ChatGPT解决这个技术问题 Extra ChatGPT

Change private static final field using Java reflection

I have a class with a private static final field that, unfortunately, I need to change it at run-time.

Using reflection I get this error: java.lang.IllegalAccessException: Can not set static final boolean field

Is there any way to change the value?

Field hack = WarpTransform2D.class.getDeclaredField("USE_HACK");
hack.setAccessible(true);
hack.set(null, true);
Such a bad idea. I'd try to get the source and recompile (or even decompile/recompile) instead.
System.out is a public static final field, but it can be changed too.
@irreputable System.out/in/err are so "special" that the Java Memory Model has to make special mention of them. They are not examples which should be followed.
well my point ws to find a hack in the between to have my app working until the lib responsible make the change at the next release so i don't need to hack anymore...
@Bill K from ten years ago: It would be GREAT to recompile it but it's on a deployed system and I just need to patch it until we can update the deployed app!

C
Community

Assuming no SecurityManager is preventing you from doing this, you can use setAccessible to get around private and resetting the modifier to get rid of final, and actually modify a private static final field.

Here's an example:

import java.lang.reflect.*;

public class EverythingIsTrue {
   static void setFinalStatic(Field field, Object newValue) throws Exception {
      field.setAccessible(true);

      Field modifiersField = Field.class.getDeclaredField("modifiers");
      modifiersField.setAccessible(true);
      modifiersField.setInt(field, field.getModifiers() & ~Modifier.FINAL);

      field.set(null, newValue);
   }
   public static void main(String args[]) throws Exception {      
      setFinalStatic(Boolean.class.getField("FALSE"), true);

      System.out.format("Everything is %s", false); // "Everything is true"
   }
}

Assuming no SecurityException is thrown, the above code prints "Everything is true".

What's actually done here is as follows:

The primitive boolean values true and false in main are autoboxed to reference type Boolean "constants" Boolean.TRUE and Boolean.FALSE

Reflection is used to change the public static final Boolean.FALSE to refer to the Boolean referred to by Boolean.TRUE

As a result, subsequently whenever a false is autoboxed to Boolean.FALSE, it refers to the same Boolean as the one refered to by Boolean.TRUE

Everything that was "false" now is "true"

Related questions

Using reflection to change static final File.separatorChar for unit testing

How to limit setAccessible to only “legitimate” uses? Has examples of messing with Integer's cache, mutating a String, etc

Has examples of messing with Integer's cache, mutating a String, etc

Caveats

Extreme care should be taken whenever you do something like this. It may not work because a SecurityManager may be present, but even if it doesn't, depending on usage pattern, it may or may not work.

JLS 17.5.3 Subsequent Modification of Final Fields In some cases, such as deserialization, the system will need to change the final fields of an object after construction. final fields can be changed via reflection and other implementation dependent means. The only pattern in which this has reasonable semantics is one in which an object is constructed and then the final fields of the object are updated. The object should not be made visible to other threads, nor should the final fields be read, until all updates to the final fields of the object are complete. Freezes of a final field occur both at the end of the constructor in which the final field is set, and immediately after each modification of a final field via reflection or other special mechanism. Even then, there are a number of complications. If a final field is initialized to a compile-time constant in the field declaration, changes to the final field may not be observed, since uses of that final field are replaced at compile time with the compile-time constant. Another problem is that the specification allows aggressive optimization of final fields. Within a thread, it is permissible to reorder reads of a final field with those modifications of a final field that do not take place in the constructor.

See also

JLS 15.28 Constant Expression It's unlikely that this technique works with a primitive private static final boolean, because it's inlineable as a compile-time constant and thus the "new" value may not be observable

It's unlikely that this technique works with a primitive private static final boolean, because it's inlineable as a compile-time constant and thus the "new" value may not be observable

Appendix: On the bitwise manipulation

Essentially,

field.getModifiers() & ~Modifier.FINAL

turns off the bit corresponding to Modifier.FINAL from field.getModifiers(). & is the bitwise-and, and ~ is the bitwise-complement.

See also

Wikipedia/Bitwise operation

Remember Constant Expressions

Still not being able to solve this?, have fallen onto depression like I did for it? Does your code looks like this?

public class A {
    private final String myVar = "Some Value";
}

Reading the comments on this answer, specially the one by @Pshemo, it reminded me that Constant Expressions are handled different so it will be impossible to modify it. Hence you will need to change your code to look like this:

public class A {
    private final String myVar;

    private A() {
        myVar = "Some Value";
    }
}

if you are not the owner of the class... I feel you!

For more details about why this behavior read this?


@thecoop, @HalfBrian: there is no doubt that this is EXTREMELY EVIL, but this example was chosen by design. My answer only shows how, in some circumstances, this is possible. The most disgusting example that I can think of is deliberate chosen with the hope that perhaps people would be instantly disgusted by instead of falling in love with the technique.
Yo, dawg. I heard you like reflection, so I reflected on field so you can reflect while you reflect.
Note that Boolean.FALSE is not private.Does this really work with "private final static" members?
@mgaert it does, but you have to use getDeclaredField() instead of getField() for target class
+1. For those who will try to change something like final String myConstant = "x"; and will fail: remember that compile time constants will be inlined by compiler so when you will write code like System.out.println(myConstant); it will be compiled as System.out.println("x"); because compiler knows value of constant at compile time. To get rid of this problem you need to initialize your constants at runtime like final String myConstant = new String("x");. Also in case of primitives like final int myField = 11 use final int myField = new Integer(11); or final Integer myField = 11;
e
erickson

If the value assigned to a static final boolean field is known at compile-time, it is a constant. Fields of primitive or String type can be compile-time constants. A constant will be inlined in any code that references the field. Since the field is not actually read at runtime, changing it then will have no effect.

The Java language specification says this:

If a field is a constant variable (§4.12.4), then deleting the keyword final or changing its value will not break compatibility with pre-existing binaries by causing them not to run, but they will not see any new value for the usage of the field unless they are recompiled. This is true even if the usage itself is not a compile-time constant expression (§15.28)

Here's an example:

class Flag {
  static final boolean FLAG = true;
}

class Checker {
  public static void main(String... argv) {
    System.out.println(Flag.FLAG);
  }
}

If you decompile Checker, you'll see that instead of referencing Flag.FLAG, the code simply pushes a value of 1 (true) onto the stack (instruction #3).

0:   getstatic       #2; //Field java/lang/System.out:Ljava/io/PrintStream;
3:   iconst_1
4:   invokevirtual   #3; //Method java/io/PrintStream.println:(Z)V
7:   return

That was my first thought, but then I remembered Java compiled at runtime, if you were to reset the bit it would simply recompile with it as a variable instead of a constant.
@Bill K - No, this doesn't refer to JIT compilation. The dependent class files will actually contain the inlined values, and no reference to the independent class. It's a pretty simple experiment to test; I will add an example.
How does this jive with @polygenelubricants 's answer where he redefines Boolean.false?--but you are correct, I've seen this behavior when things didn't recompile correctly.
@Bill K - in polygenlubricants' answer, the field is not a compile time constant. It's public static final Boolean FALSE = new Boolean(false) not public static final boolean FALSE = false
S
Stephan Markwalder

A little curiosity from the Java Language Specification, chapter 17, section 17.5.4 "Write-protected Fields":

Normally, a field that is final and static may not be modified. However, System.in, System.out, and System.err are static final fields that, for legacy reasons, must be allowed to be changed by the methods System.setIn, System.setOut, and System.setErr. We refer to these fields as being write-protected to distinguish them from ordinary final fields.

Source: http://docs.oracle.com/javase/specs/jls/se7/html/jls-17.html#jls-17.5.4


A
Aleksandr Dubinsky

I also integrated it with joor library

Just use

      Reflect.on(yourObject).set("finalFieldName", finalFieldValue);

Also I fixed an issue with override which the previous solutions seem to miss. However use this very carefully, only when there's no other good solution.


When I try this (JDK12), I get an exception: "Can not set final ___ field".
@AaronIba It's no longer allowed in Java 12+.
n
nndru

Along with top ranked answer you may use a bit simpliest approach. Apache commons FieldUtils class already has particular method that can do the stuff. Please, take a look at FieldUtils.removeFinalModifier method. You should specify target field instance and accessibility forcing flag (if you play with non-public fields). More info you can find here.


This is a much simpler solution than the currently accepted answer
Is it? Copying one method sounds like a simpler solution than importing an entire library (which is doing the same thing as the method you'd be copying).
Does not work in Java 12+: java.lang.UnsupportedOperationException: In java 12+ final cannot be removed.
S
Some Name

Even in spite of being final a field can be modified outside of static initializer and (at least JVM HotSpot) will execute the bytecode perfectly fine.

The problem is that Java compiler does not allow this, but this can be easily bypassed using objectweb.asm. Here is p̶e̶r̶f̶e̶c̶t̶l̶y̶ ̶v̶a̶l̶i̶d̶ ̶c̶l̶a̶s̶s̶f̶i̶l̶e̶ an invalid classfile from the JVMS specification standpoint, but it passes bytecode verification and then is successfully loaded and initialized under JVM HotSpot OpenJDK12:

ClassWriter cw = new ClassWriter(0);
cw.visit(Opcodes.V1_8, Opcodes.ACC_PUBLIC, "Cl", null, "java/lang/Object", null);
{
    FieldVisitor fv = cw.visitField(Opcodes.ACC_PRIVATE | Opcodes.ACC_STATIC | Opcodes.ACC_FINAL, "fld", "I", null, null);
    fv.visitEnd();
}
{
    // public void setFinalField1() { //... }
    MethodVisitor mv = cw.visitMethod(Opcodes.ACC_PUBLIC | Opcodes.ACC_STATIC, "setFinalField1", "()V", null, null);
    mv.visitMaxs(2, 1);
    mv.visitInsn(Opcodes.ICONST_5);
    mv.visitFieldInsn(Opcodes.PUTSTATIC, "Cl", "fld", "I");
    mv.visitInsn(Opcodes.RETURN);
    mv.visitEnd();
}
{
    // public void setFinalField2() { //... }
    MethodVisitor mv = cw.visitMethod(Opcodes.ACC_PUBLIC | Opcodes.ACC_STATIC, "setFinalField2", "()V", null, null);
    mv.visitMaxs(2, 1);
    mv.visitInsn(Opcodes.ICONST_2);
    mv.visitFieldInsn(Opcodes.PUTSTATIC, "Cl", "fld", "I");
    mv.visitInsn(Opcodes.RETURN);
    mv.visitEnd();
}
cw.visitEnd();

In Java, the class looks roughly speaking as follows:

public class Cl{
    private static final int fld;

    public static void setFinalField1(){
        fld = 5;
    }

    public static void setFinalField2(){
        fld = 2;
    }
}

which cannot be compiled with javac, but can be loaded and executed by JVM.

JVM HotSpot has special treatment of such classes in the sense that it prevents such "constants" from participating in constant folding. This check is done on the bytecode rewriting phase of class initialization:

// Check if any final field of the class given as parameter is modified
// outside of initializer methods of the class. Fields that are modified
// are marked with a flag. For marked fields, the compilers do not perform
// constant folding (as the field can be changed after initialization).
//
// The check is performed after verification and only if verification has
// succeeded. Therefore, the class is guaranteed to be well-formed.
InstanceKlass* klass = method->method_holder();
u2 bc_index = Bytes::get_Java_u2(bcp + prefix_length + 1);
constantPoolHandle cp(method->constants());
Symbol* ref_class_name = cp->klass_name_at(cp->klass_ref_index_at(bc_index));
if (klass->name() == ref_class_name) {
   Symbol* field_name = cp->name_ref_at(bc_index);
   Symbol* field_sig = cp->signature_ref_at(bc_index);

   fieldDescriptor fd;
   if (klass->find_field(field_name, field_sig, &fd) != NULL) {
      if (fd.access_flags().is_final()) {
         if (fd.access_flags().is_static()) {
            if (!method->is_static_initializer()) {
               fd.set_has_initialized_final_update(true);
            }
          } else {
            if (!method->is_object_initializer()) {
              fd.set_has_initialized_final_update(true);
            }
          }
        }
      }
    }
}

The only restriction that JVM HotSpot checks is that the final field should not be modified outside of the class that the final field is declared at.


this is just pure EVIL and beautiful.
I don’t agree with the “perfectly valid classfile”. The JVMS §6.5 clearly says: “Otherwise, if the resolved field is final, it must be declared in the current class or interface, and the instruction must occur in the class or interface initialization method of the current class or interface. Otherwise, an IllegalAccessError is thrown”. So this is just another case where the implementation blatantly violates the specification and has code spread over multiple places to handle what should have been rejected
@Holger Thanks for the note. I made a correction according to your note to not confuse further readers.
V
VanagaS

In case of presence of a Security Manager, one can make use of AccessController.doPrivileged

Taking the same example from accepted answer above:

import java.lang.reflect.*;

public class EverythingIsTrue {
    static void setFinalStatic(Field field, Object newValue) throws Exception {
        field.setAccessible(true);
        Field modifiersField = Field.class.getDeclaredField("modifiers");

        // wrapping setAccessible 
        AccessController.doPrivileged(new PrivilegedAction() {
            @Override
            public Object run() {
                modifiersField.setAccessible(true);
                return null;
            }
        });

        modifiersField.setInt(field, field.getModifiers() & ~Modifier.FINAL);
        field.set(null, newValue);
    }

    public static void main(String args[]) throws Exception {      
      setFinalStatic(Boolean.class.getField("FALSE"), true);
      System.out.format("Everything is %s", false); // "Everything is true"
    }
}

In lambda expression, AccessController.doPrivileged, can be simplified to:

AccessController.doPrivileged((PrivilegedAction) () -> {
    modifiersField.setAccessible(true);
    return null;
});

yes @dan1st, you are right! Please check this for a solution: stackoverflow.com/a/56043252/2546381
B
Brice

With JDK 18 this won't be possible anymore due to the reimplementation of the core reflection over invokedynamic and MethodHandles as part of JEP-416 (PR).

Quote of Mandy Chung – who is the main author of this incredible work – in the following comment. Emphasis are mine.

If the underlying field is final, a Field object has write access if and only if setAccessible(true) has succeeded for this Field object; the field is non-static; and the field's declaring class is not a hidden class; and the field's declaring class is not a record class.


h
hasskell

Just saw that question on one of the interview question, if possible to change final variable with reflection or in runtime. Got really interested, so that what I became with:

 /**
 * @author Dmitrijs Lobanovskis
 * @since 03/03/2016.
 */
public class SomeClass {

    private final String str;

    SomeClass(){
        this.str = "This is the string that never changes!";
    }

    public String getStr() {
        return str;
    }

    @Override
    public String toString() {
        return "Class name: " + getClass() + " Value: " + getStr();
    }
}

Some simple class with final String variable. So in the main class import java.lang.reflect.Field;

/**
 * @author Dmitrijs Lobanovskis
 * @since 03/03/2016.
 */
public class Main {


    public static void main(String[] args) throws Exception{

        SomeClass someClass = new SomeClass();
        System.out.println(someClass);

        Field field = someClass.getClass().getDeclaredField("str");
        field.setAccessible(true);

        field.set(someClass, "There you are");

        System.out.println(someClass);
    }
}

The output will be as follows:

Class name: class SomeClass Value: This is the string that never changes!
Class name: class SomeClass Value: There you are

Process finished with exit code 0

According to documentation https://docs.oracle.com/javase/tutorial/reflect/member/fieldValues.html


Have you seen this post?
This question asks about a static final field, so this code doesn't work. setAccessible(true) only works for setting final instance fields.
T
Tomáš Záluský

The accepted answer worked for me until deployed on JDK 1.8u91. Then I realized it failed at field.set(null, newValue); line when I had read the value via reflection before calling of setFinalStatic method.

Probably the read caused somehow different setup of Java reflection internals (namely sun.reflect.UnsafeQualifiedStaticObjectFieldAccessorImpl in failing case instead of sun.reflect.UnsafeStaticObjectFieldAccessorImpl in success case) but I didn't elaborate it further.

Since I needed to temporarily set new value based on old value and later set old value back, I changed signature little bit to provide computation function externally and also return old value:

public static <T> T assignFinalField(Object object, Class<?> clazz, String fieldName, UnaryOperator<T> newValueFunction) {
    Field f = null, ff = null;
    try {
        f = clazz.getDeclaredField(fieldName);
        final int oldM = f.getModifiers();
        final int newM = oldM & ~Modifier.FINAL;
        ff = Field.class.getDeclaredField("modifiers");
        ff.setAccessible(true);
        ff.setInt(f,newM);
        f.setAccessible(true);

        T result = (T)f.get(object);
        T newValue = newValueFunction.apply(result);

        f.set(object,newValue);
        ff.setInt(f,oldM);

        return result;
    } ...

However for general case this would not be sufficient.


d
d4vidi

Many of the answers here are useful, but I've found none of them to work on Android, in particular. I'm even a pretty hefty user of Reflect by joor, and neither it nor apache's FieldUtils - both mentioned here in some of the answers, do the trick.

Problem with Android

The fundamental reason why this is so is because on Android there's no modifiers field in the Field class, which renders any suggestion involving this code (as in the marked answer), useless:

Field modifiersField = Field.class.getDeclaredField("modifiers");
modifiersField.setAccessible(true);
modifiersField.setInt(field, field.getModifiers() & ~Modifier.FINAL);

In fact, to quote from FieldUtils.removeFinalModifier():

// Do all JREs implement Field with a private ivar called "modifiers"?
final Field modifiersField = Field.class.getDeclaredField("modifiers");

So, the answer is no...

Solution

Pretty easy - instead of modifiers, the field name is accessFlags. This does the trick:

Field accessFlagsField = Field.class.getDeclaredField("accessFlags");
accessFlagsField.setAccessible(true);
accessFlagsField.setInt(field, field.getModifiers() & ~Modifier.FINAL);

Side-note #1: this can work regardless of whether the field is static in the class, or not.

Side-note #2: Seeing that the field itself could be private, it's recommended to also enable access over the field itself, using field.setAccessible(true) (in addition to accessFlagsField.setAccessible(true).


When running the original code in Android, you will get the following error: java.lang.NoSuchFieldException: No field modifiers in class Ljava/lang/reflect/Field; (declaration of 'java.lang.reflect.Field' appears in /apex/com.android.runtime/javalib/core-oj.jar). The suggested solution works in my case. (This error currently has only one google result, so hopefully people will now find this page)
p
pringi

Since Java 12 onwards, the answers given will not work.

Here is an example on how to modify a private static final field since Java 12 (based on this answer).

  private Object modifyField(Object newFieldValue, String fieldName, Object classInstance) throws NoSuchFieldException, IllegalAccessException {
    Field field = classInstance.getClass().getDeclaredField(fieldName);
    VarHandle MODIFIERS;

    field.setAccessible(true);

    var lookup = MethodHandles.privateLookupIn(Field.class, MethodHandles.lookup());
    MODIFIERS = lookup.findVarHandle(Field.class, "modifiers", int.class);
    int mods = field.getModifiers();

    if (Modifier.isFinal(mods)) {
      MODIFIERS.set(field, mods & ~Modifier.FINAL);
    }

    Object previousValue = field.get(classInstance);
    field.set(null, newFieldValue);

    return previousValue;
  }

See this thread for more details.


This does not work with Java 16+.
@JohannesKuhn works if you add --illegal-access=permit
--illegal-access=permit was removed in Java 17.
There is no way to make this work with Java 18 due to JEP 416
You can't change static final fields. That never really worked. It may appear to work, but in the end - relying on that effect makes your code just non-portable.
P
Philip Rego

If your field is simply private you can do this:

MyClass myClass= new MyClass();
Field aField= myClass.getClass().getDeclaredField("someField");
aField.setAccessible(true);
aField.set(myClass, "newValueForAString");

and throw/handle NoSuchFieldException


t
thecoop

The whole point of a final field is that it cannot be reassigned once set. The JVM uses this guarentee to maintain consistency in various places (eg inner classes referencing outer variables). So no. Being able to do so would break the JVM!

The solution is not to declare it final in the first place.


Moreover, final has a special role in multithreaded execution - changing final values would break the Java memory model too.
And a field not declared final should not be declared static.
@Tom: In general that's probably true, but I wouldn't outlaw all static mutable variables.
@Tom: Did you ever read up why singletons are evil? I did! Now I know that they evil only in Java. And only because of the availabilty of a user defined class loader. And ever since I know all this and I don´t use user defined class loader I use singletons without regret. And so does Scala where singleton are a first class language feature — That singletons are evil is is a well known false myth.
@Martin I know that your comment is old, and perhaps your views have changed by now, but I thought I'd just add this: Singletons are evil for reasons that have nothing to do with Java. They add hidden complexity to your code. Additionally, they can make it impossible to unit test without also knowing that n singletons must also be configured first. They are the antithesis of dependency injection. Your team might make the decision that the pitfalls of having hidden complexity don't outweigh the convenience of Singletons, but many teams take the opposite stance for good reason.