7042285: G1: native memory leak during humongous object allocation
6804436: G1: heap region indices should be size_t
A series of fixes and improvements to the HeapRegionSeq class: a) replace the _regions growable array with a standard C array, b) avoid de-allocating / re-allocating HeapRegion instances when the heap shrinks / grows (fix for 7042285), c) introduce fast method to map address to HeapRegion via a "biased" array pointer, d) embed the _hrs object in G1CollectedHeap, instead of pointing to it via an indirection, e) assume that all the regions added to the HeapRegionSeq instance are contiguous, f) replace int's with size_t's for indexes (and expand that to HeapRegion as part of 6804436), g) remove unnecessary / unused methods, h) rename a couple of fields (_alloc_search_start and _seq_bottom), i) fix iterate_from() not to always start from index 0 irrespective of the region passed to it, j) add a verification method to check the HeapRegionSeq assumptions, k) always call the wrappers for _hrs.iterate(), _hrs_length(), and _hrs.at() from G1CollectedHeap, not those methods directly, and l) unify the code that expands the sequence (by either re-using or creating a new HeapRegion) and make it robust wrt to a HeapRegion allocation failing.
Reviewed-by: stefank, johnc, brutisso
This changeset extends the logging information generated by +PrintGCDetails to also print out separate size transitions for the eden, survivors, and old regions.
Reviewed-by: ysr, brutisso
The LNC array is sized before the start of a scavenge, while the heap may expand during a scavenge. With CMS, the last block of an arbitrary suffice of the LNC array may expand due to coalition with the expansion delta. We now take care not to attempt access past the end of the LNC array. LNC array code will be cleaned up and suitably encapsulated as part of the forthcoming performance RFE 7043675.
Reviewed-by: brutisso
The freeness bit was being cleared in debug code when it shouldn't have been. Also removed unused FreeChunk methods linkAfterNonNull and clearPrev.
Reviewed-by: brutisso
Do a one-step look-ahead, when sweeping free or garbage blocks, to avoid overstepping sweep limit, which may become a non-block-boundary because of a heap expansion delta coalescing with a previously co-terminal free block.
Reviewed-by: brutisso, tonyp
During an evacuation pause clear the region fields of any concurrent marking task whose local finger points into the collection set as the values in the region fields will become stale. Clearing these fields causes the concurrent mark task to claim a new region when marking restarts after the pause.
Reviewed-by: tonyp, iveresov
Fixed assertion checking code that was attempting to translate addresses past end of space for card-table slot. Also elaborated some assertion checking messages.
Reviewed-by: iveresov, jmasa, tonyp
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
There is a race in the evac failure handling code that causes the condition the assert checks not to be true. The fix is to replace the too-strong assert with a more targeted one.
Reviewed-by: johnc, ysr, jcoomes
We should only undirty cards after we decide that they are not on a young region, not before. The fix also includes improvements to the verify_dirty_region() method which print out which cards were not found dirty.
Reviewed-by: johnc, brutisso
We were calling STS join and leave during a STW pause and we are not suppoesed to. I now only call those during concurrent phase. I also added stress code in the non-product builds to force an overflows (the condition that ws uncovering the bug) to make sure it does not happen again.
Reviewed-by: johnc, brutisso
In G1 heap verification, we no longer scan perm to G1-collected heap refs as part of process_strong_roots() but rather in a separate explicit oop iteration over the perm gen. This preserves the original perm card-marks. Added a new assertion in younger_refs_iterate() to catch a simple subcase where the user may have forgotten a prior save_marks() call, as happened in the case of G1's attempt to iterate perm to G1 refs when verifying the heap before exit. The assert was deliberately weakened for ParNew+CMS and will be fixed for that combination in a future CR. Also made some (non-G1) cleanups related to code and comments obsoleted by the migration of Symbols to the native heap.
Reviewed-by: iveresov, jmasa, tonyp
Short-circuited an unnecessary double traversal of dirty card windows when iterating younger refs. Also renamed some cardtable methods for more clarity.
Reviewed-by: jmasa, stefank, poonam
It introduces non-product cmd line parameter G1DummyRegionsPerGC which indicates how many "dummy" regions to allocate at the end of each GC. This allows the G1 heap to grow artificially and makes concurrent marking cycles more frequent irrespective of what the application that is running is doing. The dummy regions will be found totally empty during cleanup so this parameter can also be used to stress the concurrent cleanup operation.
Reviewed-by: brutisso, johnc
Missing parentheses around a casted expression and some missing casts were causing build failures with the jdk6 build tools.
Reviewed-by: kvn, brutisso
Removed the unused clear argument to various cardtbale methods and unused mod_oop_in_space_iterate method. Unrelated to synopsis, added a pair of clarifying parens in AllocationStats constructor.
Reviewed-by: brutisso, jcoomes
A referent object that is only weakly reachable at the start of concurrent marking but is re-attached to the strongly reachable object graph during marking may not be marked as live. This can cause the reference object to be processed prematurely and leave dangling pointers to the referent object. Implement a read barrier for the java.lang.ref.Reference::referent field by intrinsifying the Reference.get() method, and intercepting accesses though JNI, reflection, and Unsafe, so that when a non-null referent object is read it is also logged in an SATB buffer.
Reviewed-by: kvn, iveresov, never, tonyp, dholmes
Use _max_num_q = max(discovery_degree, processing_degree), and let balance_queues() redistribute from discovery_degree to processing_degree of queues. This should also allow a more dynamic and flexible parallelism policy in the future.
Reviewed-by: jmasa, johnc
Count enable_icms() and disable_icms() events so as to prevent inteference between concurrent calls, which can cause the iCMS thread to be left stranded in icms_wait() with an unserviced request and no young allocations to unwedge it.
Reviewed-by: jcoomes, poonam
Bulk zeroing reduction didn't work with G1, because arraycopy would call pre-barriers on uninitialized oops. The solution is to have version of arraycopy stubs that don't have pre-barriers. Also refactored arraycopy stubs generation on SPARC to be more readable and reduced the number of stubs necessary in some cases.
Reviewed-by: jrose, kvn, never
Repurpose the existing flag G1PrintRegionLivenessInfo to print out the liveness distribution across the regions in the heap at the end of marking.
Reviewed-by: iveresov, jwilhelm
The epochs on the card cache array are initialized to 0 and our initial epoch also starts at 0. So, until the first GC, it might be possible to successfully "claim" a card which was in fact never initialized.
Reviewed-by: johnc, iveresov, ysr
7023151: G1: refactor the code that operates on _cur_alloc_region to be re-used for allocs by the GC threads
7018286: G1: humongous allocation attempts should take the GC locker into account
First, this change replaces the asymmetric locking scheme in the G1 slow alloc path by a summetric one. Second, it factors out the code that operates on _cur_alloc_region so that it can be re-used for allocations by the GC threads in the future.
Reviewed-by: stefank, brutisso, johnc
Manage allocation/freeing of the card cache counts and epochs arrays directly so that an allocation failure while attempting to expand these arrays does not abort the JVM. Failure to expand these arrays is not fatal.
Reviewed-by: iveresov, tonyp
Changed default value of WorkStealingYieldsBeforeSleep from 1000 to 5000. Added more information to G1 pause logging.
Reviewed-by: jwilhelm, tonyp, jmasa
