1 files changed, 99 insertions, 173 deletions

diff --git a/arch/um/kernel/irq.c b/arch/um/kernel/irq.c
index 635d44606bfe..abe8f30a521c 100644
--- a/arch/um/kernel/irq.c
+++ b/arch/um/kernel/irq.c
@@ -37,7 +37,7 @@ struct irq_reg {
 	bool pending;
 	bool wakeup;
 #ifdef CONFIG_UML_TIME_TRAVEL_SUPPORT
-	bool pending_on_resume;
+	bool pending_event;
 	void (*timetravel_handler)(int, int, void *,
 				   struct time_travel_event *);
 	struct time_travel_event event;
@@ -52,10 +52,13 @@ struct irq_entry {
 	bool sigio_workaround;
 };
 
-static DEFINE_SPINLOCK(irq_lock);
+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)
 {
@@ -84,9 +87,12 @@ static void irq_event_handler(struct time_travel_event *ev)
 {
 	struct irq_reg *reg = container_of(ev, struct irq_reg, event);
 
-	/* do nothing if suspended - just to cause a wakeup */
-	if (irqs_suspended)
+	/* do nothing if suspended; just cause a wakeup and mark as pending */
+	if (irqs_suspended) {
+		irqs_pending = true;
+		reg->pending_event = true;
 		return;
+	}
 
 	generic_handle_irq(reg->irq);
 }
@@ -110,16 +116,47 @@ static bool irq_do_timetravel_handler(struct irq_entry *entry,
 	if (!reg->event.pending)
 		return false;
 
-	if (irqs_suspended)
-		reg->pending_on_resume = true;
 	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;
+			}
+		}
+	}
+}
 #else
 static bool irq_do_timetravel_handler(struct irq_entry *entry,
 				      enum um_irq_type t)
 {
 	return false;
 }
+
+static void irq_do_pending_events(bool timetravel_handlers_only)
+{
+}
 #endif
 
 static void sigio_reg_handler(int idx, struct irq_entry *entry, enum um_irq_type t,
@@ -145,6 +182,8 @@ static void sigio_reg_handler(int idx, struct irq_entry *entry, enum um_irq_type
 	 */
 	if (timetravel_handlers_only) {
 #ifdef CONFIG_UML_TIME_TRAVEL_SUPPORT
+		reg->pending_event = true;
+		irqs_pending = true;
 		mark_sigio_pending();
 #endif
 		return;
@@ -162,6 +201,10 @@ static void _sigio_handler(struct uml_pt_regs *regs,
 	if (timetravel_handlers_only && !um_irq_timetravel_handler_used())
 		return;
 
+	/* Flush out pending events that were ignored due to time-travel. */
+	if (!irqs_suspended)
+		irq_do_pending_events(timetravel_handlers_only);
+
 	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
@@ -193,9 +236,12 @@ static void _sigio_handler(struct uml_pt_regs *regs,
 		free_irqs();
 }
 
-void sigio_handler(int sig, struct siginfo *unused_si, struct uml_pt_regs *regs)
+void sigio_handler(int sig, struct siginfo *unused_si, struct uml_pt_regs *regs,
+		   void *mc)
 {
+	preempt_disable();
 	_sigio_handler(regs, irqs_suspended);
+	preempt_enable();
 }
 
 static struct irq_entry *get_irq_entry_by_fd(int fd)
@@ -212,7 +258,7 @@ static struct irq_entry *get_irq_entry_by_fd(int fd)
 	return NULL;
 }
 
-static void free_irq_entry(struct irq_entry *to_free, bool remove)
+static void remove_irq_entry(struct irq_entry *to_free, bool remove)
 {
 	if (!to_free)
 		return;
@@ -220,7 +266,6 @@ static void free_irq_entry(struct irq_entry *to_free, bool remove)
 	if (remove)
 		os_del_epoll_fd(to_free->fd);
 	list_del(&to_free->list);
-	kfree(to_free);
 }
 
 static bool update_irq_entry(struct irq_entry *entry)
@@ -241,17 +286,19 @@ static bool update_irq_entry(struct irq_entry *entry)
 	return false;
 }
 
-static void update_or_free_irq_entry(struct irq_entry *entry)
+static struct irq_entry *update_or_remove_irq_entry(struct irq_entry *entry)
 {
-	if (!update_irq_entry(entry))
-		free_irq_entry(entry, false);
+	if (update_irq_entry(entry))
+		return NULL;
+	remove_irq_entry(entry, false);
+	return entry;
 }
 
 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_entry *irq_entry;
+	struct irq_entry *irq_entry, *to_free = NULL;
 	int err, events = os_event_mask(type);
 	unsigned long flags;
 
@@ -259,9 +306,10 @@ static int activate_fd(int irq, int fd, enum um_irq_type type, void *dev_id,
 	if (err < 0)
 		goto out;
 
-	spin_lock_irqsave(&irq_lock, flags);
+	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;
@@ -271,11 +319,22 @@ static int activate_fd(int irq, int fd, enum um_irq_type type, void *dev_id,
 		/* temporarily disable to avoid IRQ-side locking */
 		os_del_epoll_fd(fd);
 	} else {
-		irq_entry = kzalloc(sizeof(*irq_entry), GFP_ATOMIC);
-		if (!irq_entry) {
-			err = -ENOMEM;
-			goto out_unlock;
+		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);
@@ -294,12 +353,11 @@ static int activate_fd(int irq, int fd, enum um_irq_type type, void *dev_id,
 #endif
 
 	WARN_ON(!update_irq_entry(irq_entry));
-	spin_unlock_irqrestore(&irq_lock, flags);
-
-	return 0;
+	err = 0;
 out_unlock:
-	spin_unlock_irqrestore(&irq_lock, flags);
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
 out:
+	kfree(to_free);
 	return err;
 }
 
@@ -313,19 +371,20 @@ void free_irq_by_fd(int fd)
 	struct irq_entry *to_free;
 	unsigned long flags;
 
-	spin_lock_irqsave(&irq_lock, flags);
+	raw_spin_lock_irqsave(&irq_lock, flags);
 	to_free = get_irq_entry_by_fd(fd);
-	free_irq_entry(to_free, true);
-	spin_unlock_irqrestore(&irq_lock, flags);
+	remove_irq_entry(to_free, true);
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
+	kfree(to_free);
 }
 EXPORT_SYMBOL(free_irq_by_fd);
 
 static void free_irq_by_irq_and_dev(unsigned int irq, void *dev)
 {
-	struct irq_entry *entry;
+	struct irq_entry *entry, *to_free = NULL;
 	unsigned long flags;
 
-	spin_lock_irqsave(&irq_lock, flags);
+	raw_spin_lock_irqsave(&irq_lock, flags);
 	list_for_each_entry(entry, &active_fds, list) {
 		enum um_irq_type i;
 
@@ -341,12 +400,13 @@ static void free_irq_by_irq_and_dev(unsigned int irq, void *dev)
 
 			os_del_epoll_fd(entry->fd);
 			reg->events = 0;
-			update_or_free_irq_entry(entry);
+			to_free = update_or_remove_irq_entry(entry);
 			goto out;
 		}
 	}
 out:
-	spin_unlock_irqrestore(&irq_lock, flags);
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
+	kfree(to_free);
 }
 
 void deactivate_fd(int fd, int irqnum)
@@ -357,7 +417,7 @@ void deactivate_fd(int fd, int irqnum)
 
 	os_del_epoll_fd(fd);
 
-	spin_lock_irqsave(&irq_lock, flags);
+	raw_spin_lock_irqsave(&irq_lock, flags);
 	entry = get_irq_entry_by_fd(fd);
 	if (!entry)
 		goto out;
@@ -369,9 +429,10 @@ void deactivate_fd(int fd, int irqnum)
 			entry->reg[i].events = 0;
 	}
 
-	update_or_free_irq_entry(entry);
+	entry = update_or_remove_irq_entry(entry);
 out:
-	spin_unlock_irqrestore(&irq_lock, flags);
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
+	kfree(entry);
 
 	ignore_sigio_fd(fd);
 }
@@ -501,7 +562,7 @@ void um_irqs_suspend(void)
 
 	irqs_suspended = true;
 
-	spin_lock_irqsave(&irq_lock, flags);
+	raw_spin_lock_irqsave(&irq_lock, flags);
 	list_for_each_entry(entry, &active_fds, list) {
 		enum um_irq_type t;
 		bool clear = true;
@@ -534,7 +595,7 @@ void um_irqs_suspend(void)
 				!__ignore_sigio_fd(entry->fd);
 		}
 	}
-	spin_unlock_irqrestore(&irq_lock, flags);
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
 }
 
 void um_irqs_resume(void)
@@ -543,30 +604,7 @@ void um_irqs_resume(void)
 	unsigned long flags;
 
 
-	local_irq_save(flags);
-#ifdef CONFIG_UML_TIME_TRAVEL_SUPPORT
-	/*
-	 * We don't need to lock anything here since we're in resume
-	 * and nothing else is running, but have disabled IRQs so we
-	 * don't try anything else with the interrupt list from there.
-	 */
-	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];
-
-			if (reg->pending_on_resume) {
-				irq_enter();
-				generic_handle_irq(reg->irq);
-				irq_exit();
-				reg->pending_on_resume = false;
-			}
-		}
-	}
-#endif
-
-	spin_lock(&irq_lock);
+	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);
@@ -580,7 +618,7 @@ void um_irqs_resume(void)
 			}
 		}
 	}
-	spin_unlock_irqrestore(&irq_lock, flags);
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
 
 	irqs_suspended = false;
 	send_sigio_to_self();
@@ -591,7 +629,7 @@ static int normal_irq_set_wake(struct irq_data *d, unsigned int on)
 	struct irq_entry *entry;
 	unsigned long flags;
 
-	spin_lock_irqsave(&irq_lock, flags);
+	raw_spin_lock_irqsave(&irq_lock, flags);
 	list_for_each_entry(entry, &active_fds, list) {
 		enum um_irq_type t;
 
@@ -606,7 +644,7 @@ static int normal_irq_set_wake(struct irq_data *d, unsigned int on)
 		}
 	}
 unlock:
-	spin_unlock_irqrestore(&irq_lock, flags);
+	raw_spin_unlock_irqrestore(&irq_lock, flags);
 	return 0;
 }
 #else
@@ -652,115 +690,3 @@ void __init init_IRQ(void)
 	/* 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)
-{
-	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;
-}
-
-unsigned long from_irq_stack(int nested)
-{
-	struct thread_info *ti, *to;
-	unsigned long mask;
-
-	ti = current_thread_info();
-
-	pending_mask = 1;
-
-	to = ti->real_thread;
-	current->stack = to;
-	ti->real_thread = NULL;
-	*to = *ti;
-
-	mask = xchg(&pending_mask, 0);
-	return mask & ~1;
-}
-