2026-02-06 Time Contribution Pitch

[qor-lb]

Location

Online via teams: https://teams.microsoft.com/meet/47122962657518?p=lOXuhe7sD38agkkRO8

Participants

• @mikehaller
• @odra
• @basheerFZ
• @anmittag
• @AThums
• @nhzpc122
• @daeyoung-jeong-lge
• @RolandJentschETAS
• @MiWoMBTI
• @seokheehan
• @gordon9901
• @mwx2015
• @FScholPer
• @Kriszhu007
• @BlueXMarv3l
• @lavrovvalera
• @ramceb
• @umaucher
• @vinodreddy-g
• @qor-lb

Agenda

2. 9:00 - 9:15 Feature Request for Time #910 - @lavrovvalera / @qor-lb
3. 9:15 - 9:45 [Contribution Proposal] Time (BMW) #1938
4. 9:45 - 10:15 [Contribution Proposal] Time (ETAS) #2216
5. 10:15 - 10:45 [Contribution Proposal] Time (ecarx) #2426
6. 10:45 - 11:30 Grooming [Contribution Request] Time #1765

Meeting Minutes

1. [Contribution Proposal] Time (BMW) #1938 : https://eclipse-score.github.io/inc_time/pr-6/time/index.html - @lavrovvalera

   • Q Aram K: Is conversion between different time-points from different clock domains considered? (navigation has uc: system has global time, sensors have individual time - they need to be able to convert between these time points - e.g. from ptp to absolute and the other way arround)
   • A: currently not - we would require precise offsets and could enable that
   • Q Aram K: Ring buffer vs shared memory: what is the benefit of having an additional buffer in the frontend to the business logic of the application?
   • A: This is optional and only required if the latency of the receive of IPC has a too high
   • Q @BjoernAtBosch : Is the implementation already available?
   • A: not yet - should be available in 1-2 weeks
   • Q @RolandJentschETAS : how is the time to other ecus synchronized?
   • A: by utilizing the gPTP protocol on the network layer
   • Q @RolandJentschETAS : how is the error handling done to ensure that the time source is not corrupted/available?
   • A: It depends on the time base - this additionally requires a local clock available that we can trust

2. [Contribution Proposal] Time (ETAS) #2216 : Slides - @mikehaller & @BjoernAtBosch

   • Q @qor-lb: Does the impelemenation
   • A: no not yet - but could be brought to such a state
   • I @umaucher: The crucial part is the gptp deamon part to get a qualified timestamp
   • Q @lavrovvalera: does the current implementation update the hardware counter (i.e. the avb use case for video frames)?
   • A: no not yet
   • Q @lavrovvalera: where do you plan to do the qualification of the timestamp?
   • A: most likely in the ptp daemon
   • Q @lavrovvalera: are you planning to add additional meta data into the shared memory towards the business logic for diagnostic job use cases (e.g. what is the current synchronization delay)?
   • A: This use-case is not yet implemented.
   • Q @qor-lb: does the current solution utilize the pdpd from QNX?
   • A: no it does use the open source basis with extensions. it is not yet verified on QNX

3. [Contribution Proposal] Time (ecarx) #2426 : Slides - @BlueXMarv3l

   • Q @lavrovvalera: How are you ensuring consistency between the different apis that you expose - e.g. now and get accuracy qualified
   • A: @mwx2015
   • Q @lavrovvalera: Which of the components are asil B and which not?
   • A: The current implemenation is against QM, could be extended further
   • Q @anmittag: Is it planned that ecarx contributes also to other modules than time in s-core?
   • A: time is the starting point, maybe in the future more
   • Q @anmittag: Is the implementation only compatible with QNX 7.1 or also QNX 8.0?
   • A: Not yet tested on QNX 8, will do this in the future.

4. Grooming

Legend: https://github.com/orgs/eclipse-score/discussions/375#:~:text=Formal%20criteria%20for%20contribution%20selection

criterion	BMW	ETAS	ECARX
Alignment with the Feature Request.	++	+ (after clarification open question above)	++
Availability of the Source Code	+ (within next 2 weeks)	++ (copy right will be updated once accepted)	++
Technical Maturity	++	+	+
Initial Impact Assessment:	+/++	+/++	+ (can be updated after acceptance)
Supporting Artifacts:	+	+	+
Availability for QNX and Linux	++	++ (only build integration missing)	++

Next steps:

• define feature team to drive this further: @lavrovvalera
  • poll for meeting slot
  • invite via public eclipse calendar
  • @FScholPer to create slack channel for time and publish in public channel: https://app.slack.com/client/T02MS1M89UH/C0ADTA4SKUH
  • Time interest group: https://github.com/orgs/eclipse-score/discussions/1465
• find consensus within 4 weeks after availability of BMW source code

[mwx2015]

@qor-lb @lavrovvalera
“Q : How are you ensuring consistency between the different apis that you expose - e.g. now and get accuracy qualified”
Answer from ECARX :
We provided two APIs:

• AccuracyQualifier GetAccuracyQualifier()
• TimePointQualifier GetTimePointQualifier()

to satisfy the requirements as follow:

feat_req__time__vehicle_time_acc_qual_api
feat_req__time__vehicle_time_time_pt_qual

The API ' Now()' return only one timestamp value, can't ensure consistency between the 'Now()' and ' GetAccuracyQualifier() ' .

Our solution is as follows:
Return both Timestamp and synchronization status in the Now() interface, in details:

a) When the protocol stack calculates the global time point, it will synchronously update the status accordingly

b) If the synchronization times out, only the synchronization status will be updated accordingly