Re: [Xen-devel] [PATCH v4] xen/tools: Introduce QNX IFS loader

[Ian Jackson <Ian.Jackson@xxxxxxxxxxxxx>]

Oleksandr Tyshchenko writes ("Re: [PATCH v4] xen/tools: Introduce QNX IFS 
loader"):
> On Fri, Sep 26, 2014 at 5:37 PM, Ian Jackson <Ian.Jackson@xxxxxxxxxxxxx> 
> wrote:
> > Suppose that the incoming image is corrupt or malicious and
> > startup_header.startup_size and dom->kernel_size are both equal to
> > sizeof(startup_header)+1.
...
> ok. Maybe, do you mean that (stored_size == kernel_size) instead of
> (startup_size == kernel_size)?

You are right that I had failed to properly analyse the condition on
startup_size and stored_size.  I guess that just goes to show how hard
this is.

But I think there is still an attack.  Consider:
  startup_size == 3;
  stored_size == kernel_size == sizeof(startup_header);

Then the first calc_checksum gets 3 as a size argument and loops
forever scanning memory until it crashes.

Ian.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel