)]}'
{
  "commit": "f07f2a3092b70d407a009dae28b44ecc8fbcffb7",
  "tree": "6dcfd61ba4251549b8ccda7b46a37b0fc559016d",
  "parents": [
    "b4c82b428828c0ffff273a49f24a22cb4e18d485"
  ],
  "author": {
    "name": "Peter Xu",
    "email": "peterx@redhat.com",
    "time": "Fri Jun 13 10:12:12 2025 -0400"
  },
  "committer": {
    "name": "Fabiano Rosas",
    "email": "farosas@suse.de",
    "time": "Fri Jul 11 10:37:38 2025 -0300"
  },
  "message": "migration/postcopy: Initialize blocktime context only until listen\n\nBefore this patch, the blocktime context can be created very early, because\npostcopy_ram_supported_by_host() \u003c- migrate_caps_check() can happen during\nmigration object init.\n\nThe trick here is the blocktime context needs system vCPU information,\nwhich seems to be possible to change after that point.  I didn\u0027t verify it,\nbut it doesn\u0027t sound right.\n\nNow move it out and initialize the context only when postcopy listen\nstarts.  That is already during a migration so it should be guaranteed the\nvCPU topology can never change on both sides.\n\nWhile at it, assert that the ctx isn\u0027t created instead this time; the old\n\"if\" trick isn\u0027t needed when we\u0027re sure it will only happen once now.\n\nReviewed-by: Fabiano Rosas \u003cfarosas@suse.de\u003e\nLink: https://lore.kernel.org/r/20250613141217.474825-10-peterx@redhat.com\nSigned-off-by: Peter Xu \u003cpeterx@redhat.com\u003e\nSigned-off-by: Fabiano Rosas \u003cfarosas@suse.de\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "fe940f89b932869ca6d549c9af205f27eb8ed25b",
      "old_mode": 33188,
      "old_path": "migration/postcopy-ram.c",
      "new_id": "dd3615663fa38abc6c21237a83590ac00fd5f88f",
      "new_mode": 33188,
      "new_path": "migration/postcopy-ram.c"
    }
  ]
}