)]}'
{
  "commit": "b5e4a225e4102157ebf5c53ccdcb749ebe914797",
  "tree": "35cb238cad5e7cd348ee8a6c73d0945ed478c175",
  "parents": [
    "d1673c2191034a65b38c5553a19572f26b122703"
  ],
  "author": {
    "name": "Chris Kennelly",
    "email": "ckennelly@google.com",
    "time": "Mon Sep 10 11:47:15 2018 -0400"
  },
  "committer": {
    "name": "Adam Langley",
    "email": "agl@google.com",
    "time": "Mon Sep 17 21:02:59 2018 +0000"
  },
  "message": "Use sdallocx, if available, when deallocating.\n\nProviding a size hint to the allocator is substantially faster,\nespecially as we already know/need the size for OPENSSL_cleanse.\n\nWe provide a weak symbol that falls back to free when a malloc with\nsdallocx is not statically linked with BoringSSL.\n\nAlternatives considered:\n* Use dlsym():  This is prone to fail on statically linked binaries\n  without symbols.  Additionally, the extra indirection adds call\n  overhead above and beyond the linker resolved technique we\u0027re using.\n* Use CMake rules to identify whether sdallocx is available:  Once the\n  library is built, we may link against a variety of malloc\n  implementations (not all of which may have sdallocx), so we need to\n  have a fallback when the symbol is unavailable.\n\nChange-Id: I3a78e88fac5b6e5d4712aa0347d2ba6b43046e07\nReviewed-on: https://boringssl-review.googlesource.com/31784\nReviewed-by: Chris Kennelly \u003cckennelly@google.com\u003e\nReviewed-by: Adam Langley \u003cagl@google.com\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "5d45baa1e9c7cb530dff27c6c57238065d59331f",
      "old_mode": 33188,
      "old_path": "crypto/mem.c",
      "new_id": "a06061b49a0753aad4c567e94bb1587e2c6d49ea",
      "new_mode": 33188,
      "new_path": "crypto/mem.c"
    }
  ]
}