Freebsd vs. linux

[Anthony Atkielski]

Eric Kjeldergaard writes:

> Well, that's simply not true.  Windows (At least XP, can't remember if
> this happens on 2k or not) definitely states that the error that
> caused the crash is or is not a driver.  It says something to the
> effect of "The system crash was caused by a bad driver."

I suppose it may isolate messages better than its predecessors.

> You read the code, I'm sure you know where that message gets
> generated.

These operating systems contain millions of lines of code.  I haven't
read them all.  Nobody has.

> As windows is largely infallible, I'm sure they have some way of
> keeping straight which portion of the kernelspace is drivers and which
> isn't.

You overestimate the capabilities of the OS.  In many cases it is
possible to roughly or precisely localize a fault in kernel space, but
at other times it is not.  Bugs in one area of an OS (defined as
anything with kernel privileges) can cause faults in other areas.

> Yeah, they kinda are.  If they are on the operating system cd, they
> are part of the operating system.

There are lots of things on CDs that are not part of the OS.

> An operating system is just a software package.

In this context, one can think of the operating system as the aggregate
of permanent code executing with kernel privileges.

Drivers execute with these privileges but they are essentially add-ons.
They aren't written by the same people and they usually aren't written
and tested with the same rigor.

-- 
Anthony