Initial commit — leak-hunt project complete

Five bugs identified and patched in retail Asheron's Call client:
- v3b: palette refcount over-increment (3-byte NOP at two sites)
- v5: RenderSurface PurgeResource no-op stub (vtable slot 2 thunk)
- v11: two dangling-pointer crash guards (NULL-check + reorder)
- v14: CEnvCell::Destroy ClipPlaneList leak (18-byte JMP to cleanup thunk)
- v22: unpacker stale-pointer SEH guard (whole-function __try/__except)

All five ship in leakfix.dll (117 KB, SHA d282f23c…) which is loaded
by acclient.exe at process start via PE import table patching by
tools/install_leakfix.py.

Controlled 15-client fleet soak: unpatched control died at 26h with
palette exhaustion; all 14 patched clients survived past that point
and reached ≥5-day uptime.

Residual ~15 MB/h growth traced to d3d9.dll's internal slab allocator
(260KB surface backing buffers retained after Release). See REPORT.md
§10 for the full investigation; conclusion is that it's unfixable from
outside d3d9.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
acbot 2026-05-23 21:05:17 +02:00
commit 57b5e43d0e
199 changed files with 1648333 additions and 0 deletions

140
tools/position_host_scan.py Normal file
View file

@ -0,0 +1,140 @@
"""position_host_scan.py <pid>
For every Position instance found in a live PID:
- Read 64 bytes before and 128 bytes after the vtable pointer.
- Look at offsets -32, -28, -24, -20, -16, -12, -8, -4 (might be
vtables of host containers / property wrappers / list nodes).
- Look at offsets +96 (the byte right after Position ends), +100,
+104, +108 (vtables of "next instance in the same container").
- For any DWORD in those slots that falls in acclient.exe code/.rdata
range (0x00400000..0x007FFFFF), bucket it.
- Print top vtables seen at each offset.
This is the diagnostic that breaks open which OWNER class is
heap-allocating Positions but never freeing them.
"""
import ctypes
import ctypes.wintypes as wt
import struct
import sys
from collections import Counter, defaultdict
POSITION_VT = 0x00797910
ACCLIENT_MIN = 0x00400000
ACCLIENT_MAX = 0x00900000 # generous upper bound for .text/.rdata
PROCESS_VM_READ = 0x10
PROCESS_QUERY_INFORMATION = 0x400
class MBI(ctypes.Structure):
_fields_ = [('BaseAddress', ctypes.c_void_p),
('AllocationBase', ctypes.c_void_p),
('AllocationProtect', wt.DWORD),
('PartitionId', wt.WORD),
('RegionSize', ctypes.c_size_t),
('State', wt.DWORD),
('Protect', wt.DWORD),
('Type', wt.DWORD)]
k = ctypes.windll.kernel32
k.OpenProcess.argtypes = [wt.DWORD, wt.BOOL, wt.DWORD]
k.OpenProcess.restype = wt.HANDLE
k.CloseHandle.argtypes = [wt.HANDLE]
k.CloseHandle.restype = wt.BOOL
k.ReadProcessMemory.argtypes = [wt.HANDLE, wt.LPCVOID, wt.LPVOID,
ctypes.c_size_t, ctypes.POINTER(ctypes.c_size_t)]
k.ReadProcessMemory.restype = wt.BOOL
k.VirtualQueryEx.argtypes = [wt.HANDLE, ctypes.c_void_p,
ctypes.POINTER(MBI), ctypes.c_size_t]
k.VirtualQueryEx.restype = ctypes.c_size_t
# Offsets relative to Position vtable address. Positive = into Position;
# negative = backwards into containing header.
OFFSETS = [-32, -28, -24, -20, -16, -12, -8, -4,
+96, +100, +104, +108, +112, +116, +120, +124]
def scan(pid):
h = k.OpenProcess(PROCESS_VM_READ | PROCESS_QUERY_INFORMATION, False, pid)
if not h:
print(f"OpenProcess err={ctypes.get_last_error()}"); return 1
# Histogram offset -> Counter(value)
hist = defaultdict(Counter)
total = 0
# Track Position-at-N density: how many Positions hit at exact
# stride 96 from a prior Position (suggests array/list of Positions).
stride_hits = Counter()
mbi = MBI()
addr = 0
while k.VirtualQueryEx(h, addr, ctypes.byref(mbi), ctypes.sizeof(mbi)):
pr = mbi.Protect & 0xff
if (mbi.State == 0x1000 and mbi.Type == 0x20000
and pr in (0x04, 0x40)):
buf = (ctypes.c_ubyte * mbi.RegionSize)()
sz = ctypes.c_size_t(0)
if k.ReadProcessMemory(h, mbi.BaseAddress, buf,
mbi.RegionSize, ctypes.byref(sz)):
data = bytes(buf[:sz.value])
end = (len(data) // 4) * 4
# Find all Position vtable hits in this region first
pos_offs_in_region = []
for off in range(0, end, 4):
if struct.unpack_from("<I", data, off)[0] == POSITION_VT:
pos_offs_in_region.append(off)
for off in pos_offs_in_region:
total += 1
# context
for delta in OFFSETS:
target = off + delta
if 0 <= target <= len(data) - 4:
v = struct.unpack_from("<I", data, target)[0]
# only record if it looks like a code pointer (vtable)
if ACCLIENT_MIN <= v <= ACCLIENT_MAX:
hist[delta][v] += 1
else:
# also track NULL specifically
if v == 0:
hist[delta][0] += 1
# Stride: if Position at off and another at off+96, +192 etc.
pos_set = set(pos_offs_in_region)
for off in pos_offs_in_region:
for k_stride in (96, 192, 288):
if off + k_stride in pos_set:
stride_hits[k_stride] += 1
addr = (mbi.BaseAddress or 0) + mbi.RegionSize
if addr >= 0x80000000:
break
print(f"PID {pid}: {total} Position instances scanned")
print()
print("Top non-zero vtables at each offset relative to Position vt:")
for delta in OFFSETS:
c = hist[delta]
non_null = [(v, n) for v, n in c.items() if v != 0]
non_null.sort(key=lambda x: -x[1])
null_n = c.get(0, 0)
if not non_null and null_n == 0:
continue
print(f"\n offset {delta:+4d}: (null={null_n}/{total} = "
f"{100.0*null_n/max(1,total):.1f}%)")
for v, n in non_null[:6]:
pct = 100.0 * n / total
print(f" 0x{v:08x} count={n:>6} ({pct:.1f}%)")
print()
print("Position-at-stride hits (suggests array/list of Positions):")
for s, n in stride_hits.most_common():
print(f" stride {s}B: {n} adjacent Position pairs")
k.CloseHandle(h)
return 0
if __name__ == "__main__":
sys.exit(scan(int(sys.argv[1])))