Merge pull request #55 from pythonbpf/vmlinux-ir-gen

remove bitfield support and add assignment support

Author: varun-r-mallya

pythonbpf/tbaa_gen/__init__.py

@@ -71,6 +71,9 @@ def process_vmlinux_post_ast(
                     if len(field_elem) == 2:
                         field_name, field_type = field_elem
                     elif len(field_elem) == 3:
+                        raise NotImplementedError(
+                            "Bitfields are not supported in the current version"
+                        )
                         field_name, field_type, bitfield_size = field_elem
                     field_table[field_name] = [field_type, bitfield_size]
             elif hasattr(class_obj, "__annotations__"):
@@ -144,7 +147,12 @@ def process_vmlinux_post_ast(
                             process_vmlinux_post_ast(
                                 containing_type, llvm_handler, handler, processing_stack
                             )
-                            new_dep_node.set_field_ready(elem_name, True)
+                            size_of_containing_type = (
+                                handler[containing_type.__name__]
+                            ).__sizeof__()
+                            new_dep_node.set_field_ready(
+                                elem_name, True, size_of_containing_type
+                            )
                         elif containing_type.__module__ == ctypes.__name__:
                             logger.debug(f"Processing ctype internal{containing_type}")
                             new_dep_node.set_field_ready(elem_name, True)
@@ -161,7 +169,12 @@ def process_vmlinux_post_ast(
                         process_vmlinux_post_ast(
                             elem_type, llvm_handler, handler, processing_stack
                         )
-                        new_dep_node.set_field_ready(elem_name, True)
+                        size_of_containing_type = (
+                            handler[elem_type.__name__]
+                        ).__sizeof__()
+                        new_dep_node.set_field_ready(
+                            elem_name, True, size_of_containing_type
+                        )
                 else:
                     raise ValueError(
                         f"{elem_name} with type {elem_type} from module {module_name} not supported in recursive resolver"

pythonbpf/vmlinux_parser/class_handler.py

@@ -147,3 +147,23 @@ def __len__(self) -> int:
             int: The number of nodes
         """
         return len(self._nodes)
+
+    def __getitem__(self, name: str) -> DependencyNode:
+        """
+        Get a node by name using dictionary-style access.
+
+        Args:
+            name: The name of the node to retrieve
+
+        Returns:
+            DependencyNode: The node with the given name
+
+        Raises:
+            KeyError: If no node with the given name exists
+
+        Example:
+            node = handler["some-dep_node_name"]
+        """
+        if name not in self._nodes:
+            raise KeyError(f"No node with name '{name}' found")
+        return self._nodes[name]

pythonbpf/vmlinux_parser/dependency_handler.py

@@ -1,5 +1,6 @@
 from dataclasses import dataclass, field
 from typing import Dict, Any, Optional
+import ctypes
 
 
 # TODO: FIX THE FUCKING TYPE NAME CONVENTION.
@@ -13,6 +14,7 @@ class Field:
     containing_type: Optional[Any]
     type_size: Optional[int]
     bitfield_size: Optional[int]
+    offset: int
     value: Any = None
     ready: bool = False
 
@@ -60,6 +62,10 @@ def set_bitfield_size(self, bitfield_size: Any, mark_ready: bool = False) -> Non
         if mark_ready:
             self.ready = True
 
+    def set_offset(self, offset: int) -> None:
+        """Set the offset of this field"""
+        self.offset = offset
+
 
 @dataclass
 class DependencyNode:
@@ -109,6 +115,7 @@ class DependencyNode:
     depends_on: Optional[list[str]] = None
     fields: Dict[str, Field] = field(default_factory=dict)
     _ready_cache: Optional[bool] = field(default=None, repr=False)
+    current_offset: int = 0
 
     def add_field(
         self,
@@ -120,6 +127,7 @@ def add_field(
         ctype_complex_type: Optional[int] = None,
         bitfield_size: Optional[int] = None,
         ready: bool = False,
+        offset: int = 0,
     ) -> None:
         """Add a field to the node with an optional initial value and readiness state."""
         if self.depends_on is None:
@@ -133,10 +141,14 @@ def add_field(
             type_size=type_size,
             ctype_complex_type=ctype_complex_type,
             bitfield_size=bitfield_size,
+            offset=offset,
         )
         # Invalidate readiness cache
         self._ready_cache = None
 
+    def __sizeof__(self):
+        return self.current_offset
+
     def get_field(self, name: str) -> Field:
         """Get a field by name."""
         return self.fields[name]
@@ -203,15 +215,78 @@ def set_field_bitfield_size(
         # Invalidate readiness cache
         self._ready_cache = None
 
-    def set_field_ready(self, name: str, is_ready: bool = False) -> None:
+    def set_field_ready(
+        self,
+        name: str,
+        is_ready: bool = False,
+        size_of_containing_type: Optional[int] = None,
+    ) -> None:
         """Mark a field as ready or not ready."""
         if name not in self.fields:
             raise KeyError(f"Field '{name}' does not exist in node '{self.name}'")
 
         self.fields[name].set_ready(is_ready)
+        self.fields[name].set_offset(self.current_offset)
+        self.current_offset += self._calculate_size(name, size_of_containing_type)
         # Invalidate readiness cache
         self._ready_cache = None
 
+    def _calculate_size(
+        self, name: str, size_of_containing_type: Optional[int] = None
+    ) -> int:
+        processing_field = self.fields[name]
+        # size_of_field will be in bytes
+        if processing_field.type.__module__ == ctypes.__name__:
+            size_of_field = ctypes.sizeof(processing_field.type)
+            return size_of_field
+        elif processing_field.type.__module__ == "vmlinux":
+            if processing_field.ctype_complex_type is not None:
+                if issubclass(processing_field.ctype_complex_type, ctypes.Array):
+                    if processing_field.containing_type.__module__ == ctypes.__name__:
+                        if (
+                            processing_field.containing_type is not None
+                            and processing_field.type_size is not None
+                        ):
+                            size_of_field = (
+                                ctypes.sizeof(processing_field.containing_type)
+                                * processing_field.type_size
+                            )
+                        else:
+                            raise RuntimeError(
+                                f"{processing_field} has no containing_type or type_size"
+                            )
+                        return size_of_field
+                    elif processing_field.containing_type.__module__ == "vmlinux":
+                        if (
+                            size_of_containing_type is not None
+                            and processing_field.type_size is not None
+                        ):
+                            size_of_field = (
+                                size_of_containing_type * processing_field.type_size
+                            )
+                        else:
+                            raise RuntimeError(
+                                f"{processing_field} has no containing_type or type_size"
+                            )
+                        return size_of_field
+                elif issubclass(processing_field.ctype_complex_type, ctypes._Pointer):
+                    return ctypes.sizeof(ctypes.c_void_p)
+                else:
+                    raise NotImplementedError(
+                        "This subclass of ctype not supported yet"
+                    )
+            else:
+                # search up pre-created stuff and get size
+                if size_of_containing_type is None:
+                    raise RuntimeError(
+                        f"Size of containing type {size_of_containing_type} is None"
+                    )
+                return size_of_containing_type
+
+        else:
+            raise ModuleNotFoundError("Module is not supported for the operation")
+        raise RuntimeError("control should not reach here")
+
     @property
     def is_ready(self) -> bool:
         """Check if the node is ready (all fields are ready)."""

pythonbpf/vmlinux_parser/dependency_node.py

@@ -1,11 +1,11 @@
 import ast
 import logging
-from typing import List, Tuple, Dict
+from typing import List, Tuple, Any
 import importlib
 import inspect
 
 from .dependency_handler import DependencyHandler
-from .ir_generation import IRGenerator
+from .ir_gen import IRGenerator
 from .class_handler import process_vmlinux_class
 
 logger = logging.getLogger(__name__)
@@ -82,7 +82,7 @@ def vmlinux_proc(tree: ast.AST, module):
     # initialise dependency handler
     handler = DependencyHandler()
     # initialise assignment dictionary of name to type
-    assignments: Dict[str, type] = {}
+    assignments: dict[str, tuple[type, Any]] = {}
 
     if not import_statements:
         logger.info("No vmlinux imports found")
@@ -129,7 +129,19 @@ def vmlinux_proc(tree: ast.AST, module):
                 )
 
     IRGenerator(module, handler)
-
-
-def process_vmlinux_assign(node, module, assignments: Dict[str, type]):
-    raise NotImplementedError("Assignment handling has not been implemented yet")
+    return assignments
+
+
+def process_vmlinux_assign(node, module, assignments: dict[str, tuple[type, Any]]):
+    # Check if this is a simple assignment with a constant value
+    if len(node.targets) == 1 and isinstance(node.targets[0], ast.Name):
+        target_name = node.targets[0].id
+        if isinstance(node.value, ast.Constant):
+            assignments[target_name] = (type(node.value.value), node.value.value)
+            logger.info(
+                f"Added assignment: {target_name} = {node.value.value!r} of type {type(node.value.value)}"
+            )
+        else:
+            raise ValueError(f"Unsupported assignment type for {target_name}")
+    else:
+        raise ValueError("Not a simple assignment")

pythonbpf/vmlinux_parser/import_detector.py

@@ -0,0 +1,3 @@
+from .ir_generation import IRGenerator
+
+__all__ = ["IRGenerator"]

pythonbpf/vmlinux_parser/ir_gen/__init__.py

@@ -0,0 +1,36 @@
+import logging
+from pythonbpf.vmlinux_parser.dependency_handler import DependencyHandler
+
+logger = logging.getLogger(__name__)
+
+
+class IRGenerator:
+    def __init__(self, module, handler: DependencyHandler):
+        self.module = module
+        self.handler: DependencyHandler = handler
+        self.generated: list[str] = []
+        if not handler.is_ready:
+            raise ImportError(
+                "Semantic analysis of vmlinux imports failed. Cannot generate IR"
+            )
+        for struct in handler:
+            self.struct_processor(struct)
+            print()
+
+    def struct_processor(self, struct):
+        if struct.name not in self.generated:
+            print(f"IR generating for {struct.name}")
+            print(f"Struct is {struct}")
+            for dependency in struct.depends_on:
+                if dependency not in self.generated:
+                    dep_node_from_dependency = self.handler[dependency]
+                    self.struct_processor(dep_node_from_dependency)
+                    self.generated.append(dependency)
+            # write actual processor logic here after assuming all dependencies are resolved
+            # this part cannot yet resolve circular dependencies. Gets stuck on an infinite loop during that.
+            self.generated.append(struct.name)
+
+    def struct_name_generator(
+        self,
+    ) -> None:
+        pass

pythonbpf/vmlinux_parser/ir_gen/ir_generation.py

@@ -1,9 +1,13 @@
 from pythonbpf import bpf, map, section, bpfglobal, compile_to_ir
 from pythonbpf.maps import HashMap
 from pythonbpf.helper import XDP_PASS
-from vmlinux import struct_xdp_md
+from vmlinux import TASK_COMM_LEN  # noqa: F401
 from vmlinux import struct_trace_event_raw_sys_enter  # noqa: F401
-from vmlinux import struct_ring_buffer_per_cpu  # noqa: F401
+
+# from vmlinux import struct_request
+from vmlinux import struct_xdp_md
+# from vmlinux import struct_trace_event_raw_sys_enter  # noqa: F401
+# from vmlinux import struct_ring_buffer_per_cpu  # noqa: F401
 
 from ctypes import c_int64