D686 Operating Systems for Computer Scientists - Set 2 - Part 2

Correct Answer: Biometric scanning

Biometric factors like fingerprints or facial recognition verify physical identity, providing strong authentication when combined with knowledge-based passwords for access control.
Option a is incorrect because security questions are another knowledge factor like passwords, not providing true two-factor authentication with different factor types.
Option c is incorrect because PIN codes are knowledge factors similar to passwords, failing to add different authentication factor type for security.
Option d is incorrect because while tokens provide possession factor, biometrics offer stronger identity binding since physical characteristics cannot be shared.

Correct Answer: Cache consistency

Cache consistency problems occur when cached copies become stale after modifications, requiring protocols to invalidate or update client caches maintaining coherence.
Option a is incorrect because network latency affects performance but doesn't cause consistency problems; it only delays propagation of updates between systems.
Option c is incorrect because bandwidth limitation restricts data transfer rates but doesn't create consistency issues between cached copies of files.
Option d is incorrect because protocol mismatch prevents communication entirely rather than causing subtle consistency problems between successfully communicating clients' caches.

Correct Answer: Data corruption

Data corruption occurs when preemption interrupts critical sections, leaving shared data partially modified and inconsistent for other processes accessing it subsequently.
Option a is incorrect because memory leaks result from failing to free allocated memory, not from preemption during data structure updates specifically.
Option c is incorrect because stack overflow occurs from excessive recursion or local variables, unrelated to preemption timing during shared data updates.
Option d is incorrect because null pointers result from uninitialized or freed pointers, not from interrupting processes during data structure modifications particularly.

Correct Answer: Process suspension

Process suspension keeps app state in memory while stopping CPU execution, enabling quick resumption while conserving battery by eliminating processing overhead.
Option a is incorrect because termination removes apps from memory completely, requiring full restart and losing state when users return to them.
Option c is incorrect because migration moves processes between processors or systems, not relevant for battery conservation in mobile device scenarios typically.
Option d is incorrect because duplication creates process copies for redundancy or parallelism, increasing rather than decreasing resource consumption for battery saving.

Correct Answer: Random access capability

Block devices support random access to any block location with buffering, while character devices provide sequential byte streams without seeking capability.
Option a is incorrect because speed varies within device categories; the distinction is access pattern not transfer rate between block and character devices.
Option c is incorrect because network support depends on device drivers and protocols, not the fundamental block versus character device classification distinction.
Option d is incorrect because encryption can apply to both device types; it's not the distinguishing characteristic between block and character devices.

Correct Answer: System-enforced policies

MAC enforces system-wide security policies that users cannot override, unlike DAC where resource owners control access permissions at their discretion.
Option a is incorrect because user-controlled permissions characterize DAC not MAC; MAC removes discretionary control from users for security policy enforcement.
Option c is incorrect because performance impact varies by implementation; the fundamental difference is policy control not execution speed between models.
Option d is incorrect because while MAC may be complex, the defining difference is mandatory enforcement versus discretionary user control of permissions.

Correct Answer: Lower overhead

Shared memory eliminates copying data between address spaces, providing fastest IPC for frequent exchanges by allowing direct memory access between processes.
Option a is incorrect because shared memory requires careful synchronization and protection, potentially less secure than isolated message passing communication channels.
Option c is incorrect because shared memory requires complex synchronization to prevent race conditions, more complicated than message passing queue mechanisms.
Option d is incorrect because shared memory requires explicit synchronization through semaphores or locks; message passing provides implicit synchronization through send/receive.

Correct Answer: Direct access support

Indexed allocation enables direct access to any block through index lookups while avoiding contiguous allocation's fragmentation and linked allocation's sequential limitations.
Option a is incorrect because indexed allocation primarily solves access problems; external fragmentation is specifically contiguous allocation's issue not linked allocation's.
Option c is incorrect because internal fragmentation can occur in any method when block sizes don't match data; it's not specifically solved.
Option d is incorrect because file size limits exist in all methods based on addressing; indexed allocation may have different limits not elimination.

Correct Answer: Indefinite starvation

Indefinite starvation occurs when low-priority processes wait forever as new high-priority processes continuously arrive and receive CPU time first.
Option a is incorrect because convoy effect involves short processes waiting behind long ones in FCFS, not priority-based scheduling starvation issues.
Option b is incorrect because priority inversion occurs when high-priority waits for low-priority holding resources, different from starvation without resources.
Option d is incorrect because deadline miss relates to real-time scheduling failing time constraints, not general priority scheduling starvation problems specifically.

Correct Answer: Side-channel attack

Side-channel attacks exploit timing differences from speculative execution accessing forbidden memory, leaking information through cache timing analysis despite permission denial ultimately.
Option a is incorrect because buffer overflow involves writing beyond allocated bounds, not exploiting speculative execution timing for information leakage attacks.
Option c is incorrect because SQL injection targets database queries through malicious input, unrelated to processor speculative execution vulnerabilities in hardware.
Option d is incorrect because denial of service prevents legitimate access through resource exhaustion, not information leakage through speculative execution timing.

Correct Answer: Process tracing

Process tracing facilities like ptrace allow debuggers to attach, control execution, and examine state of traced processes for debugging purposes.
Option a is incorrect because network sockets enable inter-process communication over networks, not the direct process control required for debugging operations.
Option c is incorrect because file monitoring tracks file system changes but doesn't provide process execution control needed for debugging functionality requirements.
Option d is incorrect because memory mapping provides file access methods, not the process control and introspection capabilities debuggers require for operation.

Correct Answer: Split-brain syndrome

Split-brain syndrome occurs when network partitions create multiple active clusters believing others failed, causing data inconsistency when partitions heal eventually.
Option a is incorrect because clock drift involves time synchronization issues over time, not the immediate consistency problems from network partition isolation.
Option c is incorrect because resource leaks involve unreleased resources accumulating over time, not partition-induced inconsistency in distributed systems specifically occurring.
Option d is incorrect because cascade failure involves sequential component failures propagating through system, different from partition-based split-brain consistency issues fundamentally.

Correct Answer: Secure boot

Secure boot cryptographically verifies boot component signatures before loading, establishing chain of trust from firmware through operating system preventing malware injection.
Option b is incorrect because fast boot optimizes startup speed through hibernation or caching, not providing security verification of boot components loaded.
Option c is incorrect because network boot loads OS from network servers for deployment, not specifically providing security verification during boot process.
Option d is incorrect because dual boot allows multiple OS installation choices at startup, unrelated to security verification of boot components loaded.

Correct Answer: Rate monotonic scheduling

Rate monotonic scheduling assigns higher priorities to tasks with shorter periods, providing optimal fixed-priority scheduling for periodic real-time tasks theoretically.
Option a is incorrect because earliest deadline first uses dynamic priorities based on deadlines, not fixed priorities based on task periods specifically.
Option c is incorrect because lottery scheduling uses randomized proportional share allocation, inappropriate for deterministic real-time deadline requirements in systems.
Option d is incorrect because multilevel feedback queues adjust priorities based on behavior, lacking predictability required for real-time scheduling guarantees needed.

Correct Answer: Eventual consistency

Eventual consistency permits temporary divergence between replicas for availability and performance, guaranteeing convergence when updates cease, suitable for distributed systems.
Option a is incorrect because strong consistency requires immediate synchronization of all replicas, sacrificing availability and performance for consistency in distributed systems.
Option c is incorrect because causal consistency preserves causally related operation ordering, stronger than eventual but weaker than strong consistency models available.
Option d is incorrect because sequential consistency requires all processes see operations in same order, stronger requirement than eventual consistency provides typically.

Correct Answer: Software fault isolation

Software fault isolation uses compiler techniques and runtime checks to enforce memory boundaries without hardware MMU support, protecting memory in embedded systems.
Option a is incorrect because software segments still require some form of address translation or bounds checking, typically needing hardware support for efficiency.
Option b is incorrect because memory pools organize allocation but don't provide protection between different code modules without additional isolation mechanisms present.
Option c is incorrect because static allocation determines memory layout at compile time but doesn't prevent access violations during runtime execution happening.

Correct Answer: Unified memory

Unified memory architecture allows CPU and GPU to access same physical memory directly, eliminating explicit data copying between separate memory spaces.
Option a is incorrect because memory compression reduces memory usage through compression algorithms, not addressing CPU-GPU memory sharing for graphics applications specifically.
Option c is incorrect because virtual memory provides address space abstraction but doesn't specifically enable CPU-GPU memory sharing without copying data.
Option d is incorrect because cache memory speeds repeated access to data but doesn't eliminate transfers between separate CPU and GPU memory spaces.

Correct Answer: Shared kernel

Containers share host OS kernel while isolating user space, providing lighter weight virtualization than VMs which virtualize entire OS stack completely.
Option a is incorrect because hardware requirements vary by workload; the fundamental difference is architectural approach not resource needs between technologies compared.
Option c is incorrect because both containers and VMs can have various network configurations; this isn't the defining architectural difference between them.
Option d is incorrect because storage management varies by implementation in both technologies; the key difference is kernel sharing versus full OS virtualization.

Correct Answer: Concurrent access paths

Parallel file systems enable multiple clients to read/write different parts simultaneously through parallel data paths, maximizing I/O bandwidth for HPC workloads.
Option a is incorrect because encryption support exists in many file systems; it's not the distinguishing feature of parallel file systems specifically designed.
Option c is incorrect because compression algorithms can be implemented in various file systems, not unique to parallel file system architectures fundamentally.
Option d is incorrect because backup features are independent of parallel architecture; the key distinction is concurrent access scalability for performance gains.

Correct Answer: Resource limits

Resource limits (ulimits) enforce maximum consumption boundaries for various resources, preventing single processes from monopolizing system resources and affecting others negatively.
Option a is incorrect because access controls manage permissions for files and objects, not limiting computational resource consumption by processes directly enforced.
Option c is incorrect because priority levels affect scheduling order but don't enforce hard limits on resource consumption amounts by processes.
Option d is incorrect because scheduling classes determine how CPU time is allocated but don't limit total consumption of various resources comprehensively.