)]}'
{
  "commit": "2ece9ec62e5b276b4635d391c844b2ffcd10bbf1",
  "tree": "58d39b74f06a75bc34d9f4a935dc98eab85a1440",
  "parents": [
    "630e438f040c3838206b5e6717b9b5c29edf3548"
  ],
  "author": {
    "name": "Jack Wang",
    "email": "jinpu.wang@cloud.ionos.com",
    "time": "Mon Jun 21 07:53:37 2021 +0200"
  },
  "committer": {
    "name": "Jason Gunthorpe",
    "email": "jgg@nvidia.com",
    "time": "Mon Jun 21 21:02:21 2021 -0300"
  },
  "message": "RDMA/rtrs-clt: Write path fast memory registration\n\nWith fast memory registration in write path, we can reduce\nthe memory consumption by using less max_send_sge, support IO bigger\nthan 116 KB (29 segments * 4 KB) without splitting, and it also\nmake the IO path more symmetric.\n\nTo avoid some times MR reg failed, waiting for the invalidation to finish\nbefore the new mr reg. Introduce a refcount, only finish the request\nwhen both local invalidation and io reply are there.\n\nLink: https://lore.kernel.org/r/20210621055340.11789-3-jinpu.wang@ionos.com\nSigned-off-by: Jack Wang \u003cjinpu.wang@cloud.ionos.com\u003e\nSigned-off-by: Md Haris Iqbal \u003charis.iqbal@ionos.com\u003e\nSigned-off-by: Dima Stepanov \u003cdmitrii.stepanov@ionos.com\u003e\nSigned-off-by: Jason Gunthorpe \u003cjgg@nvidia.com\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "c451fae5a2b70a7a30d4ba5b5b05fd4eba34b708",
      "old_mode": 33188,
      "old_path": "drivers/infiniband/ulp/rtrs/rtrs-clt.c",
      "new_id": "82c4a12733218d091e5c95298bd111f068b221f3",
      "new_mode": 33188,
      "new_path": "drivers/infiniband/ulp/rtrs/rtrs-clt.c"
    },
    {
      "type": "modify",
      "old_id": "eed2a20ee9be7ab855c97b695815b77206dc969c",
      "old_mode": 33188,
      "old_path": "drivers/infiniband/ulp/rtrs/rtrs-clt.h",
      "new_id": "e276a2dfcf7c7a12686babe95f90e5b34abacf78",
      "new_mode": 33188,
      "new_path": "drivers/infiniband/ulp/rtrs/rtrs-clt.h"
    }
  ]
}