OS PrimerOperating systems, in depth, for senior interviews

Build the mental models first, then learn why they matter in real systems and how interviewers probe them.

Start with Virtual Memory

How to use this primer

Chapters

I. Foundations

1What an OS IsKernel vs user mode, privilege rings · System calls: the mechanism and the cost · Interrupts, exceptions and traps · Monolithic vs microkernel · The vDSO 2Processes & Threadsfork / exec / wait · Process states and the PCB (task_struct) · Threads vs processes: what is shared · Signals, SIGTERM vs SIGKILL, graceful shutdown · Zombies, orphans, reparenting, process groups and sessions 3Linking, Loading & Program StartupELF: sections and segments · Static vs dynamic linking · PLT / GOT and lazy binding · LD_PRELOAD and symbol interposition · How exec builds the address space

II. Memory

4Virtual MemoryPages, frames and address translation · Multi-level page tables · The TLB, PCID and shootdowns · Page faults and demand paging · Huge pages 5Kernel Memory Managementmmap: file-backed vs anonymous · Copy-on-write and fork · Page cache and reclaim (LRU lists, kswapd) · Swap, overcommit and the OOM killer · RSS vs VSZ vs PSS 6User-Space AllocatorsHow malloc gets memory (brk, mmap) · Free lists, size classes, arenas · Fragmentation: internal and external · jemalloc / tcmalloc / mimalloc · Why RSS does not shrink after free()7CPU Caches & NUMACache hierarchy and latency numbers · Cache lines, locality, prefetching · Coherence (MESI) and false sharing · NUMA topology and locality · Data-oriented design in practice

III. CPU & Concurrency

8Scheduling & Context SwitchesFCFS, SJF, round robin, MLFQ · Linux CFS and EEVDF · Context switch: what is saved and what it costs · Priorities, nice, real-time classes · CPU affinity and cgroup CPU limits 9Concurrency I: Locks & DeadlockRace conditions and critical sections · Mutexes, spinlocks, condition variables · Semaphores and monitors · Deadlock: conditions, prevention, detection · Priority inversion 10Concurrency II: Atomics & Memory OrderingAtomic operations and CAS · Memory ordering: acquire / release / seq_cst · The ABA problem · futex: how user-space locks block · Reader-writer locks and RCU 11Inter-Process CommunicationPipes and FIFOs · Unix domain sockets and fd passing · Shared memory · Message queues · Choosing an IPC mechanism

IV. I/O & Storage

12I/O ModelsBlocking vs non-blocking vs async · select / poll / epoll (level vs edge triggered) · io_uring · Zero-copy: sendfile, splice · The C10K problem and event loops 13File SystemsInodes, directories, links · File descriptors and the open file table · The page cache and write-back · fsync, fdatasync, O_DIRECT · Journaling and crash consistency 14The Storage StackThe block layer and I/O schedulers · HDD vs SSD vs NVMe · SSD internals: FTL, write amplification · What durability actually guarantees · RAID basics 15Networking in the KernelSockets and socket buffers · TCP states, TIME_WAIT · SYN and accept queues, backlog · From syscall to NIC: the packet path · Interrupts, NAPI, RSS, kernel bypass

V. Systems

16Time & TimersWall clock vs monotonic clock · clock_gettime and the vDSO · Timer interrupts, tickless kernels · hrtimers and timer wheels · Clock skew and NTP 17Security & IsolationUsers, permissions, capabilities · seccomp · ASLR, stack canaries, NX · Spectre, Meltdown and KPTI · Why syscalls got more expensive 18Virtualization & ContainersHypervisors: type 1 vs type 2 · Hardware virtualization and nested paging · Namespaces · cgroups · Containers vs VMs vs microVMs 19Performance & DebuggingThe USE method · perf and flame graphs · strace, ltrace, bpftrace · Reasoning about tail latency · Latency numbers every engineer should know

Capstone & Appendix

CWhat Happens When……you run ./a.out · …you call read() on a file · …a server accepts a connection · …a process runs out of memory AInterview Question BankAll questions, indexed by chapter