{
  "id": "CVE-2026-72340",
  "url": "https://poc.griffinguard.io/cve/CVE-2026-72340",
  "published_at": "2026-08-15T05:55:47.119Z",
  "title": "net: microchip: vcap: fix races on the shared Super VCAP block",
  "vendor": "Linux",
  "product": "Linux",
  "cvss": {
    "score": 7.8,
    "source": "cna",
    "vector": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H"
  },
  "kev": {
    "in_kev": false,
    "added_at": null,
    "days_after_publication": null
  },
  "first_poc": null,
  "poc_repo_count": 0,
  "exploit_count": 0,
  "suspicious_repo_count": 0,
  "edge_device": null,
  "description": "In the Linux kernel, the following vulnerability has been resolved:\n\nnet: microchip: vcap: fix races on the shared Super VCAP block\n\nThe VCAP instances on a chip are not independent, yet they are locked\nindependently. On sparx5 and lan969x the IS0 and IS2 instances are\nbacked by the same Super VCAP hardware block and share its cache and\ncommand registers: every access drives the shared VCAP_SUPER_CTRL\nregister and moves data through the shared cache registers.\n\nAccessing one instance therefore races with accessing another. The\nper-instance admin-\u003elock cannot prevent this, as each instance takes a\ndifferent lock.\n\nThe locking issue is mostly disguised by the fact that the core usage of\nthe vcap api runs under rtnl. However, the full rule dump in debugfs\ndecodes rules straight from hardware (a READ command followed by a cache\nread) and runs outside rtnl, so it races a concurrent tc-flower rule\nwrite to another Super VCAP instance.\n\nBesides corrupting the dump, the read repopulates the shared cache\nbetween the writers cache fill and its write command, so the writer\ncommits the wrong data and corrupts the hardware entry.\n\nIntroduce vcap_lock() and vcap_unlock() helpers and route every rule\nlock site in the VCAP API and its debugfs code through them. Replace the\nper-instance admin-\u003elock with a single mutex in struct vcap_control that\nserializes access to all instances. The helpers reach it through a new\nadmin-\u003evctrl back-pointer, and the clients initialise and destroy the\ncontrol lock instead of a per-instance one.\n\nNo path holds more than one instance lock, so collapsing them onto a\nsingle mutex cannot self-deadlock.",
  "cvss_by_source": {
    "cna": 7.8
  },
  "repos": [],
  "remediation": {
    "affected": [
      {
        "vendor": "Linux",
        "product": "Linux",
        "ranges": [
          "\u003e= 71c9de995260222e739020104d477af4775a6d26, \u003c 786456d0a244bbd405dfc60e4de51f8b348b9cb1",
          "\u003e= 71c9de995260222e739020104d477af4775a6d26, \u003c 49806bef9572a2e012610517bc14ed0a4db0d1fc",
          "\u003e= 71c9de995260222e739020104d477af4775a6d26, \u003c 1e71a40d101547380590db582213c1f1dce1f041",
          "\u003e= 71c9de995260222e739020104d477af4775a6d26, \u003c 952928564cc5fdb06f92d7e25c6cd2e1d816362b",
          "\u003e= 71c9de995260222e739020104d477af4775a6d26, \u003c d7a8d500d7e42837bd8dce40cb52c97c6e8706a9"
        ],
        "fixed_in": [
          "786456d0a244bbd405dfc60e4de51f8b348b9cb1",
          "49806bef9572a2e012610517bc14ed0a4db0d1fc",
          "1e71a40d101547380590db582213c1f1dce1f041",
          "952928564cc5fdb06f92d7e25c6cd2e1d816362b",
          "d7a8d500d7e42837bd8dce40cb52c97c6e8706a9"
        ]
      },
      {
        "vendor": "Linux",
        "product": "Linux",
        "ranges": [
          "6.2"
        ],
        "fixed_in": [
          "0",
          "6.6.145",
          "6.12.97",
          "6.18.40",
          "7.1.5",
          "7.2"
        ]
      }
    ],
    "upgrade": [
      "Upgrade Linux to 786456d0a244bbd405dfc60e4de51f8b348b9cb1 or 49806bef9572a2e012610517bc14ed0a4db0d1fc or 1e71a40d101547380590db582213c1f1dce1f041 or 952928564cc5fdb06f92d7e25c6cd2e1d816362b or d7a8d500d7e42837bd8dce40cb52c97c6e8706a9 or later.",
      "Upgrade Linux to 0 or 6.6.145 or 6.12.97 or 6.18.40 or 7.1.5 or 7.2 or later."
    ]
  },
  "exploit_refs": [],
  "hunt_profile": [],
  "rules": []
}
