CVE-2025-38455

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> KVM: SVM: Reject SEV{-ES} intra host migration if vCPU creation is in-flight<br /> <br /> Reject migration of SEV{-ES} state if either the source or destination VM<br /> is actively creating a vCPU, i.e. if kvm_vm_ioctl_create_vcpu() is in the<br /> section between incrementing created_vcpus and online_vcpus. The bulk of<br /> vCPU creation runs _outside_ of kvm-&gt;lock to allow creating multiple vCPUs<br /> in parallel, and so sev_info.es_active can get toggled from false=&gt;true in<br /> the destination VM after (or during) svm_vcpu_create(), resulting in an<br /> SEV{-ES} VM effectively having a non-SEV{-ES} vCPU.<br /> <br /> The issue manifests most visibly as a crash when trying to free a vCPU&amp;#39;s<br /> NULL VMSA page in an SEV-ES VM, but any number of things can go wrong.<br /> <br /> BUG: unable to handle page fault for address: ffffebde00000000<br /> #PF: supervisor read access in kernel mode<br /> #PF: error_code(0x0000) - not-present page<br /> PGD 0 P4D 0<br /> Oops: Oops: 0000 [#1] SMP KASAN NOPTI<br /> CPU: 227 UID: 0 PID: 64063 Comm: syz.5.60023 Tainted: G U O 6.15.0-smp-DEV #2 NONE<br /> Tainted: [U]=USER, [O]=OOT_MODULE<br /> Hardware name: Google, Inc. Arcadia_IT_80/Arcadia_IT_80, BIOS 12.52.0-0 10/28/2024<br /> RIP: 0010:constant_test_bit arch/x86/include/asm/bitops.h:206 [inline]<br /> RIP: 0010:arch_test_bit arch/x86/include/asm/bitops.h:238 [inline]<br /> RIP: 0010:_test_bit include/asm-generic/bitops/instrumented-non-atomic.h:142 [inline]<br /> RIP: 0010:PageHead include/linux/page-flags.h:866 [inline]<br /> RIP: 0010:___free_pages+0x3e/0x120 mm/page_alloc.c:5067<br /> Code: f7 06 40 00 00 00 75 05 45 31 ff eb 0c 66 90 4c 89 f0 4c 39 f0<br /> RSP: 0018:ffff8984551978d0 EFLAGS: 00010246<br /> RAX: 0000777f80000001 RBX: 0000000000000000 RCX: ffffffff918aeb98<br /> RDX: 0000000000000000 RSI: 0000000000000008 RDI: ffffebde00000000<br /> RBP: 0000000000000000 R08: ffffebde00000007 R09: 1ffffd7bc0000000<br /> R10: dffffc0000000000 R11: fffff97bc0000001 R12: dffffc0000000000<br /> R13: ffff8983e19751a8 R14: ffffebde00000000 R15: 1ffffd7bc0000000<br /> FS: 0000000000000000(0000) GS:ffff89ee661d3000(0000) knlGS:0000000000000000<br /> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033<br /> CR2: ffffebde00000000 CR3: 000000793ceaa000 CR4: 0000000000350ef0<br /> DR0: 0000000000000000 DR1: 0000000000000b5f DR2: 0000000000000000<br /> DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400<br /> Call Trace:<br /> <br /> sev_free_vcpu+0x413/0x630 arch/x86/kvm/svm/sev.c:3169<br /> svm_vcpu_free+0x13a/0x2a0 arch/x86/kvm/svm/svm.c:1515<br /> kvm_arch_vcpu_destroy+0x6a/0x1d0 arch/x86/kvm/x86.c:12396<br /> kvm_vcpu_destroy virt/kvm/kvm_main.c:470 [inline]<br /> kvm_destroy_vcpus+0xd1/0x300 virt/kvm/kvm_main.c:490<br /> kvm_arch_destroy_vm+0x636/0x820 arch/x86/kvm/x86.c:12895<br /> kvm_put_kvm+0xb8e/0xfb0 virt/kvm/kvm_main.c:1310<br /> kvm_vm_release+0x48/0x60 virt/kvm/kvm_main.c:1369<br /> __fput+0x3e4/0x9e0 fs/file_table.c:465<br /> task_work_run+0x1a9/0x220 kernel/task_work.c:227<br /> exit_task_work include/linux/task_work.h:40 [inline]<br /> do_exit+0x7f0/0x25b0 kernel/exit.c:953<br /> do_group_exit+0x203/0x2d0 kernel/exit.c:1102<br /> get_signal+0x1357/0x1480 kernel/signal.c:3034<br /> arch_do_signal_or_restart+0x40/0x690 arch/x86/kernel/signal.c:337<br /> exit_to_user_mode_loop kernel/entry/common.c:111 [inline]<br /> exit_to_user_mode_prepare include/linux/entry-common.h:329 [inline]<br /> __syscall_exit_to_user_mode_work kernel/entry/common.c:207 [inline]<br /> syscall_exit_to_user_mode+0x67/0xb0 kernel/entry/common.c:218<br /> do_syscall_64+0x7c/0x150 arch/x86/entry/syscall_64.c:100<br /> entry_SYSCALL_64_after_hwframe+0x76/0x7e<br /> RIP: 0033:0x7f87a898e969<br /> <br /> Modules linked in: gq(O)<br /> gsmi: Log Shutdown Reason 0x03<br /> CR2: ffffebde00000000<br /> ---[ end trace 0000000000000000 ]---<br /> <br /> Deliberately don&amp;#39;t check for a NULL VMSA when freeing the vCPU, as crashing<br /> the host is likely desirable due to the VMSA being consumed by hardware.<br /> E.g. if KVM manages to allow VMRUN on the vCPU, hardware may read/write a<br /> bogus VMSA page. Accessing P<br /> ---truncated---