System hangs on Reboot or shutdown since I upgraded linux to 5.12.

PickMeNow · 2021-05-25 19:28:34

Hello,

Since I upgraded to linux 5.12, this issue is permanent:

Every time I try to shutdown or reboot, it seems the system is shutting down properly, but then for some reason it hangs right at the end it seems, the lights on the case are kept on, screen goes black, keyboard goes black but it turns on again after some time.
With 5.11.16 I have no issues shutting down or rebooting.I have been searching how to solve this problem, found this https://bugs.archlinux.org/task/70654 unfortunately I didn't find any solutions.

My system:

MB: X570 Taichi 4.30
CPU: 3700X
GPU: 6800XT

You can see my system log here:
http://ix.io/3mNJ

If anyone has similar problem or knows how to fix, please advice.

Last edited by PickMeNow (2021-05-25 19:30:21)

seth · 2021-05-26 06:22:12

https://bbs.archlinux.org/viewtopic.php?id=262043 attributed the hanging flips to early KMS (but on intel) - the error shows up rather often for NAVI chips (even w/ older kernels)
Have you tried the non-zen kernel?
Can you shutdown from the multi-user.target (2nd link below) ie. notabyl w/o lightdm?

PickMeNow · 2021-05-26 12:11:30

Hi seth,

Yes, I have tried with non-zen kernel as well, the problem persists (tested with linux kernel instead of linux-zen).

Do you mind being more specific with your 2nd request,

I made the command to set to multi-user.target before shutting down, but i had still lightdm enabled...
I still had the same issue, had to power down pressing the power off button.
On rebooting i had to set back to graphical.target otherwise it wouldn't boot correctly (blank screen).
Do I have to install another DM, as.... SLIM? and try again with multi-user.target? I am a bit confused how I should proceed to test what you want.

seth · 2021-05-26 12:14:47

If you boot to the multi-user.target, the graphical.target (in your case lightdm) won't start and that's all that matters.

It's a far stretch because the session was so short, but does passing "amdgpu.dpm=0" to the kernel help?

PickMeNow · 2021-05-26 17:27:22

Added "amdgpu.dpm=0" to GRUB_CMDLINE_LINUX_DEFAULT="loglevel=3 quiet splash"
But wasn't able to reach the DM, black screen, seems the gpu can't work without dpm.

seth · 2021-05-26 18:54:08

That seems unlikely related, more likely https://wiki.archlinux.org/title/LightD … TTY_output ?
Do you currently use https://wiki.archlinux.org/title/Kernel … _KMS_start ?

PickMeNow · 2021-05-27 14:26:04

Yes, I do have:

[LightDM]
logind-check-graphical=true

I had to put it, because it wouldn't display anything in the beginning, ever since i changed from Vega 56 to 6800xt ( possibly very early support ). Unncessary with Vega 56.
on mkinitcpio.conf I basically have default, nothing loading on modules, not binaries or files.

Should I revert to "logind-check-graphical=false" and add amdgpu to modules?

seth · 2021-05-27 15:01:02

No, keep the check.
You can add amdgpu to the iniramfs (don't forget to re-generate it) "despite" https://bbs.archlinux.org/viewtopic.php?id=262043 (which is on intel anyway)
Then try whether disabling dpm still prevents reaching the DM (and in doubt try w/o lightdm and just the behavior on the multi-user.target)

PickMeNow · 2021-05-27 15:54:36

Adding amdgpu to modules, just made it worse, still black screen, changed the mkinitcpio and mkinitcpio -p linux-zen now fails... no idea why....

Tried installing linux-zen and linux-zen-headers, says a bunch of error messages that already exists in the system.
Cannot generate mkinitcpio for linux-zen no more.

Had to install linux and was able to boot but now I dunno how to go back to linux-zen....

seth · 2021-05-27 15:56:28

Don't paraphrase but post the actual outputs - "fails... no idea why" does not allow anyone else to determine the cause either.

PickMeNow · 2021-05-27 16:54:33

Sorry for this, meanwhile I fixed the linux-zen issue.

So tested with mkinitcio.conf with modules= ( amdgpu ), did mkinitcio -p linux-zen to update initramfs.
When I try to boot black screen, tested with both linux and linux-zen.

Rolling back to modules= () fixed the issue, but still have the persist issue with reboot and shutdown.

Last edited by PickMeNow (2021-05-27 17:22:13)

seth · 2021-05-27 21:01:44

That's rather odd. Can you please post a system journal for such a boot ("sudo journalctl -b -1" for the previous boot) as well a your xorg log (w/ a "regular" boot)?

PickMeNow · 2021-05-29 11:03:13

MODULES = (amdgpu) logs :

http://ix.io/3odQ ( Xorg.0.log.old )
http://ix.io/3odP (sudo journalctl -b -1)

W/ MODULES = (amdgpu) logs:

http://ix.io/3odQ (normal Xorg.0.log )
http://ix.io/3odV (sudo journalctl -b)

Just a few things I noticed:
When i selected archlinux in grub it usually displays the version number of the kernel plus the nvme drive where the linux system is located, this time it didn't display the nvme drive.
Another thing I noticed, it had the same behavior when I usually reboot or shutdown ( my keyboard lights go off and on ) and also noticed one of my fans completely dying and coming back.

seth · 2021-05-29 18:21:11

The xorg log is always the same /var/log/Xorg.0.log and shows a server that ran for ~300s

The journal has interesting parts
Bad:

May 29 11:43:30 ArchLinux kernel: Linux agpgart interface v0.103
May 29 11:43:30 ArchLinux kernel: [drm] amdgpu kernel modesetting enabled.
May 29 11:43:30 ArchLinux kernel: amdgpu: Ignoring ACPI CRAT on non-APU system
May 29 11:43:30 ArchLinux kernel: Virtual CRAT table created for CPU
May 29 11:43:30 ArchLinux kernel: amdgpu: Topology: Add CPU node
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: vgaarb: deactivate vga console
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: enabling device (0000 -> 0003)
May 29 11:43:30 ArchLinux kernel: [drm] initializing kernel modesetting (SIENNA_CICHLID 0x1002:0x73BF 0x1462:0x3951 0xC1).
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: Trusted Memory Zone (TMZ) feature not supported
May 29 11:43:30 ArchLinux kernel: [drm] register mmio base: 0xFCC00000
May 29 11:43:30 ArchLinux kernel: [drm] register mmio size: 1048576
May 29 11:43:30 ArchLinux kernel: [drm:amdgpu_discovery_reg_base_init [amdgpu]] *ERROR* invalid ip discovery binary signature
May 29 11:43:30 ArchLinux kernel: [drm:amdgpu_discovery_reg_base_init [amdgpu]] *ERROR* amdgpu_discovery_init failed
May 29 11:43:30 ArchLinux kernel: [drm] failed to init reg base from ip discovery table, fallback to legacy init method
May 29 11:43:30 ArchLinux kernel: [drm] add ip block number 0 <nv_common>
May 29 11:43:30 ArchLinux kernel: [drm] add ip block number 1 <gmc_v10_0>
May 29 11:43:30 ArchLinux kernel: [drm] add ip block number 2 <navi10_ih>
May 29 11:43:30 ArchLinux kernel: [drm] add ip block number 3 <psp>
May 29 11:43:30 ArchLinux kernel: [drm] add ip block number 4 <smu>
May 29 11:43:30 ArchLinux kernel: [drm] add ip block number 5 <dm>
May 29 11:43:30 ArchLinux kernel: [drm] add ip block number 6 <gfx_v10_0>
May 29 11:43:30 ArchLinux kernel: [drm] add ip block number 7 <sdma_v5_2>
May 29 11:43:30 ArchLinux kernel: [drm] add ip block number 8 <vcn_v3_0>
May 29 11:43:30 ArchLinux kernel: [drm] add ip block number 9 <jpeg_v3_0>
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: Fetched VBIOS from ROM BAR
May 29 11:43:30 ArchLinux kernel: amdgpu: ATOM BIOS: GAMINGX
May 29 11:43:30 ArchLinux kernel: [drm] VCN(0) decode is enabled in VM mode
May 29 11:43:30 ArchLinux kernel: [drm] VCN(1) decode is enabled in VM mode
May 29 11:43:30 ArchLinux kernel: [drm] VCN(0) encode is enabled in VM mode
May 29 11:43:30 ArchLinux kernel: [drm] VCN(1) encode is enabled in VM mode
May 29 11:43:30 ArchLinux kernel: [drm] JPEG decode is enabled in VM mode
May 29 11:43:30 ArchLinux kernel: [drm] GPU posting now...
May 29 11:43:30 ArchLinux kernel: tsc: Refined TSC clocksource calibration: 3599.949 MHz
May 29 11:43:30 ArchLinux kernel: clocksource: tsc: mask: 0xffffffffffffffff max_cycles: 0x33e422fe1de, max_idle_ns: 440795286378 ns
May 29 11:43:30 ArchLinux kernel: clocksource: Switched to clocksource tsc
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: HBM ECC is not presented.
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: SRAM ECC is not presented.
May 29 11:43:30 ArchLinux kernel: [drm] vm size is 262144 GB, 4 levels, block size is 9-bit, fragment size is 9-bit
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: BAR 2: releasing [mem 0x7c00000000-0x7c0fffffff 64bit pref]
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: BAR 0: releasing [mem 0x7800000000-0x7bffffffff 64bit pref]
May 29 11:43:30 ArchLinux kernel: [drm:amdgpu_device_resize_fb_bar [amdgpu]] *ERROR* Problem resizing BAR0 (-22).
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: BAR 0: assigned [mem 0x7800000000-0x7bffffffff 64bit pref]
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: BAR 2: assigned [mem 0x7c00000000-0x7c0fffffff 64bit pref]
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: VRAM: 0M 0x0000000000000000 - 0xFFFFFFFFFFFFFFFF (0M used)
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: GART: 512M 0x00007FFF00000000 - 0x00007FFF1FFFFFFF
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: AGP: 134201344M 0x0000000000000000 - 0x00007FFBFFFFFFFF
May 29 11:43:30 ArchLinux kernel: [drm] Detected VRAM RAM=0M, BAR=16384M
May 29 11:43:30 ArchLinux kernel: [drm] RAM width 256bits GDDR6
May 29 11:43:30 ArchLinux kernel: [TTM] Zone  kernel: Available graphics memory: 16397942 KiB
May 29 11:43:30 ArchLinux kernel: [TTM] Zone   dma32: Available graphics memory: 2097152 KiB
May 29 11:43:30 ArchLinux kernel: [drm] amdgpu: 0M of VRAM memory ready
May 29 11:43:30 ArchLinux kernel: [drm] amdgpu: 3072M of GTT memory ready.
May 29 11:43:30 ArchLinux kernel: [drm] GART: num cpu pages 131072, num gpu pages 131072
May 29 11:43:30 ArchLinux kernel: [drm:amdgpu_device_init.cold [amdgpu]] *ERROR* sw_init of IP block <gmc_v10_0> failed -12
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: amdgpu_device_ip_init failed
May 29 11:43:30 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: Fatal error during GPU init
May 29 11:43:30 ArchLinux kernel: amdgpu: probe of 0000:10:00.0 failed with error -12

"Good"(-ish):

May 29 11:47:05 ArchLinux kernel: [drm] amdgpu kernel modesetting enabled.
May 29 11:47:05 ArchLinux kernel: amdgpu: Ignoring ACPI CRAT on non-APU system
May 29 11:47:05 ArchLinux kernel: Virtual CRAT table created for CPU
May 29 11:47:05 ArchLinux kernel: amdgpu: Topology: Add CPU node
May 29 11:47:05 ArchLinux kernel: checking generic (7800000000 e10000) vs hw (7800000000 400000000)
May 29 11:47:05 ArchLinux kernel: fb0: switching to amdgpudrmfb from EFI VGA
May 29 11:47:05 ArchLinux kernel: Console: switching to colour dummy device 80x25
May 29 11:47:05 ArchLinux kernel: amdgpu 0000:10:00.0: vgaarb: deactivate vga console
May 29 11:47:05 ArchLinux kernel: amdgpu 0000:10:00.0: enabling device (0006 -> 0007)
May 29 11:47:05 ArchLinux kernel: [drm] initializing kernel modesetting (SIENNA_CICHLID 0x1002:0x73BF 0x1462:0x3951 0xC1).
May 29 11:47:05 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: Trusted Memory Zone (TMZ) feature not supported
May 29 11:47:05 ArchLinux kernel: [drm] register mmio base: 0xFCC00000
May 29 11:47:05 ArchLinux kernel: [drm] register mmio size: 1048576
May 29 11:47:05 ArchLinux kernel: [drm] add ip block number 0 <nv_common>
May 29 11:47:05 ArchLinux kernel: [drm] add ip block number 1 <gmc_v10_0>
May 29 11:47:05 ArchLinux kernel: [drm] add ip block number 2 <navi10_ih>
May 29 11:47:05 ArchLinux kernel: [drm] add ip block number 3 <psp>
May 29 11:47:05 ArchLinux kernel: [drm] add ip block number 4 <smu>
May 29 11:47:05 ArchLinux kernel: [drm] add ip block number 5 <dm>
May 29 11:47:05 ArchLinux kernel: [drm] add ip block number 6 <gfx_v10_0>
May 29 11:47:05 ArchLinux kernel: [drm] add ip block number 7 <sdma_v5_2>
May 29 11:47:05 ArchLinux kernel: [drm] add ip block number 8 <vcn_v3_0>
May 29 11:47:05 ArchLinux kernel: [drm] add ip block number 9 <jpeg_v3_0>
May 29 11:47:05 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: Fetched VBIOS from VFCT
May 29 11:47:05 ArchLinux kernel: amdgpu: ATOM BIOS: GAMINGX
May 29 11:47:05 ArchLinux kernel: [drm] VCN(0) decode is enabled in VM mode
May 29 11:47:05 ArchLinux kernel: [drm] VCN(1) decode is enabled in VM mode
May 29 11:47:05 ArchLinux kernel: [drm] VCN(0) encode is enabled in VM mode
May 29 11:47:05 ArchLinux kernel: [drm] VCN(1) encode is enabled in VM mode
May 29 11:47:05 ArchLinux kernel: [drm] JPEG decode is enabled in VM mode
May 29 11:47:05 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: HBM ECC is not presented.
May 29 11:47:05 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: SRAM ECC is not presented.
May 29 11:47:05 ArchLinux kernel: [drm] vm size is 262144 GB, 4 levels, block size is 9-bit, fragment size is 9-bit
May 29 11:47:05 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: VRAM: 16368M 0x0000008000000000 - 0x00000083FEFFFFFF (16368M used)
May 29 11:47:05 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: GART: 512M 0x0000000000000000 - 0x000000001FFFFFFF
May 29 11:47:05 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: AGP: 267894784M 0x0000008400000000 - 0x0000FFFFFFFFFFFF
May 29 11:47:05 ArchLinux kernel: [drm] Detected VRAM RAM=16368M, BAR=16384M
May 29 11:47:05 ArchLinux kernel: [drm] RAM width 256bits GDDR6
May 29 11:47:05 ArchLinux kernel: [TTM] Zone  kernel: Available graphics memory: 16397814 KiB
May 29 11:47:05 ArchLinux kernel: [TTM] Zone   dma32: Available graphics memory: 2097152 KiB
May 29 11:47:05 ArchLinux kernel: [drm] amdgpu: 16368M of VRAM memory ready
May 29 11:47:05 ArchLinux kernel: [drm] amdgpu: 16368M of GTT memory ready.
May 29 11:47:05 ArchLinux kernel: [drm] GART: num cpu pages 131072, num gpu pages 131072
May 29 11:47:05 ArchLinux kernel: [drm] PCIE GART of 512M enabled (table at 0x0000008000E10000).
…
May 29 11:47:07 ArchLinux kernel: [drm] Loading DMUB firmware via PSP: version=0x00000000
May 29 11:47:07 ArchLinux kernel: [drm] use_doorbell being set to: [true]
May 29 11:47:07 ArchLinux kernel: [drm] use_doorbell being set to: [true]
May 29 11:47:07 ArchLinux kernel: [drm] use_doorbell being set to: [true]
May 29 11:47:07 ArchLinux kernel: [drm] use_doorbell being set to: [true]
May 29 11:47:07 ArchLinux kernel: [drm] Found VCN firmware Version ENC: 1.13 DEC: 2 VEP: 0 Revision: 20
May 29 11:47:07 ArchLinux kernel: [drm] PSP loading VCN firmware
May 29 11:47:07 ArchLinux kernel: [drm] reserve 0xa00000 from 0x83fe000000 for PSP TMR
May 29 11:47:07 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: SECUREDISPLAY: securedisplay ta ucode is not available
May 29 11:47:07 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: use vbios provided pptable
May 29 11:47:07 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: SMU is initialized successfully!
May 29 11:47:07 ArchLinux kernel: [drm] Display Core initialized with v3.2.122!
May 29 11:47:07 ArchLinux kernel: [drm] DMUB hardware initialized: version=0x02000018
May 29 11:47:07 ArchLinux kernel: [drm] REG_WAIT timeout 1us * 100000 tries - mpc2_assert_idle_mpcc line:480
May 29 11:47:07 ArchLinux kernel: snd_hda_intel 0000:10:00.1: bound 0000:10:00.0 (ops amdgpu_dm_audio_component_bind_ops [amdgpu])
May 29 11:47:08 ArchLinux kernel: [drm] kiq ring mec 2 pipe 1 q 0
May 29 11:47:08 ArchLinux kernel: [drm] VCN decode and encode initialized successfully(under DPG Mode).
May 29 11:47:08 ArchLinux kernel: [drm] JPEG decode initialized successfully.
May 29 11:47:08 ArchLinux kernel: kfd kfd: Allocated 3969056 bytes on gart
May 29 11:47:08 ArchLinux kernel: Virtual CRAT table created for GPU
May 29 11:47:08 ArchLinux kernel: amdgpu: Topology: Add dGPU node [0x73bf:0x1002]
May 29 11:47:08 ArchLinux kernel: kfd kfd: added device 1002:73bf
May 29 11:47:08 ArchLinux kernel: amdgpu 0000:10:00.0: amdgpu: SE 4, SH per SE 2, CU per SH 10, active_cu_number 72
May 29 11:47:08 ArchLinux kernel: [drm] fb mappable at 0x780107B000
May 29 11:47:08 ArchLinux kernel: [drm] vram apper at 0x7800000000
May 29 11:47:08 ArchLinux kernel: [drm] size 14745600
May 29 11:47:08 ArchLinux kernel: [drm] fb depth is 24
May 29 11:47:08 ArchLinux kernel: [drm]    pitch is 10240
May 29 11:47:08 ArchLinux kernel: fbcon: amdgpudrmfb (fb0) is primary device
May 29 11:47:08 ArchLinux kernel: [drm] REG_WAIT timeout 1us * 100000 tries - mpc2_assert_idle_mpcc line:480

sw_init of IP block matches https://bugs.freedesktop.org/show_bug.cgi?id=110929 and https://blog.lordvan.com/blog/radeon-rx … h-on-boot/ but they're both dealing w/ SME
There're also some MCE errrors.
You don't load the micorcode updates early, do you?
https://wiki.archlinux.org/title/Microc … om_kernels

If that's not it, I'd check https://wiki.archlinux.org/title/Stress … ing_memory
(notice that you need to run it for several cycles to halfwise reliably detect issues)

Also, sanity check: no eGPU or pci passthrough involved here?

PickMeNow · 2021-05-29 18:59:45

I only have one GPU - 6800xt - it's not a eGPU or PCI Passthrough.

hmmm regarding the microcode in /boot/grub/grub.cfg i have

initrd /amd-ucode.img /initramfs-linux-zen.img

So everything should be okish... on the microcode side.

Should I add "mem_encrypt=off" to kernel parameters?

seth · 2021-05-29 19:51:42

Given https://archlinux.org/packages/core/any/amd-ucode/ is installed - yes.
"mem_encrypt=off" will disable SME, you can try, but the logs don't indicate that it's activated.

Remains the memtest… but the behavior seems a bit deterministic.

PickMeNow · 2021-05-29 23:37:59

Ran memtest for a couple of hours, didn't find any issue.

Found an interesting post on gitlab amdgpu that might be the cause, another person is experiencing the same issue, reading the comments :
https://gitlab.freedesktop.org/drm/amd/-/issues/1602

It appears to be fixed in 5.13, so I guess I just need to wait...

Arch Linux

#1 2021-05-25 19:28:34

System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#2 2021-05-26 06:22:12

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#3 2021-05-26 12:11:30

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#4 2021-05-26 12:14:47

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#5 2021-05-26 17:27:22

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#6 2021-05-26 18:54:08

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#7 2021-05-27 14:26:04

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#8 2021-05-27 15:01:02

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#9 2021-05-27 15:54:36

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#10 2021-05-27 15:56:28

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#11 2021-05-27 16:54:33

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#12 2021-05-27 21:01:44

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#13 2021-05-29 11:03:13

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#14 2021-05-29 18:21:11

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#15 2021-05-29 18:59:45

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#16 2021-05-29 19:51:42

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

#17 2021-05-29 23:37:59

Re: System hangs on Reboot or shutdown since I upgraded linux to 5.12.

Board footer