[PATCH v2 0/2] arch_timers patches to enable KVM support

@@ -27,13 +27,31 @@
 #include <asm/sched_clock.h>
 
 static unsigned long arch_timer_rate;
-static int arch_timer_ppi;
-static int arch_timer_ppi2;
+
+enum ppi_nr {
+	PHYS_SECURE_PPI,
+	PHYS_NONSECURE_PPI,
+	VIRT_PPI,
+	HYP_PPI,
+	MAX_TIMER_PPI
+};
+
+static int arch_timer_ppi[MAX_TIMER_PPI];
 
 static struct clock_event_device __percpu **arch_timer_evt;
 
 extern void init_current_timer_delay(unsigned long freq);
 
+static bool arch_timer_use_virtual = true;
+
+struct arch_timer_reg_ops {
+	void	(*reg_write)(int reg, u32 val);
+	u32	(*reg_read)(int reg);
+	cycle_t	(*counter_read)(void);
+};
+
+static struct arch_timer_reg_ops *arch_timer_reg_ops;
+
 /*
  * Architected system timer support.
  */

@@ -46,7 +64,37 @@ extern void init_current_timer_delay(unsigned long freq);
 #define ARCH_TIMER_REG_FREQ		1
 #define ARCH_TIMER_REG_TVAL		2
 
-static void arch_timer_reg_write(int reg, u32 val)
+static inline void arch_timer_reg_write(int reg, u32 val)
+{
+	arch_timer_reg_ops->reg_write(reg, val);
+}
+
+static inline u32 arch_timer_reg_read(int reg)
+{
+	return arch_timer_reg_ops->reg_read(reg);
+}
+
+static inline cycle_t arch_timer_counter_read(void)
+{
+	return arch_timer_reg_ops->counter_read();
+}
+
+static u32 arch_timer_common_reg_read(int reg)
+{
+	u32 val;
+
+	switch (reg) {
+	case ARCH_TIMER_REG_FREQ:
+		asm volatile("mrc p15, 0, %0, c14, c0, 0" : "=r" (val));
+		break;
+	default:
+		BUG();
+	}
+
+	return val;
+}
+
+static void arch_timer_phys_reg_write(int reg, u32 val)
 {
 	switch (reg) {
 	case ARCH_TIMER_REG_CTRL:

@@ -60,7 +108,7 @@ static void arch_timer_reg_write(int reg, u32 val)
 	isb();
 }
 
-static u32 arch_timer_reg_read(int reg)
+static u32 arch_timer_phys_reg_read(int reg)
 {
 	u32 val;
 

@@ -68,19 +116,78 @@ static u32 arch_timer_reg_read(int reg)
 	case ARCH_TIMER_REG_CTRL:
 		asm volatile("mrc p15, 0, %0, c14, c2, 1" : "=r" (val));
 		break;
-	case ARCH_TIMER_REG_FREQ:
-		asm volatile("mrc p15, 0, %0, c14, c0, 0" : "=r" (val));
-		break;
 	case ARCH_TIMER_REG_TVAL:
 		asm volatile("mrc p15, 0, %0, c14, c2, 0" : "=r" (val));
 		break;
 	default:
-		BUG();
+		val = arch_timer_common_reg_read(reg);
+	}
+
+	return val;
+}
+
+static inline cycle_t arch_counter_get_cntpct(void)
+{
+	u32 cvall, cvalh;
+
+	asm volatile("mrrc p15, 0, %0, %1, c14" : "=r" (cvall), "=r" (cvalh));
+
+	return ((cycle_t) cvalh << 32) | cvall;
+}
+
+static struct arch_timer_reg_ops arch_timer_phys_ops = {
+	.reg_write	= arch_timer_phys_reg_write,
+	.reg_read	= arch_timer_phys_reg_read,
+	.counter_read	= arch_counter_get_cntpct,
+};
+
+static void arch_timer_virt_reg_write(int reg, u32 val)
+{
+	switch (reg) {
+	case ARCH_TIMER_REG_CTRL:
+		asm volatile("mcr p15, 0, %0, c14, c3, 1" : : "r" (val));
+		break;
+	case ARCH_TIMER_REG_TVAL:
+		asm volatile("mcr p15, 0, %0, c14, c3, 0" : : "r" (val));
+		break;
+	}
+
+	isb();
+}
+
+static u32 arch_timer_virt_reg_read(int reg)
+{
+	u32 val;
+
+	switch (reg) {
+	case ARCH_TIMER_REG_CTRL:
+		asm volatile("mrc p15, 0, %0, c14, c3, 1" : "=r" (val));
+		break;
+	case ARCH_TIMER_REG_TVAL:
+		asm volatile("mrc p15, 0, %0, c14, c3, 0" : "=r" (val));
+		break;
+	default:
+		val = arch_timer_common_reg_read(reg);
 	}
 
 	return val;
 }
 
+static inline cycle_t arch_counter_get_cntvct(void)
+{
+	u32 cvall, cvalh;
+
+	asm volatile("mrrc p15, 1, %0, %1, c14" : "=r" (cvall), "=r" (cvalh));
+
+	return ((cycle_t) cvalh << 32) | cvall;
+}
+
+static struct arch_timer_reg_ops arch_timer_virt_ops = {
+	.reg_write	= arch_timer_virt_reg_write,
+	.reg_read	= arch_timer_virt_reg_read,
+	.counter_read	= arch_counter_get_cntvct,
+};
+
 static irqreturn_t arch_timer_handler(int irq, void *dev_id)
 {
 	struct clock_event_device *evt = *(struct clock_event_device **)dev_id;

@@ -144,16 +251,19 @@ static int __cpuinit arch_timer_setup(struct clock_event_device *clk)
 	clk->rating = 450;
 	clk->set_mode = arch_timer_set_mode;
 	clk->set_next_event = arch_timer_set_next_event;
-	clk->irq = arch_timer_ppi;
+	clk->irq = arch_timer_ppi[PHYS_SECURE_PPI];
 
 	clockevents_config_and_register(clk, arch_timer_rate,
 					0xf, 0x7fffffff);
 
 	*__this_cpu_ptr(arch_timer_evt) = clk;
 
-	enable_percpu_irq(clk->irq, 0);
-	if (arch_timer_ppi2)
-		enable_percpu_irq(arch_timer_ppi2, 0);
+	if (arch_timer_use_virtual)
+		enable_percpu_irq(arch_timer_ppi[VIRT_PPI], 0);
+	else {
+		enable_percpu_irq(arch_timer_ppi[PHYS_SECURE_PPI], 0);
+		enable_percpu_irq(arch_timer_ppi[PHYS_NONSECURE_PPI], 0);
+	}
 
 	return 0;
 }

@@ -185,43 +295,30 @@ static int arch_timer_available(void)
 		arch_timer_rate = freq;
 	}
 
-	pr_info_once("Architected local timer running at %lu.%02luMHz.\n",
-		     arch_timer_rate / 1000000, (arch_timer_rate / 10000) % 100);
+	pr_info_once("Architected local timer running at %lu.%02luMHz (%s).\n",
+		     arch_timer_rate / 1000000, (arch_timer_rate / 10000) % 100,
+		     arch_timer_use_virtual ? "virt" : "phys");
 	return 0;
 }
 
-static inline cycle_t arch_counter_get_cntpct(void)
-{
-	u32 cvall, cvalh;
-
-	asm volatile("mrrc p15, 0, %0, %1, c14" : "=r" (cvall), "=r" (cvalh));
-
-	return ((cycle_t) cvalh << 32) | cvall;
-}
-
-static inline cycle_t arch_counter_get_cntvct(void)
+static u32 notrace arch_counter_get_cnt32(void)
 {
-	u32 cvall, cvalh;
-
-	asm volatile("mrrc p15, 1, %0, %1, c14" : "=r" (cvall), "=r" (cvalh));
-
-	return ((cycle_t) cvalh << 32) | cvall;
-}
-
-static u32 notrace arch_counter_get_cntvct32(void)
-{
-	cycle_t cntvct = arch_counter_get_cntvct();
+	cycle_t cnt = arch_timer_counter_read();
 
 	/*
 	 * The sched_clock infrastructure only knows about counters
 	 * with@most 32bits. Forget about the upper 24 bits for the
 	 * time being...
 	 */
-	return (u32)(cntvct & (u32)~0);
+	return (u32)(cnt & (u32)~0);
 }
 
 static cycle_t arch_counter_read(struct clocksource *cs)
 {
+	/*
+	 * Always use the physical counter for the clocksource.
+	 * CNTHCTL.PL1PCTEN must be set to 1.
+	 */
 	return arch_counter_get_cntpct();
 }
 

@@ -245,9 +342,14 @@ static void __cpuinit arch_timer_stop(struct clock_event_device *clk)
 {
 	pr_debug("arch_timer_teardown disable IRQ%d cpu #%d\n",
 		 clk->irq, smp_processor_id());
-	disable_percpu_irq(clk->irq);
-	if (arch_timer_ppi2)
-		disable_percpu_irq(arch_timer_ppi2);
+
+	if (arch_timer_use_virtual)
+		disable_percpu_irq(arch_timer_ppi[VIRT_PPI]);
+	else {
+		disable_percpu_irq(arch_timer_ppi[PHYS_SECURE_PPI]);
+		disable_percpu_irq(arch_timer_ppi[PHYS_NONSECURE_PPI]);
+	}
+
 	arch_timer_set_mode(CLOCK_EVT_MODE_UNUSED, clk);
 }
 

@@ -261,36 +363,49 @@ static struct clock_event_device arch_timer_global_evt;
 static int __init arch_timer_register(void)
 {
 	int err;
+	int ppi;
+
+	if (arch_timer_use_virtual)
+		arch_timer_reg_ops = &arch_timer_virt_ops;
+	else
+		arch_timer_reg_ops = &arch_timer_phys_ops;
 
 	err = arch_timer_available();
 	if (err)
-		return err;
+		goto out;
 
 	arch_timer_evt = alloc_percpu(struct clock_event_device *);
-	if (!arch_timer_evt)
-		return -ENOMEM;
+	if (!arch_timer_evt) {
+		err = -ENOMEM;
+		goto out;
+	}
 
 	clocksource_register_hz(&clocksource_counter, arch_timer_rate);
 
-	err = request_percpu_irq(arch_timer_ppi, arch_timer_handler,
-				 "arch_timer", arch_timer_evt);
+	if (arch_timer_use_virtual) {
+		ppi = arch_timer_ppi[VIRT_PPI];
+		err = request_percpu_irq(ppi, arch_timer_handler,
+					 "arch_timer", arch_timer_evt);
+	} else {
+		ppi = arch_timer_ppi[PHYS_SECURE_PPI];
+		err = request_percpu_irq(ppi, arch_timer_handler,
+					 "arch_timer", arch_timer_evt);
+		if (!err) {
+			ppi = arch_timer_ppi[PHYS_NONSECURE_PPI];
+			err = request_percpu_irq(ppi, arch_timer_handler,
+						 "arch_timer", arch_timer_evt);
+			if (err)
+				free_percpu_irq(arch_timer_ppi[PHYS_SECURE_PPI],
+						arch_timer_evt);
+		}
+	}
+
 	if (err) {
 		pr_err("arch_timer: can't register interrupt %d (%d)\n",
-		       arch_timer_ppi, err);
+		       ppi, err);
 		goto out_free;
 	}
 
-	if (arch_timer_ppi2) {
-		err = request_percpu_irq(arch_timer_ppi2, arch_timer_handler,
-					 "arch_timer", arch_timer_evt);
-		if (err) {
-			pr_err("arch_timer: can't register interrupt %d (%d)\n",
-			       arch_timer_ppi2, err);
-			arch_timer_ppi2 = 0;
-			goto out_free_irq;
-		}
-	}
-
 	err = local_timer_register(&arch_timer_ops);
 	if (err) {
 		/*

@@ -310,13 +425,18 @@ static int __init arch_timer_register(void)
 	return 0;
 
 out_free_irq:
-	free_percpu_irq(arch_timer_ppi, arch_timer_evt);
-	if (arch_timer_ppi2)
-		free_percpu_irq(arch_timer_ppi2, arch_timer_evt);
+	if (arch_timer_use_virtual)
+		free_percpu_irq(arch_timer_ppi[VIRT_PPI], arch_timer_evt);
+	else {
+		free_percpu_irq(arch_timer_ppi[PHYS_SECURE_PPI],
+				arch_timer_evt);
+		free_percpu_irq(arch_timer_ppi[PHYS_NONSECURE_PPI],
+				arch_timer_evt);
+	}
 
 out_free:
 	free_percpu(arch_timer_evt);
-
+out:
 	return err;
 }
 

@@ -329,6 +449,7 @@ int __init arch_timer_of_register(void)
 {
 	struct device_node *np;
 	u32 freq;
+	int i;
 
 	np = of_find_matching_node(NULL, arch_timer_of_match);
 	if (!np) {

@@ -340,10 +461,22 @@ int __init arch_timer_of_register(void)
 	if (!of_property_read_u32(np, "clock-frequency", &freq))
 		arch_timer_rate = freq;
 
-	arch_timer_ppi = irq_of_parse_and_map(np, 0);
-	arch_timer_ppi2 = irq_of_parse_and_map(np, 1);
-	pr_info("arch_timer: found %s irqs %d %d\n",
-		np->name, arch_timer_ppi, arch_timer_ppi2);
+	for (i = PHYS_SECURE_PPI; i < MAX_TIMER_PPI; i++)
+		arch_timer_ppi[i] = irq_of_parse_and_map(np, i);
+
+	/*
+	 * If no interrupt provided for virtual timer, we'll have to
+	 * stick to the physical timer. It'd better be accessible...
+	 */
+	if (!arch_timer_ppi[VIRT_PPI]) {
+		arch_timer_use_virtual = false;
+
+		if (!arch_timer_ppi[PHYS_SECURE_PPI] ||
+		    !arch_timer_ppi[PHYS_NONSECURE_PPI]) {
+			pr_warn("arch_timer: No interrupt available, giving up\n");
+			return -EINVAL;
+		}
+	}
 
 	return arch_timer_register();
 }

@@ -356,6 +489,6 @@ int __init arch_timer_sched_clock_init(void)
 	if (err)
 		return err;
 
-	setup_sched_clock(arch_counter_get_cntvct32, 32, arch_timer_rate);
+	setup_sched_clock(arch_counter_get_cnt32, 32, arch_timer_rate);
 	return 0;
 }

@@ -2,11 +2,13 @@
 #define __ASMARM_ARCH_TIMER_H
 
 #include <asm/errno.h>
+#include <linux/clocksource.h>
 
 #ifdef CONFIG_ARM_ARCH_TIMER
 #define ARCH_HAS_READ_CURRENT_TIMER
 int arch_timer_of_register(void);
 int arch_timer_sched_clock_init(void);
+struct timecounter *arch_timer_get_timecounter(void);
 #else
 static inline int arch_timer_of_register(void)
 {

@@ -17,6 +19,11 @@ static inline int arch_timer_sched_clock_init(void)
 {
 	return -ENXIO;
 }
+
+static inline struct timecounter *arch_timer_get_timecounter(void)
+{
+	return NULL;
+}
 #endif
 
 #endif

@@ -330,6 +330,15 @@ int read_current_timer(unsigned long *timer_val)
 	return 0;
 }
 
+static cycle_t arch_counter_read_cc(const struct cyclecounter *cc)
+{
+	/*
+	 * Always use the physical counter for the clocksource.
+	 * CNTHCTL.PL1PCTEN must be set to 1.
+	 */
+	return arch_counter_get_cntpct();
+}
+
 static struct clocksource clocksource_counter = {
 	.name	= "arch_sys_counter",
 	.rating	= 400,

@@ -338,6 +347,18 @@ static struct clocksource clocksource_counter = {
 	.flags	= CLOCK_SOURCE_IS_CONTINUOUS,
 };
 
+static struct cyclecounter cyclecounter = {
+	.read	= arch_counter_read_cc,
+	.mask	= CLOCKSOURCE_MASK(56),
+};
+
+static struct timecounter timecounter;
+
+struct timecounter *arch_timer_get_timecounter(void)
+{
+	return &timecounter;
+}
+
 static void __cpuinit arch_timer_stop(struct clock_event_device *clk)
 {
 	pr_debug("arch_timer_teardown disable IRQ%d cpu #%d\n",

@@ -381,6 +402,10 @@ static int __init arch_timer_register(void)
 	}
 
 	clocksource_register_hz(&clocksource_counter, arch_timer_rate);
+	cyclecounter.mult = clocksource_counter.mult;
+	cyclecounter.shift = clocksource_counter.shift;
+	timecounter_init(&timecounter, &cyclecounter,
+			 arch_counter_get_cntpct());
 
 	if (arch_timer_use_virtual) {
 		ppi = arch_timer_ppi[VIRT_PPI];