Process Thread
Memory Separate address space Shared address space Creation Heavy (fork) Light (clone) IPC Pipes, sockets, shared memory Direct shared memory Context Switch Expensive (TLB flush) Cheap Crash Isolated Can kill entire process
Algorithm Preemptive Starvation Convoy Best For
FCFS No No Yes Batch SJF No Yes No Min avg wait SRTF Yes Yes No Interactive Round Robin Yes No No Time-sharing Priority Both Yes No Real-time MLFQ Yes No No General OS
Mutex: Binary lock, owner must unlock
Semaphore: Counter, any thread can signal
- Binary: like mutex (0 or 1)
- Counting: 0 to N (resource pool)
Critical Section Requirements:
1. Mutual Exclusion
2. Progress
3. Bounded Waiting
Four Conditions (ALL must hold):
1. Mutual Exclusion
2. Hold and Wait
3. No Preemption
4. Circular Wait
Prevention: Break one condition
Avoidance: Banker's Algorithm (safe state)
Detection: Resource Allocation Graph (cycle)
Recovery: Kill process / rollback / preempt
Paging:
- Fixed-size pages (4KB typical)
- No external fragmentation
- Page table maps VPN â PFN
- TLB caches translations
Segmentation:
- Variable-size segments
- Matches programmer's view
- External fragmentation possible
Page Replacement:
FIFO: Simple, Belady's anomaly
LRU: Optimal approximation, expensive
Clock: Practical LRU approximation
Optimal: Theoretical best, not implementable
Address Translation:
Virtual [VPN | Offset] â Physical [PFN | Offset]
Page Fault:
1. Process accesses page not in memory
2. OS finds page on disk
3. Load page into free frame
4. Update page table
5. Restart instruction
Thrashing: Working set > available frames
Solution: Reduce multiprogramming / add memory
Inode: Metadata structure for files
- Direct pointers (12): â data blocks
- Single indirect: â block of pointers
- Double indirect: â block â block of pointers
- Triple indirect: â block â block â block of pointers
Hard Link: Same inode, different name
Soft Link: Different inode, stores path
Journaling: WAL for crash recovery
- Write intent to journal
- Write data
- Mark journal complete
Mechanism Speed Use Case
Pipes Fast Parent-child Named Pipes Fast Unrelated processes Message Queue Medium Structured messages Shared Memory Fastest High-performance data Sockets Medium Network-capable Signals Fast Async notification
Context switch : Save PCB, load next PCB (~1-10 Ξs direct cost)System call : User mode â Kernel mode transitionCopy-on-write : fork() shares pages, copies on modificationZombie process : Terminated but not waited on by parentOrphan process : Parent terminated, adopted by init (PID 1)DMA : Direct Memory Access, bypasses CPU for I/ORace condition : Non-deterministic result from concurrent accessPriority inversion : High-priority waits on low-priority resource