Subject: Re: ACL's revisited
To: Wojciech Puchar <wojtek@wojtek.3miasto.net>
From: Robert Elz <kre@munnari.OZ.AU>
List: tech-kern
Date: 08/26/2001 18:48:24
    Date:        Sun, 26 Aug 2001 11:14:10 +0200 (CEST)
    From:        Wojciech Puchar <wojtek@wojtek.3miasto.net>
    Message-ID:  <Pine.NEB.4.33.0108261113540.1663-100000@wojtek.3miasto.net>

  | is it extra file per normal file?

It was an extra file per ACL (or other data needing such a thing).
That is, if you have one ACL (generic permission set) that's the same
for all your files, there'd be exactly one extra file.  If you want to
set up a different permission set for every different file, then there
would be one extra file per file.

  | what about speed?

Hasn't been implemented yet, so no idea.   But given that the data can
be cached, and would typically be shared between many files (unlike the
inode number index into a huge file model) it shouldn't be too bad.
For files where no extra ACL beyond normal unix permission bits is
needed, the cost ought not be measureable (so even if you had it enabled
in your kernel & filesystem, there should be no cost apart from kernel
size, unless you're actually using it).

kre