Source library

Memory region allocator — red team v4 (panel re-review of v5)

docs/ps1/memory-region-allocator-red-team-v4.md

This is the website shelf page for docs/ps1/memory-region-allocator-red-team-v4.md. It exists because the project has hundreds of Markdown files and the website should not pretend the interesting work only happened in the polished essays.

Why this file matters

Active PS1 reference. This is active engineering documentation for the PS1 port. It is close enough to the code that stale claims here become real bugs.

Same five-reviewer panel reconvenes for the third pass. v5 closed 19 of v3’s 22 graded findings; this review focuses on new issues introduced by the v5 fixes themselves. Loop 2 of the goal-driven iteration.

File facts

Field Value
Source path docs/ps1/memory-region-allocator-red-team-v4.md
Lines 404
Approx. words 2302
Code fences 2

Outline

  • Memory region allocator — red team v4 (panel re-review of v5)
  • Reviewer 1 — Pat, Embedded Systems Veteran
  • Closed in v5
  • New concerns
  • 🔴 P13. assert(ps1IsMainContext()) is NOT unconditional
  • 🟠 P14. MIPS COP0 read latency for ps1IsMainContext is understated
  • 🟠 P15. CRC-32 table is 1 KB — accounted in BOOT budget?
  • 🟡 P16. The builtinunreachable() after memHalt branches
  • Reviewer 2 — Sarah, SRE / Operations
  • Closed in v5
  • New concerns
  • 🟠 S11. Per-commit CI may fail on intermediate states of the bundled PR
  • 🟠 S12. The decision tree is documentation, not enforcement
  • 🟡 S13. Tag in commit-bundle vs branch tag

Read it in context