Mercurial > jdk8-mips64-public > hotspot / changeset

     1.1 --- a/src/share/vm/gc_implementation/g1/concurrentMark.cpp	Fri Jan 13 01:55:22 2012 -0800
     1.2 +++ b/src/share/vm/gc_implementation/g1/concurrentMark.cpp	Fri Jan 13 13:27:48 2012 -0800
     1.3 @@ -104,17 +104,6 @@
     1.4    return (int) (diff >> _shifter);
     1.5  }
     1.6  
     1.7 -bool CMBitMapRO::iterate(BitMapClosure* cl, MemRegion mr) {
     1.8 -  HeapWord* left  = MAX2(_bmStartWord, mr.start());
     1.9 -  HeapWord* right = MIN2(_bmStartWord + _bmWordSize, mr.end());
    1.10 -  if (right > left) {
    1.11 -    // Right-open interval [leftOffset, rightOffset).
    1.12 -    return _bm.iterate(cl, heapWordToOffset(left), heapWordToOffset(right));
    1.13 -  } else {
    1.14 -    return true;
    1.15 -  }
    1.16 -}
    1.17 -
    1.18  void CMBitMapRO::mostly_disjoint_range_union(BitMap*   from_bitmap,
    1.19                                               size_t    from_start_index,
    1.20                                               HeapWord* to_start_word,

     2.1 --- a/src/share/vm/gc_implementation/g1/concurrentMark.hpp	Fri Jan 13 01:55:22 2012 -0800
     2.2 +++ b/src/share/vm/gc_implementation/g1/concurrentMark.hpp	Fri Jan 13 13:27:48 2012 -0800
     2.3 @@ -84,8 +84,8 @@
     2.4    }
     2.5  
     2.6    // iteration
     2.7 -  bool iterate(BitMapClosure* cl) { return _bm.iterate(cl); }
     2.8 -  bool iterate(BitMapClosure* cl, MemRegion mr);
     2.9 +  inline bool iterate(BitMapClosure* cl, MemRegion mr);
    2.10 +  inline bool iterate(BitMapClosure* cl);
    2.11  
    2.12    // Return the address corresponding to the next marked bit at or after
    2.13    // "addr", and before "limit", if "limit" is non-NULL.  If there is no

     3.1 --- a/src/share/vm/gc_implementation/g1/concurrentMark.inline.hpp	Fri Jan 13 01:55:22 2012 -0800
     3.2 +++ b/src/share/vm/gc_implementation/g1/concurrentMark.inline.hpp	Fri Jan 13 13:27:48 2012 -0800
     3.3 @@ -28,6 +28,35 @@
     3.4  #include "gc_implementation/g1/concurrentMark.hpp"
     3.5  #include "gc_implementation/g1/g1CollectedHeap.inline.hpp"
     3.6  
     3.7 +inline bool CMBitMapRO::iterate(BitMapClosure* cl, MemRegion mr) {
     3.8 +  HeapWord* start_addr = MAX2(startWord(), mr.start());
     3.9 +  HeapWord* end_addr = MIN2(endWord(), mr.end());
    3.10 +
    3.11 +  if (end_addr > start_addr) {
    3.12 +    // Right-open interval [start-offset, end-offset).
    3.13 +    BitMap::idx_t start_offset = heapWordToOffset(start_addr);
    3.14 +    BitMap::idx_t end_offset = heapWordToOffset(end_addr);
    3.15 +
    3.16 +    start_offset = _bm.get_next_one_offset(start_offset, end_offset);
    3.17 +    while (start_offset < end_offset) {
    3.18 +      HeapWord* obj_addr = offsetToHeapWord(start_offset);
    3.19 +      oop obj = (oop) obj_addr;
    3.20 +      if (!cl->do_bit(start_offset)) {
    3.21 +        return false;
    3.22 +      }
    3.23 +      HeapWord* next_addr = MIN2(obj_addr + obj->size(), end_addr);
    3.24 +      BitMap::idx_t next_offset = heapWordToOffset(next_addr);
    3.25 +      start_offset = _bm.get_next_one_offset(next_offset, end_offset);
    3.26 +    }
    3.27 +  }
    3.28 +  return true;
    3.29 +}
    3.30 +
    3.31 +inline bool CMBitMapRO::iterate(BitMapClosure* cl) {
    3.32 +  MemRegion mr(startWord(), sizeInWords());
    3.33 +  return iterate(cl, mr);
    3.34 +}
    3.35 +
    3.36  inline void CMTask::push(oop obj) {
    3.37    HeapWord* objAddr = (HeapWord*) obj;
    3.38    assert(_g1h->is_in_g1_reserved(objAddr), "invariant");

     4.1 --- a/src/share/vm/utilities/bitMap.inline.hpp	Fri Jan 13 01:55:22 2012 -0800
     4.2 +++ b/src/share/vm/utilities/bitMap.inline.hpp	Fri Jan 13 13:27:48 2012 -0800
     4.3 @@ -1,5 +1,5 @@
     4.4  /*
     4.5 - * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
     4.6 + * Copyright (c) 2005, 2012, Oracle and/or its affiliates. All rights reserved.
     4.7   * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
     4.8   *
     4.9   * This code is free software; you can redistribute it and/or modify it
    4.10 @@ -178,8 +178,30 @@
    4.11      for (; !(res & 1); res_offset++) {
    4.12        res = res >> 1;
    4.13      }
    4.14 -    assert(res_offset >= l_offset &&
    4.15 -           res_offset < r_offset, "just checking");
    4.16 +
    4.17 +#ifdef ASSERT
    4.18 +    // In the following assert, if r_offset is not bitamp word aligned,
    4.19 +    // checking that res_offset is strictly less than r_offset is too
    4.20 +    // strong and will trip the assert.
    4.21 +    //
    4.22 +    // Consider the case where l_offset is bit 15 and r_offset is bit 17
    4.23 +    // of the same map word, and where bits [15:16:17:18] == [00:00:00:01].
    4.24 +    // All the bits in the range [l_offset:r_offset) are 0.
    4.25 +    // The loop that calculates res_offset, above, would yield the offset
    4.26 +    // of bit 18 because it's in the same map word as l_offset and there
    4.27 +    // is a set bit in that map word above l_offset (i.e. res != NoBits).
    4.28 +    //
    4.29 +    // In this case, however, we can assert is that res_offset is strictly
    4.30 +    // less than size() since we know that there is at least one set bit
    4.31 +    // at an offset above, but in the same map word as, r_offset.
    4.32 +    // Otherwise, if r_offset is word aligned then it will not be in the
    4.33 +    // same map word as l_offset (unless it equals l_offset). So either
    4.34 +    // there won't be a set bit between l_offset and the end of it's map
    4.35 +    // word (i.e. res == NoBits), or res_offset will be less than r_offset.
    4.36 +
    4.37 +    idx_t limit = is_word_aligned(r_offset) ? r_offset : size();
    4.38 +    assert(res_offset >= l_offset && res_offset < limit, "just checking");
    4.39 +#endif // ASSERT
    4.40      return MIN2(res_offset, r_offset);
    4.41    }
    4.42    // skip over all word length 0-bit runs