{"api_version":"1","generated_at":"2026-11-20T07:30:02+00:00","cve":"CVE-2026-99710","urls":{"html":"https://cve.report/CVE-2026-99710","api":"https://cve.report/api/cve/CVE-2026-99710.json","docs":"https://cve.report/api","cve_org":"https://www.cve.org/CVERecord?id=CVE-2026-99710","nvd":"https://nvd.nist.gov/vuln/detail/CVE-2026-99710"},"summary":{"title":"KVM: x86/mmu: Cap number of parent PTEs per SP at 1000 (the arbitrary rmap limit)","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nKVM: x86/mmu: Cap number of parent PTEs per SP at 1000 (the arbitrary rmap limit)\n\nWhen potentially reusing a shadow page, zap the shadow page if its number\nof parent PTEs exceeds KVM's (mostly) arbitrary rmap recycling threshold.\nLetting the guest create a (mostly) unbounded number of parent PTEs allows\nthe guest to coerce KVM into creating lists containing millions of PTEs,\neffectively bounded only by the maximum number of shadow pages.\n\nBecause PTE lists are simple singly linked lists, removal is O(n).  Even\nworse, if a list is constructed just so, teardown can be O(n^2), e.g. if\nSPTEs are built such that zapping child SPTEs always zaps the last entry\nin the list.  To cap things off, all paths that walk PTE lists do so\nwhile holding mmu_lock, without conditionally rescheduling.\n\nUsing a doubly linked list would not address the issue, as removal would\nremain O(n) since KVM still wouldn't know where in the list to remove from.\nRemoval could be made O(1) by storing a per-SPTE pointer, but that would\nlargely defeat the purpose of reusing SPs, and would penalize well-behaved\nguests, especially for nested TDP where aliasing page tables is uncommon.\n\nSo, cap the maximum number of parent PTEs at the rmap limit.  In practice,\nwell-behaved guests will likely never run afoul of the limit.  E.g. when\nrunning Linux as L1 with TDP, the number of parent PTEs for a single SP\ntops out at under 200 (though undoubtedly obscure workloads can push it\nhigher).  Deliberately use a \"greater than or equal\" as opposed to rmap's\nstrictly \"greater than\", as the rmap code checks the threshold immediately\n*after* inserting into the list, i.e. the limit is 1000 for rmaps, KVM just\nvery briefly blows past it.\n\nApply the check to both indirect and direct SPs, because while a fully\ndirect MMU naturally limits the number of parent PTEs to '1', direct SPs in\nindirect MMUs have no such constraint.  E.g. if the guest maps a huge PTE\ninto many locations in its page tables, and KVM is backing the hugepage\nwith a direct SP, i.e. with a series of smaller pages, then KVM could end\nup with a direct SP with many parents.\n\nNote, shadow_mmu_try_split_huge_pages() always restarts the rmap walk or\nbails entirely if it gets to the point where it tries to grab a SP, i.e.\nthere is *no* pre-existing bug that would be tripped by zapping a SP and\nthus rmaps when getting a SP.","state":"PUBLISHED","assigner":"Linux","published_at":"2026-11-20 06:18:07","updated_at":"2026-11-20 06:18:07"},"problem_types":[],"metrics":[],"references":[{"url":"https://git.kernel.org/stable/c/e4558c3d7198ed9a0adc3304681f561f4e4d80e1","name":"https://git.kernel.org/stable/c/e4558c3d7198ed9a0adc3304681f561f4e4d80e1","refsource":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","tags":[],"title":"","mime":"","httpstatus":"","archivestatus":"0"},{"url":"https://git.kernel.org/stable/c/edf87e07376054eedcd5306e0794a26e3004d97b","name":"https://git.kernel.org/stable/c/edf87e07376054eedcd5306e0794a26e3004d97b","refsource":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","tags":[],"title":"","mime":"","httpstatus":"","archivestatus":"0"},{"url":"https://git.kernel.org/stable/c/72238066aacc400df69cef4b602f083ae4737762","name":"https://git.kernel.org/stable/c/72238066aacc400df69cef4b602f083ae4737762","refsource":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","tags":[],"title":"","mime":"","httpstatus":"","archivestatus":"0"},{"url":"https://git.kernel.org/stable/c/e0752f3e543c1b67fc9115251ae6b8f151ba9f04","name":"https://git.kernel.org/stable/c/e0752f3e543c1b67fc9115251ae6b8f151ba9f04","refsource":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","tags":[],"title":"","mime":"","httpstatus":"","archivestatus":"0"},{"url":"https://www.cve.org/CVERecord?id=CVE-2026-99710","name":"CVE Program record","refsource":"CVE.ORG","tags":["canonical"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-99710","name":"NVD vulnerability detail","refsource":"NVD","tags":["canonical","analysis"]}],"affected":[{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected 1da177e4c3f41524e886b7f1b8a0c1fc7321cac2 edf87e07376054eedcd5306e0794a26e3004d97b git","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected 1da177e4c3f41524e886b7f1b8a0c1fc7321cac2 e0752f3e543c1b67fc9115251ae6b8f151ba9f04 git","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected 1da177e4c3f41524e886b7f1b8a0c1fc7321cac2 72238066aacc400df69cef4b602f083ae4737762 git","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected 1da177e4c3f41524e886b7f1b8a0c1fc7321cac2 e4558c3d7198ed9a0adc3304681f561f4e4d80e1 git","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected 6.12.118 semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected 6.18.64 semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected 7.3.5 semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 6.12.118 6.12.* semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 6.18.64 6.18.* semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 7.3.5 7.3.* semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 7.4-rc2 * original_commit_for_fix","platforms":[]}],"timeline":[],"solutions":[],"workarounds":[],"exploits":[],"credits":[],"nvd_cpes":[],"vendor_comments":[],"enrichments":{"kev":null,"epss":null,"legacy_qids":[]},"source_records":{"cve_program":{"containers":{"cna":{"affected":[{"defaultStatus":"unaffected","product":"Linux","programFiles":["arch/x86/kvm/mmu/mmu.c"],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","vendor":"Linux","versions":[{"lessThan":"edf87e07376054eedcd5306e0794a26e3004d97b","status":"affected","version":"1da177e4c3f41524e886b7f1b8a0c1fc7321cac2","versionType":"git"},{"lessThan":"e0752f3e543c1b67fc9115251ae6b8f151ba9f04","status":"affected","version":"1da177e4c3f41524e886b7f1b8a0c1fc7321cac2","versionType":"git"},{"lessThan":"72238066aacc400df69cef4b602f083ae4737762","status":"affected","version":"1da177e4c3f41524e886b7f1b8a0c1fc7321cac2","versionType":"git"},{"lessThan":"e4558c3d7198ed9a0adc3304681f561f4e4d80e1","status":"affected","version":"1da177e4c3f41524e886b7f1b8a0c1fc7321cac2","versionType":"git"},{"lessThan":"6.12.118","status":"affected","version":"0","versionType":"semver"},{"lessThan":"6.18.64","status":"affected","version":"0","versionType":"semver"},{"lessThan":"7.3.5","status":"affected","version":"0","versionType":"semver"}]},{"defaultStatus":"affected","product":"Linux","programFiles":["arch/x86/kvm/mmu/mmu.c"],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","vendor":"Linux","versions":[{"lessThanOrEqual":"6.12.*","status":"unaffected","version":"6.12.118","versionType":"semver"},{"lessThanOrEqual":"6.18.*","status":"unaffected","version":"6.18.64","versionType":"semver"},{"lessThanOrEqual":"7.3.*","status":"unaffected","version":"7.3.5","versionType":"semver"},{"lessThanOrEqual":"*","status":"unaffected","version":"7.4-rc2","versionType":"original_commit_for_fix"}]}],"cpeApplicability":[{"nodes":[{"cpeMatch":[{"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionEndExcluding":"6.12.118","vulnerable":true},{"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionEndExcluding":"6.18.64","vulnerable":true},{"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionEndExcluding":"7.3.5","vulnerable":true},{"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionEndExcluding":"7.4-rc2","vulnerable":true}],"negate":false,"operator":"OR"}]}],"descriptions":[{"lang":"en","value":"In the Linux kernel, the following vulnerability has been resolved:\n\nKVM: x86/mmu: Cap number of parent PTEs per SP at 1000 (the arbitrary rmap limit)\n\nWhen potentially reusing a shadow page, zap the shadow page if its number\nof parent PTEs exceeds KVM's (mostly) arbitrary rmap recycling threshold.\nLetting the guest create a (mostly) unbounded number of parent PTEs allows\nthe guest to coerce KVM into creating lists containing millions of PTEs,\neffectively bounded only by the maximum number of shadow pages.\n\nBecause PTE lists are simple singly linked lists, removal is O(n).  Even\nworse, if a list is constructed just so, teardown can be O(n^2), e.g. if\nSPTEs are built such that zapping child SPTEs always zaps the last entry\nin the list.  To cap things off, all paths that walk PTE lists do so\nwhile holding mmu_lock, without conditionally rescheduling.\n\nUsing a doubly linked list would not address the issue, as removal would\nremain O(n) since KVM still wouldn't know where in the list to remove from.\nRemoval could be made O(1) by storing a per-SPTE pointer, but that would\nlargely defeat the purpose of reusing SPs, and would penalize well-behaved\nguests, especially for nested TDP where aliasing page tables is uncommon.\n\nSo, cap the maximum number of parent PTEs at the rmap limit.  In practice,\nwell-behaved guests will likely never run afoul of the limit.  E.g. when\nrunning Linux as L1 with TDP, the number of parent PTEs for a single SP\ntops out at under 200 (though undoubtedly obscure workloads can push it\nhigher).  Deliberately use a \"greater than or equal\" as opposed to rmap's\nstrictly \"greater than\", as the rmap code checks the threshold immediately\n*after* inserting into the list, i.e. the limit is 1000 for rmaps, KVM just\nvery briefly blows past it.\n\nApply the check to both indirect and direct SPs, because while a fully\ndirect MMU naturally limits the number of parent PTEs to '1', direct SPs in\nindirect MMUs have no such constraint.  E.g. if the guest maps a huge PTE\ninto many locations in its page tables, and KVM is backing the hugepage\nwith a direct SP, i.e. with a series of smaller pages, then KVM could end\nup with a direct SP with many parents.\n\nNote, shadow_mmu_try_split_huge_pages() always restarts the rmap walk or\nbails entirely if it gets to the point where it tries to grab a SP, i.e.\nthere is *no* pre-existing bug that would be tripped by zapping a SP and\nthus rmaps when getting a SP."}],"providerMetadata":{"dateUpdated":"2026-11-20T05:50:50.082Z","orgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","shortName":"Linux"},"references":[{"url":"https://git.kernel.org/stable/c/edf87e07376054eedcd5306e0794a26e3004d97b"},{"url":"https://git.kernel.org/stable/c/e0752f3e543c1b67fc9115251ae6b8f151ba9f04"},{"url":"https://git.kernel.org/stable/c/72238066aacc400df69cef4b602f083ae4737762"},{"url":"https://git.kernel.org/stable/c/e4558c3d7198ed9a0adc3304681f561f4e4d80e1"}],"title":"KVM: x86/mmu: Cap number of parent PTEs per SP at 1000 (the arbitrary rmap limit)","x_generator":{"engine":"bippy-1.2.0"}}},"cveMetadata":{"assignerOrgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","assignerShortName":"Linux","cveId":"CVE-2026-99710","datePublished":"2026-11-20T05:50:50.082Z","dateReserved":"2026-09-25T10:25:14.430Z","dateUpdated":"2026-11-20T05:50:50.082Z","state":"PUBLISHED"},"dataType":"CVE_RECORD","dataVersion":"5.2"},"nvd":{"publishedDate":"2026-11-20 06:18:07","lastModifiedDate":"2026-11-20 06:18:07","problem_types":[],"metrics":[],"configurations":[]},"legacy_mitre":{"record":{"CveYear":"2026","CveId":"99710","Ordinal":"1","Title":"KVM: x86/mmu: Cap number of parent PTEs per SP at 1000 (the arbi","CVE":"CVE-2026-99710","Year":"2026"},"notes":[{"CveYear":"2026","CveId":"99710","Ordinal":"1","NoteData":"In the Linux kernel, the following vulnerability has been resolved:\n\nKVM: x86/mmu: Cap number of parent PTEs per SP at 1000 (the arbitrary rmap limit)\n\nWhen potentially reusing a shadow page, zap the shadow page if its number\nof parent PTEs exceeds KVM's (mostly) arbitrary rmap recycling threshold.\nLetting the guest create a (mostly) unbounded number of parent PTEs allows\nthe guest to coerce KVM into creating lists containing millions of PTEs,\neffectively bounded only by the maximum number of shadow pages.\n\nBecause PTE lists are simple singly linked lists, removal is O(n).  Even\nworse, if a list is constructed just so, teardown can be O(n^2), e.g. if\nSPTEs are built such that zapping child SPTEs always zaps the last entry\nin the list.  To cap things off, all paths that walk PTE lists do so\nwhile holding mmu_lock, without conditionally rescheduling.\n\nUsing a doubly linked list would not address the issue, as removal would\nremain O(n) since KVM still wouldn't know where in the list to remove from.\nRemoval could be made O(1) by storing a per-SPTE pointer, but that would\nlargely defeat the purpose of reusing SPs, and would penalize well-behaved\nguests, especially for nested TDP where aliasing page tables is uncommon.\n\nSo, cap the maximum number of parent PTEs at the rmap limit.  In practice,\nwell-behaved guests will likely never run afoul of the limit.  E.g. when\nrunning Linux as L1 with TDP, the number of parent PTEs for a single SP\ntops out at under 200 (though undoubtedly obscure workloads can push it\nhigher).  Deliberately use a \"greater than or equal\" as opposed to rmap's\nstrictly \"greater than\", as the rmap code checks the threshold immediately\n*after* inserting into the list, i.e. the limit is 1000 for rmaps, KVM just\nvery briefly blows past it.\n\nApply the check to both indirect and direct SPs, because while a fully\ndirect MMU naturally limits the number of parent PTEs to '1', direct SPs in\nindirect MMUs have no such constraint.  E.g. if the guest maps a huge PTE\ninto many locations in its page tables, and KVM is backing the hugepage\nwith a direct SP, i.e. with a series of smaller pages, then KVM could end\nup with a direct SP with many parents.\n\nNote, shadow_mmu_try_split_huge_pages() always restarts the rmap walk or\nbails entirely if it gets to the point where it tries to grab a SP, i.e.\nthere is *no* pre-existing bug that would be tripped by zapping a SP and\nthus rmaps when getting a SP.","Type":"Description","Title":"KVM: x86/mmu: Cap number of parent PTEs per SP at 1000 (the arbi"}]}}}