angr 9.2.136__py3-none-macosx_11_0_arm64.whl → 9.2.138__py3-none-macosx_11_0_arm64.whl

angr/analyses/s_liveness.py

@@ -2,7 +2,7 @@ from __future__ import annotations
 
 import networkx
 from ailment.expression import VirtualVariable
-from ailment.statement import Assignment
+from ailment.statement import Assignment, Call
 
 from angr.analyses import Analysis, register_analysis
 from angr.utils.ssa import VVarUsesCollector, phi_assignment_get_src
@@ -86,6 +86,8 @@ class SLivenessAnalysis(Analysis):
                 # handle assignments: a defined vvar is not live before the assignment
                 if isinstance(stmt, Assignment) and isinstance(stmt.dst, VirtualVariable):
                     live.discard(stmt.dst.varid)
+                elif isinstance(stmt, Call) and isinstance(stmt.ret_expr, VirtualVariable):
+                    live.discard(stmt.ret_expr.varid)
 
                 phi_expr = phi_assignment_get_src(stmt)
                 if phi_expr is not None:
@@ -136,6 +138,8 @@ class SLivenessAnalysis(Analysis):
             for stmt in reversed(block.statements):
                 if isinstance(stmt, Assignment) and isinstance(stmt.dst, VirtualVariable):
                     def_vvar = stmt.dst.varid
+                elif isinstance(stmt, Call) and isinstance(stmt.ret_expr, VirtualVariable):
+                    def_vvar = stmt.ret_expr.varid
                 else:
                     def_vvar = None
 

angr/analyses/s_propagator.py

@@ -4,6 +4,8 @@ import contextlib
 from collections.abc import Mapping
 from collections import defaultdict
 
+import networkx
+
 from ailment.block import Block
 from ailment.expression import (
     Const,
@@ -17,11 +19,13 @@ from ailment.expression import (
 from ailment.statement import Assignment, Store, Return, Jump
 
 from angr.knowledge_plugins.functions import Function
-from angr.code_location import CodeLocation
+from angr.code_location import CodeLocation, ExternalCodeLocation
 from angr.analyses import Analysis, register_analysis
 from angr.utils.ssa import (
     get_vvar_uselocs,
     get_vvar_deflocs,
+    has_ite_expr,
+    has_ite_stmt,
     is_phi_assignment,
     is_const_assignment,
     is_const_and_vvar_assignment,
@@ -51,9 +55,10 @@ class SPropagatorAnalysis(Analysis):
     def __init__(  # pylint: disable=too-many-positional-arguments
         self,
         subject: Block | Function,
-        func_graph=None,
+        func_graph: networkx.DiGraph | None = None,
         only_consts: bool = True,
         stack_pointer_tracker=None,
+        func_args: set[VirtualVariable] | None = None,
         func_addr: int | None = None,
     ):
         if isinstance(subject, Block):
@@ -69,6 +74,7 @@ class SPropagatorAnalysis(Analysis):
 
         self.func_graph = func_graph
         self.func_addr = func_addr
+        self.func_args = func_args
         self.only_consts = only_consts
         self._sp_tracker = stack_pointer_tracker
 
@@ -109,6 +115,11 @@ class SPropagatorAnalysis(Analysis):
         # find all vvar uses
         vvar_uselocs = get_vvar_uselocs(blocks.values())
 
+        # update vvar_deflocs using function arguments
+        if self.func_args:
+            for func_arg in self.func_args:
+                vvar_deflocs[func_arg] = ExternalCodeLocation()
+
         # find all ret sites and indirect jump sites
         retsites: set[tuple[int, int | None, int]] = set()
         jumpsites: set[tuple[int, int | None, int]] = set()
@@ -130,8 +141,12 @@ class SPropagatorAnalysis(Analysis):
 
             vvarid_to_vvar[vvar.varid] = vvar
             defloc = vvar_deflocs[vvar]
+            if isinstance(defloc, ExternalCodeLocation):
+                continue
+
             assert defloc.block_addr is not None
             assert defloc.stmt_idx is not None
+
             block = blocks[(defloc.block_addr, defloc.block_idx)]
             stmt = block.statements[defloc.stmt_idx]
             r, v = is_const_assignment(stmt)
@@ -147,56 +162,94 @@ class SPropagatorAnalysis(Analysis):
             if v is not None:
                 src_varids = {vvar.varid if vvar is not None else None for _, vvar in v.src_and_vvars}
                 if None not in src_varids and all(varid in const_vvars for varid in src_varids):
+                    all_int_src_varids: set[int] = {varid for varid in src_varids if varid is not None}
                     src_values = {
                         (
                             (const_vvars[varid].value, const_vvars[varid].bits)
                             if isinstance(const_vvars[varid], Const)
                             else const_vvars[varid]
                         )
-                        for varid in src_varids
+                        for varid in all_int_src_varids
                     }
                     if len(src_values) == 1:
                         # replace it!
-                        const_value = const_vvars[next(iter(src_varids))]
+                        const_value = const_vvars[next(iter(all_int_src_varids))]
                         const_vvars[vvar.varid] = const_value
                         for vvar_at_use, useloc in vvar_uselocs[vvar.varid]:
                             replacements[useloc][vvar_at_use] = const_value
 
             if self.mode == "function" and vvar.varid in vvar_uselocs:
+                if len(vvar_uselocs[vvar.varid]) <= 2 and isinstance(stmt, Assignment) and isinstance(stmt.src, Load):
+                    # do we want to propagate this Load expression if it's used for less than twice?
+                    # it's often seen in the following pattern, where propagation will be beneficial:
+                    #    v0 = Load(...)
+                    #    if (!v0) {
+                    #       v1 = v0 + 1;
+                    #    }
+                    can_replace = True
+                    for _, vvar_useloc in vvar_uselocs[vvar.varid]:
+                        if self.has_store_stmt_in_between(blocks, defloc, vvar_useloc):
+                            can_replace = False
+
+                    if can_replace:
+                        # we can propagate this load because there is no store between its def and use
+                        for vvar_used, vvar_useloc in vvar_uselocs[vvar.varid]:
+                            replacements[vvar_useloc][vvar_used] = stmt.src
+                        continue
+
                 if len(vvar_uselocs[vvar.varid]) == 1:
                     vvar_used, vvar_useloc = next(iter(vvar_uselocs[vvar.varid]))
-                    if (
-                        is_const_vvar_load_assignment(stmt)
-                        and vvar_useloc.block_addr == defloc.block_addr
-                        and vvar_useloc.block_idx == defloc.block_idx
-                        and not any(
-                            isinstance(stmt_, Store)
-                            for stmt_ in block.statements[defloc.stmt_idx + 1 : vvar_useloc.stmt_idx]
-                        )
+                    if is_const_vvar_load_assignment(stmt) and not self.has_store_stmt_in_between(
+                        blocks, defloc, vvar_useloc
                     ):
                         # we can propagate this load because there is no store between its def and use
                         replacements[vvar_useloc][vvar_used] = stmt.src
                         continue
 
                     if is_const_and_vvar_assignment(stmt):
-                        replacements[vvar_useloc][vvar_used] = stmt.src
+                        # if the useloc is a phi assignment statement, ensure that stmt.src is the same as the phi
+                        # variable
+                        assert vvar_useloc.block_addr is not None
+                        assert vvar_useloc.stmt_idx is not None
+                        useloc_stmt = blocks[(vvar_useloc.block_addr, vvar_useloc.block_idx)].statements[
+                            vvar_useloc.stmt_idx
+                        ]
+                        if is_phi_assignment(useloc_stmt):
+                            if (
+                                isinstance(stmt.src, VirtualVariable)
+                                and stmt.src.oident == useloc_stmt.dst.oident
+                                and stmt.src.category == useloc_stmt.dst.category
+                            ):
+                                replacements[vvar_useloc][vvar_used] = stmt.src
+                        else:
+                            replacements[vvar_useloc][vvar_used] = stmt.src
                         continue
 
-                elif (
-                    len(
-                        {
-                            loc
-                            for _, loc in vvar_uselocs[vvar.varid]
-                            if (loc.block_addr, loc.block_idx, loc.stmt_idx) not in (retsites | jumpsites)
-                        }
-                    )
-                    == 1
-                ):
+                else:
+                    non_exitsite_uselocs = [
+                        loc
+                        for _, loc in vvar_uselocs[vvar.varid]
+                        if (loc.block_addr, loc.block_idx, loc.stmt_idx) not in (retsites | jumpsites)
+                    ]
                     if is_const_and_vvar_assignment(stmt):
-                        # this vvar is used once if we exclude its uses at ret sites or jump sites. we can propagate it
-                        for vvar_used, vvar_useloc in vvar_uselocs[vvar.varid]:
-                            replacements[vvar_useloc][vvar_used] = stmt.src
-                        continue
+                        if len(non_exitsite_uselocs) == 1:
+                            # this vvar is used once if we exclude its uses at ret sites or jump sites. we can
+                            # propagate it
+                            for vvar_used, vvar_useloc in vvar_uselocs[vvar.varid]:
+                                replacements[vvar_useloc][vvar_used] = stmt.src
+                            continue
+
+                        if len(set(non_exitsite_uselocs)) == 1 and not has_ite_expr(stmt.src):
+                            useloc = non_exitsite_uselocs[0]
+                            assert useloc.block_addr is not None
+                            assert useloc.stmt_idx is not None
+                            useloc_stmt = blocks[(useloc.block_addr, useloc.block_idx)].statements[useloc.stmt_idx]
+                            if stmt.src.depth <= 3 and not has_ite_stmt(useloc_stmt):
+                                # remove duplicate use locs (e.g., if the variable is used multiple times by the same
+                                # statement) - but ensure stmt is simple enough
+                                for vvar_used, vvar_useloc in vvar_uselocs[vvar.varid]:
+                                    replacements[vvar_useloc][vvar_used] = stmt.src
+                                continue
 
                 # special logic for global variables: if it's used once or multiple times, and the variable is never
                 # updated before it's used, we will propagate the load
@@ -205,7 +258,11 @@ class SPropagatorAnalysis(Analysis):
                     # unpack conversions
                     while isinstance(stmt_src, Convert):
                         stmt_src = stmt_src.operand
-                    if isinstance(stmt_src, Load) and isinstance(stmt_src.addr, Const):
+                    if (
+                        isinstance(stmt_src, Load)
+                        and isinstance(stmt_src.addr, Const)
+                        and isinstance(stmt_src.addr.value, int)
+                    ):
                         gv_updated = False
                         for _vvar_used, vvar_useloc in vvar_uselocs[vvar.varid]:
                             gv_updated |= self.is_global_variable_updated(
@@ -265,6 +322,8 @@ class SPropagatorAnalysis(Analysis):
         for block_loc, tmp_and_uses in tmp_uselocs.items():
             for tmp_atom, tmp_uses in tmp_and_uses.items():
                 # take a look at the definition and propagate the definition if supported
+                assert block_loc.block_addr is not None
+
                 block = blocks[(block_loc.block_addr, block_loc.block_idx)]
                 tmp_def_stmtidx = tmp_deflocs[block_loc][tmp_atom]
 
@@ -329,6 +388,8 @@ class SPropagatorAnalysis(Analysis):
 
             start_stmt_idx = defloc.stmt_idx if block is defblock else 0  # inclusive
             end_stmt_idx = useloc.stmt_idx if block is useblock else len(block.statements)  # exclusive
+            assert start_stmt_idx is not None
+            assert end_stmt_idx is not None
 
             for idx in range(start_stmt_idx, end_stmt_idx):
                 stmt = block.statements[idx]
@@ -359,5 +420,43 @@ class SPropagatorAnalysis(Analysis):
 
         return False
 
+    def has_store_stmt_in_between(
+        self, blocks: dict[tuple[int, int | None], Block], defloc: CodeLocation, useloc: CodeLocation
+    ) -> bool:
+        assert defloc.block_addr is not None
+        assert defloc.stmt_idx is not None
+        assert useloc.block_addr is not None
+        assert useloc.stmt_idx is not None
+        assert self.func_graph is not None
+
+        use_block = blocks[(useloc.block_addr, useloc.block_idx)]
+        def_block = blocks[(defloc.block_addr, defloc.block_idx)]
+
+        # traverse the graph, go from use_block until we reach def_block, and look for Store statements
+        seen = {use_block}
+        queue = [use_block]
+        while queue:
+            block = queue.pop(0)
+
+            starting_stmt_idx, ending_stmt_idx = 0, len(block.statements)
+            if block is def_block:
+                starting_stmt_idx = defloc.stmt_idx + 1
+            if block is use_block:
+                ending_stmt_idx = useloc.stmt_idx + 1
+
+            for i in range(starting_stmt_idx, ending_stmt_idx):
+                if isinstance(block.statements[i], Store):
+                    return True
+
+            if block is def_block:
+                continue
+
+            for pred in self.func_graph.predecessors(block):
+                if pred not in seen:
+                    seen.add(pred)
+                    queue.append(pred)
+
+        return False
+
 
 register_analysis(SPropagatorAnalysis, "SPropagator")

angr/analyses/s_reaching_definitions/s_rda_model.py

@@ -15,8 +15,9 @@ class SRDAModel:
     The model for SRDA.
     """
 
-    def __init__(self, func_graph, arch):
+    def __init__(self, func_graph, func_args, arch):
         self.func_graph = func_graph
+        self.func_args = func_args
         self.arch = arch
         self.varid_to_vvar: dict[int, VirtualVariable] = {}
         self.all_vvar_definitions: dict[VirtualVariable, CodeLocation] = {}

angr/analyses/s_reaching_definitions/s_rda_view.py

@@ -4,7 +4,7 @@ import logging
 from collections import defaultdict
 
 from ailment.statement import Statement, Assignment, Call, Label
-from ailment.expression import VirtualVariable, Expression
+from ailment.expression import VirtualVariable, VirtualVariableCategory, Expression
 
 from angr.utils.ail import is_phi_assignment
 from angr.utils.graph import GraphUtils
@@ -133,6 +133,16 @@ class SRDAView:
         predicater = RegVVarPredicate(reg_offset, vvars, self.model.arch)
         self._get_vvar_by_stmt(block_addr, block_idx, stmt_idx, op_type, predicater.predicate)
 
+        if not vvars:
+            # not found - check function arguments
+            for func_arg in self.model.func_args:
+                if isinstance(func_arg, VirtualVariable):
+                    func_arg_category = func_arg.oident[0]
+                    if func_arg_category == VirtualVariableCategory.REGISTER:
+                        func_arg_regoff = func_arg.oident[1]
+                        if func_arg_regoff == reg_offset:
+                            vvars.add(func_arg)
+
         assert len(vvars) <= 1
         return next(iter(vvars), None)
 
@@ -149,6 +159,15 @@ class SRDAView:
         predicater = StackVVarPredicate(stack_offset, size, vvars)
         self._get_vvar_by_stmt(block_addr, block_idx, stmt_idx, op_type, predicater.predicate)
 
+        if not vvars:
+            # not found - check function arguments
+            for func_arg in self.model.func_args:
+                if isinstance(func_arg, VirtualVariable):
+                    func_arg_category = func_arg.oident[0]
+                    if func_arg_category == VirtualVariableCategory.STACK:
+                        func_arg_stackoff = func_arg.oident[1]
+                        if func_arg_stackoff == stack_offset and func_arg.size == size:
+                            vvars.add(func_arg)
         assert len(vvars) <= 1
         return next(iter(vvars), None)
 

angr/analyses/s_reaching_definitions/s_reaching_definitions.py

@@ -2,6 +2,7 @@ from __future__ import annotations
 
 from ailment.block import Block
 from ailment.statement import Assignment, Call, Return
+from ailment.expression import VirtualVariable
 import networkx
 
 from angr.knowledge_plugins.functions import Function
@@ -24,6 +25,7 @@ class SReachingDefinitionsAnalysis(Analysis):
         subject,
         func_addr: int | None = None,
         func_graph: networkx.DiGraph[Block] | None = None,
+        func_args: set[VirtualVariable] | None = None,
         track_tmps: bool = False,
     ):
         if isinstance(subject, Block):
@@ -39,13 +41,14 @@ class SReachingDefinitionsAnalysis(Analysis):
 
         self.func_graph = func_graph
         self.func_addr = func_addr if func_addr is not None else self.func.addr if self.func is not None else None
+        self.func_args = func_args
         self._track_tmps = track_tmps
 
         self._bp_as_gpr = False
         if self.func is not None:
             self._bp_as_gpr = self.func.info.get("bp_as_gpr", False)
 
-        self.model = SRDAModel(func_graph, self.project.arch)
+        self.model = SRDAModel(func_graph, func_args, self.project.arch)
 
         self._analyze()
 
@@ -66,6 +69,13 @@ class SReachingDefinitionsAnalysis(Analysis):
         # find all explicit vvar uses
         vvar_uselocs = get_vvar_uselocs(blocks.values())
 
+        # update vvar definitions using function arguments
+        if self.func_args:
+            for vvar in self.func_args:
+                if vvar not in vvar_deflocs:
+                    vvar_deflocs[vvar] = ExternalCodeLocation()
+            self.model.func_args = self.func_args
+
         # update model
         for vvar, defloc in vvar_deflocs.items():
             self.model.varid_to_vvar[vvar.varid] = vvar

angr/analyses/stack_pointer_tracker.py

@@ -2,6 +2,7 @@
 from __future__ import annotations
 
 from typing import Any, TYPE_CHECKING
+import contextlib
 import re
 import logging
 from collections import defaultdict
@@ -15,7 +16,6 @@ from angr.knowledge_plugins import Function
 from angr.block import BlockNode
 from angr.errors import SimTranslationError
 from .analysis import Analysis
-import contextlib
 
 try:
     import pypcode
@@ -702,21 +702,31 @@ class StackPointerTracker(Analysis, ForwardAnalysis):
             # who are we calling?
             callees = [] if self._func is None else self._find_callees(node)
             if callees:
-                if (
-                    len(callees) == 1
-                    and callees[0].info.get("is_rust_probestack", False) is True
-                    and self.project.arch.name == "AMD64"
-                ):
-                    # special-case for rust_probestack: sp = sp - rax right after returning from the call, so we need
-                    # to keep track of rax
-                    for stmt in reversed(vex_block.statements):
-                        if (
-                            isinstance(stmt, pyvex.IRStmt.Put)
-                            and stmt.offset == self.project.arch.registers["rax"][0]
-                            and isinstance(stmt.data, pyvex.IRExpr.Const)
-                        ):
-                            state.put(stmt.offset, Constant(stmt.data.con.value), force=True)
-                            break
+                if len(callees) == 1:
+                    callee = callees[0]
+
+                    if callee.info.get("is_rust_probestack", False) is True and self.project.arch.name == "AMD64":
+                        # special-case for rust_probestack: sp = sp - rax right after returning from the call, so we
+                        # need to keep track of rax
+                        for stmt in reversed(vex_block.statements):
+                            if (
+                                isinstance(stmt, pyvex.IRStmt.Put)
+                                and stmt.offset == self.project.arch.registers["rax"][0]
+                                and isinstance(stmt.data, pyvex.IRExpr.Const)
+                            ):
+                                state.put(stmt.offset, Constant(stmt.data.con.value), force=True)
+                                break
+                    elif callee.name == "__chkstk":
+                        # special-case for __chkstk: sp = sp - rax right after returning from the call, so we need to
+                        # keep track of rax
+                        for stmt in reversed(vex_block.statements):
+                            if (
+                                isinstance(stmt, pyvex.IRStmt.Put)
+                                and stmt.offset == self.project.arch.registers["rax"][0]
+                                and isinstance(stmt.data, pyvex.IRExpr.Const)
+                            ):
+                                state.put(stmt.offset, Constant(stmt.data.con.value), force=True)
+                                break
 
                 callee_cleanups = [
                     callee

angr/analyses/variable_recovery/engine_ail.py

@@ -74,17 +74,28 @@ class SimEngineVRAIL(
 
         elif dst_type is ailment.Expr.VirtualVariable:
             data = self._expr(stmt.src)
-            self._assign_to_vvar(
+            variable = self._assign_to_vvar(
                 stmt.dst, data, src=stmt.src, dst=stmt.dst, vvar_id=self._mapped_vvarid(stmt.dst.varid)
             )
 
+            if variable is not None and isinstance(stmt.src, ailment.Expr.Phi):
+                # this is a phi node - we update variable manager's phi variable tracking
+                for _, vvar in stmt.src.src_and_vvars:
+                    if vvar is not None:
+                        r = self._read_from_vvar(vvar, expr=stmt.src, vvar_id=self._mapped_vvarid(vvar.varid))
+                        if r.variable is not None:
+                            pv = self.variable_manager[self.func_addr]._phi_variables
+                            if variable not in pv:
+                                pv[variable] = set()
+                            pv[variable].add(r.variable)
+
             if stmt.dst.was_stack and isinstance(stmt.dst.stack_offset, int):
                 # store it to the stack region in case it's directly referenced later
                 self._store(
                     RichR(self.state.stack_address(stmt.dst.stack_offset)),
                     data,
                     stmt.dst.bits // self.arch.byte_width,
-                    stmt=stmt,
+                    atom=stmt.dst,
                 )
 
         else:
@@ -94,10 +105,11 @@ class SimEngineVRAIL(
         addr_r = self._expr_bv(stmt.addr)
         data = self._expr(stmt.data)
         size = stmt.size
-        self._store(addr_r, data, size, stmt=stmt)
+        self._store(addr_r, data, size, atom=stmt)
 
-    def _handle_stmt_Jump(self, stmt):
-        pass
+    def _handle_stmt_Jump(self, stmt: ailment.Stmt.Jump):
+        if not isinstance(stmt.target, ailment.Expr.Const):
+            self._expr(stmt.target)
 
     def _handle_stmt_ConditionalJump(self, stmt):
         self._expr(stmt.condition)
@@ -149,7 +161,7 @@ class SimEngineVRAIL(
                 prototype_libname = func.prototype_libname
 
         # dump the type of the return value
-        ret_ty = typevars.TypeVariable() if prototype is not None else typevars.TypeVariable()
+        ret_ty = typevars.TypeVariable()
         if isinstance(ret_ty, typeconsts.BottomType):
             ret_ty = typevars.TypeVariable()
 
@@ -218,7 +230,7 @@ class SimEngineVRAIL(
                 prototype_libname = func.prototype_libname
 
         # dump the type of the return value
-        ret_ty = typevars.TypeVariable() if prototype is not None else typevars.TypeVariable()
+        ret_ty = typevars.TypeVariable()
         if isinstance(ret_ty, typeconsts.BottomType):
             ret_ty = typevars.TypeVariable()
 

angr/analyses/variable_recovery/engine_base.py

@@ -387,7 +387,7 @@ class SimEngineVRBase(
         ) or not create_variable:
             # only store the value. don't worry about variables.
             self.vvar_region[vvar_id] = richr.data
-            return
+            return None
 
         codeloc: CodeLocation = self._codeloc()
         data = richr.data
@@ -463,10 +463,14 @@ class SimEngineVRBase(
             else:
                 typevar = self.state.typevars.get_type_variable(variable, codeloc)
             self.state.add_type_constraint(typevars.Subtype(richr.typevar, typevar))
+            # the constraint below is a default constraint that may conflict with more specific ones with different
+            # sizes; we post-process at the very end of VRA to remove conflicting default constraints.
             self.state.add_type_constraint(typevars.Subtype(typevar, typeconsts.int_type(variable.size * 8)))
 
+        return variable
+
     def _store(
-        self, richr_addr: RichR[claripy.ast.BV], data: RichR[claripy.ast.BV | claripy.ast.FP], size, stmt=None
+        self, richr_addr: RichR[claripy.ast.BV], data: RichR[claripy.ast.BV | claripy.ast.FP], size, atom=None
     ):  # pylint:disable=unused-argument
         """
 
@@ -481,19 +485,19 @@ class SimEngineVRBase(
 
         if addr.concrete:
             # fully concrete. this is a global address
-            self._store_to_global(addr.concrete_value, data, size, stmt=stmt)
+            self._store_to_global(addr.concrete_value, data, size, stmt=atom)
             stored = True
         elif self._addr_has_concrete_base(addr) and (parsed := self._parse_offsetted_addr(addr)) is not None:
             # we are storing to a concrete global address with an offset
             base_addr, offset, elem_size = parsed
-            self._store_to_global(base_addr.concrete_value, data, size, stmt=stmt, offset=offset, elem_size=elem_size)
+            self._store_to_global(base_addr.concrete_value, data, size, stmt=atom, offset=offset, elem_size=elem_size)
             stored = True
         else:
             if self.state.is_stack_address(addr):
                 stack_offset = self.state.get_stack_offset(addr)
                 if stack_offset is not None:
                     # fast path: Storing data to stack
-                    self._store_to_stack(stack_offset, data, size, stmt=stmt)
+                    self._store_to_stack(stack_offset, data, size, atom=atom)
                     stored = True
 
         if not stored:
@@ -504,21 +508,21 @@ class SimEngineVRBase(
             codeloc = self._codeloc()
             if existing_vars:
                 for existing_var, _ in list(existing_vars):
-                    self.variable_manager[self.func_addr].remove_variable_by_atom(codeloc, existing_var, stmt)
+                    self.variable_manager[self.func_addr].remove_variable_by_atom(codeloc, existing_var, atom)
 
             # storing to a location specified by a pointer whose value cannot be determined at this point
-            self._store_to_variable(richr_addr, size, stmt=stmt)
+            self._store_to_variable(richr_addr, size)
 
     def _store_to_stack(
-        self, stack_offset, data: RichR[claripy.ast.BV | claripy.ast.FP], size, offset=0, stmt=None, endness=None
+        self, stack_offset, data: RichR[claripy.ast.BV | claripy.ast.FP], size, offset=0, atom=None, endness=None
     ):
-        if stmt is None:
+        if atom is None:
             existing_vars = self.variable_manager[self.func_addr].find_variables_by_stmt(
                 self.block.addr, self.stmt_idx, "memory"
             )
         else:
             existing_vars = self.variable_manager[self.func_addr].find_variables_by_atom(
-                self.block.addr, self.stmt_idx, stmt
+                self.block.addr, self.stmt_idx, atom
             )
         if not existing_vars:
             variable = SimStackVariable(
@@ -562,7 +566,7 @@ class SimEngineVRBase(
                     var,
                     offset_into_var,
                     codeloc,
-                    atom=stmt,
+                    atom=atom,
                 )
 
             # create type constraints
@@ -673,9 +677,7 @@ class SimEngineVRBase(
                 self.state.add_type_constraint(typevars.Subtype(store_typevar, typeconsts.TopType()))
                 self.state.add_type_constraint(typevars.Subtype(data.typevar, store_typevar))
 
-    def _store_to_variable(
-        self, richr_addr: RichR[claripy.ast.BV], size: int, stmt=None
-    ):  # pylint:disable=unused-argument
+    def _store_to_variable(self, richr_addr: RichR[claripy.ast.BV], size: int):
         addr_variable = richr_addr.variable
         codeloc = self._codeloc()
 

angr/analyses/variable_recovery/engine_vex.py

@@ -74,7 +74,7 @@ class SimEngineVRVEX(
         size = stmt.data.result_size(self.tyenv) // 8
         r = self._expr(stmt.data)
 
-        self._store(addr_r, r, size, stmt=stmt)
+        self._store(addr_r, r, size, atom=stmt)
 
     def _handle_stmt_StoreG(self, stmt):
         guard = self._expr(stmt.guard)
@@ -82,7 +82,7 @@ class SimEngineVRVEX(
             addr = self._expr_bv(stmt.addr)
             size = stmt.data.result_size(self.tyenv) // 8
             data = self._expr(stmt.data)
-            self._store(addr, data, size, stmt=stmt)
+            self._store(addr, data, size, atom=stmt)
 
     def _handle_stmt_LoadG(self, stmt):
         guard = self._expr(stmt.guard)

angr/analyses/variable_recovery/variable_recovery_fast.py

@@ -20,7 +20,8 @@ from angr.knowledge_plugins import Function
 from angr.sim_variable import SimStackVariable, SimRegisterVariable, SimVariable, SimMemoryVariable
 from angr.engines.vex.claripy.irop import vexop_to_simop
 from angr.analyses import ForwardAnalysis, visitors
-from angr.analyses.typehoon.typevars import Equivalence, TypeVariable, TypeVariables
+from angr.analyses.typehoon.typevars import Equivalence, TypeVariable, TypeVariables, Subtype, DerivedTypeVariable
+from angr.analyses.typehoon.typeconsts import Int
 from .variable_recovery_base import VariableRecoveryBase, VariableRecoveryStateBase
 from .engine_vex import SimEngineVRVEX
 from .engine_ail import SimEngineVRAIL
@@ -500,6 +501,26 @@ class VariableRecoveryFast(ForwardAnalysis, VariableRecoveryBase):  # pylint:dis
                 for tv in sorted_typevars[1:]:
                     self.type_constraints[self.func_typevar].add(Equivalence(sorted_typevars[0], tv))
 
+        # remove default constraints with size conflicts
+        for func_var in self.type_constraints:
+            var_to_subtyping: dict[TypeVariable, list[Subtype]] = defaultdict(list)
+            for constraint in self.type_constraints[func_var]:
+                if isinstance(constraint, Subtype) and isinstance(constraint.sub_type, TypeVariable):
+                    var_to_subtyping[constraint.sub_type].append(constraint)
+
+            for constraints in var_to_subtyping.values():
+                if len(constraints) <= 1:
+                    continue
+                default_subtyping_constraints = set()
+                has_nondefault_subtyping_constraints = False
+                for constraint in constraints:
+                    if isinstance(constraint.super_type, Int):
+                        default_subtyping_constraints.add(constraint)
+                    elif isinstance(constraint.super_type, DerivedTypeVariable) and constraint.super_type.labels:
+                        has_nondefault_subtyping_constraints = True
+                if has_nondefault_subtyping_constraints:
+                    self.type_constraints[func_var].difference_update(default_subtyping_constraints)
+
         self.variable_manager[self.function.addr].ret_val_size = self.ret_val_size
 
         self.delayed_type_constraints = None