In cases where a full collection is triggered by an allocation failure, resizing the heap at the end of the GC should take into account the size of the pending allocation. Otherwise, we shrink the heap only to trigger an expansion to satisfy the allocation request that triggered the full gc. It is not a major issue, as the shrinking does not actually uncommit the regions, just marks these regions for uncommitting concurrently, so the expansion is just undoing this marking.

[0.621s][debug][gc,ergo ] GC(21) Running G1 Clear Bitmap with 2 workers for 2 work units.
[0.622s][debug][gc,ergo,heap] GC(21) Heap resize. Attempt heap shrinking (capacity higher than max desired capacity). Capacity: 134217728B occupancy: 14942210B live: 1744800B maximum_desired_capacity: 49807366B (70 %) Required for allocation 12845058B
[0.622s][debug][gc,ergo,heap] GC(21) Heap resize. Requested shrink amount: 84410362B aligned shrink amount: 83886080B
[0.622s][debug][gc,ergo,heap] GC(21) Heap resize. Requested shrinking amount: 80M actual shrinking amount: 80M (80 regions)
[0.623s][info ][gc ] GC(21) Pause Full (G1 Compaction Pause) 71M->1M(48M) 36.999ms
[0.623s][debug][gc,ergo,heap] Attempt heap expansion (allocation request failed). Allocation request: 102760464B
[0.623s][debug][gc,ergo,heap] Heap resize. Requested expansion amount: 98M aligned expansion amount: 99M (99 regions)