Merge tag 'kvmarm-fixes-6.14-2' of git://git.kernel.org/pub/scm/linux/kernel/git/kvmarm/kvmarm into HEAD

KVM/arm64 fixes for 6.14, take #2

- Large set of fixes for vector handling, specially in the interactions
between host and guest state. This fixes a number of bugs affecting
actual deployments, and greatly simplifies the FP/SIMD/SVE handling.
Thanks to Mark Rutland for dealing with this thankless task.

- Fix an ugly race between vcpu and vgic creation/init, resulting in
unexpected behaviours.

- Fix use of kernel VAs at EL2 when emulating timers with nVHE.

- Small set of pKVM improvements and cleanups.

arch/arm64/include/asm/kvm_emulate.h

···
 					 __cpacr_to_cptr_set(clr, set));\
 	} while (0)
 
-static __always_inline void kvm_write_cptr_el2(u64 val)
-{
-	if (has_vhe() || has_hvhe())
-		write_sysreg(val, cpacr_el1);
-	else
-		write_sysreg(val, cptr_el2);
-}
-
-/* Resets the value of cptr_el2 when returning to the host. */
-static __always_inline void __kvm_reset_cptr_el2(struct kvm *kvm)
-{
-	u64 val;
-
-	if (has_vhe()) {
-		val = (CPACR_EL1_FPEN | CPACR_EL1_ZEN_EL1EN);
-		if (cpus_have_final_cap(ARM64_SME))
-			val |= CPACR_EL1_SMEN_EL1EN;
-	} else if (has_hvhe()) {
-		val = CPACR_EL1_FPEN;
-
-		if (!kvm_has_sve(kvm) || !guest_owns_fp_regs())
-			val |= CPACR_EL1_ZEN;
-		if (cpus_have_final_cap(ARM64_SME))
-			val |= CPACR_EL1_SMEN;
-	} else {
-		val = CPTR_NVHE_EL2_RES1;
-
-		if (kvm_has_sve(kvm) && guest_owns_fp_regs())
-			val |= CPTR_EL2_TZ;
-		if (!cpus_have_final_cap(ARM64_SME))
-			val |= CPTR_EL2_TSM;
-	}
-
-	kvm_write_cptr_el2(val);
-}
-
-#ifdef __KVM_NVHE_HYPERVISOR__
-#define kvm_reset_cptr_el2(v)	__kvm_reset_cptr_el2(kern_hyp_va((v)->kvm))
-#else
-#define kvm_reset_cptr_el2(v)	__kvm_reset_cptr_el2((v)->kvm)
-#endif
-
 /*
  * Returns a 'sanitised' view of CPTR_EL2, translating from nVHE to the VHE
  * format if E2H isn't set.

arch/arm64/include/asm/kvm_host.h

···
 static inline void *pop_hyp_memcache(struct kvm_hyp_memcache *mc,
 				     void *(*to_va)(phys_addr_t phys))
 {
-	phys_addr_t *p = to_va(mc->head);
+	phys_addr_t *p = to_va(mc->head & PAGE_MASK);
 
 	if (!mc->nr_pages)
 		return NULL;
···
 struct kvm_host_data {
 #define KVM_HOST_DATA_FLAG_HAS_SPE			0
 #define KVM_HOST_DATA_FLAG_HAS_TRBE			1
-#define KVM_HOST_DATA_FLAG_HOST_SVE_ENABLED		2
-#define KVM_HOST_DATA_FLAG_HOST_SME_ENABLED		3
 #define KVM_HOST_DATA_FLAG_TRBE_ENABLED			4
 #define KVM_HOST_DATA_FLAG_EL1_TRACING_CONFIGURED	5
 	unsigned long flags;
···
 	struct kvm_cpu_context host_ctxt;
 
 	/*
-	 * All pointers in this union are hyp VA.
+	 * Hyp VA.
 	 * sve_state is only used in pKVM and if system_supports_sve().
 	 */
-	union {
-		struct user_fpsimd_state *fpsimd_state;
-		struct cpu_sve_state *sve_state;
-	};
+	struct cpu_sve_state *sve_state;
 
-	union {
-		/* HYP VA pointer to the host storage for FPMR */
-		u64	*fpmr_ptr;
-		/*
-		 * Used by pKVM only, as it needs to provide storage
-		 * for the host
-		 */
-		u64	fpmr;
-	};
+	/* Used by pKVM only. */
+	u64	fpmr;
 
 	/* Ownership of the FP regs */
 	enum {

arch/arm64/kernel/fpsimd.c

···
 }
 
 /*
- * Called by KVM when entering the guest.
- */
-void fpsimd_kvm_prepare(void)
-{
-	if (!system_supports_sve())
-		return;
-
-	/*
-	 * KVM does not save host SVE state since we can only enter
-	 * the guest from a syscall so the ABI means that only the
-	 * non-saved SVE state needs to be saved.  If we have left
-	 * SVE enabled for performance reasons then update the task
-	 * state to be FPSIMD only.
-	 */
-	get_cpu_fpsimd_context();
-
-	if (test_and_clear_thread_flag(TIF_SVE)) {
-		sve_to_fpsimd(current);
-		current->thread.fp_type = FP_STATE_FPSIMD;
-	}
-
-	put_cpu_fpsimd_context();
-}
-
-/*
  * Associate current's FPSIMD context with this cpu
  * The caller must have ownership of the cpu FPSIMD context before calling
  * this function.

arch/arm64/kvm/arch_timer.c

···
 static void kvm_timer_update_irq(struct kvm_vcpu *vcpu, bool new_level,
 				 struct arch_timer_context *timer_ctx)
 {
-	int ret;
-
 	kvm_timer_update_status(timer_ctx, new_level);
 
 	timer_ctx->irq.level = new_level;
 	trace_kvm_timer_update_irq(vcpu->vcpu_id, timer_irq(timer_ctx),
 				   timer_ctx->irq.level);
 
-	if (!userspace_irqchip(vcpu->kvm)) {
-		ret = kvm_vgic_inject_irq(vcpu->kvm, vcpu,
-					  timer_irq(timer_ctx),
-					  timer_ctx->irq.level,
-					  timer_ctx);
-		WARN_ON(ret);
-	}
+	if (userspace_irqchip(vcpu->kvm))
+		return;
+
+	kvm_vgic_inject_irq(vcpu->kvm, vcpu,
+			    timer_irq(timer_ctx),
+			    timer_ctx->irq.level,
+			    timer_ctx);
 }
 
 /* Only called for a fully emulated timer */

arch/arm64/kvm/arm.c

···
 			per_cpu_ptr_nvhe_sym(kvm_host_data, cpu)->sve_state =
 				kern_hyp_va(sve_state);
 		}
-	} else {
-		for_each_possible_cpu(cpu) {
-			struct user_fpsimd_state *fpsimd_state;
-
-			fpsimd_state = &per_cpu_ptr_nvhe_sym(kvm_host_data, cpu)->host_ctxt.fp_regs;
-			per_cpu_ptr_nvhe_sym(kvm_host_data, cpu)->fpsimd_state =
-				kern_hyp_va(fpsimd_state);
-		}
 	}
 }
 

arch/arm64/kvm/fpsimd.c

···
 	if (!system_supports_fpsimd())
 		return;
 
-	fpsimd_kvm_prepare();
-
 	/*
-	 * We will check TIF_FOREIGN_FPSTATE just before entering the
-	 * guest in kvm_arch_vcpu_ctxflush_fp() and override this to
-	 * FP_STATE_FREE if the flag set.
+	 * Ensure that any host FPSIMD/SVE/SME state is saved and unbound such
+	 * that the host kernel is responsible for restoring this state upon
+	 * return to userspace, and the hyp code doesn't need to save anything.
+	 *
+	 * When the host may use SME, fpsimd_save_and_flush_cpu_state() ensures
+	 * that PSTATE.{SM,ZA} == {0,0}.
 	 */
-	*host_data_ptr(fp_owner) = FP_STATE_HOST_OWNED;
-	*host_data_ptr(fpsimd_state) = kern_hyp_va(&current->thread.uw.fpsimd_state);
-	*host_data_ptr(fpmr_ptr) = kern_hyp_va(&current->thread.uw.fpmr);
+	fpsimd_save_and_flush_cpu_state();
+	*host_data_ptr(fp_owner) = FP_STATE_FREE;
 
-	host_data_clear_flag(HOST_SVE_ENABLED);
-	if (read_sysreg(cpacr_el1) & CPACR_EL1_ZEN_EL0EN)
-		host_data_set_flag(HOST_SVE_ENABLED);
-
-	if (system_supports_sme()) {
-		host_data_clear_flag(HOST_SME_ENABLED);
-		if (read_sysreg(cpacr_el1) & CPACR_EL1_SMEN_EL0EN)
-			host_data_set_flag(HOST_SME_ENABLED);
-
-		/*
-		 * If PSTATE.SM is enabled then save any pending FP
-		 * state and disable PSTATE.SM. If we leave PSTATE.SM
-		 * enabled and the guest does not enable SME via
-		 * CPACR_EL1.SMEN then operations that should be valid
-		 * may generate SME traps from EL1 to EL1 which we
-		 * can't intercept and which would confuse the guest.
-		 *
-		 * Do the same for PSTATE.ZA in the case where there
-		 * is state in the registers which has not already
-		 * been saved, this is very unlikely to happen.
-		 */
-		if (read_sysreg_s(SYS_SVCR) & (SVCR_SM_MASK | SVCR_ZA_MASK)) {
-			*host_data_ptr(fp_owner) = FP_STATE_FREE;
-			fpsimd_save_and_flush_cpu_state();
-		}
-	}
-
-	/*
-	 * If normal guests gain SME support, maintain this behavior for pKVM
-	 * guests, which don't support SME.
-	 */
-	WARN_ON(is_protected_kvm_enabled() && system_supports_sme() &&
-		read_sysreg_s(SYS_SVCR));
+	WARN_ON_ONCE(system_supports_sme() && read_sysreg_s(SYS_SVCR));
 }
 
 /*
···
 
 	local_irq_save(flags);
 
-	/*
-	 * If we have VHE then the Hyp code will reset CPACR_EL1 to
-	 * the default value and we need to reenable SME.
-	 */
-	if (has_vhe() && system_supports_sme()) {
-		/* Also restore EL0 state seen on entry */
-		if (host_data_test_flag(HOST_SME_ENABLED))
-			sysreg_clear_set(CPACR_EL1, 0, CPACR_EL1_SMEN);
-		else
-			sysreg_clear_set(CPACR_EL1,
-					 CPACR_EL1_SMEN_EL0EN,
-					 CPACR_EL1_SMEN_EL1EN);
-		isb();
-	}
-
 	if (guest_owns_fp_regs()) {
-		if (vcpu_has_sve(vcpu)) {
-			u64 zcr = read_sysreg_el1(SYS_ZCR);
-
-			/*
-			 * If the vCPU is in the hyp context then ZCR_EL1 is
-			 * loaded with its vEL2 counterpart.
-			 */
-			__vcpu_sys_reg(vcpu, vcpu_sve_zcr_elx(vcpu)) = zcr;
-
-			/*
-			 * Restore the VL that was saved when bound to the CPU,
-			 * which is the maximum VL for the guest. Because the
-			 * layout of the data when saving the sve state depends
-			 * on the VL, we need to use a consistent (i.e., the
-			 * maximum) VL.
-			 * Note that this means that at guest exit ZCR_EL1 is
-			 * not necessarily the same as on guest entry.
-			 *
-			 * ZCR_EL2 holds the guest hypervisor's VL when running
-			 * a nested guest, which could be smaller than the
-			 * max for the vCPU. Similar to above, we first need to
-			 * switch to a VL consistent with the layout of the
-			 * vCPU's SVE state. KVM support for NV implies VHE, so
-			 * using the ZCR_EL1 alias is safe.
-			 */
-			if (!has_vhe() || (vcpu_has_nv(vcpu) && !is_hyp_ctxt(vcpu)))
-				sve_cond_update_zcr_vq(vcpu_sve_max_vq(vcpu) - 1,
-						       SYS_ZCR_EL1);
-		}
-
 		/*
 		 * Flush (save and invalidate) the fpsimd/sve state so that if
 		 * the host tries to use fpsimd/sve, it's not using stale data
···
 		 * when needed.
 		 */
 		fpsimd_save_and_flush_cpu_state();
-	} else if (has_vhe() && system_supports_sve()) {
-		/*
-		 * The FPSIMD/SVE state in the CPU has not been touched, and we
-		 * have SVE (and VHE): CPACR_EL1 (alias CPTR_EL2) has been
-		 * reset by kvm_reset_cptr_el2() in the Hyp code, disabling SVE
-		 * for EL0.  To avoid spurious traps, restore the trap state
-		 * seen by kvm_arch_vcpu_load_fp():
-		 */
-		if (host_data_test_flag(HOST_SVE_ENABLED))
-			sysreg_clear_set(CPACR_EL1, 0, CPACR_EL1_ZEN_EL0EN);
-		else
-			sysreg_clear_set(CPACR_EL1, CPACR_EL1_ZEN_EL0EN, 0);
 	}
 
 	local_irq_restore(flags);

arch/arm64/kvm/hyp/entry.S

···
 alternative_else_nop_endif
 	mrs	x1, isr_el1
 	cbz	x1,  1f
+
+	// Ensure that __guest_enter() always provides a context
+	// synchronization event so that callers don't need ISBs for anything
+	// that would usually be synchonized by the ERET.
+	isb
 	mov	x0, #ARM_EXCEPTION_IRQ
 	ret
 

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

···
 	return __get_fault_info(vcpu->arch.fault.esr_el2, &vcpu->arch.fault);
 }
 
-static bool kvm_hyp_handle_mops(struct kvm_vcpu *vcpu, u64 *exit_code)
+static inline bool kvm_hyp_handle_mops(struct kvm_vcpu *vcpu, u64 *exit_code)
 {
 	*vcpu_pc(vcpu) = read_sysreg_el2(SYS_ELR);
 	arm64_mops_reset_regs(vcpu_gp_regs(vcpu), vcpu->arch.fault.esr_el2);
···
 			 true);
 }
 
-static void kvm_hyp_save_fpsimd_host(struct kvm_vcpu *vcpu);
+static inline void fpsimd_lazy_switch_to_guest(struct kvm_vcpu *vcpu)
+{
+	u64 zcr_el1, zcr_el2;
+
+	if (!guest_owns_fp_regs())
+		return;
+
+	if (vcpu_has_sve(vcpu)) {
+		/* A guest hypervisor may restrict the effective max VL. */
+		if (vcpu_has_nv(vcpu) && !is_hyp_ctxt(vcpu))
+			zcr_el2 = __vcpu_sys_reg(vcpu, ZCR_EL2);
+		else
+			zcr_el2 = vcpu_sve_max_vq(vcpu) - 1;
+
+		write_sysreg_el2(zcr_el2, SYS_ZCR);
+
+		zcr_el1 = __vcpu_sys_reg(vcpu, vcpu_sve_zcr_elx(vcpu));
+		write_sysreg_el1(zcr_el1, SYS_ZCR);
+	}
+}
+
+static inline void fpsimd_lazy_switch_to_host(struct kvm_vcpu *vcpu)
+{
+	u64 zcr_el1, zcr_el2;
+
+	if (!guest_owns_fp_regs())
+		return;
+
+	/*
+	 * When the guest owns the FP regs, we know that guest+hyp traps for
+	 * any FPSIMD/SVE/SME features exposed to the guest have been disabled
+	 * by either fpsimd_lazy_switch_to_guest() or kvm_hyp_handle_fpsimd()
+	 * prior to __guest_entry(). As __guest_entry() guarantees a context
+	 * synchronization event, we don't need an ISB here to avoid taking
+	 * traps for anything that was exposed to the guest.
+	 */
+	if (vcpu_has_sve(vcpu)) {
+		zcr_el1 = read_sysreg_el1(SYS_ZCR);
+		__vcpu_sys_reg(vcpu, vcpu_sve_zcr_elx(vcpu)) = zcr_el1;
+
+		/*
+		 * The guest's state is always saved using the guest's max VL.
+		 * Ensure that the host has the guest's max VL active such that
+		 * the host can save the guest's state lazily, but don't
+		 * artificially restrict the host to the guest's max VL.
+		 */
+		if (has_vhe()) {
+			zcr_el2 = vcpu_sve_max_vq(vcpu) - 1;
+			write_sysreg_el2(zcr_el2, SYS_ZCR);
+		} else {
+			zcr_el2 = sve_vq_from_vl(kvm_host_sve_max_vl) - 1;
+			write_sysreg_el2(zcr_el2, SYS_ZCR);
+
+			zcr_el1 = vcpu_sve_max_vq(vcpu) - 1;
+			write_sysreg_el1(zcr_el1, SYS_ZCR);
+		}
+	}
+}
+
+static void kvm_hyp_save_fpsimd_host(struct kvm_vcpu *vcpu)
+{
+	/*
+	 * Non-protected kvm relies on the host restoring its sve state.
+	 * Protected kvm restores the host's sve state as not to reveal that
+	 * fpsimd was used by a guest nor leak upper sve bits.
+	 */
+	if (system_supports_sve()) {
+		__hyp_sve_save_host();
+
+		/* Re-enable SVE traps if not supported for the guest vcpu. */
+		if (!vcpu_has_sve(vcpu))
+			cpacr_clear_set(CPACR_EL1_ZEN, 0);
+
+	} else {
+		__fpsimd_save_state(host_data_ptr(host_ctxt.fp_regs));
+	}
+
+	if (kvm_has_fpmr(kern_hyp_va(vcpu->kvm)))
+		*host_data_ptr(fpmr) = read_sysreg_s(SYS_FPMR);
+}
+
 
 /*
  * We trap the first access to the FP/SIMD to save the host context and
···
  * If FP/SIMD is not implemented, handle the trap and inject an undefined
  * instruction exception to the guest. Similarly for trapped SVE accesses.
  */
-static bool kvm_hyp_handle_fpsimd(struct kvm_vcpu *vcpu, u64 *exit_code)
+static inline bool kvm_hyp_handle_fpsimd(struct kvm_vcpu *vcpu, u64 *exit_code)
 {
 	bool sve_guest;
 	u8 esr_ec;
···
 	isb();
 
 	/* Write out the host state if it's in the registers */
-	if (host_owns_fp_regs())
+	if (is_protected_kvm_enabled() && host_owns_fp_regs())
 		kvm_hyp_save_fpsimd_host(vcpu);
 
 	/* Restore the guest state */
···
 	return true;
 }
 
+/* Open-coded version of timer_get_offset() to allow for kern_hyp_va() */
+static inline u64 hyp_timer_get_offset(struct arch_timer_context *ctxt)
+{
+	u64 offset = 0;
+
+	if (ctxt->offset.vm_offset)
+		offset += *kern_hyp_va(ctxt->offset.vm_offset);
+	if (ctxt->offset.vcpu_offset)
+		offset += *kern_hyp_va(ctxt->offset.vcpu_offset);
+
+	return offset;
+}
+
 static inline u64 compute_counter_value(struct arch_timer_context *ctxt)
 {
-	return arch_timer_read_cntpct_el0() - timer_get_offset(ctxt);
+	return arch_timer_read_cntpct_el0() - hyp_timer_get_offset(ctxt);
 }
 
 static bool kvm_handle_cntxct(struct kvm_vcpu *vcpu)
···
 	return true;
 }
 
-static bool kvm_hyp_handle_sysreg(struct kvm_vcpu *vcpu, u64 *exit_code)
+static inline bool kvm_hyp_handle_sysreg(struct kvm_vcpu *vcpu, u64 *exit_code)
 {
 	if (cpus_have_final_cap(ARM64_WORKAROUND_CAVIUM_TX2_219_TVM) &&
 	    handle_tx2_tvm(vcpu))
···
 	return false;
 }
 
-static bool kvm_hyp_handle_cp15_32(struct kvm_vcpu *vcpu, u64 *exit_code)
+static inline bool kvm_hyp_handle_cp15_32(struct kvm_vcpu *vcpu, u64 *exit_code)
 {
 	if (static_branch_unlikely(&vgic_v3_cpuif_trap) &&
 	    __vgic_v3_perform_cpuif_access(vcpu) == 1)
···
 	return false;
 }
 
-static bool kvm_hyp_handle_memory_fault(struct kvm_vcpu *vcpu, u64 *exit_code)
+static inline bool kvm_hyp_handle_memory_fault(struct kvm_vcpu *vcpu,
+					       u64 *exit_code)
 {
 	if (!__populate_fault_info(vcpu))
 		return true;
 
 	return false;
 }
-static bool kvm_hyp_handle_iabt_low(struct kvm_vcpu *vcpu, u64 *exit_code)
-	__alias(kvm_hyp_handle_memory_fault);
-static bool kvm_hyp_handle_watchpt_low(struct kvm_vcpu *vcpu, u64 *exit_code)
-	__alias(kvm_hyp_handle_memory_fault);
+#define kvm_hyp_handle_iabt_low		kvm_hyp_handle_memory_fault
+#define kvm_hyp_handle_watchpt_low	kvm_hyp_handle_memory_fault
 
-static bool kvm_hyp_handle_dabt_low(struct kvm_vcpu *vcpu, u64 *exit_code)
+static inline bool kvm_hyp_handle_dabt_low(struct kvm_vcpu *vcpu, u64 *exit_code)
 {
 	if (kvm_hyp_handle_memory_fault(vcpu, exit_code))
 		return true;
···
 
 typedef bool (*exit_handler_fn)(struct kvm_vcpu *, u64 *);
 
-static const exit_handler_fn *kvm_get_exit_handler_array(struct kvm_vcpu *vcpu);
-
-static void early_exit_filter(struct kvm_vcpu *vcpu, u64 *exit_code);
-
 /*
  * Allow the hypervisor to handle the exit with an exit handler if it has one.
  *
  * Returns true if the hypervisor handled the exit, and control should go back
  * to the guest, or false if it hasn't.
  */
-static inline bool kvm_hyp_handle_exit(struct kvm_vcpu *vcpu, u64 *exit_code)
+static inline bool kvm_hyp_handle_exit(struct kvm_vcpu *vcpu, u64 *exit_code,
+				       const exit_handler_fn *handlers)
 {
-	const exit_handler_fn *handlers = kvm_get_exit_handler_array(vcpu);
-	exit_handler_fn fn;
-
-	fn = handlers[kvm_vcpu_trap_get_class(vcpu)];
-
+	exit_handler_fn fn = handlers[kvm_vcpu_trap_get_class(vcpu)];
 	if (fn)
 		return fn(vcpu, exit_code);
 
···
  * the guest, false when we should restore the host state and return to the
  * main run loop.
  */
-static inline bool fixup_guest_exit(struct kvm_vcpu *vcpu, u64 *exit_code)
+static inline bool __fixup_guest_exit(struct kvm_vcpu *vcpu, u64 *exit_code,
+				      const exit_handler_fn *handlers)
 {
-	/*
-	 * Save PSTATE early so that we can evaluate the vcpu mode
-	 * early on.
-	 */
-	synchronize_vcpu_pstate(vcpu, exit_code);
-
-	/*
-	 * Check whether we want to repaint the state one way or
-	 * another.
-	 */
-	early_exit_filter(vcpu, exit_code);
-
 	if (ARM_EXCEPTION_CODE(*exit_code) != ARM_EXCEPTION_IRQ)
 		vcpu->arch.fault.esr_el2 = read_sysreg_el2(SYS_ESR);
 
···
 		goto exit;
 
 	/* Check if there's an exit handler and allow it to handle the exit. */
-	if (kvm_hyp_handle_exit(vcpu, exit_code))
+	if (kvm_hyp_handle_exit(vcpu, exit_code, handlers))
 		goto guest;
 exit:
 	/* Return to the host kernel and handle the exit */

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

···
  */
 
 #include <hyp/adjust_pc.h>
+#include <hyp/switch.h>
 
 #include <asm/pgtable-types.h>
 #include <asm/kvm_asm.h>
···
 	if (system_supports_sve())
 		__hyp_sve_restore_host();
 	else
-		__fpsimd_restore_state(*host_data_ptr(fpsimd_state));
+		__fpsimd_restore_state(host_data_ptr(host_ctxt.fp_regs));
 
 	if (has_fpmr)
 		write_sysreg_s(*host_data_ptr(fpmr), SYS_FPMR);
···
 
 		sync_hyp_vcpu(hyp_vcpu);
 	} else {
+		struct kvm_vcpu *vcpu = kern_hyp_va(host_vcpu);
+
 		/* The host is fully trusted, run its vCPU directly. */
-		ret = __kvm_vcpu_run(kern_hyp_va(host_vcpu));
+		fpsimd_lazy_switch_to_guest(vcpu);
+		ret = __kvm_vcpu_run(vcpu);
+		fpsimd_lazy_switch_to_host(vcpu);
 	}
 out:
 	cpu_reg(host_ctxt, 1) =  ret;
···
 		break;
 	case ESR_ELx_EC_SMC64:
 		handle_host_smc(host_ctxt);
-		break;
-	case ESR_ELx_EC_SVE:
-		cpacr_clear_set(0, CPACR_EL1_ZEN);
-		isb();
-		sve_cond_update_zcr_vq(sve_vq_from_vl(kvm_host_sve_max_vl) - 1,
-				       SYS_ZCR_EL2);
 		break;
 	case ESR_ELx_EC_IABT_LOW:
 	case ESR_ELx_EC_DABT_LOW:

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

···
 	ret = kvm_pgtable_get_leaf(&vm->pgt, ipa, &pte, &level);
 	if (ret)
 		return ret;
-	if (level != KVM_PGTABLE_LAST_LEVEL)
-		return -E2BIG;
 	if (!kvm_pte_valid(pte))
 		return -ENOENT;
+	if (level != KVM_PGTABLE_LAST_LEVEL)
+		return -E2BIG;
 
 	state = guest_get_page_state(pte, ipa);
 	if (state != PKVM_PAGE_SHARED_BORROWED)
···
 	return ret;
 }
 
-int __pkvm_host_relax_perms_guest(u64 gfn, struct pkvm_hyp_vcpu *vcpu, enum kvm_pgtable_prot prot)
+static void assert_host_shared_guest(struct pkvm_hyp_vm *vm, u64 ipa)
 {
-	struct pkvm_hyp_vm *vm = pkvm_hyp_vcpu_to_hyp_vm(vcpu);
-	u64 ipa = hyp_pfn_to_phys(gfn);
 	u64 phys;
 	int ret;
 
-	if (prot & ~KVM_PGTABLE_PROT_RWX)
-		return -EINVAL;
+	if (!IS_ENABLED(CONFIG_NVHE_EL2_DEBUG))
+		return;
 
 	host_lock_component();
 	guest_lock_component(vm);
 
 	ret = __check_host_shared_guest(vm, &phys, ipa);
-	if (!ret)
-		ret = kvm_pgtable_stage2_relax_perms(&vm->pgt, ipa, prot, 0);
 
 	guest_unlock_component(vm);
 	host_unlock_component();
+
+	WARN_ON(ret && ret != -ENOENT);
+}
+
+int __pkvm_host_relax_perms_guest(u64 gfn, struct pkvm_hyp_vcpu *vcpu, enum kvm_pgtable_prot prot)
+{
+	struct pkvm_hyp_vm *vm = pkvm_hyp_vcpu_to_hyp_vm(vcpu);
+	u64 ipa = hyp_pfn_to_phys(gfn);
+	int ret;
+
+	if (pkvm_hyp_vm_is_protected(vm))
+		return -EPERM;
+
+	if (prot & ~KVM_PGTABLE_PROT_RWX)
+		return -EINVAL;
+
+	assert_host_shared_guest(vm, ipa);
+	guest_lock_component(vm);
+	ret = kvm_pgtable_stage2_relax_perms(&vm->pgt, ipa, prot, 0);
+	guest_unlock_component(vm);
 
 	return ret;
 }
···
 int __pkvm_host_wrprotect_guest(u64 gfn, struct pkvm_hyp_vm *vm)
 {
 	u64 ipa = hyp_pfn_to_phys(gfn);
-	u64 phys;
 	int ret;
 
-	host_lock_component();
+	if (pkvm_hyp_vm_is_protected(vm))
+		return -EPERM;
+
+	assert_host_shared_guest(vm, ipa);
 	guest_lock_component(vm);
-
-	ret = __check_host_shared_guest(vm, &phys, ipa);
-	if (!ret)
-		ret = kvm_pgtable_stage2_wrprotect(&vm->pgt, ipa, PAGE_SIZE);
-
+	ret = kvm_pgtable_stage2_wrprotect(&vm->pgt, ipa, PAGE_SIZE);
 	guest_unlock_component(vm);
-	host_unlock_component();
 
 	return ret;
 }
···
 int __pkvm_host_test_clear_young_guest(u64 gfn, bool mkold, struct pkvm_hyp_vm *vm)
 {
 	u64 ipa = hyp_pfn_to_phys(gfn);
-	u64 phys;
 	int ret;
 
-	host_lock_component();
+	if (pkvm_hyp_vm_is_protected(vm))
+		return -EPERM;
+
+	assert_host_shared_guest(vm, ipa);
 	guest_lock_component(vm);
-
-	ret = __check_host_shared_guest(vm, &phys, ipa);
-	if (!ret)
-		ret = kvm_pgtable_stage2_test_clear_young(&vm->pgt, ipa, PAGE_SIZE, mkold);
-
+	ret = kvm_pgtable_stage2_test_clear_young(&vm->pgt, ipa, PAGE_SIZE, mkold);
 	guest_unlock_component(vm);
-	host_unlock_component();
 
 	return ret;
 }
···
 {
 	struct pkvm_hyp_vm *vm = pkvm_hyp_vcpu_to_hyp_vm(vcpu);
 	u64 ipa = hyp_pfn_to_phys(gfn);
-	u64 phys;
-	int ret;
 
-	host_lock_component();
+	if (pkvm_hyp_vm_is_protected(vm))
+		return -EPERM;
+
+	assert_host_shared_guest(vm, ipa);
 	guest_lock_component(vm);
-
-	ret = __check_host_shared_guest(vm, &phys, ipa);
-	if (!ret)
-		kvm_pgtable_stage2_mkyoung(&vm->pgt, ipa, 0);
-
+	kvm_pgtable_stage2_mkyoung(&vm->pgt, ipa, 0);
 	guest_unlock_component(vm);
-	host_unlock_component();
 
-	return ret;
+	return 0;
 }

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

···
 {
 	u64 val = CPTR_EL2_TAM;	/* Same bit irrespective of E2H */
 
+	if (!guest_owns_fp_regs())
+		__activate_traps_fpsimd32(vcpu);
+
 	if (has_hvhe()) {
 		val |= CPACR_EL1_TTA;
 
···
 			if (vcpu_has_sve(vcpu))
 				val |= CPACR_EL1_ZEN;
 		}
+
+		write_sysreg(val, cpacr_el1);
 	} else {
 		val |= CPTR_EL2_TTA | CPTR_NVHE_EL2_RES1;
 
···
 
 		if (!guest_owns_fp_regs())
 			val |= CPTR_EL2_TFP;
+
+		write_sysreg(val, cptr_el2);
 	}
+}
 
-	if (!guest_owns_fp_regs())
-		__activate_traps_fpsimd32(vcpu);
+static void __deactivate_cptr_traps(struct kvm_vcpu *vcpu)
+{
+	if (has_hvhe()) {
+		u64 val = CPACR_EL1_FPEN;
 
-	kvm_write_cptr_el2(val);
+		if (cpus_have_final_cap(ARM64_SVE))
+			val |= CPACR_EL1_ZEN;
+		if (cpus_have_final_cap(ARM64_SME))
+			val |= CPACR_EL1_SMEN;
+
+		write_sysreg(val, cpacr_el1);
+	} else {
+		u64 val = CPTR_NVHE_EL2_RES1;
+
+		if (!cpus_have_final_cap(ARM64_SVE))
+			val |= CPTR_EL2_TZ;
+		if (!cpus_have_final_cap(ARM64_SME))
+			val |= CPTR_EL2_TSM;
+
+		write_sysreg(val, cptr_el2);
+	}
 }
 
 static void __activate_traps(struct kvm_vcpu *vcpu)
···
 
 	write_sysreg(this_cpu_ptr(&kvm_init_params)->hcr_el2, hcr_el2);
 
-	kvm_reset_cptr_el2(vcpu);
+	__deactivate_cptr_traps(vcpu);
 	write_sysreg(__kvm_hyp_host_vector, vbar_el2);
 }
 
···
 		kvm_handle_pvm_sysreg(vcpu, exit_code));
 }
 
-static void kvm_hyp_save_fpsimd_host(struct kvm_vcpu *vcpu)
-{
-	/*
-	 * Non-protected kvm relies on the host restoring its sve state.
-	 * Protected kvm restores the host's sve state as not to reveal that
-	 * fpsimd was used by a guest nor leak upper sve bits.
-	 */
-	if (unlikely(is_protected_kvm_enabled() && system_supports_sve())) {
-		__hyp_sve_save_host();
-
-		/* Re-enable SVE traps if not supported for the guest vcpu. */
-		if (!vcpu_has_sve(vcpu))
-			cpacr_clear_set(CPACR_EL1_ZEN, 0);
-
-	} else {
-		__fpsimd_save_state(*host_data_ptr(fpsimd_state));
-	}
-
-	if (kvm_has_fpmr(kern_hyp_va(vcpu->kvm))) {
-		u64 val = read_sysreg_s(SYS_FPMR);
-
-		if (unlikely(is_protected_kvm_enabled()))
-			*host_data_ptr(fpmr) = val;
-		else
-			**host_data_ptr(fpmr_ptr) = val;
-	}
-}
-
 static const exit_handler_fn hyp_exit_handlers[] = {
 	[0 ... ESR_ELx_EC_MAX]		= NULL,
 	[ESR_ELx_EC_CP15_32]		= kvm_hyp_handle_cp15_32,
···
 	return hyp_exit_handlers;
 }
 
-/*
- * Some guests (e.g., protected VMs) are not be allowed to run in AArch32.
- * The ARMv8 architecture does not give the hypervisor a mechanism to prevent a
- * guest from dropping to AArch32 EL0 if implemented by the CPU. If the
- * hypervisor spots a guest in such a state ensure it is handled, and don't
- * trust the host to spot or fix it.  The check below is based on the one in
- * kvm_arch_vcpu_ioctl_run().
- *
- * Returns false if the guest ran in AArch32 when it shouldn't have, and
- * thus should exit to the host, or true if a the guest run loop can continue.
- */
-static void early_exit_filter(struct kvm_vcpu *vcpu, u64 *exit_code)
+static inline bool fixup_guest_exit(struct kvm_vcpu *vcpu, u64 *exit_code)
 {
+	const exit_handler_fn *handlers = kvm_get_exit_handler_array(vcpu);
+
+	synchronize_vcpu_pstate(vcpu, exit_code);
+
+	/*
+	 * Some guests (e.g., protected VMs) are not be allowed to run in
+	 * AArch32.  The ARMv8 architecture does not give the hypervisor a
+	 * mechanism to prevent a guest from dropping to AArch32 EL0 if
+	 * implemented by the CPU. If the hypervisor spots a guest in such a
+	 * state ensure it is handled, and don't trust the host to spot or fix
+	 * it.  The check below is based on the one in
+	 * kvm_arch_vcpu_ioctl_run().
+	 */
 	if (unlikely(vcpu_is_protected(vcpu) && vcpu_mode_is_32bit(vcpu))) {
 		/*
 		 * As we have caught the guest red-handed, decide that it isn't
···
 		*exit_code &= BIT(ARM_EXIT_WITH_SERROR_BIT);
 		*exit_code |= ARM_EXCEPTION_IL;
 	}
+
+	return __fixup_guest_exit(vcpu, exit_code, handlers);
 }
 
 /* Switch to the guest for legacy non-VHE systems */

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

···
 	write_sysreg(val, cpacr_el1);
 }
 
+static void __deactivate_cptr_traps(struct kvm_vcpu *vcpu)
+{
+	u64 val = CPACR_EL1_FPEN | CPACR_EL1_ZEN_EL1EN;
+
+	if (cpus_have_final_cap(ARM64_SME))
+		val |= CPACR_EL1_SMEN_EL1EN;
+
+	write_sysreg(val, cpacr_el1);
+}
+
 static void __activate_traps(struct kvm_vcpu *vcpu)
 {
 	u64 val;
···
 	 */
 	asm(ALTERNATIVE("nop", "isb", ARM64_WORKAROUND_SPECULATIVE_AT));
 
-	kvm_reset_cptr_el2(vcpu);
+	__deactivate_cptr_traps(vcpu);
 
 	if (!arm64_kernel_unmapped_at_el0())
 		host_vectors = __this_cpu_read(this_cpu_vector);
···
 	return true;
 }
 
-static void kvm_hyp_save_fpsimd_host(struct kvm_vcpu *vcpu)
-{
-	__fpsimd_save_state(*host_data_ptr(fpsimd_state));
-
-	if (kvm_has_fpmr(vcpu->kvm))
-		**host_data_ptr(fpmr_ptr) = read_sysreg_s(SYS_FPMR);
-}
-
 static bool kvm_hyp_handle_tlbi_el2(struct kvm_vcpu *vcpu, u64 *exit_code)
 {
 	int ret = -EINVAL;
···
 	[ESR_ELx_EC_MOPS]		= kvm_hyp_handle_mops,
 };
 
-static const exit_handler_fn *kvm_get_exit_handler_array(struct kvm_vcpu *vcpu)
+static inline bool fixup_guest_exit(struct kvm_vcpu *vcpu, u64 *exit_code)
 {
-	return hyp_exit_handlers;
-}
+	synchronize_vcpu_pstate(vcpu, exit_code);
 
-static void early_exit_filter(struct kvm_vcpu *vcpu, u64 *exit_code)
-{
 	/*
 	 * If we were in HYP context on entry, adjust the PSTATE view
 	 * so that the usual helpers work correctly.
···
 		*vcpu_cpsr(vcpu) &= ~(PSR_MODE_MASK | PSR_MODE32_BIT);
 		*vcpu_cpsr(vcpu) |= mode;
 	}
+
+	return __fixup_guest_exit(vcpu, exit_code, hyp_exit_handlers);
 }
 
 /* Switch to the guest for VHE systems running in EL2 */
···
 	guest_ctxt = &vcpu->arch.ctxt;
 
 	sysreg_save_host_state_vhe(host_ctxt);
+
+	fpsimd_lazy_switch_to_guest(vcpu);
 
 	/*
 	 * Note that ARM erratum 1165522 requires us to configure both stage 1
···
 	sysreg_save_guest_state_vhe(guest_ctxt);
 
 	__deactivate_traps(vcpu);
+
+	fpsimd_lazy_switch_to_host(vcpu);
 
 	sysreg_restore_host_state_vhe(host_ctxt);
 

arch/arm64/kvm/vgic/vgic-init.c

···
  *
  * CPU Interface:
  *
- * - kvm_vgic_vcpu_init(): initialization of static data that
- *   doesn't depend on any sizing information or emulation type. No
- *   allocation is allowed there.
+ * - kvm_vgic_vcpu_init(): initialization of static data that doesn't depend
+ *   on any sizing information. Private interrupts are allocated if not
+ *   already allocated at vgic-creation time.
  */
 
 /* EARLY INIT */
···
 }
 
 /* CREATION */
+
+static int vgic_allocate_private_irqs_locked(struct kvm_vcpu *vcpu, u32 type);
 
 /**
  * kvm_vgic_create: triggered by the instantiation of the VGIC device by
···
 
 	if (atomic_read(&kvm->online_vcpus) > kvm->max_vcpus) {
 		ret = -E2BIG;
+		goto out_unlock;
+	}
+
+	kvm_for_each_vcpu(i, vcpu, kvm) {
+		ret = vgic_allocate_private_irqs_locked(vcpu, type);
+		if (ret)
+			break;
+	}
+
+	if (ret) {
+		kvm_for_each_vcpu(i, vcpu, kvm) {
+			struct vgic_cpu *vgic_cpu = &vcpu->arch.vgic_cpu;
+			kfree(vgic_cpu->private_irqs);
+			vgic_cpu->private_irqs = NULL;
+		}
+
 		goto out_unlock;
 	}
 
···
 	return 0;
 }
 
-static int vgic_allocate_private_irqs_locked(struct kvm_vcpu *vcpu)
+static int vgic_allocate_private_irqs_locked(struct kvm_vcpu *vcpu, u32 type)
 {
 	struct vgic_cpu *vgic_cpu = &vcpu->arch.vgic_cpu;
 	int i;
···
 			/* PPIs */
 			irq->config = VGIC_CONFIG_LEVEL;
 		}
+
+		switch (type) {
+		case KVM_DEV_TYPE_ARM_VGIC_V3:
+			irq->group = 1;
+			irq->mpidr = kvm_vcpu_get_mpidr_aff(vcpu);
+			break;
+		case KVM_DEV_TYPE_ARM_VGIC_V2:
+			irq->group = 0;
+			irq->targets = BIT(vcpu->vcpu_id);
+			break;
+		}
 	}
 
 	return 0;
 }
 
-static int vgic_allocate_private_irqs(struct kvm_vcpu *vcpu)
+static int vgic_allocate_private_irqs(struct kvm_vcpu *vcpu, u32 type)
 {
 	int ret;
 
 	mutex_lock(&vcpu->kvm->arch.config_lock);
-	ret = vgic_allocate_private_irqs_locked(vcpu);
+	ret = vgic_allocate_private_irqs_locked(vcpu, type);
 	mutex_unlock(&vcpu->kvm->arch.config_lock);
 
 	return ret;
···
 	if (!irqchip_in_kernel(vcpu->kvm))
 		return 0;
 
-	ret = vgic_allocate_private_irqs(vcpu);
+	ret = vgic_allocate_private_irqs(vcpu, dist->vgic_model);
 	if (ret)
 		return ret;
 
···
 {
 	struct vgic_dist *dist = &kvm->arch.vgic;
 	struct kvm_vcpu *vcpu;
-	int ret = 0, i;
+	int ret = 0;
 	unsigned long idx;
 
 	lockdep_assert_held(&kvm->arch.config_lock);
···
 	ret = kvm_vgic_dist_init(kvm, dist->nr_spis);
 	if (ret)
 		goto out;
-
-	/* Initialize groups on CPUs created before the VGIC type was known */
-	kvm_for_each_vcpu(idx, vcpu, kvm) {
-		ret = vgic_allocate_private_irqs_locked(vcpu);
-		if (ret)
-			goto out;
-
-		for (i = 0; i < VGIC_NR_PRIVATE_IRQS; i++) {
-			struct vgic_irq *irq = vgic_get_vcpu_irq(vcpu, i);
-
-			switch (dist->vgic_model) {
-			case KVM_DEV_TYPE_ARM_VGIC_V3:
-				irq->group = 1;
-				irq->mpidr = kvm_vcpu_get_mpidr_aff(vcpu);
-				break;
-			case KVM_DEV_TYPE_ARM_VGIC_V2:
-				irq->group = 0;
-				irq->targets = 1U << idx;
-				break;
-			default:
-				ret = -EINVAL;
-			}
-
-			vgic_put_irq(kvm, irq);
-
-			if (ret)
-				goto out;
-		}
-	}
 
 	/*
 	 * If we have GICv4.1 enabled, unconditionally request enable the