ring-buffer: add NMI protection for spinlocks

Impact: prevent deadlock in NMI

The ring buffers are not yet totally lockless with writing to
the buffer. When a writer crosses a page, it grabs a per cpu spinlock
to protect against a reader. The spinlocks taken by a writer are not
to protect against other writers, since a writer can only write to
its own per cpu buffer. The spinlocks protect against readers that
can touch any cpu buffer. The writers are made to be reentrant
with the spinlocks disabling interrupts.

The problem arises when an NMI writes to the buffer, and that write
crosses a page boundary. If it grabs a spinlock, it can be racing
with another writer (since disabling interrupts does not protect
against NMIs) or with a reader on the same CPU. Luckily, most of the
users are not reentrant and protects against this issue. But if a
user of the ring buffer becomes reentrant (which is what the ring
buffers do allow), if the NMI also writes to the ring buffer then
we risk the chance of a deadlock.

This patch moves the ftrace_nmi_enter called by nmi_enter() to the
ring buffer code. It replaces the current ftrace_nmi_enter that is
used by arch specific code to arch_ftrace_nmi_enter and updates
the Kconfig to handle it.

When an NMI is called, it will set a per cpu variable in the ring buffer
code and will clear it when the NMI exits. If a write to the ring buffer
crosses page boundaries inside an NMI, a trylock is used on the spin
lock instead. If the spinlock fails to be acquired, then the entry
is discarded.

This bug appeared in the ftrace work in the RT tree, where event tracing
is reentrant. This workaround solved the deadlocks that appeared there.

Signed-off-by: Steven Rostedt <srostedt@redhat.com>

kernel/trace/ring_buffer.c

@@ -4,6 +4,7 @@
  * Copyright (C) 2008 Steven Rostedt <srostedt@redhat.com>
  */
 #include <linux/ring_buffer.h>
+#include <linux/ftrace_irq.h>
 #include <linux/spinlock.h>
 #include <linux/debugfs.h>
 #include <linux/uaccess.h>
@@ -18,6 +19,35 @@
 
 #include "trace.h"
 
+/*
+ * Since the write to the buffer is still not fully lockless,
+ * we must be careful with NMIs. The locks in the writers
+ * are taken when a write crosses to a new page. The locks
+ * protect against races with the readers (this will soon
+ * be fixed with a lockless solution).
+ *
+ * Because we can not protect against NMIs, and we want to
+ * keep traces reentrant, we need to manage what happens
+ * when we are in an NMI.
+ */
+static DEFINE_PER_CPU(int, rb_in_nmi);
+
+void ftrace_nmi_enter(void)
+{
+	__get_cpu_var(rb_in_nmi)++;
+	/* call arch specific handler too */
+	arch_ftrace_nmi_enter();
+}
+
+void ftrace_nmi_exit(void)
+{
+	arch_ftrace_nmi_exit();
+	__get_cpu_var(rb_in_nmi)--;
+	/* NMIs are not recursive */
+	WARN_ON_ONCE(__get_cpu_var(rb_in_nmi));
+}
+
+
 /*
  * A fast way to enable or disable all ring buffers is to
  * call tracing_on or tracing_off. Turning off the ring buffers
@@ -982,6 +1012,7 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
 	struct ring_buffer *buffer = cpu_buffer->buffer;
 	struct ring_buffer_event *event;
 	unsigned long flags;
+	bool lock_taken = false;
 
 	commit_page = cpu_buffer->commit_page;
 	/* we just need to protect against interrupts */
@@ -995,7 +1026,19 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
 		struct buffer_page *next_page = tail_page;
 
 		local_irq_save(flags);
-		__raw_spin_lock(&cpu_buffer->lock);
+		/*
+		 * NMIs can happen after we take the lock.
+		 * If we are in an NMI, only take the lock
+		 * if it is not already taken. Otherwise
+		 * simply fail.
+		 */
+		if (unlikely(__get_cpu_var(rb_in_nmi))) {
+			if (!__raw_spin_trylock(&cpu_buffer->lock))
+				goto out_unlock;
+		} else
+			__raw_spin_lock(&cpu_buffer->lock);
+
+		lock_taken = true;
 
 		rb_inc_page(cpu_buffer, &next_page);
 
@@ -1097,7 +1140,8 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
 	if (tail <= BUF_PAGE_SIZE)
 		local_set(&tail_page->write, tail);
 
-	__raw_spin_unlock(&cpu_buffer->lock);
+	if (likely(lock_taken))
+		__raw_spin_unlock(&cpu_buffer->lock);
 	local_irq_restore(flags);
 	return NULL;
 }