Re: [PATCH 8/8] xhci: Fix NULL pointer dereference in xhci_clear_tt_buffer_complete()

[Mathias Nyman <mathias.nyman@linux.intel.com>]

On 8.10.2019 11.15, Mathias Nyman wrote:
> On 7.10.2019 17.02, Johan Hovold wrote:
>>
>> I didn't have time to look into this myself last week, or comment on the
>> patch before Greg picked it up, but this clearly isn't the right fix.
>>
>> As your comment suggests, ep->hcpriv may indeed be NULL here if USB core
>> have allocated a new udev. But this only happens after USB has freed the
>> old usb_device and the new one happens to get the same address.
>>
> 
> You're right, that fix doesn't solve the actual issue, it avoids a few specific
> null pointer dereference cases, but leaves both root cause and several other
> use-after-free cases open.
> 
>> Note that even the usb_host_endpoint itself (ep) has then been freed and
>> reallocated since it is member of struct usb_device, and it is the
>> use-after-free that needs fixing.
>>
>> I've even been able to trigger another NULL-deref in this function
>> before a new udev has been allocated, due to the virt dev having been
>> freed by xhci_free_dev as part of usb_release_dev:
>>
>> It seems the xhci clear-tt implementation was incomplete since it did
>> not take care to wait for any ongoing work before disabling the
>> endpoint. EHCI does this in ehci_endpoint_disable(), but xhci doesn't
>> even implement that callback.
>>
> 
> So it seems, it might be possible to remove pending clear_tt work for
> most endpoints in the .drop_endpoint callbacks, but ep0 is different,
> it isn't dropped, we might need to implement the endpoint_disable()
> callback for this.
> 

I was able to reproduce the use-after-free issue by faking a endpoint halt,
and resetting the endpoint early in enumeration at Get device descriptor request.

To fix this I added the endpoint_disable() callback that will wait for
clear_tt_work to finish before returning, similar to the ehci solution.
It works in my case.

the endpoint_disable() callback is called by usb core both before dropping
xhci endpoints belonging to a interface, and separately for ep0 before udev is
freed during enumeration retry.

Both the hack that triggers the issue (LS/FS behind HS bug won't ever enumerate with this)
and the fix to prevent use after free can be found in  clear_tt_fix branch:

git://git.kernel.org/pub/scm/linux/kernel/git/mnyman/xhci.git clear_tt_fix

I'll send the fix out to the list as well, any chance you could try that out?

-Mathias