Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm

Pull KVM fixes from Paolo Bonzini:
"ARM fixes:

- Another state update on exit to userspace fix

- Prevent the creation of mixed 32/64 VMs

- Fix regression with irqbypass not restarting the guest on failed
connect

- Fix regression with debug register decoding resulting in
overlapping access

- Commit exception state on exit to usrspace

- Fix the MMU notifier return values

- Add missing 'static' qualifiers in the new host stage-2 code

x86 fixes:

- fix guest missed wakeup with assigned devices

- fix WARN reported by syzkaller

- do not use BIT() in UAPI headers

- make the kvm_amd.avic parameter bool

PPC fixes:

- make halt polling heuristics consistent with other architectures

selftests:

- various fixes

- new performance selftest memslot_perf_test

- test UFFD minor faults in demand_paging_test"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (44 commits)
selftests: kvm: fix overlapping addresses in memslot_perf_test
KVM: X86: Kill off ctxt->ud
KVM: X86: Fix warning caused by stale emulation context
KVM: X86: Use kvm_get_linear_rip() in single-step and #DB/#BP interception
KVM: x86/mmu: Fix comment mentioning skip_4k
KVM: VMX: update vcpu posted-interrupt descriptor when assigning device
KVM: rename KVM_REQ_PENDING_TIMER to KVM_REQ_UNBLOCK
KVM: x86: add start_assignment hook to kvm_x86_ops
KVM: LAPIC: Narrow the timer latency between wait_lapic_expire and world switch
selftests: kvm: do only 1 memslot_perf_test run by default
KVM: X86: Use _BITUL() macro in UAPI headers
KVM: selftests: add shared hugetlbfs backing source type
KVM: selftests: allow using UFFD minor faults for demand paging
KVM: selftests: create alias mappings when using shared memory
KVM: selftests: add shmem backing source type
KVM: selftests: refactor vm_mem_backing_src_type flags
KVM: selftests: allow different backing source types
KVM: selftests: compute correct demand paging size
KVM: selftests: simplify setup_demand_paging error handling
KVM: selftests: Print a message if /dev/kvm is missing
...

Documentation/virt/kvm/vcpu-requests.rst

···
   necessary to inform each VCPU to completely refresh the tables.  This
   request is used for that.
 
-KVM_REQ_PENDING_TIMER
+KVM_REQ_UNBLOCK
 
-  This request may be made from a timer handler run on the host on behalf
-  of a VCPU.  It informs the VCPU thread to inject a timer interrupt.
+  This request informs the vCPU to exit kvm_vcpu_block.  It is used for
+  example from timer handlers that run on the host on behalf of a vCPU,
+  or in order to update the interrupt routing and ensure that assigned
+  devices will wake up the vCPU.
 
 KVM_REQ_UNHALT
 

arch/arm64/include/asm/kvm_asm.h

···
 #define __KVM_HOST_SMCCC_FUNC___pkvm_cpu_set_vector		18
 #define __KVM_HOST_SMCCC_FUNC___pkvm_prot_finalize		19
 #define __KVM_HOST_SMCCC_FUNC___pkvm_mark_hyp			20
+#define __KVM_HOST_SMCCC_FUNC___kvm_adjust_pc			21
 
 #ifndef __ASSEMBLY__
 
···
 extern void __kvm_timer_set_cntvoff(u64 cntvoff);
 
 extern int __kvm_vcpu_run(struct kvm_vcpu *vcpu);
+
+extern void __kvm_adjust_pc(struct kvm_vcpu *vcpu);
 
 extern u64 __vgic_v3_get_gic_config(void);
 extern u64 __vgic_v3_read_vmcr(void);

arch/arm64/include/asm/kvm_emulate.h

···
 	vcpu->arch.flags |= KVM_ARM64_INCREMENT_PC;
 }
 
+static inline bool vcpu_has_feature(struct kvm_vcpu *vcpu, int feature)
+{
+	return test_bit(feature, vcpu->arch.features);
+}
+
 #endif /* __ARM64_KVM_EMULATE_H__ */

arch/arm64/kvm/arm.c

···
 			return ret;
 	}
 
-	if (run->immediate_exit)
-		return -EINTR;
-
 	vcpu_load(vcpu);
+
+	if (run->immediate_exit) {
+		ret = -EINTR;
+		goto out;
+	}
 
 	kvm_sigset_activate(vcpu);
 
···
 	}
 
 	kvm_sigset_deactivate(vcpu);
+
+out:
+	/*
+	 * In the unlikely event that we are returning to userspace
+	 * with pending exceptions or PC adjustment, commit these
+	 * adjustments in order to give userspace a consistent view of
+	 * the vcpu state. Note that this relies on __kvm_adjust_pc()
+	 * being preempt-safe on VHE.
+	 */
+	if (unlikely(vcpu->arch.flags & (KVM_ARM64_PENDING_EXCEPTION |
+					 KVM_ARM64_INCREMENT_PC)))
+		kvm_call_hyp(__kvm_adjust_pc, vcpu);
 
 	vcpu_put(vcpu);
 	return ret;

arch/arm64/kvm/hyp/exception.c

···
 	*vcpu_pc(vcpu) = vect_offset;
 }
 
-void kvm_inject_exception(struct kvm_vcpu *vcpu)
+static void kvm_inject_exception(struct kvm_vcpu *vcpu)
 {
 	if (vcpu_el1_is_32bit(vcpu)) {
 		switch (vcpu->arch.flags & KVM_ARM64_EXCEPT_MASK) {
···
 			 */
 			break;
 		}
+	}
+}
+
+/*
+ * Adjust the guest PC (and potentially exception state) depending on
+ * flags provided by the emulation code.
+ */
+void __kvm_adjust_pc(struct kvm_vcpu *vcpu)
+{
+	if (vcpu->arch.flags & KVM_ARM64_PENDING_EXCEPTION) {
+		kvm_inject_exception(vcpu);
+		vcpu->arch.flags &= ~(KVM_ARM64_PENDING_EXCEPTION |
+				      KVM_ARM64_EXCEPT_MASK);
+	} else 	if (vcpu->arch.flags & KVM_ARM64_INCREMENT_PC) {
+		kvm_skip_instr(vcpu);
+		vcpu->arch.flags &= ~KVM_ARM64_INCREMENT_PC;
 	}
 }

arch/arm64/kvm/hyp/include/hyp/adjust_pc.h

···
 #include <asm/kvm_emulate.h>
 #include <asm/kvm_host.h>
 
-void kvm_inject_exception(struct kvm_vcpu *vcpu);
-
 static inline void kvm_skip_instr(struct kvm_vcpu *vcpu)
 {
 	if (vcpu_mode_is_32bit(vcpu)) {
···
 
 	write_sysreg_el2(vcpu_gp_regs(vcpu)->pstate, SYS_SPSR);
 	write_sysreg_el2(*vcpu_pc(vcpu), SYS_ELR);
-}
-
-/*
- * Adjust the guest PC on entry, depending on flags provided by EL1
- * for the purpose of emulation (MMIO, sysreg) or exception injection.
- */
-static inline void __adjust_pc(struct kvm_vcpu *vcpu)
-{
-	if (vcpu->arch.flags & KVM_ARM64_PENDING_EXCEPTION) {
-		kvm_inject_exception(vcpu);
-		vcpu->arch.flags &= ~(KVM_ARM64_PENDING_EXCEPTION |
-				      KVM_ARM64_EXCEPT_MASK);
-	} else 	if (vcpu->arch.flags & KVM_ARM64_INCREMENT_PC) {
-		kvm_skip_instr(vcpu);
-		vcpu->arch.flags &= ~KVM_ARM64_INCREMENT_PC;
-	}
 }
 
 /*

arch/arm64/kvm/hyp/nvhe/hyp-main.c

···
 	cpu_reg(host_ctxt, 1) =  __kvm_vcpu_run(kern_hyp_va(vcpu));
 }
 
+static void handle___kvm_adjust_pc(struct kvm_cpu_context *host_ctxt)
+{
+	DECLARE_REG(struct kvm_vcpu *, vcpu, host_ctxt, 1);
+
+	__kvm_adjust_pc(kern_hyp_va(vcpu));
+}
+
 static void handle___kvm_flush_vm_context(struct kvm_cpu_context *host_ctxt)
 {
 	__kvm_flush_vm_context();
···
 
 static const hcall_t host_hcall[] = {
 	HANDLE_FUNC(__kvm_vcpu_run),
+	HANDLE_FUNC(__kvm_adjust_pc),
 	HANDLE_FUNC(__kvm_flush_vm_context),
 	HANDLE_FUNC(__kvm_tlb_flush_vmid_ipa),
 	HANDLE_FUNC(__kvm_tlb_flush_vmid),

arch/arm64/kvm/hyp/nvhe/mem_protect.c

···
 extern unsigned long hyp_nr_cpus;
 struct host_kvm host_kvm;
 
-struct hyp_pool host_s2_mem;
-struct hyp_pool host_s2_dev;
+static struct hyp_pool host_s2_mem;
+static struct hyp_pool host_s2_dev;
 
 /*
  * Copies of the host's CPU features registers holding sanitized values.

arch/arm64/kvm/hyp/nvhe/setup.c

···
 #include <nvhe/trap_handler.h>
 
 struct hyp_pool hpool;
-struct kvm_pgtable_mm_ops pkvm_pgtable_mm_ops;
 unsigned long hyp_nr_cpus;
 
 #define hyp_percpu_size ((unsigned long)__per_cpu_end - \
···
 static void *hyp_pgt_base;
 static void *host_s2_mem_pgt_base;
 static void *host_s2_dev_pgt_base;
+static struct kvm_pgtable_mm_ops pkvm_pgtable_mm_ops;
 
 static int divide_memory_pool(void *virt, unsigned long size)
 {

arch/arm64/kvm/hyp/nvhe/switch.c

···
  * Author: Marc Zyngier <[e-mail]>
  */
 
-#include <hyp/adjust_pc.h>
 #include <hyp/switch.h>
 #include <hyp/sysreg-sr.h>
 
···
 	 */
 	__debug_save_host_buffers_nvhe(vcpu);
 
-	__adjust_pc(vcpu);
+	__kvm_adjust_pc(vcpu);
 
 	/*
 	 * We must restore the 32-bit state before the sysregs, thanks

arch/arm64/kvm/hyp/vhe/switch.c

···
  * Author: Marc Zyngier <[e-mail]>
  */
 
-#include <hyp/adjust_pc.h>
 #include <hyp/switch.h>
 
 #include <linux/arm-smccc.h>
···
 	__load_guest_stage2(vcpu->arch.hw_mmu);
 	__activate_traps(vcpu);
 
-	__adjust_pc(vcpu);
+	__kvm_adjust_pc(vcpu);
 
 	sysreg_restore_guest_state_vhe(guest_ctxt);
 	__debug_switch_to_guest(vcpu);

arch/arm64/kvm/mmu.c

···
 bool kvm_unmap_gfn_range(struct kvm *kvm, struct kvm_gfn_range *range)
 {
 	if (!kvm->arch.mmu.pgt)
-		return 0;
+		return false;
 
 	__unmap_stage2_range(&kvm->arch.mmu, range->start << PAGE_SHIFT,
 			     (range->end - range->start) << PAGE_SHIFT,
 			     range->may_block);
 
-	return 0;
+	return false;
 }
 
 bool kvm_set_spte_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
···
 	kvm_pfn_t pfn = pte_pfn(range->pte);
 
 	if (!kvm->arch.mmu.pgt)
-		return 0;
+		return false;
 
 	WARN_ON(range->end - range->start != 1);
 
···
 			       PAGE_SIZE, __pfn_to_phys(pfn),
 			       KVM_PGTABLE_PROT_R, NULL);
 
-	return 0;
+	return false;
 }
 
 bool kvm_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
···
 	pte_t pte;
 
 	if (!kvm->arch.mmu.pgt)
-		return 0;
+		return false;
 
 	WARN_ON(size != PAGE_SIZE && size != PMD_SIZE && size != PUD_SIZE);
 
···
 bool kvm_test_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
 {
 	if (!kvm->arch.mmu.pgt)
-		return 0;
+		return false;
 
 	return kvm_pgtable_stage2_is_young(kvm->arch.mmu.pgt,
 					   range->start << PAGE_SHIFT);

arch/arm64/kvm/reset.c

···
 	return 0;
 }
 
+static bool vcpu_allowed_register_width(struct kvm_vcpu *vcpu)
+{
+	struct kvm_vcpu *tmp;
+	bool is32bit;
+	int i;
+
+	is32bit = vcpu_has_feature(vcpu, KVM_ARM_VCPU_EL1_32BIT);
+	if (!cpus_have_const_cap(ARM64_HAS_32BIT_EL1) && is32bit)
+		return false;
+
+	/* Check that the vcpus are either all 32bit or all 64bit */
+	kvm_for_each_vcpu(i, tmp, vcpu->kvm) {
+		if (vcpu_has_feature(tmp, KVM_ARM_VCPU_EL1_32BIT) != is32bit)
+			return false;
+	}
+
+	return true;
+}
+
 /**
  * kvm_reset_vcpu - sets core registers and sys_regs to reset value
  * @vcpu: The VCPU pointer
···
 		}
 	}
 
+	if (!vcpu_allowed_register_width(vcpu)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
 	switch (vcpu->arch.target) {
 	default:
 		if (test_bit(KVM_ARM_VCPU_EL1_32BIT, vcpu->arch.features)) {
-			if (!cpus_have_const_cap(ARM64_HAS_32BIT_EL1)) {
-				ret = -EINVAL;
-				goto out;
-			}
 			pstate = VCPU_RESET_PSTATE_SVC;
 		} else {
 			pstate = VCPU_RESET_PSTATE_EL1;

arch/arm64/kvm/sys_regs.c

···
 		     struct sys_reg_params *p,
 		     const struct sys_reg_desc *rd)
 {
-	u64 *dbg_reg = &vcpu->arch.vcpu_debug_state.dbg_bvr[rd->reg];
+	u64 *dbg_reg = &vcpu->arch.vcpu_debug_state.dbg_bvr[rd->CRm];
 
 	if (p->is_write)
 		reg_to_dbg(vcpu, p, rd, dbg_reg);
 	else
 		dbg_to_reg(vcpu, p, rd, dbg_reg);
 
-	trace_trap_reg(__func__, rd->reg, p->is_write, *dbg_reg);
+	trace_trap_reg(__func__, rd->CRm, p->is_write, *dbg_reg);
 
 	return true;
 }
···
 static int set_bvr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
 		const struct kvm_one_reg *reg, void __user *uaddr)
 {
-	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_bvr[rd->reg];
+	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_bvr[rd->CRm];
 
 	if (copy_from_user(r, uaddr, KVM_REG_SIZE(reg->id)) != 0)
 		return -EFAULT;
···
 static int get_bvr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
 	const struct kvm_one_reg *reg, void __user *uaddr)
 {
-	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_bvr[rd->reg];
+	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_bvr[rd->CRm];
 
 	if (copy_to_user(uaddr, r, KVM_REG_SIZE(reg->id)) != 0)
 		return -EFAULT;
···
 static void reset_bvr(struct kvm_vcpu *vcpu,
 		      const struct sys_reg_desc *rd)
 {
-	vcpu->arch.vcpu_debug_state.dbg_bvr[rd->reg] = rd->val;
+	vcpu->arch.vcpu_debug_state.dbg_bvr[rd->CRm] = rd->val;
 }
 
 static bool trap_bcr(struct kvm_vcpu *vcpu,
 		     struct sys_reg_params *p,
 		     const struct sys_reg_desc *rd)
 {
-	u64 *dbg_reg = &vcpu->arch.vcpu_debug_state.dbg_bcr[rd->reg];
+	u64 *dbg_reg = &vcpu->arch.vcpu_debug_state.dbg_bcr[rd->CRm];
 
 	if (p->is_write)
 		reg_to_dbg(vcpu, p, rd, dbg_reg);
 	else
 		dbg_to_reg(vcpu, p, rd, dbg_reg);
 
-	trace_trap_reg(__func__, rd->reg, p->is_write, *dbg_reg);
+	trace_trap_reg(__func__, rd->CRm, p->is_write, *dbg_reg);
 
 	return true;
 }
···
 static int set_bcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
 		const struct kvm_one_reg *reg, void __user *uaddr)
 {
-	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_bcr[rd->reg];
+	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_bcr[rd->CRm];
 
 	if (copy_from_user(r, uaddr, KVM_REG_SIZE(reg->id)) != 0)
 		return -EFAULT;
···
 static int get_bcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
 	const struct kvm_one_reg *reg, void __user *uaddr)
 {
-	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_bcr[rd->reg];
+	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_bcr[rd->CRm];
 
 	if (copy_to_user(uaddr, r, KVM_REG_SIZE(reg->id)) != 0)
 		return -EFAULT;
···
 static void reset_bcr(struct kvm_vcpu *vcpu,
 		      const struct sys_reg_desc *rd)
 {
-	vcpu->arch.vcpu_debug_state.dbg_bcr[rd->reg] = rd->val;
+	vcpu->arch.vcpu_debug_state.dbg_bcr[rd->CRm] = rd->val;
 }
 
 static bool trap_wvr(struct kvm_vcpu *vcpu,
 		     struct sys_reg_params *p,
 		     const struct sys_reg_desc *rd)
 {
-	u64 *dbg_reg = &vcpu->arch.vcpu_debug_state.dbg_wvr[rd->reg];
+	u64 *dbg_reg = &vcpu->arch.vcpu_debug_state.dbg_wvr[rd->CRm];
 
 	if (p->is_write)
 		reg_to_dbg(vcpu, p, rd, dbg_reg);
 	else
 		dbg_to_reg(vcpu, p, rd, dbg_reg);
 
-	trace_trap_reg(__func__, rd->reg, p->is_write,
-		vcpu->arch.vcpu_debug_state.dbg_wvr[rd->reg]);
+	trace_trap_reg(__func__, rd->CRm, p->is_write,
+		vcpu->arch.vcpu_debug_state.dbg_wvr[rd->CRm]);
 
 	return true;
 }
···
 static int set_wvr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
 		const struct kvm_one_reg *reg, void __user *uaddr)
 {
-	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_wvr[rd->reg];
+	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_wvr[rd->CRm];
 
 	if (copy_from_user(r, uaddr, KVM_REG_SIZE(reg->id)) != 0)
 		return -EFAULT;
···
 static int get_wvr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
 	const struct kvm_one_reg *reg, void __user *uaddr)
 {
-	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_wvr[rd->reg];
+	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_wvr[rd->CRm];
 
 	if (copy_to_user(uaddr, r, KVM_REG_SIZE(reg->id)) != 0)
 		return -EFAULT;
···
 static void reset_wvr(struct kvm_vcpu *vcpu,
 		      const struct sys_reg_desc *rd)
 {
-	vcpu->arch.vcpu_debug_state.dbg_wvr[rd->reg] = rd->val;
+	vcpu->arch.vcpu_debug_state.dbg_wvr[rd->CRm] = rd->val;
 }
 
 static bool trap_wcr(struct kvm_vcpu *vcpu,
 		     struct sys_reg_params *p,
 		     const struct sys_reg_desc *rd)
 {
-	u64 *dbg_reg = &vcpu->arch.vcpu_debug_state.dbg_wcr[rd->reg];
+	u64 *dbg_reg = &vcpu->arch.vcpu_debug_state.dbg_wcr[rd->CRm];
 
 	if (p->is_write)
 		reg_to_dbg(vcpu, p, rd, dbg_reg);
 	else
 		dbg_to_reg(vcpu, p, rd, dbg_reg);
 
-	trace_trap_reg(__func__, rd->reg, p->is_write, *dbg_reg);
+	trace_trap_reg(__func__, rd->CRm, p->is_write, *dbg_reg);
 
 	return true;
 }
···
 static int set_wcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
 		const struct kvm_one_reg *reg, void __user *uaddr)
 {
-	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_wcr[rd->reg];
+	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_wcr[rd->CRm];
 
 	if (copy_from_user(r, uaddr, KVM_REG_SIZE(reg->id)) != 0)
 		return -EFAULT;
···
 static int get_wcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
 	const struct kvm_one_reg *reg, void __user *uaddr)
 {
-	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_wcr[rd->reg];
+	__u64 *r = &vcpu->arch.vcpu_debug_state.dbg_wcr[rd->CRm];
 
 	if (copy_to_user(uaddr, r, KVM_REG_SIZE(reg->id)) != 0)
 		return -EFAULT;
···
 static void reset_wcr(struct kvm_vcpu *vcpu,
 		      const struct sys_reg_desc *rd)
 {
-	vcpu->arch.vcpu_debug_state.dbg_wcr[rd->reg] = rd->val;
+	vcpu->arch.vcpu_debug_state.dbg_wcr[rd->CRm] = rd->val;
 }
 
 static void reset_amair_el1(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r)

arch/powerpc/include/asm/kvm_host.h

···
 /* PPC-specific vcpu->requests bit members */
 #define KVM_REQ_WATCHDOG	KVM_ARCH_REQ(0)
 #define KVM_REQ_EPR_EXIT	KVM_ARCH_REQ(1)
+#define KVM_REQ_PENDING_TIMER	KVM_ARCH_REQ(2)
 
 #include <linux/mmu_notifier.h>
 

arch/powerpc/kvm/book3s_hv.c

···
 				break;
 			}
 			cur = ktime_get();
-		} while (single_task_running() && ktime_before(cur, stop));
+		} while (kvm_vcpu_can_poll(cur, stop));
 
 		spin_lock(&vc->lock);
 		vc->vcore_state = VCORE_INACTIVE;

arch/x86/include/asm/kvm-x86-ops.h

···
 KVM_X86_OP_NULL(vcpu_blocking)
 KVM_X86_OP_NULL(vcpu_unblocking)
 KVM_X86_OP_NULL(update_pi_irte)
+KVM_X86_OP_NULL(start_assignment)
 KVM_X86_OP_NULL(apicv_post_state_restore)
 KVM_X86_OP_NULL(dy_apicv_has_pending_interrupt)
 KVM_X86_OP_NULL(set_hv_timer)

arch/x86/include/asm/kvm_host.h

···
 
 	int (*update_pi_irte)(struct kvm *kvm, unsigned int host_irq,
 			      uint32_t guest_irq, bool set);
+	void (*start_assignment)(struct kvm *kvm);
 	void (*apicv_post_state_restore)(struct kvm_vcpu *vcpu);
 	bool (*dy_apicv_has_pending_interrupt)(struct kvm_vcpu *vcpu);
 

arch/x86/kvm/emulate.c

···
 	return rc;
 }
 
-int x86_decode_insn(struct x86_emulate_ctxt *ctxt, void *insn, int insn_len)
+int x86_decode_insn(struct x86_emulate_ctxt *ctxt, void *insn, int insn_len, int emulation_type)
 {
 	int rc = X86EMUL_CONTINUE;
 	int mode = ctxt->mode;
···
 
 	ctxt->execute = opcode.u.execute;
 
-	if (unlikely(ctxt->ud) && likely(!(ctxt->d & EmulateOnUD)))
+	if (unlikely(emulation_type & EMULTYPE_TRAP_UD) &&
+	    likely(!(ctxt->d & EmulateOnUD)))
 		return EMULATION_FAILED;
 
 	if (unlikely(ctxt->d &

arch/x86/kvm/hyperv.c

···
 {
 	struct kvm_hv *hv = to_kvm_hv(kvm);
 	u64 gfn;
+	int idx;
 
 	if (hv->hv_tsc_page_status == HV_TSC_PAGE_BROKEN ||
 	    hv->hv_tsc_page_status == HV_TSC_PAGE_UNSET ||
···
 	gfn = hv->hv_tsc_page >> HV_X64_MSR_TSC_REFERENCE_ADDRESS_SHIFT;
 
 	hv->tsc_ref.tsc_sequence = 0;
+
+	/*
+	 * Take the srcu lock as memslots will be accessed to check the gfn
+	 * cache generation against the memslots generation.
+	 */
+	idx = srcu_read_lock(&kvm->srcu);
 	if (kvm_write_guest(kvm, gfn_to_gpa(gfn),
 			    &hv->tsc_ref, sizeof(hv->tsc_ref.tsc_sequence)))
 		hv->hv_tsc_page_status = HV_TSC_PAGE_BROKEN;
+	srcu_read_unlock(&kvm->srcu, idx);
 
 out_unlock:
 	mutex_unlock(&hv->hv_lock);

arch/x86/kvm/kvm_emulate.h

···
 	int interruptibility;
 
 	bool perm_ok; /* do not check permissions if true */
-	bool ud;	/* inject an #UD if host doesn't support insn */
 	bool tf;	/* TF value before instruction (after for syscall/sysret) */
 
 	bool have_exception;
···
 #define X86EMUL_MODE_HOST X86EMUL_MODE_PROT64
 #endif
 
-int x86_decode_insn(struct x86_emulate_ctxt *ctxt, void *insn, int insn_len);
+int x86_decode_insn(struct x86_emulate_ctxt *ctxt, void *insn, int insn_len, int emulation_type);
 bool x86_page_table_writing_insn(struct x86_emulate_ctxt *ctxt);
 #define EMULATION_FAILED -1
 #define EMULATION_OK 0

arch/x86/kvm/lapic.c

···
 	guest_tsc = kvm_read_l1_tsc(vcpu, rdtsc());
 	apic->lapic_timer.advance_expire_delta = guest_tsc - tsc_deadline;
 
+	if (lapic_timer_advance_dynamic) {
+		adjust_lapic_timer_advance(vcpu, apic->lapic_timer.advance_expire_delta);
+		/*
+		 * If the timer fired early, reread the TSC to account for the
+		 * overhead of the above adjustment to avoid waiting longer
+		 * than is necessary.
+		 */
+		if (guest_tsc < tsc_deadline)
+			guest_tsc = kvm_read_l1_tsc(vcpu, rdtsc());
+	}
+
 	if (guest_tsc < tsc_deadline)
 		__wait_lapic_expire(vcpu, tsc_deadline - guest_tsc);
-
-	if (lapic_timer_advance_dynamic)
-		adjust_lapic_timer_advance(vcpu, apic->lapic_timer.advance_expire_delta);
 }
 
 void kvm_wait_lapic_expire(struct kvm_vcpu *vcpu)
···
 	}
 
 	atomic_inc(&apic->lapic_timer.pending);
-	kvm_make_request(KVM_REQ_PENDING_TIMER, vcpu);
+	kvm_make_request(KVM_REQ_UNBLOCK, vcpu);
 	if (from_timer_fn)
 		kvm_vcpu_kick(vcpu);
 }

arch/x86/kvm/mmu/tdp_mmu.c

···
 }
 
 /*
- * Remove write access from all the SPTEs mapping GFNs [start, end). If
- * skip_4k is set, SPTEs that map 4k pages, will not be write-protected.
- * Returns true if an SPTE has been changed and the TLBs need to be flushed.
+ * Remove write access from all SPTEs at or above min_level that map GFNs
+ * [start, end). Returns true if an SPTE has been changed and the TLBs need to
+ * be flushed.
  */
 static bool wrprot_gfn_range(struct kvm *kvm, struct kvm_mmu_page *root,
 			     gfn_t start, gfn_t end, int min_level)

arch/x86/kvm/svm/avic.c

···
 #include "svm.h"
 
 /* enable / disable AVIC */
-int avic;
-#ifdef CONFIG_X86_LOCAL_APIC
-module_param(avic, int, S_IRUGO);
-#endif
+bool avic;
+module_param(avic, bool, S_IRUGO);
 
 #define SVM_AVIC_DOORBELL	0xc001011b
 

arch/x86/kvm/svm/svm.c

···
 	}
 
 	if (avic) {
-		if (!npt_enabled ||
-		    !boot_cpu_has(X86_FEATURE_AVIC) ||
-		    !IS_ENABLED(CONFIG_X86_LOCAL_APIC)) {
+		if (!npt_enabled || !boot_cpu_has(X86_FEATURE_AVIC)) {
 			avic = false;
 		} else {
 			pr_info("AVIC enabled\n");

arch/x86/kvm/svm/svm.h

···
 
 #define VMCB_AVIC_APIC_BAR_MASK		0xFFFFFFFFFF000ULL
 
-extern int avic;
+extern bool avic;
 
 static inline void avic_update_vapic_bar(struct vcpu_svm *svm, u64 data)
 {

arch/x86/kvm/vmx/capabilities.h

···
 
 static inline bool cpu_has_vmx_posted_intr(void)
 {
-	return IS_ENABLED(CONFIG_X86_LOCAL_APIC) &&
-		vmcs_config.pin_based_exec_ctrl & PIN_BASED_POSTED_INTR;
+	return vmcs_config.pin_based_exec_ctrl & PIN_BASED_POSTED_INTR;
 }
 
 static inline bool cpu_has_load_ia32_efer(void)

arch/x86/kvm/vmx/posted_intr.c

···
 
 
 /*
+ * Bail out of the block loop if the VM has an assigned
+ * device, but the blocking vCPU didn't reconfigure the
+ * PI.NV to the wakeup vector, i.e. the assigned device
+ * came along after the initial check in pi_pre_block().
+ */
+void vmx_pi_start_assignment(struct kvm *kvm)
+{
+	if (!irq_remapping_cap(IRQ_POSTING_CAP))
+		return;
+
+	kvm_make_all_cpus_request(kvm, KVM_REQ_UNBLOCK);
+}
+
+/*
  * pi_update_irte - set IRTE for Posted-Interrupts
  *
  * @kvm: kvm

arch/x86/kvm/vmx/posted_intr.h

···
 bool pi_has_pending_interrupt(struct kvm_vcpu *vcpu);
 int pi_update_irte(struct kvm *kvm, unsigned int host_irq, uint32_t guest_irq,
 		   bool set);
+void vmx_pi_start_assignment(struct kvm *kvm);
 
 #endif /* __KVM_X86_VMX_POSTED_INTR_H */

arch/x86/kvm/vmx/vmx.c

···
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
 	struct kvm_run *kvm_run = vcpu->run;
 	u32 intr_info, ex_no, error_code;
-	unsigned long cr2, rip, dr6;
+	unsigned long cr2, dr6;
 	u32 vect_info;
 
 	vect_info = vmx->idt_vectoring_info;
···
 		vmx->vcpu.arch.event_exit_inst_len =
 			vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
 		kvm_run->exit_reason = KVM_EXIT_DEBUG;
-		rip = kvm_rip_read(vcpu);
-		kvm_run->debug.arch.pc = vmcs_readl(GUEST_CS_BASE) + rip;
+		kvm_run->debug.arch.pc = kvm_get_linear_rip(vcpu);
 		kvm_run->debug.arch.exception = ex_no;
 		break;
 	case AC_VECTOR:
···
 	.nested_ops = &vmx_nested_ops,
 
 	.update_pi_irte = pi_update_irte,
+	.start_assignment = vmx_pi_start_assignment,
 
 #ifdef CONFIG_X86_64
 	.set_hv_timer = vmx_set_hv_timer,

arch/x86/kvm/x86.c

···
 				       st->preempted & KVM_VCPU_FLUSH_TLB);
 		if (xchg(&st->preempted, 0) & KVM_VCPU_FLUSH_TLB)
 			kvm_vcpu_flush_tlb_guest(vcpu);
+	} else {
+		st->preempted = 0;
 	}
 
 	vcpu->arch.st.preempted = 0;
···
 	BUILD_BUG_ON(HF_SMM_MASK != X86EMUL_SMM_MASK);
 	BUILD_BUG_ON(HF_SMM_INSIDE_NMI_MASK != X86EMUL_SMM_INSIDE_NMI_MASK);
 
+	ctxt->interruptibility = 0;
+	ctxt->have_exception = false;
+	ctxt->exception.vector = -1;
+	ctxt->perm_ok = false;
+
 	init_decode_cache(ctxt);
 	vcpu->arch.emulate_regs_need_sync_from_vcpu = false;
 }
···
 	    kvm_vcpu_check_breakpoint(vcpu, &r))
 		return r;
 
-	ctxt->interruptibility = 0;
-	ctxt->have_exception = false;
-	ctxt->exception.vector = -1;
-	ctxt->perm_ok = false;
-
-	ctxt->ud = emulation_type & EMULTYPE_TRAP_UD;
-
-	r = x86_decode_insn(ctxt, insn, insn_len);
+	r = x86_decode_insn(ctxt, insn, insn_len, emulation_type);
 
 	trace_kvm_emulate_insn_start(vcpu);
 	++vcpu->stat.insn_emulation;
···
 	struct kvm_apic_map *map;
 
 	vcpu->stat.directed_yield_attempted++;
+
+	if (single_task_running())
+		goto no_yield;
 
 	rcu_read_lock();
 	map = rcu_dereference(vcpu->kvm->arch.apic_map);
···
 		if (r <= 0)
 			break;
 
-		kvm_clear_request(KVM_REQ_PENDING_TIMER, vcpu);
+		kvm_clear_request(KVM_REQ_UNBLOCK, vcpu);
 		if (kvm_cpu_has_pending_timer(vcpu))
 			kvm_inject_pending_timer_irqs(vcpu);
 
···
 	kvm_update_dr7(vcpu);
 
 	if (vcpu->guest_debug & KVM_GUESTDBG_SINGLESTEP)
-		vcpu->arch.singlestep_rip = kvm_rip_read(vcpu) +
-			get_segment_base(vcpu, VCPU_SREG_CS);
+		vcpu->arch.singlestep_rip = kvm_get_linear_rip(vcpu);
 
 	/*
 	 * Trigger an rflags update that will inject or remove the trace
···
 
 void kvm_arch_start_assignment(struct kvm *kvm)
 {
-	atomic_inc(&kvm->arch.assigned_device_count);
+	if (atomic_inc_return(&kvm->arch.assigned_device_count) == 1)
+		static_call_cond(kvm_x86_start_assignment)(kvm);
 }
 EXPORT_SYMBOL_GPL(kvm_arch_start_assignment);
 

include/linux/kvm_host.h

···
 #include <linux/spinlock.h>
 #include <linux/signal.h>
 #include <linux/sched.h>
+#include <linux/sched/stat.h>
 #include <linux/bug.h>
 #include <linux/minmax.h>
 #include <linux/mm.h>
···
  */
 #define KVM_REQ_TLB_FLUSH         (0 | KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
 #define KVM_REQ_MMU_RELOAD        (1 | KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
-#define KVM_REQ_PENDING_TIMER     2
+#define KVM_REQ_UNBLOCK           2
 #define KVM_REQ_UNHALT            3
 #define KVM_REQUEST_ARCH_BASE     8
 
···
 static inline bool kvm_vcpu_mapped(struct kvm_host_map *map)
 {
 	return !!map->hva;
+}
+
+static inline bool kvm_vcpu_can_poll(ktime_t cur, ktime_t stop)
+{
+	return single_task_running() && !need_resched() && ktime_before(cur, stop);
 }
 
 /*

include/uapi/linux/kvm.h

···
  * Note: you must update KVM_API_VERSION if you change this interface.
  */
 
+#include <linux/const.h>
 #include <linux/types.h>
 #include <linux/compiler.h>
 #include <linux/ioctl.h>
···
  * conversion after harvesting an entry.  Also, it must not skip any
  * dirty bits, so that dirty bits are always harvested in sequence.
  */
-#define KVM_DIRTY_GFN_F_DIRTY           BIT(0)
-#define KVM_DIRTY_GFN_F_RESET           BIT(1)
+#define KVM_DIRTY_GFN_F_DIRTY           _BITUL(0)
+#define KVM_DIRTY_GFN_F_RESET           _BITUL(1)
 #define KVM_DIRTY_GFN_F_MASK            0x3
 
 /*

tools/include/uapi/linux/kvm.h

···
  * Note: you must update KVM_API_VERSION if you change this interface.
  */
 
+#include <linux/const.h>
 #include <linux/types.h>
 #include <linux/compiler.h>
 #include <linux/ioctl.h>
···
  * conversion after harvesting an entry.  Also, it must not skip any
  * dirty bits, so that dirty bits are always harvested in sequence.
  */
-#define KVM_DIRTY_GFN_F_DIRTY           BIT(0)
-#define KVM_DIRTY_GFN_F_RESET           BIT(1)
+#define KVM_DIRTY_GFN_F_DIRTY           _BITUL(0)
+#define KVM_DIRTY_GFN_F_RESET           _BITUL(1)
 #define KVM_DIRTY_GFN_F_MASK            0x3
 
 /*

tools/testing/selftests/kvm/.gitignore

···
 /kvm_create_max_vcpus
 /kvm_page_table_test
 /memslot_modification_stress_test
+/memslot_perf_test
 /set_memory_region_test
 /steal_time

tools/testing/selftests/kvm/Makefile

···
 	UNAME_M := s390x
 endif
 
-LIBKVM = lib/assert.c lib/elf.c lib/io.c lib/kvm_util.c lib/sparsebit.c lib/test_util.c lib/guest_modes.c lib/perf_test_util.c
+LIBKVM = lib/assert.c lib/elf.c lib/io.c lib/kvm_util.c lib/rbtree.c lib/sparsebit.c lib/test_util.c lib/guest_modes.c lib/perf_test_util.c
 LIBKVM_x86_64 = lib/x86_64/processor.c lib/x86_64/vmx.c lib/x86_64/svm.c lib/x86_64/ucall.c lib/x86_64/handlers.S
 LIBKVM_aarch64 = lib/aarch64/processor.c lib/aarch64/ucall.c
 LIBKVM_s390x = lib/s390x/processor.c lib/s390x/ucall.c lib/s390x/diag318_test_handler.c
···
 TEST_GEN_PROGS_x86_64 += kvm_create_max_vcpus
 TEST_GEN_PROGS_x86_64 += kvm_page_table_test
 TEST_GEN_PROGS_x86_64 += memslot_modification_stress_test
+TEST_GEN_PROGS_x86_64 += memslot_perf_test
 TEST_GEN_PROGS_x86_64 += set_memory_region_test
 TEST_GEN_PROGS_x86_64 += steal_time
 

tools/testing/selftests/kvm/demand_paging_test.c

···
 
 #define _GNU_SOURCE /* for pipe2 */
 
+#include <inttypes.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <time.h>
···
 
 static int nr_vcpus = 1;
 static uint64_t guest_percpu_mem_size = DEFAULT_PER_VCPU_MEM_SIZE;
+static size_t demand_paging_size;
 static char *guest_data_prototype;
 
 static void *vcpu_worker(void *data)
···
 	return NULL;
 }
 
-static int handle_uffd_page_request(int uffd, uint64_t addr)
+static int handle_uffd_page_request(int uffd_mode, int uffd, uint64_t addr)
 {
-	pid_t tid;
+	pid_t tid = syscall(__NR_gettid);
 	struct timespec start;
 	struct timespec ts_diff;
-	struct uffdio_copy copy;
 	int r;
-
-	tid = syscall(__NR_gettid);
-
-	copy.src = (uint64_t)guest_data_prototype;
-	copy.dst = addr;
-	copy.len = perf_test_args.host_page_size;
-	copy.mode = 0;
 
 	clock_gettime(CLOCK_MONOTONIC, &start);
 
-	r = ioctl(uffd, UFFDIO_COPY, &copy);
-	if (r == -1) {
-		pr_info("Failed Paged in 0x%lx from thread %d with errno: %d\n",
-			addr, tid, errno);
-		return r;
+	if (uffd_mode == UFFDIO_REGISTER_MODE_MISSING) {
+		struct uffdio_copy copy;
+
+		copy.src = (uint64_t)guest_data_prototype;
+		copy.dst = addr;
+		copy.len = demand_paging_size;
+		copy.mode = 0;
+
+		r = ioctl(uffd, UFFDIO_COPY, &copy);
+		if (r == -1) {
+			pr_info("Failed UFFDIO_COPY in 0x%lx from thread %d with errno: %d\n",
+				addr, tid, errno);
+			return r;
+		}
+	} else if (uffd_mode == UFFDIO_REGISTER_MODE_MINOR) {
+		struct uffdio_continue cont = {0};
+
+		cont.range.start = addr;
+		cont.range.len = demand_paging_size;
+
+		r = ioctl(uffd, UFFDIO_CONTINUE, &cont);
+		if (r == -1) {
+			pr_info("Failed UFFDIO_CONTINUE in 0x%lx from thread %d with errno: %d\n",
+				addr, tid, errno);
+			return r;
+		}
+	} else {
+		TEST_FAIL("Invalid uffd mode %d", uffd_mode);
 	}
 
 	ts_diff = timespec_elapsed(start);
 
-	PER_PAGE_DEBUG("UFFDIO_COPY %d \t%ld ns\n", tid,
+	PER_PAGE_DEBUG("UFFD page-in %d \t%ld ns\n", tid,
 		       timespec_to_ns(ts_diff));
 	PER_PAGE_DEBUG("Paged in %ld bytes at 0x%lx from thread %d\n",
-		       perf_test_args.host_page_size, addr, tid);
+		       demand_paging_size, addr, tid);
 
 	return 0;
 }
···
 bool quit_uffd_thread;
 
 struct uffd_handler_args {
+	int uffd_mode;
 	int uffd;
 	int pipefd;
 	useconds_t delay;
···
 		if (r == -1) {
 			if (errno == EAGAIN)
 				continue;
-			pr_info("Read of uffd gor errno %d", errno);
+			pr_info("Read of uffd got errno %d\n", errno);
 			return NULL;
 		}
 
···
 		if (delay)
 			usleep(delay);
 		addr =  msg.arg.pagefault.address;
-		r = handle_uffd_page_request(uffd, addr);
+		r = handle_uffd_page_request(uffd_args->uffd_mode, uffd, addr);
 		if (r < 0)
 			return NULL;
 		pages++;
···
 	return NULL;
 }
 
-static int setup_demand_paging(struct kvm_vm *vm,
-			       pthread_t *uffd_handler_thread, int pipefd,
-			       useconds_t uffd_delay,
-			       struct uffd_handler_args *uffd_args,
-			       void *hva, uint64_t len)
+static void setup_demand_paging(struct kvm_vm *vm,
+				pthread_t *uffd_handler_thread, int pipefd,
+				int uffd_mode, useconds_t uffd_delay,
+				struct uffd_handler_args *uffd_args,
+				void *hva, void *alias, uint64_t len)
 {
+	bool is_minor = (uffd_mode == UFFDIO_REGISTER_MODE_MINOR);
 	int uffd;
 	struct uffdio_api uffdio_api;
 	struct uffdio_register uffdio_register;
+	uint64_t expected_ioctls = ((uint64_t) 1) << _UFFDIO_COPY;
+
+	PER_PAGE_DEBUG("Userfaultfd %s mode, faults resolved with %s\n",
+		       is_minor ? "MINOR" : "MISSING",
+		       is_minor ? "UFFDIO_CONINUE" : "UFFDIO_COPY");
+
+	/* In order to get minor faults, prefault via the alias. */
+	if (is_minor) {
+		size_t p;
+
+		expected_ioctls = ((uint64_t) 1) << _UFFDIO_CONTINUE;
+
+		TEST_ASSERT(alias != NULL, "Alias required for minor faults");
+		for (p = 0; p < (len / demand_paging_size); ++p) {
+			memcpy(alias + (p * demand_paging_size),
+			       guest_data_prototype, demand_paging_size);
+		}
+	}
 
 	uffd = syscall(__NR_userfaultfd, O_CLOEXEC | O_NONBLOCK);
-	if (uffd == -1) {
-		pr_info("uffd creation failed\n");
-		return -1;
-	}
+	TEST_ASSERT(uffd >= 0, "uffd creation failed, errno: %d", errno);
 
 	uffdio_api.api = UFFD_API;
 	uffdio_api.features = 0;
-	if (ioctl(uffd, UFFDIO_API, &uffdio_api) == -1) {
-		pr_info("ioctl uffdio_api failed\n");
-		return -1;
-	}
+	TEST_ASSERT(ioctl(uffd, UFFDIO_API, &uffdio_api) != -1,
+		    "ioctl UFFDIO_API failed: %" PRIu64,
+		    (uint64_t)uffdio_api.api);
 
 	uffdio_register.range.start = (uint64_t)hva;
 	uffdio_register.range.len = len;
-	uffdio_register.mode = UFFDIO_REGISTER_MODE_MISSING;
-	if (ioctl(uffd, UFFDIO_REGISTER, &uffdio_register) == -1) {
-		pr_info("ioctl uffdio_register failed\n");
-		return -1;
-	}
+	uffdio_register.mode = uffd_mode;
+	TEST_ASSERT(ioctl(uffd, UFFDIO_REGISTER, &uffdio_register) != -1,
+		    "ioctl UFFDIO_REGISTER failed");
+	TEST_ASSERT((uffdio_register.ioctls & expected_ioctls) ==
+		    expected_ioctls, "missing userfaultfd ioctls");
 
-	if ((uffdio_register.ioctls & UFFD_API_RANGE_IOCTLS) !=
-			UFFD_API_RANGE_IOCTLS) {
-		pr_info("unexpected userfaultfd ioctl set\n");
-		return -1;
-	}
-
+	uffd_args->uffd_mode = uffd_mode;
 	uffd_args->uffd = uffd;
 	uffd_args->pipefd = pipefd;
 	uffd_args->delay = uffd_delay;
···
 
 	PER_VCPU_DEBUG("Created uffd thread for HVA range [%p, %p)\n",
 		       hva, hva + len);
-
-	return 0;
 }
 
 struct test_params {
-	bool use_uffd;
+	int uffd_mode;
 	useconds_t uffd_delay;
+	enum vm_mem_backing_src_type src_type;
 	bool partition_vcpu_memory_access;
 };
 
···
 	int r;
 
 	vm = perf_test_create_vm(mode, nr_vcpus, guest_percpu_mem_size,
-				 VM_MEM_SRC_ANONYMOUS);
+				 p->src_type);
 
 	perf_test_args.wr_fract = 1;
 
-	guest_data_prototype = malloc(perf_test_args.host_page_size);
+	demand_paging_size = get_backing_src_pagesz(p->src_type);
+
+	guest_data_prototype = malloc(demand_paging_size);
 	TEST_ASSERT(guest_data_prototype,
 		    "Failed to allocate buffer for guest data pattern");
-	memset(guest_data_prototype, 0xAB, perf_test_args.host_page_size);
+	memset(guest_data_prototype, 0xAB, demand_paging_size);
 
 	vcpu_threads = malloc(nr_vcpus * sizeof(*vcpu_threads));
 	TEST_ASSERT(vcpu_threads, "Memory allocation failed");
···
 	perf_test_setup_vcpus(vm, nr_vcpus, guest_percpu_mem_size,
 			      p->partition_vcpu_memory_access);
 
-	if (p->use_uffd) {
+	if (p->uffd_mode) {
 		uffd_handler_threads =
 			malloc(nr_vcpus * sizeof(*uffd_handler_threads));
 		TEST_ASSERT(uffd_handler_threads, "Memory allocation failed");
···
 		for (vcpu_id = 0; vcpu_id < nr_vcpus; vcpu_id++) {
 			vm_paddr_t vcpu_gpa;
 			void *vcpu_hva;
+			void *vcpu_alias;
 			uint64_t vcpu_mem_size;
 
 
···
 			PER_VCPU_DEBUG("Added VCPU %d with test mem gpa [%lx, %lx)\n",
 				       vcpu_id, vcpu_gpa, vcpu_gpa + vcpu_mem_size);
 
-			/* Cache the HVA pointer of the region */
+			/* Cache the host addresses of the region */
 			vcpu_hva = addr_gpa2hva(vm, vcpu_gpa);
+			vcpu_alias = addr_gpa2alias(vm, vcpu_gpa);
 
 			/*
 			 * Set up user fault fd to handle demand paging
···
 				  O_CLOEXEC | O_NONBLOCK);
 			TEST_ASSERT(!r, "Failed to set up pipefd");
 
-			r = setup_demand_paging(vm,
-						&uffd_handler_threads[vcpu_id],
-						pipefds[vcpu_id * 2],
-						p->uffd_delay, &uffd_args[vcpu_id],
-						vcpu_hva, vcpu_mem_size);
-			if (r < 0)
-				exit(-r);
+			setup_demand_paging(vm, &uffd_handler_threads[vcpu_id],
+					    pipefds[vcpu_id * 2], p->uffd_mode,
+					    p->uffd_delay, &uffd_args[vcpu_id],
+					    vcpu_hva, vcpu_alias,
+					    vcpu_mem_size);
 		}
 	}
 
···
 
 	pr_info("All vCPU threads joined\n");
 
-	if (p->use_uffd) {
+	if (p->uffd_mode) {
 		char c;
 
 		/* Tell the user fault fd handler threads to quit */
···
 
 	free(guest_data_prototype);
 	free(vcpu_threads);
-	if (p->use_uffd) {
+	if (p->uffd_mode) {
 		free(uffd_handler_threads);
 		free(uffd_args);
 		free(pipefds);
···
 static void help(char *name)
 {
 	puts("");
-	printf("usage: %s [-h] [-m mode] [-u] [-d uffd_delay_usec]\n"
-	       "          [-b memory] [-v vcpus] [-o]\n", name);
+	printf("usage: %s [-h] [-m vm_mode] [-u uffd_mode] [-d uffd_delay_usec]\n"
+	       "          [-b memory] [-t type] [-v vcpus] [-o]\n", name);
 	guest_modes_help();
-	printf(" -u: use User Fault FD to handle vCPU page\n"
-	       "     faults.\n");
+	printf(" -u: use userfaultfd to handle vCPU page faults. Mode is a\n"
+	       "     UFFD registration mode: 'MISSING' or 'MINOR'.\n");
 	printf(" -d: add a delay in usec to the User Fault\n"
 	       "     FD handler to simulate demand paging\n"
 	       "     overheads. Ignored without -u.\n");
 	printf(" -b: specify the size of the memory region which should be\n"
 	       "     demand paged by each vCPU. e.g. 10M or 3G.\n"
 	       "     Default: 1G\n");
+	printf(" -t: The type of backing memory to use. Default: anonymous\n");
+	backing_src_help();
 	printf(" -v: specify the number of vCPUs to run.\n");
 	printf(" -o: Overlap guest memory accesses instead of partitioning\n"
 	       "     them into a separate region of memory for each vCPU.\n");
···
 {
 	int max_vcpus = kvm_check_cap(KVM_CAP_MAX_VCPUS);
 	struct test_params p = {
+		.src_type = VM_MEM_SRC_ANONYMOUS,
 		.partition_vcpu_memory_access = true,
 	};
 	int opt;
 
 	guest_modes_append_default();
 
-	while ((opt = getopt(argc, argv, "hm:ud:b:v:o")) != -1) {
+	while ((opt = getopt(argc, argv, "hm:u:d:b:t:v:o")) != -1) {
 		switch (opt) {
 		case 'm':
 			guest_modes_cmdline(optarg);
 			break;
 		case 'u':
-			p.use_uffd = true;
+			if (!strcmp("MISSING", optarg))
+				p.uffd_mode = UFFDIO_REGISTER_MODE_MISSING;
+			else if (!strcmp("MINOR", optarg))
+				p.uffd_mode = UFFDIO_REGISTER_MODE_MINOR;
+			TEST_ASSERT(p.uffd_mode, "UFFD mode must be 'MISSING' or 'MINOR'.");
 			break;
 		case 'd':
 			p.uffd_delay = strtoul(optarg, NULL, 0);
···
 			break;
 		case 'b':
 			guest_percpu_mem_size = parse_size(optarg);
+			break;
+		case 't':
+			p.src_type = parse_backing_src_type(optarg);
 			break;
 		case 'v':
 			nr_vcpus = atoi(optarg);
···
 			help(argv[0]);
 			break;
 		}
+	}
+
+	if (p.uffd_mode == UFFDIO_REGISTER_MODE_MINOR &&
+	    !backing_src_is_shared(p.src_type)) {
+		TEST_FAIL("userfaultfd MINOR mode requires shared memory; pick a different -t");
 	}
 
 	for_each_guest_mode(run_test, &p);

tools/testing/selftests/kvm/hardware_disable_test.c

···
 	TEST_ASSERT(false, "%s: [%d] child escaped the ninja\n", __func__, run);
 }
 
+void wait_for_child_setup(pid_t pid)
+{
+	/*
+	 * Wait for the child to post to the semaphore, but wake up periodically
+	 * to check if the child exited prematurely.
+	 */
+	for (;;) {
+		const struct timespec wait_period = { .tv_sec = 1 };
+		int status;
+
+		if (!sem_timedwait(sem, &wait_period))
+			return;
+
+		/* Child is still running, keep waiting. */
+		if (pid != waitpid(pid, &status, WNOHANG))
+			continue;
+
+		/*
+		 * Child is no longer running, which is not expected.
+		 *
+		 * If it exited with a non-zero status, we explicitly forward
+		 * the child's status in case it exited with KSFT_SKIP.
+		 */
+		if (WIFEXITED(status))
+			exit(WEXITSTATUS(status));
+		else
+			TEST_ASSERT(false, "Child exited unexpectedly");
+	}
+}
+
 int main(int argc, char **argv)
 {
 	uint32_t i;
···
 			run_test(i); /* This function always exits */
 
 		pr_debug("%s: [%d] waiting semaphore\n", __func__, i);
-		sem_wait(sem);
+		wait_for_child_setup(pid);
 		r = (rand() % DELAY_US_MAX) + 1;
 		pr_debug("%s: [%d] waiting %dus\n", __func__, i, r);
 		usleep(r);

tools/testing/selftests/kvm/include/kvm_util.h

···
 };
 extern const struct vm_guest_mode_params vm_guest_mode_params[];
 
+int open_kvm_dev_path_or_exit(void);
 int kvm_check_cap(long cap);
 int vm_enable_cap(struct kvm_vm *vm, struct kvm_enable_cap *cap);
 int vcpu_enable_cap(struct kvm_vm *vm, uint32_t vcpu_id,
···
 void *addr_gpa2hva(struct kvm_vm *vm, vm_paddr_t gpa);
 void *addr_gva2hva(struct kvm_vm *vm, vm_vaddr_t gva);
 vm_paddr_t addr_hva2gpa(struct kvm_vm *vm, void *hva);
+void *addr_gpa2alias(struct kvm_vm *vm, vm_paddr_t gpa);
 
 /*
  * Address Guest Virtual to Guest Physical
···
 
 unsigned int vm_get_page_size(struct kvm_vm *vm);
 unsigned int vm_get_page_shift(struct kvm_vm *vm);
-unsigned int vm_get_max_gfn(struct kvm_vm *vm);
+uint64_t vm_get_max_gfn(struct kvm_vm *vm);
 int vm_get_fd(struct kvm_vm *vm);
 
 unsigned int vm_calc_num_guest_pages(enum vm_guest_mode mode, size_t size);

tools/testing/selftests/kvm/include/test_util.h

···
 #include <errno.h>
 #include <unistd.h>
 #include <fcntl.h>
+#include <sys/mman.h>
 #include "kselftest.h"
 
 static inline int _no_printf(const char *format, ...) { return 0; }
···
 	VM_MEM_SRC_ANONYMOUS_HUGETLB_1GB,
 	VM_MEM_SRC_ANONYMOUS_HUGETLB_2GB,
 	VM_MEM_SRC_ANONYMOUS_HUGETLB_16GB,
+	VM_MEM_SRC_SHMEM,
+	VM_MEM_SRC_SHARED_HUGETLB,
 	NUM_SRC_TYPES,
 };
 
···
 size_t get_backing_src_pagesz(uint32_t i);
 void backing_src_help(void);
 enum vm_mem_backing_src_type parse_backing_src_type(const char *type_name);
+
+/*
+ * Whether or not the given source type is shared memory (as opposed to
+ * anonymous).
+ */
+static inline bool backing_src_is_shared(enum vm_mem_backing_src_type t)
+{
+	return vm_mem_backing_src_alias(t)->flag & MAP_SHARED;
+}
 
 #endif /* SELFTEST_KVM_TEST_UTIL_H */

tools/testing/selftests/kvm/lib/kvm_util.c

···
 }
 
 /*
+ * Open KVM_DEV_PATH if available, otherwise exit the entire program.
+ *
+ * Input Args:
+ *   flags - The flags to pass when opening KVM_DEV_PATH.
+ *
+ * Return:
+ *   The opened file descriptor of /dev/kvm.
+ */
+static int _open_kvm_dev_path_or_exit(int flags)
+{
+	int fd;
+
+	fd = open(KVM_DEV_PATH, flags);
+	if (fd < 0) {
+		print_skip("%s not available, is KVM loaded? (errno: %d)",
+			   KVM_DEV_PATH, errno);
+		exit(KSFT_SKIP);
+	}
+
+	return fd;
+}
+
+int open_kvm_dev_path_or_exit(void)
+{
+	return _open_kvm_dev_path_or_exit(O_RDONLY);
+}
+
+/*
  * Capability
  *
  * Input Args:
···
 	int ret;
 	int kvm_fd;
 
-	kvm_fd = open(KVM_DEV_PATH, O_RDONLY);
-	if (kvm_fd < 0)
-		exit(KSFT_SKIP);
-
+	kvm_fd = open_kvm_dev_path_or_exit();
 	ret = ioctl(kvm_fd, KVM_CHECK_EXTENSION, cap);
 	TEST_ASSERT(ret != -1, "KVM_CHECK_EXTENSION IOCTL failed,\n"
 		"  rc: %i errno: %i", ret, errno);
···
 
 static void vm_open(struct kvm_vm *vm, int perm)
 {
-	vm->kvm_fd = open(KVM_DEV_PATH, perm);
-	if (vm->kvm_fd < 0)
-		exit(KSFT_SKIP);
+	vm->kvm_fd = _open_kvm_dev_path_or_exit(perm);
 
 	if (!kvm_check_cap(KVM_CAP_IMMEDIATE_EXIT)) {
 		print_skip("immediate_exit not available");
···
 	TEST_ASSERT(vm != NULL, "Insufficient Memory");
 
 	INIT_LIST_HEAD(&vm->vcpus);
-	INIT_LIST_HEAD(&vm->userspace_mem_regions);
+	vm->regions.gpa_tree = RB_ROOT;
+	vm->regions.hva_tree = RB_ROOT;
+	hash_init(vm->regions.slot_hash);
 
 	vm->mode = mode;
 	vm->type = 0;
···
 	 */
 	uint64_t vcpu_pages = (DEFAULT_STACK_PGS + num_percpu_pages) * nr_vcpus;
 	uint64_t extra_pg_pages = (extra_mem_pages + vcpu_pages) / PTES_PER_MIN_PAGE * 2;
-	uint64_t pages = DEFAULT_GUEST_PHY_PAGES + vcpu_pages + extra_pg_pages;
+	uint64_t pages = DEFAULT_GUEST_PHY_PAGES + extra_mem_pages + vcpu_pages + extra_pg_pages;
 	struct kvm_vm *vm;
 	int i;
 
···
  */
 void kvm_vm_restart(struct kvm_vm *vmp, int perm)
 {
+	int ctr;
 	struct userspace_mem_region *region;
 
 	vm_open(vmp, perm);
 	if (vmp->has_irqchip)
 		vm_create_irqchip(vmp);
 
-	list_for_each_entry(region, &vmp->userspace_mem_regions, list) {
+	hash_for_each(vmp->regions.slot_hash, ctr, region, slot_node) {
 		int ret = ioctl(vmp->fd, KVM_SET_USER_MEMORY_REGION, &region->region);
 		TEST_ASSERT(ret == 0, "KVM_SET_USER_MEMORY_REGION IOCTL failed,\n"
 			    "  rc: %i errno: %i\n"
···
 static struct userspace_mem_region *
 userspace_mem_region_find(struct kvm_vm *vm, uint64_t start, uint64_t end)
 {
-	struct userspace_mem_region *region;
+	struct rb_node *node;
 
-	list_for_each_entry(region, &vm->userspace_mem_regions, list) {
+	for (node = vm->regions.gpa_tree.rb_node; node; ) {
+		struct userspace_mem_region *region =
+			container_of(node, struct userspace_mem_region, gpa_node);
 		uint64_t existing_start = region->region.guest_phys_addr;
 		uint64_t existing_end = region->region.guest_phys_addr
 			+ region->region.memory_size - 1;
 		if (start <= existing_end && end >= existing_start)
 			return region;
+
+		if (start < existing_start)
+			node = node->rb_left;
+		else
+			node = node->rb_right;
 	}
 
 	return NULL;
···
 }
 
 static void __vm_mem_region_delete(struct kvm_vm *vm,
-				   struct userspace_mem_region *region)
+				   struct userspace_mem_region *region,
+				   bool unlink)
 {
 	int ret;
 
-	list_del(&region->list);
+	if (unlink) {
+		rb_erase(&region->gpa_node, &vm->regions.gpa_tree);
+		rb_erase(&region->hva_node, &vm->regions.hva_tree);
+		hash_del(&region->slot_node);
+	}
 
 	region->region.memory_size = 0;
 	ret = ioctl(vm->fd, KVM_SET_USER_MEMORY_REGION, &region->region);
···
  */
 void kvm_vm_free(struct kvm_vm *vmp)
 {
-	struct userspace_mem_region *region, *tmp;
+	int ctr;
+	struct hlist_node *node;
+	struct userspace_mem_region *region;
 
 	if (vmp == NULL)
 		return;
 
 	/* Free userspace_mem_regions. */
-	list_for_each_entry_safe(region, tmp, &vmp->userspace_mem_regions, list)
-		__vm_mem_region_delete(vmp, region);
+	hash_for_each_safe(vmp->regions.slot_hash, ctr, node, region, slot_node)
+		__vm_mem_region_delete(vmp, region, false);
 
 	/* Free sparsebit arrays. */
 	sparsebit_free(&vmp->vpages_valid);
···
 	return 0;
 }
 
+static void vm_userspace_mem_region_gpa_insert(struct rb_root *gpa_tree,
+					       struct userspace_mem_region *region)
+{
+	struct rb_node **cur, *parent;
+
+	for (cur = &gpa_tree->rb_node, parent = NULL; *cur; ) {
+		struct userspace_mem_region *cregion;
+
+		cregion = container_of(*cur, typeof(*cregion), gpa_node);
+		parent = *cur;
+		if (region->region.guest_phys_addr <
+		    cregion->region.guest_phys_addr)
+			cur = &(*cur)->rb_left;
+		else {
+			TEST_ASSERT(region->region.guest_phys_addr !=
+				    cregion->region.guest_phys_addr,
+				    "Duplicate GPA in region tree");
+
+			cur = &(*cur)->rb_right;
+		}
+	}
+
+	rb_link_node(&region->gpa_node, parent, cur);
+	rb_insert_color(&region->gpa_node, gpa_tree);
+}
+
+static void vm_userspace_mem_region_hva_insert(struct rb_root *hva_tree,
+					       struct userspace_mem_region *region)
+{
+	struct rb_node **cur, *parent;
+
+	for (cur = &hva_tree->rb_node, parent = NULL; *cur; ) {
+		struct userspace_mem_region *cregion;
+
+		cregion = container_of(*cur, typeof(*cregion), hva_node);
+		parent = *cur;
+		if (region->host_mem < cregion->host_mem)
+			cur = &(*cur)->rb_left;
+		else {
+			TEST_ASSERT(region->host_mem !=
+				    cregion->host_mem,
+				    "Duplicate HVA in region tree");
+
+			cur = &(*cur)->rb_right;
+		}
+	}
+
+	rb_link_node(&region->hva_node, parent, cur);
+	rb_insert_color(&region->hva_node, hva_tree);
+}
+
 /*
  * VM Userspace Memory Region Add
  *
  * Input Args:
  *   vm - Virtual Machine
- *   backing_src - Storage source for this region.
- *                 NULL to use anonymous memory.
+ *   src_type - Storage source for this region.
+ *              NULL to use anonymous memory.
  *   guest_paddr - Starting guest physical address
  *   slot - KVM region slot
  *   npages - Number of physical pages
···
 			(uint64_t) region->region.memory_size);
 
 	/* Confirm no region with the requested slot already exists. */
-	list_for_each_entry(region, &vm->userspace_mem_regions, list) {
+	hash_for_each_possible(vm->regions.slot_hash, region, slot_node,
+			       slot) {
 		if (region->region.slot != slot)
 			continue;
 
···
 	if (alignment > 1)
 		region->mmap_size += alignment;
 
+	region->fd = -1;
+	if (backing_src_is_shared(src_type)) {
+		int memfd_flags = MFD_CLOEXEC;
+
+		if (src_type == VM_MEM_SRC_SHARED_HUGETLB)
+			memfd_flags |= MFD_HUGETLB;
+
+		region->fd = memfd_create("kvm_selftest", memfd_flags);
+		TEST_ASSERT(region->fd != -1,
+			    "memfd_create failed, errno: %i", errno);
+
+		ret = ftruncate(region->fd, region->mmap_size);
+		TEST_ASSERT(ret == 0, "ftruncate failed, errno: %i", errno);
+
+		ret = fallocate(region->fd,
+				FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE, 0,
+				region->mmap_size);
+		TEST_ASSERT(ret == 0, "fallocate failed, errno: %i", errno);
+	}
+
 	region->mmap_start = mmap(NULL, region->mmap_size,
 				  PROT_READ | PROT_WRITE,
-				  MAP_PRIVATE | MAP_ANONYMOUS
-				  | vm_mem_backing_src_alias(src_type)->flag,
-				  -1, 0);
+				  vm_mem_backing_src_alias(src_type)->flag,
+				  region->fd, 0);
 	TEST_ASSERT(region->mmap_start != MAP_FAILED,
 		    "test_malloc failed, mmap_start: %p errno: %i",
 		    region->mmap_start, errno);
···
 		ret, errno, slot, flags,
 		guest_paddr, (uint64_t) region->region.memory_size);
 
-	/* Add to linked-list of memory regions. */
-	list_add(&region->list, &vm->userspace_mem_regions);
+	/* Add to quick lookup data structures */
+	vm_userspace_mem_region_gpa_insert(&vm->regions.gpa_tree, region);
+	vm_userspace_mem_region_hva_insert(&vm->regions.hva_tree, region);
+	hash_add(vm->regions.slot_hash, &region->slot_node, slot);
+
+	/* If shared memory, create an alias. */
+	if (region->fd >= 0) {
+		region->mmap_alias = mmap(NULL, region->mmap_size,
+					  PROT_READ | PROT_WRITE,
+					  vm_mem_backing_src_alias(src_type)->flag,
+					  region->fd, 0);
+		TEST_ASSERT(region->mmap_alias != MAP_FAILED,
+			    "mmap of alias failed, errno: %i", errno);
+
+		/* Align host alias address */
+		region->host_alias = align(region->mmap_alias, alignment);
+	}
 }
 
 /*
···
 {
 	struct userspace_mem_region *region;
 
-	list_for_each_entry(region, &vm->userspace_mem_regions, list) {
+	hash_for_each_possible(vm->regions.slot_hash, region, slot_node,
+			       memslot)
 		if (region->region.slot == memslot)
 			return region;
-	}
 
 	fprintf(stderr, "No mem region with the requested slot found,\n"
 		"  requested slot: %u\n", memslot);
···
  */
 void vm_mem_region_delete(struct kvm_vm *vm, uint32_t slot)
 {
-	__vm_mem_region_delete(vm, memslot2region(vm, slot));
+	__vm_mem_region_delete(vm, memslot2region(vm, slot), true);
 }
 
 /*
···
 {
 	int dev_fd, ret;
 
-	dev_fd = open(KVM_DEV_PATH, O_RDONLY);
-	if (dev_fd < 0)
-		exit(KSFT_SKIP);
+	dev_fd = open_kvm_dev_path_or_exit();
 
 	ret = ioctl(dev_fd, KVM_GET_VCPU_MMAP_SIZE, NULL);
 	TEST_ASSERT(ret >= sizeof(struct kvm_run),
···
 	uint64_t pages = (sz >> vm->page_shift) + ((sz % vm->page_size) != 0);
 
 	virt_pgd_alloc(vm, pgd_memslot);
+	vm_paddr_t paddr = vm_phy_pages_alloc(vm, pages,
+					      KVM_UTIL_MIN_PFN * vm->page_size,
+					      data_memslot);
 
 	/*
 	 * Find an unused range of virtual page addresses of at least
···
 
 	/* Map the virtual pages. */
 	for (vm_vaddr_t vaddr = vaddr_start; pages > 0;
-		pages--, vaddr += vm->page_size) {
-		vm_paddr_t paddr;
-
-		paddr = vm_phy_page_alloc(vm,
-				KVM_UTIL_MIN_PFN * vm->page_size, data_memslot);
+		pages--, vaddr += vm->page_size, paddr += vm->page_size) {
 
 		virt_pg_map(vm, vaddr, paddr, pgd_memslot);
 
···
 {
 	struct userspace_mem_region *region;
 
-	list_for_each_entry(region, &vm->userspace_mem_regions, list) {
-		if ((gpa >= region->region.guest_phys_addr)
-			&& (gpa <= (region->region.guest_phys_addr
-				+ region->region.memory_size - 1)))
-			return (void *) ((uintptr_t) region->host_mem
-				+ (gpa - region->region.guest_phys_addr));
+	region = userspace_mem_region_find(vm, gpa, gpa);
+	if (!region) {
+		TEST_FAIL("No vm physical memory at 0x%lx", gpa);
+		return NULL;
 	}
 
-	TEST_FAIL("No vm physical memory at 0x%lx", gpa);
-	return NULL;
+	return (void *)((uintptr_t)region->host_mem
+		+ (gpa - region->region.guest_phys_addr));
 }
 
 /*
···
  */
 vm_paddr_t addr_hva2gpa(struct kvm_vm *vm, void *hva)
 {
-	struct userspace_mem_region *region;
+	struct rb_node *node;
 
-	list_for_each_entry(region, &vm->userspace_mem_regions, list) {
-		if ((hva >= region->host_mem)
-			&& (hva <= (region->host_mem
-				+ region->region.memory_size - 1)))
-			return (vm_paddr_t) ((uintptr_t)
-				region->region.guest_phys_addr
-				+ (hva - (uintptr_t) region->host_mem));
+	for (node = vm->regions.hva_tree.rb_node; node; ) {
+		struct userspace_mem_region *region =
+			container_of(node, struct userspace_mem_region, hva_node);
+
+		if (hva >= region->host_mem) {
+			if (hva <= (region->host_mem
+				+ region->region.memory_size - 1))
+				return (vm_paddr_t)((uintptr_t)
+					region->region.guest_phys_addr
+					+ (hva - (uintptr_t)region->host_mem));
+
+			node = node->rb_right;
+		} else
+			node = node->rb_left;
 	}
 
 	TEST_FAIL("No mapping to a guest physical address, hva: %p", hva);
 	return -1;
+}
+
+/*
+ * Address VM physical to Host Virtual *alias*.
+ *
+ * Input Args:
+ *   vm - Virtual Machine
+ *   gpa - VM physical address
+ *
+ * Output Args: None
+ *
+ * Return:
+ *   Equivalent address within the host virtual *alias* area, or NULL
+ *   (without failing the test) if the guest memory is not shared (so
+ *   no alias exists).
+ *
+ * When vm_create() and related functions are called with a shared memory
+ * src_type, we also create a writable, shared alias mapping of the
+ * underlying guest memory. This allows the host to manipulate guest memory
+ * without mapping that memory in the guest's address space. And, for
+ * userfaultfd-based demand paging, we can do so without triggering userfaults.
+ */
+void *addr_gpa2alias(struct kvm_vm *vm, vm_paddr_t gpa)
+{
+	struct userspace_mem_region *region;
+	uintptr_t offset;
+
+	region = userspace_mem_region_find(vm, gpa, gpa);
+	if (!region)
+		return NULL;
+
+	if (!region->host_alias)
+		return NULL;
+
+	offset = gpa - region->region.guest_phys_addr;
+	return (void *) ((uintptr_t) region->host_alias + offset);
 }
 
 /*
···
  */
 void vm_dump(FILE *stream, struct kvm_vm *vm, uint8_t indent)
 {
+	int ctr;
 	struct userspace_mem_region *region;
 	struct vcpu *vcpu;
 
···
 	fprintf(stream, "%*sfd: %i\n", indent, "", vm->fd);
 	fprintf(stream, "%*spage_size: 0x%x\n", indent, "", vm->page_size);
 	fprintf(stream, "%*sMem Regions:\n", indent, "");
-	list_for_each_entry(region, &vm->userspace_mem_regions, list) {
+	hash_for_each(vm->regions.slot_hash, ctr, region, slot_node) {
 		fprintf(stream, "%*sguest_phys: 0x%lx size: 0x%lx "
 			"host_virt: %p\n", indent + 2, "",
 			(uint64_t) region->region.guest_phys_addr,
···
 
 	if (vm == NULL) {
 		/* Ensure that the KVM vendor-specific module is loaded. */
-		f = fopen(KVM_DEV_PATH, "r");
-		TEST_ASSERT(f != NULL, "Error in opening KVM dev file: %d",
-			    errno);
-		fclose(f);
+		close(open_kvm_dev_path_or_exit());
 	}
 
 	f = fopen("/sys/module/kvm_intel/parameters/unrestricted_guest", "r");
···
 	return vm->page_shift;
 }
 
-unsigned int vm_get_max_gfn(struct kvm_vm *vm)
+uint64_t vm_get_max_gfn(struct kvm_vm *vm)
 {
 	return vm->max_gfn;
 }

tools/testing/selftests/kvm/lib/kvm_util_internal.h

···
 #ifndef SELFTEST_KVM_UTIL_INTERNAL_H
 #define SELFTEST_KVM_UTIL_INTERNAL_H
 
+#include "linux/hashtable.h"
+#include "linux/rbtree.h"
+
 #include "sparsebit.h"
 
 struct userspace_mem_region {
···
 	int fd;
 	off_t offset;
 	void *host_mem;
+	void *host_alias;
 	void *mmap_start;
+	void *mmap_alias;
 	size_t mmap_size;
-	struct list_head list;
+	struct rb_node gpa_node;
+	struct rb_node hva_node;
+	struct hlist_node slot_node;
 };
 
 struct vcpu {
···
 	struct kvm_dirty_gfn *dirty_gfns;
 	uint32_t fetch_index;
 	uint32_t dirty_gfns_count;
+};
+
+struct userspace_mem_regions {
+	struct rb_root gpa_tree;
+	struct rb_root hva_tree;
+	DECLARE_HASHTABLE(slot_hash, 9);
 };
 
 struct kvm_vm {
···
 	unsigned int va_bits;
 	uint64_t max_gfn;
 	struct list_head vcpus;
-	struct list_head userspace_mem_regions;
+	struct userspace_mem_regions regions;
 	struct sparsebit *vpages_valid;
 	struct sparsebit *vpages_mapped;
 	bool has_irqchip;

tools/testing/selftests/kvm/lib/perf_test_util.c

···
 /*
  * Copyright (C) 2020, Google LLC.
  */
+#include <inttypes.h>
 
 #include "kvm_util.h"
 #include "perf_test_util.h"
···
 	 */
 	TEST_ASSERT(guest_num_pages < vm_get_max_gfn(vm),
 		    "Requested more guest memory than address space allows.\n"
-		    "    guest pages: %lx max gfn: %x vcpus: %d wss: %lx]\n",
+		    "    guest pages: %" PRIx64 " max gfn: %" PRIx64
+		    " vcpus: %d wss: %" PRIx64 "]\n",
 		    guest_num_pages, vm_get_max_gfn(vm), vcpus,
 		    vcpu_memory_bytes);
 

tools/testing/selftests/kvm/lib/rbtree.c

···
+#include "../../../../lib/rbtree.c"

tools/testing/selftests/kvm/lib/test_util.c

···
 
 const struct vm_mem_backing_src_alias *vm_mem_backing_src_alias(uint32_t i)
 {
+	static const int anon_flags = MAP_PRIVATE | MAP_ANONYMOUS;
+	static const int anon_huge_flags = anon_flags | MAP_HUGETLB;
+
 	static const struct vm_mem_backing_src_alias aliases[] = {
 		[VM_MEM_SRC_ANONYMOUS] = {
 			.name = "anonymous",
-			.flag = 0,
+			.flag = anon_flags,
 		},
 		[VM_MEM_SRC_ANONYMOUS_THP] = {
 			.name = "anonymous_thp",
-			.flag = 0,
+			.flag = anon_flags,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB] = {
 			.name = "anonymous_hugetlb",
-			.flag = MAP_HUGETLB,
+			.flag = anon_huge_flags,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB_16KB] = {
 			.name = "anonymous_hugetlb_16kb",
-			.flag = MAP_HUGETLB | MAP_HUGE_16KB,
+			.flag = anon_huge_flags | MAP_HUGE_16KB,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB_64KB] = {
 			.name = "anonymous_hugetlb_64kb",
-			.flag = MAP_HUGETLB | MAP_HUGE_64KB,
+			.flag = anon_huge_flags | MAP_HUGE_64KB,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB_512KB] = {
 			.name = "anonymous_hugetlb_512kb",
-			.flag = MAP_HUGETLB | MAP_HUGE_512KB,
+			.flag = anon_huge_flags | MAP_HUGE_512KB,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB_1MB] = {
 			.name = "anonymous_hugetlb_1mb",
-			.flag = MAP_HUGETLB | MAP_HUGE_1MB,
+			.flag = anon_huge_flags | MAP_HUGE_1MB,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB_2MB] = {
 			.name = "anonymous_hugetlb_2mb",
-			.flag = MAP_HUGETLB | MAP_HUGE_2MB,
+			.flag = anon_huge_flags | MAP_HUGE_2MB,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB_8MB] = {
 			.name = "anonymous_hugetlb_8mb",
-			.flag = MAP_HUGETLB | MAP_HUGE_8MB,
+			.flag = anon_huge_flags | MAP_HUGE_8MB,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB_16MB] = {
 			.name = "anonymous_hugetlb_16mb",
-			.flag = MAP_HUGETLB | MAP_HUGE_16MB,
+			.flag = anon_huge_flags | MAP_HUGE_16MB,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB_32MB] = {
 			.name = "anonymous_hugetlb_32mb",
-			.flag = MAP_HUGETLB | MAP_HUGE_32MB,
+			.flag = anon_huge_flags | MAP_HUGE_32MB,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB_256MB] = {
 			.name = "anonymous_hugetlb_256mb",
-			.flag = MAP_HUGETLB | MAP_HUGE_256MB,
+			.flag = anon_huge_flags | MAP_HUGE_256MB,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB_512MB] = {
 			.name = "anonymous_hugetlb_512mb",
-			.flag = MAP_HUGETLB | MAP_HUGE_512MB,
+			.flag = anon_huge_flags | MAP_HUGE_512MB,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB_1GB] = {
 			.name = "anonymous_hugetlb_1gb",
-			.flag = MAP_HUGETLB | MAP_HUGE_1GB,
+			.flag = anon_huge_flags | MAP_HUGE_1GB,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB_2GB] = {
 			.name = "anonymous_hugetlb_2gb",
-			.flag = MAP_HUGETLB | MAP_HUGE_2GB,
+			.flag = anon_huge_flags | MAP_HUGE_2GB,
 		},
 		[VM_MEM_SRC_ANONYMOUS_HUGETLB_16GB] = {
 			.name = "anonymous_hugetlb_16gb",
-			.flag = MAP_HUGETLB | MAP_HUGE_16GB,
+			.flag = anon_huge_flags | MAP_HUGE_16GB,
+		},
+		[VM_MEM_SRC_SHMEM] = {
+			.name = "shmem",
+			.flag = MAP_SHARED,
+		},
+		[VM_MEM_SRC_SHARED_HUGETLB] = {
+			.name = "shared_hugetlb",
+			/*
+			 * No MAP_HUGETLB, we use MFD_HUGETLB instead. Since
+			 * we're using "file backed" memory, we need to specify
+			 * this when the FD is created, not when the area is
+			 * mapped.
+			 */
+			.flag = MAP_SHARED,
 		},
 	};
 	_Static_assert(ARRAY_SIZE(aliases) == NUM_SRC_TYPES,
···
 
 	switch (i) {
 	case VM_MEM_SRC_ANONYMOUS:
+	case VM_MEM_SRC_SHMEM:
 		return getpagesize();
 	case VM_MEM_SRC_ANONYMOUS_THP:
 		return get_trans_hugepagesz();
 	case VM_MEM_SRC_ANONYMOUS_HUGETLB:
+	case VM_MEM_SRC_SHARED_HUGETLB:
 		return get_def_hugetlb_pagesz();
 	default:
 		return MAP_HUGE_PAGE_SIZE(flag);

tools/testing/selftests/kvm/lib/x86_64/processor.c

···
 		return cpuid;
 
 	cpuid = allocate_kvm_cpuid2();
-	kvm_fd = open(KVM_DEV_PATH, O_RDONLY);
-	if (kvm_fd < 0)
-		exit(KSFT_SKIP);
+	kvm_fd = open_kvm_dev_path_or_exit();
 
 	ret = ioctl(kvm_fd, KVM_GET_SUPPORTED_CPUID, cpuid);
 	TEST_ASSERT(ret == 0, "KVM_GET_SUPPORTED_CPUID failed %d %d\n",
···
 
 	buffer.header.nmsrs = 1;
 	buffer.entry.index = msr_index;
-	kvm_fd = open(KVM_DEV_PATH, O_RDONLY);
-	if (kvm_fd < 0)
-		exit(KSFT_SKIP);
+	kvm_fd = open_kvm_dev_path_or_exit();
 
 	r = ioctl(kvm_fd, KVM_GET_MSRS, &buffer.header);
 	TEST_ASSERT(r == 1, "KVM_GET_MSRS IOCTL failed,\n"
···
 	struct kvm_msr_list *list;
 	int nmsrs, r, kvm_fd;
 
-	kvm_fd = open(KVM_DEV_PATH, O_RDONLY);
-	if (kvm_fd < 0)
-		exit(KSFT_SKIP);
+	kvm_fd = open_kvm_dev_path_or_exit();
 
 	nmsrs = kvm_get_num_msrs_fd(kvm_fd);
 	list = malloc(sizeof(*list) + nmsrs * sizeof(list->indices[0]));
···
 		return cpuid;
 
 	cpuid = allocate_kvm_cpuid2();
-	kvm_fd = open(KVM_DEV_PATH, O_RDONLY);
-	if (kvm_fd < 0)
-		exit(KSFT_SKIP);
+	kvm_fd = open_kvm_dev_path_or_exit();
 
 	ret = ioctl(kvm_fd, KVM_GET_SUPPORTED_HV_CPUID, cpuid);
 	TEST_ASSERT(ret == 0, "KVM_GET_SUPPORTED_HV_CPUID failed %d %d\n",

tools/testing/selftests/kvm/memslot_modification_stress_test.c

···
 };
 
 static void add_remove_memslot(struct kvm_vm *vm, useconds_t delay,
-			      uint64_t nr_modifications, uint64_t gpa)
+			       uint64_t nr_modifications)
 {
+	const uint64_t pages = 1;
+	uint64_t gpa;
 	int i;
+
+	/*
+	 * Add the dummy memslot just below the perf_test_util memslot, which is
+	 * at the top of the guest physical address space.
+	 */
+	gpa = guest_test_phys_mem - pages * vm_get_page_size(vm);
 
 	for (i = 0; i < nr_modifications; i++) {
 		usleep(delay);
 		vm_userspace_mem_region_add(vm, VM_MEM_SRC_ANONYMOUS, gpa,
-					    DUMMY_MEMSLOT_INDEX, 1, 0);
+					    DUMMY_MEMSLOT_INDEX, pages, 0);
 
 		vm_mem_region_delete(vm, DUMMY_MEMSLOT_INDEX);
 	}
···
 	pr_info("Started all vCPUs\n");
 
 	add_remove_memslot(vm, p->memslot_modification_delay,
-			   p->nr_memslot_modifications,
-			   guest_test_phys_mem +
-			   (guest_percpu_mem_size * nr_vcpus) +
-			   perf_test_args.host_page_size +
-			   perf_test_args.guest_page_size);
+			   p->nr_memslot_modifications);
 
 	run_vcpus = false;
 

tools/testing/selftests/kvm/memslot_perf_test.c

···
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * A memslot-related performance benchmark.
+ *
+ * Copyright (C) 2021 Oracle and/or its affiliates.
+ *
+ * Basic guest setup / host vCPU thread code lifted from set_memory_region_test.
+ */
+#include <pthread.h>
+#include <sched.h>
+#include <semaphore.h>
+#include <stdatomic.h>
+#include <stdbool.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/mman.h>
+#include <time.h>
+#include <unistd.h>
+
+#include <linux/compiler.h>
+
+#include <test_util.h>
+#include <kvm_util.h>
+#include <processor.h>
+
+#define VCPU_ID 0
+
+#define MEM_SIZE		((512U << 20) + 4096)
+#define MEM_SIZE_PAGES		(MEM_SIZE / 4096)
+#define MEM_GPA		0x10000000UL
+#define MEM_AUX_GPA		MEM_GPA
+#define MEM_SYNC_GPA		MEM_AUX_GPA
+#define MEM_TEST_GPA		(MEM_AUX_GPA + 4096)
+#define MEM_TEST_SIZE		(MEM_SIZE - 4096)
+static_assert(MEM_SIZE % 4096 == 0, "invalid mem size");
+static_assert(MEM_TEST_SIZE % 4096 == 0, "invalid mem test size");
+
+/*
+ * 32 MiB is max size that gets well over 100 iterations on 509 slots.
+ * Considering that each slot needs to have at least one page up to
+ * 8194 slots in use can then be tested (although with slightly
+ * limited resolution).
+ */
+#define MEM_SIZE_MAP		((32U << 20) + 4096)
+#define MEM_SIZE_MAP_PAGES	(MEM_SIZE_MAP / 4096)
+#define MEM_TEST_MAP_SIZE	(MEM_SIZE_MAP - 4096)
+#define MEM_TEST_MAP_SIZE_PAGES (MEM_TEST_MAP_SIZE / 4096)
+static_assert(MEM_SIZE_MAP % 4096 == 0, "invalid map test region size");
+static_assert(MEM_TEST_MAP_SIZE % 4096 == 0, "invalid map test region size");
+static_assert(MEM_TEST_MAP_SIZE_PAGES % 2 == 0, "invalid map test region size");
+static_assert(MEM_TEST_MAP_SIZE_PAGES > 2, "invalid map test region size");
+
+/*
+ * 128 MiB is min size that fills 32k slots with at least one page in each
+ * while at the same time gets 100+ iterations in such test
+ */
+#define MEM_TEST_UNMAP_SIZE		(128U << 20)
+#define MEM_TEST_UNMAP_SIZE_PAGES	(MEM_TEST_UNMAP_SIZE / 4096)
+/* 2 MiB chunk size like a typical huge page */
+#define MEM_TEST_UNMAP_CHUNK_PAGES	(2U << (20 - 12))
+static_assert(MEM_TEST_UNMAP_SIZE <= MEM_TEST_SIZE,
+	      "invalid unmap test region size");
+static_assert(MEM_TEST_UNMAP_SIZE % 4096 == 0,
+	      "invalid unmap test region size");
+static_assert(MEM_TEST_UNMAP_SIZE_PAGES %
+	      (2 * MEM_TEST_UNMAP_CHUNK_PAGES) == 0,
+	      "invalid unmap test region size");
+
+/*
+ * For the move active test the middle of the test area is placed on
+ * a memslot boundary: half lies in the memslot being moved, half in
+ * other memslot(s).
+ *
+ * When running this test with 32k memslots (32764, really) each memslot
+ * contains 4 pages.
+ * The last one additionally contains the remaining 21 pages of memory,
+ * for the total size of 25 pages.
+ * Hence, the maximum size here is 50 pages.
+ */
+#define MEM_TEST_MOVE_SIZE_PAGES	(50)
+#define MEM_TEST_MOVE_SIZE		(MEM_TEST_MOVE_SIZE_PAGES * 4096)
+#define MEM_TEST_MOVE_GPA_DEST		(MEM_GPA + MEM_SIZE)
+static_assert(MEM_TEST_MOVE_SIZE <= MEM_TEST_SIZE,
+	      "invalid move test region size");
+
+#define MEM_TEST_VAL_1 0x1122334455667788
+#define MEM_TEST_VAL_2 0x99AABBCCDDEEFF00
+
+struct vm_data {
+	struct kvm_vm *vm;
+	pthread_t vcpu_thread;
+	uint32_t nslots;
+	uint64_t npages;
+	uint64_t pages_per_slot;
+	void **hva_slots;
+	bool mmio_ok;
+	uint64_t mmio_gpa_min;
+	uint64_t mmio_gpa_max;
+};
+
+struct sync_area {
+	atomic_bool start_flag;
+	atomic_bool exit_flag;
+	atomic_bool sync_flag;
+	void *move_area_ptr;
+};
+
+/*
+ * Technically, we need also for the atomic bool to be address-free, which
+ * is recommended, but not strictly required, by C11 for lockless
+ * implementations.
+ * However, in practice both GCC and Clang fulfill this requirement on
+ * all KVM-supported platforms.
+ */
+static_assert(ATOMIC_BOOL_LOCK_FREE == 2, "atomic bool is not lockless");
+
+static sem_t vcpu_ready;
+
+static bool map_unmap_verify;
+
+static bool verbose;
+#define pr_info_v(...)				\
+	do {					\
+		if (verbose)			\
+			pr_info(__VA_ARGS__);	\
+	} while (0)
+
+static void *vcpu_worker(void *data)
+{
+	struct vm_data *vm = data;
+	struct kvm_run *run;
+	struct ucall uc;
+	uint64_t cmd;
+
+	run = vcpu_state(vm->vm, VCPU_ID);
+	while (1) {
+		vcpu_run(vm->vm, VCPU_ID);
+
+		if (run->exit_reason == KVM_EXIT_IO) {
+			cmd = get_ucall(vm->vm, VCPU_ID, &uc);
+			if (cmd != UCALL_SYNC)
+				break;
+
+			sem_post(&vcpu_ready);
+			continue;
+		}
+
+		if (run->exit_reason != KVM_EXIT_MMIO)
+			break;
+
+		TEST_ASSERT(vm->mmio_ok, "Unexpected mmio exit");
+		TEST_ASSERT(run->mmio.is_write, "Unexpected mmio read");
+		TEST_ASSERT(run->mmio.len == 8,
+			    "Unexpected exit mmio size = %u", run->mmio.len);
+		TEST_ASSERT(run->mmio.phys_addr >= vm->mmio_gpa_min &&
+			    run->mmio.phys_addr <= vm->mmio_gpa_max,
+			    "Unexpected exit mmio address = 0x%llx",
+			    run->mmio.phys_addr);
+	}
+
+	if (run->exit_reason == KVM_EXIT_IO && cmd == UCALL_ABORT)
+		TEST_FAIL("%s at %s:%ld, val = %lu", (const char *)uc.args[0],
+			  __FILE__, uc.args[1], uc.args[2]);
+
+	return NULL;
+}
+
+static void wait_for_vcpu(void)
+{
+	struct timespec ts;
+
+	TEST_ASSERT(!clock_gettime(CLOCK_REALTIME, &ts),
+		    "clock_gettime() failed: %d\n", errno);
+
+	ts.tv_sec += 2;
+	TEST_ASSERT(!sem_timedwait(&vcpu_ready, &ts),
+		    "sem_timedwait() failed: %d\n", errno);
+}
+
+static void *vm_gpa2hva(struct vm_data *data, uint64_t gpa, uint64_t *rempages)
+{
+	uint64_t gpage, pgoffs;
+	uint32_t slot, slotoffs;
+	void *base;
+
+	TEST_ASSERT(gpa >= MEM_GPA, "Too low gpa to translate");
+	TEST_ASSERT(gpa < MEM_GPA + data->npages * 4096,
+		    "Too high gpa to translate");
+	gpa -= MEM_GPA;
+
+	gpage = gpa / 4096;
+	pgoffs = gpa % 4096;
+	slot = min(gpage / data->pages_per_slot, (uint64_t)data->nslots - 1);
+	slotoffs = gpage - (slot * data->pages_per_slot);
+
+	if (rempages) {
+		uint64_t slotpages;
+
+		if (slot == data->nslots - 1)
+			slotpages = data->npages - slot * data->pages_per_slot;
+		else
+			slotpages = data->pages_per_slot;
+
+		TEST_ASSERT(!pgoffs,
+			    "Asking for remaining pages in slot but gpa not page aligned");
+		*rempages = slotpages - slotoffs;
+	}
+
+	base = data->hva_slots[slot];
+	return (uint8_t *)base + slotoffs * 4096 + pgoffs;
+}
+
+static uint64_t vm_slot2gpa(struct vm_data *data, uint32_t slot)
+{
+	TEST_ASSERT(slot < data->nslots, "Too high slot number");
+
+	return MEM_GPA + slot * data->pages_per_slot * 4096;
+}
+
+static struct vm_data *alloc_vm(void)
+{
+	struct vm_data *data;
+
+	data = malloc(sizeof(*data));
+	TEST_ASSERT(data, "malloc(vmdata) failed");
+
+	data->vm = NULL;
+	data->hva_slots = NULL;
+
+	return data;
+}
+
+static bool prepare_vm(struct vm_data *data, int nslots, uint64_t *maxslots,
+		       void *guest_code, uint64_t mempages,
+		       struct timespec *slot_runtime)
+{
+	uint32_t max_mem_slots;
+	uint64_t rempages;
+	uint64_t guest_addr;
+	uint32_t slot;
+	struct timespec tstart;
+	struct sync_area *sync;
+
+	max_mem_slots = kvm_check_cap(KVM_CAP_NR_MEMSLOTS);
+	TEST_ASSERT(max_mem_slots > 1,
+		    "KVM_CAP_NR_MEMSLOTS should be greater than 1");
+	TEST_ASSERT(nslots > 1 || nslots == -1,
+		    "Slot count cap should be greater than 1");
+	if (nslots != -1)
+		max_mem_slots = min(max_mem_slots, (uint32_t)nslots);
+	pr_info_v("Allowed number of memory slots: %"PRIu32"\n", max_mem_slots);
+
+	TEST_ASSERT(mempages > 1,
+		    "Can't test without any memory");
+
+	data->npages = mempages;
+	data->nslots = max_mem_slots - 1;
+	data->pages_per_slot = mempages / data->nslots;
+	if (!data->pages_per_slot) {
+		*maxslots = mempages + 1;
+		return false;
+	}
+
+	rempages = mempages % data->nslots;
+	data->hva_slots = malloc(sizeof(*data->hva_slots) * data->nslots);
+	TEST_ASSERT(data->hva_slots, "malloc() fail");
+
+	data->vm = vm_create_default(VCPU_ID, 1024, guest_code);
+
+	pr_info_v("Adding slots 1..%i, each slot with %"PRIu64" pages + %"PRIu64" extra pages last\n",
+		max_mem_slots - 1, data->pages_per_slot, rempages);
+
+	clock_gettime(CLOCK_MONOTONIC, &tstart);
+	for (slot = 1, guest_addr = MEM_GPA; slot < max_mem_slots; slot++) {
+		uint64_t npages;
+
+		npages = data->pages_per_slot;
+		if (slot == max_mem_slots - 1)
+			npages += rempages;
+
+		vm_userspace_mem_region_add(data->vm, VM_MEM_SRC_ANONYMOUS,
+					    guest_addr, slot, npages,
+					    0);
+		guest_addr += npages * 4096;
+	}
+	*slot_runtime = timespec_elapsed(tstart);
+
+	for (slot = 0, guest_addr = MEM_GPA; slot < max_mem_slots - 1; slot++) {
+		uint64_t npages;
+		uint64_t gpa;
+
+		npages = data->pages_per_slot;
+		if (slot == max_mem_slots - 2)
+			npages += rempages;
+
+		gpa = vm_phy_pages_alloc(data->vm, npages, guest_addr,
+					 slot + 1);
+		TEST_ASSERT(gpa == guest_addr,
+			    "vm_phy_pages_alloc() failed\n");
+
+		data->hva_slots[slot] = addr_gpa2hva(data->vm, guest_addr);
+		memset(data->hva_slots[slot], 0, npages * 4096);
+
+		guest_addr += npages * 4096;
+	}
+
+	virt_map(data->vm, MEM_GPA, MEM_GPA, mempages, 0);
+
+	sync = (typeof(sync))vm_gpa2hva(data, MEM_SYNC_GPA, NULL);
+	atomic_init(&sync->start_flag, false);
+	atomic_init(&sync->exit_flag, false);
+	atomic_init(&sync->sync_flag, false);
+
+	data->mmio_ok = false;
+
+	return true;
+}
+
+static void launch_vm(struct vm_data *data)
+{
+	pr_info_v("Launching the test VM\n");
+
+	pthread_create(&data->vcpu_thread, NULL, vcpu_worker, data);
+
+	/* Ensure the guest thread is spun up. */
+	wait_for_vcpu();
+}
+
+static void free_vm(struct vm_data *data)
+{
+	kvm_vm_free(data->vm);
+	free(data->hva_slots);
+	free(data);
+}
+
+static void wait_guest_exit(struct vm_data *data)
+{
+	pthread_join(data->vcpu_thread, NULL);
+}
+
+static void let_guest_run(struct sync_area *sync)
+{
+	atomic_store_explicit(&sync->start_flag, true, memory_order_release);
+}
+
+static void guest_spin_until_start(void)
+{
+	struct sync_area *sync = (typeof(sync))MEM_SYNC_GPA;
+
+	while (!atomic_load_explicit(&sync->start_flag, memory_order_acquire))
+		;
+}
+
+static void make_guest_exit(struct sync_area *sync)
+{
+	atomic_store_explicit(&sync->exit_flag, true, memory_order_release);
+}
+
+static bool _guest_should_exit(void)
+{
+	struct sync_area *sync = (typeof(sync))MEM_SYNC_GPA;
+
+	return atomic_load_explicit(&sync->exit_flag, memory_order_acquire);
+}
+
+#define guest_should_exit() unlikely(_guest_should_exit())
+
+/*
+ * noinline so we can easily see how much time the host spends waiting
+ * for the guest.
+ * For the same reason use alarm() instead of polling clock_gettime()
+ * to implement a wait timeout.
+ */
+static noinline void host_perform_sync(struct sync_area *sync)
+{
+	alarm(2);
+
+	atomic_store_explicit(&sync->sync_flag, true, memory_order_release);
+	while (atomic_load_explicit(&sync->sync_flag, memory_order_acquire))
+		;
+
+	alarm(0);
+}
+
+static bool guest_perform_sync(void)
+{
+	struct sync_area *sync = (typeof(sync))MEM_SYNC_GPA;
+	bool expected;
+
+	do {
+		if (guest_should_exit())
+			return false;
+
+		expected = true;
+	} while (!atomic_compare_exchange_weak_explicit(&sync->sync_flag,
+							&expected, false,
+							memory_order_acq_rel,
+							memory_order_relaxed));
+
+	return true;
+}
+
+static void guest_code_test_memslot_move(void)
+{
+	struct sync_area *sync = (typeof(sync))MEM_SYNC_GPA;
+	uintptr_t base = (typeof(base))READ_ONCE(sync->move_area_ptr);
+
+	GUEST_SYNC(0);
+
+	guest_spin_until_start();
+
+	while (!guest_should_exit()) {
+		uintptr_t ptr;
+
+		for (ptr = base; ptr < base + MEM_TEST_MOVE_SIZE;
+		     ptr += 4096)
+			*(uint64_t *)ptr = MEM_TEST_VAL_1;
+
+		/*
+		 * No host sync here since the MMIO exits are so expensive
+		 * that the host would spend most of its time waiting for
+		 * the guest and so instead of measuring memslot move
+		 * performance we would measure the performance and
+		 * likelihood of MMIO exits
+		 */
+	}
+
+	GUEST_DONE();
+}
+
+static void guest_code_test_memslot_map(void)
+{
+	struct sync_area *sync = (typeof(sync))MEM_SYNC_GPA;
+
+	GUEST_SYNC(0);
+
+	guest_spin_until_start();
+
+	while (1) {
+		uintptr_t ptr;
+
+		for (ptr = MEM_TEST_GPA;
+		     ptr < MEM_TEST_GPA + MEM_TEST_MAP_SIZE / 2; ptr += 4096)
+			*(uint64_t *)ptr = MEM_TEST_VAL_1;
+
+		if (!guest_perform_sync())
+			break;
+
+		for (ptr = MEM_TEST_GPA + MEM_TEST_MAP_SIZE / 2;
+		     ptr < MEM_TEST_GPA + MEM_TEST_MAP_SIZE; ptr += 4096)
+			*(uint64_t *)ptr = MEM_TEST_VAL_2;
+
+		if (!guest_perform_sync())
+			break;
+	}
+
+	GUEST_DONE();
+}
+
+static void guest_code_test_memslot_unmap(void)
+{
+	struct sync_area *sync = (typeof(sync))MEM_SYNC_GPA;
+
+	GUEST_SYNC(0);
+
+	guest_spin_until_start();
+
+	while (1) {
+		uintptr_t ptr = MEM_TEST_GPA;
+
+		/*
+		 * We can afford to access (map) just a small number of pages
+		 * per host sync as otherwise the host will spend
+		 * a significant amount of its time waiting for the guest
+		 * (instead of doing unmap operations), so this will
+		 * effectively turn this test into a map performance test.
+		 *
+		 * Just access a single page to be on the safe side.
+		 */
+		*(uint64_t *)ptr = MEM_TEST_VAL_1;
+
+		if (!guest_perform_sync())
+			break;
+
+		ptr += MEM_TEST_UNMAP_SIZE / 2;
+		*(uint64_t *)ptr = MEM_TEST_VAL_2;
+
+		if (!guest_perform_sync())
+			break;
+	}
+
+	GUEST_DONE();
+}
+
+static void guest_code_test_memslot_rw(void)
+{
+	GUEST_SYNC(0);
+
+	guest_spin_until_start();
+
+	while (1) {
+		uintptr_t ptr;
+
+		for (ptr = MEM_TEST_GPA;
+		     ptr < MEM_TEST_GPA + MEM_TEST_SIZE; ptr += 4096)
+			*(uint64_t *)ptr = MEM_TEST_VAL_1;
+
+		if (!guest_perform_sync())
+			break;
+
+		for (ptr = MEM_TEST_GPA + 4096 / 2;
+		     ptr < MEM_TEST_GPA + MEM_TEST_SIZE; ptr += 4096) {
+			uint64_t val = *(uint64_t *)ptr;
+
+			GUEST_ASSERT_1(val == MEM_TEST_VAL_2, val);
+			*(uint64_t *)ptr = 0;
+		}
+
+		if (!guest_perform_sync())
+			break;
+	}
+
+	GUEST_DONE();
+}
+
+static bool test_memslot_move_prepare(struct vm_data *data,
+				      struct sync_area *sync,
+				      uint64_t *maxslots, bool isactive)
+{
+	uint64_t movesrcgpa, movetestgpa;
+
+	movesrcgpa = vm_slot2gpa(data, data->nslots - 1);
+
+	if (isactive) {
+		uint64_t lastpages;
+
+		vm_gpa2hva(data, movesrcgpa, &lastpages);
+		if (lastpages < MEM_TEST_MOVE_SIZE_PAGES / 2) {
+			*maxslots = 0;
+			return false;
+		}
+	}
+
+	movetestgpa = movesrcgpa - (MEM_TEST_MOVE_SIZE / (isactive ? 2 : 1));
+	sync->move_area_ptr = (void *)movetestgpa;
+
+	if (isactive) {
+		data->mmio_ok = true;
+		data->mmio_gpa_min = movesrcgpa;
+		data->mmio_gpa_max = movesrcgpa + MEM_TEST_MOVE_SIZE / 2 - 1;
+	}
+
+	return true;
+}
+
+static bool test_memslot_move_prepare_active(struct vm_data *data,
+					     struct sync_area *sync,
+					     uint64_t *maxslots)
+{
+	return test_memslot_move_prepare(data, sync, maxslots, true);
+}
+
+static bool test_memslot_move_prepare_inactive(struct vm_data *data,
+					       struct sync_area *sync,
+					       uint64_t *maxslots)
+{
+	return test_memslot_move_prepare(data, sync, maxslots, false);
+}
+
+static void test_memslot_move_loop(struct vm_data *data, struct sync_area *sync)
+{
+	uint64_t movesrcgpa;
+
+	movesrcgpa = vm_slot2gpa(data, data->nslots - 1);
+	vm_mem_region_move(data->vm, data->nslots - 1 + 1,
+			   MEM_TEST_MOVE_GPA_DEST);
+	vm_mem_region_move(data->vm, data->nslots - 1 + 1, movesrcgpa);
+}
+
+static void test_memslot_do_unmap(struct vm_data *data,
+				  uint64_t offsp, uint64_t count)
+{
+	uint64_t gpa, ctr;
+
+	for (gpa = MEM_TEST_GPA + offsp * 4096, ctr = 0; ctr < count; ) {
+		uint64_t npages;
+		void *hva;
+		int ret;
+
+		hva = vm_gpa2hva(data, gpa, &npages);
+		TEST_ASSERT(npages, "Empty memory slot at gptr 0x%"PRIx64, gpa);
+		npages = min(npages, count - ctr);
+		ret = madvise(hva, npages * 4096, MADV_DONTNEED);
+		TEST_ASSERT(!ret,
+			    "madvise(%p, MADV_DONTNEED) on VM memory should not fail for gptr 0x%"PRIx64,
+			    hva, gpa);
+		ctr += npages;
+		gpa += npages * 4096;
+	}
+	TEST_ASSERT(ctr == count,
+		    "madvise(MADV_DONTNEED) should exactly cover all of the requested area");
+}
+
+static void test_memslot_map_unmap_check(struct vm_data *data,
+					 uint64_t offsp, uint64_t valexp)
+{
+	uint64_t gpa;
+	uint64_t *val;
+
+	if (!map_unmap_verify)
+		return;
+
+	gpa = MEM_TEST_GPA + offsp * 4096;
+	val = (typeof(val))vm_gpa2hva(data, gpa, NULL);
+	TEST_ASSERT(*val == valexp,
+		    "Guest written values should read back correctly before unmap (%"PRIu64" vs %"PRIu64" @ %"PRIx64")",
+		    *val, valexp, gpa);
+	*val = 0;
+}
+
+static void test_memslot_map_loop(struct vm_data *data, struct sync_area *sync)
+{
+	/*
+	 * Unmap the second half of the test area while guest writes to (maps)
+	 * the first half.
+	 */
+	test_memslot_do_unmap(data, MEM_TEST_MAP_SIZE_PAGES / 2,
+			      MEM_TEST_MAP_SIZE_PAGES / 2);
+
+	/*
+	 * Wait for the guest to finish writing the first half of the test
+	 * area, verify the written value on the first and the last page of
+	 * this area and then unmap it.
+	 * Meanwhile, the guest is writing to (mapping) the second half of
+	 * the test area.
+	 */
+	host_perform_sync(sync);
+	test_memslot_map_unmap_check(data, 0, MEM_TEST_VAL_1);
+	test_memslot_map_unmap_check(data,
+				     MEM_TEST_MAP_SIZE_PAGES / 2 - 1,
+				     MEM_TEST_VAL_1);
+	test_memslot_do_unmap(data, 0, MEM_TEST_MAP_SIZE_PAGES / 2);
+
+
+	/*
+	 * Wait for the guest to finish writing the second half of the test
+	 * area and verify the written value on the first and the last page
+	 * of this area.
+	 * The area will be unmapped at the beginning of the next loop
+	 * iteration.
+	 * Meanwhile, the guest is writing to (mapping) the first half of
+	 * the test area.
+	 */
+	host_perform_sync(sync);
+	test_memslot_map_unmap_check(data, MEM_TEST_MAP_SIZE_PAGES / 2,
+				     MEM_TEST_VAL_2);
+	test_memslot_map_unmap_check(data, MEM_TEST_MAP_SIZE_PAGES - 1,
+				     MEM_TEST_VAL_2);
+}
+
+static void test_memslot_unmap_loop_common(struct vm_data *data,
+					   struct sync_area *sync,
+					   uint64_t chunk)
+{
+	uint64_t ctr;
+
+	/*
+	 * Wait for the guest to finish mapping page(s) in the first half
+	 * of the test area, verify the written value and then perform unmap
+	 * of this area.
+	 * Meanwhile, the guest is writing to (mapping) page(s) in the second
+	 * half of the test area.
+	 */
+	host_perform_sync(sync);
+	test_memslot_map_unmap_check(data, 0, MEM_TEST_VAL_1);
+	for (ctr = 0; ctr < MEM_TEST_UNMAP_SIZE_PAGES / 2; ctr += chunk)
+		test_memslot_do_unmap(data, ctr, chunk);
+
+	/* Likewise, but for the opposite host / guest areas */
+	host_perform_sync(sync);
+	test_memslot_map_unmap_check(data, MEM_TEST_UNMAP_SIZE_PAGES / 2,
+				     MEM_TEST_VAL_2);
+	for (ctr = MEM_TEST_UNMAP_SIZE_PAGES / 2;
+	     ctr < MEM_TEST_UNMAP_SIZE_PAGES; ctr += chunk)
+		test_memslot_do_unmap(data, ctr, chunk);
+}
+
+static void test_memslot_unmap_loop(struct vm_data *data,
+				    struct sync_area *sync)
+{
+	test_memslot_unmap_loop_common(data, sync, 1);
+}
+
+static void test_memslot_unmap_loop_chunked(struct vm_data *data,
+					    struct sync_area *sync)
+{
+	test_memslot_unmap_loop_common(data, sync, MEM_TEST_UNMAP_CHUNK_PAGES);
+}
+
+static void test_memslot_rw_loop(struct vm_data *data, struct sync_area *sync)
+{
+	uint64_t gptr;
+
+	for (gptr = MEM_TEST_GPA + 4096 / 2;
+	     gptr < MEM_TEST_GPA + MEM_TEST_SIZE; gptr += 4096)
+		*(uint64_t *)vm_gpa2hva(data, gptr, NULL) = MEM_TEST_VAL_2;
+
+	host_perform_sync(sync);
+
+	for (gptr = MEM_TEST_GPA;
+	     gptr < MEM_TEST_GPA + MEM_TEST_SIZE; gptr += 4096) {
+		uint64_t *vptr = (typeof(vptr))vm_gpa2hva(data, gptr, NULL);
+		uint64_t val = *vptr;
+
+		TEST_ASSERT(val == MEM_TEST_VAL_1,
+			    "Guest written values should read back correctly (is %"PRIu64" @ %"PRIx64")",
+			    val, gptr);
+		*vptr = 0;
+	}
+
+	host_perform_sync(sync);
+}
+
+struct test_data {
+	const char *name;
+	uint64_t mem_size;
+	void (*guest_code)(void);
+	bool (*prepare)(struct vm_data *data, struct sync_area *sync,
+			uint64_t *maxslots);
+	void (*loop)(struct vm_data *data, struct sync_area *sync);
+};
+
+static bool test_execute(int nslots, uint64_t *maxslots,
+			 unsigned int maxtime,
+			 const struct test_data *tdata,
+			 uint64_t *nloops,
+			 struct timespec *slot_runtime,
+			 struct timespec *guest_runtime)
+{
+	uint64_t mem_size = tdata->mem_size ? : MEM_SIZE_PAGES;
+	struct vm_data *data;
+	struct sync_area *sync;
+	struct timespec tstart;
+	bool ret = true;
+
+	data = alloc_vm();
+	if (!prepare_vm(data, nslots, maxslots, tdata->guest_code,
+			mem_size, slot_runtime)) {
+		ret = false;
+		goto exit_free;
+	}
+
+	sync = (typeof(sync))vm_gpa2hva(data, MEM_SYNC_GPA, NULL);
+
+	if (tdata->prepare &&
+	    !tdata->prepare(data, sync, maxslots)) {
+		ret = false;
+		goto exit_free;
+	}
+
+	launch_vm(data);
+
+	clock_gettime(CLOCK_MONOTONIC, &tstart);
+	let_guest_run(sync);
+
+	while (1) {
+		*guest_runtime = timespec_elapsed(tstart);
+		if (guest_runtime->tv_sec >= maxtime)
+			break;
+
+		tdata->loop(data, sync);
+
+		(*nloops)++;
+	}
+
+	make_guest_exit(sync);
+	wait_guest_exit(data);
+
+exit_free:
+	free_vm(data);
+
+	return ret;
+}
+
+static const struct test_data tests[] = {
+	{
+		.name = "map",
+		.mem_size = MEM_SIZE_MAP_PAGES,
+		.guest_code = guest_code_test_memslot_map,
+		.loop = test_memslot_map_loop,
+	},
+	{
+		.name = "unmap",
+		.mem_size = MEM_TEST_UNMAP_SIZE_PAGES + 1,
+		.guest_code = guest_code_test_memslot_unmap,
+		.loop = test_memslot_unmap_loop,
+	},
+	{
+		.name = "unmap chunked",
+		.mem_size = MEM_TEST_UNMAP_SIZE_PAGES + 1,
+		.guest_code = guest_code_test_memslot_unmap,
+		.loop = test_memslot_unmap_loop_chunked,
+	},
+	{
+		.name = "move active area",
+		.guest_code = guest_code_test_memslot_move,
+		.prepare = test_memslot_move_prepare_active,
+		.loop = test_memslot_move_loop,
+	},
+	{
+		.name = "move inactive area",
+		.guest_code = guest_code_test_memslot_move,
+		.prepare = test_memslot_move_prepare_inactive,
+		.loop = test_memslot_move_loop,
+	},
+	{
+		.name = "RW",
+		.guest_code = guest_code_test_memslot_rw,
+		.loop = test_memslot_rw_loop
+	},
+};
+
+#define NTESTS ARRAY_SIZE(tests)
+
+struct test_args {
+	int tfirst;
+	int tlast;
+	int nslots;
+	int seconds;
+	int runs;
+};
+
+static void help(char *name, struct test_args *targs)
+{
+	int ctr;
+
+	pr_info("usage: %s [-h] [-v] [-d] [-s slots] [-f first_test] [-e last_test] [-l test_length] [-r run_count]\n",
+		name);
+	pr_info(" -h: print this help screen.\n");
+	pr_info(" -v: enable verbose mode (not for benchmarking).\n");
+	pr_info(" -d: enable extra debug checks.\n");
+	pr_info(" -s: specify memslot count cap (-1 means no cap; currently: %i)\n",
+		targs->nslots);
+	pr_info(" -f: specify the first test to run (currently: %i; max %zu)\n",
+		targs->tfirst, NTESTS - 1);
+	pr_info(" -e: specify the last test to run (currently: %i; max %zu)\n",
+		targs->tlast, NTESTS - 1);
+	pr_info(" -l: specify the test length in seconds (currently: %i)\n",
+		targs->seconds);
+	pr_info(" -r: specify the number of runs per test (currently: %i)\n",
+		targs->runs);
+
+	pr_info("\nAvailable tests:\n");
+	for (ctr = 0; ctr < NTESTS; ctr++)
+		pr_info("%d: %s\n", ctr, tests[ctr].name);
+}
+
+static bool parse_args(int argc, char *argv[],
+		       struct test_args *targs)
+{
+	int opt;
+
+	while ((opt = getopt(argc, argv, "hvds:f:e:l:r:")) != -1) {
+		switch (opt) {
+		case 'h':
+		default:
+			help(argv[0], targs);
+			return false;
+		case 'v':
+			verbose = true;
+			break;
+		case 'd':
+			map_unmap_verify = true;
+			break;
+		case 's':
+			targs->nslots = atoi(optarg);
+			if (targs->nslots <= 0 && targs->nslots != -1) {
+				pr_info("Slot count cap has to be positive or -1 for no cap\n");
+				return false;
+			}
+			break;
+		case 'f':
+			targs->tfirst = atoi(optarg);
+			if (targs->tfirst < 0) {
+				pr_info("First test to run has to be non-negative\n");
+				return false;
+			}
+			break;
+		case 'e':
+			targs->tlast = atoi(optarg);
+			if (targs->tlast < 0 || targs->tlast >= NTESTS) {
+				pr_info("Last test to run has to be non-negative and less than %zu\n",
+					NTESTS);
+				return false;
+			}
+			break;
+		case 'l':
+			targs->seconds = atoi(optarg);
+			if (targs->seconds < 0) {
+				pr_info("Test length in seconds has to be non-negative\n");
+				return false;
+			}
+			break;
+		case 'r':
+			targs->runs = atoi(optarg);
+			if (targs->runs <= 0) {
+				pr_info("Runs per test has to be positive\n");
+				return false;
+			}
+			break;
+		}
+	}
+
+	if (optind < argc) {
+		help(argv[0], targs);
+		return false;
+	}
+
+	if (targs->tfirst > targs->tlast) {
+		pr_info("First test to run cannot be greater than the last test to run\n");
+		return false;
+	}
+
+	return true;
+}
+
+struct test_result {
+	struct timespec slot_runtime, guest_runtime, iter_runtime;
+	int64_t slottimens, runtimens;
+	uint64_t nloops;
+};
+
+static bool test_loop(const struct test_data *data,
+		      const struct test_args *targs,
+		      struct test_result *rbestslottime,
+		      struct test_result *rbestruntime)
+{
+	uint64_t maxslots;
+	struct test_result result;
+
+	result.nloops = 0;
+	if (!test_execute(targs->nslots, &maxslots, targs->seconds, data,
+			  &result.nloops,
+			  &result.slot_runtime, &result.guest_runtime)) {
+		if (maxslots)
+			pr_info("Memslot count too high for this test, decrease the cap (max is %"PRIu64")\n",
+				maxslots);
+		else
+			pr_info("Memslot count may be too high for this test, try adjusting the cap\n");
+
+		return false;
+	}
+
+	pr_info("Test took %ld.%.9lds for slot setup + %ld.%.9lds all iterations\n",
+		result.slot_runtime.tv_sec, result.slot_runtime.tv_nsec,
+		result.guest_runtime.tv_sec, result.guest_runtime.tv_nsec);
+	if (!result.nloops) {
+		pr_info("No full loops done - too short test time or system too loaded?\n");
+		return true;
+	}
+
+	result.iter_runtime = timespec_div(result.guest_runtime,
+					   result.nloops);
+	pr_info("Done %"PRIu64" iterations, avg %ld.%.9lds each\n",
+		result.nloops,
+		result.iter_runtime.tv_sec,
+		result.iter_runtime.tv_nsec);
+	result.slottimens = timespec_to_ns(result.slot_runtime);
+	result.runtimens = timespec_to_ns(result.iter_runtime);
+
+	/*
+	 * Only rank the slot setup time for tests using the whole test memory
+	 * area so they are comparable
+	 */
+	if (!data->mem_size &&
+	    (!rbestslottime->slottimens ||
+	     result.slottimens < rbestslottime->slottimens))
+		*rbestslottime = result;
+	if (!rbestruntime->runtimens ||
+	    result.runtimens < rbestruntime->runtimens)
+		*rbestruntime = result;
+
+	return true;
+}
+
+int main(int argc, char *argv[])
+{
+	struct test_args targs = {
+		.tfirst = 0,
+		.tlast = NTESTS - 1,
+		.nslots = -1,
+		.seconds = 5,
+		.runs = 1,
+	};
+	struct test_result rbestslottime;
+	int tctr;
+
+	/* Tell stdout not to buffer its content */
+	setbuf(stdout, NULL);
+
+	if (!parse_args(argc, argv, &targs))
+		return -1;
+
+	rbestslottime.slottimens = 0;
+	for (tctr = targs.tfirst; tctr <= targs.tlast; tctr++) {
+		const struct test_data *data = &tests[tctr];
+		unsigned int runctr;
+		struct test_result rbestruntime;
+
+		if (tctr > targs.tfirst)
+			pr_info("\n");
+
+		pr_info("Testing %s performance with %i runs, %d seconds each\n",
+			data->name, targs.runs, targs.seconds);
+
+		rbestruntime.runtimens = 0;
+		for (runctr = 0; runctr < targs.runs; runctr++)
+			if (!test_loop(data, &targs,
+				       &rbestslottime, &rbestruntime))
+				break;
+
+		if (rbestruntime.runtimens)
+			pr_info("Best runtime result was %ld.%.9lds per iteration (with %"PRIu64" iterations)\n",
+				rbestruntime.iter_runtime.tv_sec,
+				rbestruntime.iter_runtime.tv_nsec,
+				rbestruntime.nloops);
+	}
+
+	if (rbestslottime.slottimens)
+		pr_info("Best slot setup time for the whole test area was %ld.%.9lds\n",
+			rbestslottime.slot_runtime.tv_sec,
+			rbestslottime.slot_runtime.tv_nsec);
+
+	return 0;
+}

tools/testing/selftests/kvm/x86_64/get_cpuid_test.c

···
 	u32 function;
 	u32 index;
 } mangled_cpuids[] = {
+	/*
+	 * These entries depend on the vCPU's XCR0 register and IA32_XSS MSR,
+	 * which are not controlled for by this test.
+	 */
 	{.function = 0xd, .index = 0},
+	{.function = 0xd, .index = 1},
 };
 
 static void test_guest_cpuids(struct kvm_cpuid2 *guest_cpuid)

tools/testing/selftests/kvm/x86_64/get_msr_index_features.c

···
 	int old_res, res, kvm_fd, r;
 	struct kvm_msr_list *list;
 
-	kvm_fd = open(KVM_DEV_PATH, O_RDONLY);
-	if (kvm_fd < 0)
-		exit(KSFT_SKIP);
+	kvm_fd = open_kvm_dev_path_or_exit();
 
 	old_res = kvm_num_index_msrs(kvm_fd, 0);
 	TEST_ASSERT(old_res != 0, "Expecting nmsrs to be > 0");
···
 	int res, old_res, i, kvm_fd;
 	struct kvm_msr_list *feature_list;
 
-	kvm_fd = open(KVM_DEV_PATH, O_RDONLY);
-	if (kvm_fd < 0)
-		exit(KSFT_SKIP);
+	kvm_fd = open_kvm_dev_path_or_exit();
 
 	old_res = kvm_num_feature_msrs(kvm_fd, 0);
 	TEST_ASSERT(old_res != 0, "Expecting nmsrs to be > 0");

virt/kvm/kvm_main.c

···
 {
 	return kvm_make_all_cpus_request_except(kvm, req, NULL);
 }
+EXPORT_SYMBOL_GPL(kvm_make_all_cpus_request);
 
 #ifndef CONFIG_HAVE_KVM_ARCH_TLB_FLUSH_ALL
 void kvm_flush_remote_tlbs(struct kvm *kvm)
···
 		goto out;
 	if (signal_pending(current))
 		goto out;
+	if (kvm_check_request(KVM_REQ_UNBLOCK, vcpu))
+		goto out;
 
 	ret = 0;
 out:
···
 				goto out;
 			}
 			poll_end = cur = ktime_get();
-		} while (single_task_running() && !need_resched() &&
-			 ktime_before(cur, stop));
+		} while (kvm_vcpu_can_poll(cur, stop));
 	}
 
 	prepare_to_rcuwait(&vcpu->wait);

virt/lib/irqbypass.c

···
 	if (prod->add_consumer)
 		ret = prod->add_consumer(prod, cons);
 
-	if (ret)
-		goto err_add_consumer;
-
-	ret = cons->add_producer(cons, prod);
-	if (ret)
-		goto err_add_producer;
+	if (!ret) {
+		ret = cons->add_producer(cons, prod);
+		if (ret && prod->del_consumer)
+			prod->del_consumer(prod, cons);
+	}
 
 	if (cons->start)
 		cons->start(cons);
 	if (prod->start)
 		prod->start(prod);
-err_add_producer:
-	if (prod->del_consumer)
-		prod->del_consumer(prod, cons);
-err_add_consumer:
+
 	return ret;
 }