Created attachment 145216 [details]
stacktrace

byte[]#10693 has 23M items. It is held by "JDI Target VM Interface" thread. Reassigning to debugger for further investigation.

#712424: -Xmx256m - that won't work at all. Make it at least 1024m
#725906: one mega long instance of StringReferenceImpl held in JPDAObjectWatchImpl.value - might be a duplicate of bug 240475

I've verified that this works in the current dev version.
Most likely it was fixed by issue #240475.

*** This bug has been marked as a duplicate of bug 240475 ***

FYI: The program I was testing is:

    /**
     * @param args the command line arguments
     */
    public static void main(String[] args) {
        bigStringWatch();
    }

    private static void bigStringWatch() {
        String bigString = createString(134000000);
        int l = bigString.length();
        bigString.hashCode();    // Line breakpoint here
    }

    private static String createString(int l) {
        char[] characters = new char[l];
        for (int i = 0; i < l; i++) {
            String is = Integer.toString(i, 36);
            characters[i] = is.charAt(is.length() - 1);
        }
        return new String(characters);
    }

with the line breakpoint at the comment.
When the breakpoint was hit, I've added this expression to Watches:
bigString.substring(1111, 120000000)
The IDE, after GC, was consuming less memory, that is the retained size of the String. Therefore it's fixed.