Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Mon, 17 May 2021 15:43:24 +0200
Message-ID: <s5hk0nxo4qr.wl-tiwai@suse.de>
From:   Takashi Iwai <tiwai@suse.de>
To:     Hyeonggon Yoo <42.hyeyoo@gmail.com>
Cc:     Jaroslav Kysela <perex@perex.cz>, Takashi Iwai <tiwai@suse.com>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Oliver Neukum <oneukum@suse.com>,
        Vasily Khoruzhick <anarsoul@gmail.com>,
        alsa-devel@alsa-project.org, linux-kernel@vger.kernel.org
Subject: Re: [RFC PATCH] sound: line6: Fix race condition in line6_probe
In-Reply-To: <20210517132725.GA50495@hyeyoo>
References: <20210517132725.GA50495@hyeyoo>
User-Agent: Wanderlust/2.15.9 (Almost Unreal) SEMI/1.14.6 (Maruoka)
 FLIM/1.14.9 (=?UTF-8?B?R29qxY0=?=) APEL/10.8 Emacs/25.3
 (x86_64-suse-linux-gnu) MULE/6.0 (HANACHIRUSATO)
MIME-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka")
Content-Type: text/plain; charset=US-ASCII
Precedence: bulk

On Mon, 17 May 2021 15:27:25 +0200,
Hyeonggon Yoo wrote:
> 
> syzbot reported general protection fault in midibuf_is_full.
> the cause is linemidi pointer in struct usb_line6 isn't properly
> initialized.
> 
> the pointer isn't initialized because there is race condition
> in line6_probe. it calls line6_init_cap_control first, which submits urb.
> and then it initializes it's data using private_init function.
> 
> so it's possible line6_data_received is called before it's
> data isn't initialized.
> 
> Link: https://lkml.org/lkml/2021/5/17/543
> Reported-by: syzbot+0d2b3feb0a2887862e06@syzkallerlkml..appspotmail.com
> Signed-off-by: Hyeonggon Yoo <42.hyeyoo@gmail.com>

Thanks for the patch!  It's an issue I'm tracking now, too, and you
submitted quicker.  But, unfortunately, this swap does seem yet
another potential race between the private_init call and
line6_init_cap_control().  The actually needed initialization is
line6_init_mid() call, and this can be fixed by moving to the
appropriate place instead of inside each private_init callback.

So my current fix is like below.


thanks,

Takashi

---
diff --git a/sound/usb/line6/driver.c b/sound/usb/line6/driver.c
index a030dd65eb28..9602929b7de9 100644
--- a/sound/usb/line6/driver.c
+++ b/sound/usb/line6/driver.c
@@ -699,6 +699,10 @@ static int line6_init_cap_control(struct usb_line6 *line6)
 		line6->buffer_message = kmalloc(LINE6_MIDI_MESSAGE_MAXLEN, GFP_KERNEL);
 		if (!line6->buffer_message)
 			return -ENOMEM;
+
+		ret = line6_init_midi(line6);
+		if (ret < 0)
+			return ret;
 	} else {
 		ret = line6_hwdep_init(line6);
 		if (ret < 0)
diff --git a/sound/usb/line6/pod.c b/sound/usb/line6/pod.c
index cd44cb5f1310..16e644330c4d 100644
--- a/sound/usb/line6/pod.c
+++ b/sound/usb/line6/pod.c
@@ -376,11 +376,6 @@ static int pod_init(struct usb_line6 *line6,
 	if (err < 0)
 		return err;
 
-	/* initialize MIDI subsystem: */
-	err = line6_init_midi(line6);
-	if (err < 0)
-		return err;
-
 	/* initialize PCM subsystem: */
 	err = line6_init_pcm(line6, &pod_pcm_properties);
 	if (err < 0)
diff --git a/sound/usb/line6/variax.c b/sound/usb/line6/variax.c
index ed158f04de80..1376fc405c7f 100644
--- a/sound/usb/line6/variax.c
+++ b/sound/usb/line6/variax.c
@@ -172,11 +172,6 @@ static int variax_init(struct usb_line6 *line6,
 	if (variax->buffer_activate == NULL)
 		return -ENOMEM;
 
-	/* initialize MIDI subsystem: */
-	err = line6_init_midi(&variax->line6);
-	if (err < 0)
-		return err;
-
 	/* initiate startup procedure: */
 	schedule_delayed_work(&line6->startup_work,
 			      msecs_to_jiffies(VARIAX_STARTUP_DELAY1));