GHSA-7XGX-C4JR-VFQ9
Vulnerability from github – Published: 2026-06-25 09:31 – Updated: 2026-06-25 09:31In the Linux kernel, the following vulnerability has been resolved:
mm/huge_memory: use correct flags for device private PMD entry
Commit 65edfda6f3f2 ("mm/rmap: extend rmap and migration support device-private entries") updated set_pmd_migration_entry() to use pmdp_huge_get_and_clear() in the softleaf case, but made no further adjustments to the function itself.
Therefore this function continues to incorrectly use pmd_write(), pmd_soft_dirty() and pmd_uffd_wp() to determine whether the installed migration entry should be marked writable, softdirty or uffd-wp respectively.
Whilst all are incorrect, the most problematic of these is pmd_write(), as this can lead to corrupted rmap state.
On x86-64 _PAGE_SWP_SOFT_DIRTY is aliased to _PAGE_RW. So calling pmd_write() on a softleaf will return the softdirty state encoded in the entry, assuming CONFIG_MEM_SOFT_DIRTY was enabled.
This was observed when running the hmm.hmm_device_private.anon_write_child selftest:
-
The test faults in a range then migrates it such that a device-private THP range is established.
-
The parent then migrates it to a device-private writable PMD entry whose folio is entirely AnonExclusive with entire_mapcount=1, softdirty set (accidentally correct write state).
-
The parent forks and the PMD entries are set to device-private read only entries, entire_mapcount=2, softdirty still set.
-
[BUG] The child writes to the range then migrates to RAM - intending to install non-writable migration entries - but replacing parent and child PMD mappings with WRITABLE entries due to misinterpreting the softdirty bit.
-
In remove_migration_pmd(), if !softleaf_is_migration_read(entry) we set the RMAP_EXCLUSIVE flag when calling folio_add_anon_rmap_pmd() for both parent and child, which are therefore AnonExclusive.
-
[SPLAT] Child sets migrated folio entire_mapcount=1, parent sets entire_mapcount=2 and we end up with an AnonExclusive folio with entire_mapcount=2! Assert fires in __folio_add_anon_rmap():
VM_WARN_ON_FOLIO(folio_test_large(folio) && folio_entire_mapcount(folio) > 1 && PageAnonExclusive(cur_page), folio)
This patch fixes the issue by correctly referencing the softleaf entry fields for writable, softdirty and uffd-wp in set_pmd_migration_entry().
It also only updates A/D flags if the entry is present as these are otherwise not meaningful for a softleaf entry.
This patch also flips the if (!present) { ... } else { ... } logic in set_pmd_migration_entry() so it is easier to understand, and adds some comments to make things clearer.
I was able to bisect this to commit 775465fd26a3 ("lib/test_hmm: add zone device private THP test infrastructure") which first exposes this bug as it was the commit that permitted test_hmm to generate the test.
However commit 65edfda6f3f2 ("mm/rmap: extend rmap and migration support device-private entries") is the commit that actually enabled this behaviour.
{
"affected": [],
"aliases": [
"CVE-2026-53155"
],
"database_specific": {
"cwe_ids": [],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2026-06-25T09:16:32Z",
"severity": null
},
"details": "In the Linux kernel, the following vulnerability has been resolved:\n\nmm/huge_memory: use correct flags for device private PMD entry\n\nCommit 65edfda6f3f2 (\"mm/rmap: extend rmap and migration support\ndevice-private entries\") updated set_pmd_migration_entry() to use\npmdp_huge_get_and_clear() in the softleaf case, but made no further\nadjustments to the function itself.\n\nTherefore this function continues to incorrectly use pmd_write(),\npmd_soft_dirty() and pmd_uffd_wp() to determine whether the installed\nmigration entry should be marked writable, softdirty or uffd-wp\nrespectively.\n\nWhilst all are incorrect, the most problematic of these is pmd_write(), as\nthis can lead to corrupted rmap state.\n\nOn x86-64 _PAGE_SWP_SOFT_DIRTY is aliased to _PAGE_RW. So calling\npmd_write() on a softleaf will return the softdirty state encoded in the\nentry, assuming CONFIG_MEM_SOFT_DIRTY was enabled.\n\nThis was observed when running the hmm.hmm_device_private.anon_write_child\nselftest:\n\n1. The test faults in a range then migrates it such that a device-private\n THP range is established.\n\n2. The parent then migrates it to a device-private writable PMD entry whose\n folio is entirely AnonExclusive with entire_mapcount=1, softdirty set\n (accidentally correct write state).\n\n3. The parent forks and the PMD entries are set to device-private read only\n entries, entire_mapcount=2, softdirty still set.\n\n4. [BUG] The child writes to the range then migrates to RAM - intending to\n install non-writable migration entries - but replacing parent and child\n PMD mappings with WRITABLE entries due to misinterpreting the softdirty\n bit.\n\n5. In remove_migration_pmd(), if !softleaf_is_migration_read(entry) we\n set the RMAP_EXCLUSIVE flag when calling folio_add_anon_rmap_pmd() for\n both parent and child, which are therefore AnonExclusive.\n\n6. [SPLAT] Child sets migrated folio entire_mapcount=1, parent sets\n entire_mapcount=2 and we end up with an AnonExclusive folio with\n entire_mapcount=2! Assert fires in __folio_add_anon_rmap():\n\n\t\tVM_WARN_ON_FOLIO(folio_test_large(folio) \u0026\u0026\n\t\t\t\t folio_entire_mapcount(folio) \u003e 1 \u0026\u0026\n\t\t\t\t PageAnonExclusive(cur_page), folio)\n\nThis patch fixes the issue by correctly referencing the softleaf entry\nfields for writable, softdirty and uffd-wp in set_pmd_migration_entry().\n\nIt also only updates A/D flags if the entry is present as these are\notherwise not meaningful for a softleaf entry.\n\nThis patch also flips the if (!present) { ... } else { ... } logic in\nset_pmd_migration_entry() so it is easier to understand, and adds some\ncomments to make things clearer.\n\nI was able to bisect this to commit 775465fd26a3 (\"lib/test_hmm: add zone\ndevice private THP test infrastructure\") which first exposes this bug as\nit was the commit that permitted test_hmm to generate the test.\n\nHowever commit 65edfda6f3f2 (\"mm/rmap: extend rmap and migration support\ndevice-private entries\") is the commit that actually enabled this\nbehaviour.",
"id": "GHSA-7xgx-c4jr-vfq9",
"modified": "2026-06-25T09:31:19Z",
"published": "2026-06-25T09:31:19Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-53155"
},
{
"type": "WEB",
"url": "https://git.kernel.org/stable/c/43e7f189769c512c843184a8a5892ac779a6bd90"
},
{
"type": "WEB",
"url": "https://git.kernel.org/stable/c/d7251c8d3f7cea76543abac6cf4ed15582c10846"
}
],
"schema_version": "1.4.0",
"severity": []
}
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.