RFR: 8312132: Add tracking of multiple address spaces in NMT [v89]

[Johan Sjölen]

On Fri, 17 May 2024 15:27:33 GMT, Gerard Ziemski <gziemski at openjdk.org> wrote:

>The absolute worst case scenario is O(n) and the best is O(log(n)) and since we use random numbers to balance the tree it should be closer to the best case, correct?

Right. In fact, it shouldn't be even close to O(n).

 >We cannot guarantee that it will not be triggered at some point in the future, and if it does, how useful it will be for someone to investigate? I would consider printing out a warning instead of using assert here.

To be clear: This is in a verification function only called in two specific tests meant to challenge the treap (in some sense). If we do have a large deviation then that is probably worth investigating, it's an indication that something is going on. If we have intermittent failures because we chose a very tight bound, then we might have to increase that bound slightly, or we investigate how good our PRNG is.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/18289#discussion_r1605212325