The Agenda and the Minutes for this meeting can also be found at
https://wiki.linaro.org/WorkingGroups/PowerManagement/Meetings/2010-07-28


Attendees : Amit Arora, Amit Kucheria, Bobby Bhatacharia, Srinivas
Kalaga, Vishwanath Sripathy

Minutes:
 * Review action items from last meeting
   o ACTION: Bobby to investigate android contraints (e.g. display keep-alive)
     Will target it for the next week.

   o ACTION: Srinivas to provide details of where he believes
userspace - kernel interaction is required.
     pm_qos is low prio .. so this may take a few weeks for this action
     This may take a few weeks for this action. Currently at low
priority (since pm_qos work is of low prio)

   o ACTION: Srinivas to add ftrace item to task list and send patches to linaro
     Srinivas has been working on this in last few days. We should see
something soon on this.

   o ACTION: Amit K. to inform ARM about multi-core availability
     We are not getting any multi-core boards. Vishwa should have one.
Now, the question is - will one board be enough ?
     Bobby to see possibility of getting more boards for Linaro. Not
easy, but will consider the request open.

   o ACTION: ARM to discuss giving out powertop tool
     ARM doesn't have a PowerTOP tool, so to speak, to give out.
Though they have an Eclipse based tool, which is very much an ongoing
work. Can work on making it available. But, no ETA as of now.

   o ACTION: TI to elaborate on requirement of cpuidle instrumentation task
     Vishwa : The currently "known" c-states sleep time and latency
information on OMAP boards are NOT optimal. There is a requirement for
some kind of instrumentation to measure the s/w and the h/w latencies.
Later is difficult part. And once we can get the optimal latencies, we
will have to work on optimizing them.
     NEWACTION : Vishwa to create a blue print for this. It should
mention the codepaths to be measured (eg. wakeup latencies, idle loop
to idle loop latencies for various c-states) and other details.

   o ACTION: ARM to share  internal  instrumentation flow
            (BAB: we might also align with Linaro on workload discussions)
     Bobby to make this available. But, it might take a couple of months.



 * HW availability
   o Amit Arora is following up to expedite the process to acquire the
development boards. May take some more time.
   o Vishwa has access to couple of OMAP3/4 SDP boards. Amit Kucheria
mentioned that it may be a good idea to get the publically available
boards, like the Beagleboards or the PANDA blades. Vishwa to find out
about this.
   o Amit Kucheria has a Freescale i.mx51 babbage board and a beagleboard

 * Git trees on git.linaro.org
   o We will have to host a couple of git trees in git.linaro.org
   o To start with, we may need two trees for the tools - for powertop
and powerdebug (a new tool).
   o Need access to the host system and some process needs to be followed.
   o Once we start hosting the trees, we will have to send a mail to
linaro-dev mailing list, requesting everyone to try out the code.
   o When we have patches for kernel, we may have to host a separate
kernel git tree for Power Management, which should get merged with the
linaro kernel git tree.

 * gobby / ubuntu
   o Amit Kucheria recommended everyone to start using this tool,
which is a cross-platform collaboration tool and will make things like
taking minutes much easier (everyone will be able to contribute to it)

 * Blueprint status
   o Tools blueprint : Amit Arora gave the update on the work he has
been doing on powertop and powerdebug tools. He has a few patches
ready, which he plans to host in a git tree. For the time being, we
plan to host this code as tarballs at some location, which will help
others to test it.
   o Master blueprint : Amit Kucheria had split the master blue print
into "infrastructure" and "tools" blueprints. More such blueprints
will need to be created as we go forward.
   o There are some known issues with Ondemand governor, which Vishwa
and AmitK will take it to the the community (LKML/linux-pm and make
sure to copy right people). Will need to share the constraints and ask
for feedback on how its best to implement some of the things in the
current framework.


* Others
  o Linaro's first released will be based on 2.6.35. Hence, all our
work also should be inline to this release.


ACTION ITEMS:

Carried Forward from past :
-------------------------

 * CLOSED :
   o ACTION: TI to elaborate on requirement of cpuidle instrumentation task
     This has given birth to a new ACTION item, discussed below.

 * ACTIVE :
   o ACTION: Bobby to investigate android contraints (e.g. display keep-alive)
   o ACTION: Srinivas to provide details of where he believes
userspace - kernel interaction is required.
   o ACTION: Srinivas to add ftrace item to task list and send patches to linaro
   o ACTION: Amit K. to inform ARM about multi-core availability
   o ACTION: ARM to discuss giving out powertop tool

 * DORMANT :
   o ACTION: ARM to share  internal  instrumentation flow
      (BAB: we might also align with Linaro on workload discussions)
     Might take couple of months.

New Items:
---------
   o ACTION: Vishwa to create a new blue print for the instrumentation
of cpuidle related work
   o ACTION: Amit Arora to create tarballs of his current work and
share it internally for testing
   o ACTION: Amit Kucheria and Vishwa to get inputs from community on
the issues related to Ondemand governor.



Thanks!
Amit Arora

_______________________________________________
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev

Reply via email to