Tech: Reliability, availability, serviceability, and dependability (RASD) tech-rasd@lists.riscv.org

In order to promote development of RISC-V in server domain, we need a complete specification to guide implementation of RAS in the design of SoC, firmware and OS.

Tasks in scope include:

RAS terminology interpretation:
Interpretation of RAS concept and terminology (e.g. diagnosability, recoverability, types of error).

RAS framework design:

A framework covers the full path of error handling:

• Error recording: Standard error record formats (e.g. register banks, APEI­)
• Error reporting: Error event reporting methods (e.g. exceptions, NMI, local/global interrupts)
• Error recovery: strategies adopted to handle the error (e.g. neglect/warning/recover/isolation/halt)

RAS feature support:

Engage specific RAS features into the framework:

• E2E Data protection
• error isolation
• data poisoning containment;
• advanced error reporting for PCIe