{ "dataType": "CVE_RECORD", "dataVersion": "5.2", "cveMetadata": { "cveId": "CVE-2023-54060", "assignerOrgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67", "state": "PUBLISHED", "assignerShortName": "Linux", "dateReserved": "2025-12-24T12:21:05.091Z", "datePublished": "2025-12-24T12:23:07.276Z", "dateUpdated": "2026-08-05T09:16:49.139Z" }, "containers": { "cna": { "providerMetadata": { "orgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67", "shortName": "Linux", "dateUpdated": "2026-08-05T09:16:49.139Z" }, "descriptions": [ { "lang": "en", "value": "In the Linux kernel, the following vulnerability has been resolved:\n\niommufd: Set end correctly when doing batch carry\n\nEven though the test suite covers this it somehow became obscured that\nthis wasn't working.\n\nThe test iommufd_ioas.mock_domain.access_domain_destory would blow up\nrarely.\n\nend should be set to 1 because this just pushed an item, the carry, to the\npfns list.\n\nSometimes the test would blow up with:\n\n BUG: kernel NULL pointer dereference, address: 0000000000000000\n #PF: supervisor read access in kernel mode\n #PF: error_code(0x0000) - not-present page\n PGD 0 P4D 0\n Oops: 0000 [#1] SMP\n CPU: 5 PID: 584 Comm: iommufd Not tainted 6.5.0-rc1-dirty #1236\n Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014\n RIP: 0010:batch_unpin+0xa2/0x100 [iommufd]\n Code: 17 48 81 fe ff ff 07 00 77 70 48 8b 15 b7 be 97 e2 48 85 d2 74 14 48 8b 14 fa 48 85 d2 74 0b 40 0f b6 f6 48 c1 e6 04 48 01 f2 <48> 8b 3a 48 c1 e0 06 89 ca 48 89 de 48 83 e7 f0 48 01 c7 e8 96 dc\n RSP: 0018:ffffc90001677a58 EFLAGS: 00010246\n RAX: 00007f7e2646f000 RBX: 0000000000000000 RCX: 0000000000000001\n RDX: 0000000000000000 RSI: 00000000fefc4c8d RDI: 0000000000fefc4c\n RBP: ffffc90001677a80 R08: 0000000000000048 R09: 0000000000000200\n R10: 0000000000030b98 R11: ffffffff81f3bb40 R12: 0000000000000001\n R13: ffff888101f75800 R14: ffffc90001677ad0 R15: 00000000000001fe\n FS: 00007f9323679740(0000) GS:ffff8881ba540000(0000) knlGS:0000000000000000\n CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033\n CR2: 0000000000000000 CR3: 0000000105ede003 CR4: 00000000003706a0\n DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000\n DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400\n Call Trace:\n \n ? show_regs+0x5c/0x70\n ? __die+0x1f/0x60\n ? page_fault_oops+0x15d/0x440\n ? lock_release+0xbc/0x240\n ? exc_page_fault+0x4a4/0x970\n ? asm_exc_page_fault+0x27/0x30\n ? batch_unpin+0xa2/0x100 [iommufd]\n ? batch_unpin+0xba/0x100 [iommufd]\n __iopt_area_unfill_domain+0x198/0x430 [iommufd]\n ? __mutex_lock+0x8c/0xb80\n ? __mutex_lock+0x6aa/0xb80\n ? xa_erase+0x28/0x30\n ? iopt_table_remove_domain+0x162/0x320 [iommufd]\n ? lock_release+0xbc/0x240\n iopt_area_unfill_domain+0xd/0x10 [iommufd]\n iopt_table_remove_domain+0x195/0x320 [iommufd]\n iommufd_hw_pagetable_destroy+0xb3/0x110 [iommufd]\n iommufd_object_destroy_user+0x8e/0xf0 [iommufd]\n iommufd_device_detach+0xc5/0x140 [iommufd]\n iommufd_selftest_destroy+0x1f/0x70 [iommufd]\n iommufd_object_destroy_user+0x8e/0xf0 [iommufd]\n iommufd_destroy+0x3a/0x50 [iommufd]\n iommufd_fops_ioctl+0xfb/0x170 [iommufd]\n __x64_sys_ioctl+0x40d/0x9a0\n do_syscall_64+0x3c/0x80\n entry_SYSCALL_64_after_hwframe+0x46/0xb0" } ], "metrics": [ { "cvssV3_1": { "version": "3.1", "vectorString": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H", "baseScore": 8.8, "baseSeverity": "HIGH" }, "scenarios": [ { "lang": "en", "value": "AV:L - The vulnerable code is reached through ioctls on the /dev/iommu (or /dev/vfio/vfio compat) character device — IOMMU_IOAS_MAP/COPY/UNMAP and object destroy paths, as shown by the `__x64_sys_ioctl -> iommufd_fops_ioctl` call trace. No network or physical access is involved.\nAC:L - The attacker fully controls every precondition: hugepage-backed source memory for the contiguous PFN run, the overlapping area layout via IOMMU_IOAS_COPY, and the exact unmap offset that lands the hole/used boundary mid-run. Once the layout is arranged the `end=0` desynchronization and the resulting over-walk in batch_unpin() are deterministic; only the severity of the resulting garbage varies.\nPR:L - iommufd_fops_open() and the ioctl dispatcher perform no capability check whatsoever — access is gated only by permissions on /dev/iommu, which are routinely granted to unprivileged service accounts (qemu/libvirt for device passthrough, DPDK/SPDK users). An unprivileged local user with that device access can drive the entire path.\nUI:N - The full sequence — map, copy, attach, unmap/destroy — is performed by the attacker's own process with no victim action required.\nS:C - This breaks the invariant the code explicitly documents (\"For security we must not unpin something that is still DMA mapped\"): pages still mapped into a live IOMMU domain get their refcount dropped, so they can be freed and reallocated while an assigned device — programmed by a guest VM or by VFIO userspace — retains DMA access to them. That is an IOMMU/DMA boundary bypass reaching resources beyond iommufd's security authority.\nC:H - batch_unpin() walks past `end` with no bound check, reading adjacent kernel stack (the on-stack `u64 backup[32]` array) as PFNs, and the resulting refcount underflow frees pages that remain IOMMU-mapped and are then reallocated to arbitrary consumers — giving a device or attacker read access to reused kernel/user memory.\nI:H - The over-walk performs unpin_user_page_range_dirty_lock() and set_page_dirty on attacker-influenced wild struct page pointers, yielding an arbitrary refcount-decrement primitive plus premature page free (use-after-free) — a well-known route to heap grooming, arbitrary write and privilege escalation. The inflated index also drives iommu_unmap() over IOVA ranges beyond the intended area.\nA:H - The commit message documents a reproducible NULL pointer dereference oops in batch_unpin+0xa2 during iommufd_hw_pagetable_destroy, on a non-failable destroy path that holds pages->mutex, so the crash also wedges other iommufd users. Any use-after-free from the refcount underflow crashes the kernel as well." } ] } ], "affected": [ { "product": "Linux", "vendor": "Linux", "defaultStatus": "unaffected", "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git", "programFiles": [ "drivers/iommu/iommufd/pages.c" ], "versions": [ { "version": "f394576eb11dbcd3a740fa41e577b97f0720d26e", "lessThan": "176f36a376c417b58d19f79edfce20db9317eaa2", "status": "affected", "versionType": "git" }, { "version": "f394576eb11dbcd3a740fa41e577b97f0720d26e", "lessThan": "b7c822fa6b7701b17e139f1c562fc24135880ed4", "status": "affected", "versionType": "git" } ] }, { "product": "Linux", "vendor": "Linux", "defaultStatus": "affected", "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git", "programFiles": [ "drivers/iommu/iommufd/pages.c" ], "versions": [ { "version": "6.2", "status": "affected" }, { "version": "0", "lessThan": "6.2", "status": "unaffected", "versionType": "semver" }, { "version": "6.4.8", "lessThanOrEqual": "6.4.*", "status": "unaffected", "versionType": "semver" }, { "version": "6.5", "lessThanOrEqual": "*", "status": "unaffected", "versionType": "original_commit_for_fix" } ] } ], "cpeApplicability": [ { "nodes": [ { "operator": "OR", "negate": false, "cpeMatch": [ { "vulnerable": true, "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*", "versionStartIncluding": "6.2", "versionEndExcluding": "6.4.8" }, { "vulnerable": true, "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*", "versionStartIncluding": "6.2", "versionEndExcluding": "6.5" } ] } ] } ], "references": [ { "url": "https://git.kernel.org/stable/c/176f36a376c417b58d19f79edfce20db9317eaa2" }, { "url": "https://git.kernel.org/stable/c/b7c822fa6b7701b17e139f1c562fc24135880ed4" } ], "title": "iommufd: Set end correctly when doing batch carry", "x_generator": { "engine": "bippy-1.2.0" } } } }