> Yes, I realized what the bug was.  I consider it a kernel bug.
> 

IMHO, it's both a kernel bug and a bug in strip.

While the kernel must not die when ptracing a load module that it successfully
loaded in the first place, it's also a bug in strip that it truncates the data
space without proper reflection in the a.out header.

I'm not sure whether the kernel should supply zeroed pages in case of a load
module like this, or should return with an appropriate error code.

Note that the same effect can be achieved by mmaping a region of a file that
is out of its bounds.

The problem is that currently the routine vm_fault_wire doesn't return
anything (its return type is void), so vm_map_pageable cannot supply a
meaningful return value. We haven't found a solution yet.
--
ws@TooLs.DE     (Wolfgang Solfrank, TooLs GmbH) +49-228-985800

------------------------------------------------------------------------------