Solved if this is a kernel problem, how and where to find it?

Solved issue
Y

yaximik

New Member
Joined
Nov 19, 2024
Messages
5
Reaction score
3
Credits
64
Hi everyone,
Recently, for the first time I installed RHEL9.4 (kernel 5.14.0-427.42.1.el9_4.x86_64) to replace my Scientific LInux 7.x that for many years served me greatly, but recently reached the end-of-life point. The install was totally successful, but the scheduled upgrade to 9.5 (kernel 5.14.0-503.11.1.el9_5.x86_64) proceeded only at the fallback resolution of 640x480 and max resolution after boot was only 1024x768 instead of normal 1920x1200 with Nvidia NVS300. The hardware is the same and works normally when booting to RHEL9.4 when at boot kernel 5.14.0-427.42.1.el9_4.x86_64 is selected. The low resolution issue is only observed when booting to RHEL9.5 by selecting either 5.14.0-503.11.1.el9_5.x86_64 or 5.14.0-503.14.1.el9_5.x86_64 kernel, and this is reproducible as sunrise and sunset.

Just released standalone RHEL9.5 installer produced the same outcome like the 9.5 update. The fallback resolution was dropped to right after the installation started. The same fallback observed with Centos Stream 9 latest. Both standalone 9.5 and CS9 did not allow use of the recommended graphic installation mode, because the lower part of destination disk screens were beyond reach and could not be scrolled to. In the text mode, VNC could not be used because an outside VNC client become non-responsive right after connection was established. Perhaps this was a consequence of the problem with video, as the RHEL9.4 normal install in the graphic mode had the normal usable VNC mode. While CS9 could be installed in the plain text mode and booted afterwards, although at the low resolution only, the RHEL9.5 install could not be completed even in the text mode because I could not figure out how to specify in the text mode the required authorized access to the RHEL CDN installation source, which is only available in the graphic mode or over VNC.

I extracted /tmp/syslogs by interrupting all three installs in the Rescue mode, and at first it looks like indeed a nouveau driver issue.

Normal install RHEL9.4
Code: 
[yaximik@g5nnjn1 94boot]$ more tmp/syslog | grep nouveau
04:04:30,656 INFO kernel:nouveau 0000:04:00.0: vgaarb: deactivate vga console
04:04:30,657 INFO kernel:nouveau 0000:04:00.0: NVIDIA GT218 (0a8c00b1)
04:04:30,780 INFO kernel:nouveau 0000:04:00.0: bios: version 70.18.83.00.09
04:04:30,781 INFO kernel:nouveau 0000:04:00.0: fb: 512 MiB DDR3
04:04:30,872 INFO kernel:nouveau 0000:04:00.0: DRM: VRAM: 512 MiB
04:04:30,872 INFO kernel:nouveau 0000:04:00.0: DRM: GART: 1048576 MiB
04:04:30,872 INFO kernel:nouveau 0000:04:00.0: DRM: TMDS table version 2.0
04:04:30,872 INFO kernel:nouveau 0000:04:00.0: DRM: DCB version 4.0
04:04:30,872 INFO kernel:nouveau 0000:04:00.0: DRM: DCB outp 00: 02000360 00000000
04:04:30,873 INFO kernel:nouveau 0000:04:00.0: DRM: DCB outp 01: 02000362 00020010
04:04:30,873 INFO kernel:nouveau 0000:04:00.0: DRM: DCB outp 02: 028003a6 0f220010
04:04:30,873 INFO kernel:nouveau 0000:04:00.0: DRM: DCB outp 03: 01011380 00000000
04:04:30,873 INFO kernel:nouveau 0000:04:00.0: DRM: DCB outp 04: 08011382 00020010
04:04:30,873 INFO kernel:nouveau 0000:04:00.0: DRM: DCB outp 05: 088113c6 0f220010
04:04:30,874 INFO kernel:nouveau 0000:04:00.0: DRM: DCB conn 00: 00101064
04:04:30,874 INFO kernel:nouveau 0000:04:00.0: DRM: DCB conn 01: 00202165
04:04:30,885 INFO kernel:nouveau 0000:04:00.0: DRM: MM: using COPY for buffer copies
04:04:30,902 INFO kernel:[drm] Initialized nouveau 1.4.0 20120801 for 0000:04:00.0 on minor 0
04:04:30,965 INFO kernel:fbcon: nouveaudrmfb (fb0) is primary device
04:04:31,073 INFO kernel:nouveau 0000:04:00.0: [drm] fb0: nouveaudrmfb frame buffer device
[yaximik@g5nnjn1 94boot]$
Failed RHEL9.5 install
Code: 
[yaximik@g5nnjn1 95boot]$ more tmp/syslog | grep nouveau
22:39:33,969 INFO kernel:nouveau 0000:04:00.0: vgaarb: deactivate vga console
22:39:34,343 INFO kernel:nouveau 0000:04:00.0: NVIDIA GT218 (0a8c00b1)
22:39:34,343 INFO kernel:nouveau 0000:04:00.0: bios: version 70.18.83.00.09
22:39:34,343 INFO kernel:nouveau 0000:04:00.0: fb: 512 MiB DDR3
22:39:34,344 INFO kernel:nouveau 0000:04:00.0: DRM: VRAM: 512 MiB
22:39:34,344 INFO kernel:nouveau 0000:04:00.0: DRM: GART: 1048576 MiB
22:39:34,344 INFO kernel:nouveau 0000:04:00.0: DRM: TMDS table version 2.0
22:39:34,345 INFO kernel:nouveau 0000:04:00.0: DRM: MM: using COPY for buffer copies
22:39:34,345 INFO kernel:[drm] Initialized nouveau 1.4.0 20120801 for 0000:04:00.0 on minor 0
22:39:34,345 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000000:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
22:39:34,345 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000003:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
22:39:34,346 INFO kernel:fbcon: nouveaudrmfb (fb0) is primary device
22:39:34,346 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000000:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
22:39:34,346 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000003:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
22:39:34,346 INFO kernel:nouveau 0000:04:00.0: [drm] fb0: nouveaudrmfb frame buffer device
22:39:34,379 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000000:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
22:39:34,412 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000003:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
22:39:34,454 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000000:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
22:39:34,479 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000003:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
22:39:34,523 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000000:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
22:39:34,545 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000003:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
[yaximik@g5nnjn1 95boot]$
Faulty CS9 install:
Code: 
[yaximik@g5nnjn1 CS9boot]$ more tmp/syslog | grep nouveau
02:45:16,117 INFO kernel:nouveau 0000:04:00.0: vgaarb: deactivate vga console
02:45:16,477 INFO kernel:nouveau 0000:04:00.0: NVIDIA GT218 (0a8c00b1)
02:45:16,489 INFO kernel:nouveau 0000:04:00.0: bios: version 70.18.83.00.09
02:45:16,489 INFO kernel:nouveau 0000:04:00.0: fb: 512 MiB DDR3
02:45:16,489 INFO kernel:nouveau 0000:04:00.0: DRM: VRAM: 512 MiB
02:45:16,489 INFO kernel:nouveau 0000:04:00.0: DRM: GART: 1048576 MiB
02:45:16,490 INFO kernel:nouveau 0000:04:00.0: DRM: TMDS table version 2.0
02:45:16,490 INFO kernel:nouveau 0000:04:00.0: DRM: MM: using COPY for buffer copies
02:45:16,490 INFO kernel:[drm] Initialized nouveau 1.4.0 20120801 for 0000:04:00.0 on minor 0
02:45:16,490 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000000:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
02:45:16,490 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000003:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
02:45:16,490 INFO kernel:fbcon: nouveaudrmfb (fb0) is primary device
02:45:16,490 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000000:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
02:45:16,491 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000003:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
02:45:16,491 INFO kernel:nouveau 0000:04:00.0: [drm] fb0: nouveaudrmfb frame buffer device
02:45:16,504 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000000:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
02:45:16,537 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000003:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
02:45:16,570 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000000:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
02:45:16,604 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000003:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
02:45:16,637 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000000:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
02:45:16,670 ERR kernel:nouveau 0000:04:00.0: DRM: [DRM/00000003:kmsOutp] [LOAD_DETECT data:00000154] load:07 (ret:7)
[yaximik@g5nnjn1 CS9boot]$

However, all three installations use exactly the same nouveau driver! If this is a problem with kernel or settings, how to get to that?

Could someone with the needed expertise help to find what is the problem and is there a way to fix it? I also extracted dmesg -k outputs in all three aborted installs, but dmesg -k | grep nouveau extracts are essentially the same.
 


You probably need to use xrandr
 
Tried, xrandr shows and allows to specify only available resolutions. If for 9.4 the set is from 640x480 to 1920x1200, for upgraded 9.5 it is only 640x480 to 1024x768.
 
I have found that video support in RH and Fedora perhaps most distros because it seems to be kernel related, has diminished. I had a beautiful server set up and after upgrade to newer OS version I had no video support. VNC relied on the support so it showed nothing but anydesk worked and I was able to access it with that. The problem is video support being obsoleted for my card. You may have the same issue. Might fix it by putting in a new video card. Something cheap is best.
 
The NVIDIA GT218 is now 14 years old, if nouveau drivers don't work you may have to look for legacy ones from Nvidia
 
As m'colleague @Brickwizard says, I suspect support for the GT218 has been dropped. I suspect it's "upgrade time"! My own GT710 is no longer supported by nVidia - support quit after the 470-series drivers - but nouveau still supports it quite happily (most of my Pups are on the 6.1-series LTS kernels, and these don't have a problem with it).

On some of my Puppies, I run the nouveau driver.....on others, the final supported driver; in this case, 470.256.02. For what I use the thing for - mostly video-rendering - I don't NEED the very newest, bang up-to-date drivers anyway.

As @APTI says, something cheap'n'cheerful would suffice. If you're not a gamer, or not too fussed about using it, I'd take a look at the GT1030. These are still quite reasonably priced, use a Pascal-gen chip, frequently paired with GDDR5.....and for most ordinary users provide a decent performance boost over onboard graphics. Nothing special, of course, but good enough for indie games (f that's your bag).


Mike. ;)
 
@yaximik wrote:
Could someone with the needed expertise help to find what is the problem and is there a way to fix it?
Click to expand...

The GT218 card works as intended with the nouveau driver on a machine here, as shown in the data output in the spoiler. The machine runs a much more recent kernel than the 5 series, so it's clear the kernel still supports nouveau for that card. A check of the nouveau manpage confirms that the GT218 is supported by the driver.

Whilst I'm unable to help on the RHEL front, since this machine here runs debian, the best I can do is show the details that are associated with the nouveau driver so that one can see the installations that have it working as intended at the resolution of 1920x1080 and perhaps compare the details with the equivalents in RHEL.

See the details in the spoiler.

Code: 
$ lspci -nnk | grep -i vga -A3
01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GT218 [GeForce
        Subsystem: ASUSTeK Computer Inc. Device [1043:8354]
        Kernel driver in use: nouveau
        Kernel modules: nouveau


Output of xrandr:
Code: 
$ xrandr
Screen 0: minimum 320 x 200, current 1920 x 1080, maximum 8192 x 8192
DVI-I-1 disconnected primary (normal left inverted right x axis y axis)
HDMI-1 connected 1920x1080+0+0 (normal left inverted right x axis y axis) 47
   1920x1080     60.00*+  59.96    50.00    59.94    59.93
   1680x1050     59.95    59.88
   1400x1050     59.98
<snip>

Kernel version:
Code: 
$ uname -r
6.10.11-amd64

List of loaded modules:
Code: 
$ lsmod | grep nouveau
nouveau              3088384  7
gpu_sched              65536  1 nouveau
drm_gpuvm              45056  1 nouveau
drm_exec               12288  2 drm_gpuvm,nouveau
mxm_wmi                12288  1 nouveau
drm_ttm_helper         12288  1 nouveau
ttm                   102400  2 drm_ttm_helper,nouveau
i2c_algo_bit           12288  1 nouveau
drm_display_helper    262144  1 nouveau
drm_kms_helper        253952  3 drm_display_helper,nouveau
drm                   749568  13 gpu_sched,drm_kms_helper,drm_exec,drm_gpuvm,drm_display_helper,drm_ttm_helper,ttm,nouveau
video                  77824  1 nouveau
wmi                    28672  4 video,intel_wmi_thunderbolt,mxm_wmi,nouveau
button                 24576  1 nouveau


X driver and lib installed:
Code: 
$ apt list --installed | grep nouveau
libdrm-nouveau2/testing,now 2.4.121-2 amd64 [installed,automatic]
xserver-xorg-video-nouveau/testing,now 1:1.0.17-3 amd64 [installed,automatic]

The mesa drivers and associated packages:
Code: 
$ apt list --installed |  grep mesa
libegl-mesa0/testing,now 24.1.3-2 amd64 [installed,automatic]
libgl1-mesa-dev/testing,now 24.1.3-2 amd64 [installed,automatic]
libgl1-mesa-dri/testing,now 24.1.3-2 amd64 [installed,automatic]
libglapi-mesa/testing,now 24.1.3-2 amd64 [installed,automatic]
libglu1-mesa-dev/testing,now 9.0.2-1.1+b1 amd64 [installed,automatic]
libglu1-mesa/testing,now 9.0.2-1.1+b1 amd64 [installed,automatic]
libglx-mesa0/testing,now 24.1.3-2 amd64 [installed,automatic]
mesa-utils-bin/testing,now 9.0.0-2 amd64 [installed,automatic]
mesa-utils/testing,now 9.0.0-2 amd64 [installed,automatic]
mesa-va-drivers/testing,now 24.1.3-2 amd64 [installed,automatic]
mesa-vdpau-drivers/testing,now 24.1.3-2 amd64 [installed,automatic]
mesa-vulkan-drivers/testing,now 24.1.3-2 amd64 [installed,automatic]



The vulkan packages:
Code: 
$ apt list --installed | grep vulkan
libvulkan1/testing,now 1.3.283.0-1 amd64 [installed,automatic]
mesa-vulkan-drivers/testing,now 24.1.3-2 amd64 [installed,automatic]
vulkan-tools/testing,now 1.3.283.0+dfsg1-1 amd64 [installed]

The openGL info:
Code: 
$ glxinfo -B | head -n 10
name of display: :0
display: :0  screen: 0
direct rendering: Yes
Extended renderer info (GLX_MESA_query_renderer):
    Vendor: Mesa (0x10de)
    Device: NVA8 (0xa65)
    Version: 24.2.4
    Accelerated: yes
    Video memory: 1015MB
    Unified memory: no

Info on nouveau in Xorg log file:
Code: 
$ grep nouveau Xorg.0.log
[    40.610] (II) modeset(0): [DRI2]   DRI driver: nouveau
[    40.610] (II) modeset(0): [DRI2]   VDPAU driver: nouveau
[    40.634] (II) AIGLX: Loaded and initialized nouveau


Error search in Xorg log file:
Code: 
[ben@fen ~/.local/share/xorg]$ grep EE Xorg.0.log
[    37.681] Current Operating System: Linux fen 6.10.11-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.10.11-1 (2024-09-22) x86_64
        (WW) warning, (EE) error, (NI) not implemented, (??) unknown.
[    40.612] (II) Initializing extension MIT-SCREEN-SAVER
 
Last edited:
Instead of using vnc install xrdp for remote connection, you can install it by activating the EPEL repo.
Code: 
xrdp.x86_64 : Open source remote desktop protocol (RDP) server
Then als install the following package.
Code: 
xorgxrdp.x86_64 : Implementation of xrdp backend as Xorg modules
This package allows you to connect using an Xorg session instead of an Xvnc session. Reboot, select Xorg session during login. See if that is an improvement?
 
Last edited:
Brickwizard said:
The NVIDIA GT218 is now 14 years old, if nouveau drivers don't work you may have to look for legacy ones from Nvidia
Click to expand...
The problem is, the legacy drivers for some older NVidia cards are no longer compiled for recent versions of the kernel. Essentially, they are no longer supported. So you have to rely solely on the nouveau drivers.
But for some of the really old cards, even the nouvveau drivers don't work properly.

I bought my old Windows 7 desktop PC from my employer when they decided to close our office earlier this year. (Due to most of us working 100% from home since lockdown in 2020).
My old W7 PC is a half decent i7, with plenty of RAM. But it has a really old NVidia card in it, it also has an Intel card built into the motherboard.

A few weeks ago, I tried installing Ubuntu Studio on it. But with the Nvidia card - the nouveau drivers just don't work. All I got was a black screen. I looked into installing the NVidia's proprietary, legacy drivers for the card. But the last version of the legacy drivers to support that graphics card was compiled for something like kernel version 5.15.x. It's not compiled against anything after that. So in the end, I had to disable the NVidia card in the BIOS and use the onboard Intel chipset, which actually does work!

The onboard Intel card is fine for day to day desktop usage, browsing and watching videos. But it's not great for gaming, or anything that requires serious graphics acceleration.

So at some point, I'm going to need to buy a semi-decent graphics card. Something I haven't had to do for a REALLY long time. I've started looking at graphics cards for Linux, but there are LOADS out there to look at! :/

It's a bit crappy that some older graphics cards are no longer being supported by either the proprietary drivers or the free ones, but I guess support has to be dropped for them at some point!
 
Thanks a lot to everyone who commented and provided some thoughts, especially to @osprey for posting the spoiler. I will compare similar outputs after 9.4 and 9.5 installs if there any meaningful differences. RHEL9 does not use Xorg though and is using Wayland instead, but again what is so different between 9.4 and 9.5 that makes it to malfunction in the latter case?

The last driver recommended by Nvidia for NVS300 (GT218) is 340.1080 released in 2019, but @JasKinasis pointed to the major problem with Nvidia legacy drivers, so I try to switch as the last resort, although this might not work either. If the my NVS300 (GT218) does seem to work with the current (1.4.0 20120801) version of nouveau and the _427.42.1. kernel then what is preventing the normal operation with _503.11/14.1. kernel? It may be naïve question, but could it be that monitor model and its capabilities are not recognized somehow and this triggers the fallback resolution? In Graphic settings after boot to 9.4 the model "DELLU2412M" is shown and resolution capabilities are correctly displayed, the same shows xrandr output, but after boot to 9.5 Graphic settings show "Unknown monitor "and 640x480 only, same as xrandr.

Could someone knowledgeable comment on which system/kernel module is responsible for identification of monitor model and its capabilities?
 
@yaximik when you are sure this is solved, you can mark it as such by going to your first post, and do as follows

Near bottom left of the post click Edit - (No Prefix) - Solved

Cheers

Chris Turner
wizardfromoz
 
wizardfromoz said:
@yaximik when you are sure this is solved, you can mark it as such by going to your first post, and do as follows

Near bottom left of the post click Edit - (No Prefix) - Solved

Cheers

Chris Turner
wizardfromoz
Click to expand...
Thanks, I did look how to do that but could not find.
 
You must log in or register to reply here.

Members online

Total: 594 (members: 5, guests: 589)

Latest posts

Top