RFR: 8302760: Improve liveness/remembered set verification for G1 [v3]

[Kim Barrett]

On Wed, 22 Feb 2023 09:52:29 GMT, Thomas Schatzl <tschatzl at openjdk.org> wrote:

>> Hi all,
>> 
>>   can I have reviews for this refactoring/cleanup of the liveness/remembered set verification code for G1?
>> 
>> Here is an overview of the changes
>>  - previously liveness and remembered set verification could be done separately. Thismade the code a bit more complicated than necessary; now that remembered set verification always follows liveness verification, the code can assume that and do some minor cleanup. I kept the remembered verification a separate closure (`VerifyRemSetClosure`) because the additional boiler plate code is negligible compared to the other duplication that would have been needed around failure count tracking.
>>  - factored out liveness and remembered set verification from `HeapRegion::verify`, and heavily cleaned up the resulting `HeapRegion::verify_liveness_and_remembered_set`. This includes some removal of code duplication wrt to the failure count tracking.
>> - factored out a special `G1VerificationClosure::is_oop_safe` method that adds more checking of the `Klass`; I hesitated to put this into `Universe::is_oop` as this is much more extensive checking (and I wanted implicit error reporting :))
>> - CollectedHeap::is_oop() should use `klass_raw` because otherwise it will immediately assert on oops with garbage Klass values (the new code will just return `false` as expected). 
>> - some special `oopDesc::print_name_on()` which does not crash on bad oops, similar to existing related debugging helpers.
>> 
>> Thanks,
>>   Thomas
>
> Thomas Schatzl has updated the pull request incrementally with one additional commit since the last revision:
> 
>   kbarrett review

Looks good.

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

Marked as reviewed by kbarrett (Reviewer).

PR: https://git.openjdk.org/jdk/pull/12660