xv6-Projects

About

Assignments from the Operating Systems course, Spring 2025.

Task 1. Implement System Calls

Create five new system calls for the xv6 kernel:

• getnice(pid): Retrieve the nice value (priority) of a given process.
• setnice(pid, value): Set the nice value, controlling scheduling priority (valid range: 0–39, default 20).
• ps(pid): Print process info (name, pid, state, nice value) for a given PID or all processes if pid == 0.
• meminfo(): Return available memory in bytes.
• waitpid(pid): Block the current process until the specified process exits.

🔍 Refer to the Make-System-Calls branch for full implementation.

Task 2. Implement EEVDF CPU Scheduler

Replace xv6's default round-robin scheduler with a Linux-like EEVDF (Earliest Eligible Virtual Deadline First) scheduler:

• Schedule the task with the earliest virtual deadline among eligible ones.
• Accurately compute and update vruntime, vdeadline, eligibility, and runtime per tick.
• Modify ps() to display: runtime/weight, runtime, vruntime, vdeadline, eligibility, and total_tick.
• Ensure correct handling of forked/woken-up processes.

🔍 Refer to commits on the Linux-EEVDF-Scheduler branch starting from Apr 9, 2025.

Task 3. Implement mmap(), munmap(), freemem() and Page Fault Handler

Introduce virtual memory operations and support for memory-mapped files:

• mmap(addr, length, prot, flags, fd, offset): Map files or anonymous memory to a process’s virtual address space. Supports MAP_ANONYMOUS, MAP_POPULATE, PROT_READ, and PROT_WRITE.
• Page Fault Handler: On access to an unmapped page in a mapping region, allocate a physical page and set up the mapping (with file reading if needed).
• munmap(addr): Unmap a previously mapped region and free associated resources.
• freemem(): Return the number of free physical pages in the system.

🔍 Refer to commits on the Page-Fault-Handler branch starting from May 8, 2025.

Task 4. Implement Page Replacement with Swapping

Enable swapping in xv6 to handle memory pressure using a clock algorithm and LRU list:

• Swap out pages to disk when physical memory is full, using swapwrite().
• Swap in pages upon access, using swapread().
• Track swappable pages via a circular LRU list.
• Integrate a bitmap-based swap space tracking mechanism.
• Ensure both swapped-in and swapped-out pages are managed during fork and deallocation.

🔍 Refer to commits on the Page-Replacement branch starting from May 23, 2025.

@SKKU (Sungkyunkwan University)