[RFC PATCH V2 3/8] xen: generalized event channel operations

[Wei Liu <wei.liu2@citrix.com>]

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
---
 drivers/xen/events.c |  133 ++++++++++++++++++++++++++++++++++----------------
 drivers/xen/evtchn.c |   14 +++---
 2 files changed, 98 insertions(+), 49 deletions(-)

diff --git a/drivers/xen/events.c b/drivers/xen/events.c
index dadeea4..6c1917e 100644
--- a/drivers/xen/events.c
+++ b/drivers/xen/events.c
@@ -51,6 +51,27 @@
 #include <xen/interface/hvm/hvm_op.h>
 #include <xen/interface/hvm/params.h>
 
+/* N-level event channel, starting from 2 */
+unsigned int evtchn_level = 2;
+EXPORT_SYMBOL_GPL(evtchn_level);
+unsigned int nr_event_channels;
+EXPORT_SYMBOL_GPL(nr_event_channels);
+
+struct evtchn_ops {
+	unsigned long (*active_evtchns)   (unsigned int cpu,
+					   struct shared_info *sh,
+					   unsigned int idx);
+	void          (*clear_evtchn)     (int port);
+	void          (*set_evtchn)       (int port);
+	int           (*test_evtchn)      (int port);
+	void          (*mask_evtchn)      (int port);
+	void          (*unmask_evtchn)    (int port);
+	int           (*test_and_set_mask)(int port);
+	void          (*do_upcall)        (void);
+	irqreturn_t   (*debug_interrupt)  (int irq, void *dev_id);
+};
+static struct evtchn_ops *eops;
+
 /*
  * This lock protects updates to the following mapping and reference-count
  * arrays. The lock does not need to be acquired to read the mapping tables.
@@ -113,7 +134,7 @@ static int *evtchn_to_irq;
 static unsigned long *pirq_eoi_map;
 static bool (*pirq_needs_eoi)(unsigned irq);
 
-static DEFINE_PER_CPU(unsigned long [NR_EVENT_CHANNELS/BITS_PER_LONG],
+static DEFINE_PER_CPU(unsigned long [NR_EVENT_CHANNELS_L2/BITS_PER_LONG],
 		      cpu_evtchn_mask);
 
 /* Xen will never allocate port zero for any purpose. */
@@ -285,9 +306,9 @@ static bool pirq_needs_eoi_flag(unsigned irq)
 	return info->u.pirq.flags & PIRQ_NEEDS_EOI;
 }
 
-static inline unsigned long active_evtchns(unsigned int cpu,
-					   struct shared_info *sh,
-					   unsigned int idx)
+static inline unsigned long active_evtchns_l2(unsigned int cpu,
+					      struct shared_info *sh,
+					      unsigned int idx)
 {
 	return sh->evtchn_pending[idx] &
 		per_cpu(cpu_evtchn_mask, cpu)[idx] &
@@ -327,19 +348,19 @@ static void init_evtchn_cpu_bindings(void)
 		       (i == 0) ? ~0 : 0, sizeof(*per_cpu(cpu_evtchn_mask, i)));
 }
 
-static inline void clear_evtchn(int port)
+static inline void clear_evtchn_l2(int port)
 {
 	struct shared_info *s = HYPERVISOR_shared_info;
 	sync_clear_bit(port, &s->evtchn_pending[0]);
 }
 
-static inline void set_evtchn(int port)
+static inline void set_evtchn_l2(int port)
 {
 	struct shared_info *s = HYPERVISOR_shared_info;
 	sync_set_bit(port, &s->evtchn_pending[0]);
 }
 
-static inline int test_evtchn(int port)
+static inline int test_evtchn_l2(int port)
 {
 	struct shared_info *s = HYPERVISOR_shared_info;
 	return sync_test_bit(port, &s->evtchn_pending[0]);
@@ -363,13 +384,13 @@ void notify_remote_via_irq(int irq)
 }
 EXPORT_SYMBOL_GPL(notify_remote_via_irq);
 
-static void mask_evtchn(int port)
+static void mask_evtchn_l2(int port)
 {
 	struct shared_info *s = HYPERVISOR_shared_info;
 	sync_set_bit(port, &s->evtchn_mask[0]);
 }
 
-static void unmask_evtchn(int port)
+static void unmask_evtchn_l2(int port)
 {
 	struct shared_info *s = HYPERVISOR_shared_info;
 	unsigned int cpu = get_cpu();
@@ -521,7 +542,7 @@ static void eoi_pirq(struct irq_data *data)
 	irq_move_irq(data);
 
 	if (VALID_EVTCHN(evtchn))
-		clear_evtchn(evtchn);
+		eops->clear_evtchn(evtchn);
 
 	if (pirq_needs_eoi(data->irq)) {
 		rc = HYPERVISOR_physdev_op(PHYSDEVOP_eoi, &eoi);
@@ -567,7 +588,7 @@ static unsigned int __startup_pirq(unsigned int irq)
 	info->evtchn = evtchn;
 
 out:
-	unmask_evtchn(evtchn);
+	eops->unmask_evtchn(evtchn);
 	eoi_pirq(irq_get_irq_data(irq));
 
 	return 0;
@@ -590,7 +611,7 @@ static void shutdown_pirq(struct irq_data *data)
 	if (!VALID_EVTCHN(evtchn))
 		return;
 
-	mask_evtchn(evtchn);
+	eops->mask_evtchn(evtchn);
 
 	close.port = evtchn;
 	if (HYPERVISOR_event_channel_op(EVTCHNOP_close, &close) != 0)
@@ -896,7 +917,7 @@ static int find_virq(unsigned int virq, unsigned int cpu)
 	int port, rc = -ENOENT;
 
 	memset(&status, 0, sizeof(status));
-	for (port = 0; port <= NR_EVENT_CHANNELS; port++) {
+	for (port = 0; port <= nr_event_channels; port++) {
 		status.dom = DOMID_SELF;
 		status.port = port;
 		rc = HYPERVISOR_event_channel_op(EVTCHNOP_status, &status);
@@ -1121,7 +1142,7 @@ int evtchn_get(unsigned int evtchn)
 	struct irq_info *info;
 	int err = -ENOENT;
 
-	if (evtchn >= NR_EVENT_CHANNELS)
+	if (evtchn >= nr_event_channels)
 		return -EINVAL;
 
 	mutex_lock(&irq_mapping_update_lock);
@@ -1164,7 +1185,7 @@ void xen_send_IPI_one(unsigned int cpu, enum ipi_vector vector)
 	notify_remote_via_irq(irq);
 }
 
-irqreturn_t xen_debug_interrupt(int irq, void *dev_id)
+static irqreturn_t debug_interrupt_l2(int irq, void *dev_id)
 {
 	struct shared_info *sh = HYPERVISOR_shared_info;
 	int cpu = smp_processor_id();
@@ -1210,7 +1231,7 @@ irqreturn_t xen_debug_interrupt(int irq, void *dev_id)
 		       i % 8 == 0 ? "\n   " : " ");
 
 	printk("\nlocal cpu%d mask:\n   ", cpu);
-	for (i = (NR_EVENT_CHANNELS/BITS_PER_LONG)-1; i >= 0; i--)
+	for (i = (NR_EVENT_CHANNELS_L2/BITS_PER_LONG)-1; i >= 0; i--)
 		printk("%0*lx%s", (int)(sizeof(cpu_evtchn[0])*2),
 		       cpu_evtchn[i],
 		       i % 8 == 0 ? "\n   " : " ");
@@ -1225,7 +1246,7 @@ irqreturn_t xen_debug_interrupt(int irq, void *dev_id)
 	}
 
 	printk("\npending list:\n");
-	for (i = 0; i < NR_EVENT_CHANNELS; i++) {
+	for (i = 0; i < NR_EVENT_CHANNELS_L2; i++) {
 		if (sync_test_bit(i, sh->evtchn_pending)) {
 			int word_idx = i / BITS_PER_LONG;
 			printk("  %d: event %d -> irq %d%s%s%s\n",
@@ -1245,6 +1266,11 @@ irqreturn_t xen_debug_interrupt(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
+irqreturn_t xen_debug_interrupt(int irq, void *dev_id)
+{
+	return eops->debug_interrupt(irq, dev_id);
+}
+
 static DEFINE_PER_CPU(unsigned, xed_nesting_count);
 static DEFINE_PER_CPU(unsigned int, current_word_idx);
 static DEFINE_PER_CPU(unsigned int, current_bit_idx);
@@ -1263,7 +1289,7 @@ static DEFINE_PER_CPU(unsigned int, current_bit_idx);
  * a bitset of words which contain pending event bits.  The second
  * level is a bitset of pending events themselves.
  */
-static void __xen_evtchn_do_upcall(void)
+static void do_upcall_l2(void)
 {
 	int start_word_idx, start_bit_idx;
 	int word_idx, bit_idx;
@@ -1308,7 +1334,7 @@ static void __xen_evtchn_do_upcall(void)
 			}
 			word_idx = __ffs(words);
 
-			pending_bits = active_evtchns(cpu, s, word_idx);
+			pending_bits = eops->active_evtchns(cpu, s, word_idx);
 			bit_idx = 0; /* usually scan entire word from start */
 			if (word_idx == start_word_idx) {
 				/* We scan the starting word in two parts */
@@ -1377,7 +1403,7 @@ void xen_evtchn_do_upcall(struct pt_regs *regs)
 	exit_idle();
 	irq_enter();
 
-	__xen_evtchn_do_upcall();
+	eops->do_upcall();
 
 	irq_exit();
 	set_irq_regs(old_regs);
@@ -1385,7 +1411,7 @@ void xen_evtchn_do_upcall(struct pt_regs *regs)
 
 void xen_hvm_evtchn_do_upcall(void)
 {
-	__xen_evtchn_do_upcall();
+	eops->do_upcall();
 }
 EXPORT_SYMBOL_GPL(xen_hvm_evtchn_do_upcall);
 
@@ -1459,15 +1485,14 @@ static int set_affinity_irq(struct irq_data *data, const struct cpumask *dest,
 int resend_irq_on_evtchn(unsigned int irq)
 {
 	int masked, evtchn = evtchn_from_irq(irq);
-	struct shared_info *s = HYPERVISOR_shared_info;
 
 	if (!VALID_EVTCHN(evtchn))
 		return 1;
 
-	masked = sync_test_and_set_bit(evtchn, s->evtchn_mask);
-	sync_set_bit(evtchn, s->evtchn_pending);
+	masked = eops->test_and_set_mask(evtchn);
+	eops->set_evtchn(evtchn);
 	if (!masked)
-		unmask_evtchn(evtchn);
+		eops->unmask_evtchn(evtchn);
 
 	return 1;
 }
@@ -1477,7 +1502,7 @@ static void enable_dynirq(struct irq_data *data)
 	int evtchn = evtchn_from_irq(data->irq);
 
 	if (VALID_EVTCHN(evtchn))
-		unmask_evtchn(evtchn);
+		eops->unmask_evtchn(evtchn);
 }
 
 static void disable_dynirq(struct irq_data *data)
@@ -1485,7 +1510,7 @@ static void disable_dynirq(struct irq_data *data)
 	int evtchn = evtchn_from_irq(data->irq);
 
 	if (VALID_EVTCHN(evtchn))
-		mask_evtchn(evtchn);
+		eops->mask_evtchn(evtchn);
 }
 
 static void ack_dynirq(struct irq_data *data)
@@ -1495,7 +1520,7 @@ static void ack_dynirq(struct irq_data *data)
 	irq_move_irq(data);
 
 	if (VALID_EVTCHN(evtchn))
-		clear_evtchn(evtchn);
+		eops->clear_evtchn(evtchn);
 }
 
 static void mask_ack_dynirq(struct irq_data *data)
@@ -1504,19 +1529,24 @@ static void mask_ack_dynirq(struct irq_data *data)
 	ack_dynirq(data);
 }
 
+static inline int test_and_set_mask_l2(int chn)
+{
+	struct shared_info *sh = HYPERVISOR_shared_info;
+	return sync_test_and_set_bit(chn, sh->evtchn_mask);
+}
+
 static int retrigger_dynirq(struct irq_data *data)
 {
 	int evtchn = evtchn_from_irq(data->irq);
-	struct shared_info *sh = HYPERVISOR_shared_info;
 	int ret = 0;
 
 	if (VALID_EVTCHN(evtchn)) {
 		int masked;
 
-		masked = sync_test_and_set_bit(evtchn, sh->evtchn_mask);
-		sync_set_bit(evtchn, sh->evtchn_pending);
+		masked = eops->test_and_set_mask(evtchn);
+		eops->set_evtchn(evtchn);
 		if (!masked)
-			unmask_evtchn(evtchn);
+			eops->unmask_evtchn(evtchn);
 		ret = 1;
 	}
 
@@ -1616,7 +1646,7 @@ void xen_clear_irq_pending(int irq)
 	int evtchn = evtchn_from_irq(irq);
 
 	if (VALID_EVTCHN(evtchn))
-		clear_evtchn(evtchn);
+		eops->clear_evtchn(evtchn);
 }
 EXPORT_SYMBOL(xen_clear_irq_pending);
 void xen_set_irq_pending(int irq)
@@ -1624,7 +1654,7 @@ void xen_set_irq_pending(int irq)
 	int evtchn = evtchn_from_irq(irq);
 
 	if (VALID_EVTCHN(evtchn))
-		set_evtchn(evtchn);
+		eops->set_evtchn(evtchn);
 }
 
 bool xen_test_irq_pending(int irq)
@@ -1633,7 +1663,7 @@ bool xen_test_irq_pending(int irq)
 	bool ret = false;
 
 	if (VALID_EVTCHN(evtchn))
-		ret = test_evtchn(evtchn);
+		ret = eops->test_evtchn(evtchn);
 
 	return ret;
 }
@@ -1683,14 +1713,14 @@ void xen_irq_resume(void)
 	init_evtchn_cpu_bindings();
 
 	/* New event-channel space is not 'live' yet. */
-	for (evtchn = 0; evtchn < NR_EVENT_CHANNELS; evtchn++)
-		mask_evtchn(evtchn);
+	for (evtchn = 0; evtchn < nr_event_channels; evtchn++)
+		eops->mask_evtchn(evtchn);
 
 	/* No IRQ <-> event-channel mappings. */
 	list_for_each_entry(info, &xen_irq_list_head, list)
 		info->evtchn = 0; /* zap event-channel binding */
 
-	for (evtchn = 0; evtchn < NR_EVENT_CHANNELS; evtchn++)
+	for (evtchn = 0; evtchn < nr_event_channels; evtchn++)
 		evtchn_to_irq[evtchn] = -1;
 
 	for_each_possible_cpu(cpu) {
@@ -1783,21 +1813,38 @@ void xen_callback_vector(void)
 void xen_callback_vector(void) {}
 #endif
 
+static struct evtchn_ops evtchn_ops_l2 __read_mostly = {
+	.active_evtchns    = active_evtchns_l2,
+	.clear_evtchn      = clear_evtchn_l2,
+	.set_evtchn        = set_evtchn_l2,
+	.test_evtchn       = test_evtchn_l2,
+	.mask_evtchn       = mask_evtchn_l2,
+	.unmask_evtchn     = unmask_evtchn_l2,
+	.test_and_set_mask = test_and_set_mask_l2,
+	.do_upcall         = do_upcall_l2,
+	.debug_interrupt   = debug_interrupt_l2,
+};
+
 void __init xen_init_IRQ(void)
 {
 	int i, rc;
 
-	evtchn_to_irq = kcalloc(NR_EVENT_CHANNELS, sizeof(*evtchn_to_irq),
-				    GFP_KERNEL);
+	evtchn_level = 2;
+	nr_event_channels = NR_EVENT_CHANNELS_L2;
+	eops = &evtchn_ops_l2;
+
+	/* Setup 2-level event channel */
+	evtchn_to_irq = kcalloc(nr_event_channels, sizeof(*evtchn_to_irq),
+				GFP_KERNEL);
 	BUG_ON(!evtchn_to_irq);
-	for (i = 0; i < NR_EVENT_CHANNELS; i++)
+	for (i = 0; i < nr_event_channels; i++)
 		evtchn_to_irq[i] = -1;
 
 	init_evtchn_cpu_bindings();
 
 	/* No event channels are 'live' right now. */
-	for (i = 0; i < NR_EVENT_CHANNELS; i++)
-		mask_evtchn(i);
+	for (i = 0; i < nr_event_channels; i++)
+		eops->mask_evtchn(i);
 
 	pirq_needs_eoi = pirq_needs_eoi_flag;
 
diff --git a/drivers/xen/evtchn.c b/drivers/xen/evtchn.c
index b1f60a0..975e970 100644
--- a/drivers/xen/evtchn.c
+++ b/drivers/xen/evtchn.c
@@ -53,6 +53,8 @@
 #include <xen/evtchn.h>
 #include <asm/xen/hypervisor.h>
 
+extern unsigned int nr_event_channels;
+
 struct per_user_data {
 	struct mutex bind_mutex; /* serialize bind/unbind operations */
 
@@ -232,7 +234,7 @@ static ssize_t evtchn_write(struct file *file, const char __user *buf,
 	for (i = 0; i < (count/sizeof(evtchn_port_t)); i++) {
 		unsigned port = kbuf[i];
 
-		if (port < NR_EVENT_CHANNELS &&
+		if (port < nr_event_channels &&
 		    get_port_user(port) == u &&
 		    !get_port_enabled(port)) {
 			set_port_enabled(port, true);
@@ -364,7 +366,7 @@ static long evtchn_ioctl(struct file *file,
 			break;
 
 		rc = -EINVAL;
-		if (unbind.port >= NR_EVENT_CHANNELS)
+		if (unbind.port >= nr_event_channels)
 			break;
 
 		spin_lock_irq(&port_user_lock);
@@ -392,7 +394,7 @@ static long evtchn_ioctl(struct file *file,
 		if (copy_from_user(&notify, uarg, sizeof(notify)))
 			break;
 
-		if (notify.port >= NR_EVENT_CHANNELS) {
+		if (notify.port >= nr_event_channels) {
 			rc = -EINVAL;
 		} else if (get_port_user(notify.port) != u) {
 			rc = -ENOTCONN;
@@ -482,7 +484,7 @@ static int evtchn_release(struct inode *inode, struct file *filp)
 
 	free_page((unsigned long)u->ring);
 
-	for (i = 0; i < NR_EVENT_CHANNELS; i++) {
+	for (i = 0; i < nr_event_channels; i++) {
 		if (get_port_user(i) != u)
 			continue;
 
@@ -491,7 +493,7 @@ static int evtchn_release(struct inode *inode, struct file *filp)
 
 	spin_unlock_irq(&port_user_lock);
 
-	for (i = 0; i < NR_EVENT_CHANNELS; i++) {
+	for (i = 0; i < nr_event_channels; i++) {
 		if (get_port_user(i) != u)
 			continue;
 
@@ -528,7 +530,7 @@ static int __init evtchn_init(void)
 	if (!xen_domain())
 		return -ENODEV;
 
-	port_user = kcalloc(NR_EVENT_CHANNELS, sizeof(*port_user), GFP_KERNEL);
+	port_user = kcalloc(nr_event_channels, sizeof(*port_user), GFP_KERNEL);
 	if (port_user == NULL)
 		return -ENOMEM;
 
-- 
1.7.10.4