4.2 Description of the First Documented Facts …
301
Of course, we are aware of the existence of the special security function (MAC)
designed to prevent such random recording in the AES mode through data verification; however, the authors [110] successfully cracked it, finding out in the process
that the feedback shift register used here contains only four bits of errors on the AES
CBS block, which operates independently (autonomously). Moreover, the authors
of [110] even managed to block the MAC verification commands, changing only
several bits in the so-called STAPL file, which is designed to prevent such random
recording. As we can see, too many statements of authoritative manufacturers of
critical microcircuits are not supported by the results of expert studies.
4.2.3 Brief Overview of the Method of Experimental
Detection of a Hardware Trojan in the A3P250Actel
Microcircuit
This section presents the results of analysis of the microcircuit A3P250, which is
widely used for commercial and military purposes, manufactured by Actel—another
company suffering from registered attacks of “unidentified intruders” [116, 119–
123]. The first phase of the studies performed by the authors included analysis of
a well-known sample chip of the A3P250 microcircuit, which was designed with
the help of standard packets Actel-Libeo IDE and FlashPro. As demonstrated above,
even though all operations of the JTAG protocol are undocumented for ProASIC3,
the authors of [110] still managed to generate the entire series of the required STAPL
files, which helped to easily identify all the commands usually used for execution of
various standard operations from this microcircuit. The researchers designed a special
test board, which allowed implementation of the main functions of the standard JTAG
protocol and certain other simple functions controlled by the PC software directly via
the standard RS-232 interface (Fig. 4.17a). During the studies, the chip of ProASIC3
was placed on a ZF board. During this stage, the necessary information about specific
codes of the command fields and data registers was gathered. Only standard DPA
settings were used in order to analyze the nature of information transmitted via site
channels (technical data leakage channels) from the ProASCI3 device during implementation of command decoding and standard access operations, as well as identify
other possibly undocumented commands. For this purpose, another relatively simple
prototype board with the same ZIF panel was designed (Fig. 4.17b), which was
connected to the main test board in order to be able to see the end study results on
the oscilloscope screen. The power consumption was measured using the resistor
connected to the power line of the microcircuit (20 Q) in the V power line using
an Agilent 1130A differential probe and a digital recording oscilloscope synchronized with Agilent MSO8104A. The resulting oscillograms were then analyzed using
standard MATLAB software [116, 119–124].
During this analysis, all available fields of JTAG commands in different combinations were examined, as well as all valid responses received as a result of scanning
301
Of course, we are aware of the existence of the special security function (MAC)
designed to prevent such random recording in the AES mode through data verification; however, the authors [110] successfully cracked it, finding out in the process
that the feedback shift register used here contains only four bits of errors on the AES
CBS block, which operates independently (autonomously). Moreover, the authors
of [110] even managed to block the MAC verification commands, changing only
several bits in the so-called STAPL file, which is designed to prevent such random
recording. As we can see, too many statements of authoritative manufacturers of
critical microcircuits are not supported by the results of expert studies.
4.2.3 Brief Overview of the Method of Experimental
Detection of a Hardware Trojan in the A3P250Actel
Microcircuit
This section presents the results of analysis of the microcircuit A3P250, which is
widely used for commercial and military purposes, manufactured by Actel—another
company suffering from registered attacks of “unidentified intruders” [116, 119–
123]. The first phase of the studies performed by the authors included analysis of
a well-known sample chip of the A3P250 microcircuit, which was designed with
the help of standard packets Actel-Libeo IDE and FlashPro. As demonstrated above,
even though all operations of the JTAG protocol are undocumented for ProASIC3,
the authors of [110] still managed to generate the entire series of the required STAPL
files, which helped to easily identify all the commands usually used for execution of
various standard operations from this microcircuit. The researchers designed a special
test board, which allowed implementation of the main functions of the standard JTAG
protocol and certain other simple functions controlled by the PC software directly via
the standard RS-232 interface (Fig. 4.17a). During the studies, the chip of ProASIC3
was placed on a ZF board. During this stage, the necessary information about specific
codes of the command fields and data registers was gathered. Only standard DPA
settings were used in order to analyze the nature of information transmitted via site
channels (technical data leakage channels) from the ProASCI3 device during implementation of command decoding and standard access operations, as well as identify
other possibly undocumented commands. For this purpose, another relatively simple
prototype board with the same ZIF panel was designed (Fig. 4.17b), which was
connected to the main test board in order to be able to see the end study results on
the oscilloscope screen. The power consumption was measured using the resistor
connected to the power line of the microcircuit (20 Q) in the V power line using
an Agilent 1130A differential probe and a digital recording oscilloscope synchronized with Agilent MSO8104A. The resulting oscillograms were then analyzed using
standard MATLAB software [116, 119–124].
During this analysis, all available fields of JTAG commands in different combinations were examined, as well as all valid responses received as a result of scanning