{
  "affected": [
    {
      "ranges": [
        {
          "events": [
            {
              "introduced": "b0473dcd4b1d7e2e44947e7ac1820c73a268821a"
            },
            {
              "fixed": "282d220bae5fbfc90cf0e3d5b5e42c00ad79f989"
            },
            {
              "fixed": "35551efb155e3b83445a6c3f66cb498d5efc182c"
            }
          ],
          "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
          "type": "GIT"
        }
      ]
    },
    {
      "package": {
        "ecosystem": "Linux",
        "name": "Kernel"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "7.1.0"
            },
            {
              "fixed": "7.1.6"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "database_specific": {
    "cna_assigner": "Linux",
    "osv_generated_from": "https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/68xxx/CVE-2026-68438.json"
  },
  "details": "In the Linux kernel, the following vulnerability has been resolved:\n\nsmp: Make CSD lock acquisition atomic for debug mode\n\nCommit b0473dcd4b1d (\"smp: Improve smp_call_function_single()\nCSD-lock diagnostics\") changed smp_call_function_single() so that,\nwhen CSD lock debugging is enabled, async !wait calls use the\ndestination CPU csd_data. That improves diagnostics, but it also removes\nthe single-writer property that made the old csd_lock() safe: multiple\nCPUs can now prepare the same destination CPU CSD concurrently.\n\ncsd_lock() currently waits for CSD_FLAG_LOCK to clear and then sets the\nbit with a non-atomic read-modify-write. Two senders can both see an\nunlocked CSD, set the bit, overwrite the callback fields, and enqueue\nthe same llist node. Re-adding a node that is already the queue head can\nmake node-\u003enext point to itself, leaving the target CPU stuck walking\ncall_single_queue. Later synchronous work, such as a TLB shootdown, can\nthen remain queued and trigger soft-lockup warnings or panics.\n\nKeep the single csd_lock() implementation, but when CSD lock debugging is\nenabled, acquire CSD_FLAG_LOCK with try_cmpxchg_acquire(). This makes the\ndestination CPU CSD a real atomic lock in the only configuration where it\ncan be shared by multiple remote senders, while preserving the existing\nnon-debug fast path.",
  "id": "CVE-2026-68438",
  "modified": "2026-08-18T03:30:50.585770761Z",
  "published": "2026-08-12T00:07:27.105Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://git.kernel.org/stable/c/282d220bae5fbfc90cf0e3d5b5e42c00ad79f989"
    },
    {
      "type": "WEB",
      "url": "https://git.kernel.org/stable/c/35551efb155e3b83445a6c3f66cb498d5efc182c"
    },
    {
      "type": "ADVISORY",
      "url": "https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/68xxx/CVE-2026-68438.json"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-68438"
    },
    {
      "type": "PACKAGE",
      "url": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git"
    }
  ],
  "schema_version": "1.9.0",
  "summary": "smp: Make CSD lock acquisition atomic for debug mode"
}