)]}'
{
  "log": [
    {
      "commit": "c220f981809d655fb061c92a581fd6fc4df4da5c",
      "tree": "77f494f865e9b3c24340b5761a5d762408eb3a78",
      "parents": [
        "5de31a1dd5ac6e12b37517550a609b9c0576f685"
      ],
      "author": {
        "name": "Hans Boehm",
        "email": "hboehm@google.com",
        "time": "Fri Oct 12 16:15:45 2018 -0700"
      },
      "committer": {
        "name": "Hans Boehm",
        "email": "hboehm@google.com",
        "time": "Wed Dec 12 21:03:14 2018 -0800"
      },
      "message": "Change GC triggering to use mallinfo()\n\nRevise the test of whether we should collect based on native allocation\nto be based on mallinfo.\n\nMake the GC triggering threshold for native allocations grow somewhat\nwith the Java heap size to account for the added cost of Java\ncollection. We currently compromise between that goal and a\ndesire to avoid drastic changes that might provoke new heap\ngrowth issues. It does now take more native allocations to\ntrigger a GC earlier in the normal Java GC cycle than later in the\ncycle, as it should.\n\nAdd the notifyNativeAllocations() interface to replace\nregisterNativeAllocation() in the common case in which native memory\nis completely allocated via malloc(), and thus no longer needs to be\naccounted for separately. Arrange for it to be called much more\nrarely, since its cost is higher, and we can save time by not going\nthrough JNI on every allocation.\n\nAdd 175-alloc-bignums test.\n\nCleanups/fixes:\n\nFix race in IsOutOfMemoryOnAllocation.\n\nRename max_allowed_footprint_ to avoid suggesting it\u0027s a hard limit.\nIt wasn\u0027t and it isn\u0027t. Make it atomic, since it\u0027s concurrently\nmodified.\n\nFix a few cases in which unsigned counters could become negative.\nI think the only bad consequences of this were weird log messages.\n\nFix integer sizes/signedness around GrowForUtilization.\nUse uint64_t only when we can exceed address space size, and\nsize_t otherwise. Improve overflow checking.\n\nMake allocator or collector tests check the important kinds first, since\nthey\u0027re not always known at compile time.\n\nBug: 111447610\n\nTest: Built and booted AOSP. Ran 175-alloc-big-bignums.\n\nChange-Id: I55deac86622019fb85bbd569c3ae8afab2d13d9a\n"
    }
  ]
}