{
  "containers": {
    "cna": {
      "affected": [
        {
          "defaultStatus": "unaffected",
          "product": "Linux",
          "programFiles": [
            "fs/bpf_fs_kfuncs.c"
          ],
          "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
          "vendor": "Linux",
          "versions": [
            {
              "lessThan": "7504b677c4370c9e4b153c553ff8831c093b1ff6",
              "status": "affected",
              "version": "56467292794b800164df20c076c409ac548e56ec",
              "versionType": "git"
            },
            {
              "lessThan": "640d37fbd34d03af727dbc567f48c0203d1de6b4",
              "status": "affected",
              "version": "56467292794b800164df20c076c409ac548e56ec",
              "versionType": "git"
            },
            {
              "lessThan": "35d442ed1f86465e49df3119fb898f985186db13",
              "status": "affected",
              "version": "56467292794b800164df20c076c409ac548e56ec",
              "versionType": "git"
            }
          ]
        },
        {
          "defaultStatus": "affected",
          "product": "Linux",
          "programFiles": [
            "fs/bpf_fs_kfuncs.c"
          ],
          "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
          "vendor": "Linux",
          "versions": [
            {
              "status": "affected",
              "version": "6.15"
            },
            {
              "lessThan": "6.15",
              "status": "unaffected",
              "version": "0",
              "versionType": "semver"
            },
            {
              "lessThanOrEqual": "6.18.*",
              "status": "unaffected",
              "version": "6.18.55",
              "versionType": "semver"
            },
            {
              "lessThanOrEqual": "7.2.*",
              "status": "unaffected",
              "version": "7.2.9",
              "versionType": "semver"
            },
            {
              "lessThanOrEqual": "*",
              "status": "unaffected",
              "version": "7.3",
              "versionType": "original_commit_for_fix"
            }
          ]
        }
      ],
      "cpeApplicability": [
        {
          "nodes": [
            {
              "cpeMatch": [
                {
                  "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*",
                  "versionEndExcluding": "6.18.55",
                  "versionStartIncluding": "6.15",
                  "vulnerable": true
                },
                {
                  "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*",
                  "versionEndExcluding": "7.2.9",
                  "versionStartIncluding": "6.15",
                  "vulnerable": true
                },
                {
                  "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*",
                  "versionEndExcluding": "7.3",
                  "versionStartIncluding": "6.15",
                  "vulnerable": true
                }
              ],
              "negate": false,
              "operator": "OR"
            }
          ]
        }
      ],
      "descriptions": [
        {
          "lang": "en",
          "value": "In the Linux kernel, the following vulnerability has been resolved:\n\nbpf: fs/xattr: don\u0027t assume the inode is locked in path_unlink/path_rmdir\n\nbpf_lsm_has_d_inode_locked() makes the verifier rewrite\nbpf_[set|remove]_dentry_xattr() to the _locked variants, which assume\nthat the caller already holds the inode\u0027s i_rwsem.  The path_unlink and\npath_rmdir hooks are listed, but security_path_unlink() and\nsecurity_path_rmdir() run before vfs_unlink()/vfs_rmdir() take the\nvictim inode\u0027s i_rwsem, so a sleepable BPF LSM program attached to\neither hook mutates the victim\u0027s xattrs without the lock held.\n\nDrop the two path hooks from d_inode_locked_hooks so that the verifier\nkeeps the locking bpf_[set|remove]_dentry_xattr() variants, which take\nthe lock themselves."
        }
      ],
      "metrics": [
        {
          "cvssV3_1": {
            "baseScore": 7.3,
            "baseSeverity": "HIGH",
            "vectorString": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:C/C:N/I:L/A:H",
            "version": "3.1"
          },
          "scenarios": [
            {
              "lang": "en",
              "value": "AV:L - unlink(2)/unlinkat(2)/rmdir(2) reach filename_unlinkat()/filename_rmdir() -\u003e security_path_unlink()/security_path_rmdir() -\u003e bpf_lsm_path_*(); vulnerable bpf_lsm_has_d_inode_locked() classification makes the verifier redirect bpf_[set|remove]_dentry_xattr() to _locked functions. The trigger enters locally, not via a protocol.\nAC:L - On UBIFS, one invocation deterministically fails ubifs_xattr_set()/ubifs_xattr_remove()\u0027s inode_is_locked() assertion; for races, the attacker can unlink two hardlinks in different parents concurrently, so separate parent locks do not serialize the same victim. No condition is outside the attacker\u0027s control.\nPR:L - bpf_prog_load() gates BPF_PROG_TYPE_LSM on bpf_token_capable(CAP_BPF) and CAP_PERFMON. A delegated bpffs token makes those ns_capable(token-\u003euserns, ...) checks, so a namespace-confined non-admin can load and attach the sleepable path hook; without delegation, init-namespace capabilities are required.\nUI:N - The attacker loads and attaches the BPF LSM program, then invokes unlinkat(2) or rmdir(2) on an object it can modify; no victim must mount media, open a file, or perform another action.\nS:C - With bpffs token delegation, a task confined to the token\u0027s user namespace can run the faulty BPF LSM program in the host kernel and force a shared host UBIFS read-only, crossing the container or sandbox availability boundary.\nC:N - Neither affected helper returns filesystem or kernel data: bpf_set_dentry_xattr_locked() only passes a verifier-bounded dynptr to __vfs_setxattr(), and the remove helper has no output. The missing i_rwsem is a mutation-ordering flaw, with no demonstrated OOB read, UAF, or disclosure path.\nI:L - btrfs_setxattr() explicitly relies on the inode lock to exclude a concurrent delete between its XATTR_REPLACE lookup and insert; this bug removes that exclusion, so concurrent path_unlink hooks can recreate or lose security.bpf.* metadata. Btrfs tree locking shows no arbitrary-memory-write primitive, limiting integrity impact.\nA:H - ubifs_xattr_set(..., check_lock=true) and ubifs_xattr_remove() assert inode_is_locked(); default ASSACT_RO calls ubifs_ro_mode(), sets SB_RDONLY, and denies later writes. One path hook calling the mis-selected _locked helper can therefore make a mounted UBIFS lose write availability."
            }
          ]
        }
      ],
      "providerMetadata": {
        "dateUpdated": "2026-10-24T14:09:20.159Z",
        "orgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
        "shortName": "Linux"
      },
      "references": [
        {
          "url": "https://git.kernel.org/stable/c/7504b677c4370c9e4b153c553ff8831c093b1ff6"
        },
        {
          "url": "https://git.kernel.org/stable/c/640d37fbd34d03af727dbc567f48c0203d1de6b4"
        },
        {
          "url": "https://git.kernel.org/stable/c/35d442ed1f86465e49df3119fb898f985186db13"
        }
      ],
      "title": "bpf: fs/xattr: don\u0027t assume the inode is locked in path_unlink/path_rmdir",
      "x_generator": {
        "engine": "bippy-1.2.0"
      }
    }
  },
  "cveMetadata": {
    "assignerOrgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
    "assignerShortName": "Linux",
    "cveId": "CVE-2026-98543",
    "datePublished": "2026-10-18T15:18:10.823Z",
    "dateReserved": "2026-09-25T10:25:14.356Z",
    "dateUpdated": "2026-10-24T14:09:20.159Z",
    "state": "PUBLISHED"
  },
  "dataType": "CVE_RECORD",
  "dataVersion": "5.2"
}