[Solved] xorg-server doesn't start [reason still undefined]

[l_arbalette]

Hello Gentoers !

I'm running into many troubles after a fresh install of my Gentoo (a 15 years' old HDD that died, forcing me to re-install Gentoo from scratch. Fortunately, my /home and all my data were stored on 4 others HDD running RAID1 Only Gentoo's OS has needed a reinstall)

Everything went fine (after some struggle with UEFI stuff ! ) and I was able to log into KDE/Plasma to finish my setup (some things are still not yet working well, like the sound, the printer, SSL certificates issues an issue that I've never had before, preventing me to download some sources while emerging !)

while trying to solve this littles issues and emerging some programs to restore my previous configuration (especially cudnn and nvidia cuda utils (because I'm running deep neural network python code these days : I mention it because it may be related, since it is nvidia stuff ?), I must have broken something, but I'm not able to identify what :
after a reboot, my X server no longer starts...although everything was fine just before.

I've disabled the display-manager during init process (initially launching SDDM, then LightDM before I understood that the DM was not faulty), to be able to log in as root (otherwise, the system tries to launch the display-manager that starts X, and the system freezes).
Running startx -- vt7, I get a black screen with a non-blinking cursor on the top left hand, and the whole system is frozen...I'm not even able to switch back to TTY1.

elogind has started correctly

Code: Select all

SESSION UID USER SEAT  TTY 
      2   0 root seat0 tty1

1 sessions listed.

Here is my emerge --info https://pastebin.com/T9Q5kpJQ
Here are the version installed of xorg-server, nvidia-drivers and elogind :

Code: Select all

sys-auth/elogind-246.10-r2
x11-base/xorg-server-21.1.3-r1
x11-drivers/nvidia-drivers-470.103.01

Here is my kernel config : https://pastebin.com/0WGrCq9w
Here is my dmesg : https://pastebin.com/RmeGujKE
Here is my make.conf : https://pastebin.com/iWDg5reH
Here are the service started during init process : rc-update show https://pastebin.com/e5wTpWw0
And here is the status of rc-status : https://pastebin.com/chuUh29w

And finally, here is the Xorg.0.log showing no error, but leading to a system freeze :

Code: Select all

[   718.766] 
X.Org X Server 1.21.1.3
X Protocol Version 11, Revision 0
[   718.766] Current Operating System: Linux localhost 5.15.26-gentoo #1 SMP Thu Mar 31 07:21:31 CEST 2022 x86_64
[   718.766] Kernel command line: root=PARTUUID=dff0df8d-238f-4046-8f41-55238e064439 BOOT_IMAGE=/vmlinuz-5.15.26-gentoo root=/dev/sda3 ro
[   718.766]  
[   718.766] Current version of pixman: 0.40.0
[   718.766] 	Before reporting problems, check http://wiki.x.org
	to make sure that you have the latest version.
[   718.766] Markers: (--) probed, (**) from config file, (==) default setting,
	(++) from command line, (!!) notice, (II) informational,
	(WW) warning, (EE) error, (NI) not implemented, (??) unknown.
[   718.766] (==) Log file: "/var/log/Xorg.0.log", Time: Sun Apr  3 12:37:49 2022
[   718.767] (==) Using config file: "/etc/X11/xorg.conf"
[   718.767] (==) Using system config directory "/usr/share/X11/xorg.conf.d"
[   718.768] (==) ServerLayout "Layout0"
[   718.768] (**) |-->Screen "Screen0" (0)
[   718.768] (**) |   |-->Monitor "Monitor0"
[   718.768] (**) |   |-->Device "Device0"
[   718.768] (**) |-->Input Device "Keyboard0"
[   718.768] (**) |-->Input Device "Mouse0"
[   718.768] (==) Automatically adding devices
[   718.768] (==) Automatically enabling devices
[   718.768] (==) Automatically adding GPU devices
[   718.768] (==) Automatically binding GPU devices
[   718.769] (==) Max clients allowed: 256, resource mask: 0x1fffff
[   718.769] (WW) The directory "/usr/share/fonts/TTF" does not exist.
[   718.769] 	Entry deleted from font path.
[   718.769] (WW) The directory "/usr/share/fonts/OTF" does not exist.
[   718.769] 	Entry deleted from font path.
[   718.769] (WW) The directory "/usr/share/fonts/Type1" does not exist.
[   718.769] 	Entry deleted from font path.
[   718.769] (WW) `fonts.dir' not found (or not valid) in "/usr/share/fonts/100dpi".
[   718.769] 	Entry deleted from font path.
[   718.769] 	(Run 'mkfontdir' on "/usr/share/fonts/100dpi").
[   718.769] (==) FontPath set to:
	/usr/share/fonts/misc,
	/usr/share/fonts/75dpi
[   718.769] (==) ModulePath set to "/usr/lib64/xorg/modules"
[   718.769] (WW) Hotplugging is on, devices using drivers 'kbd', 'mouse' or 'vmmouse' will be disabled.
[   718.769] (WW) Disabling Keyboard0
[   718.769] (WW) Disabling Mouse0
[   718.769] (II) Module ABI versions:
[   718.769] 	X.Org ANSI C Emulation: 0.4
[   718.769] 	X.Org Video Driver: 25.2
[   718.769] 	X.Org XInput driver : 24.4
[   718.769] 	X.Org Server Extension : 10.0
[   718.770] (++) using VT number 7

[   718.770] (II) systemd-logind: logind integration requires -keeptty and -keeptty was not provided, disabling logind integration
[   718.773] (--) PCI:*(1@0:0:0) 10de:1402:1028:1139 rev 161, Mem @ 0xf6000000/16777216, 0xe0000000/268435456, 0xf0000000/33554432, I/O @ 0x0000e000/128, BIOS @ 0x????????/131072
[   718.773] (II) LoadModule: "glx"
[   718.773] (II) Loading /usr/lib64/xorg/modules/extensions/libglx.so
[   718.775] (II) Module glx: vendor="X.Org Foundation"
[   718.775] 	compiled for 1.21.1.3, module version = 1.0.0
[   718.775] 	ABI class: X.Org Server Extension, version 10.0
[   718.775] (II) LoadModule: "nvidia"
[   718.775] (II) Loading /usr/lib64/xorg/modules/drivers/nvidia_drv.so
[   718.780] (II) Module nvidia: vendor="NVIDIA Corporation"
[   718.780] 	compiled for 1.6.99.901, module version = 1.0.0
[   718.780] 	Module class: X.Org Video Driver
[   718.781] (II) NVIDIA dlloader X Driver  470.103.01  Thu Jan  6 12:18:33 UTC 2022
[   718.781] (II) NVIDIA Unified Driver for all Supported NVIDIA GPUs
[   718.784] (II) Loading sub module "fb"
[   718.784] (II) LoadModule: "fb"
[   718.784] (II) Module "fb" already built-in
[   718.784] (II) Loading sub module "wfb"
[   718.784] (II) LoadModule: "wfb"
[   718.785] (II) Loading /usr/lib64/xorg/modules/libwfb.so
[   718.786] (II) Module wfb: vendor="X.Org Foundation"
[   718.786] 	compiled for 1.21.1.3, module version = 1.0.0
[   718.786] 	ABI class: X.Org ANSI C Emulation, version 0.4
[   718.786] (II) Loading sub module "ramdac"
[   718.786] (II) LoadModule: "ramdac"
[   718.786] (II) Module "ramdac" already built-in

as you can see, no error in this log, and after the last line, the system freezes...
Does anybody has an idea to solve this issue ?
Thanks a lot !

[l_arbalette]

I've ran the nvidia bug report utility, in case it could help.
Here is the result : https://pastebin.com/cX9Jd8a4
It includes the whole Xorg config files, lspci output with modules information.

I'm surprised by the last line, that informs that drm/*/device/driver do not exists :

Code: Select all

*** ls: ls: impossible d'accéder à '/sys/class/drm/*/device/driver': Aucun fichier ou dossier de ce type

my kernel has this config :

Code: Select all

CONFIG_DRM=y

I'm not sure it is related...

[mike155]

Please take a look at https://wiki.gentoo.org/wiki/NVIDIA/nvidia-drivers, especially at the section

Important!

The "Mark VGA/VBE/EFI FB as generic system framebuffer" option moved in kernel 5.15 with a new symbol name for all arches. This may cause a black screen or no progress shown after the loader on boot if changes are not made.

CONFIG_SYSFB_SIMPLEFB (which is "Mark VGA/VBE/EFI FB as generic system framebuffer") is NOT enabled in your kernel config.

Furthermore, there's an error message in the output of dmesg - but it's not related to the nvidia graphics issue:

Code: Select all

[    7.482604] Loading firmware: rtl_nic/rtl8168e-3.fw
[    7.482774] r8169 0000:03:00.0: Direct firmware load for rtl_nic/rtl8168e-3.fw failed with error -2
[    7.482780] r8169 0000:03:00.0: Falling back to sysfs fallback for: rtl_nic/rtl8168e-3.fw
[   70.722988] r8169 0000:03:00.0: Unable to load firmware rtl_nic/rtl8168e-3.fw (-110)

[Ionen]

SYSFB_SIMPLEFB doesn't matter if you're using FB_EFI (which is enabled and in-use looking at dmesg, meaning no problem), it also has no relation to the good workings of Xorg

gentoo-kernel-bin uses FB_EFI too and disables SIMPLEFB currently

Unfortunately I'm not spotting anything related to Xorg that seems off at the moment. The /sys/class thing is normal too (at least on a single gpu setup). Guess you could always try newer ~arch nvidia-drivers-510 (although I'm not aware of issues with 470 at the moment and it generally been more stable while some 510 issues are still waiting to be resolved, not that they affect everyone).

Edit: I mention this often, but do double-check that your user is in the video group, a login manager that works and things not working after tend to be related. ls -l /usr/bin/nvidia-modprobe should also be -rws--x--- root:video

[l_arbalette]

Thanks to both of you for your answers.
I've noticed this error in dmesg, but it seems related to my network gigabit card, which works well anyway...

I've indeed the FB_EFI enabled, after the struggle I went into with booting my system with UEFI ! probably it has nothing to do with this, but finally, it's activated !

regarding ls -l /usr/bin/nvidia-modprobe :

Code: Select all

-rws--x--- 1 root video 43560  1 avril 15:28 /usr/bin/nvidia-modprobe

so it looks ok

and i confirm that root, sddm and all my users are members of the video group.
Regarding the login manager, not sure to understand what do you mean...
but I'm able to log into root account, as well as my personal user account, under TTY1 console...

[Ionen]

Regarding the login manager, not sure to understand what do you mean...
but I'm able to log into root account, as well as my personal user account, under TTY1 console...

Oh okay, given you mentioned you had a similar issue in the other thread I thought sddm/lightdm were starting fine and it was failing after login (that's how it was described there). Nevermind that bit then.

[mike155]

SYSFB_SIMPLEFB doesn't matter if you're using FB_EFI (which is enabled and in-use looking at dmesg, meaning no problem), it also has no relation to the good workings of Xorg

@Ionen: Please edit and improve the Wiki page. Because it definitely says that SYSFB_SIMPLEFB should be enabled:

A framebuffer driver is required for rendering the Linux console (TTY) as this functionality is not provided by the proprietary nvidia driver. As shown below, set Mark VGA/VBE/EFI FB as generic system framebuffer (CONFIG_SYSFB_SIMPLEFB=y), and then enable a framebuffer driver. Common options for this are to use either simplefb (CONFIG_FB_SIMPLE=y) or efifb (CONFIG_FB_EFI=y).

I cannot check whether that's wrong or right, because I don't have an nvidia graphics card...

[l_arbalette]

Guess you could always try newer ~arch nvidia-drivers-510 (although I'm not aware of issues with 470 at the moment and it generally been more stable while some 510 issues are still waiting to be resolved, not that they affect everyone)

This is really strange : I've just updated nvidia-drivers to version #570 (~amd64 keyword) and then, startx is starting well
I don't understand why I do need the new driver between yesterday and today !

I've enabled back SDDM, and everything is back again...

Thanks for the suggestion, my issue is solved...although I still don't understand what has caused this need of the latest nvidia driver (maybe the emerge of cudnn stuffs ?)