NetBSD all versions FireWire IOCTL kernel integer overflow information disclousure
11/15/2006

Notice
===================
    This bug has been specially discovered for the Month of Kernel Bugs and to
    the Hackers to Hackers Conference III (http://www.h2hc.org.br/en/).


Summary
===================

    Firewire device is enabled by default in the GENERIC kernel.  It defines
    an IOCTL function which can be malicious called passing a negative buffer
    lenght value.  This value will bypass the lenght check (because the value
    is negative) and will be used in a copyout operation.

Systems Affected
===================

    FreeBSD     all versions
    NetBSD      all versions
    DragonFly   all versions
    TrustedBSD* all versions

Impact
===================

    This is a kernel bug and the system can be compromised by local users and 
    important system informations can be discloused (basically, a mem dump ;) )

Explanation
===================

    Firewire interface can be tunned.  It provides an ioctl function receiving
    many parameters that can be changed.

    The follow is a code fragment from (FreeBSD - dev/firewire/fwdev.c (fw_ioctl    function) || DragonFlyBSD bus/firewire/fwdev.c (fw_ioctl function) || NetBSD    - dev/ieee1394/fwdev.c (FW_IOCTL function)) file:

	if (crom_buf->len < len)
		len = crom_buf->len;
	else
		crom_buf->len = len;

	err = copyout(ptr, crom_buf->ptr, len);

    We control the crom_buf->len (it's passed as argument to the ioctl function)
    so, passing it as a negative value will bypass this if statement (our value
    is minor than the default one).

    So, our value is used in a copyout function. ptr is defined before this
    copyout as:
	 if ( fwdev == NULL ) {
		...
		ptr = malloc(CROMSIZE, M_FW, M_WAITOK);
		...
	 } else {
		ptr = (void *)&fwdev->csrrom[0];
		...
	}

    This information disclousure lead an attacker dump all the system memory.

Solution
===================

     Attached in this advisory a patch for the FreeBSD 5.5 (it's pretty simple,      so, just need to be little changed to the other BSD's)

Timelife
===================

     11/15/2006 - Advisory Public Disclousure (sorry for the developers, but we 
     are just respecting the Month of Kernel Bugs Timelife)

Acknowledgments
===================

     Filipe Balestra <filipe@balestra.com.br> and Rodrigo Rubira Branco 
     (BSDaemon) <rodrigo@kernelhacking.com> for the discovering, analysis
     and patch.

Contact Information
===================

     You can reach the authors of this advisory by mail or visiting some
     websites:
	http://www.balestra.com.br  -> Personal Website of Filipe
	http://www.risesecurity.org -> RISE Security Research (Rodrigo is member     of the RISE Security Team)
	http://www.kernelhacking.com/rodrigo -> Personal Website of Rodrigo

References
===================

     http://www.kernelhacking.com/bsdadv1.txt -> Actual version of the advisory
     http://www.risesecurity.org/RISE-2006002.txt -> Related issue

Disclaimer (taken from teso-team)
===================

     This advisory does not claim to be complete or to be usable for any
     purpose. Especially information on the vulnerable systems may be
     inaccurate or wrong. The supplied exploit is not to be used for malicious
     purposes, but for educational purposes only.

     This advisory is free for open distribution in unmodified form.