Merge branch kvm-arm64/debug-6.14 into kvmarm-master/next

* kvm-arm64/debug-6.14:
: .
: Large rework of the debug code to make it a bit less horrid,
: courtesy of Oliver Upton. From the original cover letter:
:
: "The debug code has become a bit difficult to reason about, especially
: all the hacks and bandaids for state tracking + trap configuration.
:
: This series reworks the entire mess around using a single enumeration to
: track the state of the debug registers (free, guest-owned, host-owned),
: using that to drive trap configuration and save/restore.
:
: On top of that, this series wires most of the implementation into vCPU
: load/put rather than the main KVM_RUN loop. This has been a long time
: coming for VHE, as a lot of the trap configuration and EL1 state gets
: loaded into hardware at that point anyway.
:
: The save/restore of the debug registers is simplified quite a bit as
: well. KVM will now restore the registers for *any* access rather than
: just writes, and keep doing so until the next vcpu_put() instead of
: dropping it on the floor after the next exception."
: .
KVM: arm64: Promote guest ownership for DBGxVR/DBGxCR reads
KVM: arm64: Fold DBGxVR/DBGxCR accessors into common set
KVM: arm64: Avoid reading ID_AA64DFR0_EL1 for debug save/restore
KVM: arm64: nv: Honor MDCR_EL2.TDE routing for debug exceptions
KVM: arm64: Manage software step state at load/put
KVM: arm64: Don't hijack guest context MDSCR_EL1
KVM: arm64: Compute MDCR_EL2 at vcpu_load()
KVM: arm64: Reload vCPU for accesses to OSLAR_EL1
KVM: arm64: Use debug_owner to track if debug regs need save/restore
KVM: arm64: Remove vestiges of debug_ptr
KVM: arm64: Remove debug tracepoints
KVM: arm64: Select debug state to save/restore based on debug owner
KVM: arm64: Clean up KVM_SET_GUEST_DEBUG handler
KVM: arm64: Evaluate debug owner at vcpu_load()
KVM: arm64: Write MDCR_EL2 directly from kvm_arm_setup_mdcr_el2()
KVM: arm64: Move host SME/SVE tracking flags to host data
KVM: arm64: Track presence of SPE/TRBE in kvm_host_data instead of vCPU
KVM: arm64: Get rid of __kvm_get_mdcr_el2() and related warts
KVM: arm64: Drop MDSCR_EL1_DEBUG_MASK

Signed-off-by: Marc Zyngier <[e-mail]>

arch/arm64/include/asm/kvm_asm.h

···
 enum __kvm_host_smccc_func {
 	/* Hypercalls available only prior to pKVM finalisation */
 	/* __KVM_HOST_SMCCC_FUNC___kvm_hyp_init */
-	__KVM_HOST_SMCCC_FUNC___kvm_get_mdcr_el2 = __KVM_HOST_SMCCC_FUNC___kvm_hyp_init + 1,
-	__KVM_HOST_SMCCC_FUNC___pkvm_init,
+	__KVM_HOST_SMCCC_FUNC___pkvm_init = __KVM_HOST_SMCCC_FUNC___kvm_hyp_init + 1,
 	__KVM_HOST_SMCCC_FUNC___pkvm_create_private_mapping,
 	__KVM_HOST_SMCCC_FUNC___pkvm_cpu_set_vector,
 	__KVM_HOST_SMCCC_FUNC___kvm_enable_ssbs,
···
 
 extern u64 __vgic_v3_get_gic_config(void);
 extern void __vgic_v3_init_lrs(void);
-
-extern u64 __kvm_get_mdcr_el2(void);
 
 #define __KVM_EXTABLE(from, to)						\
 	"	.pushsection	__kvm_ex_table, \"a\"\n"		\

arch/arm64/include/asm/kvm_host.h

···
  * field.
  */
 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
+	unsigned long flags;
+
 	struct kvm_cpu_context host_ctxt;
 
 	/*
···
 	 * host_debug_state contains the host registers which are
 	 * saved and restored during world switches.
 	 */
-	 struct {
+	struct {
 		/* {Break,watch}point registers */
 		struct kvm_guest_debug_arch regs;
 		/* Statistical profiling extension */
···
 		/* Values of trap registers for the host before guest entry. */
 		u64 mdcr_el2;
 	} host_debug_state;
+
+	/* Number of programmable event counters (PMCR_EL0.N) for this CPU */
+	unsigned int nr_event_counters;
+
+	/* Number of debug breakpoints/watchpoints for this CPU (minus 1) */
+	unsigned int debug_brps;
+	unsigned int debug_wrps;
 };
 
 struct kvm_host_psci_config {
···
 	 *
 	 * 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.
 	 */
-	struct kvm_guest_debug_arch *debug_ptr;
 	struct kvm_guest_debug_arch vcpu_debug_state;
 	struct kvm_guest_debug_arch external_debug_state;
+	u64 external_mdscr_el1;
+
+	enum {
+		VCPU_DEBUG_FREE,
+		VCPU_DEBUG_HOST_OWNED,
+		VCPU_DEBUG_GUEST_OWNED,
+	} debug_owner;
 
 	/* VGIC state */
 	struct vgic_cpu vgic_cpu;
 	struct arch_timer_cpu timer_cpu;
 	struct kvm_pmu pmu;
-
-	/*
-	 * Guest registers we preserve during guest debugging.
-	 *
-	 * These shadow registers are updated by the kvm_handle_sys_reg
-	 * trap handler if the guest accesses or updates them while we
-	 * are using guest debug.
-	 */
-	struct {
-		u32	mdscr_el1;
-		bool	pstate_ss;
-	} guest_debug_preserved;
 
 	/* vcpu power state */
 	struct kvm_mp_state mp_state;
···
 #define EXCEPT_AA64_EL2_IRQ	__vcpu_except_flags(5)
 #define EXCEPT_AA64_EL2_FIQ	__vcpu_except_flags(6)
 #define EXCEPT_AA64_EL2_SERR	__vcpu_except_flags(7)
-/* Guest debug is live */
-#define DEBUG_DIRTY		__vcpu_single_flag(iflags, BIT(4))
-/* Save SPE context if active  */
-#define DEBUG_STATE_SAVE_SPE	__vcpu_single_flag(iflags, BIT(5))
-/* Save TRBE context if active  */
-#define DEBUG_STATE_SAVE_TRBE	__vcpu_single_flag(iflags, BIT(6))
 
-/* SVE enabled for host EL0 */
-#define HOST_SVE_ENABLED	__vcpu_single_flag(sflags, BIT(0))
-/* SME enabled for EL0 */
-#define HOST_SME_ENABLED	__vcpu_single_flag(sflags, BIT(1))
 /* Physical CPU not in supported_cpus */
-#define ON_UNSUPPORTED_CPU	__vcpu_single_flag(sflags, BIT(2))
+#define ON_UNSUPPORTED_CPU	__vcpu_single_flag(sflags, BIT(0))
 /* WFIT instruction trapped */
-#define IN_WFIT			__vcpu_single_flag(sflags, BIT(3))
+#define IN_WFIT			__vcpu_single_flag(sflags, BIT(1))
 /* vcpu system registers loaded on physical CPU */
-#define SYSREGS_ON_CPU		__vcpu_single_flag(sflags, BIT(4))
-/* Software step state is Active-pending */
-#define DBG_SS_ACTIVE_PENDING	__vcpu_single_flag(sflags, BIT(5))
+#define SYSREGS_ON_CPU		__vcpu_single_flag(sflags, BIT(2))
+/* Software step state is Active-pending for external debug */
+#define HOST_SS_ACTIVE_PENDING	__vcpu_single_flag(sflags, BIT(3))
+/* Software step state is Active pending for guest debug */
+#define GUEST_SS_ACTIVE_PENDING __vcpu_single_flag(sflags, BIT(4))
 /* PMUSERENR for the guest EL0 is on physical CPU */
-#define PMUSERENR_ON_CPU	__vcpu_single_flag(sflags, BIT(6))
+#define PMUSERENR_ON_CPU	__vcpu_single_flag(sflags, BIT(5))
 /* WFI instruction trapped */
-#define IN_WFI			__vcpu_single_flag(sflags, BIT(7))
+#define IN_WFI			__vcpu_single_flag(sflags, BIT(6))
 
 
 /* Pointer to the vcpu's SVE FFR for sve_{save,load}_state() */
···
 	 &this_cpu_ptr_hyp_sym(kvm_host_data)->f)
 #endif
 
+#define host_data_test_flag(flag)					\
+	(test_bit(KVM_HOST_DATA_FLAG_##flag, host_data_ptr(flags)))
+#define host_data_set_flag(flag)					\
+	set_bit(KVM_HOST_DATA_FLAG_##flag, host_data_ptr(flags))
+#define host_data_clear_flag(flag)					\
+	clear_bit(KVM_HOST_DATA_FLAG_##flag, host_data_ptr(flags))
+
 /* Check whether the FP regs are owned by the guest */
 static inline bool guest_owns_fp_regs(void)
 {
···
 
 static inline void kvm_arch_sync_events(struct kvm *kvm) {}
 
-void kvm_arm_init_debug(void);
-void kvm_arm_vcpu_init_debug(struct kvm_vcpu *vcpu);
-void kvm_arm_setup_debug(struct kvm_vcpu *vcpu);
-void kvm_arm_clear_debug(struct kvm_vcpu *vcpu);
-void kvm_arm_reset_debug_ptr(struct kvm_vcpu *vcpu);
+void kvm_init_host_debug_data(void);
+void kvm_vcpu_load_debug(struct kvm_vcpu *vcpu);
+void kvm_vcpu_put_debug(struct kvm_vcpu *vcpu);
+void kvm_debug_set_guest_ownership(struct kvm_vcpu *vcpu);
+void kvm_debug_handle_oslar(struct kvm_vcpu *vcpu, u64 val);
 
 #define kvm_vcpu_os_lock_enabled(vcpu)		\
 	(!!(__vcpu_sys_reg(vcpu, OSLSR_EL1) & OSLSR_EL1_OSLK))
+
+#define kvm_debug_regs_in_use(vcpu)		\
+	((vcpu)->arch.debug_owner != VCPU_DEBUG_FREE)
+#define kvm_host_owns_debug_regs(vcpu)		\
+	((vcpu)->arch.debug_owner == VCPU_DEBUG_HOST_OWNED)
+#define kvm_guest_owns_debug_regs(vcpu)		\
+	((vcpu)->arch.debug_owner == VCPU_DEBUG_GUEST_OWNED)
 
 int kvm_arm_vcpu_arch_set_attr(struct kvm_vcpu *vcpu,
 			       struct kvm_device_attr *attr);
···
 {
 	return (!has_vhe() && attr->exclude_host);
 }
-
-/* Flags for host debug state */
-void kvm_arch_vcpu_load_debug_state_flags(struct kvm_vcpu *vcpu);
-void kvm_arch_vcpu_put_debug_state_flags(struct kvm_vcpu *vcpu);
 
 #ifdef CONFIG_KVM
 void kvm_set_pmu_events(u64 set, struct perf_event_attr *attr);

arch/arm64/include/asm/kvm_nested.h

···
 }
 
 extern bool forward_smc_trap(struct kvm_vcpu *vcpu);
+extern bool forward_debug_exception(struct kvm_vcpu *vcpu);
 extern void kvm_init_nested(struct kvm *kvm);
 extern int kvm_vcpu_init_nested(struct kvm_vcpu *vcpu);
 extern void kvm_init_nested_s2_mmu(struct kvm_s2_mmu *mmu);

arch/arm64/kvm/arm.c

···
 
 	kvm_pmu_vcpu_init(vcpu);
 
-	kvm_arm_reset_debug_ptr(vcpu);
-
 	kvm_arm_pvtime_vcpu_init(&vcpu->arch);
 
 	vcpu->arch.hw_mmu = &vcpu->kvm->arch.mmu;
···
 
 	kvm_vgic_load(vcpu);
 	kvm_timer_vcpu_load(vcpu);
+	kvm_vcpu_load_debug(vcpu);
 	if (has_vhe())
 		kvm_vcpu_load_vhe(vcpu);
 	kvm_arch_vcpu_load_fp(vcpu);
···
 
 	vcpu_set_pauth_traps(vcpu);
 
-	kvm_arch_vcpu_load_debug_state_flags(vcpu);
-
 	if (!cpumask_test_cpu(cpu, vcpu->kvm->arch.supported_cpus))
 		vcpu_set_on_unsupported_cpu(vcpu);
 }
 
 void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu)
 {
-	kvm_arch_vcpu_put_debug_state_flags(vcpu);
+	kvm_vcpu_put_debug(vcpu);
 	kvm_arch_vcpu_put_fp(vcpu);
 	if (has_vhe())
 		kvm_vcpu_put_vhe(vcpu);
···
 		return 0;
 
 	kvm_init_mpidr_data(kvm);
-
-	kvm_arm_vcpu_init_debug(vcpu);
 
 	if (likely(irqchip_in_kernel(kvm))) {
 		/*
···
 			continue;
 		}
 
-		kvm_arm_setup_debug(vcpu);
 		kvm_arch_vcpu_ctxflush_fp(vcpu);
 
 		/**************************************************************
···
 		/*
 		 * Back from guest
 		 *************************************************************/
-
-		kvm_arm_clear_debug(vcpu);
 
 		/*
 		 * We must sync the PMU state before the vgic state so
···
 static void cpu_hyp_init_context(void)
 {
 	kvm_init_host_cpu_context(host_data_ptr(host_ctxt));
+	kvm_init_host_debug_data();
 
 	if (!is_kernel_in_hyp_mode())
 		cpu_init_hyp_mode();
···
 static void cpu_hyp_init_features(void)
 {
 	cpu_set_hyp_vector();
-	kvm_arm_init_debug();
 
 	if (is_kernel_in_hyp_mode())
 		kvm_timer_init_vhe();

arch/arm64/kvm/debug.c

···
  * Debug and Guest Debug support
  *
  * Copyright (C) 2015 - Linaro Ltd
- * Author: Alex Bennée <[e-mail]>
+ * Authors: Alex Bennée <[e-mail]>
+ * 	    Oliver Upton <[e-mail]>
  */
 
 #include <linux/kvm_host.h>
···
 #include <asm/kvm_asm.h>
 #include <asm/kvm_arm.h>
 #include <asm/kvm_emulate.h>
-
-#include "trace.h"
-
-/* These are the bits of MDSCR_EL1 we may manipulate */
-#define MDSCR_EL1_DEBUG_MASK	(DBG_MDSCR_SS | \
-				DBG_MDSCR_KDE | \
-				DBG_MDSCR_MDE)
-
-static DEFINE_PER_CPU(u64, mdcr_el2);
-
-/*
- * save/restore_guest_debug_regs
- *
- * For some debug operations we need to tweak some guest registers. As
- * a result we need to save the state of those registers before we
- * make those modifications.
- *
- * Guest access to MDSCR_EL1 is trapped by the hypervisor and handled
- * after we have restored the preserved value to the main context.
- *
- * When single-step is enabled by userspace, we tweak PSTATE.SS on every
- * guest entry. Preserve PSTATE.SS so we can restore the original value
- * for the vcpu after the single-step is disabled.
- */
-static void save_guest_debug_regs(struct kvm_vcpu *vcpu)
-{
-	u64 val = vcpu_read_sys_reg(vcpu, MDSCR_EL1);
-
-	vcpu->arch.guest_debug_preserved.mdscr_el1 = val;
-
-	trace_kvm_arm_set_dreg32("Saved MDSCR_EL1",
-				vcpu->arch.guest_debug_preserved.mdscr_el1);
-
-	vcpu->arch.guest_debug_preserved.pstate_ss =
-					(*vcpu_cpsr(vcpu) & DBG_SPSR_SS);
-}
-
-static void restore_guest_debug_regs(struct kvm_vcpu *vcpu)
-{
-	u64 val = vcpu->arch.guest_debug_preserved.mdscr_el1;
-
-	vcpu_write_sys_reg(vcpu, val, MDSCR_EL1);
-
-	trace_kvm_arm_set_dreg32("Restored MDSCR_EL1",
-				vcpu_read_sys_reg(vcpu, MDSCR_EL1));
-
-	if (vcpu->arch.guest_debug_preserved.pstate_ss)
-		*vcpu_cpsr(vcpu) |= DBG_SPSR_SS;
-	else
-		*vcpu_cpsr(vcpu) &= ~DBG_SPSR_SS;
-}
-
-/**
- * kvm_arm_init_debug - grab what we need for debug
- *
- * Currently the sole task of this function is to retrieve the initial
- * value of mdcr_el2 so we can preserve MDCR_EL2.HPMN which has
- * presumably been set-up by some knowledgeable bootcode.
- *
- * It is called once per-cpu during CPU hyp initialisation.
- */
-
-void kvm_arm_init_debug(void)
-{
-	__this_cpu_write(mdcr_el2, kvm_call_hyp_ret(__kvm_get_mdcr_el2));
-}
 
 /**
  * kvm_arm_setup_mdcr_el2 - configure vcpu mdcr_el2 value
···
  */
 static void kvm_arm_setup_mdcr_el2(struct kvm_vcpu *vcpu)
 {
+	preempt_disable();
+
 	/*
 	 * This also clears MDCR_EL2_E2PB_MASK and MDCR_EL2_E2TB_MASK
 	 * to disable guest access to the profiling and trace buffers
 	 */
-	vcpu->arch.mdcr_el2 = __this_cpu_read(mdcr_el2) & MDCR_EL2_HPMN_MASK;
+	vcpu->arch.mdcr_el2 = FIELD_PREP(MDCR_EL2_HPMN,
+					 *host_data_ptr(nr_event_counters));
 	vcpu->arch.mdcr_el2 |= (MDCR_EL2_TPM |
 				MDCR_EL2_TPMS |
 				MDCR_EL2_TTRF |
···
 		vcpu->arch.mdcr_el2 |= MDCR_EL2_TDE;
 
 	/*
-	 * Trap debug register access when one of the following is true:
-	 *  - Userspace is using the hardware to debug the guest
-	 *  (KVM_GUESTDBG_USE_HW is set).
-	 *  - The guest is not using debug (DEBUG_DIRTY clear).
-	 *  - The guest has enabled the OS Lock (debug exceptions are blocked).
+	 * Trap debug registers if the guest doesn't have ownership of them.
 	 */
-	if ((vcpu->guest_debug & KVM_GUESTDBG_USE_HW) ||
-	    !vcpu_get_flag(vcpu, DEBUG_DIRTY) ||
-	    kvm_vcpu_os_lock_enabled(vcpu))
+	if (!kvm_guest_owns_debug_regs(vcpu))
 		vcpu->arch.mdcr_el2 |= MDCR_EL2_TDA;
 
-	trace_kvm_arm_set_dreg32("MDCR_EL2", vcpu->arch.mdcr_el2);
-}
+	/* Write MDCR_EL2 directly if we're already at EL2 */
+	if (has_vhe())
+		write_sysreg(vcpu->arch.mdcr_el2, mdcr_el2);
 
-/**
- * kvm_arm_vcpu_init_debug - setup vcpu debug traps
- *
- * @vcpu:	the vcpu pointer
- *
- * Set vcpu initial mdcr_el2 value.
- */
-void kvm_arm_vcpu_init_debug(struct kvm_vcpu *vcpu)
-{
-	preempt_disable();
-	kvm_arm_setup_mdcr_el2(vcpu);
 	preempt_enable();
 }
 
-/**
- * kvm_arm_reset_debug_ptr - reset the debug ptr to point to the vcpu state
- * @vcpu:	the vcpu pointer
- */
-
-void kvm_arm_reset_debug_ptr(struct kvm_vcpu *vcpu)
+void kvm_init_host_debug_data(void)
 {
-	vcpu->arch.debug_ptr = &vcpu->arch.vcpu_debug_state;
-}
+	u64 dfr0 = read_sysreg(id_aa64dfr0_el1);
 
-/**
- * kvm_arm_setup_debug - set up debug related stuff
- *
- * @vcpu:	the vcpu pointer
- *
- * This is called before each entry into the hypervisor to setup any
- * debug related registers.
- *
- * Additionally, KVM only traps guest accesses to the debug registers if
- * the guest is not actively using them (see the DEBUG_DIRTY
- * flag on vcpu->arch.iflags).  Since the guest must not interfere
- * with the hardware state when debugging the guest, we must ensure that
- * trapping is enabled whenever we are debugging the guest using the
- * debug registers.
- */
+	if (cpuid_feature_extract_signed_field(dfr0, ID_AA64DFR0_EL1_PMUVer_SHIFT) > 0)
+		*host_data_ptr(nr_event_counters) = FIELD_GET(ARMV8_PMU_PMCR_N,
+							      read_sysreg(pmcr_el0));
 
-void kvm_arm_setup_debug(struct kvm_vcpu *vcpu)
-{
-	unsigned long mdscr, orig_mdcr_el2 = vcpu->arch.mdcr_el2;
+	*host_data_ptr(debug_brps) = SYS_FIELD_GET(ID_AA64DFR0_EL1, BRPs, dfr0);
+	*host_data_ptr(debug_wrps) = SYS_FIELD_GET(ID_AA64DFR0_EL1, WRPs, dfr0);
 
-	trace_kvm_arm_setup_debug(vcpu, vcpu->guest_debug);
-
-	kvm_arm_setup_mdcr_el2(vcpu);
-
-	/* Check if we need to use the debug registers. */
-	if (vcpu->guest_debug || kvm_vcpu_os_lock_enabled(vcpu)) {
-		/* Save guest debug state */
-		save_guest_debug_regs(vcpu);
-
-		/*
-		 * Single Step (ARM ARM D2.12.3 The software step state
-		 * machine)
-		 *
-		 * If we are doing Single Step we need to manipulate
-		 * the guest's MDSCR_EL1.SS and PSTATE.SS. Once the
-		 * step has occurred the hypervisor will trap the
-		 * debug exception and we return to userspace.
-		 *
-		 * If the guest attempts to single step its userspace
-		 * we would have to deal with a trapped exception
-		 * while in the guest kernel. Because this would be
-		 * hard to unwind we suppress the guest's ability to
-		 * do so by masking MDSCR_EL.SS.
-		 *
-		 * This confuses guest debuggers which use
-		 * single-step behind the scenes but everything
-		 * returns to normal once the host is no longer
-		 * debugging the system.
-		 */
-		if (vcpu->guest_debug & KVM_GUESTDBG_SINGLESTEP) {
-			/*
-			 * If the software step state at the last guest exit
-			 * was Active-pending, we don't set DBG_SPSR_SS so
-			 * that the state is maintained (to not run another
-			 * single-step until the pending Software Step
-			 * exception is taken).
-			 */
-			if (!vcpu_get_flag(vcpu, DBG_SS_ACTIVE_PENDING))
-				*vcpu_cpsr(vcpu) |= DBG_SPSR_SS;
-			else
-				*vcpu_cpsr(vcpu) &= ~DBG_SPSR_SS;
-
-			mdscr = vcpu_read_sys_reg(vcpu, MDSCR_EL1);
-			mdscr |= DBG_MDSCR_SS;
-			vcpu_write_sys_reg(vcpu, mdscr, MDSCR_EL1);
-		} else {
-			mdscr = vcpu_read_sys_reg(vcpu, MDSCR_EL1);
-			mdscr &= ~DBG_MDSCR_SS;
-			vcpu_write_sys_reg(vcpu, mdscr, MDSCR_EL1);
-		}
-
-		trace_kvm_arm_set_dreg32("SPSR_EL2", *vcpu_cpsr(vcpu));
-
-		/*
-		 * HW Breakpoints and watchpoints
-		 *
-		 * We simply switch the debug_ptr to point to our new
-		 * external_debug_state which has been populated by the
-		 * debug ioctl. The existing DEBUG_DIRTY mechanism ensures
-		 * the registers are updated on the world switch.
-		 */
-		if (vcpu->guest_debug & KVM_GUESTDBG_USE_HW) {
-			/* Enable breakpoints/watchpoints */
-			mdscr = vcpu_read_sys_reg(vcpu, MDSCR_EL1);
-			mdscr |= DBG_MDSCR_MDE;
-			vcpu_write_sys_reg(vcpu, mdscr, MDSCR_EL1);
-
-			vcpu->arch.debug_ptr = &vcpu->arch.external_debug_state;
-			vcpu_set_flag(vcpu, DEBUG_DIRTY);
-
-			trace_kvm_arm_set_regset("BKPTS", get_num_brps(),
-						&vcpu->arch.debug_ptr->dbg_bcr[0],
-						&vcpu->arch.debug_ptr->dbg_bvr[0]);
-
-			trace_kvm_arm_set_regset("WAPTS", get_num_wrps(),
-						&vcpu->arch.debug_ptr->dbg_wcr[0],
-						&vcpu->arch.debug_ptr->dbg_wvr[0]);
-
-		/*
-		 * The OS Lock blocks debug exceptions in all ELs when it is
-		 * enabled. If the guest has enabled the OS Lock, constrain its
-		 * effects to the guest. Emulate the behavior by clearing
-		 * MDSCR_EL1.MDE. In so doing, we ensure that host debug
-		 * exceptions are unaffected by guest configuration of the OS
-		 * Lock.
-		 */
-		} else if (kvm_vcpu_os_lock_enabled(vcpu)) {
-			mdscr = vcpu_read_sys_reg(vcpu, MDSCR_EL1);
-			mdscr &= ~DBG_MDSCR_MDE;
-			vcpu_write_sys_reg(vcpu, mdscr, MDSCR_EL1);
-		}
-	}
-
-	BUG_ON(!vcpu->guest_debug &&
-		vcpu->arch.debug_ptr != &vcpu->arch.vcpu_debug_state);
-
-	/* If KDE or MDE are set, perform a full save/restore cycle. */
-	if (vcpu_read_sys_reg(vcpu, MDSCR_EL1) & (DBG_MDSCR_KDE | DBG_MDSCR_MDE))
-		vcpu_set_flag(vcpu, DEBUG_DIRTY);
-
-	/* Write mdcr_el2 changes since vcpu_load on VHE systems */
-	if (has_vhe() && orig_mdcr_el2 != vcpu->arch.mdcr_el2)
-		write_sysreg(vcpu->arch.mdcr_el2, mdcr_el2);
-
-	trace_kvm_arm_set_dreg32("MDSCR_EL1", vcpu_read_sys_reg(vcpu, MDSCR_EL1));
-}
-
-void kvm_arm_clear_debug(struct kvm_vcpu *vcpu)
-{
-	trace_kvm_arm_clear_debug(vcpu->guest_debug);
-
-	/*
-	 * Restore the guest's debug registers if we were using them.
-	 */
-	if (vcpu->guest_debug || kvm_vcpu_os_lock_enabled(vcpu)) {
-		if (vcpu->guest_debug & KVM_GUESTDBG_SINGLESTEP) {
-			if (!(*vcpu_cpsr(vcpu) & DBG_SPSR_SS))
-				/*
-				 * Mark the vcpu as ACTIVE_PENDING
-				 * until Software Step exception is taken.
-				 */
-				vcpu_set_flag(vcpu, DBG_SS_ACTIVE_PENDING);
-		}
-
-		restore_guest_debug_regs(vcpu);
-
-		/*
-		 * If we were using HW debug we need to restore the
-		 * debug_ptr to the guest debug state.
-		 */
-		if (vcpu->guest_debug & KVM_GUESTDBG_USE_HW) {
-			kvm_arm_reset_debug_ptr(vcpu);
-
-			trace_kvm_arm_set_regset("BKPTS", get_num_brps(),
-						&vcpu->arch.debug_ptr->dbg_bcr[0],
-						&vcpu->arch.debug_ptr->dbg_bvr[0]);
-
-			trace_kvm_arm_set_regset("WAPTS", get_num_wrps(),
-						&vcpu->arch.debug_ptr->dbg_wcr[0],
-						&vcpu->arch.debug_ptr->dbg_wvr[0]);
-		}
-	}
-}
-
-void kvm_arch_vcpu_load_debug_state_flags(struct kvm_vcpu *vcpu)
-{
-	u64 dfr0;
-
-	/* For VHE, there is nothing to do */
 	if (has_vhe())
 		return;
 
-	dfr0 = read_sysreg(id_aa64dfr0_el1);
-	/*
-	 * If SPE is present on this CPU and is available at current EL,
-	 * we may need to check if the host state needs to be saved.
-	 */
 	if (cpuid_feature_extract_unsigned_field(dfr0, ID_AA64DFR0_EL1_PMSVer_SHIFT) &&
-	    !(read_sysreg_s(SYS_PMBIDR_EL1) & BIT(PMBIDR_EL1_P_SHIFT)))
-		vcpu_set_flag(vcpu, DEBUG_STATE_SAVE_SPE);
+	    !(read_sysreg_s(SYS_PMBIDR_EL1) & PMBIDR_EL1_P))
+		host_data_set_flag(HAS_SPE);
 
-	/* Check if we have TRBE implemented and available at the host */
 	if (cpuid_feature_extract_unsigned_field(dfr0, ID_AA64DFR0_EL1_TraceBuffer_SHIFT) &&
 	    !(read_sysreg_s(SYS_TRBIDR_EL1) & TRBIDR_EL1_P))
-		vcpu_set_flag(vcpu, DEBUG_STATE_SAVE_TRBE);
+		host_data_set_flag(HAS_TRBE);
 }
 
-void kvm_arch_vcpu_put_debug_state_flags(struct kvm_vcpu *vcpu)
+/*
+ * Configures the 'external' MDSCR_EL1 value for the guest, i.e. when the host
+ * has taken over MDSCR_EL1.
+ *
+ *  - Userspace is single-stepping the guest, and MDSCR_EL1.SS is forced to 1.
+ *
+ *  - Userspace is using the breakpoint/watchpoint registers to debug the
+ *    guest, and MDSCR_EL1.MDE is forced to 1.
+ *
+ *  - The guest has enabled the OS Lock, and KVM is forcing MDSCR_EL1.MDE to 0,
+ *    masking all debug exceptions affected by the OS Lock.
+ */
+static void setup_external_mdscr(struct kvm_vcpu *vcpu)
 {
-	vcpu_clear_flag(vcpu, DEBUG_STATE_SAVE_SPE);
-	vcpu_clear_flag(vcpu, DEBUG_STATE_SAVE_TRBE);
+	/*
+	 * Use the guest's MDSCR_EL1 as a starting point, since there are
+	 * several other features controlled by MDSCR_EL1 that are not relevant
+	 * to the host.
+	 *
+	 * Clear the bits that KVM may use which also satisfies emulation of
+	 * the OS Lock as MDSCR_EL1.MDE is cleared.
+	 */
+	u64 mdscr = vcpu_read_sys_reg(vcpu, MDSCR_EL1) & ~(MDSCR_EL1_SS |
+							   MDSCR_EL1_MDE |
+							   MDSCR_EL1_KDE);
+
+	if (vcpu->guest_debug & KVM_GUESTDBG_SINGLESTEP)
+		mdscr |= MDSCR_EL1_SS;
+
+	if (vcpu->guest_debug & KVM_GUESTDBG_USE_HW)
+		mdscr |= MDSCR_EL1_MDE | MDSCR_EL1_KDE;
+
+	vcpu->arch.external_mdscr_el1 = mdscr;
+}
+
+void kvm_vcpu_load_debug(struct kvm_vcpu *vcpu)
+{
+	u64 mdscr;
+
+	/* Must be called before kvm_vcpu_load_vhe() */
+	KVM_BUG_ON(vcpu_get_flag(vcpu, SYSREGS_ON_CPU), vcpu->kvm);
+
+	/*
+	 * Determine which of the possible debug states we're in:
+	 *
+	 *  - VCPU_DEBUG_HOST_OWNED: KVM has taken ownership of the guest's
+	 *    breakpoint/watchpoint registers, or needs to use MDSCR_EL1 to do
+	 *    software step or emulate the effects of the OS Lock being enabled.
+	 *
+	 *  - VCPU_DEBUG_GUEST_OWNED: The guest has debug exceptions enabled, and
+	 *    the breakpoint/watchpoint registers need to be loaded eagerly.
+	 *
+	 *  - VCPU_DEBUG_FREE: Neither of the above apply, no breakpoint/watchpoint
+	 *    context needs to be loaded on the CPU.
+	 */
+	if (vcpu->guest_debug || kvm_vcpu_os_lock_enabled(vcpu)) {
+		vcpu->arch.debug_owner = VCPU_DEBUG_HOST_OWNED;
+		setup_external_mdscr(vcpu);
+
+		/*
+		 * Steal the guest's single-step state machine if userspace wants
+		 * single-step the guest.
+		 */
+		if (vcpu->guest_debug & KVM_GUESTDBG_SINGLESTEP) {
+			if (*vcpu_cpsr(vcpu) & DBG_SPSR_SS)
+				vcpu_clear_flag(vcpu, GUEST_SS_ACTIVE_PENDING);
+			else
+				vcpu_set_flag(vcpu, GUEST_SS_ACTIVE_PENDING);
+
+			if (!vcpu_get_flag(vcpu, HOST_SS_ACTIVE_PENDING))
+				*vcpu_cpsr(vcpu) |= DBG_SPSR_SS;
+			else
+				*vcpu_cpsr(vcpu) &= ~DBG_SPSR_SS;
+		}
+	} else {
+		mdscr = vcpu_read_sys_reg(vcpu, MDSCR_EL1);
+
+		if (mdscr & (MDSCR_EL1_KDE | MDSCR_EL1_MDE))
+			vcpu->arch.debug_owner = VCPU_DEBUG_GUEST_OWNED;
+		else
+			vcpu->arch.debug_owner = VCPU_DEBUG_FREE;
+	}
+
+	kvm_arm_setup_mdcr_el2(vcpu);
+}
+
+void kvm_vcpu_put_debug(struct kvm_vcpu *vcpu)
+{
+	if (likely(!(vcpu->guest_debug & KVM_GUESTDBG_SINGLESTEP)))
+		return;
+
+	/*
+	 * Save the host's software step state and restore the guest's before
+	 * potentially returning to userspace.
+	 */
+	if (!(*vcpu_cpsr(vcpu) & DBG_SPSR_SS))
+		vcpu_set_flag(vcpu, HOST_SS_ACTIVE_PENDING);
+	else
+		vcpu_clear_flag(vcpu, HOST_SS_ACTIVE_PENDING);
+
+	if (vcpu_get_flag(vcpu, GUEST_SS_ACTIVE_PENDING))
+		*vcpu_cpsr(vcpu) &= ~DBG_SPSR_SS;
+	else
+		*vcpu_cpsr(vcpu) |= DBG_SPSR_SS;
+}
+
+/*
+ * Updates ownership of the debug registers after a trapped guest access to a
+ * breakpoint/watchpoint register. Host ownership of the debug registers is of
+ * strictly higher priority, and it is the responsibility of the VMM to emulate
+ * guest debug exceptions in this configuration.
+ */
+void kvm_debug_set_guest_ownership(struct kvm_vcpu *vcpu)
+{
+	if (kvm_host_owns_debug_regs(vcpu))
+		return;
+
+	vcpu->arch.debug_owner = VCPU_DEBUG_GUEST_OWNED;
+	kvm_arm_setup_mdcr_el2(vcpu);
+}
+
+void kvm_debug_handle_oslar(struct kvm_vcpu *vcpu, u64 val)
+{
+	if (val & OSLAR_EL1_OSLK)
+		__vcpu_sys_reg(vcpu, OSLSR_EL1) |= OSLSR_EL1_OSLK;
+	else
+		__vcpu_sys_reg(vcpu, OSLSR_EL1) &= ~OSLSR_EL1_OSLK;
+
+	preempt_disable();
+	kvm_arch_vcpu_put(vcpu);
+	kvm_arch_vcpu_load(vcpu, smp_processor_id());
+	preempt_enable();
 }

arch/arm64/kvm/emulate-nested.c

···
 	return true;
 }
 
-static bool forward_traps(struct kvm_vcpu *vcpu, u64 control_bit)
+static bool __forward_traps(struct kvm_vcpu *vcpu, unsigned int reg, u64 control_bit)
 {
 	bool control_bit_set;
 
 	if (!vcpu_has_nv(vcpu))
 		return false;
 
-	control_bit_set = __vcpu_sys_reg(vcpu, HCR_EL2) & control_bit;
+	control_bit_set = __vcpu_sys_reg(vcpu, reg) & control_bit;
 	if (!is_hyp_ctxt(vcpu) && control_bit_set) {
 		kvm_inject_nested_sync(vcpu, kvm_vcpu_get_esr(vcpu));
 		return true;
···
 	return false;
 }
 
+static bool forward_hcr_traps(struct kvm_vcpu *vcpu, u64 control_bit)
+{
+	return __forward_traps(vcpu, HCR_EL2, control_bit);
+}
+
 bool forward_smc_trap(struct kvm_vcpu *vcpu)
 {
-	return forward_traps(vcpu, HCR_TSC);
+	return forward_hcr_traps(vcpu, HCR_TSC);
+}
+
+static bool forward_mdcr_traps(struct kvm_vcpu *vcpu, u64 control_bit)
+{
+	return __forward_traps(vcpu, MDCR_EL2, control_bit);
+}
+
+bool forward_debug_exception(struct kvm_vcpu *vcpu)
+{
+	return forward_mdcr_traps(vcpu, MDCR_EL2_TDE);
 }
 
 static u64 kvm_check_illegal_exception_return(struct kvm_vcpu *vcpu, u64 spsr)
···
 	 * Forward this trap to the virtual EL2 if the virtual
 	 * HCR_EL2.NV bit is set and this is coming from !EL2.
 	 */
-	if (forward_traps(vcpu, HCR_NV))
+	if (forward_hcr_traps(vcpu, HCR_NV))
 		return;
 
 	spsr = vcpu_read_sys_reg(vcpu, SPSR_EL2);

arch/arm64/kvm/fpsimd.c

···
 	*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);
 
-	vcpu_clear_flag(vcpu, HOST_SVE_ENABLED);
+	host_data_clear_flag(HOST_SVE_ENABLED);
 	if (read_sysreg(cpacr_el1) & CPACR_EL1_ZEN_EL0EN)
-		vcpu_set_flag(vcpu, HOST_SVE_ENABLED);
+		host_data_set_flag(HOST_SVE_ENABLED);
 
 	if (system_supports_sme()) {
-		vcpu_clear_flag(vcpu, HOST_SME_ENABLED);
+		host_data_clear_flag(HOST_SME_ENABLED);
 		if (read_sysreg(cpacr_el1) & CPACR_EL1_SMEN_EL0EN)
-			vcpu_set_flag(vcpu, HOST_SME_ENABLED);
+			host_data_set_flag(HOST_SME_ENABLED);
 
 		/*
 		 * If PSTATE.SM is enabled then save any pending FP
···
 	 */
 	if (has_vhe() && system_supports_sme()) {
 		/* Also restore EL0 state seen on entry */
-		if (vcpu_get_flag(vcpu, HOST_SME_ENABLED))
+		if (host_data_test_flag(HOST_SME_ENABLED))
 			sysreg_clear_set(CPACR_EL1, 0, CPACR_ELx_SMEN);
 		else
 			sysreg_clear_set(CPACR_EL1,
···
 		 * for EL0.  To avoid spurious traps, restore the trap state
 		 * seen by kvm_arch_vcpu_load_fp():
 		 */
-		if (vcpu_get_flag(vcpu, HOST_SVE_ENABLED))
+		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);

arch/arm64/kvm/guest.c

···
 int kvm_arch_vcpu_ioctl_set_guest_debug(struct kvm_vcpu *vcpu,
 					struct kvm_guest_debug *dbg)
 {
-	int ret = 0;
-
 	trace_kvm_set_guest_debug(vcpu, dbg->control);
 
-	if (dbg->control & ~KVM_GUESTDBG_VALID_MASK) {
-		ret = -EINVAL;
-		goto out;
-	}
+	if (dbg->control & ~KVM_GUESTDBG_VALID_MASK)
+		return -EINVAL;
 
-	if (dbg->control & KVM_GUESTDBG_ENABLE) {
-		vcpu->guest_debug = dbg->control;
-
-		/* Hardware assisted Break and Watch points */
-		if (vcpu->guest_debug & KVM_GUESTDBG_USE_HW) {
-			vcpu->arch.external_debug_state = dbg->arch;
-		}
-
-	} else {
-		/* If not enabled clear all flags */
+	if (!(dbg->control & KVM_GUESTDBG_ENABLE)) {
 		vcpu->guest_debug = 0;
-		vcpu_clear_flag(vcpu, DBG_SS_ACTIVE_PENDING);
+		vcpu_clear_flag(vcpu, HOST_SS_ACTIVE_PENDING);
+		return 0;
 	}
 
-out:
-	return ret;
+	vcpu->guest_debug = dbg->control;
+
+	/* Hardware assisted Break and Watch points */
+	if (vcpu->guest_debug & KVM_GUESTDBG_USE_HW)
+		vcpu->arch.external_debug_state = dbg->arch;
+
+	return 0;
 }
 
 int kvm_arm_vcpu_arch_set_attr(struct kvm_vcpu *vcpu,

arch/arm64/kvm/handle_exit.c

···
 	struct kvm_run *run = vcpu->run;
 	u64 esr = kvm_vcpu_get_esr(vcpu);
 
+	if (!vcpu->guest_debug && forward_debug_exception(vcpu))
+		return 1;
+
 	run->exit_reason = KVM_EXIT_DEBUG;
 	run->debug.arch.hsr = lower_32_bits(esr);
 	run->debug.arch.hsr_high = upper_32_bits(esr);
···
 		run->debug.arch.far = vcpu->arch.fault.far_el2;
 		break;
 	case ESR_ELx_EC_SOFTSTP_LOW:
-		vcpu_clear_flag(vcpu, DBG_SS_ACTIVE_PENDING);
+		*vcpu_cpsr(vcpu) |= DBG_SPSR_SS;
 		break;
 	}
 

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

···
 	default:	write_debug(ptr[0], reg, 0);			\
 	}
 
+static struct kvm_guest_debug_arch *__vcpu_debug_regs(struct kvm_vcpu *vcpu)
+{
+	switch (vcpu->arch.debug_owner) {
+	case VCPU_DEBUG_FREE:
+		WARN_ON_ONCE(1);
+		fallthrough;
+	case VCPU_DEBUG_GUEST_OWNED:
+		return &vcpu->arch.vcpu_debug_state;
+	case VCPU_DEBUG_HOST_OWNED:
+		return &vcpu->arch.external_debug_state;
+	}
+
+	return NULL;
+}
+
 static void __debug_save_state(struct kvm_guest_debug_arch *dbg,
 			       struct kvm_cpu_context *ctxt)
 {
-	u64 aa64dfr0;
-	int brps, wrps;
-
-	aa64dfr0 = read_sysreg(id_aa64dfr0_el1);
-	brps = (aa64dfr0 >> 12) & 0xf;
-	wrps = (aa64dfr0 >> 20) & 0xf;
+	int brps = *host_data_ptr(debug_brps);
+	int wrps = *host_data_ptr(debug_wrps);
 
 	save_debug(dbg->dbg_bcr, dbgbcr, brps);
 	save_debug(dbg->dbg_bvr, dbgbvr, brps);
···
 static void __debug_restore_state(struct kvm_guest_debug_arch *dbg,
 				  struct kvm_cpu_context *ctxt)
 {
-	u64 aa64dfr0;
-	int brps, wrps;
-
-	aa64dfr0 = read_sysreg(id_aa64dfr0_el1);
-
-	brps = (aa64dfr0 >> 12) & 0xf;
-	wrps = (aa64dfr0 >> 20) & 0xf;
+	int brps = *host_data_ptr(debug_brps);
+	int wrps = *host_data_ptr(debug_wrps);
 
 	restore_debug(dbg->dbg_bcr, dbgbcr, brps);
 	restore_debug(dbg->dbg_bvr, dbgbvr, brps);
···
 	struct kvm_guest_debug_arch *host_dbg;
 	struct kvm_guest_debug_arch *guest_dbg;
 
-	if (!vcpu_get_flag(vcpu, DEBUG_DIRTY))
+	if (!kvm_debug_regs_in_use(vcpu))
 		return;
 
 	host_ctxt = host_data_ptr(host_ctxt);
 	guest_ctxt = &vcpu->arch.ctxt;
 	host_dbg = host_data_ptr(host_debug_state.regs);
-	guest_dbg = kern_hyp_va(vcpu->arch.debug_ptr);
+	guest_dbg = __vcpu_debug_regs(vcpu);
 
 	__debug_save_state(host_dbg, host_ctxt);
 	__debug_restore_state(guest_dbg, guest_ctxt);
···
 	struct kvm_guest_debug_arch *host_dbg;
 	struct kvm_guest_debug_arch *guest_dbg;
 
-	if (!vcpu_get_flag(vcpu, DEBUG_DIRTY))
+	if (!kvm_debug_regs_in_use(vcpu))
 		return;
 
 	host_ctxt = host_data_ptr(host_ctxt);
 	guest_ctxt = &vcpu->arch.ctxt;
 	host_dbg = host_data_ptr(host_debug_state.regs);
-	guest_dbg = kern_hyp_va(vcpu->arch.debug_ptr);
+	guest_dbg = __vcpu_debug_regs(vcpu);
 
 	__debug_save_state(guest_dbg, guest_ctxt);
 	__debug_restore_state(host_dbg, host_ctxt);
-
-	vcpu_clear_flag(vcpu, DEBUG_DIRTY);
 }
 
 #endif /* __ARM64_KVM_HYP_DEBUG_SR_H__ */

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

···
 
 static inline bool ctxt_has_s1poe(struct kvm_cpu_context *ctxt);
 
+static inline struct kvm_vcpu *ctxt_to_vcpu(struct kvm_cpu_context *ctxt)
+{
+	struct kvm_vcpu *vcpu = ctxt->__hyp_running_vcpu;
+
+	if (!vcpu)
+		vcpu = container_of(ctxt, struct kvm_vcpu, arch.ctxt);
+
+	return vcpu;
+}
+
+static inline bool ctxt_is_guest(struct kvm_cpu_context *ctxt)
+{
+	return host_data_ptr(host_ctxt) != ctxt;
+}
+
+static inline u64 *ctxt_mdscr_el1(struct kvm_cpu_context *ctxt)
+{
+	struct kvm_vcpu *vcpu = ctxt_to_vcpu(ctxt);
+
+	if (ctxt_is_guest(ctxt) && kvm_host_owns_debug_regs(vcpu))
+		return &vcpu->arch.external_mdscr_el1;
+
+	return &ctxt_sys_reg(ctxt, MDSCR_EL1);
+}
+
 static inline void __sysreg_save_common_state(struct kvm_cpu_context *ctxt)
 {
-	ctxt_sys_reg(ctxt, MDSCR_EL1)	= read_sysreg(mdscr_el1);
+	*ctxt_mdscr_el1(ctxt)	= read_sysreg(mdscr_el1);
 
 	// POR_EL0 can affect uaccess, so must be saved/restored early.
 	if (ctxt_has_s1poe(ctxt))
···
 {
 	ctxt_sys_reg(ctxt, TPIDR_EL0)	= read_sysreg(tpidr_el0);
 	ctxt_sys_reg(ctxt, TPIDRRO_EL0)	= read_sysreg(tpidrro_el0);
-}
-
-static inline struct kvm_vcpu *ctxt_to_vcpu(struct kvm_cpu_context *ctxt)
-{
-	struct kvm_vcpu *vcpu = ctxt->__hyp_running_vcpu;
-
-	if (!vcpu)
-		vcpu = container_of(ctxt, struct kvm_vcpu, arch.ctxt);
-
-	return vcpu;
 }
 
 static inline bool ctxt_has_mte(struct kvm_cpu_context *ctxt)
···
 
 static inline void __sysreg_restore_common_state(struct kvm_cpu_context *ctxt)
 {
-	write_sysreg(ctxt_sys_reg(ctxt, MDSCR_EL1),  mdscr_el1);
+	write_sysreg(*ctxt_mdscr_el1(ctxt),  mdscr_el1);
 
 	// POR_EL0 can affect uaccess, so must be saved/restored early.
 	if (ctxt_has_s1poe(ctxt))
···
 	__vcpu_sys_reg(vcpu, DACR32_EL2) = read_sysreg(dacr32_el2);
 	__vcpu_sys_reg(vcpu, IFSR32_EL2) = read_sysreg(ifsr32_el2);
 
-	if (has_vhe() || vcpu_get_flag(vcpu, DEBUG_DIRTY))
+	if (has_vhe() || kvm_debug_regs_in_use(vcpu))
 		__vcpu_sys_reg(vcpu, DBGVCR32_EL2) = read_sysreg(dbgvcr32_el2);
 }
 
···
 	write_sysreg(__vcpu_sys_reg(vcpu, DACR32_EL2), dacr32_el2);
 	write_sysreg(__vcpu_sys_reg(vcpu, IFSR32_EL2), ifsr32_el2);
 
-	if (has_vhe() || vcpu_get_flag(vcpu, DEBUG_DIRTY))
+	if (has_vhe() || kvm_debug_regs_in_use(vcpu))
 		write_sysreg(__vcpu_sys_reg(vcpu, DBGVCR32_EL2), dbgvcr32_el2);
 }
 

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

···
 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))
+	if (host_data_test_flag(HAS_SPE))
 		__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))
+	if (host_data_test_flag(HAS_TRBE))
 		__debug_save_trace(host_data_ptr(host_debug_state.trfcr_el1));
 }
 
···
 
 void __debug_restore_host_buffers_nvhe(struct kvm_vcpu *vcpu)
 {
-	if (vcpu_get_flag(vcpu, DEBUG_STATE_SAVE_SPE))
+	if (host_data_test_flag(HAS_SPE))
 		__debug_restore_spe(*host_data_ptr(host_debug_state.pmscr_el1));
-	if (vcpu_get_flag(vcpu, DEBUG_STATE_SAVE_TRBE))
+	if (host_data_test_flag(HAS_TRBE))
 		__debug_restore_trace(*host_data_ptr(host_debug_state.trfcr_el1));
 }
 
 void __debug_switch_to_host(struct kvm_vcpu *vcpu)
 {
 	__debug_switch_to_host_common(vcpu);
-}
-
-u64 __kvm_get_mdcr_el2(void)
-{
-	return read_sysreg(mdcr_el2);
 }

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

···
 
 	hyp_vcpu->vcpu.arch.iflags	= host_vcpu->arch.iflags;
 
-	hyp_vcpu->vcpu.arch.debug_ptr	= kern_hyp_va(host_vcpu->arch.debug_ptr);
-
 	hyp_vcpu->vcpu.arch.vsesr_el2	= host_vcpu->arch.vsesr_el2;
 
 	hyp_vcpu->vcpu.arch.vgic_cpu.vgic_v3 = host_vcpu->arch.vgic_cpu.vgic_v3;
···
 	__vgic_v3_init_lrs();
 }
 
-static void handle___kvm_get_mdcr_el2(struct kvm_cpu_context *host_ctxt)
-{
-	cpu_reg(host_ctxt, 1) = __kvm_get_mdcr_el2();
-}
-
 static void handle___vgic_v3_save_vmcr_aprs(struct kvm_cpu_context *host_ctxt)
 {
 	DECLARE_REG(struct vgic_v3_cpu_if *, cpu_if, host_ctxt, 1);
···
 
 static const hcall_t host_hcall[] = {
 	/* ___kvm_hyp_init */
-	HANDLE_FUNC(__kvm_get_mdcr_el2),
 	HANDLE_FUNC(__pkvm_init),
 	HANDLE_FUNC(__pkvm_create_private_mapping),
 	HANDLE_FUNC(__pkvm_cpu_set_vector),

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

···
 {
 	__debug_switch_to_host_common(vcpu);
 }
-
-u64 __kvm_get_mdcr_el2(void)
-{
-	return read_sysreg(mdcr_el2);
-}

arch/arm64/kvm/sys_regs.c

···
 			   struct sys_reg_params *p,
 			   const struct sys_reg_desc *r)
 {
-	u64 oslsr;
-
 	if (!p->is_write)
 		return read_from_write_only(vcpu, p, r);
 
-	/* Forward the OSLK bit to OSLSR */
-	oslsr = __vcpu_sys_reg(vcpu, OSLSR_EL1) & ~OSLSR_EL1_OSLK;
-	if (p->regval & OSLAR_EL1_OSLK)
-		oslsr |= OSLSR_EL1_OSLK;
-
-	__vcpu_sys_reg(vcpu, OSLSR_EL1) = oslsr;
+	kvm_debug_handle_oslar(vcpu, p->regval);
 	return true;
 }
 
···
 	}
 }
 
-/*
- * We want to avoid world-switching all the DBG registers all the
- * time:
- *
- * - If we've touched any debug register, it is likely that we're
- *   going to touch more of them. It then makes sense to disable the
- *   traps and start doing the save/restore dance
- * - If debug is active (DBG_MDSCR_KDE or DBG_MDSCR_MDE set), it is
- *   then mandatory to save/restore the registers, as the guest
- *   depends on them.
- *
- * For this, we use a DIRTY bit, indicating the guest has modified the
- * debug registers, used as follow:
- *
- * On guest entry:
- * - If the dirty bit is set (because we're coming back from trapping),
- *   disable the traps, save host registers, restore guest registers.
- * - If debug is actively in use (DBG_MDSCR_KDE or DBG_MDSCR_MDE set),
- *   set the dirty bit, disable the traps, save host registers,
- *   restore guest registers.
- * - Otherwise, enable the traps
- *
- * On guest exit:
- * - If the dirty bit is set, save guest registers, restore host
- *   registers and clear the dirty bit. This ensure that the host can
- *   now use the debug registers.
- */
 static bool trap_debug_regs(struct kvm_vcpu *vcpu,
 			    struct sys_reg_params *p,
 			    const struct sys_reg_desc *r)
 {
 	access_rw(vcpu, p, r);
-	if (p->is_write)
-		vcpu_set_flag(vcpu, DEBUG_DIRTY);
 
-	trace_trap_reg(__func__, r->reg, p->is_write, p->regval);
-
+	kvm_debug_set_guest_ownership(vcpu);
 	return true;
 }
 
···
  *
  * A 32 bit write to a debug register leave top bits alone
  * A 32 bit read from a debug register only returns the bottom bits
- *
- * All writes will set the DEBUG_DIRTY flag to ensure the hyp code
- * switches between host and guest values in future.
  */
 static void reg_to_dbg(struct kvm_vcpu *vcpu,
 		       struct sys_reg_params *p,
···
 	val &= ~mask;
 	val |= (p->regval & (mask >> shift)) << shift;
 	*dbg_reg = val;
-
-	vcpu_set_flag(vcpu, DEBUG_DIRTY);
 }
 
 static void dbg_to_reg(struct kvm_vcpu *vcpu,
···
 	p->regval = (*dbg_reg & mask) >> shift;
 }
 
-static bool trap_bvr(struct kvm_vcpu *vcpu,
-		     struct sys_reg_params *p,
-		     const struct sys_reg_desc *rd)
+static u64 *demux_wb_reg(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd)
 {
-	u64 *dbg_reg = &vcpu->arch.vcpu_debug_state.dbg_bvr[rd->CRm];
+	struct kvm_guest_debug_arch *dbg = &vcpu->arch.vcpu_debug_state;
+
+	switch (rd->Op2) {
+	case 0b100:
+		return &dbg->dbg_bvr[rd->CRm];
+	case 0b101:
+		return &dbg->dbg_bcr[rd->CRm];
+	case 0b110:
+		return &dbg->dbg_wvr[rd->CRm];
+	case 0b111:
+		return &dbg->dbg_wcr[rd->CRm];
+	default:
+		KVM_BUG_ON(1, vcpu->kvm);
+		return NULL;
+	}
+}
+
+static bool trap_dbg_wb_reg(struct kvm_vcpu *vcpu, struct sys_reg_params *p,
+			    const struct sys_reg_desc *rd)
+{
+	u64 *reg = demux_wb_reg(vcpu, rd);
+
+	if (!reg)
+		return false;
 
 	if (p->is_write)
-		reg_to_dbg(vcpu, p, rd, dbg_reg);
+		reg_to_dbg(vcpu, p, rd, reg);
 	else
-		dbg_to_reg(vcpu, p, rd, dbg_reg);
+		dbg_to_reg(vcpu, p, rd, reg);
 
-	trace_trap_reg(__func__, rd->CRm, p->is_write, *dbg_reg);
-
+	kvm_debug_set_guest_ownership(vcpu);
 	return true;
 }
 
-static int set_bvr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
-		   u64 val)
+static int set_dbg_wb_reg(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
+			  u64 val)
 {
-	vcpu->arch.vcpu_debug_state.dbg_bvr[rd->CRm] = val;
+	u64 *reg = demux_wb_reg(vcpu, rd);
+
+	if (!reg)
+		return -EINVAL;
+
+	*reg = val;
 	return 0;
 }
 
-static int get_bvr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
-		   u64 *val)
+static int get_dbg_wb_reg(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
+			  u64 *val)
 {
-	*val = vcpu->arch.vcpu_debug_state.dbg_bvr[rd->CRm];
+	u64 *reg = demux_wb_reg(vcpu, rd);
+
+	if (!reg)
+		return -EINVAL;
+
+	*val = *reg;
 	return 0;
 }
 
-static u64 reset_bvr(struct kvm_vcpu *vcpu,
-		      const struct sys_reg_desc *rd)
+static u64 reset_dbg_wb_reg(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd)
 {
-	vcpu->arch.vcpu_debug_state.dbg_bvr[rd->CRm] = rd->val;
-	return rd->val;
-}
+	u64 *reg = demux_wb_reg(vcpu, rd);
 
-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->CRm];
+	/*
+	 * Bail early if we couldn't find storage for the register, the
+	 * KVM_BUG_ON() in demux_wb_reg() will prevent this VM from ever
+	 * being run.
+	 */
+	if (!reg)
+		return 0;
 
-	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->CRm, p->is_write, *dbg_reg);
-
-	return true;
-}
-
-static int set_bcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
-		   u64 val)
-{
-	vcpu->arch.vcpu_debug_state.dbg_bcr[rd->CRm] = val;
-	return 0;
-}
-
-static int get_bcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
-		   u64 *val)
-{
-	*val = vcpu->arch.vcpu_debug_state.dbg_bcr[rd->CRm];
-	return 0;
-}
-
-static u64 reset_bcr(struct kvm_vcpu *vcpu,
-		      const struct sys_reg_desc *rd)
-{
-	vcpu->arch.vcpu_debug_state.dbg_bcr[rd->CRm] = rd->val;
-	return 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->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->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,
-		   u64 val)
-{
-	vcpu->arch.vcpu_debug_state.dbg_wvr[rd->CRm] = val;
-	return 0;
-}
-
-static int get_wvr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
-		   u64 *val)
-{
-	*val = vcpu->arch.vcpu_debug_state.dbg_wvr[rd->CRm];
-	return 0;
-}
-
-static u64 reset_wvr(struct kvm_vcpu *vcpu,
-		      const struct sys_reg_desc *rd)
-{
-	vcpu->arch.vcpu_debug_state.dbg_wvr[rd->CRm] = rd->val;
-	return 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->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->CRm, p->is_write, *dbg_reg);
-
-	return true;
-}
-
-static int set_wcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
-		   u64 val)
-{
-	vcpu->arch.vcpu_debug_state.dbg_wcr[rd->CRm] = val;
-	return 0;
-}
-
-static int get_wcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
-		   u64 *val)
-{
-	*val = vcpu->arch.vcpu_debug_state.dbg_wcr[rd->CRm];
-	return 0;
-}
-
-static u64 reset_wcr(struct kvm_vcpu *vcpu,
-		      const struct sys_reg_desc *rd)
-{
-	vcpu->arch.vcpu_debug_state.dbg_wcr[rd->CRm] = rd->val;
+	*reg = rd->val;
 	return rd->val;
 }
 
···
 /* Silly macro to expand the DBG{BCR,BVR,WVR,WCR}n_EL1 registers in one go */
 #define DBG_BCR_BVR_WCR_WVR_EL1(n)					\
 	{ SYS_DESC(SYS_DBGBVRn_EL1(n)),					\
-	  trap_bvr, reset_bvr, 0, 0, get_bvr, set_bvr },		\
+	  trap_dbg_wb_reg, reset_dbg_wb_reg, 0, 0,			\
+	  get_dbg_wb_reg, set_dbg_wb_reg },				\
 	{ SYS_DESC(SYS_DBGBCRn_EL1(n)),					\
-	  trap_bcr, reset_bcr, 0, 0, get_bcr, set_bcr },		\
+	  trap_dbg_wb_reg, reset_dbg_wb_reg, 0, 0,			\
+	  get_dbg_wb_reg, set_dbg_wb_reg },				\
 	{ SYS_DESC(SYS_DBGWVRn_EL1(n)),					\
-	  trap_wvr, reset_wvr, 0, 0,  get_wvr, set_wvr },		\
+	  trap_dbg_wb_reg, reset_dbg_wb_reg, 0, 0,			\
+	  get_dbg_wb_reg, set_dbg_wb_reg },				\
 	{ SYS_DESC(SYS_DBGWCRn_EL1(n)),					\
-	  trap_wcr, reset_wcr, 0, 0,  get_wcr, set_wcr }
+	  trap_dbg_wb_reg, reset_dbg_wb_reg, 0, 0,			\
+	  get_dbg_wb_reg, set_dbg_wb_reg }
 
 #define PMU_SYS_REG(name)						\
 	SYS_DESC(SYS_##name), .reset = reset_pmu_reg,			\
···
  * None of the other registers share their location, so treat them as
  * if they were 64bit.
  */
-#define DBG_BCR_BVR_WCR_WVR(n)						      \
-	/* DBGBVRn */							      \
-	{ AA32(LO), Op1( 0), CRn( 0), CRm((n)), Op2( 4), trap_bvr, NULL, n }, \
-	/* DBGBCRn */							      \
-	{ Op1( 0), CRn( 0), CRm((n)), Op2( 5), trap_bcr, NULL, n },	      \
-	/* DBGWVRn */							      \
-	{ Op1( 0), CRn( 0), CRm((n)), Op2( 6), trap_wvr, NULL, n },	      \
-	/* DBGWCRn */							      \
-	{ Op1( 0), CRn( 0), CRm((n)), Op2( 7), trap_wcr, NULL, n }
+#define DBG_BCR_BVR_WCR_WVR(n)							\
+	/* DBGBVRn */								\
+	{ AA32(LO), Op1( 0), CRn( 0), CRm((n)), Op2( 4),			\
+	  trap_dbg_wb_reg, NULL, n },						\
+	/* DBGBCRn */								\
+	{ Op1( 0), CRn( 0), CRm((n)), Op2( 5), trap_dbg_wb_reg, NULL, n },	\
+	/* DBGWVRn */								\
+	{ Op1( 0), CRn( 0), CRm((n)), Op2( 6), trap_dbg_wb_reg, NULL, n },	\
+	/* DBGWCRn */								\
+	{ Op1( 0), CRn( 0), CRm((n)), Op2( 7), trap_dbg_wb_reg, NULL, n }
 
-#define DBGBXVR(n)							      \
-	{ AA32(HI), Op1( 0), CRn( 1), CRm((n)), Op2( 1), trap_bvr, NULL, n }
+#define DBGBXVR(n)								\
+	{ AA32(HI), Op1( 0), CRn( 1), CRm((n)), Op2( 1),			\
+	  trap_dbg_wb_reg, NULL, n }
 
 /*
  * Trapped cp14 registers. We generally ignore most of the external

arch/arm64/kvm/trace_handle_exit.h

···
 		  __entry->vcpu_pc, __entry->r0, __entry->imm)
 );
 
-TRACE_EVENT(kvm_arm_setup_debug,
-	TP_PROTO(struct kvm_vcpu *vcpu, __u32 guest_debug),
-	TP_ARGS(vcpu, guest_debug),
-
-	TP_STRUCT__entry(
-		__field(struct kvm_vcpu *, vcpu)
-		__field(__u32, guest_debug)
-	),
-
-	TP_fast_assign(
-		__entry->vcpu = vcpu;
-		__entry->guest_debug = guest_debug;
-	),
-
-	TP_printk("vcpu: %p, flags: 0x%08x", __entry->vcpu, __entry->guest_debug)
-);
-
-TRACE_EVENT(kvm_arm_clear_debug,
-	TP_PROTO(__u32 guest_debug),
-	TP_ARGS(guest_debug),
-
-	TP_STRUCT__entry(
-		__field(__u32, guest_debug)
-	),
-
-	TP_fast_assign(
-		__entry->guest_debug = guest_debug;
-	),
-
-	TP_printk("flags: 0x%08x", __entry->guest_debug)
-);
-
 /*
  * The dreg32 name is a leftover from a distant past. This will really
  * output a 64bit value...
···
 	),
 
 	TP_printk("%s: 0x%llx", __entry->name, __entry->value)
-);
-
-TRACE_DEFINE_SIZEOF(__u64);
-
-TRACE_EVENT(kvm_arm_set_regset,
-	TP_PROTO(const char *type, int len, __u64 *control, __u64 *value),
-	TP_ARGS(type, len, control, value),
-	TP_STRUCT__entry(
-		__field(const char *, name)
-		__field(int, len)
-		__array(u64, ctrls, 16)
-		__array(u64, values, 16)
-	),
-	TP_fast_assign(
-		__entry->name = type;
-		__entry->len = len;
-		memcpy(__entry->ctrls, control, len << 3);
-		memcpy(__entry->values, value, len << 3);
-	),
-	TP_printk("%d %s CTRL:%s VALUE:%s", __entry->len, __entry->name,
-		__print_array(__entry->ctrls, __entry->len, sizeof(__u64)),
-		__print_array(__entry->values, __entry->len, sizeof(__u64)))
-);
-
-TRACE_EVENT(trap_reg,
-	TP_PROTO(const char *fn, int reg, bool is_write, u64 write_value),
-	TP_ARGS(fn, reg, is_write, write_value),
-
-	TP_STRUCT__entry(
-		__field(const char *, fn)
-		__field(int, reg)
-		__field(bool, is_write)
-		__field(u64, write_value)
-	),
-
-	TP_fast_assign(
-		__entry->fn = fn;
-		__entry->reg = reg;
-		__entry->is_write = is_write;
-		__entry->write_value = write_value;
-	),
-
-	TP_printk("%s %s reg %d (0x%016llx)", __entry->fn,  __entry->is_write?"write to":"read from", __entry->reg, __entry->write_value)
 );
 
 TRACE_EVENT(kvm_handle_sys_reg,