Weekly Progress Review

From OMAPpedia

Welcome to the weekly progress review wikipage of the LDC Power Management (PM) domain team.

Contents 1 30-Aug-2010 Meeting 2 16-Aug-2010 Meeting 3 02-Aug-2010 Meeting 4 26-Jul-2010 Meeting 5 19-Jul-2010 Meeting 6 12-Jul-2010 Meeting 7 28-Jun-2010 Meeting 8 21-Jun-2010 Meeting

[edit] 30-Aug-2010 Meeting

Attendees Kevin, Santosh, Jean, Thara, Anand S, Avinash, Rajendra, Vishwa, Partha

Agenda

• System PM camp updates - Anand
• SmartReflex & DVFS Status - Thara
• ES2.0 upgrade for Kevin - Anand
• Status update from Jean

Minutes

• System PM camp updates - Anand
  • First feature group being tested - MPU CSWR, MPU OFF, SmartReflex
  • Being tested at platform drivers, MM drivers, Ducati & System use case level
  • SmartReflex seems to be working fine with the usecases ran so far
  • MPU CSWR/ MPU OFF also working fine; no major issues yet - testing ongoing
  • CORE CSWR & DVFS testing is happening
• Posting ES2.0 validated PM feature patches
  • Will be posted as features get validated
  • TI can post patches after validation. It would be preferable to have someone outside TI to validate them before merging.
  • Here is a tentative sequence for the first feature group - MPU CSWR, MPU OFF & SmartReflex
    • ES2.0 register database - Rajendra will specify the date
    • Next power domain cleanup patches; may need more review and most likely may not gate MPU CSWR/ OFF - This may be posted later
    • Next will be MPU CSWR/ MPU OFF - Santosh
    • Next will be the latency numbers profiled on TI product
    • Next will be the latency number profiled on mainline and information about factors impacting latency calculation.
  • CPUIdle latency profiling
    • Needs to be carried out before merging the patches; need to have real numbers
    • Can start with conservative numbers but need to graduate to real numbers soon
    • Would be preferred to have the numbers profiled for mainline support
    • But then there should be a list of which parameters or SW layers that impact the latencies to a large extent
    • Also one needs to mention how those numbers have been arrived at
    • Currently, TI is profiling latencies after the features are validated at system level
    • Need to plan for profiling on mainline and preparing the database on entities impacting latencies - this may require significant time
    • For early availability for people to review, one may need to post the patches and in parallel work on the profiling on mainline - Anand will get back with a timeline proposal.
• ES2.0 upgrade for Kevin - Anand
  • Kevin has a socketed PAB and Panda; no Blaze
    • So, an ES2.0 sample would do - Vijay to follow-up
    • Check for Blaze availability - Anand to follow-up
• Lauterbach for Kevin - Anand
  • Need to provide one that works on OMAP3 & 4 - Anand & Vijay
  • CRO with digital probes would be also helpful - Anand & Vijay
• Issues faced in HWMOD & runtime PM usage in debugging
  • Intention here is to gather ideas how we can improve them
  • Multiple SW layers leading to multiple indirections
  • Becoming overall more complex to debug
  • HWMOD database bugs add to complexities in debugging
  • Need to understand that the complexity is going pushed from drivers to HWMOD layer; so though it appears difficult now (being the starting phase), it will help in the long run.
  • Further the complexity is being put in more managable way through HWMOD
• SmartReflex & DVFS Status - Thara
  • Posted patches for SmartReflex for OMAP3 & 4
  • Posted patched for DVFS for OMAP3
  • Working on fixing the comments this week
  • Next will be hosting them on a tree
• Status update from Jean
  • Pushed 2 patches
  • Few comments on LO; not on LKML yet
  • Priorities for Jean to be clarified - Vijay to comment. But here is how it is looking now
    • SW tracing Framework
    • Adding more tracepoints in CPUIdle would be helpful
    • Vishwa & Jean can work together to have some more tracing points in SRAM
    • Jean can do some initial investigation work

[edit] 16-Aug-2010 Meeting

Attendees

Anand, Jean, Satya, Avinash, Thara, Vibhore

Agenda

• Updates to pm_domain_all mailing list
• Status check of different projects
• Review of last week's action items

Minutes

• Misc
  • Anand to update pm_domain_all mailing list to reflect correct addresses
  • All, please send an email to Anand if you want any change in your email address as listed in pm_domain_all (like addition or change)
• Status check of different projects
  • SmartReflex from Thara
    • V2 of SmartReflex posted to LO (covers OMAP3); it addresses all review comments received so far
    • OMAP4 specific extensions SmartReflex will be posted on 17-Aug
  • DVFS from Thara
    • V1 of DVFS patches (covering OMAP3) will be posted on 18-Aug
    • DVFS extensions for OMAP4 will be posted on 19-Aug
    • Dependent patches are SmartReflex and voltage layer patches which have been posted on 13-Aug
  • Updates from Jean
    • Wiki has been updated with lots of good information
    • Tracing for DVFS, CPUIdle, Clocks & power domains has been implemented
    • Submitted to LKML for review comments
    • UI tool is up and running
    • Jean to maintain a separate wikipage (linked from his current page) to list weekly progress
    • Patch cleaning is in progress

[edit] 02-Aug-2010 Meeting

Attendees

Vishwa, Thara, Kevin, Avinash, Patrick, Cedric, Mike T, Benoit, Jean, Santosh

Agenda

• OPP Patches upstreaming
• Device wake up latency clarification
• Review of actions items

Minutes

• OPP Patches upstreaming
  • opp layer patches are not upstreamed yet. Nishant had concerns over cpufreq seperation. It will be good to have opp layer in
  • Tony's tree (in next branch), so that all dependent patches (DVFS, SR, constraints) can be pushed.
  • Kevin to review the patches and finalise on OPP patches

• Device wake up latency clarification
  • In the latest implementation of omap_pm_set_max_dev_wakeup_lat, one more parameter has been added for this function
  • (device *req_dev). Need to understand
    • why this new param is added
    • what happens if this param is NULL.
  • Vibhore to check with Paul on this.

• Actions items review:
  • Regulator Framework
    • Benoit and Thara to review Thomas patches - on going
  • OMAP4 Core DVFS
    • Vishwa & Thara to check if DDR can be scaled between 100 & 200 MHz along with OPP (just a try to check the sequence)

yet to be tried. Thara to provide the dates.

• • Constraints Part
    • Validation of DSS Constraints - Partha to share the DSS patches under review with Vibhore for testing - no update
    • Vibhore to send out the RFC patches for review after addressing Vishwa's comments (validated on OMAP3) by 27-Jul and those validated on OMAP4 by 30-Jul - OMAP3 patches sent for Review on July 30.
    • Vibhore to test camera, USB, IVA clock information on OMAP4 first and then on OMAP3 - no update

• • SmartReflex
    • Issue seen with OPP50 where the SmartReflex compensated voltage is found to be more than that without compensation
      • Vishwa/Thara to check this gap with the Silicon validation team - no update
      • Patrick to check the availability of updated HWMOD values with the HW team - nominal samples to be available by 13Aug and **** split samples by 1st of Sept

• • HWMOD Updates from Partha
    • no updates

• Misc
  • latest plist implementation has apis to get the least priority node. Vibhore to use this API for device constraint framework
  • Jean Pihet has joined PM team and will be focusing on PM Debug and Trance enhancement activities namely
    • provide profle and debug of power management framework
      • Adding profiling in CPUFreq, CPUIdle
      • Adding HW Trace capabilities for OMAP4 using STM
    • Benoit/Patrick to provide more inputs to Jean
  • Kevin/Benoit would like to see the progress on HWMod patches for Timers/GPIO
  • HWMod adaptation for Driver should be upstreamed along with hwmod database update instead of big bang merge.

[edit] 26-Jul-2010 Meeting

Attendees

Anand, Thomas, Mike, Partha, Benoit, Vibhore, Jerome, Vishwa, Rajendra,

Agenda

• Issues with Runtime PM framework
• HWMOD issue
• Review status of various projects
• Review of last week's action items
• SmartReflex for L24.8 release

Minutes

• Review status of various projects
  • Updates from Thomas
    • Short write up provided in Wiki - others can review
    • Next step is to reintegrate SmartReflex (expected to be complicated)
      • Thomas sent the patches for review - need inputs from Kevin, Thara & Benoit; please use the pm_domain_all mailing list
      • Benoit to review by 30-Jul
  • Updates from Vishwa on SRF replacement (OPP & Constraints Part), DVFS & SmartReflex
    • DVFS & OPP Part
      • OMAP4 MPU, IVA & CORE DVFS working with limited set of drivers
      • For CORE DVFS, L3 frequency is changed but no change in DDR frequency due known HW issue
      • Vishwa & Thara to check if DDR can be scaled between 100 & 200 MHz along with OPP (just a try to check the sequence)
      • Also validated on OMAP3
    • Constraints Part
      • Device latency: Framework has been validated for OMAP3430 CORE & MPU
      • Validation of PER in progress;
      • Validation of DSS will happen - Partha to share the DSS patches under review with Vibhore for testing
      • Framework code is not expected to change for OMAP4; only database will change
      • Vibhore to send out the RFC patches for review after addressing Vishwa's comments (validated on OMAP3) by 27-Jul and those validated on OMAP4 by 30-Jul
      • Vibhore to test camera, USB, IVA clock information on OMAP4 first and then on OMAP3
      • If any issues seen on OMAP3, Vibhore will check with Benoit.
      • Interrupt latency APIs have been posted to LO after validation on OMAP3; OMAP4 validation will be done after OMAP4 CORE CSWR is available
    • Shweta has validated the throughput latency constraints on OMAP3; OMAP4 validation is in progress
    • SmartReflex
      • Validated on VDD MPU, CORE & IVA for most of the OPPs
      • Issue seen with OPP50 where the SmartReflex compensated voltage is found to be more than that without compensation
      • Vishwa to check this gap with the Silicon validation team
      • Patrick to check the availability of updated HWMOD values with the HW team
      • Received comments on OPP layer from Nishanth Menon; will be posted after DVFS is done
  • HWMOD Updates from Partha
    • Drivers at risk of completion by Jul end - McBSP, MMC, USB EHCI & OHCI, Display
    • HW supervised transition of clock domain not working for SGX; needs more investigation
  • Runtime PM limitations observed
    • Not able to call Runtime PM APIs from ISR context - faced in UART, GPIO; Partha to send an email
  • Updates from Patrick
    • Anand to review it and align with Partha before deciding on PAB vs. 3630 SDP use for OMAP4 PM measurements - still pending
    • omapconf tool update is still in progress - targeting 26-Jul; next version targeted before the camp

[edit] 19-Jul-2010 Meeting

Attendees

Anand, Thomas, Mike, Partha, Vishwa, Patrick, Rajendra, Vijay

Agenda

• Review status of various projects
• Review of last week's action items

Minutes

• Review of status of various projects
  • Updates from Thomas
    • Short write up provided in Wiki - others can review
    • Next step is to reintegrate SmartReflex (expected to be complicated)
      • Thomas will send proposal patches by 26-Jul
      • At present Thomas is based on Kevin's PM branch. Since his development work doesn't require OMAP4 support at this juncture, it's OK to develop that on OMAP3 first.
      • The latest SmartReflex patches (validated on ES2.0) will be picked up later.
  • Updates from Vishwa on SRF replacement (OPP & Constraints Part), DVFS & SmartReflex
    • Constraints Part
      • Vibhore has sent the first set of patches for latency constraints; some issue with baseline used delaying validation
      • Vishwa reviewed and provided comments
      • Shweta has started working throughput constraints
    • SmartReflex
      • Thara has rebased SmartReflex patches on kernel.org
      • Vijay to check with Nishanth Menon if he can review & ack it before Kevin comes back from vacation.
  • HWMOD Updates from Partha
    • Patches have been lined up on Kevin's tree for pushing
    • Planning to complete most of the platform and power driver HWMOD update by Jul end.
    • Partha and Rajendra to review & ack HWMOD patches for faster acceptance
    • Partha completed pulling of the latest HWMOD related patches from Paul, Benoit & others
    • Drivers at risk of completion by Jul end - USB EHCI & OHCI, Runtime PM changes for UART, Display
  • Submission for 2.6.37
    • Let's aim to push most of the PM contents in 2.6.37 merge window;
    • For that let's not drop the ball on reviewing patches and lining them up to avoid last minute rush
  • CORE CSWR updates from Rajendra
    • Work will continue after 2.6.35 migration this week
    • Also need to complete MPU CSWR & MPU OFF by Jul end
  • Updates from Patrick
    • Specified OMAP4 PM measurements details
    • Anand to review it and align with Partha before deciding on PAB vs. 3630 SDP use for OMAP4 PM measurements
    • omapconf tool update is in progress - targeting 23-Jul
  • Kevin's pm-wip/omap4-hwmod Branch
    • Need to come up with a plan to provide better OMAP4 support on Kevin's pm-wip/omap4-hwmod branch - Vijay, Satya, Anand & Kevin - Still open
    • Need to check whether HWMOD database + runtime APIs can be posted to LO to unblock developers
• Review of last week's action items
  • Covered above as a part of the updates on different projects

[edit] 12-Jul-2010 Meeting

Attendees

Anand, Partha, Thomas, Mike, Vishwa, Patrick, Avinash, Satya, Santosh

Agenda

• ES2.0 Wakeup Updates
• Review status of various projects
• Review of last week's action items

Minutes

• ES2.0 Wakeup Updates
  • Very good progress so far
  • Driver functionality up and running
  • PM frameworks validated; SmartReflex and DVFS validation in progress
  • DVFS is using SRF replacement for OPP; also been tested on OMAP3430
  • MPU CSWR testing in Suspend/ Resume & CPUIdle path done with PM defconfig; MPU OFF testing is ongoing; on track for Jul end completion
• Review of status of various projects
  • Updates from Thomas
    • On leave last week
    • Basic integration with regulator framework OK
    • Next step is to reintegrate SmartReflex - expected to be complicated
    • Discussed some basic ideas with Kevin; should be able to progress as planned
    • Will pick up latest SmartReflex patches (validated on ES2.0) once Vishwa and Thara post them
  • Updates from Vishwa on SRF replacement (OPP Part), DVFS & SmartReflex
    • Covered above
    • Patches will be posted once validation for VDD_IVA & VDD_CORE voltage domains is completed
  • Updates from Partha on HWMOD
    • Pushing out HWMOD changes for key modules is in progress - DSS, DMA, MUSB
    • DMA patches have been tested on OMAP3 only; OMAP4 testing is in progress; the intention is to make them available early for reviewing
    • Planning to post them by 18-Jul
    • Partha to pull in the latest HWMOD related patches from Paul, Benoit, Kevin for the next LDC release; needed to provide a stable test bench for others to validate their changes; timeline needs to decided
  • Updates from Santosh on MPU CSWR & MPU OFF
    • MPU CSWR testing in Suspend/ Resume & CPUIdle path done with PM defconfig
    • MPU OFF testing is ongoing; on track for Jul end completion
    • Latency measurements planned
  • Updates from Patrick
    • Updation of omapconf tool in progress
  • Kevin's PM Branch
    • Need to come up with a plan to provide better OMAP4 support on Kevin's PM branch - Vijay, Satya, Anand & Kevin
• Review of last week's action items
  • Updates embedded in the minutes of respective meetings below; please look for "12-Jul"

[edit] 28-Jun-2010 Meeting

Attendees

Anand, Partha, Avinash, Kevin, Vibhore, Thomas, Mike, Satya, Rajendra, Balaji, Thara, Vishwa, Santosh

Agenda

• Review of last week's action items
• Progress review

Minutes

• Vacation Plans
  • Kevin on vacation 01-Jul through ~ 27-Jul, 2010
  • Thomas on vacation 05-Jul through 10-Jul, 2010; Available on email
• Thermal Sensors
  • Are the thermal sensors available on ES1.0 boards? -> yes.
• Hosting developers' own git trees
  • It may take more time for the official decision to come through
  • Thara is facing problems with hosting her SmartReflex patches
  • Santosh will create temporary branch on dev.omapzoom.org (base is Kevin's pm-sr branch)
    • 12-Jul: Closed - Rebased on kernel.org and made working; no need to have a separate branch
  • Per Kevin, this is an important fundamental requirement - it should be available to developers.
  • Per Santosh, one should not host a tree just for few patches - there should be a balance
  • Individual project hosting is free, branding (like TI) may cost money.
  • Mike to conduct a tutorial on how to use the free hosting sites e.g. http://github.com/mturquette/linux-omap-android-
  • Thara to schedule on 30-Jun or 01-Jul.
    • 12-Jul: Closed - Mike gave the tutorial
• SRF Replacement - OPP Part
  • Patch for removing references to VDD ID posted to LO
  • Finished upto Extend voltage layer to have a plist for each vdd to do the device use counting action item in the plan
  • Needs re-planning as VDD CORE DVFS for OMAP3 is being pushed out
  • Thara to consider the latest patch series sent by Kevin - in connection with initialization
    • 12-Jul: Closed; cross VDD dependencies is work in progress
• SRF Replacement - Constraints Part
  • On track as per the published plan which deals with domain latency constraints
  • For throughput constraints, Vibhore has a dependency on Thara's DVFS support
  • Anand to plan this with Thara, Vibhore & Vishwa
    • 12-Jul: Closed - Latest schedule updates reflect this
  • OMAP3 will be validated first then OMAP4
• SR Voltage control migration
  • Updates are available on the omappedia wiki - so far on track
  • Kevin & Benoit are OK with the high level proposal
  • Thomas to add the proposal to the wiki.
    • 12-Jul: Open - Thomas will work on it this week
• HWMOD Updates
  • Control of Sysconfig for handling error conditions
    • Partha & Hema to post a patch proposing HWMOD API changes
      • 12-Jul: Open - Planning to post it by 15-Jul
    • This will start the conversation - doesn't guarantee its acceptance
  • GPIO
    • HWMOD final series posted; only selective cleanup needed as explained by Kevin

[edit] 21-Jun-2010 Meeting

Attendees

Anand, Benoit, Vikram, Thomas, Avinash, Kevin, Mike, Vibhore, Vishwa & few others (sorry, I might have missed)

Agenda

• Why this call?
• Finalize the weekly call timing
• Project assignments
• Progress review

Minutes

• Intention of setting up this weekly call
  • More and more developers are contributing to LO PM that includes TIers, contractors, open source consultants, and in future multiple community developers
  • Need a forum to discuss PM domain ideas, share learnings, seek guidance on issues faced etc.
  • Since LDC development also feeds into product lines directly, need to review progress of various PM projects on LDC, identify risks & escalate, if needed
  • But this is not intended to be a product call

• Preferred Timing
  • Weekly call will be on Monday 9 PM India = 10:30 AM Dallas = 8:30 AM Pacific = 6:30 PM Nice time
  • Will try to send out the agenda in advance + topics added on the fly

• Tracking
  • All tracking will be via external wiki http://omappedia.org/wiki/Power_Management_Domain_Wiki
  • You can use pm_domain_all@list.ti.com for quickly accessing all PM developers, e.g. scheduling a call
  • NOT meant to replace communication on open source mailing lists

• Present Developers and the Projects they are Working on
  • Rajendra Nayak - HWMOD Framework enhancements, ES2.0 Wakeup, CORE OFF development
  • Partha Basak - Device driver HWMOD adaptations
  • Thara Gopinath - SRF Replacement - OPP part, SmartReflex
  • Vishwa Sripathy - Helping Thara on SRF replacement part, contributing on other areas
  • Vibhore Vardhan - SRF replacement - Constraints part
  • Santosh Shilimkar - MPU OFF, CORE OFF, other areas, LDC tree handling
  • Avinash Mahadeva - Ramp up, Power domain cleanup, need to check other areas as well
  • Lesly Manuel, Shweta Gulati, Teerth Reddy - In the process of completing OMAP3630 PM development, will decide their assignments soon
  • Benoit Cousson - PM SW architect, HWMOD framework enhancements, Policy management
  • Kevin Hilman - Overall guidance on LO PM SW development
  • Thomas Petazzoni - SR Voltage Control Migration to Regulator Framework Drivers
  • Anand Sawant - Overall management, planning, risk handling
  • Jean Pihet - To be announced by next week
  • Richard Woodruff - Guidance on multiple areas related to PM SW & HW
  • Mike Turquette & Vikram Pandita - Working on Android, "users" and active contributors to LO PM

NOTE: This is in NO way the final and accurate assignment. People are encouraged to contribute to any PM (and non PM) areas based on their interests. So, the assignment will do change with time.

• Where shall we keep the docs?
  • It should be maintained on wiki.
  • Preferably avoid PDFs for ongoing developments; you can use them for completed topics or trainings etc.

• Got an idea? Got some time to contribute?
  • The wikipage http://omappedia.org/wiki/New_Ideas_Under_Consideration lists a number of improvements, project ideas.
  • You can pick up from that list or add more
  • Projects from this list, once planned, will be tracked and maintained on their wikis.

• Need a place to keep developers' own git trees
  • Today, Keith Deacon controls the permission and is project based.
  • Given that more and more developers are contributing, it may not scale well. So, the preferred way would be anyone with TI account creating on their own.
  • TI may need to pay more to gitorius.org but should be OK.
  • Done by many companies. TI can even do a branding of the same.
  • Mike will send an email detailing this further - 22-Jun-2010
    • 28-Jun-2010: Mile sent the email
  • Anand to check with Vijay on further course of actions
    • 28-Jun-2010: Discussed with Vijay, Satya and management; need some moreunderstanding before taking the final decision.

• Progress Review
  • Try to update your wiki offline so that we spend less time tracking
  • The call may need to be extended by 30 minutes in case we run over.
  • There may be separate additional meetings needed to review progress in few specific areas
  • No progress review for 21-Jun-2010 due to lack of time.