Side channels are typically viewed as attacks that leak cryptographic keys during cryptographic algorithm processing, by observation of system side effects. In this paper, we present new side channels that leak password information during X Windows keyboard processing of password input. Keylogging is one approach for stealing passwords, but current keylogging techniques require special hardware or privileged processes. However, we have found that the unprivileged operation of modifying the user key mappings for X Windows clients enables a side channel sufficient for unprivileged processes to steal that user's passwords, even enabling the attacker to gain root access via sudo. We successfully tested one version on Linux 2.6; we were able to obtain a high degree of control over the scheduler, and thus we can obtain accurate timing information. A second version (logon detection) works without depending on accurate clocks or cache effects. Thus, in addition to demonstrating new side channels, we show that (a) side channels cannot be eliminated by removing accurate clocks or hardware cache mechanisms (b) side channels are of continued concern for computer security as well as cryptographic processing.

Presents new attacks against a user workstation's trusted path mechanism. These timing attacks can cause a user's password to leak bits. The timing attacks can then be combined with network authentication protocol brute-force attacks against the remainder of the key space to obtain the user's password. We present several countermeasures against this attack. We also define a property of user systems (workstations) called Trojan horse non-persistence. Workstations that fail to have this properly are more vulnerable to the timing attack and other Trojan horse attacks.

While there has been considerable effort in creating a single sign-on solution for interoperability among authentication methods, such interoperability across authorization methods has received little attention. This paper presents a flexible distributed authorization protocol that provides the full generality of restricted proxies while supporting the functionality of and interoperability with existing authorization models including OSF DCE and SESAME V2. Our authorization protocol includes a delegation method that is well suited for certain electronic commerce applications.

The author gives an analysis of the fuzzy time covert channel reduction technique developed at the Digital Equipment Corporation (DEC) and applied to the VAX security kernel for both a single computer system and a network of hosts. Specifically, he investigates covert channels in a model of a fuzzy time system and in a network model of n fuzzy time hosts and one nonfuzzy time host. The fuzzy time operating system model predicts a covert channel with a datarate of roughly 50 b/s, given a mean time between upticks of 20 ms as in the VAX security kernel. Several countermeasures including scheduler redesign are suggested, which will close or reduce the channels that are analyzed.>

We analyze the DEC fuzzy time technique (Hu [HU]) for both a single computer system and a network of hosts. Specifically, we investigate covert channels in a model of a fuzzy time system and in a network model of n fuzzy time hosts and one nonfuzzy t

The multiple Trojan horse system is a finite-state deterministic model with a high-level transmitter Trojan horse process, a low-level receiver Trojan horse process, and m conspiring Trojan horse processes. The conspiring processes attempt to aid the downgrading of data from the transmitter to the receiver. The capacity of a multiple Trojan horse system is a measure of the amount of information the high-level process can disclose to the low-level process. Any controlled signalling system can be viewed as a multiple Trojan horse system. The author gives an optimality equation which gives a way to recursively construct optimal strategies for both the receiver and conspiring processes. Computer algorithms can be constructed to iteratively compute the capacity and optimal strategies.>