CVE-2025-21939 (GCVE-0-2025-21939)
Vulnerability from cvelistv5
Published
2025-04-01 15:41
Modified
2026-08-05 11:55
Severity ?
7.8 (High) - CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
Summary
In the Linux kernel, the following vulnerability has been resolved: drm/xe/hmm: Don't dereference struct page pointers without notifier lock The pnfs that we obtain from hmm_range_fault() point to pages that we don't have a reference on, and the guarantee that they are still in the cpu page-tables is that the notifier lock must be held and the notifier seqno is still valid. So while building the sg table and marking the pages accesses / dirty we need to hold this lock with a validated seqno. However, the lock is reclaim tainted which makes sg_alloc_table_from_pages_segment() unusable, since it internally allocates memory. Instead build the sg-table manually. For the non-iommu case this might lead to fewer coalesces, but if that's a problem it can be fixed up later in the resource cursor code. For the iommu case, the whole sg-table may still be coalesced to a single contigous device va region. This avoids marking pages that we don't own dirty and accessed, and it also avoid dereferencing struct pages that we don't own. v2: - Use assert to check whether hmm pfns are valid (Matthew Auld) - Take into account that large pages may cross range boundaries (Matthew Auld) v3: - Don't unnecessarily check for a non-freed sg-table. (Matthew Auld) - Add a missing up_read() in an error path. (Matthew Auld) (cherry picked from commit ea3e66d280ce2576664a862693d1da8fd324c317)
References
Impacted products
Vendor Product Version
Linux Linux Version: 81e058a3e7fd8593d076b4f26f7b8bb49f1d61e3
Version: 81e058a3e7fd8593d076b4f26f7b8bb49f1d61e3
Version: 81e058a3e7fd8593d076b4f26f7b8bb49f1d61e3
 Create a notification for this product.
Show details on NVD website

To clipboard 

{
  "containers": {
    "cna": {
      "affected": [
        {
          "defaultStatus": "unaffected",
          "product": "Linux",
          "programFiles": [
            "drivers/gpu/drm/xe/xe_hmm.c"
          ],
          "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
          "vendor": "Linux",
          "versions": [
            {
              "lessThan": "2a24c98f0e4cc994334598d4f3a851972064809d",
              "status": "affected",
              "version": "81e058a3e7fd8593d076b4f26f7b8bb49f1d61e3",
              "versionType": "git"
            },
            {
              "lessThan": "f9326f529da7298a95643c3267f1c0fdb0db55eb",
              "status": "affected",
              "version": "81e058a3e7fd8593d076b4f26f7b8bb49f1d61e3",
              "versionType": "git"
            },
            {
              "lessThan": "0a98219bcc961edd3388960576e4353e123b4a51",
              "status": "affected",
              "version": "81e058a3e7fd8593d076b4f26f7b8bb49f1d61e3",
              "versionType": "git"
            }
          ]
        },
        {
          "defaultStatus": "affected",
          "product": "Linux",
          "programFiles": [
            "drivers/gpu/drm/xe/xe_hmm.c"
          ],
          "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
          "vendor": "Linux",
          "versions": [
            {
              "status": "affected",
              "version": "6.10"
            },
            {
              "lessThan": "6.10",
              "status": "unaffected",
              "version": "0",
              "versionType": "semver"
            },
            {
              "lessThanOrEqual": "6.12.*",
              "status": "unaffected",
              "version": "6.12.19",
              "versionType": "semver"
            },
            {
              "lessThanOrEqual": "6.13.*",
              "status": "unaffected",
              "version": "6.13.7",
              "versionType": "semver"
            },
            {
              "lessThanOrEqual": "*",
              "status": "unaffected",
              "version": "6.14",
              "versionType": "original_commit_for_fix"
            }
          ]
        }
      ],
      "cpeApplicability": [
        {
          "nodes": [
            {
              "cpeMatch": [
                {
                  "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*",
                  "versionEndExcluding": "6.12.19",
                  "versionStartIncluding": "6.10",
                  "vulnerable": true
                },
                {
                  "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*",
                  "versionEndExcluding": "6.13.7",
                  "versionStartIncluding": "6.10",
                  "vulnerable": true
                },
                {
                  "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*",
                  "versionEndExcluding": "6.14",
                  "versionStartIncluding": "6.10",
                  "vulnerable": true
                }
              ],
              "negate": false,
              "operator": "OR"
            }
          ]
        }
      ],
      "descriptions": [
        {
          "lang": "en",
          "value": "In the Linux kernel, the following vulnerability has been resolved:\n\ndrm/xe/hmm: Don\u0027t dereference struct page pointers without notifier lock\n\nThe pnfs that we obtain from hmm_range_fault() point to pages that\nwe don\u0027t have a reference on, and the guarantee that they are still\nin the cpu page-tables is that the notifier lock must be held and the\nnotifier seqno is still valid.\n\nSo while building the sg table and marking the pages accesses / dirty\nwe need to hold this lock with a validated seqno.\n\nHowever, the lock is reclaim tainted which makes\nsg_alloc_table_from_pages_segment() unusable, since it internally\nallocates memory.\n\nInstead build the sg-table manually. For the non-iommu case\nthis might lead to fewer coalesces, but if that\u0027s a problem it can\nbe fixed up later in the resource cursor code. For the iommu case,\nthe whole sg-table may still be coalesced to a single contigous\ndevice va region.\n\nThis avoids marking pages that we don\u0027t own dirty and accessed, and\nit also avoid dereferencing struct pages that we don\u0027t own.\n\nv2:\n- Use assert to check whether hmm pfns are valid (Matthew Auld)\n- Take into account that large pages may cross range boundaries\n  (Matthew Auld)\n\nv3:\n- Don\u0027t unnecessarily check for a non-freed sg-table. (Matthew Auld)\n- Add a missing up_read() in an error path. (Matthew Auld)\n\n(cherry picked from commit ea3e66d280ce2576664a862693d1da8fd324c317)"
        }
      ],
      "metrics": [
        {
          "cvssV3_1": {
            "baseScore": 7.8,
            "baseSeverity": "HIGH",
            "vectorString": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
            "version": "3.1"
          },
          "scenarios": [
            {
              "lang": "en",
              "value": "AV:L - The vulnerable path is reached only through Intel Xe DRM ioctls (`DRM_IOCTL_XE_VM_BIND`, `DRM_IOCTL_XE_EXEC`) or GPU page faults raised by the attacker\u0027s own submitted workload, all requiring an open handle on the local `/dev/dri/renderD*` node. There is no remote or adjacent-network reachability.\nAC:L - The attacker controls both sides of the race \u2014 one thread issues the userptr bind/exec while another munmaps, MADV_DONTNEEDs, or migrates the same user range \u2014 and can widen the window arbitrarily by sizing the userptr range so the intervening `kvmalloc_array()` and `sg_alloc_table_from_pages_segment()` GFP_KERNEL allocations enter direct reclaim. No condition outside the attacker\u0027s influence is required.\nPR:L - An unprivileged local account with render-node access is sufficient; these are all DRM_RENDER_ALLOW ioctls, and render nodes are granted to ordinary logged-in desktop users, Android apps, and GPU-passthrough containers by default. No CAP_SYS_ADMIN or root is needed.\nUI:N - The attacker triggers the entire sequence from its own threads via ioctls and memory-management syscalls on its own address space. No action by any other user is required.\nS:U - The corruption occurs in kernel page/folio metadata and kernel-managed DMA mappings, within the same security authority as the vulnerable driver. Standard kernel memory corruption leading to local privilege escalation does not cross a scope boundary.\nC:H - This is a use-after-free on unreferenced pages: freed pages can be reallocated to other processes or the kernel and are still DMA-mapped DMA_BIDIRECTIONAL for the GPU and read via struct page dereferences, giving an attacker who grooms the reallocation a path to read memory belonging to other security contexts.\nI:H - `set_page_dirty_lock()` on a reallocated folio calls `mapping-\u003ea_ops-\u003edirty_folio()` through a function pointer read from memory the driver no longer owns \u2014 a control-flow hijack primitive \u2014 and also wrongly dirties foreign page-cache folios (causing writeback of unintended data) while `mark_page_accessed()` corrupts LRU lists. Per kernel UAF guidance this yields arbitrary-write/code-execution potential.\nA:H - Even without full exploitation, LRU list manipulation on a freed folio, breaking another owner\u0027s page lock via `folio_unlock()`, and refcount assertions on zero-refcount pages produce list-debug BUGs, oopses, and kernel panics. Any local user with render-node access can trigger this repeatedly."
            }
          ]
        }
      ],
      "providerMetadata": {
        "dateUpdated": "2026-08-05T11:55:29.323Z",
        "orgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
        "shortName": "Linux"
      },
      "references": [
        {
          "url": "https://git.kernel.org/stable/c/2a24c98f0e4cc994334598d4f3a851972064809d"
        },
        {
          "url": "https://git.kernel.org/stable/c/f9326f529da7298a95643c3267f1c0fdb0db55eb"
        },
        {
          "url": "https://git.kernel.org/stable/c/0a98219bcc961edd3388960576e4353e123b4a51"
        }
      ],
      "title": "drm/xe/hmm: Don\u0027t dereference struct page pointers without notifier lock",
      "x_generator": {
        "engine": "bippy-1.2.0"
      }
    }
  },
  "cveMetadata": {
    "assignerOrgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
    "assignerShortName": "Linux",
    "cveId": "CVE-2025-21939",
    "datePublished": "2025-04-01T15:41:05.393Z",
    "dateReserved": "2024-12-29T08:45:45.789Z",
    "dateUpdated": "2026-08-05T11:55:29.323Z",
    "state": "PUBLISHED"
  },
  "dataType": "CVE_RECORD",
  "dataVersion": "5.2"
}


Log in or create an account to share your comment.




Tags
Taxonomy of the tags.




Sightings

Author Source Type Date

Nomenclature

  • Seen: The vulnerability was mentioned, discussed, or seen somewhere by the user.
  • Confirmed: The vulnerability is confirmed from an analyst perspective.
  • Published Proof of Concept: A public proof of concept is available for this vulnerability.
  • Exploited: This vulnerability was exploited and seen by the user reporting the sighting.
  • Patched: This vulnerability was successfully patched by the user reporting the sighting.
  • Not exploited: This vulnerability was not exploited or seen by the user reporting the sighting.
  • Not confirmed: The user expresses doubt about the veracity of the vulnerability.
  • Not patched: This vulnerability was not successfully patched by the user reporting the sighting.