KVM: x86: Grab KVM's srcu lock when setting nested state
PATCH 1/1 KVM: x86: Grab KVM's srcu lock when setting nested state
Acquire kvm->srcu for the duration of ->set_nested_state() to fix a bug where nVMX derefences ->memslots without holding ->srcu or ->slots_lock.
The other half of nested migration, ->get_nested_state(), does not need to acquire ->srcu as it is a purely a dump of internal KVM (and CPU) state to userspace.
Detected as an RCU lockdep splat that is 100% reproducible by running KVM's state_test selftest with CONFIG_PROVE_LOCKING=y. Note that the failing function, kvm_is_visible_gfn(), is only checking the validity of a gfn, it's not actually accessing guest memory (which is more or less unsupported during vmx_set_nested_state() due to incorrect MMU state), i.e. vmx_set_nested_state() itself isn't fundamentally broken. In any case, setting nested state isn't a fast path so there's no reason to go out of our way to avoid taking ->srcu.
=============================
WARNING: suspicious RCU usage
5.4.0-rc7+ #94 Not tainted
-----------------------------
include/linux/kvm_host.h:626 suspicious rcu_dereference_check() usage!
other info that might help us debug this:
rcu_scheduler_active = 2, debug_locks = 1
1 lock held by evmcs_test/10939:
#0: ffff88826ffcb800 (&vcpu->mutex){+.+.}, at: kvm_vcpu_ioctl+0x85/0x630 [kvm]
stack backtrace: CPU: 1 PID: 10939 Comm: evmcs_test Not tainted 5.4.0-rc7+ #94 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015 Call Trace: dump_stack+0x68/0x9b kvm_is_visible_gfn+0x179/0x180 [kvm] mmu_check_root+0x11/0x30 [kvm] fast_cr3_switch+0x40/0x120 [kvm] kvm_mmu_new_cr3+0x34/0x60 [kvm] nested_vmx_load_cr3+0xbd/0x1f0 [kvm_intel] nested_vmx_enter_non_root_mode+0xab8/0x1d60 [kvm_intel] vmx_set_nested_state+0x256/0x340 [kvm_intel] kvm_arch_vcpu_ioctl+0x491/0x11a0 [kvm] kvm_vcpu_ioctl+0xde/0x630 [kvm] do_vfs_ioctl+0xa2/0x6c0 ksys_ioctl+0x66/0x70 __x64_sys_ioctl+0x16/0x20 do_syscall_64+0x54/0x200 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x7f59a2b95f47
--- arch/x86/kvm/x86.c | 3 +++ 1 file changed, 3 insertions(+)
arch/x86/kvm/x86.c +3 -0
@@ -4421,6 +4421,7 @@ long kvm_arch_vcpu_ioctl(struct file *filp,case KVM_SET_NESTED_STATE : {struct kvm_nested_state __user *user_kvm_nested_state = argp;struct kvm_nested_state kvm_state; +int idx; r = -EINVAL ;if (!kvm_x86_ops->set_nested_state)
@@ -4444,7 +4445,9 @@ long kvm_arch_vcpu_ioctl(struct file *filp, && !(kvm_state.flags & KVM_STATE_NESTED_GUEST_MODE)) break; + idx =srcu_read_lock (&vcpu->kvm->srcu); r = kvm_x86_ops->set_nested_state (vcpu, user_kvm_nested_state, &kvm_state); +srcu_read_unlock (&vcpu->kvm->srcu, idx);break ; }case KVM_GET_SUPPORTED_HV_CPUID : {
Signature · 1 line
2.24.0
Quoted 70 lines · Sean Christopherson
Acquire
kvm->srcufor the duration of ->set_nested_state() to fix a bug where nVMX derefences ->memslots without holding ->srcu or ->slots_lock.The other half of nested migration, ->get_nested_state(), does not need to acquire ->srcu as it is a purely a dump of internal KVM (and CPU) state to userspace.
Detected as an RCU lockdep splat that is 100% reproducible by running KVM's
state_testselftest withCONFIG_PROVE_LOCKING=y. Note that the failing function,kvm_is_visible_gfn(), is only checking the validity of a gfn, it's not actually accessing guest memory (which is more or less unsupported duringvmx_set_nested_state()due to incorrect MMU state), i.e.vmx_set_nested_state()itself isn't fundamentally broken. In any case, setting nested state isn't a fast path so there's no reason to go out of our way to avoid taking ->srcu.=============================
WARNING: suspicious RCU usage
5.4.0-rc7+ #94 Not tainted
-----------------------------
include/linux/kvm_host.h:626suspiciousrcu_dereference_check()usage!other info that might help us debug this:
rcu_scheduler_active= 2,debug_locks= 1
1 lock held by evmcs_test/10939:#0: ffff88826ffcb800 (&vcpu->mutex){+.+.}, at: kvm_vcpu_ioctl+0x85/0x630 [kvm]stack backtrace: CPU: 1 PID: 10939 Comm: evmcs_test Not tainted 5.4.0-rc7+ #94 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015 Call Trace: dump_stack+0x68/0x9b kvm_is_visible_gfn+0x179/0x180 [kvm] mmu_check_root+0x11/0x30 [kvm] fast_cr3_switch+0x40/0x120 [kvm] kvm_mmu_new_cr3+0x34/0x60 [kvm] nested_vmx_load_cr3+0xbd/0x1f0 [kvm_intel] nested_vmx_enter_non_root_mode+0xab8/0x1d60 [kvm_intel] vmx_set_nested_state+0x256/0x340 [kvm_intel] kvm_arch_vcpu_ioctl+0x491/0x11a0 [kvm] kvm_vcpu_ioctl+0xde/0x630 [kvm] do_vfs_ioctl+0xa2/0x6c0 ksys_ioctl+0x66/0x70 __x64_sys_ioctl+0x16/0x20 do_syscall_64+0x54/0x200 entry_SYSCALL_64_after_hwframe+0x49/0xbe RIP: 0033:0x7f59a2b95f47Fixes 8fcc4b5923af5 ("kvm: nVMX: Introduce KVM_CAP_NESTED_STATE")Signed-off-by Sean Christopherson <[email protected]>--- arch/x86/kvm/x86.c | 3 +++ 1 file changed, 3 insertions(+)arch/x86/kvm/x86.c +1
-0@@ -4421,6 +4421,7 @@ long kvm_arch_vcpu_ioctl(struct file *filp,case KVM_SET_NESTED_STATE : {struct kvm_nested_state __user *user_kvm_nested_state = argp;struct kvm_nested_state kvm_state; +int idx; r = -EINVAL ;if (!kvm_x86_ops->set_nested_state)
arch/x86/kvm/x86.c long kvm_arch_vcpu_ioctl(struct file *filp, · line 4453
Quoted 9 lines
arch/x86/kvm/x86.c +2
-0@@ -4444,7 +4445,9 @@ long kvm_arch_vcpu_ioctl(struct file *filp, && !(kvm_state.flags & KVM_STATE_NESTED_GUEST_MODE)) break; + idx =srcu_read_lock (&vcpu->kvm->srcu); r = kvm_x86_ops->set_nested_state (vcpu, user_kvm_nested_state, &kvm_state); +srcu_read_unlock (&vcpu->kvm->srcu, idx);break ; }case KVM_GET_SUPPORTED_HV_CPUID : {
Queued, thanks.
Paolo