I’ve enabled NVreg_preserveVideoMemoryAllocations
, set TemporaryFilePath
and tested the suspend and hibernate behaviour. To no avail.
Exact steps: I tried to copy the following manual on the Arch forum.
- Created a new folder in root where the files of the suspend can be stored
sudo mkdir /tmp-nvidia
- Created a file with info regarding the nvidia kernel options:
sudo nano /etc/modprobe.d/nvidia-power-management.conf
options nvidia NVreg_PreserveVideoMemoryAllocations=1 NVreg_TemporaryFilePath=/tmp-nvidia
- Reboot
- Checked the settings with:
cat /proc/driver/nvidia/params
ResmanDebugLevel: 4294967295
RmLogonRC: 1
ModifyDeviceFiles: 1
DeviceFileUID: 0
DeviceFileGID: 0
DeviceFileMode: 438
InitializeSystemMemoryAllocations: 1
UsePageAttributeTable: 4294967295
EnableMSI: 1
RegisterForACPIEvents: 1
EnablePCIeGen3: 0
MemoryPoolSize: 0
KMallocHeapMaxSize: 0
VMallocHeapMaxSize: 0
IgnoreMMIOCheck: 0
TCEBypassMode: 0
EnableStreamMemOPs: 0
EnableUserNUMAManagement: 1
NvLinkDisable: 0
RmProfilingAdminOnly: 1
PreserveVideoMemoryAllocations: 1
EnableS0ixPowerManagement: 0
S0ixPowerManagementVideoMemoryThreshold: 256
DynamicPowerManagement: 3
DynamicPowerManagementVideoMemoryThreshold: 200
RegisterPCIDriver: 1
EnablePCIERelaxedOrderingMode: 0
EnableGpuFirmware: 2
RegistryDwords: ""
RegistryDwordsPerDevice: ""
RmMsg: ""
GpuBlacklist: ""
TemporaryFilePath: "/tmp-nvidia"
ExcludedGpus: ""
The options are recognized.
Suspend: suspend journal.txt (724.1 KB)
The following details are of importance in this log:
- Lines 175-1777 and 1808-1809: The
nvidia-suspend.service
is called upon and claims to be successful. - Lines 1813-1814: Between these two lines the system was suspended.
- Lines 2042-2045: The
nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])
errors mentioned in posts above. - Line 2640: Reboot which was requried to make the system responsive again.
Hibernate: hibernate journal.txt (666.8 KB)
- Lines 103-105 and 138-139: Call and success of
nvidia-hibernate.service
. - Line 144: Actual reboot
- Lines 1688-1689: again issues with the power state from D3cold (or hot) to D0.
- Similar to before it is not a real hibernate as the system does end up with a fresh boot.
Note: I’ve found other posts that also discuss this solution for an issue that is apparently equal to ours. The advice there also surrounded the memory handling of sleep. But it did not solve their issue either. Interesting however are the remarks [1, 2] of an Nvidia moderator that there is work done on the suspend/resume behaviour of the 465.19.01 beta.
Nvidia bugreport: nvidia-bug-report.log.gz (467.0 KB)