In the Linux kernel, the following vulnerability has been resolved:
s390/vfio-ap: fix stale pqap_hook pointer on error in vfio_ap_mdev_set_kvm()
In vfio_ap_mdev_set_kvm(), kvm->arch.crypto.pqap_hook is set to
&matrix_mdev->pqap_hook before the update locks are acquired and the
mdev list is checked for a conflicting assignment. If another mdev is
already attached to the same KVM instance, the function returns -EPERM
without restoring the hook pointer, leaving kvm->arch.crypto.pqap_hook
pointing at the failing matrix_mdev instead of the mdev that legitimately
owns the KVM.
Since matrix_mdev->kvm is never set on this error path,
vfio_ap_mdev_unset_kvm() will not clean up the hook when matrix_mdev
is later closed. If matrix_mdev is subsequently freed, any PQAP
instruction executed by the guest will dereference the stale pointer
through pqap_hook_rwsem, resulting in a use-after-free.
Since kvm->arch.crypto.pqap_hook is only set in the vfio_ap_mdev_set_kvm()
function and is cleared in the vfio_ap_mdev_unset_kvm() function, a check
for 'kvm->arch.crypto.pqap_hook != NULL' is all that is needed to determine
whether it belongs to another mdev. This will alleviate the need to iterate
the matrix_dev->mdev_list list to see if the kvm object is assigned to
another mdev.This was introduced in v3 to alleviate the need to take the
mdevs_lock while iterating the list; however, this did not prevent a
potential race condition.
The pqap_hook_rwsem(write) is now performed inside
get_update_locks_for_kvm(), which is updated to acquire
pqap_hook_rwsem(write) between kvm->lock and mdevs_lock. This ordering
is consistent with the PQAP intercept path, which acquires pqap_hook_rwsem
in read mode while srcu is held under vcpu->mutex, establishing the
dependency: kvm->lock -> vcpu->mutex -> srcu -> pqap_hook_rwsem(read).
The pqap_hook_rwsem is now released inside the
release_update_locks_for_kvm(), which is updated to release
pqap_hook_rwsem(write) between mdevs_lock and kvm->lock.
Additionally, kvm_put_kvm() in vfio_ap_mdev_unset_kvm() is moved
after release_update_locks_for_kvm(). Previously it was called while
kvm->lock was held; if it were ever the last reference, kvm_destroy_vm()
would run under kvm->lock, which would deadlock.
CVSS Vector: CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H
CVSS Score: 8.8
AV:L - The bug is reached by opening vfio-ap mdev fds and linking them to a KVM instance (KVM_DEV_VFIO_FILE_ADD/BIND_IOMMUFD) so vfio_ap_mdev_open_device() calls vfio_ap_mdev_set_kvm(), then by a guest PQAP(AQIC) intercept in handle_pqap(); that is a local VFIO/KVM path with no network or USB exposure.
AC:L - An attacker who can open two vfio-ap mdevs against the same KVM deterministically hits the -EPERM conflict path that overwrites kvm->arch.crypto.pqap_hook without restoring it, then frees the failing mdev and issues PQAP from the still-running guest; the attacker controls every step, with no victim race or uncontrolled layout.
PR:L - vfio_ap_mdev_set_kvm() and the PQAP intercept have no capable() gate; /dev/kvm and /dev/vfio nodes are routinely granted to an unprivileged qemu/kvm-group tenant with assigned vfio-ap mdevs, so host root is not required, and user namespaces cannot substitute for this path.
UI:N - The attacker performs the conflicting VFIO attach, tears down the failing mdev, and executes PQAP in their own guest; prior administrator creation of vfio-ap mdevs is deployment configuration, not victim interaction during exploitation.
S:C - A guest PQAP intercept on the host calls through kvm->arch.crypto.pqap_hook into a freed ap_matrix_mdev in the host vfio-ap driver, hijacking a host kernel function pointer and crossing the KVM/VFIO guest-to-host isolation boundary (VM escape).
C:H - After the failing mdev is kvfree'd, handle_pqap() does pqap_hook = *vcpu->kvm->arch.crypto.pqap_hook, reading a function pointer from the freed ap_matrix_mdev; reclaiming that object yields an arbitrary kernel-read primitive, which kernel UAF guidance scores High.
I:H - The intercept then invokes that recovered pointer as pqap_hook(vcpu); spraying the freed ap_matrix_mdev replaces the hook with an attacker-controlled address, enabling host kernel control-flow hijack and arbitrary write.
A:H - The same stale-pointer dereference and call on the freed object causes a host kernel oops or panic when the guest issues PQAP, even if the function pointer is not fully hijacked.
| Attack Vector |
Local |
Scope |
Changed |
| Attack Complexity |
Low |
Confidentiality Impact |
High |
| Privileges Required |
Low |
Integrity Impact |
High |
| User Interaction |
None |
Availability Impact |
High |
AV:L - The bug is reached by opening vfio-ap mdev fds and linking them to a KVM instance (KVM_DEV_VFIO_FILE_ADD/BIND_IOMMUFD) so vfio_ap_mdev_open_device() calls vfio_ap_mdev_set_kvm(), then by a guest PQAP(AQIC) intercept in handle_pqap(); that is a local VFIO/KVM path with no network or USB exposure.
AC:L - An attacker who can open two vfio-ap mdevs against the same KVM deterministically hits the -EPERM conflict path that overwrites kvm->arch.crypto.pqap_hook without restoring it, then frees the failing mdev and issues PQAP from the still-running guest; the attacker controls every step, with no victim race or uncontrolled layout.
PR:L - vfio_ap_mdev_set_kvm() and the PQAP intercept have no capable() gate; /dev/kvm and /dev/vfio nodes are routinely granted to an unprivileged qemu/kvm-group tenant with assigned vfio-ap mdevs, so host root is not required, and user namespaces cannot substitute for this path.
UI:N - The attacker performs the conflicting VFIO attach, tears down the failing mdev, and executes PQAP in their own guest; prior administrator creation of vfio-ap mdevs is deployment configuration, not victim interaction during exploitation.
S:C - A guest PQAP intercept on the host calls through kvm->arch.crypto.pqap_hook into a freed ap_matrix_mdev in the host vfio-ap driver, hijacking a host kernel function pointer and crossing the KVM/VFIO guest-to-host isolation boundary (VM escape).
C:H - After the failing mdev is kvfree'd, handle_pqap() does pqap_hook = *vcpu->kvm->arch.crypto.pqap_hook, reading a function pointer from the freed ap_matrix_mdev; reclaiming that object yields an arbitrary kernel-read primitive, which kernel UAF guidance scores High.
I:H - The intercept then invokes that recovered pointer as pqap_hook(vcpu); spraying the freed ap_matrix_mdev replaces the hook with an attacker-controlled address, enabling host kernel control-flow hijack and arbitrary write.
A:H - The same stale-pointer dereference and call on the freed object causes a host kernel oops or panic when the guest issues PQAP, even if the function pointer is not fully hijacked.
CVSS 3.1