RJIT: Optimize String#empty?

3 files changed, 50 insertions, 1 deletions

diff --git a/lib/ruby_vm/rjit/insn_compiler.rb b/lib/ruby_vm/rjit/insn_compiler.rb
index 39d6f07e4c..aa37388a5b 100644
--- a/lib/ruby_vm/rjit/insn_compiler.rb
+++ b/lib/ruby_vm/rjit/insn_compiler.rb
@@ -2732,6 +2732,29 @@ module RubyVM::RJIT
     # @param jit [RubyVM::RJIT::JITState]
     # @param ctx [RubyVM::RJIT::Context]
     # @param asm [RubyVM::RJIT::Assembler]
+    def jit_rb_str_empty_p(jit, ctx, asm, argc, known_recv_class)
+      # Assume same offset to len embedded or not so we can use one code path to read the length
+      assert_equal(C.RString.offsetof(:as, :heap, :len), C.RString.offsetof(:as, :embed, :len))
+
+      recv_opnd = ctx.stack_pop(1)
+      out_opnd = ctx.stack_push
+
+      asm.comment('get string length')
+      asm.mov(:rax, recv_opnd)
+      str_len_opnd = [:rax, C.RString.offsetof(:as, :heap, :len)]
+
+      asm.cmp(str_len_opnd, 0)
+      asm.mov(:rax, Qfalse)
+      asm.mov(:rcx, Qtrue)
+      asm.cmove(:rax, :rcx)
+      asm.mov(out_opnd, :rax)
+
+      return true
+    end
+
+    # @param jit [RubyVM::RJIT::JITState]
+    # @param ctx [RubyVM::RJIT::Context]
+    # @param asm [RubyVM::RJIT::Assembler]
     def jit_rb_str_to_s(jit, ctx, asm, argc, known_recv_class)
       return false if argc != 0
       if known_recv_class == String
@@ -2826,7 +2849,7 @@ module RubyVM::RJIT
       register_cfunc_method(Integer, :===, :jit_rb_int_equal)
 
       # rb_str_to_s() methods in string.c
-      #register_cfunc_method(String, :empty?, :jit_rb_str_empty_p)
+      register_cfunc_method(String, :empty?, :jit_rb_str_empty_p)
       register_cfunc_method(String, :to_s, :jit_rb_str_to_s)
       register_cfunc_method(String, :to_str, :jit_rb_str_to_s)
       #register_cfunc_method(String, :bytesize, :jit_rb_str_bytesize)
diff --git a/rjit_c.rb b/rjit_c.rb
index f5467699ff..6d855d3ad8 100644
--- a/rjit_c.rb
+++ b/rjit_c.rb
@@ -748,6 +748,31 @@ module RubyVM::RJIT # :nodoc: all
     )
   end
 
+  def C.RString
+    @RString ||= CType::Struct.new(
+      "RString", Primitive.cexpr!("SIZEOF(struct RString)"),
+      basic: [self.RBasic, Primitive.cexpr!("OFFSETOF((*((struct RString *)NULL)), basic)")],
+      as: [CType::Union.new(
+        "", Primitive.cexpr!("SIZEOF(((struct RString *)NULL)->as)"),
+        heap: CType::Struct.new(
+          "", Primitive.cexpr!("SIZEOF(((struct RString *)NULL)->as.heap)"),
+          len: [CType::Immediate.parse("long"), Primitive.cexpr!("OFFSETOF(((struct RString *)NULL)->as.heap, len)")],
+          ptr: [CType::Pointer.new { CType::Immediate.parse("char") }, Primitive.cexpr!("OFFSETOF(((struct RString *)NULL)->as.heap, ptr)")],
+          aux: [CType::Union.new(
+            "", Primitive.cexpr!("SIZEOF(((struct RString *)NULL)->as.heap.aux)"),
+            capa: CType::Immediate.parse("long"),
+            shared: self.VALUE,
+          ), Primitive.cexpr!("OFFSETOF(((struct RString *)NULL)->as.heap, aux)")],
+        ),
+        embed: CType::Struct.new(
+          "", Primitive.cexpr!("SIZEOF(((struct RString *)NULL)->as.embed)"),
+          len: [CType::Immediate.parse("long"), Primitive.cexpr!("OFFSETOF(((struct RString *)NULL)->as.embed, len)")],
+          ary: [CType::Pointer.new { CType::Immediate.parse("char") }, Primitive.cexpr!("OFFSETOF(((struct RString *)NULL)->as.embed, ary)")],
+        ),
+      ), Primitive.cexpr!("OFFSETOF((*((struct RString *)NULL)), as)")],
+    )
+  end
+
   def C.RStruct
     @RStruct ||= CType::Struct.new(
       "RStruct", Primitive.cexpr!("SIZEOF(struct RStruct)"),
diff --git a/tool/rjit/bindgen.rb b/tool/rjit/bindgen.rb
index 8dd69109c5..3c25edad09 100755
--- a/tool/rjit/bindgen.rb
+++ b/tool/rjit/bindgen.rb
@@ -556,6 +556,7 @@ generator = BindingGenerator.new(
     RBasic
     RObject
     RStruct
+    RString
     attr_index_t
     iseq_inline_constant_cache
     iseq_inline_constant_cache_entry