Merge branch 'for-next/timers' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux into for-next/core

Conflicts:
	arch/arm64/Kconfig
	arch/arm64/include/asm/arch_timer.h

12 files changed, 185 insertions, 109 deletions

diff --git a/Documentation/arm64/silicon-errata.txt b/Documentation/arm64/silicon-errata.txt
index d1e2bb801e1b..d5a124d7e242 100644
--- a/Documentation/arm64/silicon-errata.txt
+++ b/Documentation/arm64/silicon-errata.txt
@@ -61,6 +61,7 @@ stable kernels.
 | ARM            | Cortex-A76      | #1188873        | ARM64_ERRATUM_1188873       |
 | ARM            | Cortex-A76      | #1165522        | ARM64_ERRATUM_1165522       |
 | ARM            | Cortex-A76      | #1286807        | ARM64_ERRATUM_1286807       |
+| ARM            | Neoverse-N1     | #1188873        | ARM64_ERRATUM_1188873       |
 | ARM            | MMU-500         | #841119,#826419 | N/A                         |
 |                |                 |                 |                             |
 | Cavium         | ThunderX ITS    | #22375, #24313  | CAVIUM_ERRATUM_22375        |
diff --git a/arch/arm/include/asm/arch_timer.h b/arch/arm/include/asm/arch_timer.h
index 0a8d7bba2cb0..4b66ecd6be99 100644
--- a/arch/arm/include/asm/arch_timer.h
+++ b/arch/arm/include/asm/arch_timer.h
@@ -11,6 +11,10 @@
 #include <clocksource/arm_arch_timer.h>
 
 #ifdef CONFIG_ARM_ARCH_TIMER
+/* 32bit ARM doesn't know anything about timer errata... */
+#define has_erratum_handler(h)		(false)
+#define erratum_handler(h)		(arch_timer_##h)
+
 int arch_timer_arch_init(void);
 
 /*
@@ -79,7 +83,7 @@ static inline u32 arch_timer_get_cntfrq(void)
 	return val;
 }
 
-static inline u64 arch_counter_get_cntpct(void)
+static inline u64 __arch_counter_get_cntpct(void)
 {
 	u64 cval;
 
@@ -88,7 +92,12 @@ static inline u64 arch_counter_get_cntpct(void)
 	return cval;
 }
 
-static inline u64 arch_counter_get_cntvct(void)
+static inline u64 __arch_counter_get_cntpct_stable(void)
+{
+	return __arch_counter_get_cntpct();
+}
+
+static inline u64 __arch_counter_get_cntvct(void)
 {
 	u64 cval;
 
@@ -97,6 +106,11 @@ static inline u64 arch_counter_get_cntvct(void)
 	return cval;
 }
 
+static inline u64 __arch_counter_get_cntvct_stable(void)
+{
+	return __arch_counter_get_cntvct();
+}
+
 static inline u32 arch_timer_get_cntkctl(void)
 {
 	u32 cntkctl;
diff --git a/arch/arm/include/asm/cp15.h b/arch/arm/include/asm/cp15.h
index 07e27f212dc7..d2453e2d3f1f 100644
--- a/arch/arm/include/asm/cp15.h
+++ b/arch/arm/include/asm/cp15.h
@@ -68,6 +68,8 @@
 #define BPIALL				__ACCESS_CP15(c7, 0, c5, 6)
 #define ICIALLU				__ACCESS_CP15(c7, 0, c5, 0)
 
+#define CNTVCT				__ACCESS_CP15_64(1, c14)
+
 extern unsigned long cr_alignment;	/* defined in entry-armv.S */
 
 static inline unsigned long get_cr(void)
diff --git a/arch/arm/vdso/vgettimeofday.c b/arch/arm/vdso/vgettimeofday.c
index a9dd619c6c29..7bdbf5d5c47d 100644
--- a/arch/arm/vdso/vgettimeofday.c
+++ b/arch/arm/vdso/vgettimeofday.c
@@ -18,9 +18,9 @@
 #include <linux/compiler.h>
 #include <linux/hrtimer.h>
 #include <linux/time.h>
-#include <asm/arch_timer.h>
 #include <asm/barrier.h>
 #include <asm/bug.h>
+#include <asm/cp15.h>
 #include <asm/page.h>
 #include <asm/unistd.h>
 #include <asm/vdso_datapage.h>
@@ -123,7 +123,8 @@ static notrace u64 get_ns(struct vdso_data *vdata)
 	u64 cycle_now;
 	u64 nsec;
 
-	cycle_now = arch_counter_get_cntvct();
+	isb();
+	cycle_now = read_sysreg(CNTVCT);
 
 	cycle_delta = (cycle_now - vdata->cs_cycle_last) & vdata->cs_mask;
 
diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index b5c6943225ce..1c0cb5131c2a 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -477,15 +477,17 @@ config ARM64_ERRATUM_1024718
 	  If unsure, say Y.
 
 config ARM64_ERRATUM_1188873
-	bool "Cortex-A76: MRC read following MRRC read of specific Generic Timer in AArch32 might give incorrect result"
+	bool "Cortex-A76/Neoverse-N1: MRC read following MRRC read of specific Generic Timer in AArch32 might give incorrect result"
 	default y
+	depends on COMPAT
 	select ARM_ARCH_TIMER_OOL_WORKAROUND
 	help
-	  This option adds a workaround for ARM Cortex-A76 erratum 1188873.
+	  This option adds a workaround for ARM Cortex-A76/Neoverse-N1
+	  erratum 1188873.
 
-	  Affected Cortex-A76 cores (r0p0, r1p0, r2p0) could cause
-	  register corruption when accessing the timer registers from
-	  AArch32 userspace.
+	  Affected Cortex-A76/Neoverse-N1 cores (r0p0, r1p0, r2p0) could
+	  cause register corruption when accessing the timer registers
+	  from AArch32 userspace.
 
 	  If unsure, say Y.
 
diff --git a/arch/arm64/include/asm/arch_timer.h b/arch/arm64/include/asm/arch_timer.h
index 93e07512b4b6..b7bca1ae09e6 100644
--- a/arch/arm64/include/asm/arch_timer.h
+++ b/arch/arm64/include/asm/arch_timer.h
@@ -31,11 +31,23 @@
 #include <clocksource/arm_arch_timer.h>
 
 #if IS_ENABLED(CONFIG_ARM_ARCH_TIMER_OOL_WORKAROUND)
-extern struct static_key_false arch_timer_read_ool_enabled;
-#define needs_unstable_timer_counter_workaround() \
-	static_branch_unlikely(&arch_timer_read_ool_enabled)
+#define has_erratum_handler(h)						\
+	({								\
+		const struct arch_timer_erratum_workaround *__wa;	\
+		__wa = __this_cpu_read(timer_unstable_counter_workaround); \
+		(__wa && __wa->h);					\
+	})
+
+#define erratum_handler(h)						\
+	({								\
+		const struct arch_timer_erratum_workaround *__wa;	\
+		__wa = __this_cpu_read(timer_unstable_counter_workaround); \
+		(__wa && __wa->h) ? __wa->h : arch_timer_##h;		\
+	})
+
 #else
-#define needs_unstable_timer_counter_workaround()  false
+#define has_erratum_handler(h)			   false
+#define erratum_handler(h)			   (arch_timer_##h)
 #endif
 
 enum arch_timer_erratum_match_type {
@@ -61,23 +73,37 @@ struct arch_timer_erratum_workaround {
 DECLARE_PER_CPU(const struct arch_timer_erratum_workaround *,
 		timer_unstable_counter_workaround);
 
+/* inline sysreg accessors that make erratum_handler() work */
+static inline notrace u32 arch_timer_read_cntp_tval_el0(void)
+{
+	return read_sysreg(cntp_tval_el0);
+}
+
+static inline notrace u32 arch_timer_read_cntv_tval_el0(void)
+{
+	return read_sysreg(cntv_tval_el0);
+}
+
+static inline notrace u64 arch_timer_read_cntpct_el0(void)
+{
+	return read_sysreg(cntpct_el0);
+}
+
+static inline notrace u64 arch_timer_read_cntvct_el0(void)
+{
+	return read_sysreg(cntvct_el0);
+}
+
 #define arch_timer_reg_read_stable(reg)					\
-({									\
-	u64 _val;							\
-	if (needs_unstable_timer_counter_workaround()) {		\
-		const struct arch_timer_erratum_workaround *wa;		\
+	({								\
+		u64 _val;						\
+									\
 		preempt_disable_notrace();				\
-		wa = __this_cpu_read(timer_unstable_counter_workaround); \
-		if (wa && wa->read_##reg)				\
-			_val = wa->read_##reg();			\
-		else							\
-			_val = read_sysreg(reg);			\
+		_val = erratum_handler(read_ ## reg)();			\
 		preempt_enable_notrace();				\
-	} else {							\
-		_val = read_sysreg(reg);				\
-	}								\
-	_val;								\
-})
+									\
+		_val;							\
+	})
 
 /*
  * These register accessors are marked inline so the compiler can
@@ -167,7 +193,7 @@ static inline void arch_timer_set_cntkctl(u32 cntkctl)
 	: "=r" (tmp) : "r" (_val));					\
 } while (0)
 
-static inline u64 arch_counter_get_cntpct(void)
+static inline u64 __arch_counter_get_cntpct_stable(void)
 {
 	u64 cnt;
 
@@ -177,7 +203,17 @@ static inline u64 arch_counter_get_cntpct(void)
 	return cnt;
 }
 
-static inline u64 arch_counter_get_cntvct(void)
+static inline u64 __arch_counter_get_cntpct(void)
+{
+	u64 cnt;
+
+	isb();
+	cnt = read_sysreg(cntpct_el0);
+	arch_counter_enforce_ordering(cnt);
+	return cnt;
+}
+
+static inline u64 __arch_counter_get_cntvct_stable(void)
 {
 	u64 cnt;
 
@@ -187,6 +223,16 @@ static inline u64 arch_counter_get_cntvct(void)
 	return cnt;
 }
 
+static inline u64 __arch_counter_get_cntvct(void)
+{
+	u64 cnt;
+
+	isb();
+	cnt = read_sysreg(cntvct_el0);
+	arch_counter_enforce_ordering(cnt);
+	return cnt;
+}
+
 #undef arch_counter_enforce_ordering
 
 static inline int arch_timer_arch_init(void)
diff --git a/arch/arm64/include/asm/cputype.h b/arch/arm64/include/asm/cputype.h
index 5f1437099b99..2602bae334fb 100644
--- a/arch/arm64/include/asm/cputype.h
+++ b/arch/arm64/include/asm/cputype.h
@@ -89,6 +89,7 @@
 #define ARM_CPU_PART_CORTEX_A35		0xD04
 #define ARM_CPU_PART_CORTEX_A55		0xD05
 #define ARM_CPU_PART_CORTEX_A76		0xD0B
+#define ARM_CPU_PART_NEOVERSE_N1	0xD0C
 
 #define APM_CPU_PART_POTENZA		0x000
 
@@ -118,6 +119,7 @@
 #define MIDR_CORTEX_A35 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A35)
 #define MIDR_CORTEX_A55 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A55)
 #define MIDR_CORTEX_A76	MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_CORTEX_A76)
+#define MIDR_NEOVERSE_N1 MIDR_CPU_MODEL(ARM_CPU_IMP_ARM, ARM_CPU_PART_NEOVERSE_N1)
 #define MIDR_THUNDERX	MIDR_CPU_MODEL(ARM_CPU_IMP_CAVIUM, CAVIUM_CPU_PART_THUNDERX)
 #define MIDR_THUNDERX_81XX MIDR_CPU_MODEL(ARM_CPU_IMP_CAVIUM, CAVIUM_CPU_PART_THUNDERX_81XX)
 #define MIDR_THUNDERX_83XX MIDR_CPU_MODEL(ARM_CPU_IMP_CAVIUM, CAVIUM_CPU_PART_THUNDERX_83XX)
diff --git a/arch/arm64/kernel/cpu_errata.c b/arch/arm64/kernel/cpu_errata.c
index 1b9ce0fdd81d..e88d4e7bdfc7 100644
--- a/arch/arm64/kernel/cpu_errata.c
+++ b/arch/arm64/kernel/cpu_errata.c
@@ -682,6 +682,16 @@ static const struct midr_range workaround_clean_cache[] = {
 };
 #endif
 
+#ifdef CONFIG_ARM64_ERRATUM_1188873
+static const struct midr_range erratum_1188873_list[] = {
+	/* Cortex-A76 r0p0 to r2p0 */
+	MIDR_RANGE(MIDR_CORTEX_A76, 0, 0, 2, 0),
+	/* Neoverse-N1 r0p0 to r2p0 */
+	MIDR_RANGE(MIDR_NEOVERSE_N1, 0, 0, 2, 0),
+	{},
+};
+#endif
+
 const struct arm64_cpu_capabilities arm64_errata[] = {
 #ifdef CONFIG_ARM64_WORKAROUND_CLEAN_CACHE
 	{
@@ -801,10 +811,9 @@ const struct arm64_cpu_capabilities arm64_errata[] = {
 	},
 #ifdef CONFIG_ARM64_ERRATUM_1188873
 	{
-		/* Cortex-A76 r0p0 to r2p0 */
 		.desc = "ARM erratum 1188873",
 		.capability = ARM64_WORKAROUND_1188873,
-		ERRATA_MIDR_RANGE(MIDR_CORTEX_A76, 0, 0, 2, 0),
+		ERRATA_MIDR_RANGE_LIST(erratum_1188873_list),
 	},
 #endif
 #ifdef CONFIG_ARM64_ERRATUM_1165522
diff --git a/arch/arm64/kernel/entry.S b/arch/arm64/kernel/entry.S
index c50a7a75f2e0..1a7811b7e3c4 100644
--- a/arch/arm64/kernel/entry.S
+++ b/arch/arm64/kernel/entry.S
@@ -336,6 +336,21 @@ alternative_if ARM64_WORKAROUND_845719
 alternative_else_nop_endif
 #endif
 3:
+#ifdef CONFIG_ARM64_ERRATUM_1188873
+alternative_if_not ARM64_WORKAROUND_1188873
+	b	4f
+alternative_else_nop_endif
+	/*
+	 * if (x22.mode32 == cntkctl_el1.el0vcten)
+	 *     cntkctl_el1.el0vcten = ~cntkctl_el1.el0vcten
+	 */
+	mrs	x1, cntkctl_el1
+	eon	x0, x1, x22, lsr #3
+	tbz	x0, #1, 4f
+	eor	x1, x1, #2	// ARCH_TIMER_USR_VCT_ACCESS_EN
+	msr	cntkctl_el1, x1
+4:
+#endif
 	apply_ssbd 0, x0, x1
 	.endif
 
@@ -362,11 +377,11 @@ alternative_else_nop_endif
 	.if	\el == 0
 alternative_insn eret, nop, ARM64_UNMAP_KERNEL_AT_EL0
 #ifdef CONFIG_UNMAP_KERNEL_AT_EL0
-	bne	4f
+	bne	5f
 	msr	far_el1, x30
 	tramp_alias	x30, tramp_exit_native
 	br	x30
-4:
+5:
 	tramp_alias	x30, tramp_exit_compat
 	br	x30
 #endif
diff --git a/arch/arm64/kernel/traps.c b/arch/arm64/kernel/traps.c
index 21e73954762c..48432027969c 100644
--- a/arch/arm64/kernel/traps.c
+++ b/arch/arm64/kernel/traps.c
@@ -496,7 +496,7 @@ static void cntvct_read_handler(unsigned int esr, struct pt_regs *regs)
 {
 	int rt = ESR_ELx_SYS64_ISS_RT(esr);
 
-	pt_regs_write_reg(regs, rt, arch_counter_get_cntvct());
+	pt_regs_write_reg(regs, rt, arch_timer_read_counter());
 	arm64_skip_faulting_instruction(regs, AARCH64_INSN_SIZE);
 }
 
@@ -668,7 +668,7 @@ static void compat_cntvct_read_handler(unsigned int esr, struct pt_regs *regs)
 {
 	int rt = (esr & ESR_ELx_CP15_64_ISS_RT_MASK) >> ESR_ELx_CP15_64_ISS_RT_SHIFT;
 	int rt2 = (esr & ESR_ELx_CP15_64_ISS_RT2_MASK) >> ESR_ELx_CP15_64_ISS_RT2_SHIFT;
-	u64 val = arch_counter_get_cntvct();
+	u64 val = arch_timer_read_counter();
 
 	pt_regs_write_reg(regs, rt, lower_32_bits(val));
 	pt_regs_write_reg(regs, rt2, upper_32_bits(val));
diff --git a/drivers/clocksource/arm_arch_timer.c b/drivers/clocksource/arm_arch_timer.c
index 6cc8aff83805..9a2f23df7610 100644
--- a/drivers/clocksource/arm_arch_timer.c
+++ b/drivers/clocksource/arm_arch_timer.c
@@ -152,6 +152,26 @@ u32 arch_timer_reg_read(int access, enum arch_timer_reg reg,
 	return val;
 }
 
+static u64 arch_counter_get_cntpct_stable(void)
+{
+	return __arch_counter_get_cntpct_stable();
+}
+
+static u64 arch_counter_get_cntpct(void)
+{
+	return __arch_counter_get_cntpct();
+}
+
+static u64 arch_counter_get_cntvct_stable(void)
+{
+	return __arch_counter_get_cntvct_stable();
+}
+
+static u64 arch_counter_get_cntvct(void)
+{
+	return __arch_counter_get_cntvct();
+}
+
 /*
  * Default to cp15 based access because arm64 uses this function for
  * sched_clock() before DT is probed and the cp15 method is guaranteed
@@ -319,13 +339,6 @@ static u64 notrace arm64_858921_read_cntvct_el0(void)
 }
 #endif
 
-#ifdef CONFIG_ARM64_ERRATUM_1188873
-static u64 notrace arm64_1188873_read_cntvct_el0(void)
-{
-	return read_sysreg(cntvct_el0);
-}
-#endif
-
 #ifdef CONFIG_SUN50I_ERRATUM_UNKNOWN1
 /*
  * The low bits of the counter registers are indeterminate while bit 10 or
@@ -372,8 +385,7 @@ static u32 notrace sun50i_a64_read_cntv_tval_el0(void)
 DEFINE_PER_CPU(const struct arch_timer_erratum_workaround *, timer_unstable_counter_workaround);
 EXPORT_SYMBOL_GPL(timer_unstable_counter_workaround);
 
-DEFINE_STATIC_KEY_FALSE(arch_timer_read_ool_enabled);
-EXPORT_SYMBOL_GPL(arch_timer_read_ool_enabled);
+static atomic_t timer_unstable_counter_workaround_in_use = ATOMIC_INIT(0);
 
 static void erratum_set_next_event_tval_generic(const int access, unsigned long evt,
 						struct clock_event_device *clk)
@@ -457,14 +469,6 @@ static const struct arch_timer_erratum_workaround ool_workarounds[] = {
 		.read_cntvct_el0 = arm64_858921_read_cntvct_el0,
 	},
 #endif
-#ifdef CONFIG_ARM64_ERRATUM_1188873
-	{
-		.match_type = ate_match_local_cap_id,
-		.id = (void *)ARM64_WORKAROUND_1188873,
-		.desc = "ARM erratum 1188873",
-		.read_cntvct_el0 = arm64_1188873_read_cntvct_el0,
-	},
-#endif
 #ifdef CONFIG_SUN50I_ERRATUM_UNKNOWN1
 	{
 		.match_type = ate_match_dt,
@@ -552,11 +556,8 @@ void arch_timer_enable_workaround(const struct arch_timer_erratum_workaround *wa
 			per_cpu(timer_unstable_counter_workaround, i) = wa;
 	}
 
-	/*
-	 * Use the locked version, as we're called from the CPU
-	 * hotplug framework. Otherwise, we end-up in deadlock-land.
-	 */
-	static_branch_enable_cpuslocked(&arch_timer_read_ool_enabled);
+	if (wa->read_cntvct_el0 || wa->read_cntpct_el0)
+		atomic_set(&timer_unstable_counter_workaround_in_use, 1);
 
 	/*
 	 * Don't use the vdso fastpath if errata require using the
@@ -573,7 +574,7 @@ void arch_timer_enable_workaround(const struct arch_timer_erratum_workaround *wa
 static void arch_timer_check_ool_workaround(enum arch_timer_erratum_match_type type,
 					    void *arg)
 {
-	const struct arch_timer_erratum_workaround *wa;
+	const struct arch_timer_erratum_workaround *wa, *__wa;
 	ate_match_fn_t match_fn = NULL;
 	bool local = false;
 
@@ -597,53 +598,32 @@ static void arch_timer_check_ool_workaround(enum arch_timer_erratum_match_type t
 	if (!wa)
 		return;
 
-	if (needs_unstable_timer_counter_workaround()) {
-		const struct arch_timer_erratum_workaround *__wa;
-		__wa = __this_cpu_read(timer_unstable_counter_workaround);
-		if (__wa && wa != __wa)
-			pr_warn("Can't enable workaround for %s (clashes with %s\n)",
-				wa->desc, __wa->desc);
+	__wa = __this_cpu_read(timer_unstable_counter_workaround);
+	if (__wa && wa != __wa)
+		pr_warn("Can't enable workaround for %s (clashes with %s\n)",
+			wa->desc, __wa->desc);
 
-		if (__wa)
-			return;
-	}
+	if (__wa)
+		return;
 
 	arch_timer_enable_workaround(wa, local);
 	pr_info("Enabling %s workaround for %s\n",
 		local ? "local" : "global", wa->desc);
 }
 
-#define erratum_handler(fn, r, ...)					\
-({									\
-	bool __val;							\
-	if (needs_unstable_timer_counter_workaround()) {		\
-		const struct arch_timer_erratum_workaround *__wa;	\
-		__wa = __this_cpu_read(timer_unstable_counter_workaround); \
-		if (__wa && __wa->fn) {					\
-			r = __wa->fn(__VA_ARGS__);			\
-			__val = true;					\
-		} else {						\
-			__val = false;					\
-		}							\
-	} else {							\
-		__val = false;						\
-	}								\
-	__val;								\
-})
-
 static bool arch_timer_this_cpu_has_cntvct_wa(void)
 {
-	const struct arch_timer_erratum_workaround *wa;
+	return has_erratum_handler(read_cntvct_el0);
+}
 
-	wa = __this_cpu_read(timer_unstable_counter_workaround);
-	return wa && wa->read_cntvct_el0;
+static bool arch_timer_counter_has_wa(void)
+{
+	return atomic_read(&timer_unstable_counter_workaround_in_use);
 }
 #else
 #define arch_timer_check_ool_workaround(t,a)		do { } while(0)
-#define erratum_set_next_event_tval_virt(...)		({BUG(); 0;})
-#define erratum_set_next_event_tval_phys(...)		({BUG(); 0;})
-#define erratum_handler(fn, r, ...)			({false;})
 #define arch_timer_this_cpu_has_cntvct_wa()		({false;})
+#define arch_timer_counter_has_wa()			({false;})
 #endif /* CONFIG_ARM_ARCH_TIMER_OOL_WORKAROUND */
 
 static __always_inline irqreturn_t timer_handler(const int access,
@@ -736,11 +716,6 @@ static __always_inline void set_next_event(const int access, unsigned long evt,
 static int arch_timer_set_next_event_virt(unsigned long evt,
 					  struct clock_event_device *clk)
 {
-	int ret;
-
-	if (erratum_handler(set_next_event_virt, ret, evt, clk))
-		return ret;
-
 	set_next_event(ARCH_TIMER_VIRT_ACCESS, evt, clk);
 	return 0;
 }
@@ -748,11 +723,6 @@ static int arch_timer_set_next_event_virt(unsigned long evt,
 static int arch_timer_set_next_event_phys(unsigned long evt,
 					  struct clock_event_device *clk)
 {
-	int ret;
-
-	if (erratum_handler(set_next_event_phys, ret, evt, clk))
-		return ret;
-
 	set_next_event(ARCH_TIMER_PHYS_ACCESS, evt, clk);
 	return 0;
 }
@@ -777,6 +747,10 @@ static void __arch_timer_setup(unsigned type,
 	clk->features = CLOCK_EVT_FEAT_ONESHOT;
 
 	if (type == ARCH_TIMER_TYPE_CP15) {
+		typeof(clk->set_next_event) sne;
+
+		arch_timer_check_ool_workaround(ate_match_local_cap_id, NULL);
+
 		if (arch_timer_c3stop)
 			clk->features |= CLOCK_EVT_FEAT_C3STOP;
 		clk->name = "arch_sys_timer";
@@ -787,20 +761,20 @@ static void __arch_timer_setup(unsigned type,
 		case ARCH_TIMER_VIRT_PPI:
 			clk->set_state_shutdown = arch_timer_shutdown_virt;
 			clk->set_state_oneshot_stopped = arch_timer_shutdown_virt;
-			clk->set_next_event = arch_timer_set_next_event_virt;
+			sne = erratum_handler(set_next_event_virt);
 			break;
 		case ARCH_TIMER_PHYS_SECURE_PPI:
 		case ARCH_TIMER_PHYS_NONSECURE_PPI:
 		case ARCH_TIMER_HYP_PPI:
 			clk->set_state_shutdown = arch_timer_shutdown_phys;
 			clk->set_state_oneshot_stopped = arch_timer_shutdown_phys;
-			clk->set_next_event = arch_timer_set_next_event_phys;
+			sne = erratum_handler(set_next_event_phys);
 			break;
 		default:
 			BUG();
 		}
 
-		arch_timer_check_ool_workaround(ate_match_local_cap_id, NULL);
+		clk->set_next_event = sne;
 	} else {
 		clk->features |= CLOCK_EVT_FEAT_DYNIRQ;
 		clk->name = "arch_mem_timer";
@@ -1002,12 +976,22 @@ static void __init arch_counter_register(unsigned type)
 
 	/* Register the CP15 based counter if we have one */
 	if (type & ARCH_TIMER_TYPE_CP15) {
+		u64 (*rd)(void);
+
 		if ((IS_ENABLED(CONFIG_ARM64) && !is_hyp_mode_available()) ||
-		    arch_timer_uses_ppi == ARCH_TIMER_VIRT_PPI)
-			arch_timer_read_counter = arch_counter_get_cntvct;
-		else
-			arch_timer_read_counter = arch_counter_get_cntpct;
+		    arch_timer_uses_ppi == ARCH_TIMER_VIRT_PPI) {
+			if (arch_timer_counter_has_wa())
+				rd = arch_counter_get_cntvct_stable;
+			else
+				rd = arch_counter_get_cntvct;
+		} else {
+			if (arch_timer_counter_has_wa())
+				rd = arch_counter_get_cntpct_stable;
+			else
+				rd = arch_counter_get_cntpct;
+		}
 
+		arch_timer_read_counter = rd;
 		clocksource_counter.archdata.vdso_direct = vdso_default;
 	} else {
 		arch_timer_read_counter = arch_counter_get_cntvct_mem;
diff --git a/drivers/watchdog/sbsa_gwdt.c b/drivers/watchdog/sbsa_gwdt.c
index e8bd9887c566..e221e47396ab 100644
--- a/drivers/watchdog/sbsa_gwdt.c
+++ b/drivers/watchdog/sbsa_gwdt.c
@@ -161,7 +161,7 @@ static unsigned int sbsa_gwdt_get_timeleft(struct watchdog_device *wdd)
 		timeleft += readl(gwdt->control_base + SBSA_GWDT_WOR);
 
 	timeleft += lo_hi_readq(gwdt->control_base + SBSA_GWDT_WCV) -
-		    arch_counter_get_cntvct();
+		    arch_timer_read_counter();
 
 	do_div(timeleft, gwdt->clk);