SSTF I/O Scheduler for Linux

This project implements a parametrizable SSTF (Shortest Seek Time First) I/O scheduler as a Linux kernel module, along with a user-space C program (sector_read) to benchmark disk I/O performance.

It was developed for the Operating Systems Construction course at PUCRS, under Professor Sérgio Johann Filho

Features

• Custom kernel module (sstf-iosched) for Linux 4.13.
• Adjustable parameters:
  • queue_depth (max requests before dispatch, e.g., 20–100),
  • max_wait_ms (max wait time in ms, e.g., 20–5000),
  • debug (enable kernel debug logs).
• User-space benchmark tool (sector_read) for random read/write operations.
• Includes scripts and Makefile for building and running in a QEMU + Buildroot environment.
• Academic report comparing SSTF and NOOP schedulers under varied workloads.

For a detailed explanation of the logic, and results, please refer to the 🔗report*

How to Run

1. Copy files to /modules/sstf in your Linux build.
2. Build:

   make
   

3. Export Linux source (if using Buildroot):

   export LINUX_OVERRIDE_SRCDIR=`pwd`/../linux-4.13.9/
   make
   

4. Inside QEMU:

   modprobe sstf-iosched.ko queue_depth=20 max_wait_ms=3000 debug=1
   echo sstf > /sys/block/sdb/queue/scheduler
   ./bin/sector_read 512 2000000 250 50 5
   

Authors

• Bruno Bavaresco Zaffari
• Emanuel Nogueira de Barros
• Thiago Müller de Oliveira

License

This project is licensed under the GNU General Public License v2.0 (GPLv2).