Complete documentation coverage: add final module docstrings

Co-authored-by: varun-r-mallya <100590632+varun-r-mallya@users.noreply.github.com>

TODO.md

@@ -1,13 +0,0 @@
-## Short term
-
-- Implement enough functionality to port the BCC tutorial examples in PythonBPF
-- Add all maps
-- XDP support in pylibbpf
-- ringbuf support
-- Add oneline IfExpr conditionals (wishlist)
-
-## Long term
-
-- Refactor the codebase to be better than a hackathon project
-- Port to C++ and use actual LLVM?
-- Fix struct_kioctx issue in the vmlinux transpiler

pythonbpf/__init__.py

@@ -1,3 +1,10 @@
+"""
+PythonBPF - A Python frontend for eBPF programs.
+
+This package provides decorators and compilation tools to write BPF programs
+in Python syntax and compile them to eBPF bytecode that can run in the kernel.
+"""
+
 from .decorators import bpf, map, section, bpfglobal, struct
 from .codegen import compile_to_ir, compile, BPF
 

pythonbpf/allocation_pass.py

@@ -1,191 +0,0 @@
-import ast
-import logging
-
-from llvmlite import ir
-from dataclasses import dataclass
-from typing import Any
-from pythonbpf.helper import HelperHandlerRegistry
-from pythonbpf.type_deducer import ctypes_to_ir
-
-logger = logging.getLogger(__name__)
-
-
-@dataclass
-class LocalSymbol:
-    var: ir.AllocaInstr
-    ir_type: ir.Type
-    metadata: Any = None
-
-    def __iter__(self):
-        yield self.var
-        yield self.ir_type
-        yield self.metadata
-
-
-def _is_helper_call(call_node):
-    """Check if a call node is a BPF helper function call."""
-    if isinstance(call_node.func, ast.Name):
-        # Exclude print from requiring temps (handles f-strings differently)
-        func_name = call_node.func.id
-        return HelperHandlerRegistry.has_handler(func_name) and func_name != "print"
-
-    elif isinstance(call_node.func, ast.Attribute):
-        return HelperHandlerRegistry.has_handler(call_node.func.attr)
-
-    return False
-
-
-def handle_assign_allocation(builder, stmt, local_sym_tab, structs_sym_tab):
-    """Handle memory allocation for assignment statements."""
-
-    # Validate assignment
-    if len(stmt.targets) != 1:
-        logger.warning("Multi-target assignment not supported, skipping allocation")
-        return
-
-    target = stmt.targets[0]
-
-    # Skip non-name targets (e.g., struct field assignments)
-    if isinstance(target, ast.Attribute):
-        logger.debug(f"Struct field assignment to {target.attr}, no allocation needed")
-        return
-
-    if not isinstance(target, ast.Name):
-        logger.warning(f"Unsupported assignment target type: {type(target).__name__}")
-        return
-
-    var_name = target.id
-    rval = stmt.value
-
-    # Skip if already allocated
-    if var_name in local_sym_tab:
-        logger.debug(f"Variable {var_name} already allocated, skipping")
-        return
-
-    # Determine type and allocate based on rval
-    if isinstance(rval, ast.Call):
-        _allocate_for_call(builder, var_name, rval, local_sym_tab, structs_sym_tab)
-    elif isinstance(rval, ast.Constant):
-        _allocate_for_constant(builder, var_name, rval, local_sym_tab)
-    elif isinstance(rval, ast.BinOp):
-        _allocate_for_binop(builder, var_name, local_sym_tab)
-    else:
-        logger.warning(
-            f"Unsupported assignment value type for {var_name}: {type(rval).__name__}"
-        )
-
-
-def _allocate_for_call(builder, var_name, rval, local_sym_tab, structs_sym_tab):
-    """Allocate memory for variable assigned from a call."""
-
-    if isinstance(rval.func, ast.Name):
-        call_type = rval.func.id
-
-        # C type constructors
-        if call_type in ("c_int32", "c_int64", "c_uint32", "c_uint64"):
-            ir_type = ctypes_to_ir(call_type)
-            var = builder.alloca(ir_type, name=var_name)
-            var.align = ir_type.width // 8
-            local_sym_tab[var_name] = LocalSymbol(var, ir_type)
-            logger.info(f"Pre-allocated {var_name} as {call_type}")
-
-        # Helper functions
-        elif HelperHandlerRegistry.has_handler(call_type):
-            ir_type = ir.IntType(64)  # Assume i64 return type
-            var = builder.alloca(ir_type, name=var_name)
-            var.align = 8
-            local_sym_tab[var_name] = LocalSymbol(var, ir_type)
-            logger.info(f"Pre-allocated {var_name} for helper {call_type}")
-
-        # Deref function
-        elif call_type == "deref":
-            ir_type = ir.IntType(64)  # Assume i64 return type
-            var = builder.alloca(ir_type, name=var_name)
-            var.align = 8
-            local_sym_tab[var_name] = LocalSymbol(var, ir_type)
-            logger.info(f"Pre-allocated {var_name} for deref")
-
-        # Struct constructors
-        elif call_type in structs_sym_tab:
-            struct_info = structs_sym_tab[call_type]
-            var = builder.alloca(struct_info.ir_type, name=var_name)
-            local_sym_tab[var_name] = LocalSymbol(var, struct_info.ir_type, call_type)
-            logger.info(f"Pre-allocated {var_name} for struct {call_type}")
-
-        else:
-            logger.warning(f"Unknown call type for allocation: {call_type}")
-
-    elif isinstance(rval.func, ast.Attribute):
-        # Map method calls - need double allocation for ptr handling
-        _allocate_for_map_method(builder, var_name, local_sym_tab)
-
-    else:
-        logger.warning(f"Unsupported call function type for {var_name}")
-
-
-def _allocate_for_map_method(builder, var_name, local_sym_tab):
-    """Allocate memory for variable assigned from map method (double alloc)."""
-
-    # Main variable (pointer to pointer)
-    ir_type = ir.PointerType(ir.IntType(64))
-    var = builder.alloca(ir_type, name=var_name)
-    local_sym_tab[var_name] = LocalSymbol(var, ir_type)
-
-    # Temporary variable for computed values
-    tmp_ir_type = ir.IntType(64)
-    var_tmp = builder.alloca(tmp_ir_type, name=f"{var_name}_tmp")
-    local_sym_tab[f"{var_name}_tmp"] = LocalSymbol(var_tmp, tmp_ir_type)
-
-    logger.info(f"Pre-allocated {var_name} and {var_name}_tmp for map method")
-
-
-def _allocate_for_constant(builder, var_name, rval, local_sym_tab):
-    """Allocate memory for variable assigned from a constant."""
-
-    if isinstance(rval.value, bool):
-        ir_type = ir.IntType(1)
-        var = builder.alloca(ir_type, name=var_name)
-        var.align = 1
-        local_sym_tab[var_name] = LocalSymbol(var, ir_type)
-        logger.info(f"Pre-allocated {var_name} as bool")
-
-    elif isinstance(rval.value, int):
-        ir_type = ir.IntType(64)
-        var = builder.alloca(ir_type, name=var_name)
-        var.align = 8
-        local_sym_tab[var_name] = LocalSymbol(var, ir_type)
-        logger.info(f"Pre-allocated {var_name} as i64")
-
-    elif isinstance(rval.value, str):
-        ir_type = ir.PointerType(ir.IntType(8))
-        var = builder.alloca(ir_type, name=var_name)
-        var.align = 8
-        local_sym_tab[var_name] = LocalSymbol(var, ir_type)
-        logger.info(f"Pre-allocated {var_name} as string")
-
-    else:
-        logger.warning(
-            f"Unsupported constant type for {var_name}: {type(rval.value).__name__}"
-        )
-
-
-def _allocate_for_binop(builder, var_name, local_sym_tab):
-    """Allocate memory for variable assigned from a binary operation."""
-    ir_type = ir.IntType(64)  # Assume i64 result
-    var = builder.alloca(ir_type, name=var_name)
-    var.align = 8
-    local_sym_tab[var_name] = LocalSymbol(var, ir_type)
-    logger.info(f"Pre-allocated {var_name} for binop result")
-
-
-def allocate_temp_pool(builder, max_temps, local_sym_tab):
-    """Allocate the temporary scratch space pool for helper arguments."""
-    if max_temps == 0:
-        return
-
-    logger.info(f"Allocating temp pool of {max_temps} variables")
-    for i in range(max_temps):
-        temp_name = f"__helper_temp_{i}"
-        temp_var = builder.alloca(ir.IntType(64), name=temp_name)
-        temp_var.align = 8
-        local_sym_tab[temp_name] = LocalSymbol(temp_var, ir.IntType(64))

pythonbpf/assign_pass.py

@@ -1,108 +0,0 @@
-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

@@ -1,71 +1,62 @@
+"""
+Binary operations handling for BPF programs.
+
+This module provides functions to handle binary operations (add, subtract,
+multiply, etc.) and emit the corresponding LLVM IR instructions.
+"""
+
 import ast
 from llvmlite import ir
 from logging import Logger
 import logging
 
-from pythonbpf.expr import get_base_type_and_depth, deref_to_depth, eval_expr
-
 logger: Logger = logging.getLogger(__name__)
 
 
-def get_operand_value(
-    func, module, operand, builder, local_sym_tab, map_sym_tab, structs_sym_tab=None
-):
+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}")
+
+
+def get_operand_value(operand, builder, local_sym_tab):
     """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:
-            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
+            return recursive_dereferencer(local_sym_tab[operand.id].var, builder)
         raise ValueError(f"Undefined variable: {operand.id}")
     elif isinstance(operand, ast.Constant):
         if isinstance(operand.value, int):
-            cst = ir.Constant(ir.IntType(64), int(operand.value))
-            return cst
+            return ir.Constant(ir.IntType(64), operand.value)
         raise TypeError(f"Unsupported constant type: {type(operand.value)}")
     elif isinstance(operand, ast.BinOp):
-        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
+        return handle_binary_op_impl(operand, builder, local_sym_tab)
     raise TypeError(f"Unsupported operand type: {type(operand)}")
 
 
-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(
-        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}")
+def handle_binary_op_impl(rval, builder, local_sym_tab):
+    """
+    Handle binary operations and emit corresponding LLVM IR instructions.
     
-    # 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))
+    Args:
+        rval: The AST BinOp node representing the binary operation
+        builder: LLVM IR builder for emitting instructions
+        local_sym_tab: Symbol table mapping variable names to their IR representations
+    
+    Returns:
+        The LLVM IR value representing the result of the binary operation
+    """
+    op = rval.op
+    left = get_operand_value(rval.left, builder, local_sym_tab)
+    right = get_operand_value(rval.right, builder, local_sym_tab)
+    logger.info(f"left is {left}, right is {right}, op is {op}")
 
     # Map AST operation nodes to LLVM IR builder methods
     op_map = {
@@ -89,19 +80,20 @@ def handle_binary_op_impl(
         raise SyntaxError("Unsupported binary operation")
 
 
-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
-    )
+def handle_binary_op(rval, builder, var_name, local_sym_tab):
+    """
+    Handle binary operations and optionally store the result to a variable.
+    
+    Args:
+        rval: The AST BinOp node representing the binary operation
+        builder: LLVM IR builder for emitting instructions
+        var_name: Optional variable name to store the result
+        local_sym_tab: Symbol table mapping variable names to their IR representations
+    
+    Returns:
+        A tuple of (result_value, result_type)
+    """
+    result = handle_binary_op_impl(rval, builder, local_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/codegen.py

@@ -1,3 +1,11 @@
+"""
+Code generation module for PythonBPF.
+
+This module handles the conversion of Python BPF programs to LLVM IR and
+object files. It provides the main compilation pipeline from Python AST
+to BPF bytecode.
+"""
+
 import ast
 from llvmlite import ir
 from .license_pass import license_processing
@@ -37,6 +45,14 @@ def find_bpf_chunks(tree):
 
 
 def processor(source_code, filename, module):
+    """
+    Process Python source code and convert BPF-decorated functions to LLVM IR.
+    
+    Args:
+        source_code: The Python source code to process
+        filename: The name of the source file
+        module: The LLVM IR module to populate
+    """
     tree = ast.parse(source_code, filename)
     logger.debug(ast.dump(tree, indent=4))
 
@@ -56,6 +72,17 @@ def processor(source_code, filename, module):
 
 
 def compile_to_ir(filename: str, output: str, loglevel=logging.INFO):
+    """
+    Compile a Python BPF program to LLVM IR.
+    
+    Args:
+        filename: Path to the Python source file containing BPF programs
+        output: Path where the LLVM IR (.ll) file will be written
+        loglevel: Logging level for compilation messages
+    
+    Returns:
+        Path to the generated LLVM IR file
+    """
     logging.basicConfig(
         level=loglevel, format="%(asctime)s [%(levelname)s] %(name)s: %(message)s"
     )
@@ -129,6 +156,18 @@ def compile_to_ir(filename: str, output: str, loglevel=logging.INFO):
 
 
 def compile(loglevel=logging.INFO) -> bool:
+    """
+    Compile the calling Python BPF program to an object file.
+    
+    This function should be called from a Python file containing BPF programs.
+    It will compile the calling file to LLVM IR and then to a BPF object file.
+    
+    Args:
+        loglevel: Logging level for compilation messages
+    
+    Returns:
+        True if compilation succeeded, False otherwise
+    """
     # Look one level up the stack to the caller of this function
     caller_frame = inspect.stack()[1]
     caller_file = Path(caller_frame.filename).resolve()
@@ -162,6 +201,18 @@ def compile(loglevel=logging.INFO) -> bool:
 
 
 def BPF(loglevel=logging.INFO) -> BpfProgram:
+    """
+    Compile the calling Python BPF program and return a BpfProgram object.
+    
+    This function compiles the calling file's BPF programs to an object file
+    and loads it into a BpfProgram object for immediate use.
+    
+    Args:
+        loglevel: Logging level for compilation messages
+    
+    Returns:
+        A BpfProgram object that can be used to load and attach BPF programs
+    """
     caller_frame = inspect.stack()[1]
     src = inspect.getsource(caller_frame.frame)
     with tempfile.NamedTemporaryFile(

pythonbpf/debuginfo/__init__.py

@@ -1,3 +1,5 @@
+"""Debug information generation for BPF programs (DWARF/BTF)."""
+
 from .dwarf_constants import *  # noqa: F403
 from .dtypes import *  # noqa: F403
 from .debug_info_generator import DebugInfoGenerator

pythonbpf/debuginfo/debug_info_generator.py

@@ -8,11 +8,31 @@ from typing import Any, List
 
 
 class DebugInfoGenerator:
+    """
+    Generator for DWARF/BTF debug information in LLVM IR modules.
+    
+    This class provides methods to create debug metadata for BPF programs,
+    including types, structs, globals, and compilation units.
+    """
+    
     def __init__(self, module):
+        """
+        Initialize the debug info generator.
+        
+        Args:
+            module: LLVM IR module to attach debug info to
+        """
         self.module = module
         self._type_cache = {}  # Cache for common debug types
 
     def generate_file_metadata(self, filename, dirname):
+        """
+        Generate file metadata for debug info.
+        
+        Args:
+            filename: Name of the source file
+            dirname: Directory containing the source file
+        """
         self.module._file_metadata = self.module.add_debug_info(
             "DIFile",
             {  # type: ignore
@@ -24,6 +44,15 @@ class DebugInfoGenerator:
     def generate_debug_cu(
         self, language, producer: str, is_optimized: bool, is_distinct: bool
     ):
+        """
+        Generate debug compile unit metadata.
+        
+        Args:
+            language: DWARF language code (e.g., DW_LANG_C11)
+            producer: Compiler/producer string
+            is_optimized: Whether the code is optimized
+            is_distinct: Whether the compile unit should be distinct
+        """
         self.module._debug_compile_unit = self.module.add_debug_info(
             "DICompileUnit",
             {  # type: ignore
@@ -83,6 +112,16 @@ class DebugInfoGenerator:
 
     @staticmethod
     def _compute_array_size(base_type: Any, count: int) -> int:
+        """
+        Compute the size of an array in bits.
+        
+        Args:
+            base_type: The base type of the array
+            count: Number of elements in the array
+        
+        Returns:
+            Total size in bits
+        """
         # Extract size from base_type if possible
         # For simplicity, assuming base_type has a size attribute
         return getattr(base_type, "size", 32) * count

pythonbpf/debuginfo/dtypes.py

@@ -1,7 +1,10 @@
+"""Debug information types and constants."""
+
 import llvmlite.ir as ir
 
 
 class DwarfBehaviorEnum:
+    """DWARF module flag behavior constants for LLVM."""
     ERROR_IF_MISMATCH = ir.Constant(ir.IntType(32), 1)
     WARNING_IF_MISMATCH = ir.Constant(ir.IntType(32), 2)
     OVERRIDE_USE_LARGEST = ir.Constant(ir.IntType(32), 7)

pythonbpf/debuginfo/dwarf_constants.py

@@ -1,3 +1,9 @@
+"""
+DWARF debugging format constants.
+
+Generated constants from dwarf.h for use in debug information generation.
+"""
+
 # generated constants from dwarf.h
 
 DW_UT_compile = 0x01

pythonbpf/decorators.py

@@ -1,3 +1,11 @@
+"""
+Decorators for marking BPF functions, maps, structs, and globals.
+
+This module provides the core decorators used to annotate Python code
+for BPF compilation.
+"""
+
+
 def bpf(func):
     """Decorator to mark a function for BPF compilation."""
     func._is_bpf = True
@@ -23,7 +31,17 @@ def struct(cls):
 
 
 def section(name: str):
+    """
+    Decorator to specify the ELF section name for a BPF program.
+    
+    Args:
+        name: The section name (e.g., 'xdp', 'tracepoint/syscalls/sys_enter_execve')
+    
+    Returns:
+        A decorator function that marks the function with the section name
+    """
     def wrapper(fn):
+        """Decorator that sets the section name on the function."""
         fn._section = name
         return fn
 

pythonbpf/expr/__init__.py

@@ -1,10 +1,6 @@
-from .expr_pass import eval_expr, handle_expr
-from .type_normalization import convert_to_bool, get_base_type_and_depth, deref_to_depth
+"""Expression evaluation and processing for BPF programs."""
 
-__all__ = [
-    "eval_expr",
-    "handle_expr",
-    "convert_to_bool",
-    "get_base_type_and_depth",
-    "deref_to_depth",
-]
+from .expr_pass import eval_expr, handle_expr
+from .type_normalization import convert_to_bool
+
+__all__ = ["eval_expr", "handle_expr", "convert_to_bool"]

pythonbpf/expr/expr_pass.py

@@ -1,3 +1,11 @@
+"""
+Expression evaluation and LLVM IR generation.
+
+This module handles the evaluation of Python expressions in BPF programs,
+including variables, constants, function calls, comparisons, boolean
+operations, and more.
+"""
+
 import ast
 from llvmlite import ir
 from logging import Logger
@@ -26,7 +34,7 @@ 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(f"Unsupported constant type {ast.dump(expr)}")
+        logger.error("Unsupported constant type")
         return None
 
 
@@ -176,28 +184,21 @@ def _handle_unary_op(
     structs_sym_tab=None,
 ):
     """Handle ast.UnaryOp expressions."""
-    if not isinstance(expr.op, ast.Not) and not isinstance(expr.op, ast.USub):
-        logger.error("Only 'not' and '-' unary operators are supported")
+    if not isinstance(expr.op, ast.Not):
+        logger.error("Only 'not' unary operator is supported")
         return None
 
-    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
+    operand = eval_expr(
+        func, module, builder, expr.operand, local_sym_tab, map_sym_tab, structs_sym_tab
     )
     if operand is None:
         logger.error("Failed to evaluate operand for unary operation")
         return None
 
-    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)
+    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)
 
 
 def _handle_and_op(func, builder, expr, local_sym_tab, map_sym_tab, structs_sym_tab):
@@ -339,6 +340,21 @@ def eval_expr(
     map_sym_tab,
     structs_sym_tab=None,
 ):
+    """
+    Evaluate an expression and return its LLVM IR value and type.
+    
+    Args:
+        func: The LLVM IR function being built
+        module: The LLVM IR module
+        builder: LLVM IR builder
+        expr: The AST expression node to evaluate
+        local_sym_tab: Local symbol table
+        map_sym_tab: Map symbol table
+        structs_sym_tab: Struct symbol table
+    
+    Returns:
+        A tuple of (value, type) or None if evaluation fails
+    """
     logger.info(f"Evaluating expression: {ast.dump(expr)}")
     if isinstance(expr, ast.Name):
         return _handle_name_expr(expr, local_sym_tab, builder)
@@ -409,16 +425,7 @@ def eval_expr(
     elif isinstance(expr, ast.BinOp):
         from pythonbpf.binary_ops import handle_binary_op
 
-        return handle_binary_op(
-            func,
-            module,
-            expr,
-            builder,
-            None,
-            local_sym_tab,
-            map_sym_tab,
-            structs_sym_tab,
-        )
+        return handle_binary_op(expr, builder, None, local_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

@@ -1,3 +1,10 @@
+"""
+Type normalization and comparison operations for expressions.
+
+This module provides utilities for normalizing types between expressions,
+handling pointer dereferencing, and generating comparison operations.
+"""
+
 from llvmlite import ir
 import logging
 import ast
@@ -16,8 +23,16 @@ COMPARISON_OPS = {
 }
 
 
-def get_base_type_and_depth(ir_type):
-    """Get the base type for pointer types."""
+def _get_base_type_and_depth(ir_type):
+    """
+    Get the base type and pointer depth for an LLVM IR type.
+    
+    Args:
+        ir_type: The LLVM IR type to analyze
+    
+    Returns:
+        A tuple of (base_type, depth) where depth is the number of pointer levels
+    """
     cur_type = ir_type
     depth = 0
     while isinstance(cur_type, ir.PointerType):
@@ -26,8 +41,19 @@ def get_base_type_and_depth(ir_type):
     return cur_type, depth
 
 
-def deref_to_depth(func, builder, val, target_depth):
-    """Dereference a pointer to a certain depth."""
+def _deref_to_depth(func, builder, val, target_depth):
+    """
+    Dereference a pointer to a certain depth with null checks.
+    
+    Args:
+        func: The LLVM IR function being built
+        builder: LLVM IR builder
+        val: The pointer value to dereference
+        target_depth: Number of levels to dereference
+    
+    Returns:
+        The dereferenced value, or None if dereferencing fails
+    """
 
     cur_val = val
     cur_type = val.type
@@ -73,7 +99,18 @@ def deref_to_depth(func, builder, val, target_depth):
 
 
 def _normalize_types(func, builder, lhs, rhs):
-    """Normalize types for comparison."""
+    """
+    Normalize types for comparison by casting or dereferencing as needed.
+    
+    Args:
+        func: The LLVM IR function being built
+        builder: LLVM IR builder
+        lhs: Left-hand side value
+        rhs: Right-hand side value
+    
+    Returns:
+        A tuple of (normalized_lhs, normalized_rhs) or (None, None) on error
+    """
 
     logger.info(f"Normalizing types: {lhs.type} vs {rhs.type}")
     if isinstance(lhs.type, ir.IntType) and isinstance(rhs.type, ir.IntType):
@@ -88,18 +125,27 @@ 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)
 
 
 def convert_to_bool(builder, val):
-    """Convert a value to boolean."""
+    """
+    Convert an LLVM IR value to a boolean (i1) type.
+    
+    Args:
+        builder: LLVM IR builder
+        val: The value to convert
+    
+    Returns:
+        An i1 boolean value
+    """
     if val.type == ir.IntType(1):
         return val
     if isinstance(val.type, ir.PointerType):
@@ -110,7 +156,19 @@ def convert_to_bool(builder, val):
 
 
 def handle_comparator(func, builder, op, lhs, rhs):
-    """Handle comparison operations."""
+    """
+    Handle comparison operations between two values.
+    
+    Args:
+        func: The LLVM IR function being built
+        builder: LLVM IR builder
+        op: The AST comparison operator node
+        lhs: Left-hand side value
+        rhs: Right-hand side value
+    
+    Returns:
+        A tuple of (result, ir.IntType(1)) or None on error
+    """
 
     if lhs.type != rhs.type:
         lhs, rhs = _normalize_types(func, builder, lhs, rhs)

pythonbpf/functions/__init__.py

@@ -1,3 +1,5 @@
+"""BPF function processing and LLVM IR generation."""
+
 from .functions_pass import func_proc
 
 __all__ = ["func_proc"]

pythonbpf/functions/func_registry_handlers.py

@@ -1,3 +1,5 @@
+"""Registry for statement handler functions."""
+
 from typing import Dict
 
 
@@ -11,6 +13,7 @@ class StatementHandlerRegistry:
         """Register a handler for a specific statement type."""
 
         def decorator(handler):
+            """Decorator that registers the handler."""
             cls._handlers[stmt_type] = handler
             return handler
 

pythonbpf/functions/functions_pass.py

@@ -1,18 +1,21 @@
+"""
+BPF function processing and LLVM IR generation.
+
+This module handles the core function processing, converting Python function
+definitions into LLVM IR for BPF programs. It manages local variables,
+control flow, and statement processing.
+"""
+
 from llvmlite import ir
 import ast
 import logging
+from typing import Any
+from dataclasses import dataclass
 
-from pythonbpf.helper import (
-    HelperHandlerRegistry,
-    reset_scratch_pool,
-)
+from pythonbpf.helper import HelperHandlerRegistry, handle_helper_call
 from pythonbpf.type_deducer import ctypes_to_ir
+from pythonbpf.binary_ops import handle_binary_op
 from pythonbpf.expr import eval_expr, handle_expr, convert_to_bool
-from pythonbpf.assign_pass import (
-    handle_variable_assignment,
-    handle_struct_field_assignment,
-)
-from pythonbpf.allocation_pass import handle_assign_allocation, allocate_temp_pool
 
 from .return_utils import _handle_none_return, _handle_xdp_return, _is_xdp_name
 
@@ -20,6 +23,27 @@ from .return_utils import _handle_none_return, _handle_xdp_return, _is_xdp_name
 logger = logging.getLogger(__name__)
 
 
+@dataclass
+class LocalSymbol:
+    """
+    Represents a local variable in a BPF function.
+    
+    Attributes:
+        var: LLVM IR alloca instruction for the variable
+        ir_type: LLVM IR type of the variable
+        metadata: Optional metadata (e.g., struct type name)
+    """
+    var: ir.AllocaInstr
+    ir_type: ir.Type
+    metadata: Any = None
+
+    def __iter__(self):
+        """Support tuple unpacking of LocalSymbol."""
+        yield self.var
+        yield self.ir_type
+        yield self.metadata
+
+
 def get_probe_string(func_node):
     """Extract the probe string from the decorator of the function node."""
     # TODO: right now we have the whole string in the section decorator
@@ -41,54 +65,216 @@ def handle_assign(
     func, module, builder, stmt, map_sym_tab, local_sym_tab, structs_sym_tab
 ):
     """Handle assignment statements in the function body."""
-
-    # TODO: Support this later
-    # GH #37
     if len(stmt.targets) != 1:
-        logger.error("Multi-target assignment is not supported for now")
+        logger.info("Unsupported multiassignment")
         return
 
+    num_types = ("c_int32", "c_int64", "c_uint32", "c_uint64")
+
     target = stmt.targets[0]
+    logger.info(f"Handling assignment to {ast.dump(target)}")
+    if not isinstance(target, ast.Name) and not isinstance(target, ast.Attribute):
+        logger.info("Unsupported assignment target")
+        return
+    var_name = target.id if isinstance(target, ast.Name) else target.value.id
     rval = stmt.value
-
-    if isinstance(target, ast.Name):
-        # NOTE: Simple variable assignment case: x = 5
-        var_name = target.id
-        result = handle_variable_assignment(
-            func,
-            module,
-            builder,
-            var_name,
-            rval,
-            local_sym_tab,
-            map_sym_tab,
-            structs_sym_tab,
-        )
-        if not result:
-            logger.error(f"Failed to handle assignment to {var_name}")
-        return
-
     if isinstance(target, ast.Attribute):
-        # NOTE: Struct field assignment case: pkt.field = value
-        handle_struct_field_assignment(
-            func,
-            module,
-            builder,
-            target,
-            rval,
-            local_sym_tab,
-            map_sym_tab,
-            structs_sym_tab,
-        )
-        return
-
-    # Unsupported target type
-    logger.error(f"Unsupported assignment target: {ast.dump(target)}")
+        # struct field assignment
+        field_name = target.attr
+        if var_name in local_sym_tab:
+            struct_type = local_sym_tab[var_name].metadata
+            struct_info = structs_sym_tab[struct_type]
+            if field_name in struct_info.fields:
+                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 isinstance(struct_info.field_type(field_name), ir.ArrayType) and val[
+                    1
+                ] == ir.PointerType(ir.IntType(8)):
+                    # TODO: Figure it out, not a priority rn
+                    # Special case for string assignment to char array
+                    # str_len = struct_info["field_types"][field_idx].count
+                    # assign_string_to_array(builder, field_ptr, val[0], str_len)
+                    # print(f"Assigned to struct field {var_name}.{field_name}")
+                    pass
+                if val is None:
+                    logger.info("Failed to evaluate struct field assignment")
+                    return
+                logger.info(field_ptr)
+                builder.store(val[0], field_ptr)
+                logger.info(f"Assigned to struct field {var_name}.{field_name}")
+                return
+    elif isinstance(rval, ast.Constant):
+        if isinstance(rval.value, bool):
+            if rval.value:
+                builder.store(
+                    ir.Constant(ir.IntType(1), 1), local_sym_tab[var_name].var
+                )
+            else:
+                builder.store(
+                    ir.Constant(ir.IntType(1), 0), local_sym_tab[var_name].var
+                )
+            logger.info(f"Assigned constant {rval.value} to {var_name}")
+        elif isinstance(rval.value, int):
+            # Assume c_int64 for now
+            # var = builder.alloca(ir.IntType(64), name=var_name)
+            # var.align = 8
+            builder.store(
+                ir.Constant(ir.IntType(64), rval.value), local_sym_tab[var_name].var
+            )
+            logger.info(f"Assigned constant {rval.value} to {var_name}")
+        elif isinstance(rval.value, str):
+            str_val = rval.value.encode("utf-8") + b"\x00"
+            str_const = ir.Constant(
+                ir.ArrayType(ir.IntType(8), len(str_val)), bytearray(str_val)
+            )
+            global_str = ir.GlobalVariable(
+                module, str_const.type, name=f"{var_name}_str"
+            )
+            global_str.linkage = "internal"
+            global_str.global_constant = True
+            global_str.initializer = str_const
+            str_ptr = builder.bitcast(global_str, ir.PointerType(ir.IntType(8)))
+            builder.store(str_ptr, local_sym_tab[var_name].var)
+            logger.info(f"Assigned string constant '{rval.value}' to {var_name}")
+        else:
+            logger.info("Unsupported constant type")
+    elif isinstance(rval, ast.Call):
+        if isinstance(rval.func, ast.Name):
+            call_type = rval.func.id
+            logger.info(f"Assignment call type: {call_type}")
+            if (
+                call_type in num_types
+                and len(rval.args) == 1
+                and isinstance(rval.args[0], ast.Constant)
+                and isinstance(rval.args[0].value, int)
+            ):
+                ir_type = ctypes_to_ir(call_type)
+                # var = builder.alloca(ir_type, name=var_name)
+                # var.align = ir_type.width // 8
+                builder.store(
+                    ir.Constant(ir_type, rval.args[0].value),
+                    local_sym_tab[var_name].var,
+                )
+                logger.info(
+                    f"Assigned {call_type} constant {rval.args[0].value} to {var_name}"
+                )
+            elif HelperHandlerRegistry.has_handler(call_type):
+                # var = builder.alloca(ir.IntType(64), name=var_name)
+                # var.align = 8
+                val = handle_helper_call(
+                    rval,
+                    module,
+                    builder,
+                    func,
+                    local_sym_tab,
+                    map_sym_tab,
+                    structs_sym_tab,
+                )
+                builder.store(val[0], local_sym_tab[var_name].var)
+                logger.info(f"Assigned constant {rval.func.id} to {var_name}")
+            elif call_type == "deref" and len(rval.args) == 1:
+                logger.info(f"Handling deref assignment {ast.dump(rval)}")
+                val = eval_expr(
+                    func,
+                    module,
+                    builder,
+                    rval,
+                    local_sym_tab,
+                    map_sym_tab,
+                    structs_sym_tab,
+                )
+                if val is None:
+                    logger.info("Failed to evaluate deref argument")
+                    return
+                logger.info(f"Dereferenced value: {val}, storing in {var_name}")
+                builder.store(val[0], local_sym_tab[var_name].var)
+                logger.info(f"Dereferenced and assigned to {var_name}")
+            elif call_type in structs_sym_tab and len(rval.args) == 0:
+                struct_info = structs_sym_tab[call_type]
+                ir_type = struct_info.ir_type
+                # var = builder.alloca(ir_type, name=var_name)
+                # Null init
+                builder.store(ir.Constant(ir_type, None), local_sym_tab[var_name].var)
+                logger.info(f"Assigned struct {call_type} to {var_name}")
+            else:
+                logger.info(f"Unsupported assignment call type: {call_type}")
+        elif isinstance(rval.func, ast.Attribute):
+            logger.info(f"Assignment call attribute: {ast.dump(rval.func)}")
+            if isinstance(rval.func.value, ast.Name):
+                if rval.func.value.id in map_sym_tab:
+                    map_name = rval.func.value.id
+                    method_name = rval.func.attr
+                    if HelperHandlerRegistry.has_handler(method_name):
+                        val = handle_helper_call(
+                            rval,
+                            module,
+                            builder,
+                            func,
+                            local_sym_tab,
+                            map_sym_tab,
+                            structs_sym_tab,
+                        )
+                        builder.store(val[0], local_sym_tab[var_name].var)
+                else:
+                    # TODO: probably a struct access
+                    logger.info(f"TODO STRUCT ACCESS {ast.dump(rval)}")
+            elif isinstance(rval.func.value, ast.Call) and isinstance(
+                rval.func.value.func, ast.Name
+            ):
+                map_name = rval.func.value.func.id
+                method_name = rval.func.attr
+                if map_name in map_sym_tab:
+                    if HelperHandlerRegistry.has_handler(method_name):
+                        val = handle_helper_call(
+                            rval,
+                            module,
+                            builder,
+                            func,
+                            local_sym_tab,
+                            map_sym_tab,
+                            structs_sym_tab,
+                        )
+                        # var = builder.alloca(ir.IntType(64), name=var_name)
+                        # var.align = 8
+                        builder.store(val[0], local_sym_tab[var_name].var)
+            else:
+                logger.info("Unsupported assignment call structure")
+        else:
+            logger.info("Unsupported assignment call function type")
+    elif isinstance(rval, ast.BinOp):
+        handle_binary_op(rval, builder, var_name, local_sym_tab)
+    else:
+        logger.info("Unsupported assignment value type")
 
 
 def handle_cond(
     func, module, builder, cond, local_sym_tab, map_sym_tab, structs_sym_tab=None
 ):
+    """
+    Evaluate a condition expression and convert it to a boolean value.
+    
+    Args:
+        func: The LLVM IR function being built
+        module: The LLVM IR module
+        builder: LLVM IR builder
+        cond: The AST condition node to evaluate
+        local_sym_tab: Local symbol table
+        map_sym_tab: Map symbol table
+        structs_sym_tab: Struct symbol table
+    
+    Returns:
+        LLVM IR boolean value representing the condition result
+    """
     val = eval_expr(
         func, module, builder, cond, local_sym_tab, map_sym_tab, structs_sym_tab
     )[0]
@@ -144,6 +330,18 @@ def handle_if(
 
 
 def handle_return(builder, stmt, local_sym_tab, ret_type):
+    """
+    Handle return statements in BPF functions.
+    
+    Args:
+        builder: LLVM IR builder
+        stmt: The AST Return node
+        local_sym_tab: Local symbol table
+        ret_type: Expected return type
+    
+    Returns:
+        True if a return was emitted, False otherwise
+    """
     logger.info(f"Handling return statement: {ast.dump(stmt)}")
     if stmt.value is None:
         return _handle_none_return(builder)
@@ -175,8 +373,24 @@ def process_stmt(
     did_return,
     ret_type=ir.IntType(64),
 ):
+    """
+    Process a single statement in a BPF function.
+    
+    Args:
+        func: The LLVM IR function being built
+        module: The LLVM IR module
+        builder: LLVM IR builder
+        stmt: The AST statement node to process
+        local_sym_tab: Local symbol table
+        map_sym_tab: Map symbol table
+        structs_sym_tab: Struct symbol table
+        did_return: Whether a return has been emitted
+        ret_type: Expected return type
+    
+    Returns:
+        True if a return was emitted, False otherwise
+    """
     logger.info(f"Processing statement: {ast.dump(stmt)}")
-    reset_scratch_pool()
     if isinstance(stmt, ast.Expr):
         handle_expr(
             func,
@@ -207,107 +421,138 @@ def process_stmt(
     return did_return
 
 
-def handle_if_allocation(
-    module, builder, stmt, func, ret_type, map_sym_tab, local_sym_tab, structs_sym_tab
-):
-    """Recursively handle allocations in if/else branches."""
-    if stmt.body:
-        allocate_mem(
-            module,
-            builder,
-            stmt.body,
-            func,
-            ret_type,
-            map_sym_tab,
-            local_sym_tab,
-            structs_sym_tab,
-        )
-    if stmt.orelse:
-        allocate_mem(
-            module,
-            builder,
-            stmt.orelse,
-            func,
-            ret_type,
-            map_sym_tab,
-            local_sym_tab,
-            structs_sym_tab,
-        )
-
-
-def count_temps_in_call(call_node, local_sym_tab):
-    """Count the number of temporary variables needed for a function call."""
-
-    count = 0
-    is_helper = False
-
-    # NOTE: We exclude print calls for now
-    if isinstance(call_node.func, ast.Name):
-        if (
-            HelperHandlerRegistry.has_handler(call_node.func.id)
-            and call_node.func.id != "print"
-        ):
-            is_helper = True
-    elif isinstance(call_node.func, ast.Attribute):
-        if HelperHandlerRegistry.has_handler(call_node.func.attr):
-            is_helper = True
-
-    if not is_helper:
-        return 0
-
-    for arg in call_node.args:
-        # NOTE: Count all non-name arguments
-        # For struct fields, if it is being passed as an argument,
-        # The struct object should already exist in the local_sym_tab
-        if not isinstance(arg, ast.Name) and not (
-            isinstance(arg, ast.Attribute) and arg.value.id in local_sym_tab
-        ):
-            count += 1
-
-    return count
-
-
 def allocate_mem(
     module, builder, body, func, ret_type, map_sym_tab, local_sym_tab, structs_sym_tab
 ):
-    max_temps_needed = 0
+    """
+    Pre-allocate stack memory for local variables in a BPF function.
     
-    def update_max_temps_for_stmt(stmt):
-        nonlocal max_temps_needed
-        temps_needed = 0
+    This function scans the function body and creates alloca instructions
+    for all local variables before processing the function statements.
     
-        if isinstance(stmt, ast.If):
-            for s in stmt.body:
-                update_max_temps_for_stmt(s)
-            for s in stmt.orelse:
-                update_max_temps_for_stmt(s)
-            return
-
-        for node in ast.walk(stmt):
-            if isinstance(node, ast.Call):
-                temps_needed += count_temps_in_call(node, local_sym_tab)
-        max_temps_needed = max(max_temps_needed, temps_needed)
+    Args:
+        module: The LLVM IR module
+        builder: LLVM IR builder
+        body: List of AST statements in the function body
+        func: The LLVM IR function being built
+        ret_type: Expected return type
+        map_sym_tab: Map symbol table
+        local_sym_tab: Local symbol table to populate
+        structs_sym_tab: Struct symbol table
     
+    Returns:
+        Updated local symbol table
+    """
     for stmt in body:
-        update_max_temps_for_stmt(stmt)
-
-        # Handle allocations
+        has_metadata = False
         if isinstance(stmt, ast.If):
-            handle_if_allocation(
-                module,
-                builder,
-                stmt,
-                func,
-                ret_type,
-                map_sym_tab,
-                local_sym_tab,
-                structs_sym_tab,
-            )
+            if stmt.body:
+                local_sym_tab = allocate_mem(
+                    module,
+                    builder,
+                    stmt.body,
+                    func,
+                    ret_type,
+                    map_sym_tab,
+                    local_sym_tab,
+                    structs_sym_tab,
+                )
+            if stmt.orelse:
+                local_sym_tab = allocate_mem(
+                    module,
+                    builder,
+                    stmt.orelse,
+                    func,
+                    ret_type,
+                    map_sym_tab,
+                    local_sym_tab,
+                    structs_sym_tab,
+                )
         elif isinstance(stmt, ast.Assign):
-            handle_assign_allocation(builder, stmt, local_sym_tab, structs_sym_tab)
-
-    allocate_temp_pool(builder, max_temps_needed, local_sym_tab)
+            if len(stmt.targets) != 1:
+                logger.info("Unsupported multiassignment")
+                continue
+            target = stmt.targets[0]
+            if not isinstance(target, ast.Name):
+                logger.info("Unsupported assignment target")
+                continue
+            var_name = target.id
+            rval = stmt.value
+            if var_name in local_sym_tab:
+                logger.info(f"Variable {var_name} already allocated")
+                continue
+            if isinstance(rval, ast.Call):
+                if isinstance(rval.func, ast.Name):
+                    call_type = rval.func.id
+                    if call_type in ("c_int32", "c_int64", "c_uint32", "c_uint64"):
+                        ir_type = ctypes_to_ir(call_type)
+                        var = builder.alloca(ir_type, name=var_name)
+                        var.align = ir_type.width // 8
+                        logger.info(
+                            f"Pre-allocated variable {var_name} of type {call_type}"
+                        )
+                    elif HelperHandlerRegistry.has_handler(call_type):
+                        # Assume return type is int64 for now
+                        ir_type = ir.IntType(64)
+                        var = builder.alloca(ir_type, name=var_name)
+                        var.align = ir_type.width // 8
+                        logger.info(f"Pre-allocated variable {var_name} for helper")
+                    elif call_type == "deref" and len(rval.args) == 1:
+                        # Assume return type is int64 for now
+                        ir_type = ir.IntType(64)
+                        var = builder.alloca(ir_type, name=var_name)
+                        var.align = ir_type.width // 8
+                        logger.info(f"Pre-allocated variable {var_name} for deref")
+                    elif call_type in structs_sym_tab:
+                        struct_info = structs_sym_tab[call_type]
+                        ir_type = struct_info.ir_type
+                        var = builder.alloca(ir_type, name=var_name)
+                        has_metadata = True
+                        logger.info(
+                            f"Pre-allocated variable {var_name} for struct {call_type}"
+                        )
+                elif isinstance(rval.func, ast.Attribute):
+                    ir_type = ir.PointerType(ir.IntType(64))
+                    var = builder.alloca(ir_type, name=var_name)
+                    # var.align = ir_type.width // 8
+                    logger.info(f"Pre-allocated variable {var_name} for map")
+                else:
+                    logger.info("Unsupported assignment call function type")
+                    continue
+            elif isinstance(rval, ast.Constant):
+                if isinstance(rval.value, bool):
+                    ir_type = ir.IntType(1)
+                    var = builder.alloca(ir_type, name=var_name)
+                    var.align = 1
+                    logger.info(f"Pre-allocated variable {var_name} of type c_bool")
+                elif isinstance(rval.value, int):
+                    # Assume c_int64 for now
+                    ir_type = ir.IntType(64)
+                    var = builder.alloca(ir_type, name=var_name)
+                    var.align = ir_type.width // 8
+                    logger.info(f"Pre-allocated variable {var_name} of type c_int64")
+                elif isinstance(rval.value, str):
+                    ir_type = ir.PointerType(ir.IntType(8))
+                    var = builder.alloca(ir_type, name=var_name)
+                    var.align = 8
+                    logger.info(f"Pre-allocated variable {var_name} of type string")
+                else:
+                    logger.info("Unsupported constant type")
+                    continue
+            elif isinstance(rval, ast.BinOp):
+                # Assume c_int64 for now
+                ir_type = ir.IntType(64)
+                var = builder.alloca(ir_type, name=var_name)
+                var.align = ir_type.width // 8
+                logger.info(f"Pre-allocated variable {var_name} of type c_int64")
+            else:
+                logger.info("Unsupported assignment value type")
+                continue
 
+            if has_metadata:
+                local_sym_tab[var_name] = LocalSymbol(var, ir_type, call_type)
+            else:
+                local_sym_tab[var_name] = LocalSymbol(var, ir_type)
     return local_sym_tab
 
 
@@ -391,6 +636,16 @@ def process_bpf_chunk(func_node, module, return_type, map_sym_tab, structs_sym_t
 
 
 def func_proc(tree, module, chunks, map_sym_tab, structs_sym_tab):
+    """
+    Process all BPF function chunks and generate LLVM IR.
+    
+    Args:
+        tree: The Python AST (not used in current implementation)
+        module: The LLVM IR module to add functions to
+        chunks: List of AST function nodes decorated with @bpf
+        map_sym_tab: Map symbol table
+        structs_sym_tab: Struct symbol table
+    """
     for func_node in chunks:
         is_global = False
         for decorator in func_node.decorator_list:
@@ -416,6 +671,18 @@ def func_proc(tree, module, chunks, map_sym_tab, structs_sym_tab):
 
 
 def infer_return_type(func_node: ast.FunctionDef):
+    """
+    Infer the return type of a BPF function from annotations or return statements.
+    
+    Args:
+        func_node: The AST function node
+    
+    Returns:
+        String representation of the return type (e.g., 'c_int64')
+    
+    Raises:
+        TypeError: If func_node is not a FunctionDef
+    """
     if not isinstance(func_node, (ast.FunctionDef, ast.AsyncFunctionDef)):
         raise TypeError("Expected ast.FunctionDef")
     if func_node.returns is not None:
@@ -434,6 +701,7 @@ def infer_return_type(func_node: ast.FunctionDef):
     found_type = None
 
     def _expr_type(e):
+        """Helper function to extract type from an expression."""
         if e is None:
             return "None"
         if isinstance(e, ast.Constant):

pythonbpf/functions/return_utils.py

@@ -1,3 +1,10 @@
+"""
+Utility functions for handling return statements in BPF functions.
+
+Provides handlers for different types of returns including XDP actions,
+None returns, and standard returns.
+"""
+
 import logging
 import ast
 

pythonbpf/globals_pass.py

@@ -1,3 +1,10 @@
+"""
+Global variables and compiler metadata processing.
+
+This module handles BPF global variables and emits the @llvm.compiler.used
+metadata to prevent LLVM from optimizing away important symbols.
+"""
+
 from llvmlite import ir
 import ast
 
@@ -12,6 +19,16 @@ global_sym_tab = []
 
 
 def populate_global_symbol_table(tree, module: ir.Module):
+    """
+    Populate the global symbol table with BPF functions, maps, and globals.
+    
+    Args:
+        tree: The Python AST to scan for global symbols
+        module: The LLVM IR module (not used in current implementation)
+    
+    Returns:
+        False (legacy return value)
+    """
     for node in tree.body:
         if isinstance(node, ast.FunctionDef):
             for dec in node.decorator_list:
@@ -33,6 +50,17 @@ def populate_global_symbol_table(tree, module: ir.Module):
 
 
 def emit_global(module: ir.Module, node, name):
+    """
+    Emit a BPF global variable into the LLVM IR module.
+    
+    Args:
+        module: The LLVM IR module to add the global variable to
+        node: The AST function node containing the global definition
+        name: The name of the global variable
+    
+    Returns:
+        The created global variable
+    """
     logger.info(f"global identifier {name} processing")
     # deduce LLVM type from the annotated return
     if not isinstance(node.returns, ast.Name):
@@ -117,7 +145,11 @@ def globals_processing(tree, module):
 
 def emit_llvm_compiler_used(module: ir.Module, names: list[str]):
     """
-    Emit the @llvm.compiler.used global given a list of function/global names.
+    Emit the @llvm.compiler.used global to prevent LLVM from optimizing away symbols.
+    
+    Args:
+        module: The LLVM IR module to add the compiler.used metadata to
+        names: List of function/global names that must be preserved
     """
     ptr_ty = ir.PointerType()
     used_array_ty = ir.ArrayType(ptr_ty, len(names))
@@ -138,6 +170,13 @@ def emit_llvm_compiler_used(module: ir.Module, names: list[str]):
 
 
 def globals_list_creation(tree, module: ir.Module):
+    """
+    Collect all BPF symbols and emit @llvm.compiler.used metadata.
+    
+    Args:
+        tree: The Python AST to scan for symbols
+        module: The LLVM IR module to add metadata to
+    """
     collected = ["LICENSE"]
 
     for node in tree.body:

pythonbpf/helper/__init__.py

@@ -1,10 +1,11 @@
-from .helper_utils import HelperHandlerRegistry, reset_scratch_pool
+"""BPF helper functions and handlers."""
+
+from .helper_utils import HelperHandlerRegistry
 from .bpf_helper_handler import handle_helper_call
 from .helpers import ktime, pid, deref, XDP_DROP, XDP_PASS
 
 __all__ = [
     "HelperHandlerRegistry",
-    "reset_scratch_pool",
     "handle_helper_call",
     "ktime",
     "pid",

pythonbpf/helper/bpf_helper_handler.py

@@ -1,3 +1,11 @@
+"""
+BPF helper function handlers for LLVM IR emission.
+
+This module provides handlers for various BPF helper functions, emitting
+the appropriate LLVM IR to call kernel BPF helpers like map operations,
+printing, time functions, etc.
+"""
+
 import ast
 from llvmlite import ir
 from enum import Enum
@@ -16,6 +24,7 @@ logger: Logger = logging.getLogger(__name__)
 
 
 class BPFHelperID(Enum):
+    """Enumeration of BPF helper function IDs."""
     BPF_MAP_LOOKUP_ELEM = 1
     BPF_MAP_UPDATE_ELEM = 2
     BPF_MAP_DELETE_ELEM = 3
@@ -34,7 +43,6 @@ def bpf_ktime_get_ns_emitter(
     func,
     local_sym_tab=None,
     struct_sym_tab=None,
-    map_sym_tab=None,
 ):
     """
     Emit LLVM IR for bpf_ktime_get_ns helper function call.
@@ -57,7 +65,6 @@ def bpf_map_lookup_elem_emitter(
     func,
     local_sym_tab=None,
     struct_sym_tab=None,
-    map_sym_tab=None,
 ):
     """
     Emit LLVM IR for bpf_map_lookup_elem helper function call.
@@ -66,17 +73,11 @@ def bpf_map_lookup_elem_emitter(
         raise ValueError(
             f"Map lookup expects exactly one argument (key), got {len(call.args)}"
         )
-    key_ptr = get_or_create_ptr_from_arg(
-        func, module, call.args[0], builder, local_sym_tab, map_sym_tab, struct_sym_tab
-    )
+    key_ptr = get_or_create_ptr_from_arg(call.args[0], builder, local_sym_tab)
     map_void_ptr = builder.bitcast(map_ptr, ir.PointerType())
 
-    # TODO: I have changed the return type to i64*, as we are
-    # allocating space for that type in allocate_mem. This is
-    # temporary, and we will honour other widths later. But this
-    # allows us to have cool binary ops on the returned value.
     fn_type = ir.FunctionType(
-        ir.PointerType(ir.IntType(64)),  # Return type: void*
+        ir.PointerType(),  # Return type: void*
         [ir.PointerType(), ir.PointerType()],  # Args: (void*, void*)
         var_arg=False,
     )
@@ -99,7 +100,6 @@ def bpf_printk_emitter(
     func,
     local_sym_tab=None,
     struct_sym_tab=None,
-    map_sym_tab=None,
 ):
     """Emit LLVM IR for bpf_printk helper function call."""
     if not hasattr(func, "_fmt_counter"):
@@ -147,7 +147,6 @@ def bpf_map_update_elem_emitter(
     func,
     local_sym_tab=None,
     struct_sym_tab=None,
-    map_sym_tab=None,
 ):
     """
     Emit LLVM IR for bpf_map_update_elem helper function call.
@@ -162,12 +161,8 @@ def bpf_map_update_elem_emitter(
     value_arg = call.args[1]
     flags_arg = call.args[2] if len(call.args) > 2 else None
 
-    key_ptr = get_or_create_ptr_from_arg(
-        func, module, key_arg, builder, local_sym_tab, map_sym_tab, struct_sym_tab
-    )
-    value_ptr = get_or_create_ptr_from_arg(
-        func, module, value_arg, builder, local_sym_tab, map_sym_tab, struct_sym_tab
-    )
+    key_ptr = get_or_create_ptr_from_arg(key_arg, builder, local_sym_tab)
+    value_ptr = get_or_create_ptr_from_arg(value_arg, builder, local_sym_tab)
     flags_val = get_flags_val(flags_arg, builder, local_sym_tab)
 
     map_void_ptr = builder.bitcast(map_ptr, ir.PointerType())
@@ -202,7 +197,6 @@ def bpf_map_delete_elem_emitter(
     func,
     local_sym_tab=None,
     struct_sym_tab=None,
-    map_sym_tab=None,
 ):
     """
     Emit LLVM IR for bpf_map_delete_elem helper function call.
@@ -212,9 +206,7 @@ def bpf_map_delete_elem_emitter(
         raise ValueError(
             f"Map delete expects exactly one argument (key), got {len(call.args)}"
         )
-    key_ptr = get_or_create_ptr_from_arg(
-        func, module, call.args[0], builder, local_sym_tab, map_sym_tab, struct_sym_tab
-    )
+    key_ptr = get_or_create_ptr_from_arg(call.args[0], builder, local_sym_tab)
     map_void_ptr = builder.bitcast(map_ptr, ir.PointerType())
 
     # Define function type for bpf_map_delete_elem
@@ -242,7 +234,6 @@ def bpf_get_current_pid_tgid_emitter(
     func,
     local_sym_tab=None,
     struct_sym_tab=None,
-    map_sym_tab=None,
 ):
     """
     Emit LLVM IR for bpf_get_current_pid_tgid helper function call.
@@ -269,8 +260,12 @@ def bpf_perf_event_output_handler(
     func,
     local_sym_tab=None,
     struct_sym_tab=None,
-    map_sym_tab=None,
 ):
+    """
+    Emit LLVM IR for bpf_perf_event_output helper function call.
+    
+    This allows sending data to userspace via a perf event array.
+    """
     if len(call.args) != 1:
         raise ValueError(
             f"Perf event output expects exactly one argument, got {len(call.args)}"
@@ -321,6 +316,7 @@ def handle_helper_call(
 
     # Helper function to get map pointer and invoke handler
     def invoke_helper(method_name, map_ptr=None):
+        """Helper function to look up and invoke a registered handler."""
         handler = HelperHandlerRegistry.get_handler(method_name)
         if not handler:
             raise NotImplementedError(
@@ -334,7 +330,6 @@ def handle_helper_call(
             func,
             local_sym_tab,
             struct_sym_tab,
-            map_sym_tab,
         )
 
     # Handle direct function calls (e.g., print(), ktime())

pythonbpf/helper/helper_utils.py

@@ -1,10 +1,17 @@
+"""
+Utility functions for BPF helper function handling.
+
+This module provides utility functions for processing BPF helper function
+calls, including argument handling, string formatting for bpf_printk,
+and a registry for helper function handlers.
+"""
+
 import ast
 import logging
 from collections.abc import Callable
 
 from llvmlite import ir
-from pythonbpf.expr import eval_expr, get_base_type_and_depth, deref_to_depth
-from pythonbpf.binary_ops import get_operand_value
+from pythonbpf.expr import eval_expr
 
 logger = logging.getLogger(__name__)
 
@@ -19,6 +26,7 @@ class HelperHandlerRegistry:
         """Decorator to register a handler function for a helper"""
 
         def decorator(func):
+            """Decorator that registers the handler function."""
             cls._handlers[helper_name] = func
             return func
 
@@ -35,88 +43,88 @@ class HelperHandlerRegistry:
         return helper_name in cls._handlers
 
 
-class ScratchPoolManager:
-    """Manage the temporary helper variables in local_sym_tab"""
-
-    def __init__(self):
-        self._counter = 0
-
-    @property
-    def counter(self):
-        return self._counter
-
-    def reset(self):
-        self._counter = 0
-        logger.debug("Scratch pool counter reset to 0")
-
-    def get_next_temp(self, local_sym_tab):
-        temp_name = f"__helper_temp_{self._counter}"
-        self._counter += 1
-
-        if temp_name not in local_sym_tab:
-            raise ValueError(
-                f"Scratch pool exhausted or inadequate: {temp_name}. "
-                f"Current counter: {self._counter}"
-            )
-
-        return local_sym_tab[temp_name].var, temp_name
-
-
-_temp_pool_manager = ScratchPoolManager()  # Singleton instance
-
-
-def reset_scratch_pool():
-    """Reset the scratch pool counter"""
-    _temp_pool_manager.reset()
-
-
 def get_var_ptr_from_name(var_name, local_sym_tab):
-    """Get a pointer to a variable from the symbol table."""
+    """
+    Get a pointer to a variable from the symbol table.
+    
+    Args:
+        var_name: Name of the variable to look up
+        local_sym_tab: Local symbol table
+    
+    Returns:
+        Pointer to the variable
+    
+    Raises:
+        ValueError: If the variable is not found
+    """
     if local_sym_tab and var_name in local_sym_tab:
         return local_sym_tab[var_name].var
     raise ValueError(f"Variable '{var_name}' not found in local symbol table")
 
 
-def create_int_constant_ptr(value, builder, local_sym_tab, int_width=64):
-    """Create a pointer to an integer constant."""
+def create_int_constant_ptr(value, builder, int_width=64):
+    """
+    Create a pointer to an integer constant.
     
+    Args:
+        value: The integer value
+        builder: LLVM IR builder
+        int_width: Width of the integer in bits (default: 64)
+    
+    Returns:
+        Pointer to the allocated integer constant
+    """
     # Default to 64-bit integer
-    ptr, temp_name = _temp_pool_manager.get_next_temp(local_sym_tab)
-    logger.info(f"Using temp variable '{temp_name}' for int constant {value}")
-    const_val = ir.Constant(ir.IntType(int_width), value)
-    builder.store(const_val, ptr)
+    int_type = ir.IntType(int_width)
+    ptr = builder.alloca(int_type)
+    ptr.align = int_type.width // 8
+    builder.store(ir.Constant(int_type, value), ptr)
     return ptr
 
 
-def get_or_create_ptr_from_arg(
-    func, module, arg, builder, local_sym_tab, map_sym_tab, struct_sym_tab=None
-):
-    """Extract or create pointer from the call arguments."""
+def get_or_create_ptr_from_arg(arg, builder, local_sym_tab):
+    """
+    Extract or create pointer from call arguments.
+    
+    Args:
+        arg: The AST argument node
+        builder: LLVM IR builder
+        local_sym_tab: Local symbol table
+    
+    Returns:
+        Pointer to the argument value
+    
+    Raises:
+        NotImplementedError: If the argument type is not supported
+    """
 
     if isinstance(arg, ast.Name):
         ptr = get_var_ptr_from_name(arg.id, local_sym_tab)
     elif isinstance(arg, ast.Constant) and isinstance(arg.value, int):
-        ptr = create_int_constant_ptr(arg.value, builder, local_sym_tab)
+        ptr = create_int_constant_ptr(arg.value, builder)
     else:
-        # Evaluate the expression and store the result in a temp variable
-        val = get_operand_value(
-            func, module, arg, builder, local_sym_tab, map_sym_tab, struct_sym_tab
+        raise NotImplementedError(
+            "Only simple variable names are supported as args in map helpers."
         )
-        if val is None:
-            raise ValueError("Failed to evaluate expression for helper arg.")
-
-        # NOTE: We assume the result is an int64 for now
-        # if isinstance(arg, ast.Attribute):
-        # return val
-        ptr, temp_name = _temp_pool_manager.get_next_temp(local_sym_tab)
-        logger.info(f"Using temp variable '{temp_name}' for expression result")
-        builder.store(val, ptr)
-
     return ptr
 
 
 def get_flags_val(arg, builder, local_sym_tab):
-    """Extract or create flags value from the call arguments."""
+    """
+    Extract or create flags value from call arguments.
+    
+    Args:
+        arg: The AST argument node for flags
+        builder: LLVM IR builder
+        local_sym_tab: Local symbol table
+    
+    Returns:
+        Integer flags value or LLVM IR value
+    
+    Raises:
+        ValueError: If a variable is not found in symbol table
+        NotImplementedError: If the argument type is not supported
+    """
     if not arg:
         return 0
 
@@ -135,7 +143,18 @@ def get_flags_val(arg, builder, local_sym_tab):
 
 
 def simple_string_print(string_value, module, builder, func):
-    """Prepare arguments for bpf_printk from a simple string value"""
+    """
+    Prepare arguments for bpf_printk from a simple string value.
+    
+    Args:
+        string_value: The string to print
+        module: LLVM IR module
+        builder: LLVM IR builder
+        func: The LLVM IR function being built
+    
+    Returns:
+        List of arguments for bpf_printk
+    """
     fmt_str = string_value + "\n\0"
     fmt_ptr = _create_format_string_global(fmt_str, func, module, builder)
 
@@ -151,7 +170,23 @@ def handle_fstring_print(
     local_sym_tab=None,
     struct_sym_tab=None,
 ):
-    """Handle f-string formatting for bpf_printk emitter."""
+    """
+    Handle f-string formatting for bpf_printk emitter.
+    
+    Args:
+        joined_str: AST JoinedStr node representing the f-string
+        module: LLVM IR module
+        builder: LLVM IR builder
+        func: The LLVM IR function being built
+        local_sym_tab: Local symbol table
+        struct_sym_tab: Struct symbol table
+    
+    Returns:
+        List of arguments for bpf_printk
+    
+    Raises:
+        NotImplementedError: If f-string contains unsupported value types
+    """
     fmt_parts = []
     exprs = []
 
@@ -274,27 +309,10 @@ def _populate_fval(ftype, node, fmt_parts, exprs):
             raise NotImplementedError(
                 f"Unsupported integer width in f-string: {ftype.width}"
             )
-    elif isinstance(ftype, ir.PointerType):
-        target, depth = get_base_type_and_depth(ftype)
-        if isinstance(target, ir.IntType):
-            if target.width == 64:
-                fmt_parts.append("%lld")
-                exprs.append(node)
-            elif target.width == 32:
-                fmt_parts.append("%d")
-                exprs.append(node)
-            elif target.width == 8 and depth == 1:
-                # NOTE: Assume i8* is a string
-                fmt_parts.append("%s")
-                exprs.append(node)
-            else:
-                raise NotImplementedError(
-                    f"Unsupported pointer target type in f-string: {target}"
-                )
-        else:
-            raise NotImplementedError(
-                f"Unsupported pointer target type in f-string: {target}"
-            )
+    elif ftype == ir.PointerType(ir.IntType(8)):
+        # NOTE: We assume i8* is a string
+        fmt_parts.append("%s")
+        exprs.append(node)
     else:
         raise NotImplementedError(f"Unsupported field type in f-string: {ftype}")
 
@@ -331,20 +349,7 @@ def _prepare_expr_args(expr, func, module, builder, local_sym_tab, struct_sym_ta
 
     if val:
         if isinstance(val.type, ir.PointerType):
-            target, depth = get_base_type_and_depth(val.type)
-            if isinstance(target, ir.IntType):
-                if target.width >= 32:
-                    val = deref_to_depth(func, builder, val, depth)
-                    val = builder.sext(val, ir.IntType(64))
-                elif target.width == 8 and depth == 1:
-                    # NOTE: i8* is string, no need to deref
-                    pass
-
-            else:
-                logger.warning(
-                    "Only int and ptr supported in bpf_printk args. Others default to 0."
-                )
-                val = ir.Constant(ir.IntType(64), 0)
+            val = builder.ptrtoint(val, ir.IntType(64))
         elif isinstance(val.type, ir.IntType):
             if val.type.width < 64:
                 val = builder.sext(val, ir.IntType(64))

pythonbpf/helper/helpers.py

@@ -1,11 +1,31 @@
+"""
+BPF helper function stubs for Python type hints.
+
+This module provides Python stub functions that represent BPF helper functions.
+These stubs are used for type checking and will be replaced with actual BPF
+helper calls during compilation.
+"""
+
 import ctypes
 
 
 def ktime():
+    """
+    Get the current kernel time in nanoseconds.
+    
+    Returns:
+        A c_int64 stub value (actual implementation is in BPF runtime)
+    """
     return ctypes.c_int64(0)
 
 
 def pid():
+    """
+    Get the current process ID (PID).
+    
+    Returns:
+        A c_int32 stub value (actual implementation is in BPF runtime)
+    """
     return ctypes.c_int32(0)
 
 

pythonbpf/license_pass.py

@@ -1,3 +1,10 @@
+"""
+LICENSE global variable processing for BPF programs.
+
+This module handles the processing of the LICENSE function which is required
+for BPF programs to declare their license (typically "GPL").
+"""
+
 from llvmlite import ir
 import ast
 from logging import Logger
@@ -7,6 +14,16 @@ logger: Logger = logging.getLogger(__name__)
 
 
 def emit_license(module: ir.Module, license_str: str):
+    """
+    Emit a LICENSE global variable into the LLVM IR module.
+    
+    Args:
+        module: The LLVM IR module to add the LICENSE variable to
+        license_str: The license string (e.g., 'GPL')
+    
+    Returns:
+        The created global variable
+    """
     license_bytes = license_str.encode("utf8") + b"\x00"
     elems = [ir.Constant(ir.IntType(8), b) for b in license_bytes]
     ty = ir.ArrayType(ir.IntType(8), len(elems))

pythonbpf/maps/__init__.py

@@ -1,3 +1,5 @@
+"""BPF map types and processing."""
+
 from .maps import HashMap, PerfEventArray, RingBuf
 from .maps_pass import maps_proc
 

pythonbpf/maps/maps.py

@@ -1,18 +1,59 @@
+"""
+BPF map type definitions for Python type hints.
+
+This module provides Python classes that represent BPF map types.
+These are used for type checking and map definition; the actual BPF maps
+are generated as LLVM IR during compilation.
+"""
+
 # This file provides type  and function hints only and does not actually give any functionality.
 class HashMap:
+    """
+    A BPF hash map for storing key-value pairs.
+    
+    This is a type hint class used during compilation. The actual BPF map
+    implementation is generated as LLVM IR.
+    """
+    
     def __init__(self, key, value, max_entries):
+        """
+        Initialize a HashMap definition.
+        
+        Args:
+            key: The ctypes type for keys (e.g., c_int64)
+            value: The ctypes type for values (e.g., c_int64)
+            max_entries: Maximum number of entries the map can hold
+        """
         self.key = key
         self.value = value
         self.max_entries = max_entries
         self.entries = {}
 
     def lookup(self, key):
+        """
+        Look up a value by key in the map.
+        
+        Args:
+            key: The key to look up
+        
+        Returns:
+            The value if found, None otherwise
+        """
         if key in self.entries:
             return self.entries[key]
         else:
             return None
 
     def delete(self, key):
+        """
+        Delete an entry from the map by key.
+        
+        Args:
+            key: The key to delete
+        
+        Raises:
+            KeyError: If the key is not found in the map
+        """
         if key in self.entries:
             del self.entries[key]
         else:
@@ -20,6 +61,17 @@ class HashMap:
 
     # TODO: define the flags that can be added
     def update(self, key, value, flags=None):
+        """
+        Update or insert a key-value pair in the map.
+        
+        Args:
+            key: The key to update
+            value: The new value
+            flags: Optional flags for update behavior
+        
+        Raises:
+            KeyError: If the key is not found in the map
+        """
         if key in self.entries:
             self.entries[key] = value
         else:
@@ -27,25 +79,76 @@ class HashMap:
 
 
 class PerfEventArray:
+    """
+    A BPF perf event array for sending data to userspace.
+    
+    This is a type hint class used during compilation.
+    """
+    
     def __init__(self, key_size, value_size):
+        """
+        Initialize a PerfEventArray definition.
+        
+        Args:
+            key_size: The size/type for keys
+            value_size: The size/type for values
+        """
         self.key_type = key_size
         self.value_type = value_size
         self.entries = {}
 
     def output(self, data):
+        """
+        Output data to the perf event array.
+        
+        Args:
+            data: The data to output
+        """
         pass  # Placeholder for output method
 
 
 class RingBuf:
+    """
+    A BPF ring buffer for efficient data transfer to userspace.
+    
+    This is a type hint class used during compilation.
+    """
+    
     def __init__(self, max_entries):
+        """
+        Initialize a RingBuf definition.
+        
+        Args:
+            max_entries: Maximum number of entries the ring buffer can hold
+        """
         self.max_entries = max_entries
 
     def reserve(self, size: int, flags=0):
+        """
+        Reserve space in the ring buffer.
+        
+        Args:
+            size: Size in bytes to reserve
+            flags: Optional reservation flags
+        
+        Returns:
+            0 as a placeholder (actual implementation is in BPF runtime)
+        
+        Raises:
+            ValueError: If size exceeds max_entries
+        """
         if size > self.max_entries:
             raise ValueError("size cannot be greater than set maximum entries")
         return 0
 
     def submit(self, data, flags=0):
+        """
+        Submit data to the ring buffer.
+        
+        Args:
+            data: The data to submit
+            flags: Optional submission flags
+        """
         pass
 
     # add discard, output and also give names to flags and stuff

pythonbpf/maps/maps_pass.py

@@ -1,3 +1,10 @@
+"""
+BPF map processing and LLVM IR generation.
+
+This module handles the processing of BPF map definitions decorated with @map,
+converting them to appropriate LLVM IR global variables with BTF debug info.
+"""
+
 import ast
 from logging import Logger
 from llvmlite import ir
@@ -20,6 +27,15 @@ def maps_proc(tree, module, chunks):
 
 
 def is_map(func_node):
+    """
+    Check if a function node is decorated with @map.
+    
+    Args:
+        func_node: The AST function node to check
+    
+    Returns:
+        True if the function is decorated with @map, False otherwise
+    """
     return any(
         isinstance(decorator, ast.Name) and decorator.id == "map"
         for decorator in func_node.decorator_list
@@ -27,6 +43,7 @@ def is_map(func_node):
 
 
 class BPFMapType(Enum):
+    """Enumeration of BPF map types."""
     UNSPEC = 0
     HASH = 1
     ARRAY = 2
@@ -65,7 +82,17 @@ class BPFMapType(Enum):
 
 
 def create_bpf_map(module, map_name, map_params):
-    """Create a BPF map in the module with given parameters and debug info"""
+    """
+    Create a BPF map in the module with given parameters and debug info.
+    
+    Args:
+        module: The LLVM IR module to add the map to
+        map_name: The name of the BPF map
+        map_params: Dictionary of map parameters (type, key_size, value_size, max_entries)
+    
+    Returns:
+        The created global variable representing the map
+    """
 
     # Create the anonymous struct type for BPF map
     map_struct_type = ir.LiteralStructType(

pythonbpf/maps/maps_utils.py

@@ -1,3 +1,5 @@
+"""Registry for BPF map processor functions."""
+
 from collections.abc import Callable
 from typing import Any
 
@@ -12,6 +14,7 @@ class MapProcessorRegistry:
         """Decorator to register a processor function for a map type"""
 
         def decorator(func):
+            """Decorator that registers the processor function."""
             cls._processors[map_type_name] = func
             return func
 

pythonbpf/structs/__init__.py

@@ -1,3 +1,5 @@
+"""Struct processing for BPF programs."""
+
 from .structs_pass import structs_proc
 
 __all__ = ["structs_proc"]

pythonbpf/structs/struct_type.py

@@ -1,19 +1,72 @@
+"""
+Struct type wrapper for BPF structs.
+
+This module provides a wrapper class for LLVM IR struct types with
+helper methods for field access and manipulation.
+"""
+
 from llvmlite import ir
 
 
 class StructType:
+    """
+    Wrapper class for LLVM IR struct types with field access helpers.
+    
+    Attributes:
+        ir_type: The LLVM IR struct type
+        fields: Dictionary mapping field names to their types
+        size: Total size of the struct in bytes
+    """
+    
     def __init__(self, ir_type, fields, size):
+        """
+        Initialize a StructType.
+        
+        Args:
+            ir_type: The LLVM IR struct type
+            fields: Dictionary mapping field names to their types
+            size: Total size of the struct in bytes
+        """
         self.ir_type = ir_type
         self.fields = fields
         self.size = size
 
     def field_idx(self, field_name):
+        """
+        Get the index of a field in the struct.
+        
+        Args:
+            field_name: The name of the field
+        
+        Returns:
+            The zero-based index of the field
+        """
         return list(self.fields.keys()).index(field_name)
 
     def field_type(self, field_name):
+        """
+        Get the LLVM IR type of a field.
+        
+        Args:
+            field_name: The name of the field
+        
+        Returns:
+            The LLVM IR type of the field
+        """
         return self.fields[field_name]
 
     def gep(self, builder, ptr, field_name):
+        """
+        Generate a GEP (GetElementPtr) instruction to access a struct field.
+        
+        Args:
+            builder: LLVM IR builder
+            ptr: Pointer to the struct
+            field_name: Name of the field to access
+        
+        Returns:
+            A pointer to the field
+        """
         idx = self.field_idx(field_name)
         return builder.gep(
             ptr,
@@ -22,6 +75,18 @@ class StructType:
         )
 
     def field_size(self, field_name):
+        """
+        Calculate the size of a field in bytes.
+        
+        Args:
+            field_name: The name of the field
+        
+        Returns:
+            The size of the field in bytes
+        
+        Raises:
+            TypeError: If the field type is not supported
+        """
         fld = self.fields[field_name]
         if isinstance(fld, ir.ArrayType):
             return fld.count * (fld.element.width // 8)

pythonbpf/structs/structs_pass.py

@@ -1,3 +1,10 @@
+"""
+BPF struct processing and LLVM IR type generation.
+
+This module handles the processing of Python classes decorated with @struct,
+converting them to LLVM IR struct types for use in BPF programs.
+"""
+
 import ast
 import logging
 from llvmlite import ir
@@ -26,6 +33,15 @@ def structs_proc(tree, module, chunks):
 
 
 def is_bpf_struct(cls_node):
+    """
+    Check if a class node is decorated with @struct.
+    
+    Args:
+        cls_node: The AST class node to check
+    
+    Returns:
+        True if the class is decorated with @struct, False otherwise
+    """
     return any(
         isinstance(decorator, ast.Name) and decorator.id == "struct"
         for decorator in cls_node.decorator_list
@@ -33,7 +49,16 @@ def is_bpf_struct(cls_node):
 
 
 def process_bpf_struct(cls_node, module):
-    """Process a single BPF struct definition"""
+    """
+    Process a single BPF struct definition and create its LLVM IR representation.
+    
+    Args:
+        cls_node: The AST class node representing the struct
+        module: The LLVM IR module (not used in current implementation)
+    
+    Returns:
+        A StructType object containing the struct's type information
+    """
 
     fields = parse_struct_fields(cls_node)
     field_types = list(fields.values())
@@ -44,7 +69,18 @@ def process_bpf_struct(cls_node, module):
 
 
 def parse_struct_fields(cls_node):
-    """Parse fields of a struct class node"""
+    """
+    Parse fields of a struct class node.
+    
+    Args:
+        cls_node: The AST class node representing the struct
+    
+    Returns:
+        A dictionary mapping field names to their LLVM IR types
+    
+    Raises:
+        TypeError: If a field has an unsupported type annotation
+    """
     fields = {}
 
     for item in cls_node.body:
@@ -57,7 +93,18 @@ def parse_struct_fields(cls_node):
 
 
 def get_type_from_ann(annotation):
-    """Convert an AST annotation node to an LLVM IR type for struct fields"""
+    """
+    Convert an AST annotation node to an LLVM IR type for struct fields.
+    
+    Args:
+        annotation: The AST annotation node (e.g., c_int64, str(32))
+    
+    Returns:
+        The corresponding LLVM IR type
+    
+    Raises:
+        TypeError: If the annotation type is not supported
+    """
     if isinstance(annotation, ast.Call) and isinstance(annotation.func, ast.Name):
         if annotation.func.id == "str":
             # Char array
@@ -72,7 +119,15 @@ def get_type_from_ann(annotation):
 
 
 def calc_struct_size(field_types):
-    """Calculate total size of the struct with alignment and padding"""
+    """
+    Calculate total size of the struct with alignment and padding.
+    
+    Args:
+        field_types: List of LLVM IR types for each field
+    
+    Returns:
+        The total size of the struct in bytes
+    """
     curr_offset = 0
     for ftype in field_types:
         if isinstance(ftype, ir.IntType):

pythonbpf/type_deducer.py

@@ -1,3 +1,10 @@
+"""
+Type mapping from Python ctypes to LLVM IR types.
+
+This module provides utilities to convert Python ctypes type names
+to their corresponding LLVM IR representations.
+"""
+
 from llvmlite import ir
 
 # TODO: THIS IS NOT SUPPOSED TO MATCH STRINGS :skull:
@@ -19,10 +26,31 @@ mapping = {
 
 
 def ctypes_to_ir(ctype: str):
+    """
+    Convert a ctypes type name to its corresponding LLVM IR type.
+    
+    Args:
+        ctype: String name of the ctypes type (e.g., 'c_int64', 'c_void_p')
+    
+    Returns:
+        The corresponding LLVM IR type
+    
+    Raises:
+        NotImplementedError: If the ctype is not supported
+    """
     if ctype in mapping:
         return mapping[ctype]
     raise NotImplementedError(f"No mapping for {ctype}")
 
 
 def is_ctypes(ctype: str) -> bool:
+    """
+    Check if a given type name is a supported ctypes type.
+    
+    Args:
+        ctype: String name of the type to check
+    
+    Returns:
+        True if the type is a supported ctypes type, False otherwise
+    """
     return ctype in mapping

tests/failing_tests/assign/retype.py

@@ -1,39 +0,0 @@
-from pythonbpf import bpf, map, section, bpfglobal, compile
-from ctypes import c_void_p, c_int64, c_uint64
-from pythonbpf.maps import HashMap
-
-
-# NOTE: This example tries to reinterpret the variable `x` to a different type.
-# We do not allow this for now, as stack allocations are typed and have to be
-# done in the first basic block. Allowing re-interpretation would require
-# re-allocation of stack space (possibly in a new basic block), which is not
-# supported in eBPF yet.
-# We can allow bitcasts in cases where the width of the types is the same in
-# the future. But for now, we do not allow any re-interpretation of variables.
-
-@bpf
-@map
-def last() -> HashMap:
-    return HashMap(key=c_uint64, value=c_uint64, max_entries=3)
-
-
-@bpf
-@section("tracepoint/syscalls/sys_enter_execve")
-def hello_world(ctx: c_void_p) -> c_int64:
-    last.update(0, 1)
-    x = last.lookup(0)
-    x = 20
-    if x == 2:
-        print("Hello, World!")
-    else:
-        print("Goodbye, World!")
-    return
-
-
-@bpf
-@bpfglobal
-def LICENSE() -> str:
-    return "GPL"
-
-
-compile()

tests/passing_tests/assign/comprehensive.py

@@ -1,69 +0,0 @@
-from pythonbpf import bpf, map, section, bpfglobal, compile, struct
-from ctypes import c_void_p, c_int64, c_int32, c_uint64
-from pythonbpf.maps import HashMap
-from pythonbpf.helper import ktime
-
-
-# NOTE: This is a comprehensive test combining struct, helper, and map features
-# Please note that at line 50, though we have used an absurd expression to test
-# the compiler, it is recommended to use named variables to reduce the amount of
-# scratch space that needs to be allocated.
-
-@bpf
-@struct
-class data_t:
-    pid: c_uint64
-    ts: c_uint64
-
-
-@bpf
-@map
-def last() -> HashMap:
-    return HashMap(key=c_uint64, value=c_uint64, max_entries=3)
-
-
-@bpf
-@section("tracepoint/syscalls/sys_enter_execve")
-def hello_world(ctx: c_void_p) -> c_int64:
-    dat = data_t()
-    dat.pid = 123
-    dat.pid = dat.pid + 1
-    print(f"pid is {dat.pid}")
-    tu = 9
-    last.update(0, tu)
-    last.update(1, -last.lookup(0))
-    x = last.lookup(0)
-    print(f"Map value at index 0: {x}")
-    x = x + c_int32(1)
-    print(f"x after adding 32-bit 1 is {x}")
-    x = ktime() - 121
-    print(f"ktime - 121 is {x}")
-    x = last.lookup(0)
-    x = x + 1
-    print(f"x is {x}")
-    if x == 10:
-        jat = data_t()
-        jat.ts = 456
-        print(f"Hello, World!, ts is {jat.ts}")
-        a = last.lookup(0)
-        print(f"a is {a}")
-        last.update(9, 9)
-        last.update(0, last.lookup(last.lookup(0)) +
-                    last.lookup(last.lookup(0)) + last.lookup(last.lookup(0)))
-        z = last.lookup(0)
-        print(f"new map val at index 0 is {z}")
-    else:
-        a = last.lookup(0)
-        print("Goodbye, World!")
-    c = last.lookup(1 - 1)
-    print(f"c is {c}")
-    return
-
-
-@bpf
-@bpfglobal
-def LICENSE() -> str:
-    return "GPL"
-
-
-compile()

tests/passing_tests/assign/cst_var_binop.py

@@ -1,27 +0,0 @@
-from pythonbpf import bpf, section, bpfglobal, compile
-from ctypes import c_void_p, c_int64
-
-
-@bpf
-@section("tracepoint/syscalls/sys_enter_execve")
-def hello_world(ctx: c_void_p) -> c_int64:
-    x = 1
-    print(f"Initial x: {x}")
-    a = 20
-    x = a
-    print(f"Updated x with a: {x}")
-    x = (x + x) * 3
-    if x == 2:
-        print("Hello, World!")
-    else:
-        print(f"Goodbye, World! {x}")
-    return
-
-
-@bpf
-@bpfglobal
-def LICENSE() -> str:
-    return "GPL"
-
-
-compile()

tests/passing_tests/assign/helper.py

@@ -1,34 +0,0 @@
-from pythonbpf import bpf, map, section, bpfglobal, compile
-from ctypes import c_void_p, c_int64, c_uint64
-from pythonbpf.maps import HashMap
-
-# NOTE: An example of i64** assignment with binops on the RHS
-
-
-@bpf
-@map
-def last() -> HashMap:
-    return HashMap(key=c_uint64, value=c_uint64, max_entries=3)
-
-
-@bpf
-@section("tracepoint/syscalls/sys_enter_execve")
-def hello_world(ctx: c_void_p) -> c_int64:
-    last.update(0, 1)
-    x = last.lookup(0)
-    print(f"{x}")
-    x = x + 1
-    if x == 2:
-        print("Hello, World!")
-    else:
-        print("Goodbye, World!")
-    return
-
-
-@bpf
-@bpfglobal
-def LICENSE() -> str:
-    return "GPL"
-
-
-compile()

tests/passing_tests/assign/struct_and_helper_binops.py

@@ -1,40 +0,0 @@
-from pythonbpf import bpf, section, bpfglobal, compile, struct
-from ctypes import c_void_p, c_int64, c_uint64
-from pythonbpf.helper import ktime
-
-
-@bpf
-@struct
-class data_t:
-    pid: c_uint64
-    ts: c_uint64
-
-
-@bpf
-@section("tracepoint/syscalls/sys_enter_execve")
-def hello_world(ctx: c_void_p) -> c_int64:
-    dat = data_t()
-    dat.pid = 123
-    dat.pid = dat.pid + 1
-    print(f"pid is {dat.pid}")
-    x = ktime() - 121
-    print(f"ktime is {x}")
-    x = 1
-    x = x + 1
-    print(f"x is {x}")
-    if x == 2:
-        jat = data_t()
-        jat.ts = 456
-        print(f"Hello, World!, ts is {jat.ts}")
-    else:
-        print("Goodbye, World!")
-    return
-
-
-@bpf
-@bpfglobal
-def LICENSE() -> str:
-    return "GPL"
-
-
-compile()

tools/vmlinux-gen.py

@@ -243,6 +243,17 @@ class BTFConverter:
             data
         )
 
+        # below to replace those c_bool with bitfield greater than 8
+        def repl(m):
+            name, bits = m.groups()
+            return f"('{name}', ctypes.c_uint32, {bits})" if int(bits) > 8 else m.group(0)
+
+        data = re.sub(
+            r"\('([^']+)',\s*ctypes\.c_bool,\s*(\d+)\)",
+            repl,
+            data
+        )
+
         # Remove ctypes. prefix from invalid entries
         invalid_ctypes = ["bpf_iter_state", "_cache_type", "fs_context_purpose"]
         for name in invalid_ctypes: