Re: init is pid 1 vs. initrd
H. Peter Anvin (hpa@transmeta.com)
Fri, 17 Apr 1998 15:55:46 -0700 (PDT)
> H. Peter Anvin <hpa@transmeta.com> wrote:
> > That will, unfortunately, not release the initrd memory.
>
> Yes, you'd need a mechanism for that. Three possibilities
> come to mind:
>
> (1) magic in the initrd system and it goes away when the last
> referencing process exits. This is evil, I think [what
> triggers it?]
>
> (2) An explicit command which enables similar magic.
>
> (3) An explicit command which only frees the initrd memory if
> nothing is still around which refers to it (returns an error
> otherwise).
>
> I'm not sure which would be simplest to implement.
>
> I'm not sure if there are any subtle references which would have
> to be forcibly broken.
>
This doesn't solve the main problem, which is that if you use chroot,
the very existence of the mounted filesystem on top of it will count
as a reference, and hence the root will never be freed.
Let's face it that the initrd pivot is an exceptional even, and let's
treat it as such.
-hpa
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu