Commit 517e6a30 authored by Peter Zijlstra's avatar Peter Zijlstra
Browse files

perf: Fix perf_pending_task() UaF 



Per syzbot it is possible for perf_pending_task() to run after the
event is free()'d. There are two related but distinct cases:

 - the task_work was already queued before destroying the event;
 - destroying the event itself queues the task_work.

The first cannot be solved using task_work_cancel() since
perf_release() itself might be called from a task_work (____fput),
which means the current->task_works list is already empty and
task_work_cancel() won't be able to find the perf_pending_task()
entry.

The simplest alternative is extending the perf_event lifetime to cover
the task_work.

The second is just silly, queueing a task_work while you know the
event is going away makes no sense and is easily avoided by
re-arranging how the event is marked STATE_DEAD and ensuring it goes
through STATE_OFF on the way down.

Reported-by: default avatar <syzbot+9228d6098455bb209ec8@syzkaller.appspotmail.com>
Signed-off-by: default avatarPeter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: default avatarMarco Elver <elver@google.com>
parent 030a976e

kernel/events/core.c

+13 −4
Original line number Diff line number Diff line
@@ -2291,6 +2291,7 @@ event_sched_out(struct perf_event *event, 
		    !event->pending_work) {

			event->pending_work = 1;

			dec = false;

			WARN_ON_ONCE(!atomic_long_inc_not_zero(&event->refcount));

			task_work_add(current, &event->pending_task, TWA_RESUME);

		}

		if (dec)
@@ -2336,6 +2337,7 @@ group_sched_out(struct perf_event *group_event, 


#define DETACH_GROUP	0x01UL

#define DETACH_CHILD	0x02UL

#define DETACH_DEAD	0x04UL



/*

 * Cross CPU call to remove a performance event
@@ -2356,12 +2358,20 @@ __perf_remove_from_context(struct perf_event *event, 
		update_cgrp_time_from_cpuctx(cpuctx, false);

	}



	/*

	 * Ensure event_sched_out() switches to OFF, at the very least

	 * this avoids raising perf_pending_task() at this time.

	 */

	if (flags & DETACH_DEAD)

		event->pending_disable = 1;

	event_sched_out(event, cpuctx, ctx);

	if (flags & DETACH_GROUP)

		perf_group_detach(event);

	if (flags & DETACH_CHILD)

		perf_child_detach(event);

	list_del_event(event, ctx);

	if (flags & DETACH_DEAD)

		event->state = PERF_EVENT_STATE_DEAD;



	if (!ctx->nr_events && ctx->is_active) {

		if (ctx == &cpuctx->ctx)
@@ -5121,9 +5131,7 @@ int perf_event_release_kernel(struct perf_event *event) 


	ctx = perf_event_ctx_lock(event);

	WARN_ON_ONCE(ctx->parent_ctx);

	perf_remove_from_context(event, DETACH_GROUP);



	raw_spin_lock_irq(&ctx->lock);

	/*

	 * Mark this event as STATE_DEAD, there is no external reference to it

	 * anymore.
@@ -5135,8 +5143,7 @@ int perf_event_release_kernel(struct perf_event *event) 
	 * Thus this guarantees that we will in fact observe and kill _ALL_

	 * child events.

	 */

	event->state = PERF_EVENT_STATE_DEAD;

	raw_spin_unlock_irq(&ctx->lock);

	perf_remove_from_context(event, DETACH_GROUP|DETACH_DEAD);



	perf_event_ctx_unlock(event, ctx);
@@ -6577,6 +6584,8 @@ static void perf_pending_task(struct callback_head *head) 
	if (rctx >= 0)

		perf_swevent_put_recursion_context(rctx);

	preempt_enable_notrace();



	put_event(event);

}



#ifdef CONFIG_GUEST_PERF_EVENTS