{
  "schema_version": "1.6.1",
  "id": "RPI-CVE-2024-41070",
  "modified": "2026-10-09T11:39:44Z",
  "upstream": [
    "CVE-2024-41070"
  ],
  "published": "2024-07-29T14:58:20Z",
  "summary": "KVM: PPC: Book3S HV: Prevent UAF in kvm_spapr_tce_attach_iommu_group()",
  "details": "In the Linux kernel, the following vulnerability has been resolved:\n\nKVM: PPC: Book3S HV: Prevent UAF in kvm_spapr_tce_attach_iommu_group()\n\nAl reported a possible use-after-free (UAF) in kvm_spapr_tce_attach_iommu_group().\n\nIt looks up `stt` from tablefd, but then continues to use it after doing\nfdput() on the returned fd. After the fdput() the tablefd is free to be\nclosed by another thread. The close calls kvm_spapr_tce_release() and\nthen release_spapr_tce_table() (via call_rcu()) which frees `stt`.\n\nAlthough there are calls to rcu_read_lock() in\nkvm_spapr_tce_attach_iommu_group() they are not sufficient to prevent\nthe UAF, because `stt` is used outside the locked regions.\n\nWith an artifcial delay after the fdput() and a userspace program which\ntriggers the race, KASAN detects the UAF:\n\n  BUG: KASAN: slab-use-after-free in kvm_spapr_tce_attach_iommu_group+0x298/0x720 [kvm]\n  Read of size 4 at addr c000200027552c30 by task kvm-vfio/2505\n  CPU: 54 PID: 2505 Comm: kvm-vfio Not tainted 6.10.0-rc3-next-20240612-dirty #1\n  Hardware name: 8335-GTH POWER9 0x4e1202 opal:skiboot-v6.5.3-35-g1851b2a06 PowerNV\n  Call Trace:\n    dump_stack_lvl+0xb4/0x108 (unreliable)\n    print_report+0x2b4/0x6ec\n    kasan_report+0x118/0x2b0\n    __asan_load4+0xb8/0xd0\n    kvm_spapr_tce_attach_iommu_group+0x298/0x720 [kvm]\n    kvm_vfio_set_attr+0x524/0xac0 [kvm]\n    kvm_device_ioctl+0x144/0x240 [kvm]\n    sys_ioctl+0x62c/0x1810\n    system_call_exception+0x190/0x440\n    system_call_vectored_common+0x15c/0x2ec\n  ...\n  Freed by task 0:\n   ...\n   kfree+0xec/0x3e0\n   release_spapr_tce_table+0xd4/0x11c [kvm]\n   rcu_core+0x568/0x16a0\n   handle_softirqs+0x23c/0x920\n   do_softirq_own_stack+0x6c/0x90\n   do_softirq_own_stack+0x58/0x90\n   __irq_exit_rcu+0x218/0x2d0\n   irq_exit+0x30/0x80\n   arch_local_irq_restore+0x128/0x230\n   arch_local_irq_enable+0x1c/0x30\n   cpuidle_enter_state+0x134/0x5cc\n   cpuidle_enter+0x6c/0xb0\n   call_cpuidle+0x7c/0x100\n   do_idle+0x394/0x410\n   cpu_startup_entry+0x60/0x70\n   start_secondary+0x3fc/0x410\n   start_secondary_prolog+0x10/0x14\n\nFix it by delaying the fdput() until `stt` is no longer in use, which\nis effectively the entire function. To keep the patch minimal add a call\nto fdput() at each of the existing return paths. Future work can convert\nthe function to goto or __cleanup style cleanup.\n\nWith the fix in place the test case no longer triggers the UAF.",
  "severity": [
    {
      "type": "CVSS_V3",
      "score": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H"
    }
  ],
  "affected": [
    {
      "package": {
        "ecosystem": "Raspberry Pi OS:12",
        "name": "linux"
      },
      "versions": [
        "1:6.1.47-1+rpt4"
      ],
      "ecosystem_specific": {
        "binaries": [
          {
            "binary_name": "linux-image-6.1.0-rpi3-rpi-2712",
            "binary_version": "1:6.1.47-1+rpt4"
          },
          {
            "binary_name": "linux-image-6.1.0-rpi3-rpi-v6",
            "binary_version": "1:6.1.47-1+rpt4"
          },
          {
            "binary_name": "linux-image-6.1.0-rpi3-rpi-v7",
            "binary_version": "1:6.1.47-1+rpt4"
          },
          {
            "binary_name": "linux-image-6.1.0-rpi3-rpi-v7l",
            "binary_version": "1:6.1.47-1+rpt4"
          },
          {
            "binary_name": "linux-image-6.1.0-rpi3-rpi-v8",
            "binary_version": "1:6.1.47-1+rpt4"
          }
        ]
      },
      "database_specific": {
        "per_version": {
          "1:6.1.47-1+rpt4": {
            "linux_commit": "655fc658a15ae7a6f37103754adb39ba52a9a14e",
            "upstream_version": "6.1.47",
            "assessment": "commit ancestry (strak rule)",
            "upstream_fixed": "6.1.101",
            "compiled_filter": false
          }
        }
      }
    }
  ],
  "references": [
    {
      "type": "ADVISORY",
      "url": "https://www.cve.org/CVERecord?id=CVE-2024-41070"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/be847bb20c809de8ac124431b556f244400b0491"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/4cdf6926f443c84f680213c7aafbe6f91a5fcbc0"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/b26c8c85463ef27a522d24fcd05651f0bb039e47"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/5f856023971f97fff74cfaf21b48ec320147b50a"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/82c7a4cf14aa866f8f7f09e662b02eddc49ee0bf"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/9975f93c760a32453d7639cf6fcf3f73b4e71ffe"
    },
    {
      "type": "FIX",
      "url": "https://git.kernel.org/stable/c/a986fa57fd81a1430e00b3c6cf8a325d6f894a63"
    }
  ],
  "database_specific": {
    "source": "https://git.kernel.org/pub/scm/linux/security/vulns.git",
    "upstream_modified": "2026-08-05T11:52:36Z",
    "severity": {
      "source": "kernel CNA",
      "cvss_v3_score": 7.8,
      "cvss_v3_severity": "High",
      "scenario": "AV:L - The vulnerable code is reached only through `ioctl(KVM_SET_DEVICE_ATTR, KVM_DEV_VFIO_GROUP_SET_SPAPR_TCE)` on a KVM VFIO device fd, requiring local access to /dev/kvm and /dev/vfio on a POWER Book3S HV host. No remote or adjacent-network input reaches this path.\nAC:L - The attacker owns both sides of the race \u2014 one thread issues the ioctl while another calls `close(tablefd)` \u2014 and the window between `fdput()` and the uses of `stt` contains a sleepable `kzalloc(GFP_KERNEL)` that can be stretched arbitrarily under attacker-induced memory pressure. Failed attempts merely return an errno, so the race can be retried indefinitely, and the attacker also chooses the slab bucket via `KVM_CREATE_SPAPR_TCE_64`'s `size` argument.\nPR:L - No `capable()` or `CAP_*` check exists anywhere along the path in `book3s_64_vio.c` or `virt/kvm/vfio.c`; the attacker only needs file access to /dev/kvm and a VFIO group node, which libvirt routinely grants to the unprivileged QEMU user in the standard PPC64 device-passthrough deployment. A compromised VMM process that has escaped a guest into QEMU holds exactly these descriptors, so real root is not required.\nUI:N - Both the ioctl and the racing `close()` are performed entirely by the attacker's own threads. No administrator or victim action is involved.\nS:U - The attacker is a host-local userspace process and the corrupted memory is host kernel heap, so the vulnerable and impacted components share one security authority. A guest cannot reach `kvm_vfio_set_attr`, so this is not a guest-to-host escape.\nC:H - The freed `stt` is read after free (`page_shift`, `offset`, `size`), and the `next->prev = new` write derived from freed-memory contents gives an arbitrary kernel-write primitive that is readily converted into arbitrary kernel-memory disclosure. Attacker-controlled slab-bucket selection makes the reallocation reliable enough to build a read primitive.\nI:H - `list_add_rcu(&stit->next, &stt->iommu_tables)` writes into the freed object and additionally performs `next->prev = new` where `next` is loaded from attacker-reallocated memory, yielding a write of a kernel heap pointer to an arbitrary address. With `__list_add_valid()` a no-op on default configs, this is a classic path to control-flow hijack and privilege escalation.\nA:H - The slab use-after-free corrupts the kernel heap and the wild `next->prev` store into an unrelated address readily produces an oops or panic; on `CONFIG_LIST_HARDENED` kernels the corrupted list triggers a BUG. The attacker can repeat the trigger at will to keep the host down."
    }
  }
}