Experimental System Call Tracer:
================================

This is just a proof of concept. As an example it traces `open` syscall
for specified pid using kprobes mechanism. This can be extended so that
a userspace program can specify a set of syscalls to trace. The module
creates two sysfs files.

/sys/kernel/esct/pid: Write a pid to this and all open syscalls made by
that pid will be traced.

/sys/kernel/esct/trace: Read this file to get trace for the pid
specified above.

forker.c is a quick demo program which forks, the child runs ls while
parent pulls the trace from /sys/kernel/esct/trace and writes it to
/tmp/syscall.trace.

Compiling:
==========

From the directory containing esct.c and Makefile, run:

make -C /path/to/linux/source M=`pwd`

This will produce esct.ko which you can insmod.

Compile forker with: gcc -Wall -o forker forker.c

First insmod esct.ko then run forker as root.

NOTE:
=====

Just a note, this is totally experimental, and hacked together over a
busy weekend. There are improvements that can be made. A key one will
be for the module to export data using netlink sockets rather than the
sysfs file. On code level, there might be bugs too as this was done
in short time.