Bindgen immediate types with builtin

Merged: https://github.com/ruby/ruby/pull/6418

lib/mjit/c_type.rb

@@ -27,7 +27,7 @@ module RubyVM::MJIT
     end
 
     module Immediate
-      # @param fiddle_type [Integer] Fiddle::TYPE_*
+      # @param fiddle_type [Integer]
       def self.new(fiddle_type)
         name = Fiddle.constants.find do |const|
           const.start_with?('TYPE_') && Fiddle.const_get(const) == fiddle_type.abs
@@ -40,6 +40,20 @@ module RubyVM::MJIT
           CPointer::Immediate.define(fiddle_type)
         end
       end
+
+      # @param type [String]
+      def self.parse(ctype)
+        new(Fiddle::Importer.parse_ctype(ctype))
+      end
+
+      def self.find(size, unsigned)
+        fiddle_type = TYPE_MAP.fetch(size)
+        fiddle_type = -fiddle_type if unsigned
+        new(fiddle_type)
+      end
+
+      TYPE_MAP = Fiddle::PackInfo::SIZE_MAP.map { |type, size| [size, type.abs] }.to_h
+      private_constant :TYPE_MAP
     end
 
     module Bool

mjit_c.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 # Part of this file is generated by tool/mjit/bindgen.rb.
 # Run `make mjit-bindgen` to update code between "MJIT bindgen begin" and "MJIT bindgen end".
 module RubyVM::MJIT
@@ -175,14 +176,10 @@ module RubyVM::MJIT
     @RB_BUILTIN ||= self.rb_builtin_function
   end
 
-  def C.VALUE
-    @VALUE ||= CType::Immediate.new(-5)
-  end
-
   def C.compile_branch
     @compile_branch ||= CType::Struct.new(
       "compile_branch", 8,
-      stack_size: [0, CType::Immediate.new(-4)],
+      stack_size: [0, CType::Immediate.parse("unsigned int")],
       finish_p: [32, self._Bool],
     )
   end
@@ -191,16 +188,16 @@ module RubyVM::MJIT
     @compile_status ||= CType::Struct.new(
       "compile_status", 120,
       success: [0, self._Bool],
-      stack_size_for_pos: [64, CType::Pointer.new { CType::Immediate.new(4) }],
+      stack_size_for_pos: [64, CType::Pointer.new { CType::Immediate.parse("int") }],
       local_stack_p: [128, self._Bool],
       is_entries: [192, CType::Pointer.new { self.iseq_inline_storage_entry }],
-      cc_entries_index: [256, CType::Immediate.new(4)],
+      cc_entries_index: [256, CType::Immediate.parse("int")],
       compiled_iseq: [320, CType::Pointer.new { self.rb_iseq_constant_body }],
-      compiled_id: [384, CType::Immediate.new(4)],
+      compiled_id: [384, CType::Immediate.parse("int")],
       compile_info: [448, CType::Pointer.new { self.rb_mjit_compile_info }],
       merge_ivar_guards_p: [512, self._Bool],
       ivar_serial: [576, self.rb_serial_t],
-      max_ivar_index: [640, CType::Immediate.new(-5)],
+      max_ivar_index: [640, CType::Immediate.parse("size_t")],
       inlined_iseqs: [704, CType::Pointer.new { CType::Pointer.new { self.rb_iseq_constant_body } }],
       inline_context: [768, self.inlined_call_context],
     )
@@ -209,10 +206,10 @@ module RubyVM::MJIT
   def C.inlined_call_context
     @inlined_call_context ||= CType::Struct.new(
       "inlined_call_context", 24,
-      orig_argc: [0, CType::Immediate.new(4)],
+      orig_argc: [0, CType::Immediate.parse("int")],
       me: [64, self.VALUE],
-      param_size: [128, CType::Immediate.new(4)],
+      param_size: [128, CType::Immediate.parse("int")],
-      local_size: [160, CType::Immediate.new(4)],
+      local_size: [160, CType::Immediate.parse("int")],
     )
   end
 
@@ -262,11 +259,11 @@ module RubyVM::MJIT
       save_temps: [8, self._Bool],
       warnings: [16, self._Bool],
       debug: [24, self._Bool],
-      debug_flags: [64, CType::Pointer.new { CType::Immediate.new(2) }],
+      debug_flags: [64, CType::Pointer.new { CType::Immediate.parse("char") }],
       wait: [128, self._Bool],
-      min_calls: [160, CType::Immediate.new(-4)],
+      min_calls: [160, CType::Immediate.parse("unsigned int")],
-      verbose: [192, CType::Immediate.new(4)],
+      verbose: [192, CType::Immediate.parse("int")],
-      max_cache_size: [224, CType::Immediate.new(4)],
+      max_cache_size: [224, CType::Immediate.parse("int")],
       pause: [256, self._Bool],
       custom: [264, self._Bool],
     )
@@ -275,11 +272,11 @@ module RubyVM::MJIT
   def C.rb_builtin_function
     @rb_builtin_function ||= CType::Struct.new(
       "rb_builtin_function", 32,
-      func_ptr: [0, CType::Pointer.new { CType::Immediate.new(0) }],
+      func_ptr: [0, CType::Pointer.new { CType::Immediate.parse("void") }],
-      argc: [64, CType::Immediate.new(4)],
+      argc: [64, CType::Immediate.parse("int")],
-      index: [96, CType::Immediate.new(4)],
+      index: [96, CType::Immediate.parse("int")],
-      name: [128, CType::Pointer.new { CType::Immediate.new(2) }],
+      name: [128, CType::Pointer.new { CType::Immediate.parse("char") }],
-      compiler: [192, CType::Immediate.new(1)],
+      compiler: [192, CType::Immediate.parse("void *")],
     )
   end
 
@@ -311,7 +308,7 @@ module RubyVM::MJIT
       call_: [192, self.vm_call_handler],
       aux_: [256, CType::Union.new(
         "", 8,
-        attr_index: CType::Immediate.new(-4),
+        attr_index: CType::Immediate.parse("unsigned int"),
         method_missing_reason: self.method_missing_reason,
         v: self.VALUE,
       )],
@@ -337,9 +334,9 @@ module RubyVM::MJIT
       iseq: [128, CType::Pointer.new { self.rb_iseq_t }],
       self: [192, self.VALUE],
       ep: [256, CType::Pointer.new { self.VALUE }],
-      block_code: [320, CType::Pointer.new { CType::Immediate.new(0) }],
+      block_code: [320, CType::Pointer.new { CType::Immediate.parse("void") }],
       __bp__: [384, CType::Pointer.new { self.VALUE }],
-      jit_return: [448, CType::Pointer.new { CType::Immediate.new(0) }],
+      jit_return: [448, CType::Pointer.new { CType::Immediate.parse("void") }],
     )
   end
 
@@ -358,7 +355,7 @@ module RubyVM::MJIT
     @rb_execution_context_struct ||= CType::Struct.new(
       "rb_execution_context_struct", 1,
       vm_stack: [-1, CType::Pointer.new { self.VALUE }],
-      vm_stack_size: [-1, CType::Immediate.new(-5)],
+      vm_stack_size: [-1, CType::Immediate.parse("size_t")],
       cfp: [-1, CType::Pointer.new { self.rb_control_frame_t }],
       tag: [-1, CType::Pointer.new { self.rb_vm_tag }],
       interrupt_flag: [-1, self.rb_atomic_t],
@@ -374,15 +371,15 @@ module RubyVM::MJIT
       trace_arg: [-1, CType::Pointer.new { self.rb_trace_arg_struct }],
       errinfo: [-1, self.VALUE],
       passed_block_handler: [-1, self.VALUE],
-      raised_flag: [-1, CType::Immediate.new(-2)],
+      raised_flag: [-1, CType::Immediate.parse("uint8_t")],
       method_missing_reason: [-1, self.method_missing_reason],
       private_const_reference: [-1, self.VALUE],
       machine: [-1, CType::Struct.new(
         "", 1,
         stack_start: [-1, CType::Pointer.new { self.VALUE }],
         stack_end: [-1, CType::Pointer.new { self.VALUE }],
-        stack_maxsize: [-1, CType::Immediate.new(-5)],
+        stack_maxsize: [-1, CType::Immediate.parse("size_t")],
-        regs: [-1, CType::Immediate.new(4)],
+        regs: [-1, CType::Immediate.parse("int")],
       )],
     )
   end
@@ -395,7 +392,7 @@ module RubyVM::MJIT
     @rb_iseq_constant_body ||= CType::Struct.new(
       "rb_iseq_constant_body", 336,
       type: [0, self.rb_iseq_type],
-      iseq_size: [32, CType::Immediate.new(-4)],
+      iseq_size: [32, CType::Immediate.parse("unsigned int")],
       iseq_encoded: [64, CType::Pointer.new { self.VALUE }],
       param: [128, CType::Struct.new(
         "", 48,
@@ -412,13 +409,13 @@ module RubyVM::MJIT
           accepts_no_kwarg: [8, CType::BitField.new(1, 0)],
           ruby2_keywords: [9, CType::BitField.new(1, 1)],
         )],
-        size: [32, CType::Immediate.new(-4)],
+        size: [32, CType::Immediate.parse("unsigned int")],
-        lead_num: [64, CType::Immediate.new(4)],
+        lead_num: [64, CType::Immediate.parse("int")],
-        opt_num: [96, CType::Immediate.new(4)],
+        opt_num: [96, CType::Immediate.parse("int")],
-        rest_start: [128, CType::Immediate.new(4)],
+        rest_start: [128, CType::Immediate.parse("int")],
-        post_start: [160, CType::Immediate.new(4)],
+        post_start: [160, CType::Immediate.parse("int")],
-        post_num: [192, CType::Immediate.new(4)],
+        post_num: [192, CType::Immediate.parse("int")],
-        block_start: [224, CType::Immediate.new(4)],
+        block_start: [224, CType::Immediate.parse("int")],
         opt_table: [256, CType::Pointer.new { self.VALUE }],
         keyword: [320, CType::Pointer.new { self.rb_iseq_param_keyword }],
       )],
@@ -438,13 +435,13 @@ module RubyVM::MJIT
         pc2branchindex: [192, self.VALUE],
         original_iseq: [256, CType::Pointer.new { self.VALUE }],
       )],
-      local_table_size: [1920, CType::Immediate.new(-4)],
+      local_table_size: [1920, CType::Immediate.parse("unsigned int")],
-      ic_size: [1952, CType::Immediate.new(-4)],
+      ic_size: [1952, CType::Immediate.parse("unsigned int")],
-      ise_size: [1984, CType::Immediate.new(-4)],
+      ise_size: [1984, CType::Immediate.parse("unsigned int")],
-      ivc_size: [2016, CType::Immediate.new(-4)],
+      ivc_size: [2016, CType::Immediate.parse("unsigned int")],
-      icvarc_size: [2048, CType::Immediate.new(-4)],
+      icvarc_size: [2048, CType::Immediate.parse("unsigned int")],
-      ci_size: [2080, CType::Immediate.new(-4)],
+      ci_size: [2080, CType::Immediate.parse("unsigned int")],
-      stack_max: [2112, CType::Immediate.new(-4)],
+      stack_max: [2112, CType::Immediate.parse("unsigned int")],
       mark_bits: [2176, CType::Union.new(
         "", 8,
         list: CType::Pointer.new { self.iseq_bits_t },
@@ -454,10 +451,10 @@ module RubyVM::MJIT
       builtin_inline_p: [2248, self._Bool],
       outer_variables: [2304, CType::Pointer.new { self.rb_id_table }],
       mandatory_only_iseq: [2368, CType::Pointer.new { self.rb_iseq_t }],
-      jit_func: [2432, CType::Immediate.new(1)],
+      jit_func: [2432, CType::Immediate.parse("void *")],
-      total_calls: [2496, CType::Immediate.new(-5)],
+      total_calls: [2496, CType::Immediate.parse("unsigned long")],
       jit_unit: [2560, CType::Pointer.new { self.rb_mjit_unit }],
-      yjit_payload: [2624, CType::Pointer.new { CType::Immediate.new(0) }],
+      yjit_payload: [2624, CType::Pointer.new { CType::Immediate.parse("void") }],
     )
   end
 
@@ -468,7 +465,7 @@ module RubyVM::MJIT
       base_label: [64, self.VALUE, true],
       label: [128, self.VALUE, true],
       first_lineno: [192, self.VALUE, true],
-      node_id: [256, CType::Immediate.new(4)],
+      node_id: [256, CType::Immediate.parse("int")],
       code_location: [288, self.rb_code_location_t],
     )
   end
@@ -485,7 +482,7 @@ module RubyVM::MJIT
         loader: CType::Struct.new(
           "", 16,
           obj: [0, self.VALUE],
-          index: [64, CType::Immediate.new(4)],
+          index: [64, CType::Immediate.parse("int")],
         ),
         exec: CType::Struct.new(
           "", 16,
@@ -503,7 +500,7 @@ module RubyVM::MJIT
   def C.rb_iv_index_tbl_entry
     @rb_iv_index_tbl_entry ||= CType::Struct.new(
       "rb_iv_index_tbl_entry", 24,
-      index: [0, CType::Immediate.new(-4)],
+      index: [0, CType::Immediate.parse("uint32_t")],
       class_serial: [64, self.rb_serial_t],
       class_value: [128, self.VALUE],
     )
@@ -528,7 +525,7 @@ module RubyVM::MJIT
         optimized: self.rb_method_optimized_t,
       )],
       original_id: [256, self.ID],
-      method_serial: [320, CType::Immediate.new(-5)],
+      method_serial: [320, CType::Immediate.parse("uintptr_t")],
     )
   end
 
@@ -541,7 +538,7 @@ module RubyVM::MJIT
   end
 
   def C.rb_method_type_t
-    @rb_method_type_t ||= CType::Immediate.new(4)
+    @rb_method_type_t ||= CType::Immediate.parse("int")
   end
 
   def C.rb_mjit_compile_info
@@ -559,19 +556,23 @@ module RubyVM::MJIT
     @rb_mjit_unit ||= CType::Struct.new(
       "rb_mjit_unit", 64,
       unode: [0, self.ccan_list_node],
-      id: [128, CType::Immediate.new(4)],
+      id: [128, CType::Immediate.parse("int")],
-      handle: [192, CType::Pointer.new { CType::Immediate.new(0) }],
+      handle: [192, CType::Pointer.new { CType::Immediate.parse("void") }],
       iseq: [256, CType::Pointer.new { self.rb_iseq_t }],
       used_code_p: [320, self._Bool],
       compact_p: [328, self._Bool],
       compile_info: [336, self.rb_mjit_compile_info],
       cc_entries: [384, CType::Pointer.new { CType::Pointer.new { self.rb_callcache } }],
-      cc_entries_size: [448, CType::Immediate.new(-4)],
+      cc_entries_size: [448, CType::Immediate.parse("unsigned int")],
     )
   end
 
   def C.rb_serial_t
-    @rb_serial_t ||= CType::Immediate.new(-6)
+    @rb_serial_t ||= CType::Immediate.parse("unsigned long long")
+  end
+
+  def C.VALUE
+    @VALUE ||= CType::Immediate.find(Primitive.cexpr!("SIZEOF(VALUE)"), Primitive.cexpr!("SIGNED_TYPE_P(VALUE)"))
   end
 
   def C._Bool

mjit_compiler.c

@@ -164,6 +164,9 @@ has_cache_for_send(rb_execution_context_t *ec, VALUE self, VALUE cc_addr, VALUE
 
 extern bool rb_splat_or_kwargs_p(const struct rb_callinfo *restrict ci);
 
+#define SIZEOF(type) RB_SIZE2NUM(sizeof(type))
+#define SIGNED_TYPE_P(type) RBOOL((type)(-1) < (type)(0))
+
 #include "mjit_c.rbinc"
 #include "mjit_compiler.rbinc"
 

tool/mjit/bindgen.rb

@@ -114,14 +114,16 @@ class BindingGenerator
   # @param src_path [String]
   # @param uses [Array<String>]
   # @param ints [Array<String>]
-  # @param types [Array<String>] Imported types
+  # @param types [Array<String>]
+  # @param dynamic_types [Array<String>] #ifdef-dependent immediate types, which need Primitive.cexpr! for type detection
   # @param ruby_fields [Hash{ Symbol => Array<String> }] Struct VALUE fields that are considered Ruby objects
-  def initialize(src_path:, uses:, ints:, types:, ruby_fields:)
+  def initialize(src_path:, uses:, ints:, types:, dynamic_types:, ruby_fields:)
     @preamble, @postamble = split_ambles(src_path)
     @src = String.new
     @uses = uses.sort
     @ints = ints.sort
     @types = types.sort
+    @dynamic_types = dynamic_types.sort
     @ruby_fields = ruby_fields.transform_keys(&:to_s)
     @references = Set.new
   end
@@ -159,8 +161,19 @@ class BindingGenerator
       println
     end
 
+    # Define dynamic types
+    @dynamic_types.each do |type|
+      unless generate_node(nodes_index[type])&.start_with?('CType::Immediate')
+        raise "Non-immediate type is given to dynamic_types: #{type}"
+      end
+      println "  def C.#{type}"
+      println "    @#{type} ||= CType::Immediate.find(Primitive.cexpr!(\"SIZEOF(#{type})\"), Primitive.cexpr!(\"SIGNED_TYPE_P(#{type})\"))"
+      println "  end"
+      println
+    end
+
     # Leave a stub for types that are referenced but not targeted
-    (@references - @types).each do |type|
+    (@references - @types - @dynamic_types).each do |type|
       println "  def C.#{type}"
       println "    #{DEFAULTS[type]}"
       println "  end"
@@ -257,7 +270,12 @@ class BindingGenerator
         push_target(type)
         "self.#{type}"
       else
-        "CType::Immediate.new(#{ctype})"
+        # Convert any function pointers to void* to workaround FILE* vs int*
+        if ctype == Fiddle::TYPE_VOIDP
+          "CType::Immediate.parse(\"void *\")"
+        else
+          "CType::Immediate.parse(#{type.dump})"
+        end
       end
     end
   end
@@ -317,7 +335,6 @@ generator = BindingGenerator.new(
     IC
     IVC
     RB_BUILTIN
-    VALUE
     compile_branch
     compile_status
     inlined_call_context
@@ -347,6 +364,9 @@ generator = BindingGenerator.new(
     rb_mjit_unit
     rb_serial_t
   ],
+  dynamic_types: %w[
+    VALUE
+  ],
   ruby_fields: {
     rb_iseq_location_struct: %w[
       base_label