archive/jdk
1
0
Fork 0
mirror of https://github.com/openjdk/jdk.git synced 2025-09-22 12:04:39 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

6906727: UseCompressedOops: some card-marking fixes related to object arrays

Browse source 
Introduced a new write_ref_array(HeapWords* start, size_t count) method that does the requisite MemRegion range calculation so (some of the) clients of the erstwhile write_ref_array(MemRegion mr) do not need to worry. This removed all external uses of array_size(), which was also simplified and made private. Asserts were added to catch other possible issues. Further, less essential, fixes stemming from this investigation are deferred to CR 6904516 (to follow shortly in hs17).

Reviewed-by: kvn, coleenp, jmasa
This commit is contained in:
Y. Srinivas Ramakrishna 2009-12-03 15:01:57 -08:00
parent 5a239d996b
commit d875a8e0f1
8 changed files with 74 additions and 36 deletions
Download patch file Download diff file Expand all files Collapse all files

30
hotspot/src/share/vm/memory/barrierSet.inline.hpp
View file

@ -43,6 +43,8 @@ void BarrierSet::write_ref_field(void* field, oop new_val) {
}
void BarrierSet::write_ref_array(MemRegion mr) {
assert((HeapWord*)align_size_down((uintptr_t)mr.start(), HeapWordSize) == mr.start() , "Unaligned start");
assert((HeapWord*)align_size_up ((uintptr_t)mr.end(), HeapWordSize) == mr.end(), "Unaligned end" );
if (kind() == CardTableModRef) {
((CardTableModRefBS*)this)->inline_write_ref_array(mr);
} else {
@ -50,6 +52,34 @@ void BarrierSet::write_ref_array(MemRegion mr) {
}
}
// count is number of array elements being written
void BarrierSet::write_ref_array(HeapWord* start, size_t count) {
assert(count <= (size_t)max_intx, "count too large");
HeapWord* end = (HeapWord*)((char*)start + (count*heapOopSize));
// In the case of compressed oops, start and end may potentially be misaligned;
// so we need to conservatively align the first downward (this is not
// strictly necessary for current uses, but a case of good hygiene and,
// if you will, aesthetics) and the second upward (this is essential for
// current uses) to a HeapWord boundary, so we mark all cards overlapping
// this write. In the event that this evolves in the future to calling a
// logging barrier of narrow oop granularity, like the pre-barrier for G1
// (mentioned here merely by way of example), we will need to change this
// interface, much like the pre-barrier one above, so it is "exactly precise"
// (if i may be allowed the adverbial redundancy for emphasis) and does not
// include narrow oop slots not included in the original write interval.
HeapWord* aligned_start = (HeapWord*)align_size_down((uintptr_t)start, HeapWordSize);
HeapWord* aligned_end = (HeapWord*)align_size_up ((uintptr_t)end, HeapWordSize);
// If compressed oops were not being used, these should already be aligned
assert(UseCompressedOops || (aligned_start == start && aligned_end == end),
"Expected heap word alignment of start and end");
#if 0
warning("Post:\t" INTPTR_FORMAT "[" SIZE_FORMAT "] : [" INTPTR_FORMAT","INTPTR_FORMAT")\t",
start, count, aligned_start, aligned_end);
#endif
write_ref_array_work(MemRegion(aligned_start, aligned_end));
}
void BarrierSet::write_region(MemRegion mr) {
if (kind() == CardTableModRef) {
((CardTableModRefBS*)this)->inline_write_region(mr);