[SOLVED] Graphics does not come up after Ceres update/upgrade.

mr12345 · 2023-09-21 13:35:54

After todays update/upgrade of my Ceres setup the gui login page (SLiM) does not come up.Just the recovery console.
The upgrade list was huge... I did not save it, sorry. I do remember there were some Mesa upgrades.

After login to my user account and "X' command:

[   244.951] (II) modesetting: Driver for Modesetting Kernel Drivers: kms
[   244.951] (II) FBDEV: driver for framebuffer: fbdev
[   244.951] (II) VESA: driver for VESA chipsets: vesa
[   244.951] (EE) 
Fatal server error:
[   244.953] (EE) xf86OpenConsole: Cannot open virtual console 7 (Permission denied)
[   244.953] (EE) 
[   244.954] (EE) 
Please consult the The X.Org Foundation support 
	 at http://wiki.x.org
 for help. 
[   244.956] (EE) Please also check the log file at "/home/user/.local/share/xorg/Xorg.0.log" for additional information.
[   244.956] (EE) 
[   244.957] (WW) xf86CloseConsole: KDSETMODE failed: Bad file descriptor
[   244.957] (WW) xf86CloseConsole: VT_GETMODE failed: Bad file descriptor
[   244.957] (EE) Server terminated with error (1). Closing log file.

Under root:

[    43.730] (II) modesetting: Driver for Modesetting Kernel Drivers: kms
[    43.730] (II) FBDEV: driver for framebuffer: fbdev
[    43.730] (II) VESA: driver for VESA chipsets: vesa
[    43.740] (WW) Falling back to old probe method for modesetting
[    43.741] (WW) Falling back to old probe method for fbdev
[    43.741] (II) Loading sub module "fbdevhw"
[    43.741] (II) LoadModule: "fbdevhw"
[    43.741] (II) Loading /usr/lib/xorg/modules/libfbdevhw.so
[    43.741] (II) Module fbdevhw: vendor="X.Org Foundation"
[    43.741] 	compiled for 1.21.1.8, module version = 0.0.2
[    43.741] 	ABI class: X.Org Video Driver, version 25.2
[    43.741] (II) AMDGPU(0): Creating default Display subsection in Screen section
	"Default Screen Section" for depth/fbbpp 24/32
[    43.741] (==) AMDGPU(0): Depth 24, (--) framebuffer bpp 32
[    43.741] (II) AMDGPU(0): Pixel depth = 24 bits stored in 4 bytes (32 bpp pixmaps)
[    43.741] (==) AMDGPU(0): Default visual is TrueColor
[    43.741] (II) Applying OutputClass "AMDgpu" options to /dev/dri/card0
[    43.741] (==) AMDGPU(0): RGB weight 888
[    43.741] (II) AMDGPU(0): Using 8 bits per RGB (8 bit DAC)
[    43.741] (--) AMDGPU(0): Chipset: "AMD Radeon RX 6700 XT" (ChipID = 0x73df)
[    43.741] (II) Loading sub module "fb"
[    43.741] (II) LoadModule: "fb"
[    43.742] (II) Module "fb" already built-in
[    43.742] (II) Loading sub module "dri2"
[    43.742] (II) LoadModule: "dri2"
[    43.742] (II) Module "dri2" already built-in
[    43.800] (II) Loading sub module "glamoregl"
[    43.800] (II) LoadModule: "glamoregl"
[    43.800] (II) Loading /usr/lib/xorg/modules/libglamoregl.so
[    43.806] (II) Module glamoregl: vendor="X.Org Foundation"
[    43.806] 	compiled for 1.21.1.8, module version = 1.0.1
[    43.806] 	ABI class: X.Org ANSI C Emulation, version 0.4
[    43.813] (EE) 
[    43.813] (EE) Backtrace:
[    43.813] (EE) 0: /usr/lib/xorg/Xorg (OsLookupColor+0x14d) [0x55c4625151ed]
[    43.814] (EE) 1: /lib/x86_64-linux-gnu/libc.so.6 (__sigaction+0x40) [0x7fdca3e5a510]
[    43.814] (EE) 2: /usr/local/lib/x86_64-linux-gnu/libglapi.so.0 (_glapi_add_dispatch+0x82) [0x7fdca3b152f2]
[    43.815] (EE) 3: /usr/local/lib/x86_64-linux-gnu/dri/radeonsi_dri.so (__driDriverGetExtensions_d3d12+0x7a0bb) [0x7fdca153229b]
[    43.815] (EE) 4: /lib/x86_64-linux-gnu/libc.so.6 (pthread_mutexattr_settype+0x107) [0x7fdca3eab267]
[    43.815] (EE) 5: /usr/local/lib/x86_64-linux-gnu/dri/radeonsi_dri.so (__driDriverGetExtensions_d3d12+0x1e97ca) [0x7fdca16a19aa]
[    43.815] (EE) 6: /usr/local/lib/x86_64-linux-gnu/dri/radeonsi_dri.so (__driDriverGetExtensions_d3d12+0xca31c) [0x7fdca15824fc]
[    43.816] (EE) 7: /usr/local/lib/x86_64-linux-gnu/dri/radeonsi_dri.so (__driDriverGetExtensions_d3d12+0x64fe) [0x7fdca14be6de]
[    43.816] (EE) 8: /usr/local/lib/x86_64-linux-gnu/dri/radeonsi_dri.so (__driDriverGetExtensions_d3d12+0xa02f) [0x7fdca14c220f]
[    43.816] (EE) 9: /usr/local/lib/x86_64-linux-gnu/libEGL.so.1 (dri2_create_context+0x2c4) [0x7fdc9a096c34]
[    43.816] (EE) 10: /usr/local/lib/x86_64-linux-gnu/libEGL.so.1 (eglCreateContext+0xc0) [0x7fdc9a08b6f0]
[    43.816] (EE) 11: /usr/lib/xorg/modules/libglamoregl.so (glamor_egl_init+0x2a9) [0x7fdca1301679]
[    43.817] (EE) unw_get_proc_name failed: no unwind info found [-10]
[    43.817] (EE) 12: /usr/lib/xorg/modules/drivers/amdgpu_drv.so (?+0x0) [0x7fdca39190b9]
[    43.817] (EE) unw_get_proc_name failed: no unwind info found [-10]
[    43.817] (EE) 13: /usr/lib/xorg/modules/drivers/amdgpu_drv.so (?+0x0) [0x7fdca390e25a]
[    43.817] (EE) 14: /usr/lib/xorg/Xorg (InitOutput+0x95a) [0x55c4623e38ba]
[    43.817] (EE) 15: /usr/lib/xorg/Xorg (InitFonts+0x1ce) [0x55c4623a3fce]
[    43.818] (EE) 16: /lib/x86_64-linux-gnu/libc.so.6 (__libc_init_first+0x8a) [0x7fdca3e456ca]
[    43.818] (EE) 17: /lib/x86_64-linux-gnu/libc.so.6 (__libc_start_main+0x85) [0x7fdca3e45785]
[    43.818] (EE) 18: /usr/lib/xorg/Xorg (_start+0x21) [0x55c46238d1e1]
[    43.818] (EE) 
[    43.818] (EE) Segmentation fault at address 0x0
[    43.818] (EE) 
Fatal server error:
[    43.819] (EE) Caught signal 11 (Segmentation fault). Server aborting
[    43.819] (EE) 
[    43.819] (EE) 
Please consult the The X.Org Foundation support 
	 at http://wiki.x.org
 for help. 
[    43.819] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information.
[    43.819] (EE) 
[    43.846] (EE) Server terminated with error (1). Closing log file.

"nomodeset" kind of helps. Slim shows up as usual but I get all windows without window frames (xfwm4 instant crash and core dump) and some other numerous issues...

Any thoughts?

+ Actually I can show the upgrade history if its needed.
+ Adding "mate-desktop-environment" did resolve windows usabilty problem still I have to use "nomodeset".

Last edited by mr12345 (2023-09-21 17:49:21)

mr12345 · 2023-09-22 15:38:47

Switched to excalibur and downgraded Mesa 23.2.0RC3 -> 23.1.7. Bunch of the installed apps were removed along with it. Now it seems to be work fine. I will be checking the release number of Mesa next time before installing any updates. RCs are not so good I guess.

mr12345 · 2023-09-29 11:40:40

Just a little add up.
Today I had tripped just the same way again! 🤡
I tried to run an old OpenGL i386 game. The console thrown an error like “libgl.so.1 not found”. Not a big deal, I thought (my brain definitely was hanged up at that moment), and thrown back the “sudo apt install libgl1-mesa-dri:i386 libgl1:i386”. After that, the game performed fine, and then I thought about apt's recommendation to reboot… And guess what… 🤦 I literally had upgraded back to broken Mesa 23.2.0RC3-3 which is today on the Excalibur! The 23.1.7 is no longer available…
Having in mind that the Ceres has got 23.2.0RC4-1 today, I thought it's worth of a try, switched back to it and upgraded. But no luck so far. Still nothing but the recovery console.
What did save the situation? I had built Mesa from the source. Acknowledgments to the great Mesa folks! I am using 23.2.1, and it works like a charm.

Last edited by mr12345 (2023-09-29 11:48:02)

mr12345 · 2023-12-15 09:22:03

Today I had almost the same issue again with 23.3.1-1 right after my Devuan-Ceres setup upgrade. To restore my desktop, I had to build Mesa from the source again. I did not expect the non-RC release would be broken, at least for me.
I have not changed my hardware a bit, it is still the same.
On xfwm4 run attempt, I was getting this:

$ xfwm4
libGL error: No matching fbConfigs or visuals found
libGL error: failed to load driver: radeonsi
libGL error: No matching fbConfigs or visuals found
libGL error: failed to load driver: radeonsi
libGL error: No matching fbConfigs or visuals found
libGL error: failed to load driver: swrast

mr12345 · 2023-12-16 19:08:42

23.3.1-3 is better, thanks. Still, sometimes there are pink glitches (pink frames) in Firefox video playback.
Whereas Chromium just crashes upon start. Same thing I had with Mesa 23.3.1 which I built from the source:

[6419:6419:1216/213853.439618:ERROR:gpu_process_host.cc(990)] GPU process exited unexpectedly: exit_code=139
Cannot find target for triple amdgcn-- Unable to find target for this triple (no targets are registered)
[6419:6419:1216/213853.542270:ERROR:gpu_process_host.cc(990)] GPU process exited unexpectedly: exit_code=139
[6419:6419:1216/213853.542291:FATAL:gpu_data_manager_impl_private.cc(448)] GPU process isn't usable. Goodbye.
Trace/breakpoint trap (core dumped)

I did clean up the "/home/user/.config/chromium/Default/GPUCache/". It sometimes helped in the past, but not today.
AFAIK, “FreeTube” and “Session messenger” got Chromium under their hood. Their GUI is doing well. I have not even touched their GPUCache/ .

Last edited by mr12345 (2023-12-16 19:25:53)

mr12345 · 2023-12-24 14:33:44

As soon as I removed “mesa-VA-drivers” package, Firefox and Chromium issues have been resolved, sort of.
I have got the clue on the Chromium`s GitHub page, when I saw "No Hardware Acceleration Video Decode under AMD (vaapi is broken)".
Somehow I had not “mesa-VA-drivers” installed before 2023-12-15. This is why I did not know VA API is broken.

+ I did upgrade to 23.3.1-4 before “mesa-VA-drivers” removal.

Last edited by mr12345 (2023-12-24 14:48:10)

GlennW · 2023-12-24 22:30:35

Hi, thanks for taking the time to report on those issues.

I enjoy the read. rc = release candidate, and experimental is kinda the same but on a distro level.

I used to surf the caldron of magaiea and cooker of mandriva before that...

I found it quite exciting at times.

All the best.

mr12345 · 2023-12-25 19:37:03

Thanks, GlennW. Glad to see my tiny blog is of help. I like to explore Linux too. All the best to you too!
About the non-RC, my first thought was the 23.3.1 is broken at all, and I would need to downgrade back to 23.2.1. I was a bit stunned after I successfully built 23.3.1 from the source. I was not sure if it is worth a try at all…

Back to business

As soon as I removed “mesa-VA-drivers” package, I realized that HW FFmpeg VA API transcoding which I discovered about a week ago and like very much was not available anymore. I did reinstallation and continue to search.
First, I found Firefox “media.rdd-ffmpeg.enabled” flag and toggled it to false in the about:config. This corrected “pink frames”. I thought it would brake YouTube video playback, actually it did not. 1440p60 runs fine with CPU load about 5-10%. I am down with it.
I almost despaired to find anything to prop up Chromium, and then it appeared in my view, shiny and efficient! XD

LIBVA_DRIVER_NAME="vdpau" ./chrome

This seems is good enough too if you do not want to touch flags:

LIBVA_DRIVER_NAME="vdpau" ./librewolf

Despite Firefox runtime errors:

Crash Annotation GraphicsCriticalError: |[0][GFX1-]: vaapitest: ERROR (t=0.382356) [GFX1-]: vaapitest: ERROR
Crash Annotation GraphicsCriticalError: |[0][GFX1-]: vaapitest: ERROR (t=0.382356) |[1][GFX1-]: vaapitest: VA-API test failed: failed to initialise VAAPI connection.
 (t=0.382356) [GFX1-]: vaapitest: VA-API test failed: failed to initialise VAAPI connection

Last edited by mr12345 (2023-12-25 19:49:16)

mr12345 · 2024-01-10 14:24:33

Yes! Couple of days ago I have had a tough upgrade of my Ceres too!
First, I noticed that my system can not find modprobe. With help of this Forum search, I found this. But it was not enough to install a fresh kernel completely, so I had to read more of the installation output. lvm2, dmsetup (both at /usr/share/initramfs-tools/hooks/) were missing some of their rules (I took a look at their scripts). So I made more links:

ln -s /usr/lib/udev/rules.d/56-lvm.rules /lib/udev/rules.d/56-lvm.rules
ln -s /usr/lib/udev/rules.d/69-lvm.rules /lib/udev/rules.d/69-lvm.rules
ln -s /usr/lib/udev/rules.d/55-dm.rules /lib/udev/rules.d/55-dm.rules
ln -s /usr/lib/udev/rules.d/60-persistent-storage-dm.rules /lib/udev/rules.d/60-persistent-storage-dm.rules
ln -s /usr/lib/udev/rules.d/95-dm-notify.rules /lib/udev/rules.d/95-dm-notify.rules

Ok, my kernel has been installed successfully! But no graphics so far.
This time boot log advised me that grep, egrep, sed were not at /bin/. Actually, I found them at /usr/bin/ and created a bit more links.
And at last it was enough to boot up to graphic mode!

I believe evading usrmerge will have its consequences sometime in the future… Oh, well…

mr12345 · 2024-02-24 07:33:54

And another broken Mesa 3D build (libgl1-mesa-dri 24.0.1-1). Well, at least for me and my RX 6700XT. As usual, after a simple upgrade (not that simple actually; several more links have been added) my desktop kind of came up, but xfwm4 had not. Again the source (same 24.0.1), compilation, reboot and now it is all fine.

Although 24.0.1-1 is ok for QEMU/KVM and UHD 630.

The officially official Devuan Forum!

#1 2023-09-21 13:35:54

[SOLVED] Graphics does not come up after Ceres update/upgrade.

#2 2023-09-22 15:38:47

Re: [SOLVED] Graphics does not come up after Ceres update/upgrade.

#3 2023-09-29 11:40:40

Re: [SOLVED] Graphics does not come up after Ceres update/upgrade.

#4 2023-12-15 09:22:03

Re: [SOLVED] Graphics does not come up after Ceres update/upgrade.

#5 2023-12-16 19:08:42

Re: [SOLVED] Graphics does not come up after Ceres update/upgrade.

#6 2023-12-24 14:33:44

Re: [SOLVED] Graphics does not come up after Ceres update/upgrade.

#7 2023-12-24 22:30:35

Re: [SOLVED] Graphics does not come up after Ceres update/upgrade.

#8 2023-12-25 19:37:03

Re: [SOLVED] Graphics does not come up after Ceres update/upgrade.

#9 2024-01-10 14:24:33

Re: [SOLVED] Graphics does not come up after Ceres update/upgrade.

#10 2024-02-24 07:33:54

Re: [SOLVED] Graphics does not come up after Ceres update/upgrade.

Board footer