)]}'
{
  "commit": "e1051cbaad46dc98967bee2e00a697d7f82b6658",
  "tree": "489b9c4b84490310f360e205a0ce9567fec11527",
  "parents": [
    "558540feea2ad57e9b8daecce1cb522df7d9266a"
  ],
  "author": {
    "name": "Nicolas Capens",
    "email": "capn@google.com",
    "time": "Tue Sep 01 09:45:17 2020 -0400"
  },
  "committer": {
    "name": "Nicolas Capens",
    "email": "nicolascapens@google.com",
    "time": "Tue Sep 01 20:40:21 2020 +0000"
  },
  "message": "Fix undefined behavior in minifloat conversion\n\nShifting a 32-bit value by 32 or more is undefined behavior in C++. This\npreviously happened in this code when converting a 32-bit float value to\nan 11- or 10-bit minifloat which is too small to be represented as a\ndenormal, and should produce 0 instead.\n\nInstead of going through this arithmetic for denormals, just test\nwhether the input value is too small to produce a valid denormal, and\nreturn 0 instead.\n\nBug: chromium:1117433\nChange-Id: I8149996fb6d66d328db45725c4cdb81dc7826a10\nReviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/48069\nPresubmit-Ready: Nicolas Capens \u003cnicolascapens@google.com\u003e\nTested-by: Nicolas Capens \u003cnicolascapens@google.com\u003e\nKokoro-Result: kokoro \u003cnoreply+kokoro@google.com\u003e\nReviewed-by: Alexis Hétu \u003csugoi@google.com\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "577514185eaee07f7da07f282fd283251544ecea",
      "old_mode": 33188,
      "old_path": "src/System/Half.hpp",
      "new_id": "84025ef002359f493a642b08663a555d2275c7b8",
      "new_mode": 33188,
      "new_path": "src/System/Half.hpp"
    }
  ]
}