Merge pull request #68 from pythonbpf/request-struct

Support enough machinery to make request struct work

Author: varun-r-mallya

pythonbpf/allocation_pass.py

@@ -118,6 +118,18 @@ def _allocate_for_call(
             local_sym_tab[var_name] = LocalSymbol(var, struct_info.ir_type, call_type)
             logger.info(f"Pre-allocated {var_name} for struct {call_type}")
 
+        elif VmlinuxHandlerRegistry.is_vmlinux_struct(call_type):
+            # When calling struct_name(pointer), we're doing a cast, not construction
+            # So we allocate as a pointer (i64) not as the actual struct
+            var = builder.alloca(ir.IntType(64), name=var_name)
+            var.align = 8
+            local_sym_tab[var_name] = LocalSymbol(
+                var, ir.IntType(64), VmlinuxHandlerRegistry.get_struct_type(call_type)
+            )
+            logger.info(
+                f"Pre-allocated {var_name} for vmlinux struct pointer cast to {call_type}"
+            )
+
         else:
             logger.warning(f"Unknown call type for allocation: {call_type}")
 
@@ -325,13 +337,6 @@ def _allocate_for_attribute(builder, var_name, rval, local_sym_tab, structs_sym_
                 VmlinuxHandlerRegistry.get_field_type(vmlinux_struct_name, field_name)
             )
             field_ir, field = field_type
-            # TODO: For now, we only support integer type allocations.
-            # This always assumes first argument of function to be the context struct
-            base_ptr = builder.function.args[0]
-            local_sym_tab[
-                struct_var
-            ].var = base_ptr  # This is repurposing of var to store the pointer of the base type
-            local_sym_tab[struct_var].ir_type = field_ir
 
             # Determine the actual IR type based on the field's type
             actual_ir_type = None
@@ -386,12 +391,14 @@ def _allocate_for_attribute(builder, var_name, rval, local_sym_tab, structs_sym_
                 )
                 actual_ir_type = ir.IntType(64)
 
-            # Allocate with the actual IR type, not the GlobalVariable
+            # Allocate with the actual IR type
             var = _allocate_with_type(builder, var_name, actual_ir_type)
-            local_sym_tab[var_name] = LocalSymbol(var, actual_ir_type, field)
+            local_sym_tab[var_name] = LocalSymbol(
+                var, actual_ir_type, field
+            )  # <-- Store Field metadata
 
             logger.info(
-                f"Pre-allocated {var_name} from vmlinux struct {vmlinux_struct_name}.{field_name}"
+                f"Pre-allocated {var_name} as {actual_ir_type} from vmlinux struct {vmlinux_struct_name}.{field_name}"
             )
             return
         else:

pythonbpf/assign_pass.py

@@ -1,5 +1,7 @@
 import ast
 import logging
+from inspect import isclass
+
 from llvmlite import ir
 from pythonbpf.expr import eval_expr
 from pythonbpf.helper import emit_probe_read_kernel_str_call
@@ -148,8 +150,30 @@ def handle_variable_assignment(
         return False
 
     val, val_type = val_result
-    logger.info(f"Evaluated value for {var_name}: {val} of type {val_type}, {var_type}")
+    logger.info(
+        f"Evaluated value for {var_name}: {val} of type {val_type}, expected {var_type}"
+    )
+
     if val_type != var_type:
+        # Handle vmlinux struct pointers - they're represented as Python classes but are i64 pointers
+        if isclass(val_type) and (val_type.__module__ == "vmlinux"):
+            logger.info("Handling vmlinux struct pointer assignment")
+            # vmlinux struct pointers: val is a pointer, need to convert to i64
+            if isinstance(var_type, ir.IntType) and var_type.width == 64:
+                # Convert pointer to i64 using ptrtoint
+                if isinstance(val.type, ir.PointerType):
+                    val = builder.ptrtoint(val, ir.IntType(64))
+                    logger.info(
+                        "Converted vmlinux struct pointer to i64 using ptrtoint"
+                    )
+                builder.store(val, var_ptr)
+                logger.info(f"Assigned vmlinux struct pointer to {var_name} (i64)")
+                return True
+            else:
+                logger.error(
+                    f"Type mismatch: vmlinux struct pointer requires i64, got {var_type}"
+                )
+                return False
         if isinstance(val_type, Field):
             logger.info("Handling assignment to struct field")
             # Special handling for struct_xdp_md i32 fields that are zero-extended to i64

pythonbpf/expr/expr_pass.py

@@ -12,8 +12,8 @@
     get_base_type_and_depth,
     deref_to_depth,
 )
-from pythonbpf.vmlinux_parser.assignment_info import Field
 from .vmlinux_registry import VmlinuxHandlerRegistry
+from ..vmlinux_parser.dependency_node import Field
 
 logger: Logger = logging.getLogger(__name__)
 
@@ -173,8 +173,16 @@ def _handle_attribute_expr(
                     return vmlinux_result
                 else:
                     raise RuntimeError("Vmlinux struct did not process successfully")
-            metadata = structs_sym_tab[var_metadata]
-            if attr_name in metadata.fields:
+
+            elif isinstance(var_metadata, Field):
+                logger.error(
+                    f"Cannot access field '{attr_name}' on already-loaded field value '{var_name}'"
+                )
+                return None
+
+            # Regular user-defined struct
+            metadata = structs_sym_tab.get(var_metadata)
+            if metadata and attr_name in metadata.fields:
                 gep = metadata.gep(builder, var_ptr, attr_name)
                 val = builder.load(gep)
                 field_type = metadata.field_type(attr_name)
@@ -609,6 +617,66 @@ def _handle_boolean_op(
         return None
 
 
+# ============================================================================
+# VMLinux casting
+# ============================================================================
+
+
+def _handle_vmlinux_cast(
+    func,
+    module,
+    builder,
+    expr,
+    local_sym_tab,
+    map_sym_tab,
+    structs_sym_tab=None,
+):
+    # handle expressions such as struct_request(ctx.di) where struct_request is a vmlinux
+    # struct and ctx.di is a pointer to a struct but is actually represented as a c_uint64
+    # which needs to be cast to a pointer. This is also a field of another vmlinux struct
+    """Handle vmlinux struct cast expressions like struct_request(ctx.di)."""
+    if len(expr.args) != 1:
+        logger.info("vmlinux struct cast takes exactly one argument")
+        return None
+
+    # Get the struct name
+    struct_name = expr.func.id
+
+    # Evaluate the argument (e.g., ctx.di which is a c_uint64)
+    arg_result = eval_expr(
+        func,
+        module,
+        builder,
+        expr.args[0],
+        local_sym_tab,
+        map_sym_tab,
+        structs_sym_tab,
+    )
+
+    if arg_result is None:
+        logger.info("Failed to evaluate argument to vmlinux struct cast")
+        return None
+
+    arg_val, arg_type = arg_result
+    # Get the vmlinux struct type
+    vmlinux_struct_type = VmlinuxHandlerRegistry.get_struct_type(struct_name)
+    if vmlinux_struct_type is None:
+        logger.error(f"Failed to get vmlinux struct type for {struct_name}")
+        return None
+    # Cast the integer/value to a pointer to the struct
+    # If arg_val is an integer type, we need to inttoptr it
+    ptr_type = ir.PointerType()
+    # TODO: add a integer check here later
+    if ctypes_to_ir(arg_type.type.__name__):
+        # Cast integer to pointer
+        casted_ptr = builder.inttoptr(arg_val, ptr_type)
+    else:
+        logger.error(f"Unsupported type for vmlinux cast: {arg_type}")
+        return None
+
+    return casted_ptr, vmlinux_struct_type
+
+
 # ============================================================================
 # Expression Dispatcher
 # ============================================================================
@@ -629,6 +697,18 @@ def eval_expr(
     elif isinstance(expr, ast.Constant):
         return _handle_constant_expr(module, builder, expr)
     elif isinstance(expr, ast.Call):
+        if isinstance(expr.func, ast.Name) and VmlinuxHandlerRegistry.is_vmlinux_struct(
+            expr.func.id
+        ):
+            return _handle_vmlinux_cast(
+                func,
+                module,
+                builder,
+                expr,
+                local_sym_tab,
+                map_sym_tab,
+                structs_sym_tab,
+            )
         if isinstance(expr.func, ast.Name) and expr.func.id == "deref":
             return _handle_deref_call(expr, local_sym_tab, builder)
 

pythonbpf/helper/__init__.py

@@ -1,6 +1,10 @@
 from .helper_registry import HelperHandlerRegistry
 from .helper_utils import reset_scratch_pool
-from .bpf_helper_handler import handle_helper_call, emit_probe_read_kernel_str_call
+from .bpf_helper_handler import (
+    handle_helper_call,
+    emit_probe_read_kernel_str_call,
+    emit_probe_read_kernel_call,
+)
 from .helpers import (
     ktime,
     pid,
@@ -74,6 +78,7 @@ def helper_call_handler(
     "reset_scratch_pool",
     "handle_helper_call",
     "emit_probe_read_kernel_str_call",
+    "emit_probe_read_kernel_call",
     "ktime",
     "pid",
     "deref",

pythonbpf/helper/bpf_helper_handler.py

@@ -34,6 +34,7 @@ class BPFHelperID(Enum):
     BPF_PERF_EVENT_OUTPUT = 25
     BPF_GET_STACK = 67
     BPF_PROBE_READ_KERNEL_STR = 115
+    BPF_PROBE_READ_KERNEL = 113
     BPF_RINGBUF_OUTPUT = 130
     BPF_RINGBUF_RESERVE = 131
     BPF_RINGBUF_SUBMIT = 132
@@ -574,6 +575,75 @@ def bpf_probe_read_kernel_str_emitter(
     return result, ir.IntType(64)
 
 
+def emit_probe_read_kernel_call(builder, dst_ptr, dst_size, src_ptr):
+    """Emit LLVM IR call to bpf_probe_read_kernel"""
+
+    fn_type = ir.FunctionType(
+        ir.IntType(64),
+        [ir.PointerType(), ir.IntType(32), ir.PointerType()],
+        var_arg=False,
+    )
+    fn_ptr = builder.inttoptr(
+        ir.Constant(ir.IntType(64), BPFHelperID.BPF_PROBE_READ_KERNEL.value),
+        ir.PointerType(fn_type),
+    )
+
+    result = builder.call(
+        fn_ptr,
+        [
+            builder.bitcast(dst_ptr, ir.PointerType()),
+            ir.Constant(ir.IntType(32), dst_size),
+            builder.bitcast(src_ptr, ir.PointerType()),
+        ],
+        tail=False,
+    )
+
+    logger.info(f"Emitted bpf_probe_read_kernel (size={dst_size})")
+    return result
+
+
+@HelperHandlerRegistry.register(
+    "probe_read_kernel",
+    param_types=[
+        ir.PointerType(ir.IntType(8)),
+        ir.PointerType(ir.IntType(8)),
+    ],
+    return_type=ir.IntType(64),
+)
+def bpf_probe_read_kernel_emitter(
+    call,
+    map_ptr,
+    module,
+    builder,
+    func,
+    local_sym_tab=None,
+    struct_sym_tab=None,
+    map_sym_tab=None,
+):
+    """Emit LLVM IR for bpf_probe_read_kernel helper."""
+
+    if len(call.args) != 2:
+        raise ValueError(
+            f"probe_read_kernel expects 2 args (dst, src), got {len(call.args)}"
+        )
+
+    # Get destination buffer (char array -> i8*)
+    dst_ptr, dst_size = get_or_create_ptr_from_arg(
+        func, module, call.args[0], builder, local_sym_tab, map_sym_tab, struct_sym_tab
+    )
+
+    # Get source pointer (evaluate expression)
+    src_ptr, src_type = get_ptr_from_arg(
+        call.args[1], func, module, builder, local_sym_tab, map_sym_tab, struct_sym_tab
+    )
+
+    # Emit the helper call
+    result = emit_probe_read_kernel_call(builder, dst_ptr, dst_size, src_ptr)
+
+    logger.info(f"Emitted bpf_probe_read_kernel (size={dst_size})")
+    return result, ir.IntType(64)
+
+
 @HelperHandlerRegistry.register(
     "random",
     param_types=[],