system starts, but display seems to deactivate

seth · 2024-10-09 12:50:19

Compile a 6.11 kernel w/ CONFIG_FB_RADEON explicitly enabled and see whether that has any effect.
https://wiki.archlinux.org/title/Kernel … ild_system

sukolyn · 2024-10-10 08:35:05

I've compiled kernel, but I still don't have display. zgrep -i radeon /proc/config.gz shows the same output telling FB_RAEON is not set.
before compiling, I added CONFIG_FB_RADEON=y at the beginning of ~/build/linux/config.
should I've had to write it under # Frame buffer Devices ?

here's last boot journal : http://0x0.st/X6oC.txt

seth · 2024-10-10 13:07:18

https://gitlab.archlinux.org/archlinux/ … eads#L7049
How exactly did you go about rebuilding the kernel?
Following https://wiki.archlinux.org/title/Kernel … ild_system you either have to inject a pre-configured config file or you could

sed 's/.*CONFIG_FB_RADEON.*/CONFIG_FB_RADEON=y/g' .config

in the prepare() function of the PKGBUILD

sukolyn · 2024-10-10 14:31:52

from https://wiki.archlinux.org/title/Kernel … _prepare(), I've got no patch to apply, or change kernel build conf.

I've modified ~/build/linux/config by adding CONFIG_FB_RADEON=y at the beginning of the file, then updpkgsums, and makepkg -s.
I'm going to remove "#...RADEON is not set" (as it is commented I thought it has no effect), move there "CONFIG_FB_RADEON=y",
and rebuild

I'll tell you what happened when kernel has finished c...o...m...p...i...l...i...n...g...

seth · 2024-10-10 14:44:03

The commented line is inert, I assume you also installed the kernel package after building it?

sukolyn · 2024-10-10 15:55:34

yes, i did. linux-break is its name, as shown in journal.

seth · 2024-10-10 17:57:21

But "zgrep -i radeon /proc/config.gz" still doesn't reflect the config change?
Did you typo it?

sukolyn · 2024-10-10 19:14:55

zgrep doesn't show changes.

I think I didn't typo:

$ head -2 ~/build/linux/config
#sukolyn add
CONFIG_FB_RADEON=y

seth · 2024-10-10 19:23:09

grep -i radeon ~/build/linux/linux-6.11.2/.config

sukolyn · 2024-10-10 19:51:17

$ grep -i radeon ~/build/linux/src/linux-6.11.2/.config
CONFIG_DRM_RADEON=m
CONFIG_DRM_RADEON_USERPTR=y
# CONFIG_FB_RADEON is not set

I will rebuild tomorrow.

sukolyn · 2024-10-11 19:47:23

guess what ? I've got the same problem with another machine including Nvidia GK208B [GeForce GT 730] and nouveau using linux 6.11.3

seth · 2024-10-11 20:17:41

Journal?
Though kelper is also supported by https://aur.archlinux.org/packages/nvidia-470xx-dkms/

sukolyn · 2024-10-11 21:56:50

I'd like to know what you think about the situation.
I mean, do you think it's normal to have to do all those things to just access a TTY ?

seth · 2024-10-11 22:07:39

Certainly not and has not been w/ previous kernels, has it?

As mentioned, there's something off w/ the 6.11 frambuffer handling.
For radeon I can see why it might have simply been forgotten and there's also some fuzz about the (OOT) nvidia drivers.

For (in tree) nouveau I'd have to see the journal to tell you whether the problems are perhaps related.

sukolyn · 2024-10-12 13:00:36

about the Radeon
I've compiled linux 6.11.3 with FB_RADEON set to y into config file, and it is still set at boot

$ zgrep -i radeon /proc/config.gz
CONFIG_DRM_RADEON=m
CONFIG_DRM_RADEON_USERPTR=y
CONFIG_FB_RADEON=y
CONFIG_FB_RADEON_I2C=y
CONFIG_FB_RADEON_BACKLIGHT=y
# CONFIG_FB_RADEON_DEBUG is not set

and here is the journal : http://0x0.st/X6_X.txt

about the Nvidia
here is the log of failing boot : http://0x0.st/X6_u.txt

seth · 2024-10-12 20:45:48

radeonfb doesn't show up
You've "oct. 12 14:23:48 archlinux kernel: radeon 0000:01:00.0: [drm] Cannot find any crtc or sizes"
It's all geared towards

oct. 12 14:23:47 archlinux kernel: fbcon: Deferring console take-over
oct. 12 14:23:47 archlinux kernel: simple-framebuffer simple-framebuffer.0: [drm] fb0: simpledrmdrmfb frame buffer device
oct. 12 14:23:47 archlinux kernel: fbcon: Taking over console

nouveau is the exact same thing, nouveaufb doesn't load, "[drm] Cannot find any crtc or sizes" and the simplydumb device runs the show.

Grrr..

zgrep -i console /proc/config.gz | curl -F 'file=@-' 0x0.st

sukolyn · 2024-10-12 23:03:08

https://0x0.st/X6Ox.txt

what is in LTS but that doesn't appear in this zgrep is : CONFIG_HW_CONSOLE=y

seth · 2024-10-13 07:40:44

I had hoped fro deviances along

CONFIG_FRAMEBUFFER_CONSOLE_DETECT_PRIMARY=y
CONFIG_FRAMEBUFFER_CONSOLE_ROTATION=y
CONFIG_FRAMEBUFFER_CONSOLE_DEFERRED_TAKEOVER=y

but nope.

https://www.spinics.net/lists/kernel/msg5057950.html

zgrep -i _vt /proc/config.gz

Edit, fuck - one underscore too many.

Last edited by seth (2024-10-13 07:41:05)

sukolyn · 2024-10-15 08:58:58

sorry for being late, here's the output :

# uname -sr
Linux 6.11.3-arch1-1

# zgrep -i _vt /proc/config.gz
CONFIG_IPV6_VTI=m
CONFIG_VT=y
CONFIG_VT_CONSOLE=y
CONFIG_VT_CONSOLE_SLEEP=y
CONFIG_VT_HW_CONSOLE_BINDING=y
CONFIG_TCG_VTPM_PROXY=m
CONFIG_SENSORS_VT1211=m
CONFIG_SENSORS_VT8231=m
# CONFIG_FB_VT8623 is not set
# CONFIG_HYPERV_VTL_MODE is not set
CONFIG_VT6655=m
CONFIG_VT6656=m

seth · 2024-10-15 15:13:32

"CONFIG_VT=y"
There's your new key for CONFIG_HW_CONSOLE, so that's not it.

With a little luck you're facing https://gitlab.freedesktop.org/drm/amd/-/issues/3665 and https://aur.archlinux.org/packages/linux-drm-tip-git has a fix for that…

sukolyn · 2024-10-15 15:33:29

https://gitlab.freedesktop.org/drm/amd/-/issues/3665 says it's been patched about a week ago in 6.12 rc1
so, it should be fixed in 6.12; though I don't know when this version of the kernel will be available in Arch, I think I will just hang with LTS until then.

thank you for all.

seth · 2024-10-15 19:41:35

To be clear

With a little luck you're facing https://gitlab.freedesktop.org/drm/amd/-/issues/3665

I'd not just bet on that. If you want to, try the drm-tip kernel to see whether you can expect this to be fixed.

sukolyn · 2024-10-17 13:04:52

f***

here's drm-tip-git's journal : http://0x0.st/XIQO.txt

seth · 2024-10-17 13:52:40

File a bug upstream.
You can https://wiki.gentoo.org/wiki/Kernel_git-bisect - the framebuffer code seems to have been damaged considerably and there're A LOT of related commits - blind guessing will likely not work

I'll do nevertheless

 drm/radeon: remove radeon_connector_edid() and stop using edid_blob_ptr

Did we ever try to inject the edid?
https://wiki.archlinux.org/title/Kernel … s_and_EDID
You can copy it from the LTS kernel boot and /sys/class/drm/card*/edid

sukolyn · 2024-10-17 15:23:33

here's what I did with no success though:
boot on LTS
copy /sys/class/drm/card0/card0-HDMI-A-1/edid to /usr/lib/firmware/edid/HDMIedid.bin
reboot
edit grub drm-tip-git's options
add drm.edid_firmware=card0-HDMI-A-1:edid/HDMIedid.bin as kernel parameter
boot
and see my screen going black, and remain black.
here's its journal: http://0x0.st/XIe9.txt

did I miss something ?

File a bug upstream.

how do I do this ?

Arch Linux

#26 2024-10-09 12:50:19

Re: system starts, but display seems to deactivate

#27 2024-10-10 08:35:05

Re: system starts, but display seems to deactivate

#28 2024-10-10 13:07:18

Re: system starts, but display seems to deactivate

#29 2024-10-10 14:31:52

Re: system starts, but display seems to deactivate

#30 2024-10-10 14:44:03

Re: system starts, but display seems to deactivate

#31 2024-10-10 15:55:34

Re: system starts, but display seems to deactivate

#32 2024-10-10 17:57:21

Re: system starts, but display seems to deactivate

#33 2024-10-10 19:14:55

Re: system starts, but display seems to deactivate

#34 2024-10-10 19:23:09

Re: system starts, but display seems to deactivate

#35 2024-10-10 19:51:17

Re: system starts, but display seems to deactivate

#36 2024-10-11 19:47:23

Re: system starts, but display seems to deactivate

#37 2024-10-11 20:17:41

Re: system starts, but display seems to deactivate

#38 2024-10-11 21:56:50

Re: system starts, but display seems to deactivate

#39 2024-10-11 22:07:39

Re: system starts, but display seems to deactivate

#40 2024-10-12 13:00:36

Re: system starts, but display seems to deactivate

#41 2024-10-12 20:45:48

Re: system starts, but display seems to deactivate

#42 2024-10-12 23:03:08

Re: system starts, but display seems to deactivate

#43 2024-10-13 07:40:44

Re: system starts, but display seems to deactivate

#44 2024-10-15 08:58:58

Re: system starts, but display seems to deactivate

#45 2024-10-15 15:13:32

Re: system starts, but display seems to deactivate

#46 2024-10-15 15:33:29

Re: system starts, but display seems to deactivate

#47 2024-10-15 19:41:35

Re: system starts, but display seems to deactivate

#48 2024-10-17 13:04:52

Re: system starts, but display seems to deactivate

#49 2024-10-17 13:52:40

Re: system starts, but display seems to deactivate

#50 2024-10-17 15:23:33

Re: system starts, but display seems to deactivate

Board footer