WeirdUtils/.github/workflows/wp24b2-inplace-gc.yml

name: Build WeirdPerformance 2.4-B2 in-place GC test

on:
  push:
    branches:
      - test/wp24b2-gc-inplace-sweep
    paths:
      - 'experiments/wp24b-gc/**'
      - '.github/workflows/wp24b2-inplace-gc.yml'
  workflow_dispatch:

permissions:
  contents: read

jobs:
  build:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: mlugg/setup-zig@v2
        with:
          version: '0.17.0-dev.1970+67f39b551'
      - name: Build and ABI-validate x86 DLL
        working-directory: experiments/wp24b-gc
        run: |
          set -euo pipefail
          zig build --fetch
          python3 - <<'PY'
          from pathlib import Path
          matches = list(Path("zig-pkg").glob("zhook-*/src/zhook.zig"))
          if len(matches) != 1:
              raise SystemExit(f"expected one pinned zhook source, found {len(matches)}")
          p = matches[0]
          text = p.read_text(encoding="utf-8")
          old = "var patch: [MAX_STOLEN]u8 = .{0x90} ** MAX_STOLEN;"
          new = "var patch: [MAX_STOLEN]u8 = @splat(0x90);"
          if old not in text:
              raise SystemExit("expected pinned zhook repeat initializer not found")
          p.write_text(text.replace(old, new), encoding="utf-8")
          PY

          zig build -Doptimize=small
          DLL=zig-out/bin/weirdperformance_gc24b2.dll
          test -s "$DLL"
          file "$DLL" | tee BINARY_INFO.txt
          file "$DLL" | grep -Eq 'PE32.*Intel (80386|i386)'
          strings "$DLL" | grep -q '2.4-B2-gc-inplace-safe-sweep'

          if command -v llvm-objdump >/dev/null 2>&1; then
            llvm-objdump -d --x86-asm-syntax=intel "$DLL" > ABI_DISASM.txt
          else
            objdump -d -M intel "$DLL" > ABI_DISASM.txt
          fi

          python3 - <<'PY'
          import re
          from pathlib import Path

          asm = Path("ABI_DISASM.txt").read_text(encoding="utf-8", errors="replace").lower()

          def require(pattern, label):
              if not re.search(pattern, asm, re.S):
                  raise SystemExit(f"ABI validation failed: {label}")

          def callback_snippet(target, window):
              m = re.search(rf"mov\s+edx,\s*0x{target}.{{0,240}}?push\s+0x([0-9a-f]+)", asm, re.S)
              if not m:
                  raise SystemExit(f"ABI validation failed: callback for 0x{target} not found")
              cb = m.group(1).lstrip("0") or "0"
              pos = asm.find(cb + ":")
              if pos < 0:
                  raise SystemExit(f"ABI validation failed: callback body {cb} not found")
              return asm[pos:pos+window]

          # Collector callback: same proven B1 code shape, L copied from ECX.
          collector = callback_snippet("6f7340", 900)
          if not re.search(r"mov\s+esi,\s*ecx", collector):
              raise SystemExit("ABI validation failed: collector callback does not consume L from ECX")

          # B2 lua_close has no helper call before callOriginal: Zig can legally
          # leave the incoming fastcall ECX untouched and call the trampoline
          # directly. Accept either explicit preservation or verified no write
          # to ECX before the first call instruction.
          close = callback_snippet("6f6ef0", 800)
          first_call = re.search(r"\bcall\b", close)
          if not first_call:
              raise SystemExit("ABI validation failed: lua_close trampoline call missing")
          before_call = close[:first_call.start()]
          ecx_saved = re.search(
              r"(?:mov\s+(?:esi|edi|ebx|ebp|eax|edx),\s*ecx|push\s+ecx|mov\s+(?:dword ptr\s*)?\[[^\]]+\],\s*ecx)",
              before_call,
          )
          ecx_written = re.search(
              r"\b(?:mov|lea|xor|and|or|add|sub|imul|inc|dec|pop)\s+ecx\b",
              before_call,
          )
          if not ecx_saved and ecx_written:
              print("lua_close callback snippet:")
              print(close)
              raise SystemExit("ABI validation failed: lua_close clobbers incoming ECX before trampoline")

          # Native helper ABI observed directly in WoW 5875:
          # lua_gc_remove_objects: ECX=L, EDX=&list, stack arg all=0.
          require(
              r"lea\s+edx,\s*\[edi\s*\+\s*0x14\].{0,200}?"
              r"mov\s+eax,\s*0x6f7210.{0,160}?"
              r"mov\s+ecx,\s*esi.{0,160}?"
              r"push\s+0x0.{0,100}?call\s+eax",
              "rootudata lua_gc_remove_objects ABI",
          )

          # At least one dynamic-list remove call must load L into ECX.
          require(
              r"mov\s+eax,\s*0x6f7210.{0,240}?mov\s+ecx,\s*e[a-z]{2}.{0,200}?call\s+eax",
              "chunk lua_gc_remove_objects fastcall callsite",
          )

          require(
              r"mov\s+eax,\s*0x6f72f0.{0,160}?"
              r"mov\s+ecx,\s*esi.{0,120}?"
              r"xor\s+edx,\s*edx.{0,100}?call\s+eax",
              "lua_gc_sweep_all_lists fastcall ABI",
          )

          for addr, label in (
              ("6f73e0", "lua_gc_full_collection"),
              ("6f7370", "lua_gc_shrink_memory"),
              ("6f7080", "luaCallUserDataGC"),
          ):
              require(
                  rf"mov\s+eax,\s*0x{addr}.{{0,180}}?mov\s+ecx,\s*e[a-z]{{2}}.{{0,120}}?call\s+eax",
                  f"{label} ECX ABI",
              )

          print("ABI validation: PASS")
          PY

          sha256sum "$DLL" | tee SHA256SUMS.txt
      - uses: actions/upload-artifact@v4
        if: always()
        with:
          name: WeirdPerformance-2.4-B2-GC-InPlace-Safe-Sweep
          path: |
            experiments/wp24b-gc/zig-out/bin/weirdperformance_gc24b2.dll
            experiments/wp24b-gc/SHA256SUMS.txt
            experiments/wp24b-gc/BINARY_INFO.txt
            experiments/wp24b-gc/ABI_DISASM.txt
          if-no-files-found: warn