Power Mode Changes#
Power Mode and VM State API Changes#
Target Changes |
Backward Compatible |
Platform |
OS |
|---|---|---|---|
In DriveOS 7.x, power mode and VM functional-state management follows
the new Power Management architecture built on the PM Server and
NvDVMS. The power-management architecture is common to both Linux and
QNX guest VMs. SoC operating-point (SOC_OP) transitions are driven
through the PM Server client library ( |
No |
NSR SR |
QNX Linux |
Migration Path
Orin to Thor
Dimensity C-X1 to Dimensity C-X1
Migration Release Path
QNX: 6.x to 7.x
Linux: 7.0 to 7.2
Migration Rationale
DriveOS 7.x introduces a new Power Management architecture that changes the complete set of power-mode and VM-state methods:
The PM Server (running on HVRTOS) owns the system power state machine and orchestrates SoC operating-point (SOC_OP) transitions. The PM Server is accessed only through its client library
libnvpwrmgmt(theNvPwrMgmt*APIs) over IVC.The LCM Server and LCM Client framework coordinates SOC_OP transitions (notify, prepare, start, and wait-for-completion phases) with the PM Server and propagates the corresponding per-VM operations.
NvDVMS is the authoritative owner of the per-VM functional state (
NvDvmsState) and VM power profile (NvDvmsVmOp). Applications and integration code drive VM state and power-profile transitions through the NvDVMS client library usingnvdvms_set_vm_state()andnvdvms_set_power_profile().Power state is modeled as an
IP_OPtoVM_OPtoSOC_OPhierarchy with a defined VM state machine (init, init done, operational, suspend/resume, deinit prepare, deinit, and reinit).
As a result, the power-management interfaces used in 6.x are replaced, and existing power-mode and VM-state handling must be reworked to use the PM Server and NvDVMS APIs.
Steps to Migrate
Adopt the new power-management interfaces. For additional information, refer to the “PM Architecture Integration” topic in the NVIDIA DriveOS Developer Guide.