Re: startup code

[Espen Skoglund]

[Johan Rydberg]
> The question is where should most of the work be done: in the parent
> or in the client [1], i.e. the startup code?

[...]

> This means that the parent process must do the actual parsing of the
> program headers of the executable, and pass information about it to
> the client.  If it sees that the program is a shell-script, it is
> responsible for launching the interpreter.

Intuitively, this suggests to me that the startup code should be in
the client and have the parent just pass along the
code/module/whatever that does the real parsing and setup.  By doing
it this way one takes a lot of complexity out of the parent (making it
more robust), makes the system more extensible, makes the startup code
execute with least privilege, gets all the work accounted to the
client, and obviates the need for the parent to actually trust the
startup code.

Just my 2 cents.

        eSk