)]}'
{
  "commit": "775f1686a3df68bd20370f1fabc6273883e2c5d2",
  "tree": "2ac4c57922d756e79bfbd0087cceb1940ec03644",
  "parents": [
    "9ae129017c7fc1fa638d6beedd3802b515ca692b"
  ],
  "author": {
    "name": "Thomas Schwinge",
    "email": "thomas@codesourcery.com",
    "time": "Thu Mar 26 17:34:01 2020 +0100"
  },
  "committer": {
    "name": "Thomas Schwinge",
    "email": "thomas@codesourcery.com",
    "time": "Thu Mar 26 17:34:01 2020 +0100"
  },
  "message": "Really fix og9 \"Fix hang when running oacc exec with CUDA 9.0 nvprof\"\n\nIn my yesterday\u0027s commit 9ae129017c7fc1fa638d6beedd3802b515ca692b \u0027Fix og9 \"Fix\nhang when running oacc exec with CUDA 9.0 nvprof\"\u0027, I wrongly blamed the og9\n\"Fix hang when running oacc exec with CUDA 9.0 nvprof\" to be responsible for\nthe hang being back, when in fact it\u0027s only the later og9 \"OpenACC Profiling\nInterface (incomplete)\" backport from trunk that introduced the problem.\n\nThe confusing thing was that the og9 \"Fix hang when running oacc exec with CUDA\n9.0 nvprof\" commit appears *before* the og9 \"OpenACC Profiling Interface\n(incomplete)\" backport that it relates to.\n\nAnd, in addition to that, I pushed the wrong (incomplete) version of my fix.\n\n\tlibgomp/\n\t* oacc-init.c (acc_init_1): Move other \u0027acc_init_state\u0027 logic to\n\twhere it belongs.\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "75c459179981502c4c346fa91aea153c5e54df4e",
      "old_mode": 33188,
      "old_path": "libgomp/ChangeLog.omp",
      "new_id": "922e00fbff5b5e3fac28e7343b02fec4fb9a4cb0",
      "new_mode": 33188,
      "new_path": "libgomp/ChangeLog.omp"
    },
    {
      "type": "modify",
      "old_id": "765fa2f3b9521acbd602da9f00a82160ef320c41",
      "old_mode": 33188,
      "old_path": "libgomp/oacc-init.c",
      "new_id": "40c14fa9bf2b3c54b3c47d50fe5c6892f5595a1f",
      "new_mode": 33188,
      "new_path": "libgomp/oacc-init.c"
    }
  ]
}