Refactor assignment statement handling and the typing mechanism around it

pythonbpf/assign_pass.py

@@ -0,0 +1,108 @@
+import ast
+import logging
+from llvmlite import ir
+from pythonbpf.expr import eval_expr
+
+logger = logging.getLogger(__name__)
+
+
+def handle_struct_field_assignment(
+    func, module, builder, target, rval, local_sym_tab, map_sym_tab, structs_sym_tab
+):
+    """Handle struct field assignment (obj.field = value)."""
+
+    var_name = target.value.id
+    field_name = target.attr
+
+    if var_name not in local_sym_tab:
+        logger.error(f"Variable '{var_name}' not found in symbol table")
+        return
+
+    struct_type = local_sym_tab[var_name].metadata
+    struct_info = structs_sym_tab[struct_type]
+
+    if field_name not in struct_info.fields:
+        logger.error(f"Field '{field_name}' not found in struct '{struct_type}'")
+        return
+
+    # Get field pointer and evaluate value
+    field_ptr = struct_info.gep(builder, local_sym_tab[var_name].var, field_name)
+    val = eval_expr(
+        func, module, builder, rval, local_sym_tab, map_sym_tab, structs_sym_tab
+    )
+
+    if val is None:
+        logger.error(f"Failed to evaluate value for {var_name}.{field_name}")
+        return
+
+    # TODO: Handle string assignment to char array (not a priority)
+    field_type = struct_info.field_type(field_name)
+    if isinstance(field_type, ir.ArrayType) and val[1] == ir.PointerType(ir.IntType(8)):
+        logger.warning(
+            f"String to char array assignment not implemented for {var_name}.{field_name}"
+        )
+        return
+
+    # Store the value
+    builder.store(val[0], field_ptr)
+    logger.info(f"Assigned to struct field {var_name}.{field_name}")
+
+
+def handle_variable_assignment(
+    func, module, builder, var_name, rval, local_sym_tab, map_sym_tab, structs_sym_tab
+):
+    """Handle single named variable assignment."""
+
+    if var_name not in local_sym_tab:
+        logger.error(f"Variable {var_name} not declared.")
+        return False
+
+    var_ptr = local_sym_tab[var_name].var
+    var_type = local_sym_tab[var_name].ir_type
+
+    # NOTE: Special case for struct initialization
+    if isinstance(rval, ast.Call) and isinstance(rval.func, ast.Name):
+        struct_name = rval.func.id
+        if struct_name in structs_sym_tab and len(rval.args) == 0:
+            struct_info = structs_sym_tab[struct_name]
+            ir_struct = struct_info.ir_type
+
+            builder.store(ir.Constant(ir_struct, None), var_ptr)
+            logger.info(f"Initialized struct {struct_name} for variable {var_name}")
+            return True
+
+    val_result = eval_expr(
+        func, module, builder, rval, local_sym_tab, map_sym_tab, structs_sym_tab
+    )
+    if val_result is None:
+        logger.error(f"Failed to evaluate value for {var_name}")
+        return False
+
+    val, val_type = val_result
+    logger.info(f"Evaluated value for {var_name}: {val} of type {val_type}, {var_type}")
+    if val_type != var_type:
+        if isinstance(val_type, ir.IntType) and isinstance(var_type, ir.IntType):
+            # Allow implicit int widening
+            if val_type.width < var_type.width:
+                val = builder.sext(val, var_type)
+                logger.info(f"Implicitly widened int for variable {var_name}")
+            elif val_type.width > var_type.width:
+                val = builder.trunc(val, var_type)
+                logger.info(f"Implicitly truncated int for variable {var_name}")
+        elif isinstance(val_type, ir.IntType) and isinstance(var_type, ir.PointerType):
+            # NOTE: This is assignment to a PTR_TO_MAP_VALUE_OR_NULL
+            logger.info(
+                f"Creating temporary variable for pointer assignment to {var_name}"
+            )
+            var_ptr_tmp = local_sym_tab[f"{var_name}_tmp"].var
+            builder.store(val, var_ptr_tmp)
+            val = var_ptr_tmp
+        else:
+            logger.error(
+                f"Type mismatch for variable {var_name}: {val_type} vs {var_type}"
+            )
+            return False
+
+    builder.store(val, var_ptr)
+    logger.info(f"Assigned value to variable {var_name}")
+    return True

pythonbpf/binary_ops.py

@@ -3,43 +3,70 @@
 from logging import Logger
 import logging
 
-logger: Logger = logging.getLogger(__name__)
-
+from pythonbpf.expr import get_base_type_and_depth, deref_to_depth, eval_expr
 
-def recursive_dereferencer(var, builder):
-    """dereference until primitive type comes out"""
-    # TODO: Not worrying about stack overflow for now
-    logger.info(f"Dereferencing {var}, type is {var.type}")
-    if isinstance(var.type, ir.PointerType):
-        a = builder.load(var)
-        return recursive_dereferencer(a, builder)
-    elif isinstance(var.type, ir.IntType):
-        return var
-    else:
-        raise TypeError(f"Unsupported type for dereferencing: {var.type}")
+logger: Logger = logging.getLogger(__name__)
 
 
-def get_operand_value(operand, builder, local_sym_tab):
+def get_operand_value(
+    func, module, operand, builder, local_sym_tab, map_sym_tab, structs_sym_tab=None
+):
     """Extract the value from an operand, handling variables and constants."""
+    logger.info(f"Getting operand value for: {ast.dump(operand)}")
     if isinstance(operand, ast.Name):
         if operand.id in local_sym_tab:
-            return recursive_dereferencer(local_sym_tab[operand.id].var, builder)
+            var = local_sym_tab[operand.id].var
+            var_type = var.type
+            base_type, depth = get_base_type_and_depth(var_type)
+            logger.info(f"var is {var}, base_type is {base_type}, depth is {depth}")
+            val = deref_to_depth(func, builder, var, depth)
+            return val
         raise ValueError(f"Undefined variable: {operand.id}")
     elif isinstance(operand, ast.Constant):
         if isinstance(operand.value, int):
-            return ir.Constant(ir.IntType(64), operand.value)
+            cst = ir.Constant(ir.IntType(64), int(operand.value))
+            return cst
         raise TypeError(f"Unsupported constant type: {type(operand.value)}")
     elif isinstance(operand, ast.BinOp):
-        return handle_binary_op_impl(operand, builder, local_sym_tab)
+        res = handle_binary_op_impl(
+            func, module, operand, builder, local_sym_tab, map_sym_tab, structs_sym_tab
+        )
+        return res
+    else:
+        res = eval_expr(
+            func, module, builder, operand, local_sym_tab, map_sym_tab, structs_sym_tab
+        )
+        if res is None:
+            raise ValueError(f"Failed to evaluate call expression: {operand}")
+        val, _ = res
+        logger.info(f"Evaluated expr to {val} of type {val.type}")
+        base_type, depth = get_base_type_and_depth(val.type)
+        if depth > 0:
+            val = deref_to_depth(func, builder, val, depth)
+        return val
     raise TypeError(f"Unsupported operand type: {type(operand)}")
 
 
-def handle_binary_op_impl(rval, builder, local_sym_tab):
+def handle_binary_op_impl(
+    func, module, rval, builder, local_sym_tab, map_sym_tab, structs_sym_tab=None
+):
     op = rval.op
-    left = get_operand_value(rval.left, builder, local_sym_tab)
-    right = get_operand_value(rval.right, builder, local_sym_tab)
+    left = get_operand_value(
+        func, module, rval.left, builder, local_sym_tab, map_sym_tab, structs_sym_tab
+    )
+    right = get_operand_value(
+        func, module, rval.right, builder, local_sym_tab, map_sym_tab, structs_sym_tab
+    )
     logger.info(f"left is {left}, right is {right}, op is {op}")
 
+    # NOTE: Before doing the operation, if the operands are integers
+    # we always extend them to i64. The assignment to LHS will take
+    # care of truncation if needed.
+    if isinstance(left.type, ir.IntType) and left.type.width < 64:
+        left = builder.sext(left, ir.IntType(64))
+    if isinstance(right.type, ir.IntType) and right.type.width < 64:
+        right = builder.sext(right, ir.IntType(64))
+
     # Map AST operation nodes to LLVM IR builder methods
     op_map = {
         ast.Add: builder.add,
@@ -62,8 +89,19 @@ def handle_binary_op_impl(rval, builder, local_sym_tab):
         raise SyntaxError("Unsupported binary operation")
 
 
-def handle_binary_op(rval, builder, var_name, local_sym_tab):
-    result = handle_binary_op_impl(rval, builder, local_sym_tab)
+def handle_binary_op(
+    func,
+    module,
+    rval,
+    builder,
+    var_name,
+    local_sym_tab,
+    map_sym_tab,
+    structs_sym_tab=None,
+):
+    result = handle_binary_op_impl(
+        func, module, rval, builder, local_sym_tab, map_sym_tab, structs_sym_tab
+    )
     if var_name and var_name in local_sym_tab:
         logger.info(
             f"Storing result {result} into variable {local_sym_tab[var_name].var}"

pythonbpf/expr/__init__.py

@@ -1,4 +1,10 @@
 from .expr_pass import eval_expr, handle_expr
-from .type_normalization import convert_to_bool
+from .type_normalization import convert_to_bool, get_base_type_and_depth, deref_to_depth
 
-__all__ = ["eval_expr", "handle_expr", "convert_to_bool"]
+__all__ = [
+    "eval_expr",
+    "handle_expr",
+    "convert_to_bool",
+    "get_base_type_and_depth",
+    "deref_to_depth",
+]

pythonbpf/expr/expr_pass.py

@@ -26,7 +26,7 @@ def _handle_constant_expr(expr: ast.Constant):
     if isinstance(expr.value, int) or isinstance(expr.value, bool):
         return ir.Constant(ir.IntType(64), int(expr.value)), ir.IntType(64)
     else:
-        logger.error("Unsupported constant type")
+        logger.error(f"Unsupported constant type {ast.dump(expr)}")
         return None
 
 
@@ -176,21 +176,28 @@ def _handle_unary_op(
     structs_sym_tab=None,
 ):
     """Handle ast.UnaryOp expressions."""
-    if not isinstance(expr.op, ast.Not):
-        logger.error("Only 'not' unary operator is supported")
+    if not isinstance(expr.op, ast.Not) and not isinstance(expr.op, ast.USub):
+        logger.error("Only 'not' and '-' unary operators are supported")
         return None
 
-    operand = eval_expr(
-        func, module, builder, expr.operand, local_sym_tab, map_sym_tab, structs_sym_tab
+    from pythonbpf.binary_ops import get_operand_value
+
+    operand = get_operand_value(
+        func, module, expr.operand, builder, local_sym_tab, map_sym_tab, structs_sym_tab
     )
     if operand is None:
         logger.error("Failed to evaluate operand for unary operation")
         return None
 
-    operand_val, operand_type = operand
-    true_const = ir.Constant(ir.IntType(1), 1)
-    result = builder.xor(convert_to_bool(builder, operand_val), true_const)
-    return result, ir.IntType(1)
+    if isinstance(expr.op, ast.Not):
+        true_const = ir.Constant(ir.IntType(1), 1)
+        result = builder.xor(convert_to_bool(builder, operand), true_const)
+        return result, ir.IntType(1)
+    elif isinstance(expr.op, ast.USub):
+        # Multiply by -1
+        neg_one = ir.Constant(ir.IntType(64), -1)
+        result = builder.mul(operand, neg_one)
+        return result, ir.IntType(64)
 
 
 def _handle_and_op(func, builder, expr, local_sym_tab, map_sym_tab, structs_sym_tab):
@@ -402,7 +409,16 @@ def eval_expr(
     elif isinstance(expr, ast.BinOp):
         from pythonbpf.binary_ops import handle_binary_op
 
-        return handle_binary_op(expr, builder, None, local_sym_tab)
+        return handle_binary_op(
+            func,
+            module,
+            expr,
+            builder,
+            None,
+            local_sym_tab,
+            map_sym_tab,
+            structs_sym_tab,
+        )
     elif isinstance(expr, ast.Compare):
         return _handle_compare(
             func, module, builder, expr, local_sym_tab, map_sym_tab, structs_sym_tab

pythonbpf/expr/type_normalization.py

@@ -16,7 +16,7 @@
 }
 
 
-def _get_base_type_and_depth(ir_type):
+def get_base_type_and_depth(ir_type):
     """Get the base type for pointer types."""
     cur_type = ir_type
     depth = 0
@@ -26,7 +26,7 @@ def _get_base_type_and_depth(ir_type):
     return cur_type, depth
 
 
-def _deref_to_depth(func, builder, val, target_depth):
+def deref_to_depth(func, builder, val, target_depth):
     """Dereference a pointer to a certain depth."""
 
     cur_val = val
@@ -88,13 +88,13 @@ def _normalize_types(func, builder, lhs, rhs):
         logger.error(f"Type mismatch: {lhs.type} vs {rhs.type}")
         return None, None
     else:
-        lhs_base, lhs_depth = _get_base_type_and_depth(lhs.type)
-        rhs_base, rhs_depth = _get_base_type_and_depth(rhs.type)
+        lhs_base, lhs_depth = get_base_type_and_depth(lhs.type)
+        rhs_base, rhs_depth = get_base_type_and_depth(rhs.type)
         if lhs_base == rhs_base:
             if lhs_depth < rhs_depth:
-                rhs = _deref_to_depth(func, builder, rhs, rhs_depth - lhs_depth)
+                rhs = deref_to_depth(func, builder, rhs, rhs_depth - lhs_depth)
             elif rhs_depth < lhs_depth:
-                lhs = _deref_to_depth(func, builder, lhs, lhs_depth - rhs_depth)
+                lhs = deref_to_depth(func, builder, lhs, lhs_depth - rhs_depth)
             return _normalize_types(func, builder, lhs, rhs)