1berry/openjdk
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

8349766: GenShen: Bad progress after degen does not always need full gc

Browse Source 
Reviewed-by: wkemper
This commit is contained in:
Kelvin Nilsen 2025-02-27 18:40:20 +00:00
parent 2fd7156110
commit 3ae80bfb60
2 changed files with 20 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
src/hotspot/share/gc/shenandoah/shenandoahDegeneratedGC.cpp
View File

@ -49,7 +49,8 @@ ShenandoahDegenGC::ShenandoahDegenGC(ShenandoahDegenPoint degen_point, Shenandoa
ShenandoahGC(),
_degen_point(degen_point),
_generation(generation),
_abbreviated(false) {
_abbreviated(false),
_consecutive_degen_with_bad_progress(0) {
}
bool ShenandoahDegenGC::collect(GCCause::Cause cause) {
@ -305,9 +306,24 @@ void ShenandoahDegenGC::op_degenerated() {
metrics.snap_after();
// Check for futility and fail. There is no reason to do several back-to-back Degenerated cycles,
// because that probably means the heap is overloaded and/or fragmented.
// The most common scenario for lack of good progress following a degenerated GC is an accumulation of floating
// garbage during the most recently aborted concurrent GC effort. With generational GC, it is far more effective to
// reclaim this floating garbage with another degenerated cycle (which focuses on young generation and might require
// a pause of 200 ms) rather than a full GC cycle (which may require over 2 seconds with a 10 GB old generation).
//
// In generational mode, we'll only upgrade to full GC if we've done two degen cycles in a row and both indicated
// bad progress. In non-generational mode, we'll preserve the original behavior, which is to upgrade to full
// immediately following a degenerated cycle with bad progress. This preserves original behavior of non-generational
// Shenandoah so as to avoid introducing "surprising new behavior." It also makes less sense with non-generational
// Shenandoah to replace a full GC with a degenerated GC, because both have similar pause times in non-generational
// mode.
if (!metrics.is_good_progress(_generation)) {
_consecutive_degen_with_bad_progress++;
} else {
_consecutive_degen_with_bad_progress = 0;
}
if (!heap->mode()->is_generational() ||
((heap->shenandoah_policy()->consecutive_degenerated_gc_count() > 1) && (_consecutive_degen_with_bad_progress >= 2))) {
heap->cancel_gc(GCCause::_shenandoah_upgrade_to_full_gc);
op_degenerated_futile();
} else {

1
src/hotspot/share/gc/shenandoah/shenandoahDegeneratedGC.hpp
View File

@ -36,6 +36,7 @@ private:
const ShenandoahDegenPoint _degen_point;
ShenandoahGeneration* _generation;
bool _abbreviated;
size_t _consecutive_degen_with_bad_progress;
public:
ShenandoahDegenGC(ShenandoahDegenPoint degen_point, ShenandoahGeneration* generation);