Merge tag 'kvmarm-fixes-6.16-2' of git://git.kernel.org/pub/scm/linux…

…/kernel/git/kvmarm/kvmarm into HEAD

KVM/arm64 fixes for 6.16, take #2

- Rework of system register accessors for system registers that are
  directly writen to memory, so that sanitisation of the in-memory
  value happens at the correct time (after the read, or before the
  write). For convenience, RMW-style accessors are also provided.

- Multiple fixes for the so-called "arch-timer-edge-cases' selftest,
  which was always broken.

arch/arm64/include/asm/kvm_host.h

@@ -1107,14 +1107,36 @@ static inline u64 *___ctxt_sys_reg(const struct kvm_cpu_context *ctxt, int r)
 #define ctxt_sys_reg(c,r)	(*__ctxt_sys_reg(c,r))
 
 u64 kvm_vcpu_apply_reg_masks(const struct kvm_vcpu *, enum vcpu_sysreg, u64);
+
+#define __vcpu_assign_sys_reg(v, r, val)				\
+	do {								\
+		const struct kvm_cpu_context *ctxt = &(v)->arch.ctxt;	\
+		u64 __v = (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_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;	\
-		u64 *__r = __ctxt_sys_reg(ctxt, (r));			\
+		u64 __v = ctxt_sys_reg(ctxt, (r));			\
 		if (vcpu_has_nv((v)) && (r) >= __SANITISED_REG_START__)	\
-			*__r = kvm_vcpu_apply_reg_masks((v), (r), *__r);\
-		__r;							\
-	}))
+			__v = kvm_vcpu_apply_reg_masks((v), (r), __v);	\
+		__v;							\
+	})
 
 u64 vcpu_read_sys_reg(const struct kvm_vcpu *vcpu, int reg);
 void vcpu_write_sys_reg(struct kvm_vcpu *vcpu, u64 val, int reg);

arch/arm64/kvm/arch_timer.c

@@ -108,16 +108,16 @@ static void timer_set_ctl(struct arch_timer_context *ctxt, u32 ctl)
 
 	switch(arch_timer_ctx_index(ctxt)) {
 	case TIMER_VTIMER:
-		__vcpu_sys_reg(vcpu, CNTV_CTL_EL0) = ctl;
+		__vcpu_assign_sys_reg(vcpu, CNTV_CTL_EL0, ctl);
 		break;
 	case TIMER_PTIMER:
-		__vcpu_sys_reg(vcpu, CNTP_CTL_EL0) = ctl;
+		__vcpu_assign_sys_reg(vcpu, CNTP_CTL_EL0, ctl);
 		break;
 	case TIMER_HVTIMER:
-		__vcpu_sys_reg(vcpu, CNTHV_CTL_EL2) = ctl;
+		__vcpu_assign_sys_reg(vcpu, CNTHV_CTL_EL2, ctl);
 		break;
 	case TIMER_HPTIMER:
-		__vcpu_sys_reg(vcpu, CNTHP_CTL_EL2) = ctl;
+		__vcpu_assign_sys_reg(vcpu, CNTHP_CTL_EL2, ctl);
 		break;
 	default:
 		WARN_ON(1);
@@ -130,16 +130,16 @@ static void timer_set_cval(struct arch_timer_context *ctxt, u64 cval)
 
 	switch(arch_timer_ctx_index(ctxt)) {
 	case TIMER_VTIMER:
-		__vcpu_sys_reg(vcpu, CNTV_CVAL_EL0) = cval;
+		__vcpu_assign_sys_reg(vcpu, CNTV_CVAL_EL0, cval);
 		break;
 	case TIMER_PTIMER:
-		__vcpu_sys_reg(vcpu, CNTP_CVAL_EL0) = cval;
+		__vcpu_assign_sys_reg(vcpu, CNTP_CVAL_EL0, cval);
 		break;
 	case TIMER_HVTIMER:
-		__vcpu_sys_reg(vcpu, CNTHV_CVAL_EL2) = cval;
+		__vcpu_assign_sys_reg(vcpu, CNTHV_CVAL_EL2, cval);
 		break;
 	case TIMER_HPTIMER:
-		__vcpu_sys_reg(vcpu, CNTHP_CVAL_EL2) = cval;
+		__vcpu_assign_sys_reg(vcpu, CNTHP_CVAL_EL2, cval);
 		break;
 	default:
 		WARN_ON(1);
@@ -1036,7 +1036,7 @@ void kvm_timer_vcpu_reset(struct kvm_vcpu *vcpu)
 	if (vcpu_has_nv(vcpu)) {
 		struct arch_timer_offset *offs = &vcpu_vtimer(vcpu)->offset;
 
-		offs->vcpu_offset = &__vcpu_sys_reg(vcpu, CNTVOFF_EL2);
+		offs->vcpu_offset = __ctxt_sys_reg(&vcpu->arch.ctxt, CNTVOFF_EL2);
 		offs->vm_offset = &vcpu->kvm->arch.timer_data.poffset;
 	}
 

arch/arm64/kvm/debug.c

@@ -216,9 +216,9 @@ void kvm_debug_set_guest_ownership(struct kvm_vcpu *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;
+		__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/fpsimd.c

@@ -103,8 +103,8 @@ void kvm_arch_vcpu_ctxsync_fp(struct kvm_vcpu *vcpu)
 		fp_state.sve_state = vcpu->arch.sve_state;
 		fp_state.sve_vl = vcpu->arch.sve_max_vl;
 		fp_state.sme_state = NULL;
-		fp_state.svcr = &__vcpu_sys_reg(vcpu, SVCR);
-		fp_state.fpmr = &__vcpu_sys_reg(vcpu, FPMR);
+		fp_state.svcr = __ctxt_sys_reg(&vcpu->arch.ctxt, SVCR);
+		fp_state.fpmr = __ctxt_sys_reg(&vcpu->arch.ctxt, FPMR);
 		fp_state.fp_type = &vcpu->arch.fp_type;
 
 		if (vcpu_has_sve(vcpu))

arch/arm64/kvm/hyp/exception.c

@@ -37,7 +37,7 @@ static inline void __vcpu_write_sys_reg(struct kvm_vcpu *vcpu, u64 val, int reg)
 	if (unlikely(vcpu_has_nv(vcpu)))
 		vcpu_write_sys_reg(vcpu, val, reg);
 	else if (!__vcpu_write_sys_reg_to_cpu(val, reg))
-		__vcpu_sys_reg(vcpu, reg) = val;
+		__vcpu_assign_sys_reg(vcpu, reg, val);
 }
 
 static void __vcpu_write_spsr(struct kvm_vcpu *vcpu, unsigned long target_mode,
@@ -51,7 +51,7 @@ static void __vcpu_write_spsr(struct kvm_vcpu *vcpu, unsigned long target_mode,
 	} else if (has_vhe()) {
 		write_sysreg_el1(val, SYS_SPSR);
 	} else {
-		__vcpu_sys_reg(vcpu, SPSR_EL1) = val;
+		__vcpu_assign_sys_reg(vcpu, SPSR_EL1, val);
 	}
 }
 

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

@@ -45,7 +45,7 @@ static inline void __fpsimd_save_fpexc32(struct kvm_vcpu *vcpu)
 	if (!vcpu_el1_is_32bit(vcpu))
 		return;
 
-	__vcpu_sys_reg(vcpu, FPEXC32_EL2) = read_sysreg(fpexc32_el2);
+	__vcpu_assign_sys_reg(vcpu, FPEXC32_EL2, read_sysreg(fpexc32_el2));
 }
 
 static inline void __activate_traps_fpsimd32(struct kvm_vcpu *vcpu)
@@ -456,7 +456,7 @@ static inline void fpsimd_lazy_switch_to_host(struct kvm_vcpu *vcpu)
 	 */
 	if (vcpu_has_sve(vcpu)) {
 		zcr_el1 = read_sysreg_el1(SYS_ZCR);
-		__vcpu_sys_reg(vcpu, vcpu_sve_zcr_elx(vcpu)) = zcr_el1;
+		__vcpu_assign_sys_reg(vcpu, vcpu_sve_zcr_elx(vcpu), zcr_el1);
 
 		/*
 		 * The guest's state is always saved using the guest's max VL.

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

@@ -307,11 +307,11 @@ static inline void __sysreg32_save_state(struct kvm_vcpu *vcpu)
 	vcpu->arch.ctxt.spsr_irq = read_sysreg(spsr_irq);
 	vcpu->arch.ctxt.spsr_fiq = read_sysreg(spsr_fiq);
 
-	__vcpu_sys_reg(vcpu, DACR32_EL2) = read_sysreg(dacr32_el2);
-	__vcpu_sys_reg(vcpu, IFSR32_EL2) = read_sysreg(ifsr32_el2);
+	__vcpu_assign_sys_reg(vcpu, DACR32_EL2, read_sysreg(dacr32_el2));
+	__vcpu_assign_sys_reg(vcpu, IFSR32_EL2, read_sysreg(ifsr32_el2));
 
 	if (has_vhe() || kvm_debug_regs_in_use(vcpu))
-		__vcpu_sys_reg(vcpu, DBGVCR32_EL2) = read_sysreg(dbgvcr32_el2);
+		__vcpu_assign_sys_reg(vcpu, DBGVCR32_EL2, read_sysreg(dbgvcr32_el2));
 }
 
 static inline void __sysreg32_restore_state(struct kvm_vcpu *vcpu)

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

@@ -26,7 +26,7 @@ void __kvm_hyp_host_forward_smc(struct kvm_cpu_context *host_ctxt);
 
 static void __hyp_sve_save_guest(struct kvm_vcpu *vcpu)
 {
-	__vcpu_sys_reg(vcpu, ZCR_EL1) = read_sysreg_el1(SYS_ZCR);
+	__vcpu_assign_sys_reg(vcpu, ZCR_EL1, read_sysreg_el1(SYS_ZCR));
 	/*
 	 * On saving/restoring guest sve state, always use the maximum VL for
 	 * the guest. The layout of the data when saving the sve state depends
@@ -79,7 +79,7 @@ static void fpsimd_sve_sync(struct kvm_vcpu *vcpu)
 
 	has_fpmr = kvm_has_fpmr(kern_hyp_va(vcpu->kvm));
 	if (has_fpmr)
-		__vcpu_sys_reg(vcpu, FPMR) = read_sysreg_s(SYS_FPMR);
+		__vcpu_assign_sys_reg(vcpu, FPMR, read_sysreg_s(SYS_FPMR));
 
 	if (system_supports_sve())
 		__hyp_sve_restore_host();

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

@@ -223,9 +223,9 @@ static void __deactivate_traps(struct kvm_vcpu *vcpu)
 		 */
 		val = read_sysreg_el0(SYS_CNTP_CVAL);
 		if (map.direct_ptimer == vcpu_ptimer(vcpu))
-			__vcpu_sys_reg(vcpu, CNTP_CVAL_EL0) = val;
+			__vcpu_assign_sys_reg(vcpu, CNTP_CVAL_EL0, val);
 		if (map.direct_ptimer == vcpu_hptimer(vcpu))
-			__vcpu_sys_reg(vcpu, CNTHP_CVAL_EL2) = val;
+			__vcpu_assign_sys_reg(vcpu, CNTHP_CVAL_EL2, val);
 
 		offset = read_sysreg_s(SYS_CNTPOFF_EL2);
 

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

@@ -18,17 +18,17 @@
 static void __sysreg_save_vel2_state(struct kvm_vcpu *vcpu)
 {
 	/* These registers are common with EL1 */
-	__vcpu_sys_reg(vcpu, PAR_EL1)	= read_sysreg(par_el1);
-	__vcpu_sys_reg(vcpu, TPIDR_EL1)	= read_sysreg(tpidr_el1);
-
-	__vcpu_sys_reg(vcpu, ESR_EL2)	= read_sysreg_el1(SYS_ESR);
-	__vcpu_sys_reg(vcpu, AFSR0_EL2)	= read_sysreg_el1(SYS_AFSR0);
-	__vcpu_sys_reg(vcpu, AFSR1_EL2)	= read_sysreg_el1(SYS_AFSR1);
-	__vcpu_sys_reg(vcpu, FAR_EL2)	= read_sysreg_el1(SYS_FAR);
-	__vcpu_sys_reg(vcpu, MAIR_EL2)	= read_sysreg_el1(SYS_MAIR);
-	__vcpu_sys_reg(vcpu, VBAR_EL2)	= read_sysreg_el1(SYS_VBAR);
-	__vcpu_sys_reg(vcpu, CONTEXTIDR_EL2) = read_sysreg_el1(SYS_CONTEXTIDR);
-	__vcpu_sys_reg(vcpu, AMAIR_EL2)	= read_sysreg_el1(SYS_AMAIR);
+	__vcpu_assign_sys_reg(vcpu, PAR_EL1,	 read_sysreg(par_el1));
+	__vcpu_assign_sys_reg(vcpu, TPIDR_EL1,	 read_sysreg(tpidr_el1));
+
+	__vcpu_assign_sys_reg(vcpu, ESR_EL2,	 read_sysreg_el1(SYS_ESR));
+	__vcpu_assign_sys_reg(vcpu, AFSR0_EL2,	 read_sysreg_el1(SYS_AFSR0));
+	__vcpu_assign_sys_reg(vcpu, AFSR1_EL2,	 read_sysreg_el1(SYS_AFSR1));
+	__vcpu_assign_sys_reg(vcpu, FAR_EL2,	 read_sysreg_el1(SYS_FAR));
+	__vcpu_assign_sys_reg(vcpu, MAIR_EL2,	 read_sysreg_el1(SYS_MAIR));
+	__vcpu_assign_sys_reg(vcpu, VBAR_EL2,	 read_sysreg_el1(SYS_VBAR));
+	__vcpu_assign_sys_reg(vcpu, CONTEXTIDR_EL2, read_sysreg_el1(SYS_CONTEXTIDR));
+	__vcpu_assign_sys_reg(vcpu, AMAIR_EL2,	 read_sysreg_el1(SYS_AMAIR));
 
 	/*
 	 * In VHE mode those registers are compatible between EL1 and EL2,
@@ -46,21 +46,21 @@ static void __sysreg_save_vel2_state(struct kvm_vcpu *vcpu)
 		 * are always trapped, ensuring that the in-memory
 		 * copy is always up-to-date. A small blessing...
 		 */
-		__vcpu_sys_reg(vcpu, SCTLR_EL2)	= read_sysreg_el1(SYS_SCTLR);
-		__vcpu_sys_reg(vcpu, TTBR0_EL2)	= read_sysreg_el1(SYS_TTBR0);
-		__vcpu_sys_reg(vcpu, TTBR1_EL2)	= read_sysreg_el1(SYS_TTBR1);
-		__vcpu_sys_reg(vcpu, TCR_EL2)	= read_sysreg_el1(SYS_TCR);
+		__vcpu_assign_sys_reg(vcpu, SCTLR_EL2,	 read_sysreg_el1(SYS_SCTLR));
+		__vcpu_assign_sys_reg(vcpu, TTBR0_EL2,	 read_sysreg_el1(SYS_TTBR0));
+		__vcpu_assign_sys_reg(vcpu, TTBR1_EL2,	 read_sysreg_el1(SYS_TTBR1));
+		__vcpu_assign_sys_reg(vcpu, TCR_EL2,	 read_sysreg_el1(SYS_TCR));
 
 		if (ctxt_has_tcrx(&vcpu->arch.ctxt)) {
-			__vcpu_sys_reg(vcpu, TCR2_EL2) = read_sysreg_el1(SYS_TCR2);
+			__vcpu_assign_sys_reg(vcpu, TCR2_EL2, read_sysreg_el1(SYS_TCR2));
 
 			if (ctxt_has_s1pie(&vcpu->arch.ctxt)) {
-				__vcpu_sys_reg(vcpu, PIRE0_EL2) = read_sysreg_el1(SYS_PIRE0);
-				__vcpu_sys_reg(vcpu, PIR_EL2) = read_sysreg_el1(SYS_PIR);
+				__vcpu_assign_sys_reg(vcpu, PIRE0_EL2, read_sysreg_el1(SYS_PIRE0));
+				__vcpu_assign_sys_reg(vcpu, PIR_EL2, read_sysreg_el1(SYS_PIR));
 			}
 
 			if (ctxt_has_s1poe(&vcpu->arch.ctxt))
-				__vcpu_sys_reg(vcpu, POR_EL2) = read_sysreg_el1(SYS_POR);
+				__vcpu_assign_sys_reg(vcpu, POR_EL2, read_sysreg_el1(SYS_POR));
 		}
 
 		/*
@@ -70,13 +70,13 @@ static void __sysreg_save_vel2_state(struct kvm_vcpu *vcpu)
 		 */
 		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_sys_reg(vcpu, SP_EL2)	= read_sysreg(sp_el1);
-	__vcpu_sys_reg(vcpu, ELR_EL2)	= read_sysreg_el1(SYS_ELR);
-	__vcpu_sys_reg(vcpu, SPSR_EL2)	= read_sysreg_el1(SYS_SPSR);
+	__vcpu_assign_sys_reg(vcpu, SP_EL2,	 read_sysreg(sp_el1));
+	__vcpu_assign_sys_reg(vcpu, ELR_EL2,	 read_sysreg_el1(SYS_ELR));
+	__vcpu_assign_sys_reg(vcpu, SPSR_EL2,	 read_sysreg_el1(SYS_SPSR));
 }
 
 static void __sysreg_restore_vel2_state(struct kvm_vcpu *vcpu)

arch/arm64/kvm/nested.c

@@ -1757,7 +1757,7 @@ int kvm_init_nv_sysregs(struct kvm_vcpu *vcpu)
 
 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

@@ -178,7 +178,7 @@ static void kvm_pmu_set_pmc_value(struct kvm_pmc *pmc, u64 val, bool force)
 		val |= lower_32_bits(val);
 	}
 
-	__vcpu_sys_reg(vcpu, reg) = val;
+	__vcpu_assign_sys_reg(vcpu, reg, val);
 
 	/* Recreate the perf event to reflect the updated sample_period */
 	kvm_pmu_create_perf_event(pmc);
@@ -204,7 +204,7 @@ void kvm_pmu_set_counter_value(struct kvm_vcpu *vcpu, u64 select_idx, u64 val)
 void kvm_pmu_set_counter_value_user(struct kvm_vcpu *vcpu, u64 select_idx, u64 val)
 {
 	kvm_pmu_release_perf_event(kvm_vcpu_idx_to_pmc(vcpu, select_idx));
-	__vcpu_sys_reg(vcpu, counter_index_to_reg(select_idx)) = val;
+	__vcpu_assign_sys_reg(vcpu, counter_index_to_reg(select_idx), val);
 	kvm_make_request(KVM_REQ_RELOAD_PMU, vcpu);
 }
 
@@ -239,7 +239,7 @@ static void kvm_pmu_stop_counter(struct kvm_pmc *pmc)
 
 	reg = counter_index_to_reg(pmc->idx);
 
-	__vcpu_sys_reg(vcpu, reg) = val;
+	__vcpu_assign_sys_reg(vcpu, reg, val);
 
 	kvm_pmu_release_perf_event(pmc);
 }
@@ -503,14 +503,14 @@ static void kvm_pmu_counter_increment(struct kvm_vcpu *vcpu,
 		reg = __vcpu_sys_reg(vcpu, counter_index_to_reg(i)) + 1;
 		if (!kvm_pmc_is_64bit(pmc))
 			reg = lower_32_bits(reg);
-		__vcpu_sys_reg(vcpu, counter_index_to_reg(i)) = reg;
+		__vcpu_assign_sys_reg(vcpu, counter_index_to_reg(i), reg);
 
 		/* No overflow? move on */
 		if (kvm_pmc_has_64bit_overflow(pmc) ? reg : lower_32_bits(reg))
 			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),
@@ -556,7 +556,7 @@ static void kvm_pmu_perf_overflow(struct perf_event *perf_event,
 	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),
@@ -602,7 +602,7 @@ void kvm_pmu_handle_pmcr(struct kvm_vcpu *vcpu, u64 val)
 		kvm_make_request(KVM_REQ_RELOAD_PMU, vcpu);
 
 	/* The reset bits don't indicate any state, and shouldn't be saved. */
-	__vcpu_sys_reg(vcpu, PMCR_EL0) = val & ~(ARMV8_PMU_PMCR_C | ARMV8_PMU_PMCR_P);
+	__vcpu_assign_sys_reg(vcpu, PMCR_EL0, (val & ~(ARMV8_PMU_PMCR_C | ARMV8_PMU_PMCR_P)));
 
 	if (val & ARMV8_PMU_PMCR_C)
 		kvm_pmu_set_counter_value(vcpu, ARMV8_PMU_CYCLE_IDX, 0);
@@ -779,7 +779,7 @@ void kvm_pmu_set_counter_event_type(struct kvm_vcpu *vcpu, u64 data,
 	u64 reg;
 
 	reg = counter_index_to_evtreg(pmc->idx);
-	__vcpu_sys_reg(vcpu, reg) = data & kvm_pmu_evtyper_mask(vcpu->kvm);
+	__vcpu_assign_sys_reg(vcpu, reg, (data & kvm_pmu_evtyper_mask(vcpu->kvm)));
 
 	kvm_pmu_create_perf_event(pmc);
 }
@@ -914,9 +914,9 @@ void kvm_vcpu_reload_pmu(struct kvm_vcpu *vcpu)
 {
 	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);
 }
@@ -1038,7 +1038,7 @@ static void kvm_arm_set_nr_counters(struct kvm *kvm, unsigned int nr)
 			u64 val = __vcpu_sys_reg(vcpu, MDCR_EL2);
 			val &= ~MDCR_EL2_HPMN;
 			val |= FIELD_PREP(MDCR_EL2_HPMN, kvm->arch.nr_pmu_counters);
-			__vcpu_sys_reg(vcpu, MDCR_EL2) = val;
+			__vcpu_assign_sys_reg(vcpu, MDCR_EL2, val);
 		}
 	}
 }