| Message ID | 20190523063632.GF19655@blackberry (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | KVM: PPC: Book3S: Fix potential deadlocks | expand |
On 5/23/19 8:36 AM, Paul Mackerras wrote: > Currently the HV KVM code takes the kvm->lock around calls to > kvm_for_each_vcpu() and kvm_get_vcpu_by_id() (which can call > kvm_for_each_vcpu() internally). However, that leads to a lock > order inversion problem, because these are called in contexts where > the vcpu mutex is held, but the vcpu mutexes nest within kvm->lock > according to Documentation/virtual/kvm/locking.txt. Hence there > is a possibility of deadlock. > > To fix this, we simply don't take the kvm->lock mutex around these > calls. This is safe because the implementations of kvm_for_each_vcpu() > and kvm_get_vcpu_by_id() have been designed to be able to be called > locklessly. Yes. Reviewed-by: Cédric Le Goater <clg@kaod.org> Thanks, C. > > Signed-off-by: Paul Mackerras <paulus@ozlabs.org> > --- > arch/powerpc/kvm/book3s_hv.c | 9 +-------- > 1 file changed, 1 insertion(+), 8 deletions(-) > > diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c > index b1c0a9b..27054d3 100644 > --- a/arch/powerpc/kvm/book3s_hv.c > +++ b/arch/powerpc/kvm/book3s_hv.c > @@ -446,12 +446,7 @@ static void kvmppc_dump_regs(struct kvm_vcpu *vcpu) > > static struct kvm_vcpu *kvmppc_find_vcpu(struct kvm *kvm, int id) > { > - struct kvm_vcpu *ret; > - > - mutex_lock(&kvm->lock); > - ret = kvm_get_vcpu_by_id(kvm, id); > - mutex_unlock(&kvm->lock); > - return ret; > + return kvm_get_vcpu_by_id(kvm, id); > } > > static void init_vpa(struct kvm_vcpu *vcpu, struct lppaca *vpa) > @@ -1583,7 +1578,6 @@ static void kvmppc_set_lpcr(struct kvm_vcpu *vcpu, u64 new_lpcr, > struct kvmppc_vcore *vc = vcpu->arch.vcore; > u64 mask; > > - mutex_lock(&kvm->lock); > spin_lock(&vc->lock); > /* > * If ILE (interrupt little-endian) has changed, update the > @@ -1623,7 +1617,6 @@ static void kvmppc_set_lpcr(struct kvm_vcpu *vcpu, u64 new_lpcr, > mask &= 0xFFFFFFFF; > vc->lpcr = (vc->lpcr & ~mask) | (new_lpcr & mask); > spin_unlock(&vc->lock); > - mutex_unlock(&kvm->lock); > } > > static int kvmppc_get_one_reg_hv(struct kvm_vcpu *vcpu, u64 id, >
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c index b1c0a9b..27054d3 100644 --- a/arch/powerpc/kvm/book3s_hv.c +++ b/arch/powerpc/kvm/book3s_hv.c @@ -446,12 +446,7 @@ static void kvmppc_dump_regs(struct kvm_vcpu *vcpu) static struct kvm_vcpu *kvmppc_find_vcpu(struct kvm *kvm, int id) { - struct kvm_vcpu *ret; - - mutex_lock(&kvm->lock); - ret = kvm_get_vcpu_by_id(kvm, id); - mutex_unlock(&kvm->lock); - return ret; + return kvm_get_vcpu_by_id(kvm, id); } static void init_vpa(struct kvm_vcpu *vcpu, struct lppaca *vpa) @@ -1583,7 +1578,6 @@ static void kvmppc_set_lpcr(struct kvm_vcpu *vcpu, u64 new_lpcr, struct kvmppc_vcore *vc = vcpu->arch.vcore; u64 mask; - mutex_lock(&kvm->lock); spin_lock(&vc->lock); /* * If ILE (interrupt little-endian) has changed, update the @@ -1623,7 +1617,6 @@ static void kvmppc_set_lpcr(struct kvm_vcpu *vcpu, u64 new_lpcr, mask &= 0xFFFFFFFF; vc->lpcr = (vc->lpcr & ~mask) | (new_lpcr & mask); spin_unlock(&vc->lock); - mutex_unlock(&kvm->lock); } static int kvmppc_get_one_reg_hv(struct kvm_vcpu *vcpu, u64 id,
Currently the HV KVM code takes the kvm->lock around calls to kvm_for_each_vcpu() and kvm_get_vcpu_by_id() (which can call kvm_for_each_vcpu() internally). However, that leads to a lock order inversion problem, because these are called in contexts where the vcpu mutex is held, but the vcpu mutexes nest within kvm->lock according to Documentation/virtual/kvm/locking.txt. Hence there is a possibility of deadlock. To fix this, we simply don't take the kvm->lock mutex around these calls. This is safe because the implementations of kvm_for_each_vcpu() and kvm_get_vcpu_by_id() have been designed to be able to be called locklessly. Signed-off-by: Paul Mackerras <paulus@ozlabs.org> --- arch/powerpc/kvm/book3s_hv.c | 9 +-------- 1 file changed, 1 insertion(+), 8 deletions(-)