UNIX hacker and security officer wrote:
> 
>  * In other words, is there no getcwd()-like functionality in the kernel? 
                                         ^^^^^
> You would, indeed, have to walk up the tree yourself.  getcwd() or getwd()
> (It's been a while since I've used the call, so pardon the ambiguity)
> are written in user space and have never been system calls.

I'm well aware that getcwd() is not a syscall. Please look at what I
wrote. The fact that memcpy(3) is implemented in user space does not
imply that similar functionality is not availible to kernel code,
perhaps with different name and arguments (eg, bcopy()).

> I'm not altogether sure why you'd want to keep current-directory information
> around anyway.  You'd have core files which were large and unwieldy to
> access, and this would not necessarily be a good thing.

*Huh?* I have absolutely no idea what this means. What does this have to
do with the size of corefiles?