[linux-pm] Ottawa Linux Power Management Summit, June 25-26, 2007

Len Brown lenb at kernel.org
Wed May 30 21:37:06 PDT 2007


Monday, June 25th
Tuesday, June 26th

We have the "Garden Suite" in Les Suites Ottawa at our disposal 9AM - 6PM (thanks Intel).
The room has 20 seats.  It also has natural light:-)


Hopefully small enough that we can hold single threaded round-table discussions.

We'll also have a projector so folks can make some short presentations to get people
in the room on the same page.


I volunteer to take notes.
After the meetings I'll send them to the attendees for review.
With Mark, I also volunteer to present them at the OLS PM BOF,
and send them to linux-pm, and LWN.


I believe the 12 below will attend -- please correct me if I'm mistaken.
I didn't include folks who said "maybe," or folks who "might drop-in."
(order is arbitrary here, based on copy/paste from my mailbox)
If you want a better "personal summary" for the notes than
I conjured up here, please sent it to me and I'll get it right next time.

Len Brown <len.brown at intel.com>
	Linux Kernel ACPI Maintainer
	Willing to give a 30-min ACPI PM presentation,
	though if attendees are mostly focused on non-ACPI embedded,
	happy to skip doing so.

Dave Jones <davej at redhat.com>
	Fedora Maintainer, Linux Kernel Cpufreq Maintainer
	Len wants Dave to lead a candid discussion on
	the lessons learned from cpufreq.

Mark Gross <mgross at linux.intel.com>
	Intel OTC embedded Linux team

Tariq Shureih <tariq.shureih at intel.com>
	Intel OTC MID power policy manager

"Pedersen Klaus.K (Nokia-M/Helsinki)" <klaus.k.pedersen at nokia.com>
Igor Stoppa <igor.stoppa at nokia.com>
Poussa Sakari <sakari.poussa at nokia.com>
	Will make a short presentation about latest developments
	in Nokia N800 power management area.

Matthew Locke <matt at nomadgs.com>
"Eugeny S. Mints" <eugeny.mints at gmail.com>

	Our primary interest is unified 
	cross platforms pm API to control pm hardware (covering both x86 and embedded), 
	constraints (including  latencies) engine, pm subsystem/drivers interface, API 
	for use case driven pm.

"Scott E. Preece" <preece at motorola.com>

	I would love to see agreement on a way to structure PM so that different
	kinds of systems could better meet their different goals...

Vivek Kashyap <kashyapv at us.ibm.com>
Kenneth <kenroz at us.ibm.com>
	We are interested in cross 
	platform (PowerPC, x86) power management, pm controls and policies, 
	pm based scheduling, pm in presence of hypervisors, and power scaled 
	cpu utilization statistics.

Agenda Suggestions:

With 3+ weeks to go, it is time to firm up the agenda.
To seed the discussion, here are some notes from my mailbox:

From: Poussa Sakari <sakari.poussa at nokia.com>
we in this context mean people involved in Nokia Internet Tablet
power management development (Igor, Klaus and myself).

- Presentation of DVFS in Nokia N800 including
  - integration with clock framework and needed enhancements
  - role of the voltage and other frameworks

- Future power management challenges in Nokia Internet Tablets
  - next omap and new requirements 
  - power aware applications and policies
  - pm interface between kernel and pm aware applications

- Current budgeting in embedded USB host
  - how to balance the current flow from battery to usb peripheral
- How to use the clk fw to express constraints
  - between clocks
  - between clocks and voltages

- How to address power domains with timeouts
  - simple bitfields management example

From: Amit Kucheria
1. Unified interface to dynamic/runtime PM for laptop/server and
embedded platforms - Does it make sense?
2. Enhancements to Linux driver model to handle dynamic PM
3. Where should policy be located? Drivers or userspace or a mix of both?
4. Interfaces to userspace policy manager
5. Can x86 device drivers and bus drivers do more runtime PM? Similar
to some of the USB work being carried on by Alan/David.
6. What can be done to ensure Suspend To RAM _always_ works? Video
drivers are the big problem now, anything else?
7. Using wakeup latency as a parameter to decide how deep we want to sleep

More information about the linux-pm mailing list