archive/jdk
1
0
Fork 0
mirror of https://github.com/openjdk/jdk.git synced 2025-09-21 03:24:38 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

6883834: ParNew: assert(!_g->to()->is_in_reserved(obj),"Scanning field twice?") with LargeObjects tests

Browse source 
Fixed process_chunk_boundaries(), used for parallel card scanning when using ParNew/CMS, so as to prevent double-scanning, or worse, non-scanning of imprecisely marked objects exceeding parallel chunk size. Made some sizing parameters for parallel card scanning diagnostic, disabled ParallelGCRetainPLAB, and elaborated and clarified some comments.

Reviewed-by: stefank, johnc
This commit is contained in:
Y. Srinivas Ramakrishna 2011-05-10 00:33:21 -07:00
parent 90ab9d500c
commit 8e56205189
8 changed files with 297 additions and 140 deletions
Download patch file Download diff file Expand all files Collapse all files

14
hotspot/src/share/vm/memory/space.cpp
View file

@ -97,6 +97,14 @@ void DirtyCardToOopClosure::walk_mem_region(MemRegion mr,
}
}
// We get called with "mr" representing the dirty region
// that we want to process. Because of imprecise marking,
// we may need to extend the incoming "mr" to the right,
// and scan more. However, because we may already have
// scanned some of that extended region, we may need to
// trim its right-end back some so we do not scan what
// we (or another worker thread) may already have scanned
// or planning to scan.
void DirtyCardToOopClosure::do_MemRegion(MemRegion mr) {
// Some collectors need to do special things whenever their dirty
@ -148,7 +156,7 @@ void DirtyCardToOopClosure::do_MemRegion(MemRegion mr) {
// e.g. the dirty card region is entirely in a now free object
// -- something that could happen with a concurrent sweeper.
bottom = MIN2(bottom, top);
mr = MemRegion(bottom, top);
MemRegion extended_mr = MemRegion(bottom, top);
assert(bottom <= top &&
(_precision != CardTableModRefBS::ObjHeadPreciseArray ||
_min_done == NULL ||
@ -156,8 +164,8 @@ void DirtyCardToOopClosure::do_MemRegion(MemRegion mr) {
"overlap!");
// Walk the region if it is not empty; otherwise there is nothing to do.
if (!mr.is_empty()) {
walk_mem_region(mr, bottom_obj, top);
if (!extended_mr.is_empty()) {
walk_mem_region(extended_mr, bottom_obj, top);
}
// An idempotent closure might be applied in any order, so we don't