUtoo has quit [Ping timeout: 240 seconds]
GNUtoo has joined #linux-amlogic
<narmstrong>
B1oHazard: this explains what the LibreELEC guy found !
<narmstrong>
(I’m AFK until tomorrow morning Paris time)
<B1oHazard>
narmstrong yes, 1 min will try
jbrunet has quit [Ping timeout: 250 seconds]
jbrunet has joined #linux-amlogic
GNUtoo has quit [Ping timeout: 260 seconds]
GNUtoo has joined #linux-amlogic
<B1oHazard>
narmstrong yes, it work after removing "> 0" using odroid-n2 defconfig
<narmstrong>
\o/
<B1oHazard>
but with gtking defconfig black screen :)
<narmstrong>
Typo error and the self tests didn’t look more than 2 resets so it still worked...
<narmstrong>
Oh, it must be another error then
<B1oHazard>
My device GT-King, but I do all tests with odroid-n2 defconfig, cuz in v2020.10 GT-King didnt supported
<B1oHazard>
narmstrong tried to compile master using beelink-gtking defconfig with removing the « > 0 » in reset-uclass.c and pure master branch. With your fix, screen didn't work only in u-boot, I can see LibreELEC splash screen and Kodi. Without your fix screen didn't work in both u-boot and LE.
<narmstrong>
Ok this means something is missing in uboot
<narmstrong>
Thanks I’ll look and post fixes tomorrow :-)