Understanding Java byte code and the class file format
Download as PPTX, PDF19 likes4,214 views
The document explains the Java bytecode and class file format, detailing various aspects such as source code examples, bytecode instructions, stack slot utilization, and local variable arrays. It also discusses method invocation types and dynamic method handling, including exceptions and type safety during bytecode verification. Additionally, it touches on the use of libraries for dynamic bytecode manipulation and service class design patterns.
![source code byte code
package pkg;
class Bar {
void foo() {
return;
}
}
RETURN
0x0000 foo ()V
0 1
foo
()V
pkg/Bar
java/lang/Object
0x0000
pkg/Bar
java/lang/Object
[]
[]
0xCAFEBABE
0x0052
0000:
0001:
0002:
0003:
0x0000
0x0002
0x0003
0x0000
0x0000
0x0000 0x0000 0x0001
0xB1
0x0000 0x0001
UTF-8[foo]
UTF-8[()V]
UTF-8[pkg/Bar]
UTF-8[java/lang/Object]
0x0004](https://image.slidesharecdn.com/bytecode-140724051149-phpapp01/85/Understanding-Java-byte-code-and-the-class-file-format-21-320.jpg)
Understanding Java byte code and the class file format
- 1. Understanding Java byte code and the class file format
- 2. 0xCAFEBABE source code byte code JVM javac scalac groovyc jrubyc JIT compilerinterpreter class loader creates reads runs
- 3. source code byte code void foo() { } RETURNreturn; 0xB1
- 4. source code byte code int foo() { return 1 + 2; } ICONST_1 ICONST_2 IADD operand stack 1 2 13 IRETURN 0x04 0x05 0x60 0xAC
- 5. source code byte code ICONST_2 IADD IRETURN BIPUSH int foo() { return 11 + 2; } operand stack 11 2 1113 0x05 0x60 0xAC 0x10 110x0B
- 6. source code byte code int foo(int i) { return i + 1; } ILOAD_1 ICONST_1 IADD IRETURN operand stack local variable array 1 : i 0 : this i 1 ii + 1 0x1B 0x04 0x60 0xAC
- 7. source code byte code ILOAD_1 ICONST_1 IADD operand stack local variable array 1 : i 0 : this 1 i int foo(int i) { int i2 = i + 1; return i2; } ISTORE_2 ILOAD_2 IRETURN 2 : i + 1 ii + 1 2 :2 : i + 1 0x1B 0x04 0x60 0x3D 0x1C 0xAC
- 8. source code byte code long foo(long i) { return i + 1L; } LLOAD_1 LCONST_1 LADD LRETURN operand stack local variable array 2 : i (cn.) 1 : i 0 : this i (cn.) i 0x1F 0x0A 0x61 0xAD i (cn.) i 1L 1L (cn.) i + 1L i + 1L (cn.)
- 9. source code byte code short foo(short i) { return (short) (i + 1); } ILOAD_1 ICONST_1 IADD I2S IRETURN operand stack local variable array 1 : i 0 : this 1 iii + 1 0x1B 0x04 0x60 0x93 0xAC
- 10. Java type JVM type (non-array) JVM descriptor stack slots boolean I Z 1 byte B 1 short S 1 char C 1 int I 1 long L J 2 float F F 1 double D D 2 void - V 0 java.lang.Object A Ljava/lang/Object; 1
- 11. source code byte code pkg/Bar.class void foo() { return; } RETURN package pkg; class Bar { } constant pool (i.a. UTF-8) 0x0000 0x0000: foo 0x0001: ()V 0x0002: pkg/Bar 0xB1 0x0000 0x0000foo ()V()V0x0001 0x00000 1 10x0001 operand stack ///////////////// local variable array 0 : this
- 12. source code byte code package pkg; class Bar { public static void foo() { return; } } RETURN 0x0009 operand stack ///////////////// local variable array ///////////////// 0 0 Modifier Value static 0x0008 public 0x0001 0x0009 constant pool 0xB1 0x0000 0x0000 0x0000 0x0001foo ()V
- 13. source code byte code package pkg; class Bar { int foo(int i) { return foo(i + 1); } } ILOAD_1 ICONST_1 IADD INVOKEVIRTUAL pkg/Bar foo (I)I ALOAD_0 IRETURN operand stack local variable array 1 : i 0 : this this i 1 this ii + 1 foo(i + 1) 0x2A 0x1B 0x04 0x60 0xAC 0xB6 constant pool 0x0002 0x0000 0x0001
- 14. source code byte code package pkg; class Bar { static int foo(int i) { return foo(i + 1); } } ILOAD_0 ICONST_1 IADD INVOKESTATIC pkg/Bar foo (I)I IRETURN operand stack local variable array 0 : i i 1 ii + 1foo(i + 1) 0x1A 0x04 0x60 0xB8 0x0002 0x0001 0x0000 0xAC constant pool
- 15. source code byte code package pkg; class Bar { @Override public String toString() { return super.toString(); } } ALOAD_0 INVOKESPECIAL java/lang/Object toString ()Ljava/lang/String; ARETURN operand stack local variable array 0 : this thistoString() 0x1A 0xB7 0x0004 0x0005 0x0006 0xAC constant pool
- 16. INVOKESPECIAL INVOKEVIRTUAL INVOKESTATIC INVOKEINTERFACE INVOKEDYNAMIC Invokes a static method. pkg/Bar foo ()V pkg/Bar foo ()V Invokes the most-specific version of an inherited method on a non-interface class. pkg/Bar foo ()V Invokes a super class‘s version of an inherited method. Invokes a “constructor method”. Invokes a private method. Invokes an interface default method (Java 8). pkg/Bar foo ()V Invokes an interface method. (Similar to INVOKEVIRTUAL but without virtual method table index optimization.) foo ()V bootstrap Queries the given bootstrap method for locating a method implementation at runtime. (MethodHandle: Combines a specific method and an INVOKE* instruction.)
- 17. void foo() { ???(); // invokedynamic } foo() ? qux() bootstrap() bar() void foo() { bar(); } In theory, this could be achieved by using reflection. However, using invokedynamic offers a more native approach. This is especially important when dealing with primitive types (double boxing). Used to implement lambda expressions, more important for dynamic languages.
- 18. source code byte code 0x1A 0x06 0xA4 0x03 0xAC 0x1A 0x04 0x60 0xB8 0x0002 0x0000 0x0001 0xAC package pkg; class Bar { static int foo(int i) { if (i > 3) { return 0; } else { return foo(i + 1); } } } ILOAD_0 ICONST_3 IF_ICMPLE ICONST_0 IRETURN ILOAD_0 ICONST_1 IADD INVOKESTATIC IRETURN pkg/Bar foo (I)I X X: operand stack local variable array 0 : i i 3 i 1 i + 1foo(i + 1) constant pool 0x0008 0008:
- 19. java.lang.VerifyError: Inconsistent stackmap frames at branch target XXX in method pkg.Bar.foo(I)I at offset YYY type inferencing (until Java 5, failover for Java 6) Follow any possible path of jump instructions: Assert the consistency of the contents of the operand stack and local variable array for any possible path. Inference might require several iterations over a method‘s byte code. Legacy: -XX:-UseSplitVerifier type checking (from Java 6) The verifier rejects malformed class files and byte code: • compliance to class file format • type safety • access violations Require the otherwise infered information about the contents of the operand stack and the local variable array to be embedded in the code as stack map frames at any target of a jump instruction. Checking requires exactly one iteration over a method‘s byte code.
- 20. source code byte code int foo() { try { return 1; } catch (Exception e) { return 0; } } ICONST_1 IRETURN ICONST_0 IRETURN exception table: 0000: 0001: 0002: 0003: <from> <to> <catch> java/lang/Exception0x0000 0x0001 0x0002 constant pool 0x0005 0x04 0xAC 0x03 0xAC
- 21. source code byte code package pkg; class Bar { void foo() { return; } } RETURN 0x0000 foo ()V 0 1 foo ()V pkg/Bar java/lang/Object 0x0000 pkg/Bar java/lang/Object [] [] 0xCAFEBABE 0x0052 0000: 0001: 0002: 0003: 0x0000 0x0002 0x0003 0x0000 0x0000 0x0000 0x0000 0x0001 0xB1 0x0000 0x0001 UTF-8[foo] UTF-8[()V] UTF-8[pkg/Bar] UTF-8[java/lang/Object] 0x0004
- 22. class Service { @Secured(user = "ADMIN") void deleteEverything() { if(!"ADMIN".equals(UserHolder.user)) { throw new IllegalStateException("Wrong user"); } // delete everything... } } redefine class (build time, agent) create subclass (Liskov substitution) class SecuredService extends Service { @Override void deleteEverything() { if(!"ADMIN".equals(UserHolder.user)) { throw new IllegalStateException("Wrong user"); } super.deleteEverything(); } } class Service { @Secured(user = "ADMIN") void deleteEverything() { // delete everything... } }
- 23. Class<?> dynamicType = new ByteBuddy() .subclass(Object.class) .method(named("toString")) .intercept(value("Hello World!")) .make() .load(getClass().getClassLoader(), ClassLoadingStrategy.Default.WRAPPER) .getLoaded(); assertThat(dynamicType.newInstance().toString(), is("Hello World!"));