Re: [gentoo-user] SOLVED Can't get the GUI to stay up for more than a minute or so before crashing

[Michael <confabulate@kintzios.com>]

[-- Attachment #1: Type: text/plain, Size: 7027 bytes --]

On Wednesday, 26 June 2024 01:28:47 BST Dale wrote:
> Michael wrote:

> > The above message indicates the same problem you had experienced before
> > you
> > reinstalled.  The monitor is not sending its EDID table, or the card can't
> > read it.
> > 
> > Your Xorg sets a default dummy resolution of 640 x 480, because it can't
> > find anything connected to the card.
> > 
> > Things I would try, until someone who can grok nvidia contributes better
> > ideas:
> > 
> > Eliminate the hardware being the cause of the problem, e.g.: try a
> > different cable, different monitor, then try the same card (with same
> > drivers and same kernel settings) on your other PC.  If this proves
> > there's nothing wrong with the cable, card, or kernel settings:
> > 
> > 1. Try different ports and restart display-manager each time.
> > 
> > 2. Add these two lines at the bottom of /usr/share/sddm/scripts/Xsetup:
> > 
> > xrandr --setprovideroutputsource modesetting NVIDIA-0
> > xrandr --auto
> > 
> > Again restart display-manager.
> > 
> > 3. Add a file /etc/X11/xorg.conf.d/20nvidia.conf
> > 
> > Section "Device"
> > 
> >    Identifier  "nvidia"
> >    Driver      "nvidia"
> >    BusID       "PCI:9:0:0"
> >    Option "UseEDID" "false" ## Try this too ##
> > 
> > EndSection
> > 
> > Again restart display-manager.
> > 
> > Every time you try a setting and it doesn't produce the goods, revert it
> > before you try the next thing.  Make notes and keep an eye on your logs in
> > case you spot a difference.
> > 
> > If none of these tweaks work, then you can try capturing the EDID table
> > and creating a file for the card to load.
[snip ...]

> I was
> even thinking of moving my main rig monitor to the new rig and see what
> it did.  I'd already tried a different card so didn't see any need in
> repeating that.  Then I had a thought.  Why is it saying port DP-3?  Why
> is it not port DP-0?

Your PC indicated DFP-3 was what it had booted at - from your Xorg.0.log:

[    44.311] (--) NVIDIA(0): Valid display device(s) on GPU-0 at PCI:9:0:0
[    44.311] (--) NVIDIA(0):     DFP-0
[    44.311] (--) NVIDIA(0):     DFP-1
[    44.311] (--) NVIDIA(0):     DFP-2
[    44.311] (--) NVIDIA(0):     DFP-3 (boot)
[    44.311] (--) NVIDIA(0):     DFP-4
[    44.311] (--) NVIDIA(0):     DFP-5
[    44.311] (--) NVIDIA(0):     DFP-6
[    44.311] (--) NVIDIA(0):     DFP-7

which is the display device connector type nvidia identifies the monitor being 
connected to.  However, then it prints this discouraging message:

[    44.312] (--) NVIDIA(GPU-0): 
[    44.332] (--) NVIDIA(GPU-0): DFP-3: disconnected  <== This ===
[    44.332] (--) NVIDIA(GPU-0): DFP-3: Internal TMDS
[    44.332] (--) NVIDIA(GPU-0): DFP-3: 165.0 MHz maximum pixel clock
[    44.332] (--) NVIDIA(GPU-0): 


> I thought the first port was the one on the
> bottom.  Turns out, the top port is the first one.  So, I moved the
> cable to the first port, DP-0.

I thought you had already tried this prior to reinstalling, when I had 
suggested to try different ports.


> I booted the rig up, started DM, got the
> login screen as usual and guess what was next, a complete desktop.  I
> changed it to not power off or switch to a screensaver so that it would
> stay on and I could keep a eye on it. I heated up supper, ate, typing
> this reply and it is still running, in 1080P no less. 

YES!  :-D


> Now tell me this, why would it not work on DP-3 or DP-2 when I tried
> those earlier on?  Does one always have to have a monitor connected to
> DP-0 first then others as monitors are added? 

It may have something to do with auto-detecting PNP display devices, like 
DisplayPort monitor devices.  There is a HPD pin (Hot Plug Detect) on the DP, 
which lets the card know if a monitor turns off.  This seems to cause the 
card's driver to detect the display as "disconnected", which then disables the 
port.

The question is why would the monitor turn off.  Well, it might be taking too 
long for the card to walk from DP-1 to DP-3, by which time the monitor has 
gone to sleep to save energy.  If the monitor is on DP-1, then it doesn't get 
a chance to do this.

Alternatively, the Quadro P1000 video card being a 'pro' graphics card may 
have been designed with the assumption a monitor (the primary monitor) is 
*always* connected on the first port, or else the PC is configured as a 
headless server - I don't know really.

I think if you capture and feed manually the EDID table to the card's driver, 
it may work differently - but again, I have no experience with Nvidia.  By 
accident or good fortune I've always had 'linux-friendly' AMD-Radeon cards on 
my PCs.

One thing I have noticed with my DisplayPort monitor, it needs to be powered 
on while the PC boots up/shuts down.  If the monitor is switched off it will 
not get detected after boot and also the shutdown process is cancelled.  :-/


> Now comes the next question.  To move just KDE stuff over, desktop
> settings and such.  ~/.local and .config.  Are those the big ones?

Yes, if we're talking about plasma and kde applications.

You'll also need:

.mozilla
.ssh
.gnupg
.gkrellm2

and others, if you want to keep their settings the same across PCs.

> Also, I have a .kde4 directory, that's no longer used right?  I think it
> died ages ago.  I forgot all about that thing.

Last time kde4 was touched on my system was in 2022 - perhaps I changed some 
theme/panel colors or some such.  I thought it was defunct as a directory for 
some year now:

~ $ ls -la .kde4/share/apps/color-schemes/
total 23
drwxr-xr-x 2 michael michael 3452 Oct  1  2022 .
drwxr-xr-x 3 michael michael 3452 Jun 14  2019 ..
-rw-r--r-- 1 michael michael 3519 Jun 16  2019 Breeze.colors
-rw-r--r-- 1 michael michael 3806 Jun 15  2019 BreezeDark.colors
-rw-r--r-- 1 michael michael 4474 Oct  1  2022 Oxygen.colors

~ $ ls -la .kde4/share/config/
total 11
drwxr-xr-x 2 michael michael 3452 Oct  1  2022 .
drwxr-xr-x 4 michael michael 3452 Jun 14  2019 ..
-rw------- 1 michael michael 3403 Oct  1  2022 kdeglobals


> I'll copy the other
> stuff over at some point but just want to play with the big stuff at the
> moment. 
> 
> In your list, #1 would have been the fix.  It also turns out, it was
> me.  I plugged the cable in the wrong port.  No idea why everything else
> worked fine tho.  All the boot media worked just fine.  This is a large
> thread over something so simple.  ;-) 

Well, if I had more experience with Nvidia, or if you tried each and every 
port as I erroneously assumed you had done, it may not have taken this long!  
;-)


> Thanks so much for all the help.  The main rig is still sitting there at
> 1080P waiting on me.  Finally, after over $1,000 spent, days of
> installing, twice, and a lot of testing, a working computer.  :-D  :-D 
> 
> Dale
> 
> :-)  :-) 

Could you please send me your Xorg.0.log again with a working and powered up 
monitor, off-list to keep the noise down.  I'd be interested to see what it 
reports now the monitor is connected to DP-1 and initialized correctly.

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]