Re: [Bug 212995] New: A possible divide by zero in qh_urb_transaction

[Greg KH <greg@kroah.com>]

On Sat, May 08, 2021 at 11:14:25AM +0000, bugzilla-daemon@bugzilla.kernel.org wrote:
> https://bugzilla.kernel.org/show_bug.cgi?id=212995
> 
>             Bug ID: 212995
>            Summary: A possible divide by zero in qh_urb_transaction
>            Product: Drivers
>            Version: 2.5
>     Kernel Version: 5.12.2
>           Hardware: All
>                 OS: Linux
>               Tree: Mainline
>             Status: NEW
>           Severity: normal
>           Priority: P1
>          Component: USB
>           Assignee: drivers_usb@kernel-bugs.kernel.org
>           Reporter: yguoaz@gmail.com
>         Regression: No
> 
> In the file drivers/usb/host/ehci-q.c, the function qh_urb_transaction has the
> following code (link to the code location:
> https://github.com/torvalds/linux/blob/dd860052c99b1e088352bdd4fb7aef46f8d2ef47/drivers/usb/host/ehci-q.c#L715):
> 
> maxpacket = usb_maxpacket(urb->dev, urb->pipe, !is_input);
> ...
> if (usb_pipeout(urb->pipe) && (urb->transfer_flags & URB_ZERO_PACKET)
>                            && !(urb->transfer_buffer_length % maxpacket))
> 
> However, the function usb_maxpacket may retun 0, leading to a possible divide
> zero problem in the remainder operation:

Please feel free to submit a patch to the developers, but first see how
it could be possible for usb_maxpacket to ever return 0 for this.  Are
you sure it's not already sanitized by the time it gets here?

thanks,

greg k-h