Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:5 as permitted sender) client-ip=2620:137:e000::3:5;
MIME-Version: 1.0
References: <20231117072131.2886406-1-khtsai@google.com> <a4a129a3-e271-acbb-949c-534a8e1627ee@linux.intel.com>
 <CAKzKK0rnx+tSFAj6N-U_vcAZ_5P=Hx_Kb97NFkdPMHs8dR3Ukw@mail.gmail.com>
 <a970f296-da67-9a80-ab2f-a94fd16e0fd9@linux.intel.com> <51b8fc3d-25ef-1ab3-d744-8d851a133828@linux.intel.com>
In-Reply-To: <51b8fc3d-25ef-1ab3-d744-8d851a133828@linux.intel.com>
From:   Kuen-Han Tsai <khtsai@google.com>
Date:   Tue, 28 Nov 2023 23:01:17 +0800
Message-ID: <CAKzKK0r_a_8uWTxjupDwwHH9hyCUd_NNTojL=209FiEzrcK9fw@mail.gmail.com>
Subject: Re: [PATCH] xhci: fix null pointer deref for xhci_urb_enqueue
To:     Mathias Nyman <mathias.nyman@linux.intel.com>
Cc:     mathias.nyman@intel.com, gregkh@linuxfoundation.org,
        linux-usb@vger.kernel.org, linux-kernel@vger.kernel.org,
        alan Stern <stern@rowland.harvard.edu>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Thank you so much for fixing the issue, Mathias!

> I moved the max packet checks away from xhci_urb_enqueue() and fixed up the locking.
> I can't trigger the original issue, but I tested it by setting incorrect initial max packet
> size values.

I added a 3-seconds delay within xhci_check_maxpacket(). When I saw
the max packet size was being checked, I removed the USB device to
trigger the race problem.

[  172.392813][ T1960] [khtsai] xhci_check_maxpacket, before,
slot_id=2, devs[slot_id]=000000003cb76fec
[  174.290601][   T20] usb 2-1: USB disconnect, device number 2
[  174.290608][   T20] usb 2-1.2: USB disconnect, device number 3
[  174.297180][   T20] [khtsai] xhci_free_dev, ret=1
[  174.305010][  T133] usb usb3: USB disconnect, device number 1
[  174.316346][   T20] [khtsai] xhci_free_dev, ret=1
[  175.458962][ T1960] [khtsai] xhci_check_maxpacket, after,
slot_id=2, devs[slot_id]=0000000000000000
[  175.460835][ T1960] Unable to handle kernel NULL pointer
dereference at virtual address 0000000000000010

> If you have the chance to test this with your setup I'd appreciate it.

Sure, I will definitely help verify it. However, I believe the race
problem won't happen as your patch already removes max packet checks
from xhci_urb_enqueue() and also protects sections using the
xhci->devs[slot_id] virtual device.

> patches found here:
> git://git.kernel.org/pub/scm/linux/kernel/git/mnyman/xhci.git fix_urb_enqueue_locking
> https://git.kernel.org/pub/scm/linux/kernel/git/mnyman/xhci.git/log/?h=fix_urb_enqueue_locking

I'll add them to this thread as well