{
  "affected": [
    {
      "ranges": [
        {
          "events": [
            {
              "introduced": "8ea636848aca35b9f97c5b5dee30225cf2dd0fe6"
            },
            {
              "fixed": "c76b8abce575e0c6e4096957220b4515ed847d89"
            },
            {
              "fixed": "a6f0643e4f63cfaa0d5d4a69de4f132eac4b8fe4"
            }
          ],
          "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
          "type": "GIT"
        }
      ]
    },
    {
      "package": {
        "ecosystem": "Linux",
        "name": "Kernel"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "5.10.0"
            },
            {
              "fixed": "7.1.4"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "database_specific": {
    "cna_assigner": "Linux",
    "osv_generated_from": "https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/64xxx/CVE-2026-64192.json"
  },
  "details": "In the Linux kernel, the following vulnerability has been resolved:\n\nbpf: Reject BPF_MAP_TYPE_INODE_STORAGE creation if BPF LSM is uninitialized\n\nWhen CONFIG_BPF_LSM=y is set, BPF inode storage maps\n(BPF_MAP_TYPE_INODE_STORAGE) are compiled into the kernel. However,\nif the BPF LSM is not explicitly enabled at boot time (e.g. omitted\nfrom the \"lsm=\" boot parameter), lsm_prepare() is never executed for\nthe BPF LSM.\n\nConsequently, the BPF inode security blob offset\n(bpf_lsm_blob_sizes.lbs_inode) is never initialized and remains at\nits default compiled size of 8 bytes instead of being updated to a\nvalid offset past the reserved struct rcu_head (typically 16 bytes\nor more).\n\nWhen a privileged user creates and updates a BPF_MAP_TYPE_INODE_STORAGE\nmap, bpf_inode() evaluates inode-\u003ei_security + 8. This erroneously\naliases the struct rcu_head.func callback pointer at the beginning\nof the inode-\u003ei_security blob. During subsequent map element cleanup\nor inode destruction, writing NULL to owner_storage clears the queued\nRCU callback pointer. When rcu_do_batch() later executes the queued\ncallback, it attempts an instruction fetch at address 0x0, triggering\nan immediate kernel panic.\n\nFix this by introducing a global bpf_lsm_initialized boolean flag\nmarked with __ro_after_init. Set this flag to true inside bpf_lsm_init()\nwhen the LSM framework successfully registers the BPF LSM. Gate map\nallocation in inode_storage_map_alloc() on this flag, returning\n-EOPNOTSUPP if the BPF LSM is in turn uninitialized.\n\nThis fail-fast approach prevents userspace from allocating inode\nstorage maps when the supporting BPF LSM infrastructure is absent,\navoiding zombie map states.",
  "id": "CVE-2026-64192",
  "modified": "2026-07-22T05:30:00.037421135Z",
  "published": "2026-07-20T16:27:51.720Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://git.kernel.org/stable/c/a6f0643e4f63cfaa0d5d4a69de4f132eac4b8fe4"
    },
    {
      "type": "WEB",
      "url": "https://git.kernel.org/stable/c/c76b8abce575e0c6e4096957220b4515ed847d89"
    },
    {
      "type": "ADVISORY",
      "url": "https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/64xxx/CVE-2026-64192.json"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-64192"
    },
    {
      "type": "PACKAGE",
      "url": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git"
    }
  ],
  "schema_version": "1.8.0",
  "summary": "bpf: Reject BPF_MAP_TYPE_INODE_STORAGE creation if BPF LSM is uninitialized"
}