)]}'
{
  "commit": "f6fe9fea995249ecc2cd72975d803fbf4d512c02",
  "tree": "2df9f78d56a24c7a1e32dd0778b09cff83fcef9b",
  "parents": [
    "34e104b897da6e144a5f34e7c5eebf8a4c4d9d59"
  ],
  "author": {
    "name": "Yuan Liu",
    "email": "yuan1.liu@intel.com",
    "time": "Mon Jun 10 18:21:09 2024 +0800"
  },
  "committer": {
    "name": "Fabiano Rosas",
    "email": "farosas@suse.de",
    "time": "Fri Jun 14 14:01:29 2024 -0300"
  },
  "message": "migration/multifd: implement qpl compression and decompression\n\nQPL compression and decompression will use IAA hardware path if the IAA\nhardware is available. Otherwise the QPL library software path is used.\n\nThe hardware path will automatically fall back to QPL software path if\nthe IAA queues are busy. In some scenarios, this may happen frequently,\nsuch as configuring 4 channels but only one IAA device is available. In\nthe case of insufficient IAA hardware resources, retry and fallback can\nhelp optimize performance:\n\n 1. Retry + SW fallback:\n    total time: 14649 ms\n    downtime: 25 ms\n    throughput: 17666.57 mbps\n    pages-per-second: 1509647\n\n 2. No fallback, always wait for work queues to become available\n    total time: 18381 ms\n    downtime: 25 ms\n    throughput: 13698.65 mbps\n    pages-per-second: 859607\n\nIf both the hardware and software paths fail, the uncompressed page is\nsent directly.\n\nSigned-off-by: Yuan Liu \u003cyuan1.liu@intel.com\u003e\nReviewed-by: Nanhai Zou \u003cnanhai.zou@intel.com\u003e\nReviewed-by: Fabiano Rosas \u003cfarosas@suse.de\u003e\nSigned-off-by: Fabiano Rosas \u003cfarosas@suse.de\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "6791a204d55ffa074071f73905332a9fa080d4e6",
      "old_mode": 33188,
      "old_path": "migration/multifd-qpl.c",
      "new_id": "9265098ee799f625e2391fade95802fb71b3ea2a",
      "new_mode": 33188,
      "new_path": "migration/multifd-qpl.c"
    }
  ]
}