Deep Dive — Study Catalog

Reference Books

Part I — System Boot & Initialization

ULK Ch.1 · LKD Ch.1

• §1.1Bootloader & U-Boot
• §1.2Kernel Boot Process (start_kernel → PID 1)

Part II — Memory Management

ULK Ch.2,8,9 · LKD Ch.12,15,16

• §2.1Physical Memory Layout (NUMA / Zones / struct page)
• §2.2Buddy Allocator
• §2.3Slab / Slub Allocator
• §2.4Virtual Memory & Paging (4-level PT, TLB, Huge Pages)
• §2.5Process Address Space (mm_struct, VMA, mmap, COW)
• §2.6Page Cache & Buffer Cache (LRU/MRU, writeback)
• §2.7Memory-Mapped I/O & DMA (ioremap, IOMMU)

Part III — Process Management & Scheduling

LKD Ch.3,4,5 · ULK Ch.3,6,11

• §3.1Process & Thread Data Structures (task_struct)
• §3.2Process Lifecycle (fork → exec → exit)
• §3.3CPU Scheduling (CFS, SCHED_FIFO, load balancing)
• §3.4Preemption & Context Switch
• §3.5Interrupts & Bottom Halves (IRQ → softirq → NAPI)
• §3.6Kernel Timers (jiffies, timer wheel, hrtimer, NO_HZ)
• §3.7Kernel Data Structures (list, hlist, rbtree, xarray, bitmap)

Part IV — System Calls & User-Kernel Interface

LKD Ch.5 · ULK Ch.10

• §4.1System Call Mechanism (syscall, vDSO)
• §4.2Key Networking Syscalls (socket, epoll, io_uring)
• §4.3ioctl & Netlink
• §4.4Sync / Async I/O (blocking, O_NONBLOCK, epoll, io_uring, sendfile)
• §4.5Signals

Part V — Interprocess Communication

LKD Ch.17 · ULK Ch.19

• §5.1Pipes & FIFOs
• §5.2POSIX IPC (msg queues, shared memory, semaphores)
• §5.3Futex (fast userspace mutex)
• §5.4Unix Domain Sockets

Part VI — Parallel Programming & Synchronization

PPH all · LKD Ch.9,10 · critical interview section

• §6.1Hardware Foundations (pipeline, store buffer, false sharing, NUMA)
• §6.2Cache Coherence Protocols (MESI / MOESI)
• §6.3Memory Barriers & Ordering (smp_mb, acquire/release, C11 model)
• §6.4Atomic Operations (CAS, fetch-add, ABA problem)
• §6.5Counting & Per-CPU Counters
• §6.6Locking Design Patterns (lockdep, ordering, granularity)
• §6.7Spinlock Internals (ticket → MCS → qspinlock)
• §6.8Sleeping Locks (mutex, rwsem, semaphore)
• §6.9Sequence Locks (seqlock_t)
• §6.10RCU Deep Dive (grace period, call_rcu, SRCU)
• §6.11Lock-Free & Wait-Free Algorithms (Treiber stack, MS queue, hazard ptrs)
• §6.12Per-CPU Data & Data Ownership
• §6.13Deferred Processing (refcount, RCU reclaim, workqueue)
• §6.14Parallel Design Patterns (batch, sharding, pipeline, map-reduce)
• §6.15Debugging (lockdep, KCSAN, TSan, perf lock)
• §6.16Linux Kernel Memory Model (LKMM, litmus tests)

Part VII — Network Stack Internals

LKN all · ULNI all · TCPIP Vol.1&2

• §7.1Socket Layer (struct socket → sock → tcp_sock)
• §7.2sk_buff — Core Data Structure (GSO, GRO, zero-copy)
• §7.3Network Device Layer (net_device, NAPI, TX/RX path)
• §7.4Traffic Control & QoS (Qdisc, HTB, cls_flower)
• §7.5IP Layer (ip_rcv, fragmentation, ICMP)
• §7.6Routing Subsystem (fib_table, LPM, fib_trie, policy routing)
• §7.7Neighbor Subsystem (ARP, NDP, proxy ARP, GARP)
• §7.8TCP/IP Deep Dive (state machine, CUBIC, BBR, SYN Cookie)
• §7.9UDP & Multicast socket
• §7.10Netfilter & NAT (5 hooks, conntrack, DNAT/SNAT)
• §7.11Network Bridge (FDB, MAC learning, STP)
• §7.12GRE & VXLAN Tunnels
• §7.13IPv6 (NDP, RA/RS, DHCPv6, extension headers)
• §7.14Multicast (IGMP, MLD, PIM-SM, RPF)
• §7.15DHCP Protocol & Proxy DHCP (DORA, DHCPv6)
• §7.16Routing Protocols (RIP, OSPF, BGP, BIRD)

Part VIII — Network Drivers & Hardware Interface

LDD Ch.17 · ULNI Ch.8,9 · DPDK docs

• §8.1PCI / PCIe Device Driver (BARs, MSI-X, pci_probe)
• §8.2Network Driver Model (TX/RX descriptor rings, DMA, TSO)
• §8.3SR-IOV (PF / VF, per-VF flow classification)
• §8.4DDIO (Intel Data Direct I/O)
• §8.5Character Devices (cdev, file_operations, UIO)
• §8.6Block Devices (gendisk, blk-mq, struct bio)

Part IX — DPDK (Userspace Networking) — Deep Dive

DPDKPG all · critical interview section

• §9.1Why DPDK: Kernel Bypass Architecture (UIO vs VFIO)
• §9.2DPDK Startup Flow (rte_eal_init 10 steps)
• §9.3PMD Poll Mode Driver (descriptor rings, DD bit, doorbell)
• §9.4NIC Multi-Queue, RSS & Flow Classification (rte_flow)
• §9.5Memory: Huge Pages, Mempool & Mbuf
• §9.6NUMA-Aware Programming
• §9.7Lock-Free Ring Buffer (rte_ring — CAS algorithm)
• §9.8Cache Optimization (alignment, prefetch, DDIO, WC)
• §9.9ACL & LPM Classification (DIR-24-8, rte_hash cuckoo)
• §9.10SR-IOV in DPDK (VF pass-through, VFIO)
• §9.11Hardware Offload (checksum, TSO, LRO, VXLAN offload)
• §9.12SIMD in DPDK (SSE4.2, AVX2, AVX-512)
• §9.13Virtio, vhost-user & vDPA
• §9.14Pipeline & Run-to-Completion Models
• §9.15DPDK Hot Upgrade (multi-process, shared memory)
• §9.16Session Management (fast path / slow path)
• §9.17QoS in DPDK (rte_meter, rte_sched, distributed rate limit)
• §9.18DPDK Bonding & Link Aggregation
• §9.19OVS-DPDK (EMC, dpcls, megaflow, VXLAN)

Part X — XDP (eXpress Data Path)

kernel docs · eBPF/XDP reference

• §10.1XDP Architecture & Motivation (XDP vs DPDK comparison)
• §10.2XDP Modes (native, generic, offloaded)
• §10.3AF_XDP Socket (UMEM, fill/completion/RX/TX rings)
• §10.4XDP Maps & Data Structures (BPF map types)
• §10.5XDP Use Cases (DDoS, Katran LB, Cilium, monitoring)

Part XI — Storage & File Systems

LKD Ch.13,14 · EXT4 docs

• §11.1VFS (super_block, inode, dentry, address_space)
• §11.2ext4 File System (extents, JBD2, delayed alloc)
• §11.3Block Layer (struct bio, blk-mq, I/O schedulers, O_DIRECT)
• §11.4B+ Tree (split/merge, lock coupling)
• §11.5Page Cache Management (LRU/MRU, double-write buffer)
• §11.6Disk Free Space Management (bitmap, buddy)
• §11.7Coroutines & Cooperative Scheduling (async I/O + ucontext)

Part XII — Virtualization & Containers

LKD Ch.18 · KVM docs · LXC/Docker

• §12.1CPU Virtualization (KVM/VMX, VMCS, VMEXIT)
• §12.2Memory Virtualization (EPT, balloon, KSM)
• §12.3VM Live Migration (dirty page tracking, pre-copy)
• §12.4Containers: Namespace & Cgroup v2
• §12.5Container Networking — CNI (veth, VXLAN, Cilium)
• §12.6Storage Virtualization (virtio-blk, vhost-user-blk, NVMe PT)

Part XIII — Hardware Communication & PCIe

PCIe spec · DPDK docs · LDD Ch.15

• §13.1PCIe Architecture (TLP, credits, LTSSM, lanes)
• §13.2Memory-Mapped BAR Access (MMIO, write-combining)
• §13.3DMA & IOMMU (bus address, VT-d, VFIO)
• §13.4NUMA & CPU Topology (libnuma, lscpu)

Part XIV — Debugging & Profiling Tools

LKD Ch.18 · perf docs · kernel tracing

• §14.1Network Debugging (tcpdump, ss, iperf3, ethtool)
• §14.2System Call Tracing (strace, ltrace, perf trace)
• §14.3Performance Profiling (perf stat/record/top, flamegraph)
• §14.4Kernel Debugging (gdb+KGDB, crash, KASAN, OOPS)
• §14.5Kernel Tracing (ftrace, kprobe, eBPF/bcc, bpftrace)
• §14.6Memory Debugging (valgrind, kmemleak, /proc/slabinfo)

Part XV — Distributed Systems (Raft)

Ongaro & Ousterhout

• §15.1Raft Consensus (leader election, log replication, safety)
• §15.2Raft Practical Concerns (snapshot, membership change, log replay)
• §15.3Performance Optimizations (pipelining, batching, 100K IOPS)

Part XVI — Load Balancing & Consistent Hashing

Resume projects · DPDK LPM/ACL

• §16.1Load Balancing Algorithms (RR, ECMP, consistent hashing)
• §16.2NAT Gateway Internals (SNAT port selection, HA)
• §16.3Session Synchronization (state machine replication, delta sync)

Part XVII — C Standard Library, Toolchain & ABI

CSAPP · APUE · TLPI · Linkers and Loaders · Expert C · interview-critical

• §17.1C libc Landscape (glibc vs musl vs uClibc vs Bionic vs newlib vs dietlibc — size/perf/license trade-offs)
• §17.2User → Kernel Path: libc Wrapper → syscall Instruction → kernel handler (errno convention, INTERNAL_SYSCALL macros)
• §17.3vDSO & Pseudo Syscalls (clock_gettime/gettimeofday/time/getcpu run in user space — no ring switch; auxv AT_SYSINFO_EHDR)
• §17.4Userspace Memory Allocator (ptmalloc2 arenas/bins/fastbins/tcache; brk vs mmap threshold M_MMAP_THRESHOLD; fragmentation)
• §17.5Allocator Alternatives (jemalloc, tcmalloc, mimalloc, scudo) — per-thread cache, size classes, why DPDK uses rte_malloc instead
• §17.6stdio Buffering (full/line/unbuffered, FILE*, setvbuf, fflush — fork() double-flush pitfall)
• §17.7POSIX Thread Library (NPTL, pthread_create → clone(CLONE_VM|...), 1:1 model, futex-backed mutex, pthread_key TSD)
• §17.8Thread-Local Storage (__thread / _Thread_local, ELF .tdata/.tbss, TLS models GD/LD/IE/LE, FS/GS register)
• §17.9Dynamic Linker / Loader (ld.so, PLT/GOT, lazy binding via _dl_runtime_resolve, LD_PRELOAD, LD_LIBRARY_PATH, RPATH vs RUNPATH)
• §17.10Runtime Loading (dlopen/dlsym/dlclose, RTLD_LAZY vs RTLD_NOW, RTLD_GLOBAL, plugin pattern)
• §17.11ELF File Format (header, PHT vs SHT, segments LOAD/DYNAMIC/INTERP, sections .text/.rodata/.data/.bss/.init_array, symbol/relocation tables)
• §17.12C Compilation Pipeline (cpp preprocess → cc1 compile → as assemble → ld link; gcc -E/-S/-c/-o; -save-temps; intermediate .i/.s/.o)
• §17.13Static vs Dynamic Linking (.a archive ar/ranlib, .so, -static, ldconfig cache /etc/ld.so.cache, soname/version, symbol resolution order)
• §17.14Position Independent Code (PIC -fPIC for .so, PIE -fPIE for ASLR, GOT/PLT indirection cost, copy relocations)
• §17.15Cross Compilation (target triplet arch-vendor-os-libc e.g. aarch64-linux-gnu, --host/--build/--target, sysroot, multilib)
• §17.16Cross Toolchains (crosstool-NG, Buildroot, Yocto/OE, Linaro GCC; how to build a glibc/musl cross toolchain from scratch)
• §17.17Library Inspection Tools (ldd, ldconfig -p, nm, objdump -d/-h/-r, readelf -a/-d/-l, strings, file, strip, ar, c++filt, addr2line, size, ranlib)
• §17.18Runtime Tracing (strace -e trace=... syscalls, ltrace library calls, LD_DEBUG=symbols/bindings/files, perf trace)
• §17.19Calling Conventions / ABI (SysV AMD64: rdi/rsi/rdx/rcx/r8/r9 + xmm0–7, red zone 128B, stack 16-byte alignment; x86 cdecl/stdcall; ARM AAPCS r0–r3)
• §17.20C ↔ C++ ABI (Itanium C++ ABI, name mangling _ZN..., vtable/RTTI layout, extern "C", ABI breakage between GCC versions)
• §17.21errno & Reentrancy (errno is per-thread macro → *__errno_location(); _r reentrant variants; async-signal-safe set)
• §17.22Atomics & Memory Model (C11 <stdatomic.h>, atomic_*, memory_order_relaxed/acquire/release/seq_cst, libatomic fallback)
• §17.23glibc Internals (rtld bootstrap, hidden_def, IFUNC dispatch e.g. memcpy AVX/SSE, symbol versioning GLIBC_2.x, vtable hardening)
• §17.24Security Hardening (FORTIFY_SOURCE=2/3, -fstack-protector-strong canary, ASLR, RELRO partial/full, NX/DEP, CFI, _GLIBCXX_ASSERTIONS)
• §17.25Common C Libraries — System: libpthread, librt, libdl, libm, libresolv, libcrypt (now libxcrypt)
• §17.26Common C Libraries — Crypto/Net: OpenSSL (libssl/libcrypto), libcurl, c-ares, libpcap, libnl (netlink), libnftnl, libnfnetlink
• §17.27Common C Libraries — Event Loops: libev, libevent, libuv (cross-platform), libaio, liburing (io_uring)
• §17.28Common C Libraries — DPDK/Kernel-bypass: librte_eal/mempool/mbuf/ring/hash/lpm, libnuma, libhugetlbfs, libpci, liburcu
• §17.29Common C Libraries — eBPF: libbpf, libxdp, libelf, libdebuginfod, BCC (Python-bound)
• §17.30Common C Libraries — Compression: zlib, libbz2, liblzma, lz4, libzstd, snappy-c
• §17.31Common C Libraries — Serialization: protobuf-c, msgpack-c, libcjson, libxml2, libyaml
• §17.32Common C Libraries — Data Structures: glib (GLib), libavl, libcuckoo, judy arrays
• §17.33Sanitizers & Static Analysis (ASan/MSan/TSan/UBSan, valgrind memcheck/helgrind/cachegrind, clang-tidy, cppcheck, scan-build)
• §17.34Build Systems & Package Discovery (Make, CMake, Meson, autotools; pkg-config --cflags/--libs, .pc files)

Part XVIII — Threads, Processes & User-Space Synchronization

TLPI Ch.24-30 · APUE Ch.8,11,12 · Butenhof · interview-critical

• §18.1Process vs Thread vs Coroutine (memory model, fault isolation, switch cost, scheduling unit)
• §18.2fork() Semantics (COW, FD inheritance, mmap inheritance, atfork handlers, fork+threads pitfall)
• §18.3vfork vs fork vs clone() (CLONE_VM/FILES/SIGHAND/THREAD/NEWPID/NEWNS — building blocks of threads & containers)
• §18.4exec Family (execl/execv/execvp/execve, environment passing, FD_CLOEXEC, fexecve, exec + posix_spawn)
• §18.5Process Lifecycle (states R/S/D/Z/T/X, wait/waitpid/waitid, SIGCHLD, zombie/orphan, init/subreaper)
• §18.6Daemon Process Pattern (double-fork, setsid, chdir, umask, redirect std fds, PID file, syslog)
• §18.7Process Groups, Sessions & Job Control (setpgid, setsid, tcsetpgrp, controlling terminal, SIGHUP)
• §18.8Resource Limits & Accounting (setrlimit/prlimit, getrusage, RLIMIT_NOFILE/STACK/AS/CORE, ulimit)
• §18.9Privilege Management (uid/euid/suid/fsuid, setuid/seteuid, capabilities CAP_*, cap_set_proc)
• §18.10pthread Lifecycle (pthread_create/join/detach, pthread_attr_t — stack size, sched policy, contention scope)
• §18.11pthread Mutex (NORMAL/RECURSIVE/ERRORCHECK/ROBUST, PRIO_INHERIT, PROCESS_SHARED for shm)
• §18.12pthread Condition Variable (predicate loop, wait/signal/broadcast, spurious wakeup, lost-wakeup pitfall)
• §18.13pthread RWLock (reader/writer preference, recursive-read deadlock, when RCU is better)
• §18.14pthread Spinlock & Barrier (busy-wait when context switch is more expensive, PROCESS_SHARED)
• §18.15pthread Cancel & Cleanup (cancellation points, deferred vs async, pthread_cleanup_push/pop, RAII analog in C)
• §18.16pthread_once & Thread-Specific Data (pthread_once init, pthread_key_create + destructor on exit)
• §18.17Signal Handling in Multi-Threaded Programs (sigwait, pthread_sigmask, signalfd, async-signal-safe set)
• §18.18Producer-Consumer Patterns (bounded buffer, sem + mutex, blocking queue, SPSC lock-free ring)
• §18.19Reader-Writer Patterns (preferring readers vs writers, seqlock alternative, RCU alternative)
• §18.20Thread Pool Design (fixed/dynamic sizing, bounded queue, work stealing, graceful shutdown)
• §18.21Multi-Process Server Models (Apache prefork, Nginx master+worker, fork-per-connection, pipe dispatch)
• §18.22Multi-Thread Server Models (thread-per-connection, thread pool + epoll/io_uring, M:N hybrid)
• §18.23Concurrency Model Comparison (process / thread / coroutine / event-loop / actor — pros, cons, use cases)
• §18.24Deadlock Prevention (lock ordering, try-lock with backoff, hierarchical mutex, deadlock detection)
• §18.25Priority Inversion & PI Futex (classic Mars Pathfinder bug, PI mutex, ceiling protocol)
• §18.26Inter-Process Synchronization (PROCESS_SHARED mutex/sem in shm, POSIX named sem, fcntl/flock file lock, robust mutex on owner death)
• §18.27Userspace Coroutines & Fibers (ucontext makecontext/swapcontext, libco/libtask, stackful vs stackless, ABI cost)
• §18.28Modern Coroutine Models (Go M:N goroutines + scheduler, C++20 coroutines, Rust async/await — comparison to C)

Part XIX — File Descriptors, File I/O & Unix Plumbing

TLPI Ch.4-19,44-50,63 · APUE Ch.3,4,14,15,17 · interview-critical

• §19.1File Descriptor Model (per-process FD table → open file description → inode; ref-counting on dup/fork)
• §19.2open() Flags (O_RDONLY/WRONLY/RDWR, O_APPEND atomicity, O_CREAT|O_EXCL TOCTOU-safe create, O_NONBLOCK, O_DIRECT bypass page cache, O_SYNC/O_DSYNC, O_TMPFILE, O_PATH, O_CLOEXEC, O_NOFOLLOW)
• §19.3read/write Family (read/write short-count rules, pread/pwrite no-offset-race, readv/writev scatter-gather, copy_file_range in-kernel copy)
• §19.4lseek & Sparse Files (SEEK_SET/CUR/END, SEEK_HOLE/SEEK_DATA, fallocate/FALLOC_FL_PUNCH_HOLE)
• §19.5dup / dup2 / dup3 (FD aliasing, redirect stdout to a pipe, FD_CLOEXEC propagation rules, fork/exec interaction)
• §19.6Pipe Internals (pipe(2)/pipe2, 64 KB kernel ring, PIPE_BUF=4096 atomic write boundary, SIGPIPE/EPIPE on broken reader)
• §19.7Named Pipes / FIFOs (mkfifo, blocking until both ends opened, multi-writer atomicity, use cases vs Unix sockets)
• §19.8Zero-Copy Plumbing (sendfile fd→fd, splice pipe-mediated zero-copy, tee duplicate pipe data, vmsplice user pages → pipe)
• §19.9fcntl (F_GETFL/F_SETFL toggle O_NONBLOCK, F_DUPFD_CLOEXEC, F_SETPIPE_SZ resize pipe, F_NOTIFY directory changes — legacy)
• §19.10File Locking (BSD flock whole-file, fcntl record locks per-process pitfall, OFD locks F_OFD_SETLK per-FD safe, mandatory vs advisory, lockf)
• §19.11mmap & Memory-Mapped I/O (MAP_SHARED vs PRIVATE COW, MAP_ANONYMOUS, MAP_POPULATE prefault, MAP_HUGETLB, MAP_FIXED hazards, msync/madvise WILLNEED/DONTNEED/HUGEPAGE)
• §19.12Directory Operations (opendir/readdir/closedir, raw getdents64, scandir, telldir/seekdir/rewinddir; openat/AT_FDCWD safe relative ops)
• §19.13File Metadata (stat/fstat/lstat/statx, st_dev/st_ino identity, atime/mtime/ctime, relatime/noatime mount opts, nanosecond timestamps)
• §19.14Permissions & Ownership (chmod/fchmod/chown/fchown, umask inheritance, setuid/setgid/sticky bits, ACL setfacl/getfacl, file capabilities)
• §19.15Links & Path Operations (hard link vs symlink, readlink, rename atomicity guarantees, renameat2 RENAME_EXCHANGE/RENAME_NOREPLACE, unlink-while-open delayed delete)
• §19.16Filesystem Notification (inotify watch descriptors + epoll, fanotify content access / permission events, dnotify legacy, recursive watch design)
• §19.17Special Files (eventfd counter for IPC wakeups, signalfd synchronous signal receive, timerfd timer-as-FD, memfd_create + memfd seals — sealable shared memory)
• §19.18Async I/O (POSIX aio_* user-thread emulation, Linux native io_submit/io_getevents quirks, libaio O_DIRECT only, why everyone moved to io_uring)
• §19.19Filesystem Containment (chroot escapes, pivot_root for containers, openat O_PATH FD-rooted ops, mount namespaces interaction)
• §19.20Extended Attributes (user.* / security.* / system.* / trusted.* namespaces, getxattr/setxattr/listxattr, used by SELinux/Smack and file caps)
• §19.21stdio vs Raw FD (FILE* buffering full/line/none, fdopen/fileno/freopen, fork-double-flush, mixing fwrite+write breaks order)
• §19.22tty / pty (pseudo-terminal master/slave, openpty/forkpty, raw vs cooked mode, used by ssh/screen/tmux/Docker exec)
• §19.23/proc & /sys File Interfaces (counters via read, tunables via write, /proc/self/fd, /proc/PID/maps VMA dump, /sys/class/net devices)
• §19.24epoll Deep Dive (epoll_create1, ADD/MOD/DEL, EPOLLET edge-trigger drain rule, EPOLLONESHOT, EPOLLEXCLUSIVE thundering herd, vs select/poll O(n))
• §19.25io_uring Deep Dive (SQ/CQ shared rings, SQPOLL kernel-side polling, IORING_OP_* op set, fixed buffers/registered files, multishot, IOSQE_LINK chains)

Appendix — Key Kernel Data Structures

Appendix — C libc Implementations

Appendix — Toolchain & Library Inspection Tools

Appendix — Common C Libraries Cheat Sheet

Appendix — C Compilation Pipeline & Cross-Build

Appendix — vDSO Pseudo-Syscalls (no kernel switch)