KVM: arm64: Add RMW specific sysreg accessor

In a number of cases, we perform a Read-Modify-Write operation on
a system register, meaning that we would apply the RESx masks twice.

Instead, provide a new accessor that performs this RMW operation,
allowing the masks to be applied exactly once per operation.

Reviewed-by: Miguel Luis <[e-mail]>
Reviewed-by: Oliver Upton <[e-mail]>
Link: https://lore.kernel.org/r/[e-mail]
Signed-off-by: Marc Zyngier <[e-mail]>

arch/arm64/include/asm/kvm_host.h

···
 		ctxt_sys_reg(ctxt, (r)) = __v;				\
 	} while (0)
 
+#define __vcpu_rmw_sys_reg(v, r, op, val)				\
+	do {								\
+		const struct kvm_cpu_context *ctxt = &(v)->arch.ctxt;	\
+		u64 __v = ctxt_sys_reg(ctxt, (r));			\
+		__v op (val);						\
+		if (vcpu_has_nv((v)) && (r) >= __SANITISED_REG_START__)	\
+			__v = kvm_vcpu_apply_reg_masks((v), (r), __v);	\
+									\
+		ctxt_sys_reg(ctxt, (r)) = __v;				\
+	} while (0)
+
 #define __vcpu_sys_reg(v,r)						\
 	(*({								\
 		const struct kvm_cpu_context *ctxt = &(v)->arch.ctxt;	\

arch/arm64/kvm/debug.c

···
 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;
+		__vcpu_rmw_sys_reg(vcpu, OSLSR_EL1, |=, OSLSR_EL1_OSLK);
 	else
-		__vcpu_sys_reg(vcpu, OSLSR_EL1) &= ~OSLSR_EL1_OSLK;
+		__vcpu_rmw_sys_reg(vcpu, OSLSR_EL1, &=, ~OSLSR_EL1_OSLK);
 
 	preempt_disable();
 	kvm_arch_vcpu_put(vcpu);

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

···
 		 */
 		val = read_sysreg_el1(SYS_CNTKCTL);
 		val &= CNTKCTL_VALID_BITS;
-		__vcpu_sys_reg(vcpu, CNTHCTL_EL2) &= ~CNTKCTL_VALID_BITS;
-		__vcpu_sys_reg(vcpu, CNTHCTL_EL2) |= val;
+		__vcpu_rmw_sys_reg(vcpu, CNTHCTL_EL2, &=, ~CNTKCTL_VALID_BITS);
+		__vcpu_rmw_sys_reg(vcpu, CNTHCTL_EL2, |=, val);
 	}
 
 	__vcpu_assign_sys_reg(vcpu, SP_EL2,	 read_sysreg(sp_el1));

arch/arm64/kvm/nested.c

···
 
 out:
 	for (enum vcpu_sysreg sr = __SANITISED_REG_START__; sr < NR_SYS_REGS; sr++)
-		(void)__vcpu_sys_reg(vcpu, sr);
+		__vcpu_rmw_sys_reg(vcpu, sr, |=, 0);
 
 	return 0;
 }

arch/arm64/kvm/pmu-emul.c

···
 			continue;
 
 		/* Mark overflow */
-		__vcpu_sys_reg(vcpu, PMOVSSET_EL0) |= BIT(i);
+		__vcpu_rmw_sys_reg(vcpu, PMOVSSET_EL0, |=, BIT(i));
 
 		if (kvm_pmu_counter_can_chain(pmc))
 			kvm_pmu_counter_increment(vcpu, BIT(i + 1),
···
 	perf_event->attr.sample_period = period;
 	perf_event->hw.sample_period = period;
 
-	__vcpu_sys_reg(vcpu, PMOVSSET_EL0) |= BIT(idx);
+	__vcpu_rmw_sys_reg(vcpu, PMOVSSET_EL0, |=, BIT(idx));
 
 	if (kvm_pmu_counter_can_chain(pmc))
 		kvm_pmu_counter_increment(vcpu, BIT(idx + 1),
···
 {
 	u64 mask = kvm_pmu_implemented_counter_mask(vcpu);
 
-	__vcpu_sys_reg(vcpu, PMOVSSET_EL0) &= mask;
-	__vcpu_sys_reg(vcpu, PMINTENSET_EL1) &= mask;
-	__vcpu_sys_reg(vcpu, PMCNTENSET_EL0) &= mask;
+	__vcpu_rmw_sys_reg(vcpu, PMOVSSET_EL0, &=, mask);
+	__vcpu_rmw_sys_reg(vcpu, PMINTENSET_EL1, &=, mask);
+	__vcpu_rmw_sys_reg(vcpu, PMCNTENSET_EL0, &=, mask);
 
 	kvm_pmu_reprogram_counter_mask(vcpu, mask);
 }

arch/arm64/kvm/sys_regs.c

···
 		mask |= GENMASK(n - 1, 0);
 
 	reset_unknown(vcpu, r);
-	__vcpu_sys_reg(vcpu, r->reg) &= mask;
+	__vcpu_rmw_sys_reg(vcpu, r->reg, &=, mask);
 
 	return __vcpu_sys_reg(vcpu, r->reg);
 }
···
 static u64 reset_pmevcntr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r)
 {
 	reset_unknown(vcpu, r);
-	__vcpu_sys_reg(vcpu, r->reg) &= GENMASK(31, 0);
+	__vcpu_rmw_sys_reg(vcpu, r->reg, &=, GENMASK(31, 0));
 
 	return __vcpu_sys_reg(vcpu, r->reg);
 }
···
 		return 0;
 
 	reset_unknown(vcpu, r);
-	__vcpu_sys_reg(vcpu, r->reg) &= kvm_pmu_evtyper_mask(vcpu->kvm);
+	__vcpu_rmw_sys_reg(vcpu, r->reg, &=, kvm_pmu_evtyper_mask(vcpu->kvm));
 
 	return __vcpu_sys_reg(vcpu, r->reg);
 }
···
 static u64 reset_pmselr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r)
 {
 	reset_unknown(vcpu, r);
-	__vcpu_sys_reg(vcpu, r->reg) &= PMSELR_EL0_SEL_MASK;
+	__vcpu_rmw_sys_reg(vcpu, r->reg, &=, PMSELR_EL0_SEL_MASK);
 
 	return __vcpu_sys_reg(vcpu, r->reg);
 }
···
 		val = p->regval & mask;
 		if (r->Op2 & 0x1)
 			/* accessing PMCNTENSET_EL0 */
-			__vcpu_sys_reg(vcpu, PMCNTENSET_EL0) |= val;
+			__vcpu_rmw_sys_reg(vcpu, PMCNTENSET_EL0, |=, val);
 		else
 			/* accessing PMCNTENCLR_EL0 */
-			__vcpu_sys_reg(vcpu, PMCNTENSET_EL0) &= ~val;
+			__vcpu_rmw_sys_reg(vcpu, PMCNTENSET_EL0, &=, ~val);
 
 		kvm_pmu_reprogram_counter_mask(vcpu, val);
 	} else {
···
 
 		if (r->Op2 & 0x1)
 			/* accessing PMINTENSET_EL1 */
-			__vcpu_sys_reg(vcpu, PMINTENSET_EL1) |= val;
+			__vcpu_rmw_sys_reg(vcpu, PMINTENSET_EL1, |=, val);
 		else
 			/* accessing PMINTENCLR_EL1 */
-			__vcpu_sys_reg(vcpu, PMINTENSET_EL1) &= ~val;
+			__vcpu_rmw_sys_reg(vcpu, PMINTENSET_EL1, &=, ~val);
 	} else {
 		p->regval = __vcpu_sys_reg(vcpu, PMINTENSET_EL1);
 	}
···
 	if (p->is_write) {
 		if (r->CRm & 0x2)
 			/* accessing PMOVSSET_EL0 */
-			__vcpu_sys_reg(vcpu, PMOVSSET_EL0) |= (p->regval & mask);
+			__vcpu_rmw_sys_reg(vcpu, PMOVSSET_EL0, |=, (p->regval & mask));
 		else
 			/* accessing PMOVSCLR_EL0 */
-			__vcpu_sys_reg(vcpu, PMOVSSET_EL0) &= ~(p->regval & mask);
+			__vcpu_rmw_sys_reg(vcpu, PMOVSSET_EL0, &=, ~(p->regval & mask));
 	} else {
 		p->regval = __vcpu_sys_reg(vcpu, PMOVSSET_EL0);
 	}
···
 			r->reset(vcpu, r);
 
 		if (r->reg >= __SANITISED_REG_START__ && r->reg < NR_SYS_REGS)
-			(void)__vcpu_sys_reg(vcpu, r->reg);
+			__vcpu_rmw_sys_reg(vcpu, r->reg, |=, 0);
 	}
 
 	set_bit(KVM_ARCH_FLAG_ID_REGS_INITIALIZED, &kvm->arch.flags);