ELF emitter: EM_NDS32 ET_EXEC, PT_LOAD per region at true addresses; readelf-validated; tests (structure + segment fidelity); MTK_FW_DIR env override

Makefile

@@ -2,7 +2,7 @@ BLOBS ?= $(wildcard /lib/firmware/mediatek/mt79*_wm.bin) \
         $(wildcard /lib/firmware/mediatek/mt79*_wa.bin) \
         $(wildcard /lib/firmware/mediatek/mt79*_rom_patch.bin)
 
-.PHONY: help extract test
+.PHONY: help extract test elf
 
 help:
 	@echo "make extract  - carve all MediaTek Connac2 blobs found on this host"
@@ -13,3 +13,6 @@ extract:
 
 test:
 	python3 -m unittest discover -s tests -v
+
+elf:
+	python3 tools/mtk_fw_elf.py $(BLOBS) -o extracted

tests/test_elf.py

@@ -0,0 +1,78 @@
+"""ELF emitter tests: structural validity + segment fidelity.
+
+Validates with readelf where available and always with a programmatic
+re-parse against the container's own region table.
+"""
+import os
+import struct
+import subprocess
+import sys
+import unittest
+from pathlib import Path
+
+REPO = Path(__file__).resolve().parent.parent
+sys.path.insert(0, str(REPO / 'tools'))
+import mtk_fw_elf as E  # noqa: E402
+
+FW = Path(os.environ.get('MTK_FW_DIR', '/lib/firmware/mediatek'))
+
+
+def blobs():
+    return sorted(FW.glob('mt79*_wm.bin')) + sorted(FW.glob('mt79*_wa.bin')) \
+        + sorted(FW.glob('mt79*_rom_patch.bin'))
+
+
+class Emitter(unittest.TestCase):
+
+    def test_structure_and_fidelity(self):
+        found = blobs()
+        if not found:
+            self.skipTest('no mediatek blobs')
+        for p in found:
+            with self.subTest(blob=p.name):
+                d = p.read_bytes()
+                if 'patch' in p.name:
+                    regions = list(E.patch_regions(d))
+                else:
+                    regions = list(E.ram_regions(d))
+                regions = [(a, dat) for a, dat in regions if dat]
+                elf = E.build_elf(regions)
+                # ehdr sanity
+                self.assertEqual(elf[:4], b'\x7fELF')
+                etype, mach, _, entry, phoff = struct.unpack_from('<HHIII', elf, 16)
+                self.assertEqual((etype, mach), (2, 167))
+                self.assertEqual(phoff, 52)
+                phentsize, phnum = struct.unpack_from('<HH', elf, 42)
+                self.assertEqual((phentsize, phnum), (32, len(regions)))
+                # segment fidelity
+                for i, (addr, dat) in enumerate(regions):
+                    p_type, p_off, p_va, p_pa, p_filesz, p_memsz = \
+                        struct.unpack_from('<IIIIII', elf, phoff + i * 32)
+                    self.assertEqual((p_type, p_va, p_pa), (1, addr, addr))
+                    self.assertEqual((p_filesz, p_memsz), (len(dat),) * 2)
+                    self.assertEqual(elf[p_off:p_off + p_filesz], dat)
+
+    def test_readelf_accepts(self):
+        found = blobs()
+        if not found:
+            self.skipTest('no mediatek blobs')
+        if subprocess.run(['which', 'readelf'], capture_output=True).returncode:
+            self.skipTest('readelf unavailable')
+        import tempfile
+        for p in found[:4]:
+            with self.subTest(blob=p.name):
+                d = p.read_bytes()
+                regs = list(E.patch_regions(d) if 'patch' in p.name
+                            else E.ram_regions(d))
+                elf = E.build_elf(regs)
+                with tempfile.NamedTemporaryFile(suffix='.elf') as f:
+                    f.write(elf)
+                    f.flush()
+                    r = subprocess.run(['readelf', '-l', f.name],
+                                       capture_output=True, text=True)
+                    self.assertEqual(r.returncode, 0, r.stderr)
+                    self.assertIn('LOAD', r.stdout)
+
+
+if __name__ == '__main__':
+    unittest.main()

tests/test_extract.py

@@ -5,13 +5,14 @@ Golden values observed from linux-firmware snapshot 2026-04-20; when a value
 here fails against a newer snapshot, the blob changed — update the golden
 value *and* note it in the changelog, don't silently absorb it.
 """
+import os
 import subprocess
 import sys
 import unittest
 from pathlib import Path
 
 REPO = Path(__file__).resolve().parent.parent
-FW = Path('/lib/firmware/mediatek')
+FW = Path(os.environ.get('MTK_FW_DIR', '/lib/firmware/mediatek'))
 
 # filename -> (chip_id, eco, n_region, hidden_trailer_string_or_None)
 GOLDEN_WM = {

tools/mtk_fw_elf.py

@@ -0,0 +1,98 @@
+#!/usr/bin/env python3
+"""Emit a spec-correct 32-bit LE ELF (e_machine=EM_NDS32) from a MediaTek
+Connac2 WiFi firmware container, one PT_LOAD segment per region at true
+load addresses.
+
+Structure is validated by readelf; disassembly requires Ghidra (official
+NDS32 module) or an Andes nds32le-elf toolchain — Debian binutils dropped
+NDS32. Entry point is unknown pre-RE (e_entry=0); set it in your tool after
+import.
+
+NON_DL regions are included as loadable segments by default; they are
+host-side data per the driver, not downloaded to the MCU.
+"""
+import struct
+import sys
+from pathlib import Path
+
+sys.path.insert(0, str(Path(__file__).resolve().parent))
+from mtk_fw_extract import PATCH_HDR, PATCH_SEC, REGION, TRAILER  # noqa: E402
+
+EM_NDS32 = 167
+EHDR = struct.Struct('<16sHHIIIIIHHHHHH')  # 52 bytes
+PHDR = struct.Struct('<IIIIIIII')          # 32 bytes
+PT_LOAD, PF_RWX = 1, 7
+
+
+def ram_regions(d: bytes):
+    """Yield (addr, data) for each region of a RAM container."""
+    t = TRAILER.unpack_from(d, len(d) - TRAILER.size)
+    n = t[2]
+    table = len(d) - TRAILER.size - n * REGION.size
+    off = 0
+    for i in range(n):
+        r = REGION.unpack_from(d, table + i * REGION.size)
+        addr, ln = r[4], r[5]
+        data = d[off:off + ln]
+        off += ln
+        yield addr, data
+
+
+def patch_regions(d: bytes):
+    """Yield (addr, data) for each section of a patch container."""
+    h = PATCH_HDR.unpack_from(d, 0)
+    n = h[9]
+    for i in range(n):
+        s = PATCH_SEC.unpack_from(d, PATCH_HDR.size + i * PATCH_SEC.size)
+        typ, offs, size = s[0], s[1], s[2]
+        addr = s[3]
+        yield addr, d[offs:offs + size]
+
+
+def build_elf(regions) -> bytes:
+    regions = [(a, dat) for a, dat in regions if dat]
+    n = len(regions)
+    phoff = EHDR.size
+    data_off = phoff + n * PHDR.size
+    eh = EHDR.pack(
+        b'\x7fELF' + bytes([1, 1, 1, 0]) + b'\x00' * 8,  # ELFCLASS32/LSB/v1
+        2, EM_NDS32, 1,
+        regions[0][0] if regions else 0,   # e_entry (first region; see doc)
+        phoff, 0, 0,
+        EHDR.size, PHDR.size, n,
+        40, 0, 0)
+    out = bytearray(eh)
+    offset = data_off
+    for addr, dat in regions:
+        out += PHDR.pack(PT_LOAD, offset, addr, addr, len(dat), len(dat),
+                         PF_RWX, 0x1000)
+        offset += len(dat)
+    for _, dat in regions:
+        out += dat
+    return bytes(out)
+
+
+def main():
+    if len(sys.argv) < 2:
+        sys.exit(f'usage: {sys.argv[0]} <blob> [<blob>...] [-o outdir]')
+    args, outdir = [], Path('extracted')
+    it = iter(sys.argv[1:])
+    for a in it:
+        if a == '-o':
+            outdir = Path(next(it))
+        else:
+            args.append(a)
+    for path in args:
+        p = Path(path)
+        d = p.read_bytes()
+        regions = patch_regions(d) if 'patch' in p.name else ram_regions(d)
+        elf = build_elf(regions)
+        out = outdir / p.stem / f'{p.stem}.elf'
+        out.parent.mkdir(parents=True, exist_ok=True)
+        out.write_bytes(elf)
+        n_seg = struct.unpack_from('<H', elf, 44)[0]  # e_phnum
+        print(f'{p.name} -> {out} ({len(elf)} bytes, {n_seg} PT_LOAD)')
+
+
+if __name__ == '__main__':
+    main()