Intermittent boot problem, display is whorl of colors and unreadable

CraigD · 2020-01-26 23:54:39

My display is a TV connected via HDMI to the Arch system. I am having a intermittent but persistent problem with booting. On problem boots the display is a twisted rainbow of colors and text is difficult or impossible to read. At other times the system boots fine. Once it boots correctly it runs flawlessly until reboot even when left on for several days.

I was concerned about a failing power supply so I tested voltages with a digital voltmeter and all were well within spec but without a load.

I did manage to login on a bad boot, open a terminal session and blindly log dmesg and journalctl -b. I looked for error messages in both files and came up with this from journalctl -b. It means nothing to me and may not even be important but I am trying to figure this out.

grep error journalctl.log

Jan 26 14:52:51 Arch kernel: platform regulatory.0: Direct firmware load for regulatory.db failed with error -2
Jan 26 15:01:48 Arch ksmserver[3402]: Qt: Session management error: networkIdsList argument is NULL
Jan 26 15:01:49 Arch kwin_x11[3415]: qt.qpa.xcb: QXcbConnection: XCB error: 3 (BadWindow), sequence: 919, resource id: 33554445, major code: 3 (GetWindowAttributes), minor code: 0
Jan 26 15:01:49 Arch kwin_x11[3415]: qt.qpa.xcb: QXcbConnection: XCB error: 9 (BadDrawable), sequence: 920, resource id: 33554445, major code: 14 (GetGeometry), minor code: 0
Jan 26 15:01:50 Arch kwin_x11[3415]: qt.qpa.xcb: QXcbConnection: XCB error: 3 (BadWindow), sequence: 1630, resource id: 12582917, major code: 18 (ChangeProperty), minor code: 0
Jan 26 15:01:51 Arch kwin_x11[3415]: qt.qpa.xcb: QXcbConnection: XCB error: 3 (BadWindow), sequence: 2065, resource id: 4194338, major code: 18 (ChangeProperty), minor code: 0
Jan 26 15:01:52 Arch kwin_x11[3415]: qt.qpa.xcb: QXcbConnection: XCB error: 3 (BadWindow), sequence: 2710, resource id: 33554604, major code: 3 (GetWindowAttributes), minor code: 0

And here are the err from dmesg

grep err dmesg.log1

[    0.061503] ACPI: IRQ0 used by override.
[    0.061504] ACPI: IRQ9 used by override.
[    0.622364] ACPI: Using IOAPIC for interrupt routing
[    0.656581] ACPI: PCI Interrupt Link [LNKA] (IRQs 3 4 *5 6 7 9 10 11 12 14 15)
[    0.656652] ACPI: PCI Interrupt Link [LNKB] (IRQs 3 4 5 6 7 9 *10 11 12 14 15)
[    0.656722] ACPI: PCI Interrupt Link [LNKC] (IRQs 3 4 5 6 7 9 10 *11 12 14 15)
[    0.656792] ACPI: PCI Interrupt Link [LNKD] (IRQs 3 4 5 6 7 9 10 11 *12 14 15)
[    0.656861] ACPI: PCI Interrupt Link [LNKE] (IRQs 3 4 5 6 7 9 10 11 12 14 15) *0, disabled.
[    0.656932] ACPI: PCI Interrupt Link [LNKF] (IRQs 3 4 5 6 7 *9 10 11 12 14 15)
[    0.657002] ACPI: PCI Interrupt Link [LNK0] (IRQs 3 4 5 6 *7 9 10 11 12 14 15)
[    0.657071] ACPI: PCI Interrupt Link [LNK1] (IRQs *3 4 5 6 7 9 10 11 12 14 15)
[    0.685059] pci 0000:00:02.0: BIOS left Intel GPU interrupts enabled; disabling
[    0.863737] fbcon: Deferring console take-over
[    5.204612] e1000e 0000:07:00.0: Interrupt Throttling Rate (ints/sec) set to dynamic conservative mode
[    5.349097] i801_smbus 0000:00:1f.3: SMBus using PCI interrupt
[    6.359441] platform regulatory.0: Direct firmware load for regulatory.db failed with error -2
[  352.099754] perf: interrupt took too long (2502 > 2500), lowering kernel.perf_event_max_sample_rate to 79800
[  543.319778] perf: interrupt took too long (3137 > 3127), lowering kernel.perf_event_max_sample_rate to 63600

Craig

Last edited by CraigD (2020-01-27 00:35:14)

WorMzy · 2020-01-26 23:58:18

Please don't post partial logs: https://bbs.archlinux.org/viewtopic.php?id=57855

And please use code tags: https://wiki.archlinux.org/index.php/Co … s_and_code

CraigD · 2020-01-27 00:37:29

Sorry for my oversight.

I may misunderstand but I can't find nopaste and installation of gnopaster from the AUR failed.

git clone https://aur.archlinux.org/gnopaster.git
Cloning into 'gnopaster'...
remote: Enumerating objects: 8, done.
remote: Counting objects: 100% (8/8), done.
remote: Compressing objects: 100% (7/7), done.
remote: Total 8 (delta 1), reused 7 (delta 1)
Unpacking objects: 100% (8/8), 1.51 KiB | 774.00 KiB/s, done.
[craig@Arch ~]$ cd gnopaster/
[craig@Arch gnopaster]$ makepkg -se
==> Making package: gnopaster 0.0.9-2 (Sun 26 Jan 2020 04:27:09 PM PST)
==> Checking runtime dependencies...
==> Installing missing dependencies...
[sudo] password for craig: 
resolving dependencies...
looking for conflicting packages...

Packages (1) perl-config-simple-4.59-10

Total Download Size:   0.02 MiB
Total Installed Size:  0.06 MiB

:: Proceed with installation? [Y/n] y
:: Retrieving packages...
 perl-config-simple-4.59-10-any                                        25.3 KiB  1267 KiB/s 00:00 [#########################################################] 100%
(1/1) checking keys in keyring                                                                    [#########################################################] 100%
(1/1) checking package integrity                                                                  [#########################################################] 100%
(1/1) loading package files                                                                       [#########################################################] 100%
(1/1) checking for file conflicts                                                                 [#########################################################] 100%
(1/1) checking available disk space                                                               [#########################################################] 100%
:: Processing package changes...
(1/1) installing perl-config-simple                                                               [#########################################################] 100%
:: Running post-transaction hooks...
(1/2) Arming ConditionNeedsUpdate...
(2/2) Warn about old perl modules
==> Checking buildtime dependencies...
==> WARNING: Using existing $srcdir/ tree
==> Entering fakeroot environment...
==> Starting package()...
install: cannot stat 'gnopaster-0.0.9.pl': No such file or directory
==> ERROR: A failure occurred in package().
    Aborting...

seth · 2020-01-27 08:39:00

Did you notice that the linked post is 12 years old? Details about internet services might no longer be valid
https://wiki.archlinux.org/index.php/Pastebin

WorMzy · 2020-01-27 08:52:11

Hm, yeah some of the examples in that post might be a bit dated -- nopaste got dropped from the repos a couple of years ago because it was dead upstream.

CraigD · 2020-01-27 17:14:58

Thanks for the current info. I hope I am posting the logs correctly now.

To add another bit of info. I have always rebooted right away when the bad boot happened. This morning I was busy making coffee and other things when the system booted with the whorled screen. By the time I got back to it 10 minutes later it was normal appearing and at the login screen. Not sure what to make of that.

dmesg from current boot
http://ix.io/28vl

journalctl -b
http://ix.io/28vm

Craig

seth · 2020-01-27 18:15:09

The system boots "quiet", ie. there should be no output before SDDM at all.
Please post your xorg log, in case you've xf86-video-intel installed, remove that (and possible config files referencing the intel driver in /{etc,usr/share}/X11/xorg.conf*

CraigD · 2020-01-28 00:30:05

I removed xf86-video-intel. No reference to the file in /usr/share/X11/xorg.conf.d/*

I also noted that the whorl on abnormal boot persisted until the screen slept, but was normal on wakeup.

Here are the Xorg logs.

Xorg.0.log
http://ix.io/28xi

Xorg.0.log.old
http://ix.io/28xj

Thanks,

Craig

Last edited by CraigD (2020-01-28 01:06:07)

seth · 2020-01-28 07:05:54

CraigD wrote:

I removed xf86-video-intel.

Did that cause a change to the situation?

If not: if your remove the "quiet" parameter from the kernel command line, do you initially get a stable ouptut that only turns into a "whorl of colors" once SDDM starts?

CraigD · 2020-01-28 15:11:12

Removing xf86-video-intel made no noticeable change.

I don't recall setting up quiet/silent boot.

On boot after the bios I see as best I can tell it flashes by quickly
screen 1 Grub menu with 3 or 4 sec pause

screen 2 Starting Linux
Starting ram disc

screen 3 Starting Arch.....
Root clean...

flash and what I believe is screen 3 repeated but in a smaller font.

I can temporarily remove the quiet boot by pressing e and removing the quiet comment but it reverts to quiet on the next boot and I can't figure out how to make it permanent. But, everything displayed normal until sddm.

Also, to be clear everything proceeds as normal up to and including the above. It goes wonky when the plasma sign in screen appears. AND if I don't touch anything and wait for the screen to sleep, the sign in screen is normal on wake up.

Craig

Last edited by CraigD (2020-01-28 15:50:01)

seth · 2020-01-28 15:41:15

I don't recall setting up quiet/silent boot.

It is most unfortunately the default setting.

Your description however points to SDDM as the problem, check whether
a) there're crashes or other suspicious SDDM related entries in the journal, https://github.com/sddm/sddm/issues/1179
b) this is maybe a DRI3 issue, https://wiki.archlinux.org/index.php/In … RI3_issues (nb. that the intel and modesetting drivers adhere to different configurations, probably because the modesetting driver cares about wayland)

On a far cry: try https://wiki.archlinux.org/index.php/Ke … _KMS_start

CraigD · 2020-01-29 15:34:26

Seth,

You are above my head on the last response. However, I now believe I lead you astray. I decided to try an experiment and disabled sddm.service and installed and enabled lightdm.service and lightdm-gtk-greeter. Reboot with the same experience.

I permanently disabled quiet regenerating grub.cfg. After several reboots I noticed that on a clean boot after grub the screen displays activity then pauses clears the screen then displays more activity pauses without clearing the screen, then completes and goes to login screen. The pause without refreshing last line says "Started login Service", additional lines display but I can't read it before a perfect login screen appears.

On a bad boot everything is the same for the first full screen after grub, however, the second screen fills all at once with no pause at "Started login Service' AND the display only list things like started, reached, etc but nothing further. In other words a good boot might say and I'm making this up "starting network service" a bad boot just says "starting"

I don't know if it will help but I captured dmesg and journalctl logs after a good boot and a bad boot.

dmesg bad boot
http://ix.io/28DL

dmesg good boot
http://ix.io/28DN

journalctl bad boot
http://ix.io/28DR

journalctl good boot
http://ix.io/28DS

Craig

Last edited by CraigD (2020-01-29 15:37:43)

seth · 2020-01-29 16:35:38

That's before SDDM (and w/o with lightdm), please try the early KMS config and also https://wiki.archlinux.org/index.php/In … flickering

CraigD · 2020-01-31 05:17:41

Seth,

I was optimistic with 3 reboots in a row actually completing successfully, but it is still happening. I did the early KMS and flickering changes

Flickering

/etc/default/grub
GRUB_CMDLINE_LINUX_DEFAULT="loglevel=3 i915.enable_psr=0"
sudo grub-mkconfig -o /boot/grub/grub.cfg

Early KMS add line

/etc/mkinitcpio.conf 
MODULES=(i915)

Craig

seth · 2020-01-31 08:19:49

Did you try the lts kernel?

(Also you needed to run mkinicpio after editing mkinitcpio.conf, it's not clear whether you did that)

CraigD · 2020-02-01 04:40:49

Good catch, I forgot to run sudo mkinitcpio -p linux.

However, even afterwards I am getting bad boots.

Are you suggesting switching from linux 5.4.15.arch1-1to linux-lts 4.19.99-1?

CraigD · 2020-02-01 05:48:38

Just installed linux-lts, two clean boots then a bad one.

c00ter · 2020-02-01 14:22:05

OP, have you explored kernel entropy and random number generation? Have you tried Haveged or Rng-tools?
https://wiki.archlinux.org/index.php/Ra … generation
https://wiki.archlinux.org/index.php/Haveged
https://wiki.archlinux.org/index.php/Rng-tools

Please also read ArchWiki note for SDDM:
"A low entropy pool can cause long SDDM load time (Bug report). See Random number generation for suggestions to increase the entropy pool."

regards

CraigD · 2020-02-09 05:56:53

An update that may or may not help someone figure out what is going on.

I installed Mint Cinnamon dual boot with Arch. Mint uses lightdm display console and I am using sddm in Arch. I continue to have intermittent whorled/garbled login screens in both Arch and Mint. In Mint when this happens I put the computer into suspend then immediately unsuspend and the login screen is perfect every time after unsuspending.

Craig

Last edited by CraigD (2020-02-09 05:57:36)

Arch Linux

#1 2020-01-26 23:54:39

Intermittent boot problem, display is whorl of colors and unreadable

#2 2020-01-26 23:58:18

Re: Intermittent boot problem, display is whorl of colors and unreadable

#3 2020-01-27 00:37:29

Re: Intermittent boot problem, display is whorl of colors and unreadable

#4 2020-01-27 08:39:00

Re: Intermittent boot problem, display is whorl of colors and unreadable

#5 2020-01-27 08:52:11

Re: Intermittent boot problem, display is whorl of colors and unreadable

#6 2020-01-27 17:14:58

Re: Intermittent boot problem, display is whorl of colors and unreadable

#7 2020-01-27 18:15:09

Re: Intermittent boot problem, display is whorl of colors and unreadable

#8 2020-01-28 00:30:05

Re: Intermittent boot problem, display is whorl of colors and unreadable

#9 2020-01-28 07:05:54

Re: Intermittent boot problem, display is whorl of colors and unreadable

#10 2020-01-28 15:11:12

Re: Intermittent boot problem, display is whorl of colors and unreadable

#11 2020-01-28 15:41:15

Re: Intermittent boot problem, display is whorl of colors and unreadable

#12 2020-01-29 15:34:26

Re: Intermittent boot problem, display is whorl of colors and unreadable

#13 2020-01-29 16:35:38

Re: Intermittent boot problem, display is whorl of colors and unreadable

#14 2020-01-31 05:17:41

Re: Intermittent boot problem, display is whorl of colors and unreadable

#15 2020-01-31 08:19:49

Re: Intermittent boot problem, display is whorl of colors and unreadable

#16 2020-02-01 04:40:49

Re: Intermittent boot problem, display is whorl of colors and unreadable

#17 2020-02-01 05:48:38

Re: Intermittent boot problem, display is whorl of colors and unreadable

#18 2020-02-01 14:22:05

Re: Intermittent boot problem, display is whorl of colors and unreadable

#19 2020-02-09 05:56:53

Re: Intermittent boot problem, display is whorl of colors and unreadable

Board footer