1 files changed, 554 insertions, 295 deletions

diff --git a/arch/um/kernel/irq.c b/arch/um/kernel/irq.c
index 36e12f0cefd5..f4b13f15a9c1 100644
--- a/arch/um/kernel/irq.c
+++ b/arch/um/kernel/irq.c
@@ -1,6 +1,8 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
+ * Copyright (C) 2017 - Cambridge Greys Ltd
+ * Copyright (C) 2011 - 2014 Cisco Systems Inc
  * Copyright (C) 2000 - 2007 Jeff Dike (jdike@{addtoit,linux.intel}.com)
- * Licensed under the GPL
  * Derived (i.e. mostly copied) from arch/i386/kernel/irq.c:
  *	Copyright (C) 1992, 1998 Linus Torvalds, Ingo Molnar
  */
@@ -16,245 +18,424 @@
 #include <as-layout.h>
 #include <kern_util.h>
 #include <os.h>
+#include <irq_user.h>
+#include <irq_kern.h>
+#include <linux/time-internal.h>
 
+DEFINE_PER_CPU_SHARED_ALIGNED(irq_cpustat_t, irq_stat);
+
+#define irq_stats(x)		(&per_cpu(irq_stat, x))
+
+/* When epoll triggers we do not know why it did so
+ * we can also have different IRQs for read and write.
+ * This is why we keep a small irq_reg array for each fd -
+ * one entry per IRQ type
+ */
+struct irq_reg {
+	void *id;
+	int irq;
+	/* it's cheaper to store this than to query it */
+	int events;
+	bool active;
+	bool pending;
+	bool wakeup;
+#ifdef CONFIG_UML_TIME_TRAVEL_SUPPORT
+	bool pending_event;
+	void (*timetravel_handler)(int, int, void *,
+				   struct time_travel_event *);
+	struct time_travel_event event;
+#endif
+};
+
+struct irq_entry {
+	struct list_head list;
+	int fd;
+	struct irq_reg reg[NUM_IRQ_TYPES];
+	bool suspended;
+	bool sigio_workaround;
+};
+
+static DEFINE_RAW_SPINLOCK(irq_lock);
+static LIST_HEAD(active_fds);
+static DECLARE_BITMAP(irqs_allocated, UM_LAST_SIGNAL_IRQ);
+static bool irqs_suspended;
+#ifdef CONFIG_UML_TIME_TRAVEL_SUPPORT
+static bool irqs_pending;
+#endif
+
+static void irq_io_loop(struct irq_reg *irq, struct uml_pt_regs *regs)
+{
 /*
- * This list is accessed under irq_lock, except in sigio_handler,
- * where it is safe from being modified.  IRQ handlers won't change it -
- * if an IRQ source has vanished, it will be freed by free_irqs just
- * before returning from sigio_handler.  That will process a separate
- * list of irqs to free, with its own locking, coming back here to
- * remove list elements, taking the irq_lock to do so.
+ * irq->active guards against reentry
+ * irq->pending accumulates pending requests
+ * if pending is raised the irq_handler is re-run
+ * until pending is cleared
  */
-static struct irq_fd *active_fds = NULL;
-static struct irq_fd **last_irq_ptr = &active_fds;
+	if (irq->active) {
+		irq->active = false;
 
-extern void free_irqs(void);
+		do {
+			irq->pending = false;
+			do_IRQ(irq->irq, regs);
+		} while (irq->pending);
+
+		irq->active = true;
+	} else {
+		irq->pending = true;
+	}
+}
 
-void sigio_handler(int sig, struct siginfo *unused_si, struct uml_pt_regs *regs)
+#ifdef CONFIG_UML_TIME_TRAVEL_SUPPORT
+static void irq_event_handler(struct time_travel_event *ev)
 {
-	struct irq_fd *irq_fd;
-	int n;
+	struct irq_reg *reg = container_of(ev, struct irq_reg, event);
 
-	if (smp_sigio_handler())
+	/* do nothing if suspended; just cause a wakeup and mark as pending */
+	if (irqs_suspended) {
+		irqs_pending = true;
+		reg->pending_event = true;
 		return;
+	}
 
-	while (1) {
-		n = os_waiting_for_events(active_fds);
-		if (n <= 0) {
-			if (n == -EINTR)
-				continue;
-			else break;
-		}
+	generic_handle_irq(reg->irq);
+}
+
+static bool irq_do_timetravel_handler(struct irq_entry *entry,
+				      enum um_irq_type t)
+{
+	struct irq_reg *reg = &entry->reg[t];
+
+	if (!reg->timetravel_handler)
+		return false;
+
+	/*
+	 * Handle all messages - we might get multiple even while
+	 * interrupts are already suspended, due to suspend order
+	 * etc. Note that time_travel_add_irq_event() will not add
+	 * an event twice, if it's pending already "first wins".
+	 */
+	reg->timetravel_handler(reg->irq, entry->fd, reg->id, &reg->event);
 
-		for (irq_fd = active_fds; irq_fd != NULL;
-		     irq_fd = irq_fd->next) {
-			if (irq_fd->current_events != 0) {
-				irq_fd->current_events = 0;
-				do_IRQ(irq_fd->irq, regs);
+	if (!reg->event.pending)
+		return false;
+
+	return true;
+}
+
+static void irq_do_pending_events(bool timetravel_handlers_only)
+{
+	struct irq_entry *entry;
+
+	if (!irqs_pending || timetravel_handlers_only)
+		return;
+
+	irqs_pending = false;
+
+	list_for_each_entry(entry, &active_fds, list) {
+		enum um_irq_type t;
+
+		for (t = 0; t < NUM_IRQ_TYPES; t++) {
+			struct irq_reg *reg = &entry->reg[t];
+
+			/*
+			 * Any timetravel_handler was invoked already, just
+			 * directly run the IRQ.
+			 */
+			if (reg->pending_event) {
+				irq_enter();
+				generic_handle_irq(reg->irq);
+				irq_exit();
+				reg->pending_event = false;
 			}
 		}
 	}
-
-	free_irqs();
+}
+#else
+static bool irq_do_timetravel_handler(struct irq_entry *entry,
+				      enum um_irq_type t)
+{
+	return false;
 }
 
-static DEFINE_SPINLOCK(irq_lock);
+static void irq_do_pending_events(bool timetravel_handlers_only)
+{
+}
+#endif
 
-static int activate_fd(int irq, int fd, int type, void *dev_id)
+static void sigio_reg_handler(int idx, struct irq_entry *entry, enum um_irq_type t,
+			      struct uml_pt_regs *regs,
+			      bool timetravel_handlers_only)
 {
-	struct pollfd *tmp_pfd;
-	struct irq_fd *new_fd, *irq_fd;
-	unsigned long flags;
-	int events, err, n;
+	struct irq_reg *reg = &entry->reg[t];
 
-	err = os_set_fd_async(fd);
-	if (err < 0)
-		goto out;
+	if (!reg->events)
+		return;
 
-	err = -ENOMEM;
-	new_fd = kmalloc(sizeof(struct irq_fd), GFP_KERNEL);
-	if (new_fd == NULL)
-		goto out;
+	if (os_epoll_triggered(idx, reg->events) <= 0)
+		return;
 
-	if (type == IRQ_READ)
-		events = UM_POLLIN | UM_POLLPRI;
-	else events = UM_POLLOUT;
-	*new_fd = ((struct irq_fd) { .next  		= NULL,
-				     .id 		= dev_id,
-				     .fd 		= fd,
-				     .type 		= type,
-				     .irq 		= irq,
-				     .events 		= events,
-				     .current_events 	= 0 } );
-
-	err = -EBUSY;
-	spin_lock_irqsave(&irq_lock, flags);
-	for (irq_fd = active_fds; irq_fd != NULL; irq_fd = irq_fd->next) {
-		if ((irq_fd->fd == fd) && (irq_fd->type == type)) {
-			printk(KERN_ERR "Registering fd %d twice\n", fd);
-			printk(KERN_ERR "Irqs : %d, %d\n", irq_fd->irq, irq);
-			printk(KERN_ERR "Ids : 0x%p, 0x%p\n", irq_fd->id,
-			       dev_id);
-			goto out_unlock;
-		}
+	if (irq_do_timetravel_handler(entry, t))
+		return;
+
+	/*
+	 * If we're called to only run time-travel handlers then don't
+	 * actually proceed but mark sigio as pending (if applicable).
+	 * For suspend/resume, timetravel_handlers_only may be true
+	 * despite time-travel not being configured and used.
+	 */
+	if (timetravel_handlers_only) {
+#ifdef CONFIG_UML_TIME_TRAVEL_SUPPORT
+		reg->pending_event = true;
+		irqs_pending = true;
+		mark_sigio_pending();
+#endif
+		return;
 	}
 
-	if (type == IRQ_WRITE)
-		fd = -1;
+	irq_io_loop(reg, regs);
+}
 
-	tmp_pfd = NULL;
-	n = 0;
+static void _sigio_handler(struct uml_pt_regs *regs,
+			   bool timetravel_handlers_only)
+{
+	struct irq_entry *irq_entry;
+	int n, i;
 
-	while (1) {
-		n = os_create_pollfd(fd, events, tmp_pfd, n);
-		if (n == 0)
-			break;
-
-		/*
-		 * n > 0
-		 * It means we couldn't put new pollfd to current pollfds
-		 * and tmp_fds is NULL or too small for new pollfds array.
-		 * Needed size is equal to n as minimum.
-		 *
-		 * Here we have to drop the lock in order to call
-		 * kmalloc, which might sleep.
-		 * If something else came in and changed the pollfds array
-		 * so we will not be able to put new pollfd struct to pollfds
-		 * then we free the buffer tmp_fds and try again.
-		 */
-		spin_unlock_irqrestore(&irq_lock, flags);
-		kfree(tmp_pfd);
+	if (timetravel_handlers_only && !um_irq_timetravel_handler_used())
+		return;
 
-		tmp_pfd = kmalloc(n, GFP_KERNEL);
-		if (tmp_pfd == NULL)
-			goto out_kfree;
+	/* Flush out pending events that were ignored due to time-travel. */
+	if (!irqs_suspended)
+		irq_do_pending_events(timetravel_handlers_only);
 
-		spin_lock_irqsave(&irq_lock, flags);
-	}
+	while (1) {
+		/* This is now lockless - epoll keeps back-referencesto the irqs
+		 * which have trigger it so there is no need to walk the irq
+		 * list and lock it every time. We avoid locking by turning off
+		 * IO for a specific fd by executing os_del_epoll_fd(fd) before
+		 * we do any changes to the actual data structures
+		 */
+		n = os_waiting_for_events_epoll();
 
-	*last_irq_ptr = new_fd;
-	last_irq_ptr = &new_fd->next;
+		if (n <= 0) {
+			if (n == -EINTR)
+				continue;
+			else
+				break;
+		}
 
-	spin_unlock_irqrestore(&irq_lock, flags);
+		for (i = 0; i < n ; i++) {
+			enum um_irq_type t;
 
-	/*
-	 * This calls activate_fd, so it has to be outside the critical
-	 * section.
-	 */
-	maybe_sigio_broken(fd, (type == IRQ_READ));
+			irq_entry = os_epoll_get_data_pointer(i);
 
-	return 0;
+			for (t = 0; t < NUM_IRQ_TYPES; t++)
+				sigio_reg_handler(i, irq_entry, t, regs,
+						  timetravel_handlers_only);
+		}
+	}
 
- out_unlock:
-	spin_unlock_irqrestore(&irq_lock, flags);
- out_kfree:
-	kfree(new_fd);
- out:
-	return err;
+	if (!timetravel_handlers_only)
+		free_irqs();
 }
 
-static void free_irq_by_cb(int (*test)(struct irq_fd *, void *), void *arg)
+void sigio_handler(int sig, struct siginfo *unused_si, struct uml_pt_regs *regs,
+		   void *mc)
 {
-	unsigned long flags;
-
-	spin_lock_irqsave(&irq_lock, flags);
-	os_free_irq_by_cb(test, arg, active_fds, &last_irq_ptr);
-	spin_unlock_irqrestore(&irq_lock, flags);
+	preempt_disable();
+	_sigio_handler(regs, irqs_suspended);
+	preempt_enable();
 }
 
-struct irq_and_dev {
-	int irq;
-	void *dev;
-};
-
-static int same_irq_and_dev(struct irq_fd *irq, void *d)
+static struct irq_entry *get_irq_entry_by_fd(int fd)
 {
-	struct irq_and_dev *data = d;
+	struct irq_entry *walk;
 
-	return ((irq->irq == data->irq) && (irq->id == data->dev));
+	lockdep_assert_held(&irq_lock);
+
+	list_for_each_entry(walk, &active_fds, list) {
+		if (walk->fd == fd)
+			return walk;
+	}
+
+	return NULL;
 }
 
-static void free_irq_by_irq_and_dev(unsigned int irq, void *dev)
+static void remove_irq_entry(struct irq_entry *to_free, bool remove)
 {
-	struct irq_and_dev data = ((struct irq_and_dev) { .irq  = irq,
-							  .dev  = dev });
+	if (!to_free)
+		return;
 
-	free_irq_by_cb(same_irq_and_dev, &data);
+	if (remove)
+		os_del_epoll_fd(to_free->fd);
+	list_del(&to_free->list);
 }
 
-static int same_fd(struct irq_fd *irq, void *fd)
+static bool update_irq_entry(struct irq_entry *entry)
 {
-	return (irq->fd == *((int *)fd));
+	enum um_irq_type i;
+	int events = 0;
+
+	for (i = 0; i < NUM_IRQ_TYPES; i++)
+		events |= entry->reg[i].events;
+
+	if (events) {
+		/* will modify (instead of add) if needed */
+		os_add_epoll_fd(events, entry->fd, entry);
+		return true;
+	}
+
+	os_del_epoll_fd(entry->fd);
+	return false;
 }
 
-void free_irq_by_fd(int fd)
+static struct irq_entry *update_or_remove_irq_entry(struct irq_entry *entry)
 {
-	free_irq_by_cb(same_fd, &fd);
+	if (update_irq_entry(entry))
+		return NULL;
+	remove_irq_entry(entry, false);
+	return entry;
 }
 
-/* Must be called with irq_lock held */
-static struct irq_fd *find_irq_by_fd(int fd, int irqnum, int *index_out)
+static int activate_fd(int irq, int fd, enum um_irq_type type, void *dev_id,
+		       void (*timetravel_handler)(int, int, void *,
+						  struct time_travel_event *))
 {
-	struct irq_fd *irq;
-	int i = 0;
-	int fdi;
+	struct irq_entry *irq_entry, *to_free = NULL;
+	int err, events = os_event_mask(type);
+	unsigned long flags;
 
-	for (irq = active_fds; irq != NULL; irq = irq->next) {
-		if ((irq->fd == fd) && (irq->irq == irqnum))
-			break;
-		i++;
-	}
-	if (irq == NULL) {
-		printk(KERN_ERR "find_irq_by_fd doesn't have descriptor %d\n",
-		       fd);
+	err = os_set_fd_async(fd);
+	if (err < 0)
 		goto out;
+
+	raw_spin_lock_irqsave(&irq_lock, flags);
+	irq_entry = get_irq_entry_by_fd(fd);
+	if (irq_entry) {
+already:
+		/* cannot register the same FD twice with the same type */
+		if (WARN_ON(irq_entry->reg[type].events)) {
+			err = -EALREADY;
+			goto out_unlock;
+		}
+
+		/* temporarily disable to avoid IRQ-side locking */
+		os_del_epoll_fd(fd);
+	} else {
+		struct irq_entry *new;
+
+		/* don't restore interrupts */
+		raw_spin_unlock(&irq_lock);
+		new = kzalloc(sizeof(*irq_entry), GFP_ATOMIC);
+		if (!new) {
+			local_irq_restore(flags);
+			return -ENOMEM;
+		}
+		raw_spin_lock(&irq_lock);
+		irq_entry = get_irq_entry_by_fd(fd);
+		if (irq_entry) {
+			to_free = new;
+			goto already;
+		}
+		irq_entry = new;
+		irq_entry->fd = fd;
+		list_add_tail(&irq_entry->list, &active_fds);
+		maybe_sigio_broken(fd);
 	}
-	fdi = os_get_pollfd(i);
-	if ((fdi != -1) && (fdi != fd)) {
-		printk(KERN_ERR "find_irq_by_fd - mismatch between active_fds "
-		       "and pollfds, fd %d vs %d, need %d\n", irq->fd,
-		       fdi, fd);
-		irq = NULL;
-		goto out;
+
+	irq_entry->reg[type].id = dev_id;
+	irq_entry->reg[type].irq = irq;
+	irq_entry->reg[type].active = true;
+	irq_entry->reg[type].events = events;
+
+#ifdef CONFIG_UML_TIME_TRAVEL_SUPPORT
+	if (um_irq_timetravel_handler_used()) {
+		irq_entry->reg[type].timetravel_handler = timetravel_handler;
+		irq_entry->reg[type].event.fn = irq_event_handler;
 	}
-	*index_out = i;
- out:
-	return irq;
+#endif
+
+	WARN_ON(!update_irq_entry(irq_entry));
+	err = 0;
+out_unlock:
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
+out:
+	kfree(to_free);
+	return err;
+}
+
+/*
+ * Remove the entry or entries for a specific FD, if you
+ * don't want to remove all the possible entries then use
+ * um_free_irq() or deactivate_fd() instead.
+ */
+void free_irq_by_fd(int fd)
+{
+	struct irq_entry *to_free;
+	unsigned long flags;
+
+	raw_spin_lock_irqsave(&irq_lock, flags);
+	to_free = get_irq_entry_by_fd(fd);
+	remove_irq_entry(to_free, true);
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
+	kfree(to_free);
 }
+EXPORT_SYMBOL(free_irq_by_fd);
 
-void reactivate_fd(int fd, int irqnum)
+static void free_irq_by_irq_and_dev(unsigned int irq, void *dev)
 {
-	struct irq_fd *irq;
+	struct irq_entry *entry, *to_free = NULL;
 	unsigned long flags;
-	int i;
 
-	spin_lock_irqsave(&irq_lock, flags);
-	irq = find_irq_by_fd(fd, irqnum, &i);
-	if (irq == NULL) {
-		spin_unlock_irqrestore(&irq_lock, flags);
-		return;
-	}
-	os_set_pollfd(i, irq->fd);
-	spin_unlock_irqrestore(&irq_lock, flags);
+	raw_spin_lock_irqsave(&irq_lock, flags);
+	list_for_each_entry(entry, &active_fds, list) {
+		enum um_irq_type i;
 
-	add_sigio_fd(fd);
+		for (i = 0; i < NUM_IRQ_TYPES; i++) {
+			struct irq_reg *reg = &entry->reg[i];
+
+			if (!reg->events)
+				continue;
+			if (reg->irq != irq)
+				continue;
+			if (reg->id != dev)
+				continue;
+
+			os_del_epoll_fd(entry->fd);
+			reg->events = 0;
+			to_free = update_or_remove_irq_entry(entry);
+			goto out;
+		}
+	}
+out:
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
+	kfree(to_free);
 }
 
 void deactivate_fd(int fd, int irqnum)
 {
-	struct irq_fd *irq;
+	struct irq_entry *entry;
 	unsigned long flags;
-	int i;
+	enum um_irq_type i;
 
-	spin_lock_irqsave(&irq_lock, flags);
-	irq = find_irq_by_fd(fd, irqnum, &i);
-	if (irq == NULL) {
-		spin_unlock_irqrestore(&irq_lock, flags);
-		return;
+	os_del_epoll_fd(fd);
+
+	raw_spin_lock_irqsave(&irq_lock, flags);
+	entry = get_irq_entry_by_fd(fd);
+	if (!entry)
+		goto out;
+
+	for (i = 0; i < NUM_IRQ_TYPES; i++) {
+		if (!entry->reg[i].events)
+			continue;
+		if (entry->reg[i].irq == irqnum)
+			entry->reg[i].events = 0;
 	}
 
-	os_set_pollfd(i, -1);
-	spin_unlock_irqrestore(&irq_lock, flags);
+	entry = update_or_remove_irq_entry(entry);
+out:
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
+	kfree(entry);
 
 	ignore_sigio_fd(fd);
 }
@@ -268,17 +449,18 @@ EXPORT_SYMBOL(deactivate_fd);
  */
 int deactivate_all_fds(void)
 {
-	struct irq_fd *irq;
-	int err;
+	struct irq_entry *entry;
 
-	for (irq = active_fds; irq != NULL; irq = irq->next) {
-		err = os_clear_fd_async(irq->fd);
-		if (err)
-			return err;
-	}
-	/* If there is a signal already queued, after unblocking ignore it */
+	/* Stop IO. The IRQ loop has no lock so this is our
+	 * only way of making sure we are safe to dispose
+	 * of all IRQ handlers
+	 */
 	os_set_ioignore();
 
+	/* we can no longer call kfree() here so just deactivate */
+	list_for_each_entry(entry, &active_fds, list)
+		os_del_epoll_fd(entry->fd);
+	os_close_epoll_fd();
 	return 0;
 }
 
@@ -297,31 +479,180 @@ unsigned int do_IRQ(int irq, struct uml_pt_regs *regs)
 	return 1;
 }
 
-void um_free_irq(unsigned int irq, void *dev)
+void um_free_irq(int irq, void *dev)
 {
+	if (WARN(irq < 0 || irq > UM_LAST_SIGNAL_IRQ,
+		 "freeing invalid irq %d", irq))
+		return;
+
 	free_irq_by_irq_and_dev(irq, dev);
 	free_irq(irq, dev);
+	clear_bit(irq, irqs_allocated);
 }
 EXPORT_SYMBOL(um_free_irq);
 
-int um_request_irq(unsigned int irq, int fd, int type,
-		   irq_handler_t handler,
-		   unsigned long irqflags, const char * devname,
-		   void *dev_id)
+static int
+_um_request_irq(int irq, int fd, enum um_irq_type type,
+		irq_handler_t handler, unsigned long irqflags,
+		const char *devname, void *dev_id,
+		void (*timetravel_handler)(int, int, void *,
+					   struct time_travel_event *))
 {
 	int err;
 
+	if (irq == UM_IRQ_ALLOC) {
+		int i;
+
+		for (i = UM_FIRST_DYN_IRQ; i < NR_IRQS; i++) {
+			if (!test_and_set_bit(i, irqs_allocated)) {
+				irq = i;
+				break;
+			}
+		}
+	}
+
+	if (irq < 0)
+		return -ENOSPC;
+
 	if (fd != -1) {
-		err = activate_fd(irq, fd, type, dev_id);
+		err = activate_fd(irq, fd, type, dev_id, timetravel_handler);
 		if (err)
-			return err;
+			goto error;
 	}
 
-	return request_irq(irq, handler, irqflags, devname, dev_id);
+	err = request_irq(irq, handler, irqflags, devname, dev_id);
+	if (err < 0)
+		goto error;
+
+	return irq;
+error:
+	clear_bit(irq, irqs_allocated);
+	return err;
 }
 
+int um_request_irq(int irq, int fd, enum um_irq_type type,
+		   irq_handler_t handler, unsigned long irqflags,
+		   const char *devname, void *dev_id)
+{
+	return _um_request_irq(irq, fd, type, handler, irqflags,
+			       devname, dev_id, NULL);
+}
 EXPORT_SYMBOL(um_request_irq);
-EXPORT_SYMBOL(reactivate_fd);
+
+#ifdef CONFIG_UML_TIME_TRAVEL_SUPPORT
+int um_request_irq_tt(int irq, int fd, enum um_irq_type type,
+		      irq_handler_t handler, unsigned long irqflags,
+		      const char *devname, void *dev_id,
+		      void (*timetravel_handler)(int, int, void *,
+						 struct time_travel_event *))
+{
+	return _um_request_irq(irq, fd, type, handler, irqflags,
+			       devname, dev_id, timetravel_handler);
+}
+EXPORT_SYMBOL(um_request_irq_tt);
+
+void sigio_run_timetravel_handlers(void)
+{
+	_sigio_handler(NULL, true);
+}
+#endif
+
+#ifdef CONFIG_PM_SLEEP
+void um_irqs_suspend(void)
+{
+	struct irq_entry *entry;
+	unsigned long flags;
+
+	irqs_suspended = true;
+
+	raw_spin_lock_irqsave(&irq_lock, flags);
+	list_for_each_entry(entry, &active_fds, list) {
+		enum um_irq_type t;
+		bool clear = true;
+
+		for (t = 0; t < NUM_IRQ_TYPES; t++) {
+			if (!entry->reg[t].events)
+				continue;
+
+			/*
+			 * For the SIGIO_WRITE_IRQ, which is used to handle the
+			 * SIGIO workaround thread, we need special handling:
+			 * enable wake for it itself, but below we tell it about
+			 * any FDs that should be suspended.
+			 */
+			if (entry->reg[t].wakeup ||
+			    entry->reg[t].irq == SIGIO_WRITE_IRQ
+#ifdef CONFIG_UML_TIME_TRAVEL_SUPPORT
+			    || entry->reg[t].timetravel_handler
+#endif
+			    ) {
+				clear = false;
+				break;
+			}
+		}
+
+		if (clear) {
+			entry->suspended = true;
+			os_clear_fd_async(entry->fd);
+			entry->sigio_workaround =
+				!__ignore_sigio_fd(entry->fd);
+		}
+	}
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
+}
+
+void um_irqs_resume(void)
+{
+	struct irq_entry *entry;
+	unsigned long flags;
+
+
+	raw_spin_lock_irqsave(&irq_lock, flags);
+	list_for_each_entry(entry, &active_fds, list) {
+		if (entry->suspended) {
+			int err = os_set_fd_async(entry->fd);
+
+			WARN(err < 0, "os_set_fd_async returned %d\n", err);
+			entry->suspended = false;
+
+			if (entry->sigio_workaround) {
+				err = __add_sigio_fd(entry->fd);
+				WARN(err < 0, "add_sigio_returned %d\n", err);
+			}
+		}
+	}
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
+
+	irqs_suspended = false;
+	send_sigio_to_self();
+}
+
+static int normal_irq_set_wake(struct irq_data *d, unsigned int on)
+{
+	struct irq_entry *entry;
+	unsigned long flags;
+
+	raw_spin_lock_irqsave(&irq_lock, flags);
+	list_for_each_entry(entry, &active_fds, list) {
+		enum um_irq_type t;
+
+		for (t = 0; t < NUM_IRQ_TYPES; t++) {
+			if (!entry->reg[t].events)
+				continue;
+
+			if (entry->reg[t].irq != d->irq)
+				continue;
+			entry->reg[t].wakeup = on;
+			goto unlock;
+		}
+	}
+unlock:
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
+	return 0;
+}
+#else
+#define normal_irq_set_wake NULL
+#endif
 
 /*
  * irq_chip must define at least enable/disable and ack when
@@ -331,139 +662,67 @@ static void dummy(struct irq_data *d)
 {
 }
 
-/* This is used for everything else than the timer. */
+/* This is used for everything other than the timer. */
 static struct irq_chip normal_irq_type = {
 	.name = "SIGIO",
 	.irq_disable = dummy,
 	.irq_enable = dummy,
 	.irq_ack = dummy,
+	.irq_mask = dummy,
+	.irq_unmask = dummy,
+	.irq_set_wake = normal_irq_set_wake,
 };
 
-static struct irq_chip SIGVTALRM_irq_type = {
-	.name = "SIGVTALRM",
+static struct irq_chip alarm_irq_type = {
+	.name = "SIGALRM",
 	.irq_disable = dummy,
 	.irq_enable = dummy,
 	.irq_ack = dummy,
+	.irq_mask = dummy,
+	.irq_unmask = dummy,
 };
 
 void __init init_IRQ(void)
 {
 	int i;
 
-	irq_set_chip_and_handler(TIMER_IRQ, &SIGVTALRM_irq_type, handle_edge_irq);
+	irq_set_chip_and_handler(TIMER_IRQ, &alarm_irq_type, handle_percpu_irq);
 
-	for (i = 1; i < NR_IRQS; i++)
+	for (i = 1; i < UM_LAST_SIGNAL_IRQ; i++)
 		irq_set_chip_and_handler(i, &normal_irq_type, handle_edge_irq);
+	/* Initialize EPOLL Loop */
+	os_setup_epoll();
 }
 
-/*
- * IRQ stack entry and exit:
- *
- * Unlike i386, UML doesn't receive IRQs on the normal kernel stack
- * and switch over to the IRQ stack after some preparation.  We use
- * sigaltstack to receive signals on a separate stack from the start.
- * These two functions make sure the rest of the kernel won't be too
- * upset by being on a different stack.  The IRQ stack has a
- * thread_info structure at the bottom so that current et al continue
- * to work.
- *
- * to_irq_stack copies the current task's thread_info to the IRQ stack
- * thread_info and sets the tasks's stack to point to the IRQ stack.
- *
- * from_irq_stack copies the thread_info struct back (flags may have
- * been modified) and resets the task's stack pointer.
- *
- * Tricky bits -
- *
- * What happens when two signals race each other?  UML doesn't block
- * signals with sigprocmask, SA_DEFER, or sa_mask, so a second signal
- * could arrive while a previous one is still setting up the
- * thread_info.
- *
- * There are three cases -
- *     The first interrupt on the stack - sets up the thread_info and
- * handles the interrupt
- *     A nested interrupt interrupting the copying of the thread_info -
- * can't handle the interrupt, as the stack is in an unknown state
- *     A nested interrupt not interrupting the copying of the
- * thread_info - doesn't do any setup, just handles the interrupt
- *
- * The first job is to figure out whether we interrupted stack setup.
- * This is done by xchging the signal mask with thread_info->pending.
- * If the value that comes back is zero, then there is no setup in
- * progress, and the interrupt can be handled.  If the value is
- * non-zero, then there is stack setup in progress.  In order to have
- * the interrupt handled, we leave our signal in the mask, and it will
- * be handled by the upper handler after it has set up the stack.
- *
- * Next is to figure out whether we are the outer handler or a nested
- * one.  As part of setting up the stack, thread_info->real_thread is
- * set to non-NULL (and is reset to NULL on exit).  This is the
- * nesting indicator.  If it is non-NULL, then the stack is already
- * set up and the handler can run.
- */
-
-static unsigned long pending_mask;
-
-unsigned long to_irq_stack(unsigned long *mask_out)
+int __init arch_probe_nr_irqs(void)
 {
-	struct thread_info *ti;
-	unsigned long mask, old;
-	int nested;
-
-	mask = xchg(&pending_mask, *mask_out);
-	if (mask != 0) {
-		/*
-		 * If any interrupts come in at this point, we want to
-		 * make sure that their bits aren't lost by our
-		 * putting our bit in.  So, this loop accumulates bits
-		 * until xchg returns the same value that we put in.
-		 * When that happens, there were no new interrupts,
-		 * and pending_mask contains a bit for each interrupt
-		 * that came in.
-		 */
-		old = *mask_out;
-		do {
-			old |= mask;
-			mask = xchg(&pending_mask, old);
-		} while (mask != old);
-		return 1;
-	}
-
-	ti = current_thread_info();
-	nested = (ti->real_thread != NULL);
-	if (!nested) {
-		struct task_struct *task;
-		struct thread_info *tti;
-
-		task = cpu_tasks[ti->cpu].task;
-		tti = task_thread_info(task);
-
-		*ti = *tti;
-		ti->real_thread = tti;
-		task->stack = ti;
-	}
-
-	mask = xchg(&pending_mask, 0);
-	*mask_out |= mask | nested;
-	return 0;
+	return NR_IRQS;
 }
 
-unsigned long from_irq_stack(int nested)
+void sigchld_handler(int sig, struct siginfo *unused_si,
+		     struct uml_pt_regs *regs, void *mc)
 {
-	struct thread_info *ti, *to;
-	unsigned long mask;
+	do_IRQ(SIGCHLD_IRQ, regs);
+}
 
-	ti = current_thread_info();
+/*
+ * /proc/interrupts printing for arch specific interrupts
+ */
+int arch_show_interrupts(struct seq_file *p, int prec)
+{
+#if IS_ENABLED(CONFIG_SMP)
+	int cpu;
 
-	pending_mask = 1;
+	seq_printf(p, "%*s: ", prec, "RES");
+	for_each_online_cpu(cpu)
+		seq_printf(p, "%10u ", irq_stats(cpu)->irq_resched_count);
+	seq_puts(p, "  Rescheduling interrupts\n");
 
-	to = ti->real_thread;
-	current->stack = to;
-	ti->real_thread = NULL;
-	*to = *ti;
+	seq_printf(p, "%*s: ", prec, "CAL");
+	for_each_online_cpu(cpu)
+		seq_printf(p, "%10u ", irq_stats(cpu)->irq_call_count);
+	seq_puts(p, "  Function call interrupts\n");
+#endif
 
-	mask = xchg(&pending_mask, 0);
-	return mask & ~1;
+	return 0;
 }
-