How miryala10sathvika/xv-6-and-networking works
Project type and purpose
• This repo is not a single deployed application but a collection of coursework/lab components:
• “graphs”: Python-based graph-generation for scheduler data
• “initial-xv6”: a RISC-V port of the xv6 teaching OS (monolithic kernel + userland)
• “networks”: C implementations of basic TCP/UDP clients/servers, RPC over TCP/UDP, and a reliable transport protocol (parts A & B)
• “tester”: scripts to automate building and testing the xv6 image
• The root README focuses on a custom TCP-like transport and flow-control design (networks lab).
File structure
• Top-level folders group independent subprojects (graphs, initial-xv6, networks, tester).
• “initial-xv6/src/kernel” vs. “initial-xv6/src/user” signals a monolithic kernel with user processes.
• “networks/partA” & “networks/partB” separate different networking exercises (basic, RPC, reliable transport).
• Build scripts: Makefiles in initial-xv6, Python in graphs, shell scripts in tester.
• Configuration: xv6 uses its own Makefile and fs.img; networks C code compiled via gcc.
README for additional insights
• The root README outlines how the custom transport deviates from real TCP (fixed ISN, fixed timer, no dynamic window).
• It also proposes enhancements (add window field, RTT estimation, slow-start, congestion control heuristics).
Detail without overthinking
• Break each subproject into logical blocks (e.g., “Kernel: fs, proc, trap, virtio”, “Net: basic transport, RPC stub, reliable layer”).
• How data and control cross boundaries (syscalls, network sockets, script-→build).
• Custom protocol tweaks (fixed ISN, fixed timer, proposed window-field).