Re: no sleep with pm-utils & nvidia 565.77

dummy@example.com (meridian) — Wed, 29 Jan 2025 19:26:13 +0000

I had the same problem, with the same driver and a RTX A400, but also with poweroff and logout issues.

All is good now: I deinstalled the 565.77 version and installed the devuan one (535.216.01 currently), even if nvidia-detect said that this card is unsupported (it works perfectly).

I use your 99-nvidia file (modified a little), with the pm-suspend command. Hibernation is OK too with pm-hibernate. NVreg_PreserveVideoMemoryAllocations is in use.

In /etc/modprobe.d came a longer nvidia.conf.

Many thanks hwd for your file, I didn't know how to do without systemd…

The command line for the installation was:

 apt install firmware-misc-nonfree nvidia-kernel-dkms nvidia-driver nvidia-cuda-dev nvidia-settings nvidia-smi nvidia-xconfig nvidia-opencl-icd nvidia-opencl-common nvidia-detect nvidia-suspend-common

In the sources.list I added:

deb https://deb.devuan.org/merged daedalus main non-free-firmware non-free contrib

Re: no sleep with pm-utils & nvidia 565.77

dummy@example.com (hwd) — Wed, 25 Dec 2024 17:29:41 +0000

Ok ... I found out that in order to fire pm-utils hooks, it is necessary to trigger the whole thing using pm-suspend.

Unfortunately this led to the same result, even though I can see in /var/log/pm-suspend.log that all hooks are fired, including mine from /etc/pm/sleep.d

So it seems that it's not pm-utils but the driver.

I guess I have to go deeper down the rabbit hole

no sleep with pm-utils & nvidia 565.77

dummy@example.com (hwd) — Wed, 25 Dec 2024 11:32:29 +0000

Hi folks,

I've been trying for 2 days to get my setup to go into standby mode (S2R) and have been unsuccessful in my attempts.

It is a system with a self compiled kernel 6.12.6 from the .config of ‘linux-image-6.1.0-28-amd64’ (make bindeb-pkg) and the proprietary nvidia drivers version 565.77 for a RTX 3070/8Gb.

So far everything is running.

KDE/Plasma on X11 runs normally, gaming with wine works great ... just not sending the computer to sleep.

To achieve this I have done the following so far:

pm-utils (1.4.1-19) installed

cat /proc/cmdline 

BOOT_IMAGE=/boot/vmlinuz-6.12.6-hwd root=UUID=3b0bb9e5-6471-4d97-a584-5799ea851273 ro resume=UUID=39b90cbc-cc65-4006-bd3d-9f9305d10db7 nvidia-drm. modeset=1 nvidia.NVreg_PreserveVideoMemoryAllocations=1 nvidia.NVreg_TemporaryFilePath=/var/tmp

cat /etc/pm/sleep.d/99-nvidia 
#!/bin/sh

logger ‘pm-sleep.d-99-nvidia’

case ‘$1’ in

    suspend|hibernate)
	logger ‘SLEEP’ && /usr/bin/nvidia-sleep.sh suspend
        ;;

    resume|thaw)
	logger ‘RESUME’ && /usr/bin/nvidia-sleep.sh resume
        ;;

esac

However, if I now send my system to sleep using echo freeze > /sys/power/state, the screens go off briefly, but then come back on again immediately and the system greets me with the login screen.
dmesg says:

[11399.192259] PM: suspend entry (s2idle)
[11399.205067] Filesystems sync: 0.012 seconds
[11400.050938] Freezing user space processes
[11400.052735] Freezing user space processes completed (elapsed 0.001 seconds)
[11400.052737] OOM killer disabled.
[11400.052738] Freezing remaining freezable tasks
[11406.019704] Freezing remaining freezable tasks completed (elapsed 5.966 seconds)
[11406.019712] printk: Suspending console(s) (use no_console_suspend to debug)
[11406.028557] r8169 0000:0c:00.0 eth1: Link is Down
[11406.036832] r8169 0000:0b:00.0 eth0: Link is Down
[11406.040375] sd 8:0:0:0: [sdb] Synchronizing SCSI cache
[11406.044838] NVRM: GPU 0000:01:00.0: PreserveVideoMemoryAllocations module parameter is set. System Power Management attempted without driver procfs suspend interface. Please refer to the 'Configuring Power Management Support' section in the driver README.
[11406.044840] nvidia 0000:01:00.0: PM: pci_pm_suspend(): nv_pmops_suspend [nvidia] returns -5
[11406.044970] nvidia 0000:01:00.0: PM: dpm_run_callback(): pci_pm_suspend returns -5
[11406.044974] nvidia 0000:01:00.0: PM: failed to suspend async: error -5
[11406.057238] sd 0:0:0:0: [sda] Synchronizing SCSI cache
[11406.210880] PM: Some devices failed to suspend, or early wake event detected
[11406.216183] nvme nvme0: D3 entry latency set to 10 seconds
[11406.217214] nvme nvme2: D3 entry latency set to 10 seconds
[11406.218043] nvme nvme2: 8/0/0 default/read/poll queues
[11406.218096] nvme nvme0: 16/0/0 default/read/poll queues
[11406.218180] nvme nvme2: ctrl returned bogus length: 16 for NVME_NIDT_EUI64
[11406.218221] nvme nvme2: Ignoring bogus Namespace Identifiers
[11406.377399] r8169 0000:0b:00.0 eth0: Link is Down
[11406.387071] r8169 0000:0c:00.0 eth1: Link is Down
[11406.517806] ata6: SATA link down (SStatus 0 SControl 300)
[11406.517828] ata5: SATA link down (SStatus 0 SControl 300)
[11406.517845] ata4: SATA link down (SStatus 0 SControl 300)
[11406.517863] ata3: SATA link down (SStatus 0 SControl 300)
[11406.517892] usb 1-5.3: reset high-speed USB device number 6 using xhci_hcd
[11406.934994] usb 1-5.3.3: reset full-speed USB device number 8 using xhci_hcd
[11407.119860] cp210x ttyUSB0: cp210x converter now disconnected from ttyUSB0
[11407.119874] cp210x 1-5.3.3:1.0: device disconnected
[11407.121088] cp210x 1-5.3.3:1.0: cp210x converter detected
[11407.122098] usb 1-5.3.3: cp210x converter now attached to ttyUSB0
[11407.122176] OOM killer enabled.
[11407.122177] Restarting tasks ... done.
[11407.123380] random: crng reseeded on system resumption
[11407.123413] PM: suspend exit
[11409.479482] r8169 0000:0c:00.0 eth1: Link is Up - 1Gbps/Full - flow control off
[11409.935543] r8169 0000:0b:00.0 eth0: Link is Up - 2.5Gbps/Full - flow control rx/tx

In the kernel log, however, I cannot find any lines from the script /etc/pm/sleep.d/99-nvidia
Neither that the script was executed (logger ‘pm-sleep.d-99-nvidia’), nor that it entered any branch in the case tree (logger ‘SLEEP’/logger ‘RESUME’)

Accordingly, I am assuming an error in my pm-utils configuration. However, I have no idea where to start here.
I first thought that I forgot to start the service, but there is no init-script for the pm-utils in /etc/init.d

Maybe one of you has an idea how I can execute the hook.

Many thanks in advance!

Dev1 Galaxy Forum / no sleep with pm-utils & nvidia 565.77

Re: no sleep with pm-utils & nvidia 565.77

Re: no sleep with pm-utils & nvidia 565.77

no sleep with pm-utils & nvidia 565.77