Cannot shutdown AMD laptop with linux6.6-6.6.31_1

[Sqvid]

Is this a new report?

Yes

System Info

Void 6.6.31_1 x86_64 AuthenticAMD uptodate rrFFFF

Package(s) Affected

linux6.6-6.6.31_1

Does a report exist for this bug with the project's home (upstream) and/or another distro?

No response

Expected behaviour

After runit winds down services the laptop should switch off.

Actual behaviour

Upgrading to kernel 6.6.31_1 prevents the computer from shutting down. Runit seems to stop all services correctly and the screen goes black as expected; however, the keyboard backlight and power light stay on and the fans still run.

##System info:
CPU: AMD Ryzen 7 7840HS w/ Radeon 780M Graphics
GPU: Nvidia GeForce RTX 4060 (disabled and powered off)
WM: Sway

Downgrading to 6.6.30_1 or below fixes this issue and the laptop shuts down immediately as expected.

Not sure how to extract useful debug info but would appreciate any tips.

Steps to reproduce

1. Use effected kernel (though it may be hardware dependent).
2. sudo poweroff now.
3. Read runit messages of services being stopped.
4. Screen goes black.
5. Laptop stays on.
6. Cry.

[blacklightpy]

I have kernel 6.6.29_1 and my system did not poweroff when I tried shutting it down with with my AGS widgets. I didn't debug whether it was the widgets or the system though, and I'm not in a mood to shutdown now.

[1is7ac3]

I had this problem on my two Nvidia Optimus laptops: (Intel + Nvidia) and (Amd + Nvidia) while I was in the process of resuming the use of Void Linux, so it only gave me the error while I was using Nouveau. When I finished configuring the system with the Nvidia drivers, it no longer appeared.

[funk443]

I encountered this problem after I updated the kernel this afternoon. Exactly the same problem like @Sqvid.

My system spec:

• uname -a output: Linux void-btw 6.6.31_1 #1 SMP PREEMPT_DYNAMIC Sat May 18 01:07:55 UTC 2024 x86_64 GNU/Linux
• CPU: AMD Ryzen 7 Pro 7840u w/ Radeon 780M Graphics
• GPU: Intgrated Radeon 780M Graphics

Since I usually clear caches after an update, so I cannot downgrade the kernel. But I tried installing linux6.9 and it works just fine.

[tomalexander]

I'm on arch linux but I have the same behavior. I went in my cache and reverted to test:

linux-lts-6.6.30-2-x86_64.pkg.tar.zst : shuts down fine
linux-lts-6.6.31-1-x86_64.pkg.tar.zst : does not shut down

[artluix]

Facing same issue on Laptop with 8845hs during installation (no UI is installed)

[thenbe]

I found this thread through a search engine (I'm on nixOS not void linux), but I have the same issue on nixOS on a framework amd laptop (AMD Ryzen 7 7840U). Shutdown appears to work at first, where the systemd logs zoom right past like always. Then the screen goes black, but the power "stays on" (power led stays lit) indefinitely. I need to manually hold the power button for several seconds to force the shutdown to go through. Rolling back the kernel from 6.6.31 to 6.6.30 fixes the issue.

[superm1]

Anyone affected by this can you check two things:

1. Is 6.9.1 affected?
2. If 6.9.1 isn't affected it's probably an incomplete backport. It's best to bisect 6.6.30 to 6.6.31 to find which commit caused it.

[superm1]

CC @knurd

[knurd]

Warning, the following is a totally wild guess. But there was one report about a shutdown problem caused by a commit that went into 6.6.31 as well; wonder if that might be related. https://lore.kernel.org/all/CAE4VaREzY%2Ba2PvQJYJbfh8DwB4OP7kucZG-e28H22xyWob1w_A@mail.gmail.com/

[lectrode]

Bisected between 6.6.30 and 6.6.31 to find the problem commit:

$ git bisect start --no-checkout
status: waiting for both good and bad commits
$ git bisect good 5697d159afef8c475f13a0b7b85f09bd4578106c
status: waiting for bad commit, 1 good commit known
$ git bisect bad e3d332aaf898ed755b29c8cdf59be2cfba1cac4b
Bisecting: 154 revisions left to test after this (roughly 7 steps)
[6466a0f6d235c8a18c602cb587160d7e49876db9] uio_hv_generic: Don't free decrypted memory
$ git bisect good
Bisecting: 77 revisions left to test after this (roughly 6 steps)
[7a54e5052bde582fd0e7677334fe7a5be92e242c] usb: gadget: uvc: use correct buffer size when parsing configfs lists
$ git bisect bad
Bisecting: 38 revisions left to test after this (roughly 5 steps)
[2ee2fc6786bc5ff3c24798624ea3806c9662c26f] selftests/net: convert test_bridge_neigh_suppress.sh to run it in unique namespace
$ git bisect good
Bisecting: 19 revisions left to test after this (roughly 4 steps)
[7019a64165186fd3fb5ba928a6347558dc560093] net: dsa: mv88e6xxx: add phylink_get_caps for the mv88e6320/21 family
$ git bisect good
Bisecting: 9 revisions left to test after this (roughly 3 steps)
[0c7ed3ed35eec9138b88d42217b5a6b9a62bda4d] drm/amdgpu: once more fix the call oder in amdgpu_ttm_move() v2
$ git bisect bad
Bisecting: 4 revisions left to test after this (roughly 2 steps)
[95ca7c90eaf5ea8a8460536535101e3e81160e2a] gpiolib: cdev: Fix use after free in lineinfo_changed_notify
$ git bisect bad
Bisecting: 2 revisions left to test after this (roughly 1 step)
[fa2d2e2d8eae03acf49229793f2a6fddede92c4d] drm/meson: dw-hdmi: add bandgap setting for g12
$ git bisect good
Bisecting: 0 revisions left to test after this (roughly 1 step)
[6c83a8f236ece78d5c2e60ae3dcfd1a64509410e] dm/amd/pm: Fix problems with reboot/shutdown for some SMU 13.0.4/13.0.11 users
$ git bisect bad
Bisecting: 0 revisions left to test after this (roughly 0 steps)
[1b9e8de483bcc230f6e922bdfa9d1c186c27dd3b] drm/connector: Add \n to message about demoting connector force-probes
$ git bisect good
6c83a8f236ece78d5c2e60ae3dcfd1a64509410e is the first bad commit
commit 6c83a8f236ece78d5c2e60ae3dcfd1a64509410e
Author: Mario Limonciello <mario.limonciello@amd.com>
Date:   Thu May 2 13:32:17 2024 -0500

    dm/amd/pm: Fix problems with reboot/shutdown for some SMU 13.0.4/13.0.11 users
    
    [ Upstream commit cd94d1b182d2986378550c9087571991bfee01d4 ]
    
    Limit the workaround introduced by commit 31729e8c21ec ("drm/amd/pm: fixes
    a random hang in S4 for SMU v13.0.4/11") to only run in the s4 path.
    
    Cc: Tim Huang <Tim.Huang@amd.com>
    Fixes: 31729e8c21ec ("drm/amd/pm: fixes a random hang in S4 for SMU v13.0.4/11")
    Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3351
    Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
    Acked-by: Alex Deucher <alexander.deucher@amd.com>
    Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
    Signed-off-by: Sasha Levin <sashal@kernel.org>

 drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_4_ppt.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

[superm1]

Ok... Does it also fail on 6.9?

[tomalexander]

It's best to bisect 6.6.30 to 6.6.31 to find which commit caused it.

I have bisected it to this commit in https://kernel.googlesource.com/pub/scm/linux/kernel/git/stable/linux.git :

commit 6c83a8f236ece78d5c2e60ae3dcfd1a64509410e (HEAD)
Author: Mario Limonciello <mario.limonciello@amd.com>
Date:   Thu May 2 14:32:17 2024

    dm/amd/pm: Fix problems with reboot/shutdown for some SMU 13.0.4/13.0.11 users

    [ Upstream commit cd94d1b182d2986378550c9087571991bfee01d4 ]

    Limit the workaround introduced by commit 31729e8c21ec ("drm/amd/pm: fixes
    a random hang in S4 for SMU v13.0.4/11") to only run in the s4 path.

    Cc: Tim Huang <Tim.Huang@amd.com>
    Fixes: 31729e8c21ec ("drm/amd/pm: fixes a random hang in S4 for SMU v13.0.4/11")
    Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3351
    Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
    Acked-by: Alex Deucher <alexander.deucher@amd.com>
    Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
    Signed-off-by: Sasha Levin <sashal@kernel.org>

[lectrode]

The issue is not present on 6.9.1. I would agree it looks like an incomplete/failed backport.
@tomalexander good to have confirmation