...
Rationale
- Separate EDK2 kernel from silicon code
- Increase modularity of system.
- Decrease code complexity for security.
Scope
- Scope of the project including limitations.
Goals/Objectives
Detailed description of the Goals of project.
SAN19 Presentation https://static.sched.com/hosted_files/linaroconnectsandiego/11/SAN1-508%20Open%20Source%20UEFI%20Firmware%20V3.pdf
Rationale
- Many enterprise companies have a lot of hardware products, the life cycle of hardware can be more than 10 years, especially for telecoms industry. Linux has been used widely in this market, Linux kernel has an LTS management project, so it can match the long term support requirements, this proposal is to work towards the same with EDK2.
- Intel effectively has LTS concept with its UDKxxxx releases, we wish to replicate this on ARM.
- Decreased code complexity allows for easier audits and backports of essential fixes to core code.
- Increase the usage of standardized interfaces between layers in the firmware.
Scope
- EDK2 "kernel code only"
- Silicon and platform code is out of scope of this proposal (except SBSA QEMU Machine as a reference implementation)
Goals/Objectives
- Phase 1
- Separate EDK2 kernel from silicon code
- Separate Silicon code from platform code
- Increase the modularity of the code, make advanced features installable extras.
- Phase 2
- Establish LTS team for EDK2 in conjunction with Intel
- Backport CVE (and other security patches) to EDK2 kernel code (and relevant open source silicon/platform)
Target Hardware Platforms
- Implement for SBSA QEMU Machine as the simplest hardware platform
Deliverables
- List of final outputs of project
Validation Plan
How the projects success with be tested.
Resourcing
...
- Patches for EDK2 to create the minimal platform concept on ARM
- Regular releases tied to stable tag on EDK2
- Selected releases pushed to LTS project for maintenance.
Validation Plan
- Server ACS to be run against firmware on QEMU SBSA Machine
- Other relevant testsuites to be investigated
Resourcing
- Phase 1
- 1 Futurewei LT engineer + 1 additional member engineer
- Fast server class machine for running QEMU based CI
- Member target machines(s) for minimal firmware testing.
- Phase 2
- 2 Member engineers from Linaro + engineers from Intel for LTS team
- Multiple server class machines for QEMU based CI.
- Member target machine(s) for LTS firmware testing.
High-level Timeline
Expected timeline for project
- Pre-project preperation SAN19-BUD20 cycle
- Project kickoff BUD20
- Phase 1 implementation BUD20 + six months cycle
- Phase 2 implementation Post BUD20 connect + six months cycle.
High-level Risks and Mitigation Plans
...
- Only high-level risks which need to be taken into consideration when approving a Project Proposal should be noted here. Detailed risks and issues are tracked during project delivery
| Risk | Mitigation | |
|---|---|---|
| The sky is green | prepare stocks of dulux blueIntel does not want to form cross platform LTS team | Increase the resources in people needed for phase 2 implementation |
Reference
Any external references.