[mypyc] Introduce LoadMem (python#9211)

Relates mypyc/mypyc#741 Introduce LoadMem IR to read a memory address and convert it into a designated-type value. (The address would mostly be in py_ssize_t for now.) Part of efforts to implement len primitives: *(Py_ssize_t*)(ob + size_offset).
johnjohndoe · Jul 28, 2020 · 997b5f2 · 997b5f2
1 parent 65186ae
commit 997b5f2
Show file tree

Hide file tree

Showing 5 changed files with 56 additions and 4 deletions.
diff --git a/mypyc/analysis/dataflow.py b/mypyc/analysis/dataflow.py
@@ -9,7 +9,7 @@
     BasicBlock, OpVisitor, Assign, LoadInt, LoadErrorValue, RegisterOp, Goto, Branch, Return, Call,
     Environment, Box, Unbox, Cast, Op, Unreachable, TupleGet, TupleSet, GetAttr, SetAttr,
     LoadStatic, InitStatic, PrimitiveOp, MethodCall, RaiseStandardError, CallC, LoadGlobal,
-    Truncate, BinaryIntOp
+    Truncate, BinaryIntOp, LoadMem
 )
 
 
@@ -208,6 +208,9 @@ def visit_load_global(self, op: LoadGlobal) -> GenAndKill:
     def visit_binary_int_op(self, op: BinaryIntOp) -> GenAndKill:
         return self.visit_register_op(op)
 
+    def visit_load_mem(self, op: LoadMem) -> GenAndKill:
+        return self.visit_register_op(op)
+
 
 class DefinedVisitor(BaseAnalysisVisitor):
     """Visitor for finding defined registers.

diff --git a/mypyc/codegen/emitfunc.py b/mypyc/codegen/emitfunc.py
@@ -12,7 +12,7 @@
     LoadStatic, InitStatic, TupleGet, TupleSet, Call, IncRef, DecRef, Box, Cast, Unbox,
     BasicBlock, Value, MethodCall, PrimitiveOp, EmitterInterface, Unreachable, NAMESPACE_STATIC,
     NAMESPACE_TYPE, NAMESPACE_MODULE, RaiseStandardError, CallC, LoadGlobal, Truncate,
-    BinaryIntOp
+    BinaryIntOp, LoadMem
 )
 from mypyc.ir.rtypes import (
     RType, RTuple, is_tagged, is_int32_rprimitive, is_int64_rprimitive
@@ -464,6 +464,13 @@ def visit_binary_int_op(self, op: BinaryIntOp) -> None:
         self.emit_line('%s = %s%s %s %s%s;' % (dest, lhs_cast, lhs,
                                                op.op_str[op.op], rhs_cast, rhs))
 
+    def visit_load_mem(self, op: LoadMem) -> None:
+        dest = self.reg(op)
+        src = self.reg(op.src)
+        # TODO: we shouldn't dereference to type that are pointer type so far
+        type = self.ctype(op.type)
+        self.emit_line('%s = *(%s *)%s;' % (dest, type, src))
+
     # Helpers
 
     def label(self, label: BasicBlock) -> str:

diff --git a/mypyc/ir/ops.py b/mypyc/ir/ops.py
@@ -25,7 +25,7 @@
 from mypyc.ir.rtypes import (
     RType, RInstance, RTuple, RVoid, is_bool_rprimitive, is_int_rprimitive,
     is_short_int_rprimitive, is_none_rprimitive, object_rprimitive, bool_rprimitive,
-    short_int_rprimitive, int_rprimitive, void_rtype
+    short_int_rprimitive, int_rprimitive, void_rtype, is_c_py_ssize_t_rprimitive
 )
 from mypyc.common import short_name
 
@@ -1347,6 +1347,31 @@ def accept(self, visitor: 'OpVisitor[T]') -> T:
         return visitor.visit_binary_int_op(self)
 
 
+class LoadMem(RegisterOp):
+    """Reading a memory location
+
+    type ret = *(type*)src
+    """
+    error_kind = ERR_NEVER
+
+    def __init__(self, type: RType, src: Value, line: int = -1) -> None:
+        super().__init__(line)
+        self.type = type
+        # TODO: for now we enforce that the src memory address should be Py_ssize_t
+        #       later we should also support same width unsigned int
+        assert is_c_py_ssize_t_rprimitive(src.type)
+        self.src = src
+
+    def sources(self) -> List[Value]:
+        return [self.src]
+
+    def to_str(self, env: Environment) -> str:
+        return env.format("%r = load_mem %r :: %r*", self, self.src, self.type)
+
+    def accept(self, visitor: 'OpVisitor[T]') -> T:
+        return visitor.visit_load_mem(self)
+
+
 @trait
 class OpVisitor(Generic[T]):
     """Generic visitor over ops (uses the visitor design pattern)."""
@@ -1453,6 +1478,10 @@ def visit_load_global(self, op: LoadGlobal) -> T:
     def visit_binary_int_op(self, op: BinaryIntOp) -> T:
         raise NotImplementedError
 
+    @abstractmethod
+    def visit_load_mem(self, op: LoadMem) -> T:
+        raise NotImplementedError
+
 
 # TODO: Should this live somewhere else?
 LiteralsMap = Dict[Tuple[Type[object], Union[int, float, str, bytes, complex]], str]

diff --git a/mypyc/ir/rtypes.py b/mypyc/ir/rtypes.py
@@ -301,6 +301,10 @@ def is_int64_rprimitive(rtype: RType) -> bool:
     return rtype is int64_rprimitive
 
 
+def is_c_py_ssize_t_rprimitive(rtype: RType) -> bool:
+    return rtype is c_pyssize_t_rprimitive
+
+
 def is_float_rprimitive(rtype: RType) -> bool:
     return isinstance(rtype, RPrimitive) and rtype.name == 'builtins.float'
 

diff --git a/mypyc/test/test_emitfunc.py b/mypyc/test/test_emitfunc.py
@@ -10,7 +10,7 @@
 from mypyc.ir.ops import (
     Environment, BasicBlock, Goto, Return, LoadInt, Assign, IncRef, DecRef, Branch,
     Call, Unbox, Box, TupleGet, GetAttr, PrimitiveOp, RegisterOp,
-    SetAttr, Op, Value, CallC, BinaryIntOp
+    SetAttr, Op, Value, CallC, BinaryIntOp, LoadMem
 )
 from mypyc.ir.rtypes import (
     RTuple, RInstance, int_rprimitive, bool_rprimitive, list_rprimitive,
@@ -33,6 +33,7 @@
 from mypyc.primitives.int_ops import int_neg_op
 from mypyc.subtype import is_subtype
 from mypyc.namegen import NameGenerator
+from mypyc.common import IS_32_BIT_PLATFORM
 
 
 class TestFunctionEmitterVisitor(unittest.TestCase):
@@ -273,6 +274,14 @@ def test_binary_int_op(self) -> None:
         self.assert_emit(BinaryIntOp(bool_rprimitive, self.i64, self.i64_1, BinaryIntOp.ULT, 1),
                          """cpy_r_r04 = (uint64_t)cpy_r_i64 < (uint64_t)cpy_r_i64_1;""")
 
+    def test_load_mem(self) -> None:
+        if IS_32_BIT_PLATFORM:
+            self.assert_emit(LoadMem(bool_rprimitive, self.i32),
+                             """cpy_r_r0 = *(char *)cpy_r_i32;""")
+        else:
+            self.assert_emit(LoadMem(bool_rprimitive, self.i64),
+                             """cpy_r_r0 = *(char *)cpy_r_i64;""")
+
     def assert_emit(self, op: Op, expected: str) -> None:
         self.emitter.fragments = []
         self.declarations.fragments = []