Kernel/1.9 Planning: Difference between revisions

From Yocto Project
Jump to navigationJump to search
 
(One intermediate revision by one other user not shown)
Line 6: Line 6:
*** Verify SCC/CFG to remove duplication
*** Verify SCC/CFG to remove duplication
*** Refactor intel-common & clean
*** Refactor intel-common & clean
*** LTSI update, -rt refresh, 3.14/3.19 removed
** Kernel Workflow
*** Streamline kernel configuration handling, remove meta data from kernel tree
*** Allow configuration fragments for any time of kernel, make linux-yocto a thin shim
** Kernel Packaging
*** Allow Multiple installed kernels, multi kernel build
** Remove deprecated BSPs
** Remove deprecated BSPs
*** Sugarbay, Crownbay
*** Sugarbay, Crownbay
Line 20: Line 26:
* Continue to improve Developer Workflow
* Continue to improve Developer Workflow


==1.8 Priorities==
==1.9 Commitments==
* Developer Workflow
** Silent failure, CONFIG_ missing/overridden/irrelevant - Fixed
* Automation and QA Testing
** Kernel manual guides
*** Testing the process within the Manual
*** Tooling would change and process would break
*** This will need to be pushed to 1.9
* Meta-Data cleanup
** intel-common - factor our platforms, don't just include the entire BSP scc
* BSP Updates
* linux-yocto-dev by M1
* <strike>Deployment</strike> Mark Hatle owns this now
* BSP Refactor
** Not including whole BSPs
** purge remaining MACHINE_ARCH packages for meta-intel
** runtime BSP customization (aka HDMI audio on the NUC, etc.)
* Continue Poky Tiny
** Still needs to work
** Need to add tiny variant for Intel Core BSP
* Future
** Autobuilder detection of config changes


==1.8 Commitments==
* The MUT branch will be built using linux-yocto-dev on a regular basis on the Autobuilder starting at M1.
* The versioned kernel recipe will be submitted at RC1 and before feature-freeze.
* Two weeks prior to the feature freeze date, the kernel team will send a Kernel Status Report detailing the current state of linux-yocto-dev, including the most likely kernel version (98% confidence) based on the known inputs and enumerate any unknowns. The report will include a list of known issues from the MUT builds which will reduce the surprise and scramble following the merge. The issue list is expected to be minimal given the prior runs in MUT. Factors influencing the final linux-yocto kernel version include the base kernel version for PREEMPT_RT, LTSI, stable, etc. which are beyond our control. Depending on the priorities for the release, any of these may impact the final kernel version decision. Should new or late arriving information necessitate a change to the kernel version, it will be raised with the CCB and re will need to weigh the risk (which will be much less with the MUT builds) and possible schedule slippage with the benefits (such as aligning with LTSI or PREEMPT_RT). Note that we cannot always meet all requirements of schedule and external tree support and at each release we must choose between them.


=1.9 Kernel Team Bugs=
=1.9 Kernel Team Bugs=
Line 92: Line 73:
}}
}}


==1.8 Targeted Bugs==
==1.9 Targeted Bugs==
===1.9 M1===
===1.9 M1===
====Completed Bugs====
====Completed Bugs====
Line 247: Line 228:
   |noresultsmessage="No matching bugs found."
   |noresultsmessage="No matching bugs found."
}}
}}
=Previous Release 1.8 Info=
==1.8 Priorities==
* Developer Workflow
** Silent failure, CONFIG_ missing/overridden/irrelevant - Fixed
* Automation and QA Testing
** Kernel manual guides
*** Testing the process within the Manual
*** Tooling would change and process would break
*** This will need to be pushed to 1.9
* Meta-Data cleanup
** intel-common - factor our platforms, don't just include the entire BSP scc
* BSP Updates
* linux-yocto-dev by M1
* <strike>Deployment</strike> Mark Hatle owns this now
* BSP Refactor
** Not including whole BSPs
** purge remaining MACHINE_ARCH packages for meta-intel
** runtime BSP customization (aka HDMI audio on the NUC, etc.)
* Continue Poky Tiny
** Still needs to work
** Need to add tiny variant for Intel Core BSP
* Future
** Autobuilder detection of config changes
==1.8 Commitments==
* The MUT branch will be built using linux-yocto-dev on a regular basis on the Autobuilder starting at M1.
* The versioned kernel recipe will be submitted at RC1 and before feature-freeze.
* Two weeks prior to the feature freeze date, the kernel team will send a Kernel Status Report detailing the current state of linux-yocto-dev, including the most likely kernel version (98% confidence) based on the known inputs and enumerate any unknowns. The report will include a list of known issues from the MUT builds which will reduce the surprise and scramble following the merge. The issue list is expected to be minimal given the prior runs in MUT. Factors influencing the final linux-yocto kernel version include the base kernel version for PREEMPT_RT, LTSI, stable, etc. which are beyond our control. Depending on the priorities for the release, any of these may impact the final kernel version decision. Should new or late arriving information necessitate a change to the kernel version, it will be raised with the CCB and re will need to weigh the risk (which will be much less with the MUT builds) and possible schedule slippage with the benefits (such as aligning with LTSI or PREEMPT_RT). Note that we cannot always meet all requirements of schedule and external tree support and at each release we must choose between them.

Latest revision as of 19:18, 22 May 2015

1.9 Planning

1.9 Priorities

  • BSP Updates & Consolidation
    • Update Linux Yocto
      • Verify SCC/CFG to remove duplication
      • Refactor intel-common & clean
      • LTSI update, -rt refresh, 3.14/3.19 removed
    • Kernel Workflow
      • Streamline kernel configuration handling, remove meta data from kernel tree
      • Allow configuration fragments for any time of kernel, make linux-yocto a thin shim
    • Kernel Packaging
      • Allow Multiple installed kernels, multi kernel build
    • Remove deprecated BSPs
      • Sugarbay, Crownbay
      • FRI2, NUC, eMenlow
      • Crystalforest, Jasperforest
    • Incorporate Quark Core BSP
    • Investigate Silvermont as part of Core2
  • Runtime BSP Customization
    • Graphics
    • Audio (HDMI Audio on NUC)
  • Continue Poky Tiny
    • Still needs to work
    • Need to add tiny variant for Intel Core BSP
  • Continue to improve Developer Workflow

1.9 Commitments

1.9 Kernel Team Bugs

1.9 Bug Status by Milestone

1.9 
VERIFIED (1)
1
1.9 M3 
VERIFIED (1)
1
total2

1.9 Open Bug Priority by Milestone

no bugzilla tickets were found

1.9 Weekly Activity

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0

1.9 Resolved Bugs

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0

1.9 Targeted Bugs

1.9 M1

Completed Bugs

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0

Scheduled Bugs

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0

1.9 M2

Completed Bugs

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0

Scheduled Bugs

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0

1.9 M3

Completed Bugs

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0
Darren Hart
8069linux-yocto 3.14 build failures with gcc 5.2normalMedium+1.9 M3Darren HartVERIFIED
 1      0

Scheduled Bugs

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0

1.9 M4

Completed Bugs

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0

Scheduled Bugs

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0

1.9 Backlog

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0

Previous Backlog (!1.9,!Future)

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0

Meta-Intel Backlog (!Future)

IDSummary (3 tasks) SeverityPMilestoneAssigneeStatusWhiteboardE
 0      0
Chee Yang
13548meta-intel do_testimage tasks failednormalMedium5.99Chee YangIN PROGRESS IMPLEMENTATION
 1      0
Unassigned
11816secure boot implementation using systemd-boot bootloaderenhancementMedium+5.99UnassignedNEW4
12755Separate boot config recipes are at odds with setting APPEND at the image levelnormalMedium+5.99UnassignedNEW
 2      4

Previous Backlog

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0

Future

IDSummarySeverityPMilestoneAssigneeStatusWhiteboardE
 0      0

Previous Release 1.8 Info

1.8 Priorities

  • Developer Workflow
    • Silent failure, CONFIG_ missing/overridden/irrelevant - Fixed
  • Automation and QA Testing
    • Kernel manual guides
      • Testing the process within the Manual
      • Tooling would change and process would break
      • This will need to be pushed to 1.9
  • Meta-Data cleanup
    • intel-common - factor our platforms, don't just include the entire BSP scc
  • BSP Updates
  • linux-yocto-dev by M1
  • Deployment Mark Hatle owns this now
  • BSP Refactor
    • Not including whole BSPs
    • purge remaining MACHINE_ARCH packages for meta-intel
    • runtime BSP customization (aka HDMI audio on the NUC, etc.)
  • Continue Poky Tiny
    • Still needs to work
    • Need to add tiny variant for Intel Core BSP
  • Future
    • Autobuilder detection of config changes

1.8 Commitments

  • The MUT branch will be built using linux-yocto-dev on a regular basis on the Autobuilder starting at M1.
  • The versioned kernel recipe will be submitted at RC1 and before feature-freeze.
  • Two weeks prior to the feature freeze date, the kernel team will send a Kernel Status Report detailing the current state of linux-yocto-dev, including the most likely kernel version (98% confidence) based on the known inputs and enumerate any unknowns. The report will include a list of known issues from the MUT builds which will reduce the surprise and scramble following the merge. The issue list is expected to be minimal given the prior runs in MUT. Factors influencing the final linux-yocto kernel version include the base kernel version for PREEMPT_RT, LTSI, stable, etc. which are beyond our control. Depending on the priorities for the release, any of these may impact the final kernel version decision. Should new or late arriving information necessitate a change to the kernel version, it will be raised with the CCB and re will need to weigh the risk (which will be much less with the MUT builds) and possible schedule slippage with the benefits (such as aligning with LTSI or PREEMPT_RT). Note that we cannot always meet all requirements of schedule and external tree support and at each release we must choose between them.