[PATCH] perf/aux: Fix AUX buffer serialization

[Peter Zijlstra <peterz@xxxxxxxxxxxxx>]

Ole reported that event->mmap_mutex is strictly insufficient to
serialize the AUX buffer, add a per RB mutex to fully serialize it.

Note that in the lock order comment the perf_event::mmap_mutex order
was already wrong, that is, it nesting under mmap_lock is not new with
this patch.

Fixes: 45bfb2e50471 ("perf: Add AUX area to ring buffer for raw data streams")
Reported-by: Ole <ole@xxxxxxxxxxxxxxx>
Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
Signed-off-by: Ingo Molnar <mingo@xxxxxxxxxx>

diff --git a/kernel/events/core.c b/kernel/events/core.c
index c973e3c11e03..8a6c6bbcd658 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -1255,8 +1255,9 @@ static void put_ctx(struct perf_event_context *ctx)
  *	  perf_event_context::mutex
  *	    perf_event::child_mutex;
  *	      perf_event_context::lock
- *	    perf_event::mmap_mutex
  *	    mmap_lock
+ *	      perf_event::mmap_mutex
+ *	        perf_buffer::aux_mutex
  *	      perf_addr_filters_head::lock
  *
  *    cpu_hotplug_lock
@@ -6373,12 +6374,11 @@ static void perf_mmap_close(struct vm_area_struct *vma)
 		event->pmu->event_unmapped(event, vma->vm_mm);
 
 	/*
-	 * rb->aux_mmap_count will always drop before rb->mmap_count and
-	 * event->mmap_count, so it is ok to use event->mmap_mutex to
-	 * serialize with perf_mmap here.
+	 * The AUX buffer is strictly a sub-buffer, serialize using aux_mutex
+	 * to avoid complications.
 	 */
 	if (rb_has_aux(rb) && vma->vm_pgoff == rb->aux_pgoff &&
-	    atomic_dec_and_mutex_lock(&rb->aux_mmap_count, &event->mmap_mutex)) {
+	    atomic_dec_and_mutex_lock(&rb->aux_mmap_count, &rb->aux_mutex)) {
 		/*
 		 * Stop all AUX events that are writing to this buffer,
 		 * so that we can free its AUX pages and corresponding PMU
@@ -6395,7 +6395,7 @@ static void perf_mmap_close(struct vm_area_struct *vma)
 		rb_free_aux(rb);
 		WARN_ON_ONCE(refcount_read(&rb->aux_refcount));
 
-		mutex_unlock(&event->mmap_mutex);
+		mutex_unlock(&rb->aux_mutex);
 	}
 
 	if (atomic_dec_and_test(&rb->mmap_count))
@@ -6483,6 +6483,7 @@ static int perf_mmap(struct file *file, struct vm_area_struct *vma)
 	struct perf_event *event = file->private_data;
 	unsigned long user_locked, user_lock_limit;
 	struct user_struct *user = current_user();
+	struct mutex *aux_mutex = NULL;
 	struct perf_buffer *rb = NULL;
 	unsigned long locked, lock_limit;
 	unsigned long vma_size;
@@ -6531,6 +6532,9 @@ static int perf_mmap(struct file *file, struct vm_area_struct *vma)
 		if (!rb)
 			goto aux_unlock;
 
+		aux_mutex = &rb->aux_mutex;
+		mutex_lock(aux_mutex);
+
 		aux_offset = READ_ONCE(rb->user_page->aux_offset);
 		aux_size = READ_ONCE(rb->user_page->aux_size);
 
@@ -6681,6 +6685,8 @@ static int perf_mmap(struct file *file, struct vm_area_struct *vma)
 		atomic_dec(&rb->mmap_count);
 	}
 aux_unlock:
+	if (aux_mutex)
+		mutex_unlock(aux_mutex);
 	mutex_unlock(&event->mmap_mutex);
 
 	/*
diff --git a/kernel/events/internal.h b/kernel/events/internal.h
index 451514442a1b..e072d995d670 100644
--- a/kernel/events/internal.h
+++ b/kernel/events/internal.h
@@ -40,6 +40,7 @@ struct perf_buffer {
 	struct user_struct		*mmap_user;
 
 	/* AUX area */
+	struct mutex			aux_mutex;
 	long				aux_head;
 	unsigned int			aux_nest;
 	long				aux_wakeup;	/* last aux_watermark boundary crossed by aux_head */
diff --git a/kernel/events/ring_buffer.c b/kernel/events/ring_buffer.c
index 8cadf97bc290..4f46f688d0d4 100644
--- a/kernel/events/ring_buffer.c
+++ b/kernel/events/ring_buffer.c
@@ -337,6 +337,8 @@ ring_buffer_init(struct perf_buffer *rb, long watermark, int flags)
 	 */
 	if (!rb->nr_pages)
 		rb->paused = 1;
+
+	mutex_init(&rb->aux_mutex);
 }
 
 void perf_aux_output_flag(struct perf_output_handle *handle, u64 flags)