Jon Haslam's Weblog

Well, it's been a while to say the least but I think it's about time to put fingers to keys again and see what comes forth ...

I've mentioned before some of the great things that can be achieved with destructive actions. Indeed many good examples are popping up - check out this cool example from Chris Gerhards blog. Today we'll just take a quick look at another one (I actually have a bunch which I'll try and write-up over the next few weeks).

In the UK we have run a series of events called Make-It-Fly which I've been involved with. Last week I did a session on DTrace which I hope everyone who was present enjoyed (I did anyway!). At the events I usually do quite a bit of hands on demo with most of it being hand cranked. However, one of the scripts I use that always gets a laugh is the following one which I don't hand crank as I can never remember the offsets ... If you've ever wanted to get uname(1) to return something different to normal then this is what you need:

#!/usr/sbin/dtrace -s

#pragma D option destructive

syscall::uname:entry
{
        self->addr = arg0;
}

syscall::uname:return
{
        copyoutstr("SunOS", self->addr, 257);
        copyoutstr("PowerPC", self->addr+257, 257);
        copyoutstr("5.5.1", self->addr+(257*2), 257);
        copyoutstr("gate:1996-12-01", self->addr+(257*3), 257);
        copyoutstr("PPC", self->addr+(257*4), 257);
}

Before we have:

# uname -a
SunOS homer 5.10 SunOS_Development sun4u sparc SUNW,Ultra-5_10

and like magic we morph into something else when the above script is ran:

# uname -a
SunOS PowerPC 5.5.1 gate:1996-12-01 PPC sparc SUNW,Ultra-5_10

Here at Sun we often test pieces of software on versions of Solaris that return something different to that which the software is expecting. Previously I would LD_PRELOAD a library in with my own uname hack. Now I can not only do this without bothering the application but I can present different uname information to different applications/users/whatever as I can predicate accordingly!

Note, that the above script isn't quite complete as it returns the incorrect ISA information. This is me being idle and a bit of twiddling with sysinfo() is all that's needed. Maybe another day.