Hardware Security Day 4 – Physical Attacks, Fault Injection, and Secure Boot

0.0(0)
Studied by 0 people
call kaiCall Kai
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
GameKnowt Play
Card Sorting

1/38

flashcard set

Earn XP

Description and Tags

Weeks 9–10 and the Week 3 VoltPillager case study. Compare physical attack categories, explain how glitches can cause hardware faults, and practice connecting each fault to its security impact and a matching defense.

39 Terms

1
New cards

Physical attack

A hardware security attack that uses physical access or interaction to observe a device or interfere with its operation.

2
New cards

Non-invasive attack

Uses external interfaces or measurements while leaving the chip package intact. External power measurement is an example.

3
New cards

Semi-invasive attack

Exposes the die and interacts with it without direct electrical contact with internal circuitry. Optical fault injection is an example.

4
New cards

Invasive attack

Directly probes or modifies structures inside a chip. Microprobing and focused ion beam edits are examples.

5
New cards

Decapsulation

Opening or removing a chip's protective package to expose the die.

6
New cards

Deprocessing

Removing layers from an exposed chip to reveal internal structures.

7
New cards

Microprobing

Using a fine probe to make electrical contact with an internal chip node for measurement or manipulation.

8
New cards

Focused ion beam (FIB)

A tool that uses a focused beam of ions to remove material or modify structures within a chip.

9
New cards

Fault injection

Deliberately disturbing a device so that it produces an incorrect result or changes its state or control flow.

10
New cards

Fault injection vs. side-channel observation

Fault injection changes device behavior. Side-channel observation infers information from effects the device produces during operation.

11
New cards

Glitch

A brief disturbance in a physical condition such as clock timing or supply voltage.

12
New cards

Clock glitch

A brief clock disturbance that can leave a circuit too little time to finish a computation before its result is captured.

13
New cards

Voltage glitch

A brief change in supply voltage that can disrupt circuit operation. Undervolting can slow switching enough to cause timing errors.

14
New cards

Optical fault injection

Using a pulse of light to disturb circuitry in a localized region of a chip.

15
New cards

Timing violation

A signal arrives too late or changes too close to the moment it is captured. The circuit may record the wrong value.

16
New cards

Fault model

A description of the effect a disturbance has on computation. Examples include a bit flip or an instruction skip.

17
New cards

Bit flip

A fault that changes a bit from 0 to 1 or from 1 to 0.

18
New cards

Instruction skip

A fault effect in which an instruction fails to execute as expected. Skipping a branch can change which code runs next.

19
New cards

Delayed-write fault

A write reaches its destination later than expected. A later operation may observe an old value.

20
New cards

Physical method vs. fault model vs. security impact

A voltage glitch is a physical method. An instruction skip is a possible fault model. Failure to enforce a security check is a possible impact.

21
New cards

Why a glitch can have different outcomes

The effect depends on which operation is active and how the device responds. The device may continue normally or crash or produce a fault.

22
New cards

Secure boot

A boot process in which a trusted stage verifies the next stage before allowing it to run.

23
New cards

Hardware root of trust

A trusted hardware-backed starting point for security decisions such as verifying the first boot stage.

24
New cards

Chain of trust

Each trusted boot stage verifies the next stage before transferring control to it.

25
New cards

Signature verification in secure boot

A check that the image was signed by an authorized signer and has remained intact since signing.

26
New cards

Verification vs. enforcement

Verification determines whether an image is valid. Enforcement uses that result to decide whether execution is permitted.

27
New cards

Secure boot control-flow fault

A fault can disrupt enforcement even when signature verification correctly detects an invalid image.

28
New cards

Fail-closed behavior

The device blocks boot or halts safely if verification fails or its result cannot be trusted.

29
New cards

Redundant verification

Repeating a security check or checking its result independently so a single fault is less likely to cause unsafe execution.

30
New cards

Common-mode fault

A single disturbance affects multiple safeguards in the same way. Identical redundant checks can therefore fail together.

31
New cards

Fault sensor

Hardware that detects conditions such as abnormal voltage or clock behavior and can trigger a protective response.

32
New cards

Control-flow hardening

Designing critical code so that a skipped instruction or branch is less likely to permit an unauthorized action.

33
New cards

Intel SGX enclave

A protected execution environment designed to isolate code and data from other software on the system.

34
New cards

Voltage regulator

A component that supplies and adjusts the voltage delivered to the processor.

35
New cards

Serial Voltage Identification (SVID)

The communication interface between the processor and its voltage regulator discussed in the VoltPillager case.

36
New cards

VoltPillager

A board-level case study showing that hardware undervolting through the SVID path can induce faults in computations running inside Intel SGX enclaves.

37
New cards

Plundervolt

A case study showing that software-controlled undervolting can induce faults in Intel SGX computations.

38
New cards

VoltPillager vs. Plundervolt

VoltPillager targets the board-level voltage control path. Plundervolt uses a software-accessible voltage control path.

39
New cards

VoltPillager defense lesson

Restricting software voltage controls addresses the software path. Board-level access to the regulator interface is a separate part of the threat model.