RFR: 8347459: C2: missing transformation for chain of shifts/multiplications by constants [v6]

[Emanuel Peter]

On Fri, 7 Mar 2025 13:49:05 GMT, Marc Chevalier <duke at openjdk.org> wrote:

>> test/hotspot/jtreg/compiler/c2/irTests/LShiftLNodeIdealizationTests.java line 222:
>> 
>>> 220:     public long testDoubleShift9(long x) {
>>> 221:         return (x << 62L) << 3L;
>>> 222:     }
>> 
>> I see that you have quite a few examples here with fixed constants. It would be good to extend this with random constants. You can do that with `static final` fields, as they are constant by the time we JIT compile.
>> 
>> 
>> private static final int CON0 = RANDOM.nextInt();
>> private static final int CON1 = RANDOM.nextInt();
>> 
>>     @Test
>>     public long test(int x) {
>>         return (x << CON0) << CON1;
>>     }
>> 
>> I would give you "bonus points" if you use `Generators.java`, because that produces more interesting constants.
>
> I can also do that, but I think using the constants is more valuable than only random numbers: most of random int or long won't have some corner properties I try to cover here. On top of that, with the comments, they help enumerating clearly different cases to look at. So, I'd be happy to add some randomization, but I strongly feel I should keep the constant ones as well.

Absolutely, please keep the cases you already have, and add some randomized cases on top of it.

> most of random int or long won't have some corner properties I try to cover here

Right, that's why it is better to use `Generators.java` rather than just `Random`. With `Generators`, we make sure to generate more special values, such as powers of 2. And powers of 2 multiplication can for example be converted to shift, and that seems to be quite relevant here ;)

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/23728#discussion_r1985097087