java - Constructor bytecode

Question

The ASM guide talks about constructors:

package pkg;
public class Bean {
  private int f;
  public int getF() {
      return this.f;
  }
  public void setF(int f) {
      this.f = f;
  }
}

The Bean class also has a default public constructor which is generated by the compiler, since no explicit constructor was defined by the programmer. This default public constructor is generated as Bean() { super(); }. The bytecode of this constructor is the following:

ALOAD 0
INVOKESPECIAL java/lang/Object <init> ()V
RETURN

The first instruction pushes this on the operand stack. The second instruction pops this value from the stack, and calls the <init> method defined in the Object class. This corresponds to the super() call, i.e. a call to the constructor of the super class, Object. You can see here that constructors are named differently in compiled and source classes: in compiled classes they are always named <init>, while in source classes they have the name of the class in which they are defined. Finally the last instruction returns to the caller.

How is the value of this already known to the JVM before the first instruction of the constructor?

Answer 1

At the JVM level, first the object is allocated, uninitialized, then the constructor is invoked on that object. The constructor is more-or-less an instance method executed on the uninitialized object.

Even in the Java language, this exists and has all its fields at the first line of the constructor.