RFR: 8302215: G1: Last-ditch Full GC should do serial compaction for tail regions in per thread compaction points. [v2]

[Thomas Schatzl]

On Wed, 15 Feb 2023 11:27:12 GMT, Ivan Walulya <iwalulya at openjdk.org> wrote:

>> Hi all, 
>> 
>> Please review this change to call `G1FullCollector::phase2c_prepare_serial_compaction` when `do_maximal_compaction`. For maximum compaction, we re-prepare all objects in the "tail" regions of the per thread compaction points.
>> 
>> Testing: Tier 1-3.
>> 
>> Thanks
>
> Ivan Walulya has updated the pull request incrementally with one additional commit since the last revision:
> 
>   albert review

src/hotspot/share/gc/g1/g1FullGCCompactionPoint.cpp line 150:

> 148:   }
> 149: 
> 150:   int start_index = find_sorted(start);

Using regular iteration with an index, and potentially splitting it into two parts (one until `r->hrm_index() >= bottom`, and one afterwards) I think avoids the need for `find_sorted`. At least it seems this is the only user.

Something like:
[u]int index = 0;
while (_compaction_regions.at(i)->hrm_index() < bottom) {
  index++;
}
// If there isn't a straight copy operation in `GrowableArray`...
for (int copy_index = index; copy_index < _compaction_regions.length(); copy_index++) {
  cp->add(_compaction_regions.at(copy_index));
}

_compaction_regions.trunc_to(index);

src/hotspot/share/gc/g1/g1FullGCCompactionPoint.cpp line 151:

> 149: 
> 150:   int start_index = find_sorted(start);
> 151:   assert(start_index >= 0, "Should have atleast one region");

s/atleast/at least/

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

PR: https://git.openjdk.org/jdk/pull/12529