ALSA: line6: Fix write on zero-sized buffer

BugLink: https://bugs.launchpad.net/bugs/1860681

commit 34501219 upstream.

LINE6 drivers allocate the buffers based on the value returned from
usb_maxpacket() calls.  The manipulated device may return zero for
this, and this results in the kmalloc() with zero size (and it may
succeed) while the other part of the driver code writes the packet
data with the fixed size -- which eventually overwrites.

This patch adds a simple sanity check for the invalid buffer size for
avoiding that problem.

Reported-by: syzbot+219f00fb49874dcaea17@syzkaller.appspotmail.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
[bwh: Backported to 4.4: Driver doesn't support asymmetrical packet
 sizes, so only check snd_line6_pcm::max_packet_size]
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com>
Signed-off-by: Khalid Elmously <khalid.elmously@canonical.com>

sound/usb/line6/pcm.c

@@ -529,6 +529,11 @@ int line6_init_pcm(struct usb_line6 *line6,
 				usb_rcvisocpipe(line6->usbdev, ep_read), 0),
 			usb_maxpacket(line6->usbdev,
 				usb_sndisocpipe(line6->usbdev, ep_write), 1));
+	if (!line6pcm->max_packet_size) {
+		dev_err(line6pcm->line6->ifcdev,
+			"cannot get proper max packet size\n");
+		return -EINVAL;
+	}
 
 	spin_lock_init(&line6pcm->out.lock);
 	spin_lock_init(&line6pcm->in.lock);