From nobody Fri Sep 19 00:54:50 2025 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 04316C352A1 for ; Wed, 30 Nov 2022 10:02:55 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S235308AbiK3KCy (ORCPT ); Wed, 30 Nov 2022 05:02:54 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:33562 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S235099AbiK3KCv (ORCPT ); Wed, 30 Nov 2022 05:02:51 -0500 Received: from galois.linutronix.de (Galois.linutronix.de [IPv6:2a0a:51c0:0:12e:550::1]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id CED5F1009; Wed, 30 Nov 2022 02:02:46 -0800 (PST) Date: Wed, 30 Nov 2022 10:02:43 -0000 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linutronix.de; s=2020; t=1669802565; h=from:from:sender:sender:reply-to:reply-to:subject:subject:date:date: message-id:message-id:to:to:cc:cc:mime-version:mime-version: content-type:content-type: content-transfer-encoding:content-transfer-encoding; bh=VHa/C1sCb9t0gkcdPzHNek3kN6cA/XbDouIGUpO2Jh0=; b=wDBcog33gzS4nuVQ2SXAoVBzqKGSpln78xCKHSNL7QXSzDPb3++aQRK5SuXG5DWSVhDH5T g4WdDZRf5byf4dpyDUp9BAP0vqB2H7J+BMhpGsQhSmGyEI/w3rwV0w0WDmzttH2AdpA2wX wbxufUOZah+OO+1Sw3BQ/6A/7twavcJV4vnF8kJh4dtTqoFmy64jXgDzo6lbK+Jj7rNvl1 q4UZRjH2qv2CaeRMJ4lx1xSUKQV3vuiPDJPpTNswfCexZKhcVUGsdC4H3u8TbpxN5plwVD H0cauGLq5nuWVHgREqg4ekWUoTWI7QiGDGybPDGNz1ju5IXjAcSeUcFxGaeNHA== DKIM-Signature: v=1; a=ed25519-sha256; c=relaxed/relaxed; d=linutronix.de; s=2020e; t=1669802565; h=from:from:sender:sender:reply-to:reply-to:subject:subject:date:date: message-id:message-id:to:to:cc:cc:mime-version:mime-version: content-type:content-type: content-transfer-encoding:content-transfer-encoding; bh=VHa/C1sCb9t0gkcdPzHNek3kN6cA/XbDouIGUpO2Jh0=; b=1Z8dlZojWtiexjhvYkR/+tCeG5WyFxk7nTbUJObfWkBkQRjVtyUESyonCGNNhufUlN2BxG Adt3FSPK7UI42ICw== From: "tip-bot2 for Peter Zijlstra" Sender: tip-bot2@linutronix.de Reply-to: linux-kernel@vger.kernel.org To: linux-tip-commits@vger.kernel.org Subject: [tip: perf/urgent] perf: Fix perf_pending_task() UaF Cc: syzbot+9228d6098455bb209ec8@syzkaller.appspotmail.com, "Peter Zijlstra (Intel)" , Marco Elver , x86@kernel.org, linux-kernel@vger.kernel.org MIME-Version: 1.0 Message-ID: <166980256366.4906.11174941629243947427.tip-bot2@tip-bot2> Robot-ID: Robot-Unsubscribe: Contact to get blacklisted from these emails Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org The following commit has been merged into the perf/urgent branch of tip: Commit-ID: 517e6a301f34613bff24a8e35b5455884f2d83d8 Gitweb: https://git.kernel.org/tip/517e6a301f34613bff24a8e35b5455884= f2d83d8 Author: Peter Zijlstra AuthorDate: Thu, 24 Nov 2022 12:49:12 +01:00 Committer: Peter Zijlstra CommitterDate: Tue, 29 Nov 2022 17:42:49 +01:00 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: syzbot+9228d6098455bb209ec8@syzkaller.appspotmail.com Signed-off-by: Peter Zijlstra (Intel) Tested-by: Marco Elver --- kernel/events/core.c | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-) diff --git a/kernel/events/core.c b/kernel/events/core.c index 9d15d2d..ad82479 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -2291,6 +2291,7 @@ event_sched_out(struct perf_event *event, !event->pending_work) { event->pending_work =3D 1; dec =3D 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, =20 #define DETACH_GROUP 0x01UL #define DETACH_CHILD 0x02UL +#define DETACH_DEAD 0x04UL =20 /* * 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); } =20 + /* + * 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 =3D 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 =3D PERF_EVENT_STATE_DEAD; =20 if (!ctx->nr_events && ctx->is_active) { if (ctx =3D=3D &cpuctx->ctx) @@ -5121,9 +5131,7 @@ int perf_event_release_kernel(struct perf_event *even= t) =20 ctx =3D perf_event_ctx_lock(event); WARN_ON_ONCE(ctx->parent_ctx); - perf_remove_from_context(event, DETACH_GROUP); =20 - 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 *even= t) * Thus this guarantees that we will in fact observe and kill _ALL_ * child events. */ - event->state =3D PERF_EVENT_STATE_DEAD; - raw_spin_unlock_irq(&ctx->lock); + perf_remove_from_context(event, DETACH_GROUP|DETACH_DEAD); =20 perf_event_ctx_unlock(event, ctx); =20 @@ -6577,6 +6584,8 @@ static void perf_pending_task(struct callback_head *h= ead) if (rctx >=3D 0) perf_swevent_put_recursion_context(rctx); preempt_enable_notrace(); + + put_event(event); } =20 #ifdef CONFIG_GUEST_PERF_EVENTS