Create the following source file:

import java.util.function.IntSupplier;
import java.util.stream.IntStream;

class Test {
private int compute() {
IntSupplier s1 = () -> IntStream.range(0, 10000).map(v -> 1).sum();
IntSupplier s2 = () -> IntStream.range(0, 10000).map(v -> 1).sum();
IntSupplier s3 = () -> IntStream.range(0, 10000).map(v -> 1).sum();
return s1.getAsInt() + s2.getAsInt() + s3.getAsInt();
}
}

Compile it with using openjdk 17.0.1 2021-10-19, without debug info:

$ javac Test.java

Then observe that lambdas were deduplicated using javap -p:

$ javap -p Test.class
Compiled from "Test.java"
class Test {
  Test();
  private int compute();
  private static int lambda$compute$1();
  private static int lambda$compute$0(int);
}

There are only two lambda bodies, despite we have six in the source code. This is expected behavior (see JDK-8200301).

Do the same with java 18 or later (java 22 is also affected). Now, the lambdas are not deduplicated:

$ javap -p Test.class
Compiled from "Test.java"
class Test {
  Test();
  private int compute();
  private static int lambda$compute$5();
  private static int lambda$compute$4(int);
  private static int lambda$compute$3();
  private static int lambda$compute$2(int);
  private static int lambda$compute$1();
  private static int lambda$compute$0(int);
}

This is an unexpected regression.