SRM Notes for 19-06-2025

[glopesdev]

Agenda

• 🚀Small Issue (10 min)
• 🚅Medium Issue (20min)
• 🐢Large Issue (45min)

Discussion 🚅

• Add semantic version register #68
• SRM Notes for 27-03-2025 · harp-tech · Discussion #129
• Support variable-length registers in device.yml register specifications #116
• Define a string interfaceType and establish documentation for well-known types #118
• Use a single version number in device repositories #145
• Consider revising naming of top-level folders #146

[glopesdev]

Proposal Main Points

• Add hash of device.yml to R_VERSION register

  • Default value of 0 implies no validation should be performed (we could decide whether this should be an option in the client generated code or not)

• Use R_WHOAMI + hash to validate device interface (must be identical)

• Clarify role of R_WHOAMI (hardware target)

  • Unique hardware assembly (PCB + assembly)
    • Any firmware variation building on same hardware target gets assigned same WHOAMI
    • Some devices are "breadboard" devices made to be hacked (e.g. hobgoblin)
  • Purely virtual devices have R_WHOAMI = 0 (no hardware target)
    • Consider that whoAmI: 0 is used in the generic device to skip validation

• Where is this hash stored?

  • One proposal is R_VERSION would store a minimal hash
  • Consider whether to store R_WHOAMI contents there

• Proposed new R_VERSION register (assuming unified release pipeline) gets us down to 32 bytes:

   section Id
   HARDWARE_VERSION   :hardware     3 bytes
   CORE_VERSION       :core         3 bytes
   FIRMWARE_VERSION   :firmware     3 bytes
   SDK_IDENTIFIER     :architecture 3 bytes
   DEVICE_YML_HASH    :sha1         20 bytes