xv-DOS

xv6 - Dependent Operating System

An extension of the MIT's xv6 Operating System for RISC-V.
Read the original readme here. The original code (at the time of the fork) can be found in the riscv branch.

Installation

You can follow the install instructions here. (Skip the Athena part)

Running the OS

$ make clean
$ make qemu SCHEDULER=[RR/PBS/FCFS] CPUS=[N_CPU]

default scheduler is RR (Round-robbin).

Modifications

Here are the modifications made to the original xv6:

Modification 1: Syscall Tracing

We aim to intercept and record the system calls which are called by a process during its execution. We define a user program called strace which uses the system call trace.

User Program

Running:

$ strace [mask] [command] [args]

Implementation:

void strace(int mask, char *command, char *args[]);

First, we fork the current process. The parent process waits for the child process. In the child process, before we execute the command using exec, we run the trace syscall with mask as a parameter.

System Call

int trace(int mask);

There is a 'trace_mask' defined for the proc struct, which is by default set to 0. Every time a process is forked, the child inherits its parent's trace_mask.

When this syscall is called, it takes the input argument which it sets to the proc->trace_mask.

For every set bit, the syscall() function in syscall.c prints the information for the syscall corresponding to that bit.

Modification 2: Scheduling

First Come - First Served

We compare the creation time of each process (which is stored in proc->ctime and initialized to 0 when the process is allocated in the table).

Then we schedule the process with the minimum creation time which is currently in the table.

Since this is non-preempted, a condition is added to ignore the yield() when FCFS is defined.

Priority Based Scheduler

Instead of time, we compare the priorities. Static priority (default to 60) can be changed (explained below) by the user. Dynamic priority is calculated and compared:

dynamic_pr = max(0, min(100, static_pr - niceness + 5))

where niceness is defined as

10*(ntime)/(ntime+rtime)

here ntime (nap time) and rtime (run time) are the ticks spent in sleeping state since the last call/running state in total, stored in proc->ntime and proc->rtime.

Set Priority

User program:

$ setpriority [priority] [pid]

which calls the set_priority system call which sets the static priority to the given value and resets ntime to 0 and niceness to 5.

int set_priority(int new_priority, int proc_id);

Modification 3: procdump

Procdump is triggered when Ctrl+P is pressed. It lists out information about all the processes.

For FCFS, procdump acts in the same manner as in the case of RR (default case).

PBS

pid priority state rtime wtime nrun

pid and state are already defined.

• priority is just the dynamic priority as defined in PBS.
• rtime is the run time as defined in niceness.
• wtime (wait time) is calculated as

wtime = ltime - rtime - stime

here ltime is the total lifetime of the process and stime is the total sleep time since creation (different from ntime).

• nrun (number of times the process was picked by the scheduler) is stored in proc->nrun initialized to 0 at the creation of the process. Every time the status of a process is changed from runnable to running, it is incremented.

---

This was built as a part of the Operating Systems and Networks course, Monsoon 2021. The problem statement is given here.