diff options
| author | Sheng Yang <sheng@linux.intel.com> | 2010-06-30 12:25:15 +0800 |
|---|---|---|
| committer | Avi Kivity <avi@redhat.com> | 2010-08-01 10:47:21 +0300 |
| commit | f5f48ee15c2ee3e44cf429e34b16c6fa9b900246 (patch) | |
| tree | 75496197219d9aeedd3317fa007cc3b2e414c5da /arch/x86/kvm/x86.c | |
| parent | cf3e3d3e19868ca01da163200bbfc687523df0fc (diff) | |
KVM: VMX: Execute WBINVD to keep data consistency with assigned devices
Some guest device driver may leverage the "Non-Snoop" I/O, and explicitly
WBINVD or CLFLUSH to a RAM space. Since migration may occur before WBINVD or
CLFLUSH, we need to maintain data consistency either by:
1: flushing cache (wbinvd) when the guest is scheduled out if there is no
wbinvd exit, or
2: execute wbinvd on all dirty physical CPUs when guest wbinvd exits.
Signed-off-by: Yaozu (Eddie) Dong <eddie.dong@intel.com>
Signed-off-by: Sheng Yang <sheng@linux.intel.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Diffstat (limited to 'arch/x86/kvm/x86.c')
| -rw-r--r-- | arch/x86/kvm/x86.c | 41 |
1 files changed, 41 insertions, 0 deletions
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 27322d341232..3d72fc067059 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -1783,8 +1783,28 @@ out: return r; } +static void wbinvd_ipi(void *garbage) +{ + wbinvd(); +} + +static bool need_emulate_wbinvd(struct kvm_vcpu *vcpu) +{ + return vcpu->kvm->arch.iommu_domain && + !(vcpu->kvm->arch.iommu_flags & KVM_IOMMU_CACHE_COHERENCY); +} + void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu) { + /* Address WBINVD may be executed by guest */ + if (need_emulate_wbinvd(vcpu)) { + if (kvm_x86_ops->has_wbinvd_exit()) + cpumask_set_cpu(cpu, vcpu->arch.wbinvd_dirty_mask); + else if (vcpu->cpu != -1 && vcpu->cpu != cpu) + smp_call_function_single(vcpu->cpu, + wbinvd_ipi, NULL, 1); + } + kvm_x86_ops->vcpu_load(vcpu, cpu); if (unlikely(per_cpu(cpu_tsc_khz, cpu) == 0)) { unsigned long khz = cpufreq_quick_get(cpu); @@ -3660,6 +3680,21 @@ int emulate_invlpg(struct kvm_vcpu *vcpu, gva_t address) return X86EMUL_CONTINUE; } +int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu) +{ + if (!need_emulate_wbinvd(vcpu)) + return X86EMUL_CONTINUE; + + if (kvm_x86_ops->has_wbinvd_exit()) { + smp_call_function_many(vcpu->arch.wbinvd_dirty_mask, + wbinvd_ipi, NULL, 1); + cpumask_clear(vcpu->arch.wbinvd_dirty_mask); + } + wbinvd(); + return X86EMUL_CONTINUE; +} +EXPORT_SYMBOL_GPL(kvm_emulate_wbinvd); + int emulate_clts(struct kvm_vcpu *vcpu) { kvm_x86_ops->set_cr0(vcpu, kvm_read_cr0_bits(vcpu, ~X86_CR0_TS)); @@ -5263,6 +5298,7 @@ void kvm_arch_vcpu_free(struct kvm_vcpu *vcpu) vcpu->arch.time_page = NULL; } + free_cpumask_var(vcpu->arch.wbinvd_dirty_mask); fx_free(vcpu); kvm_x86_ops->vcpu_free(vcpu); } @@ -5392,7 +5428,12 @@ int kvm_arch_vcpu_init(struct kvm_vcpu *vcpu) } vcpu->arch.mcg_cap = KVM_MAX_MCE_BANKS; + if (!zalloc_cpumask_var(&vcpu->arch.wbinvd_dirty_mask, GFP_KERNEL)) + goto fail_free_mce_banks; + return 0; +fail_free_mce_banks: + kfree(vcpu->arch.mce_banks); fail_free_lapic: kvm_free_lapic(vcpu); fail_mmu_destroy: |