Thermal Management Active Architecture

Registered by Steve Jahnke on 2011-04-19

Any user-space process can die, be killed, or starve. Need a way to make sure the Thermal Manager is still alive and if not, be able to restart or notify the system. Research, define, architect and implement. Do we have a Kernel-level component (as part of the Thermal Framework or simply a separate driver?) that detect a "heartbeat" from the Thermal Manager and restarts the process if not detected?

Blueprint information

Status:
Complete
Approver:
Amit Kucheria
Priority:
Medium
Drafter:
Steve Jahnke
Direction:
Approved
Assignee:
Steve Jahnke
Definition:
Approved
Series goal:
Accepted for trunk
Implementation:
Informational Informational
Milestone target:
milestone icon 2011.06
Started by
Amit Kucheria on 2011-06-03
Completed by
Amit Kucheria on 2011-06-03

Related branches

Sprints

Whiteboard

RESULT:
  The thermal framework, or any user space governor, will not need to concern itself with user-space starvation or any other user-space system issue. The thermal framework will contain a default governor as a back-up in kernel space if the user-space application has an issue. Further, if an overheating condition occurs resulting in unstable kernel operation (as possibility), the system hardware must be able to capture the overheat event directly and reset the system accordingly.

(?)

Work Items

Work items:
[sjahnke] : Determine if the framework needs to worry about user-space issues: DONE

Dependency tree

* Blueprints in grey have been implemented.

This blueprint contains Public information 
Everyone can see this information.