Merge branch kvm-arm64/host_data into kvmarm-master/next

* kvm-arm64/host_data:
  : .
  : Rationalise the host-specific data to live as part of the per-CPU state.
  :
  : From the cover letter:
  :
  : "It appears that over the years, we have accumulated a lot of cruft in
  : the kvm_vcpu_arch structure. Part of the gunk is data that is strictly
  : host CPU specific, and this result in two main problems:
  :
  : - the structure itself is stupidly large, over 8kB. With the
  :   arch-agnostic kvm_vcpu, we're above 10kB, which is insane. This has
  :   some ripple effects, as we need physically contiguous allocation to
  :   be able to map it at EL2 for !VHE. There is more to it though, as
  :   some data structures, although per-vcpu, could be allocated
  :   separately.
  :
  : - We lose track of the life-cycle of this data, because we're
  :   guaranteed that it will be around forever and we start relying on
  :   wrong assumptions. This is becoming a maintenance burden.
  :
  : This series rectifies some of these things, starting with the two main
  : offenders: debug and FP, a lot of which gets pushed out to the per-CPU
  : host structure. Indeed, their lifetime really isn't that of the vcpu,
  : but tied to the physical CPU the vpcu runs on.
  :
  : This results in a small reduction of the vcpu size, but mainly a much
  : clearer understanding of the life-cycle of these structures."
  : .
  KVM: arm64: Move management of __hyp_running_vcpu to load/put on VHE
  KVM: arm64: Exclude FP ownership from kvm_vcpu_arch
  KVM: arm64: Exclude host_fpsimd_state pointer from kvm_vcpu_arch
  KVM: arm64: Exclude mdcr_el2_host from kvm_vcpu_arch
  KVM: arm64: Exclude host_debug_data from vcpu_arch
  KVM: arm64: Add accessor for per-CPU state

Signed-off-by: Marc Zyngier <maz@kernel.org>

Author: Marc Zyngier

arch/arm64/include/asm/kvm_emulate.h

@@ -588,15 +588,15 @@ static __always_inline u64 kvm_get_reset_cptr_el2(struct kvm_vcpu *vcpu)
 		val = (CPACR_EL1_FPEN_EL0EN | CPACR_EL1_FPEN_EL1EN);
 
 		if (!vcpu_has_sve(vcpu) ||
-		    (vcpu->arch.fp_state != FP_STATE_GUEST_OWNED))
+		    (*host_data_ptr(fp_owner) != FP_STATE_GUEST_OWNED))
 			val |= CPACR_EL1_ZEN_EL1EN | CPACR_EL1_ZEN_EL0EN;
 		if (cpus_have_final_cap(ARM64_SME))
 			val |= CPACR_EL1_SMEN_EL1EN | CPACR_EL1_SMEN_EL0EN;
 	} else {
 		val = CPTR_NVHE_EL2_RES1;
 
 		if (vcpu_has_sve(vcpu) &&
-		    (vcpu->arch.fp_state == FP_STATE_GUEST_OWNED))
+		    (*host_data_ptr(fp_owner) == FP_STATE_GUEST_OWNED))
 			val |= CPTR_EL2_TZ;
 		if (cpus_have_final_cap(ARM64_SME))
 			val &= ~CPTR_EL2_TSM;

arch/arm64/include/asm/kvm_host.h

@@ -530,8 +530,42 @@ struct kvm_cpu_context {
 	u64 *vncr_array;
 };
 
+/*
+ * This structure is instantiated on a per-CPU basis, and contains
+ * data that is:
+ *
+ * - tied to a single physical CPU, and
+ * - either have a lifetime that does not extend past vcpu_put()
+ * - or is an invariant for the lifetime of the system
+ *
+ * Use host_data_ptr(field) as a way to access a pointer to such a
+ * field.
+ */
 struct kvm_host_data {
 	struct kvm_cpu_context host_ctxt;
+	struct user_fpsimd_state *fpsimd_state;	/* hyp VA */
+
+	/* Ownership of the FP regs */
+	enum {
+		FP_STATE_FREE,
+		FP_STATE_HOST_OWNED,
+		FP_STATE_GUEST_OWNED,
+	} fp_owner;
+
+	/*
+	 * host_debug_state contains the host registers which are
+	 * saved and restored during world switches.
+	 */
+	 struct {
+		/* {Break,watch}point registers */
+		struct kvm_guest_debug_arch regs;
+		/* Statistical profiling extension */
+		u64 pmscr_el1;
+		/* Self-hosted trace */
+		u64 trfcr_el1;
+		/* Values of trap registers for the host before guest entry. */
+		u64 mdcr_el2;
+	} host_debug_state;
 };
 
 struct kvm_host_psci_config {
@@ -592,19 +626,9 @@ struct kvm_vcpu_arch {
 	u64 mdcr_el2;
 	u64 cptr_el2;
 
-	/* Values of trap registers for the host before guest entry. */
-	u64 mdcr_el2_host;
-
 	/* Exception Information */
 	struct kvm_vcpu_fault_info fault;
 
-	/* Ownership of the FP regs */
-	enum {
-		FP_STATE_FREE,
-		FP_STATE_HOST_OWNED,
-		FP_STATE_GUEST_OWNED,
-	} fp_state;
-
 	/* Configuration flags, set once and for all before the vcpu can run */
 	u8 cflags;
 
@@ -627,11 +651,10 @@ struct kvm_vcpu_arch {
 	 * We maintain more than a single set of debug registers to support
 	 * debugging the guest from the host and to maintain separate host and
 	 * guest state during world switches. vcpu_debug_state are the debug
-	 * registers of the vcpu as the guest sees them.  host_debug_state are
-	 * the host registers which are saved and restored during
-	 * world switches. external_debug_state contains the debug
-	 * values we want to debug the guest. This is set via the
-	 * KVM_SET_GUEST_DEBUG ioctl.
+	 * registers of the vcpu as the guest sees them.
+	 *
+	 * external_debug_state contains the debug values we want to debug the
+	 * guest. This is set via the KVM_SET_GUEST_DEBUG ioctl.
 	 *
 	 * debug_ptr points to the set of debug registers that should be loaded
 	 * onto the hardware when running the guest.
@@ -640,18 +663,8 @@ struct kvm_vcpu_arch {
 	struct kvm_guest_debug_arch vcpu_debug_state;
 	struct kvm_guest_debug_arch external_debug_state;
 
-	struct user_fpsimd_state *host_fpsimd_state;	/* hyp VA */
 	struct task_struct *parent_task;
 
-	struct {
-		/* {Break,watch}point registers */
-		struct kvm_guest_debug_arch regs;
-		/* Statistical profiling extension */
-		u64 pmscr_el1;
-		/* Self-hosted trace */
-		u64 trfcr_el1;
-	} host_debug_state;
-
 	/* VGIC state */
 	struct vgic_cpu vgic_cpu;
 	struct arch_timer_cpu timer_cpu;
@@ -1168,6 +1181,32 @@ struct kvm_vcpu *kvm_mpidr_to_vcpu(struct kvm *kvm, unsigned long mpidr);
 
 DECLARE_KVM_HYP_PER_CPU(struct kvm_host_data, kvm_host_data);
 
+/*
+ * How we access per-CPU host data depends on the where we access it from,
+ * and the mode we're in:
+ *
+ * - VHE and nVHE hypervisor bits use their locally defined instance
+ *
+ * - the rest of the kernel use either the VHE or nVHE one, depending on
+ *   the mode we're running in.
+ *
+ *   Unless we're in protected mode, fully deprivileged, and the nVHE
+ *   per-CPU stuff is exclusively accessible to the protected EL2 code.
+ *   In this case, the EL1 code uses the *VHE* data as its private state
+ *   (which makes sense in a way as there shouldn't be any shared state
+ *   between the host and the hypervisor).
+ *
+ * Yes, this is all totally trivial. Shoot me now.
+ */
+#if defined(__KVM_NVHE_HYPERVISOR__) || defined(__KVM_VHE_HYPERVISOR__)
+#define host_data_ptr(f)	(&this_cpu_ptr(&kvm_host_data)->f)
+#else
+#define host_data_ptr(f)						\
+	(static_branch_unlikely(&kvm_protected_mode_initialized) ?	\
+	 &this_cpu_ptr(&kvm_host_data)->f :				\
+	 &this_cpu_ptr_hyp_sym(kvm_host_data)->f)
+#endif
+
 static inline void kvm_init_host_cpu_context(struct kvm_cpu_context *cpu_ctxt)
 {
 	/* The host's MPIDR is immutable, so let's set it up at boot time */

arch/arm64/kvm/arm.c

@@ -378,12 +378,6 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
 
 	vcpu->arch.mmu_page_cache.gfp_zero = __GFP_ZERO;
 
-	/*
-	 * Default value for the FP state, will be overloaded at load
-	 * time if we support FP (pretty likely)
-	 */
-	vcpu->arch.fp_state = FP_STATE_FREE;
-
 	/* Set up the timer */
 	kvm_timer_vcpu_init(vcpu);
 
@@ -1971,7 +1965,7 @@ static void cpu_set_hyp_vector(void)
 
 static void cpu_hyp_init_context(void)
 {
-	kvm_init_host_cpu_context(&this_cpu_ptr_hyp_sym(kvm_host_data)->host_ctxt);
+	kvm_init_host_cpu_context(host_data_ptr(host_ctxt));
 
 	if (!is_kernel_in_hyp_mode())
 		cpu_init_hyp_mode();

arch/arm64/kvm/fpsimd.c

@@ -49,8 +49,6 @@ int kvm_arch_vcpu_run_map_fp(struct kvm_vcpu *vcpu)
 	if (ret)
 		return ret;
 
-	vcpu->arch.host_fpsimd_state = kern_hyp_va(fpsimd);
-
 	/*
 	 * We need to keep current's task_struct pinned until its data has been
 	 * unshared with the hypervisor to make sure it is not re-used by the
@@ -86,7 +84,8 @@ void kvm_arch_vcpu_load_fp(struct kvm_vcpu *vcpu)
 	 * guest in kvm_arch_vcpu_ctxflush_fp() and override this to
 	 * FP_STATE_FREE if the flag set.
 	 */
-	vcpu->arch.fp_state = FP_STATE_HOST_OWNED;
+	*host_data_ptr(fp_owner) = FP_STATE_HOST_OWNED;
+	*host_data_ptr(fpsimd_state) = kern_hyp_va(&current->thread.uw.fpsimd_state);
 
 	vcpu_clear_flag(vcpu, HOST_SVE_ENABLED);
 	if (read_sysreg(cpacr_el1) & CPACR_EL1_ZEN_EL0EN)
@@ -110,7 +109,7 @@ void kvm_arch_vcpu_load_fp(struct kvm_vcpu *vcpu)
 		 * been saved, this is very unlikely to happen.
 		 */
 		if (read_sysreg_s(SYS_SVCR) & (SVCR_SM_MASK | SVCR_ZA_MASK)) {
-			vcpu->arch.fp_state = FP_STATE_FREE;
+			*host_data_ptr(fp_owner) = FP_STATE_FREE;
 			fpsimd_save_and_flush_cpu_state();
 		}
 	}
@@ -126,7 +125,7 @@ void kvm_arch_vcpu_load_fp(struct kvm_vcpu *vcpu)
 void kvm_arch_vcpu_ctxflush_fp(struct kvm_vcpu *vcpu)
 {
 	if (test_thread_flag(TIF_FOREIGN_FPSTATE))
-		vcpu->arch.fp_state = FP_STATE_FREE;
+		*host_data_ptr(fp_owner) = FP_STATE_FREE;
 }
 
 /*
@@ -142,7 +141,7 @@ void kvm_arch_vcpu_ctxsync_fp(struct kvm_vcpu *vcpu)
 
 	WARN_ON_ONCE(!irqs_disabled());
 
-	if (vcpu->arch.fp_state == FP_STATE_GUEST_OWNED) {
+	if (*host_data_ptr(fp_owner) == FP_STATE_GUEST_OWNED) {
 
 		/*
 		 * Currently we do not support SME guests so SVCR is
@@ -196,7 +195,7 @@ void kvm_arch_vcpu_put_fp(struct kvm_vcpu *vcpu)
 		isb();
 	}
 
-	if (vcpu->arch.fp_state == FP_STATE_GUEST_OWNED) {
+	if (*host_data_ptr(fp_owner) == FP_STATE_GUEST_OWNED) {
 		if (vcpu_has_sve(vcpu)) {
 			__vcpu_sys_reg(vcpu, ZCR_EL1) = read_sysreg_el1(SYS_ZCR);
 

arch/arm64/kvm/hyp/include/hyp/debug-sr.h

@@ -135,9 +135,9 @@ static inline void __debug_switch_to_guest_common(struct kvm_vcpu *vcpu)
 	if (!vcpu_get_flag(vcpu, DEBUG_DIRTY))
 		return;
 
-	host_ctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
+	host_ctxt = host_data_ptr(host_ctxt);
 	guest_ctxt = &vcpu->arch.ctxt;
-	host_dbg = &vcpu->arch.host_debug_state.regs;
+	host_dbg = host_data_ptr(host_debug_state.regs);
 	guest_dbg = kern_hyp_va(vcpu->arch.debug_ptr);
 
 	__debug_save_state(host_dbg, host_ctxt);
@@ -154,9 +154,9 @@ static inline void __debug_switch_to_host_common(struct kvm_vcpu *vcpu)
 	if (!vcpu_get_flag(vcpu, DEBUG_DIRTY))
 		return;
 
-	host_ctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
+	host_ctxt = host_data_ptr(host_ctxt);
 	guest_ctxt = &vcpu->arch.ctxt;
-	host_dbg = &vcpu->arch.host_debug_state.regs;
+	host_dbg = host_data_ptr(host_debug_state.regs);
 	guest_dbg = kern_hyp_va(vcpu->arch.debug_ptr);
 
 	__debug_save_state(guest_dbg, guest_ctxt);

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

@@ -42,7 +42,7 @@ extern struct kvm_exception_table_entry __stop___kvm_ex_table;
 /* Check whether the FP regs are owned by the guest */
 static inline bool guest_owns_fp_regs(struct kvm_vcpu *vcpu)
 {
-	return vcpu->arch.fp_state == FP_STATE_GUEST_OWNED;
+	return *host_data_ptr(fp_owner) == FP_STATE_GUEST_OWNED;
 }
 
 /* Save the 32-bit only FPSIMD system register state */
@@ -155,7 +155,7 @@ static inline bool cpu_has_amu(void)
 
 static inline void __activate_traps_hfgxtr(struct kvm_vcpu *vcpu)
 {
-	struct kvm_cpu_context *hctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
+	struct kvm_cpu_context *hctxt = host_data_ptr(host_ctxt);
 	struct kvm *kvm = kern_hyp_va(vcpu->kvm);
 
 	CHECK_FGT_MASKS(HFGRTR_EL2);
@@ -191,7 +191,7 @@ static inline void __activate_traps_hfgxtr(struct kvm_vcpu *vcpu)
 
 static inline void __deactivate_traps_hfgxtr(struct kvm_vcpu *vcpu)
 {
-	struct kvm_cpu_context *hctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
+	struct kvm_cpu_context *hctxt = host_data_ptr(host_ctxt);
 	struct kvm *kvm = kern_hyp_va(vcpu->kvm);
 
 	if (!cpus_have_final_cap(ARM64_HAS_FGT))
@@ -226,13 +226,13 @@ static inline void __activate_traps_common(struct kvm_vcpu *vcpu)
 
 		write_sysreg(0, pmselr_el0);
 
-		hctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
+		hctxt = host_data_ptr(host_ctxt);
 		ctxt_sys_reg(hctxt, PMUSERENR_EL0) = read_sysreg(pmuserenr_el0);
 		write_sysreg(ARMV8_PMU_USERENR_MASK, pmuserenr_el0);
 		vcpu_set_flag(vcpu, PMUSERENR_ON_CPU);
 	}
 
-	vcpu->arch.mdcr_el2_host = read_sysreg(mdcr_el2);
+	*host_data_ptr(host_debug_state.mdcr_el2) = read_sysreg(mdcr_el2);
 	write_sysreg(vcpu->arch.mdcr_el2, mdcr_el2);
 
 	if (cpus_have_final_cap(ARM64_HAS_HCX)) {
@@ -254,13 +254,13 @@ static inline void __activate_traps_common(struct kvm_vcpu *vcpu)
 
 static inline void __deactivate_traps_common(struct kvm_vcpu *vcpu)
 {
-	write_sysreg(vcpu->arch.mdcr_el2_host, mdcr_el2);
+	write_sysreg(*host_data_ptr(host_debug_state.mdcr_el2), mdcr_el2);
 
 	write_sysreg(0, hstr_el2);
 	if (kvm_arm_support_pmu_v3()) {
 		struct kvm_cpu_context *hctxt;
 
-		hctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
+		hctxt = host_data_ptr(host_ctxt);
 		write_sysreg(ctxt_sys_reg(hctxt, PMUSERENR_EL0), pmuserenr_el0);
 		vcpu_clear_flag(vcpu, PMUSERENR_ON_CPU);
 	}
@@ -376,8 +376,8 @@ static bool kvm_hyp_handle_fpsimd(struct kvm_vcpu *vcpu, u64 *exit_code)
 	isb();
 
 	/* Write out the host state if it's in the registers */
-	if (vcpu->arch.fp_state == FP_STATE_HOST_OWNED)
-		__fpsimd_save_state(vcpu->arch.host_fpsimd_state);
+	if (*host_data_ptr(fp_owner) == FP_STATE_HOST_OWNED)
+		__fpsimd_save_state(*host_data_ptr(fpsimd_state));
 
 	/* Restore the guest state */
 	if (sve_guest)
@@ -389,7 +389,7 @@ static bool kvm_hyp_handle_fpsimd(struct kvm_vcpu *vcpu, u64 *exit_code)
 	if (!(read_sysreg(hcr_el2) & HCR_RW))
 		write_sysreg(__vcpu_sys_reg(vcpu, FPEXC32_EL2), fpexc32_el2);
 
-	vcpu->arch.fp_state = FP_STATE_GUEST_OWNED;
+	*host_data_ptr(fp_owner) = FP_STATE_GUEST_OWNED;
 
 	return true;
 }

arch/arm64/kvm/hyp/nvhe/debug-sr.c

@@ -83,10 +83,10 @@ void __debug_save_host_buffers_nvhe(struct kvm_vcpu *vcpu)
 {
 	/* Disable and flush SPE data generation */
 	if (vcpu_get_flag(vcpu, DEBUG_STATE_SAVE_SPE))
-		__debug_save_spe(&vcpu->arch.host_debug_state.pmscr_el1);
+		__debug_save_spe(host_data_ptr(host_debug_state.pmscr_el1));
 	/* Disable and flush Self-Hosted Trace generation */
 	if (vcpu_get_flag(vcpu, DEBUG_STATE_SAVE_TRBE))
-		__debug_save_trace(&vcpu->arch.host_debug_state.trfcr_el1);
+		__debug_save_trace(host_data_ptr(host_debug_state.trfcr_el1));
 }
 
 void __debug_switch_to_guest(struct kvm_vcpu *vcpu)
@@ -97,9 +97,9 @@ void __debug_switch_to_guest(struct kvm_vcpu *vcpu)
 void __debug_restore_host_buffers_nvhe(struct kvm_vcpu *vcpu)
 {
 	if (vcpu_get_flag(vcpu, DEBUG_STATE_SAVE_SPE))
-		__debug_restore_spe(vcpu->arch.host_debug_state.pmscr_el1);
+		__debug_restore_spe(*host_data_ptr(host_debug_state.pmscr_el1));
 	if (vcpu_get_flag(vcpu, DEBUG_STATE_SAVE_TRBE))
-		__debug_restore_trace(vcpu->arch.host_debug_state.trfcr_el1);
+		__debug_restore_trace(*host_data_ptr(host_debug_state.trfcr_el1));
 }
 
 void __debug_switch_to_host(struct kvm_vcpu *vcpu)

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

@@ -39,10 +39,8 @@ static void flush_hyp_vcpu(struct pkvm_hyp_vcpu *hyp_vcpu)
 	hyp_vcpu->vcpu.arch.cptr_el2	= host_vcpu->arch.cptr_el2;
 
 	hyp_vcpu->vcpu.arch.iflags	= host_vcpu->arch.iflags;
-	hyp_vcpu->vcpu.arch.fp_state	= host_vcpu->arch.fp_state;
 
 	hyp_vcpu->vcpu.arch.debug_ptr	= kern_hyp_va(host_vcpu->arch.debug_ptr);
-	hyp_vcpu->vcpu.arch.host_fpsimd_state = host_vcpu->arch.host_fpsimd_state;
 
 	hyp_vcpu->vcpu.arch.vsesr_el2	= host_vcpu->arch.vsesr_el2;
 
@@ -64,7 +62,6 @@ static void sync_hyp_vcpu(struct pkvm_hyp_vcpu *hyp_vcpu)
 	host_vcpu->arch.fault		= hyp_vcpu->vcpu.arch.fault;
 
 	host_vcpu->arch.iflags		= hyp_vcpu->vcpu.arch.iflags;
-	host_vcpu->arch.fp_state	= hyp_vcpu->vcpu.arch.fp_state;
 
 	host_cpu_if->vgic_hcr		= hyp_cpu_if->vgic_hcr;
 	for (i = 0; i < hyp_cpu_if->used_lrs; ++i)

arch/arm64/kvm/hyp/nvhe/psci-relay.c

@@ -205,7 +205,7 @@ asmlinkage void __noreturn __kvm_host_psci_cpu_entry(bool is_cpu_on)
 	struct psci_boot_args *boot_args;
 	struct kvm_cpu_context *host_ctxt;
 
-	host_ctxt = &this_cpu_ptr(&kvm_host_data)->host_ctxt;
+	host_ctxt = host_data_ptr(host_ctxt);
 
 	if (is_cpu_on)
 		boot_args = this_cpu_ptr(&cpu_on_args);

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

@@ -257,8 +257,7 @@ static int fix_hyp_pgtable_refcnt(void)
 
 void __noreturn __pkvm_init_finalise(void)
 {
-	struct kvm_host_data *host_data = this_cpu_ptr(&kvm_host_data);
-	struct kvm_cpu_context *host_ctxt = &host_data->host_ctxt;
+	struct kvm_cpu_context *host_ctxt = host_data_ptr(host_ctxt);
 	unsigned long nr_pages, reserved_pages, pfn;
 	int ret;