CVE-2021-47092

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> KVM: VMX: Always clear vmx-&gt;fail on emulation_required<br /> <br /> Revert a relatively recent change that set vmx-&gt;fail if the vCPU is in L2<br /> and emulation_required is true, as that behavior is completely bogus.<br /> Setting vmx-&gt;fail and synthesizing a VM-Exit is contradictory and wrong:<br /> <br /> (a) it&amp;#39;s impossible to have both a VM-Fail and VM-Exit<br /> (b) vmcs.EXIT_REASON is not modified on VM-Fail<br /> (c) emulation_required refers to guest state and guest state checks are<br /> always VM-Exits, not VM-Fails.<br /> <br /> For KVM specifically, emulation_required is handled before nested exits<br /> in __vmx_handle_exit(), thus setting vmx-&gt;fail has no immediate effect,<br /> i.e. KVM calls into handle_invalid_guest_state() and vmx-&gt;fail is ignored.<br /> Setting vmx-&gt;fail can ultimately result in a WARN in nested_vmx_vmexit()<br /> firing when tearing down the VM as KVM never expects vmx-&gt;fail to be set<br /> when L2 is active, KVM always reflects those errors into L1.<br /> <br /> ------------[ cut here ]------------<br /> WARNING: CPU: 0 PID: 21158 at arch/x86/kvm/vmx/nested.c:4548<br /> nested_vmx_vmexit+0x16bd/0x17e0<br /> arch/x86/kvm/vmx/nested.c:4547<br /> Modules linked in:<br /> CPU: 0 PID: 21158 Comm: syz-executor.1 Not tainted 5.16.0-rc3-syzkaller #0<br /> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011<br /> RIP: 0010:nested_vmx_vmexit+0x16bd/0x17e0 arch/x86/kvm/vmx/nested.c:4547<br /> Code: 0b e9 2e f8 ff ff e8 57 b3 5d 00 0f 0b e9 00 f1 ff ff 89 e9 80<br /> Call Trace:<br /> vmx_leave_nested arch/x86/kvm/vmx/nested.c:6220 [inline]<br /> nested_vmx_free_vcpu+0x83/0xc0 arch/x86/kvm/vmx/nested.c:330<br /> vmx_free_vcpu+0x11f/0x2a0 arch/x86/kvm/vmx/vmx.c:6799<br /> kvm_arch_vcpu_destroy+0x6b/0x240 arch/x86/kvm/x86.c:10989<br /> kvm_vcpu_destroy+0x29/0x90 arch/x86/kvm/../../../virt/kvm/kvm_main.c:441<br /> kvm_free_vcpus arch/x86/kvm/x86.c:11426 [inline]<br /> kvm_arch_destroy_vm+0x3ef/0x6b0 arch/x86/kvm/x86.c:11545<br /> kvm_destroy_vm arch/x86/kvm/../../../virt/kvm/kvm_main.c:1189 [inline]<br /> kvm_put_kvm+0x751/0xe40 arch/x86/kvm/../../../virt/kvm/kvm_main.c:1220<br /> kvm_vcpu_release+0x53/0x60 arch/x86/kvm/../../../virt/kvm/kvm_main.c:3489<br /> __fput+0x3fc/0x870 fs/file_table.c:280<br /> task_work_run+0x146/0x1c0 kernel/task_work.c:164<br /> exit_task_work include/linux/task_work.h:32 [inline]<br /> do_exit+0x705/0x24f0 kernel/exit.c:832<br /> do_group_exit+0x168/0x2d0 kernel/exit.c:929<br /> get_signal+0x1740/0x2120 kernel/signal.c:2852<br /> arch_do_signal_or_restart+0x9c/0x730 arch/x86/kernel/signal.c:868<br /> handle_signal_work kernel/entry/common.c:148 [inline]<br /> exit_to_user_mode_loop kernel/entry/common.c:172 [inline]<br /> exit_to_user_mode_prepare+0x191/0x220 kernel/entry/common.c:207<br /> __syscall_exit_to_user_mode_work kernel/entry/common.c:289 [inline]<br /> syscall_exit_to_user_mode+0x2e/0x70 kernel/entry/common.c:300<br /> do_syscall_64+0x53/0xd0 arch/x86/entry/common.c:86<br /> entry_SYSCALL_64_after_hwframe+0x44/0xae