Merge remote-tracking branch 'kvm/master' into HEAD

Pick commit fdba608f15e2 ("KVM: VMX: Wake vCPU when delivering posted
IRQ even if vCPU == this vCPU").  In addition to fixing a bug, it
also aligns the non-nested and nested usage of triggering posted
interrupts, allowing for additional cleanups.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

10 files changed, 107 insertions, 52 deletions

diff --git a/arch/x86/include/asm/kvm-x86-ops.h b/arch/x86/include/asm/kvm-x86-ops.h
index cefe1d81e2e8..9e50da3ed01a 100644
--- a/arch/x86/include/asm/kvm-x86-ops.h
+++ b/arch/x86/include/asm/kvm-x86-ops.h
@@ -47,6 +47,7 @@ KVM_X86_OP(set_dr7)
 KVM_X86_OP(cache_reg)
 KVM_X86_OP(get_rflags)
 KVM_X86_OP(set_rflags)
+KVM_X86_OP(get_if_flag)
 KVM_X86_OP(tlb_flush_all)
 KVM_X86_OP(tlb_flush_current)
 KVM_X86_OP_NULL(tlb_remote_flush)
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index d5fede05eb5f..344ced0b6b68 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -97,7 +97,7 @@
 	KVM_ARCH_REQ_FLAGS(25, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
 #define KVM_REQ_TLB_FLUSH_CURRENT	KVM_ARCH_REQ(26)
 #define KVM_REQ_TLB_FLUSH_GUEST \
-	KVM_ARCH_REQ_FLAGS(27, KVM_REQUEST_NO_WAKEUP)
+	KVM_ARCH_REQ_FLAGS(27, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
 #define KVM_REQ_APF_READY		KVM_ARCH_REQ(28)
 #define KVM_REQ_MSR_FILTER_CHANGED	KVM_ARCH_REQ(29)
 #define KVM_REQ_UPDATE_CPU_DIRTY_LOGGING \
@@ -1354,6 +1354,7 @@ struct kvm_x86_ops {
 	void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
 	unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
 	void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
+	bool (*get_if_flag)(struct kvm_vcpu *vcpu);
 
 	void (*tlb_flush_all)(struct kvm_vcpu *vcpu);
 	void (*tlb_flush_current)(struct kvm_vcpu *vcpu);
diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c
index a91424ed436d..6e38a7d22e97 100644
--- a/arch/x86/kvm/hyperv.c
+++ b/arch/x86/kvm/hyperv.c
@@ -1923,11 +1923,13 @@ static u64 kvm_hv_send_ipi(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc, bool
 
 		all_cpus = send_ipi_ex.vp_set.format == HV_GENERIC_SET_ALL;
 
+		if (all_cpus)
+			goto check_and_send_ipi;
+
 		if (!sparse_banks_len)
 			goto ret_success;
 
-		if (!all_cpus &&
-		    kvm_read_guest(kvm,
+		if (kvm_read_guest(kvm,
 				   hc->ingpa + offsetof(struct hv_send_ipi_ex,
 							vp_set.bank_contents),
 				   sparse_banks,
@@ -1935,6 +1937,7 @@ static u64 kvm_hv_send_ipi(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc, bool
 			return HV_STATUS_INVALID_HYPERCALL_INPUT;
 	}
 
+check_and_send_ipi:
 	if ((vector < HV_IPI_LOW_VECTOR) || (vector > HV_IPI_HIGH_VECTOR))
 		return HV_STATUS_INVALID_HYPERCALL_INPUT;
 
diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
index 1ccee4d17481..1d275e9d76b5 100644
--- a/arch/x86/kvm/mmu/mmu.c
+++ b/arch/x86/kvm/mmu/mmu.c
@@ -3971,7 +3971,21 @@ out_retry:
 static bool is_page_fault_stale(struct kvm_vcpu *vcpu,
 				struct kvm_page_fault *fault, int mmu_seq)
 {
-	if (is_obsolete_sp(vcpu->kvm, to_shadow_page(vcpu->arch.mmu->root_hpa)))
+	struct kvm_mmu_page *sp = to_shadow_page(vcpu->arch.mmu->root_hpa);
+
+	/* Special roots, e.g. pae_root, are not backed by shadow pages. */
+	if (sp && is_obsolete_sp(vcpu->kvm, sp))
+		return true;
+
+	/*
+	 * Roots without an associated shadow page are considered invalid if
+	 * there is a pending request to free obsolete roots.  The request is
+	 * only a hint that the current root _may_ be obsolete and needs to be
+	 * reloaded, e.g. if the guest frees a PGD that KVM is tracking as a
+	 * previous root, then __kvm_mmu_prepare_zap_page() signals all vCPUs
+	 * to reload even if no vCPU is actively using the root.
+	 */
+	if (!sp && kvm_test_request(KVM_REQ_MMU_RELOAD, vcpu))
 		return true;
 
 	return fault->slot &&
diff --git a/arch/x86/kvm/mmu/tdp_iter.c b/arch/x86/kvm/mmu/tdp_iter.c
index b3ed302c1a35..caa96c270b95 100644
--- a/arch/x86/kvm/mmu/tdp_iter.c
+++ b/arch/x86/kvm/mmu/tdp_iter.c
@@ -26,6 +26,7 @@ static gfn_t round_gfn_for_level(gfn_t gfn, int level)
  */
 void tdp_iter_restart(struct tdp_iter *iter)
 {
+	iter->yielded = false;
 	iter->yielded_gfn = iter->next_last_level_gfn;
 	iter->level = iter->root_level;
 
@@ -160,6 +161,11 @@ static bool try_step_up(struct tdp_iter *iter)
  */
 void tdp_iter_next(struct tdp_iter *iter)
 {
+	if (iter->yielded) {
+		tdp_iter_restart(iter);
+		return;
+	}
+
 	if (try_step_down(iter))
 		return;
 
diff --git a/arch/x86/kvm/mmu/tdp_iter.h b/arch/x86/kvm/mmu/tdp_iter.h
index b1748b988d3a..e19cabbcb65c 100644
--- a/arch/x86/kvm/mmu/tdp_iter.h
+++ b/arch/x86/kvm/mmu/tdp_iter.h
@@ -45,6 +45,12 @@ struct tdp_iter {
 	 * iterator walks off the end of the paging structure.
 	 */
 	bool valid;
+	/*
+	 * True if KVM dropped mmu_lock and yielded in the middle of a walk, in
+	 * which case tdp_iter_next() needs to restart the walk at the root
+	 * level instead of advancing to the next entry.
+	 */
+	bool yielded;
 };
 
 /*
diff --git a/arch/x86/kvm/mmu/tdp_mmu.c b/arch/x86/kvm/mmu/tdp_mmu.c
index b69e47e68307..7b1bc816b7c3 100644
--- a/arch/x86/kvm/mmu/tdp_mmu.c
+++ b/arch/x86/kvm/mmu/tdp_mmu.c
@@ -502,6 +502,8 @@ static inline bool tdp_mmu_set_spte_atomic(struct kvm *kvm,
 					   struct tdp_iter *iter,
 					   u64 new_spte)
 {
+	WARN_ON_ONCE(iter->yielded);
+
 	lockdep_assert_held_read(&kvm->mmu_lock);
 
 	/*
@@ -575,6 +577,8 @@ static inline void __tdp_mmu_set_spte(struct kvm *kvm, struct tdp_iter *iter,
 				      u64 new_spte, bool record_acc_track,
 				      bool record_dirty_log)
 {
+	WARN_ON_ONCE(iter->yielded);
+
 	lockdep_assert_held_write(&kvm->mmu_lock);
 
 	/*
@@ -640,18 +644,19 @@ static inline void tdp_mmu_set_spte_no_dirty_log(struct kvm *kvm,
  * If this function should yield and flush is set, it will perform a remote
  * TLB flush before yielding.
  *
- * If this function yields, it will also reset the tdp_iter's walk over the
- * paging structure and the calling function should skip to the next
- * iteration to allow the iterator to continue its traversal from the
- * paging structure root.
+ * If this function yields, iter->yielded is set and the caller must skip to
+ * the next iteration, where tdp_iter_next() will reset the tdp_iter's walk
+ * over the paging structures to allow the iterator to continue its traversal
+ * from the paging structure root.
  *
- * Return true if this function yielded and the iterator's traversal was reset.
- * Return false if a yield was not needed.
+ * Returns true if this function yielded.
  */
-static inline bool tdp_mmu_iter_cond_resched(struct kvm *kvm,
-					     struct tdp_iter *iter, bool flush,
-					     bool shared)
+static inline bool __must_check tdp_mmu_iter_cond_resched(struct kvm *kvm,
+							  struct tdp_iter *iter,
+							  bool flush, bool shared)
 {
+	WARN_ON(iter->yielded);
+
 	/* Ensure forward progress has been made before yielding. */
 	if (iter->next_last_level_gfn == iter->yielded_gfn)
 		return false;
@@ -671,12 +676,10 @@ static inline bool tdp_mmu_iter_cond_resched(struct kvm *kvm,
 
 		WARN_ON(iter->gfn > iter->next_last_level_gfn);
 
-		tdp_iter_restart(iter);
-
-		return true;
+		iter->yielded = true;
 	}
 
-	return false;
+	return iter->yielded;
 }
 
 /*
diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
index 208566f63bce..5557867dcb69 100644
--- a/arch/x86/kvm/svm/svm.c
+++ b/arch/x86/kvm/svm/svm.c
@@ -1594,6 +1594,15 @@ static void svm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags)
 	to_svm(vcpu)->vmcb->save.rflags = rflags;
 }
 
+static bool svm_get_if_flag(struct kvm_vcpu *vcpu)
+{
+	struct vmcb *vmcb = to_svm(vcpu)->vmcb;
+
+	return sev_es_guest(vcpu->kvm)
+		? vmcb->control.int_state & SVM_GUEST_INTERRUPT_MASK
+		: kvm_get_rflags(vcpu) & X86_EFLAGS_IF;
+}
+
 static void svm_cache_reg(struct kvm_vcpu *vcpu, enum kvm_reg reg)
 {
 	kvm_register_mark_available(vcpu, reg);
@@ -3583,14 +3592,7 @@ bool svm_interrupt_blocked(struct kvm_vcpu *vcpu)
 	if (!gif_set(svm))
 		return true;
 
-	if (sev_es_guest(vcpu->kvm)) {
-		/*
-		 * SEV-ES guests to not expose RFLAGS. Use the VMCB interrupt mask
-		 * bit to determine the state of the IF flag.
-		 */
-		if (!(vmcb->control.int_state & SVM_GUEST_INTERRUPT_MASK))
-			return true;
-	} else if (is_guest_mode(vcpu)) {
+	if (is_guest_mode(vcpu)) {
 		/* As long as interrupts are being delivered...  */
 		if ((svm->nested.ctl.int_ctl & V_INTR_MASKING_MASK)
 		    ? !(svm->vmcb01.ptr->save.rflags & X86_EFLAGS_IF)
@@ -3601,7 +3603,7 @@ bool svm_interrupt_blocked(struct kvm_vcpu *vcpu)
 		if (nested_exit_on_intr(svm))
 			return false;
 	} else {
-		if (!(kvm_get_rflags(vcpu) & X86_EFLAGS_IF))
+		if (!svm_get_if_flag(vcpu))
 			return true;
 	}
 
@@ -4634,6 +4636,7 @@ static struct kvm_x86_ops svm_x86_ops __initdata = {
 	.cache_reg = svm_cache_reg,
 	.get_rflags = svm_get_rflags,
 	.set_rflags = svm_set_rflags,
+	.get_if_flag = svm_get_if_flag,
 
 	.tlb_flush_all = svm_flush_tlb,
 	.tlb_flush_current = svm_flush_tlb,
diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
index 63615d242bdf..9bf65e5e4840 100644
--- a/arch/x86/kvm/vmx/vmx.c
+++ b/arch/x86/kvm/vmx/vmx.c
@@ -1372,6 +1372,11 @@ void vmx_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags)
 		vmx->emulation_required = vmx_emulation_required(vcpu);
 }
 
+static bool vmx_get_if_flag(struct kvm_vcpu *vcpu)
+{
+	return vmx_get_rflags(vcpu) & X86_EFLAGS_IF;
+}
+
 u32 vmx_get_interrupt_shadow(struct kvm_vcpu *vcpu)
 {
 	u32 interruptibility = vmcs_read32(GUEST_INTERRUPTIBILITY_INFO);
@@ -3995,8 +4000,7 @@ static int vmx_deliver_posted_interrupt(struct kvm_vcpu *vcpu, int vector)
 	 * guaranteed to see PID.ON=1 and sync the PIR to IRR if triggering a
 	 * posted interrupt "fails" because vcpu->mode != IN_GUEST_MODE.
 	 */
-	if (vcpu != kvm_get_running_vcpu() &&
-	    !kvm_vcpu_trigger_posted_interrupt(vcpu, false))
+	if (!kvm_vcpu_trigger_posted_interrupt(vcpu, false))
 		kvm_vcpu_kick(vcpu);
 
 	return 0;
@@ -5921,18 +5925,14 @@ static int __vmx_handle_exit(struct kvm_vcpu *vcpu, fastpath_t exit_fastpath)
 		vmx_flush_pml_buffer(vcpu);
 
 	/*
-	 * We should never reach this point with a pending nested VM-Enter, and
-	 * more specifically emulation of L2 due to invalid guest state (see
-	 * below) should never happen as that means we incorrectly allowed a
-	 * nested VM-Enter with an invalid vmcs12.
+	 * KVM should never reach this point with a pending nested VM-Enter.
+	 * More specifically, short-circuiting VM-Entry to emulate L2 due to
+	 * invalid guest state should never happen as that means KVM knowingly
+	 * allowed a nested VM-Enter with an invalid vmcs12.  More below.
 	 */
 	if (KVM_BUG_ON(vmx->nested.nested_run_pending, vcpu->kvm))
 		return -EIO;
 
-	/* If guest state is invalid, start emulating */
-	if (vmx->emulation_required)
-		return handle_invalid_guest_state(vcpu);
-
 	if (is_guest_mode(vcpu)) {
 		/*
 		 * PML is never enabled when running L2, bail immediately if a
@@ -5954,10 +5954,30 @@ static int __vmx_handle_exit(struct kvm_vcpu *vcpu, fastpath_t exit_fastpath)
 		 */
 		nested_mark_vmcs12_pages_dirty(vcpu);
 
+		/*
+		 * Synthesize a triple fault if L2 state is invalid.  In normal
+		 * operation, nested VM-Enter rejects any attempt to enter L2
+		 * with invalid state.  However, those checks are skipped if
+		 * state is being stuffed via RSM or KVM_SET_NESTED_STATE.  If
+		 * L2 state is invalid, it means either L1 modified SMRAM state
+		 * or userspace provided bad state.  Synthesize TRIPLE_FAULT as
+		 * doing so is architecturally allowed in the RSM case, and is
+		 * the least awful solution for the userspace case without
+		 * risking false positives.
+		 */
+		if (vmx->emulation_required) {
+			nested_vmx_vmexit(vcpu, EXIT_REASON_TRIPLE_FAULT, 0, 0);
+			return 1;
+		}
+
 		if (nested_vmx_reflect_vmexit(vcpu))
 			return 1;
 	}
 
+	/* If guest state is invalid, start emulating.  L2 is handled above. */
+	if (vmx->emulation_required)
+		return handle_invalid_guest_state(vcpu);
+
 	if (exit_reason.failed_vmentry) {
 		dump_vmcs(vcpu);
 		vcpu->run->exit_reason = KVM_EXIT_FAIL_ENTRY;
@@ -6652,9 +6672,7 @@ static fastpath_t vmx_vcpu_run(struct kvm_vcpu *vcpu)
 	 * consistency check VM-Exit due to invalid guest state and bail.
 	 */
 	if (unlikely(vmx->emulation_required)) {
-
-		/* We don't emulate invalid state of a nested guest */
-		vmx->fail = is_guest_mode(vcpu);
+		vmx->fail = 0;
 
 		vmx->exit_reason.full = EXIT_REASON_INVALID_STATE;
 		vmx->exit_reason.failed_vmentry = 1;
@@ -7609,6 +7627,7 @@ static struct kvm_x86_ops vmx_x86_ops __initdata = {
 	.cache_reg = vmx_cache_reg,
 	.get_rflags = vmx_get_rflags,
 	.set_rflags = vmx_set_rflags,
+	.get_if_flag = vmx_get_if_flag,
 
 	.tlb_flush_all = vmx_flush_tlb_all,
 	.tlb_flush_current = vmx_flush_tlb_current,
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 26cb3a4cd0e9..8912f44c640d 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -906,7 +906,8 @@ int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
 	    !load_pdptrs(vcpu, kvm_read_cr3(vcpu)))
 		return 1;
 
-	if (!(cr0 & X86_CR0_PG) && kvm_read_cr4_bits(vcpu, X86_CR4_PCIDE))
+	if (!(cr0 & X86_CR0_PG) &&
+	    (is_64_bit_mode(vcpu) || kvm_read_cr4_bits(vcpu, X86_CR4_PCIDE)))
 		return 1;
 
 	static_call(kvm_x86_set_cr0)(vcpu, cr0);
@@ -1343,7 +1344,7 @@ static const u32 msrs_to_save_all[] = {
 	MSR_IA32_UMWAIT_CONTROL,
 
 	MSR_ARCH_PERFMON_FIXED_CTR0, MSR_ARCH_PERFMON_FIXED_CTR1,
-	MSR_ARCH_PERFMON_FIXED_CTR0 + 2, MSR_ARCH_PERFMON_FIXED_CTR0 + 3,
+	MSR_ARCH_PERFMON_FIXED_CTR0 + 2,
 	MSR_CORE_PERF_FIXED_CTR_CTRL, MSR_CORE_PERF_GLOBAL_STATUS,
 	MSR_CORE_PERF_GLOBAL_CTRL, MSR_CORE_PERF_GLOBAL_OVF_CTRL,
 	MSR_ARCH_PERFMON_PERFCTR0, MSR_ARCH_PERFMON_PERFCTR1,
@@ -3424,7 +3425,7 @@ int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
 
 		if (!msr_info->host_initiated)
 			return 1;
-		if (guest_cpuid_has(vcpu, X86_FEATURE_PDCM) && kvm_get_msr_feature(&msr_ent))
+		if (kvm_get_msr_feature(&msr_ent))
 			return 1;
 		if (data & ~msr_ent.data)
 			return 1;
@@ -7144,7 +7145,13 @@ static int emulator_pio_in(struct kvm_vcpu *vcpu, int size,
 			   unsigned short port, void *val, unsigned int count)
 {
 	if (vcpu->arch.pio.count) {
-		/* Complete previous iteration.  */
+		/*
+		 * Complete a previous iteration that required userspace I/O.
+		 * Note, @count isn't guaranteed to match pio.count as userspace
+		 * can modify ECX before rerunning the vCPU.  Ignore any such
+		 * shenanigans as KVM doesn't support modifying the rep count,
+		 * and the emulator ensures @count doesn't overflow the buffer.
+		 */
 	} else {
 		int r = __emulator_pio_in(vcpu, size, port, count);
 		if (!r)
@@ -7153,7 +7160,6 @@ static int emulator_pio_in(struct kvm_vcpu *vcpu, int size,
 		/* Results already available, fall through.  */
 	}
 
-	WARN_ON(count != vcpu->arch.pio.count);
 	complete_emulator_pio_in(vcpu, val);
 	return 1;
 }
@@ -9043,14 +9049,7 @@ static void post_kvm_run_save(struct kvm_vcpu *vcpu)
 {
 	struct kvm_run *kvm_run = vcpu->run;
 
-	/*
-	 * if_flag is obsolete and useless, so do not bother
-	 * setting it for SEV-ES guests.  Userspace can just
-	 * use kvm_run->ready_for_interrupt_injection.
-	 */
-	kvm_run->if_flag = !vcpu->arch.guest_state_protected
-		&& (kvm_get_rflags(vcpu) & X86_EFLAGS_IF) != 0;
-
+	kvm_run->if_flag = static_call(kvm_x86_get_if_flag)(vcpu);
 	kvm_run->cr8 = kvm_get_cr8(vcpu);
 	kvm_run->apic_base = kvm_get_apic_base(vcpu);