Re: LSM hooks

From: Serge E. Hallyn (hallyn@private)
Date: Thu Mar 31 2005 - 19:34:52 PST

Quoting Chris Wright (chrisw@private):
> * John Richard Moser (nigelenki@private) wrote:
> > So, Which version of Linux will first implement stacking in LSM as per
> > Serge Hallyn's patches?
> None are ready yet.  Serge is still wading through performance testing.
> There's no telling about merging without a magic eightball, a handle on
> the performance issues, and some bonafide users.

Oh, just to keep anyone interested up to date:  It turns out nearly all
of the inordinate performance degredation I was seeing in the last set
of results which I reported was due to a prefetch weirdness on my ppc64
test system.  In particular, the hlist_for_each_entry macro automatically
prefetches tmp->next.  Since my tests were done with selinux+capability,
it was the case that tmp->next was always NULL (which always causes a
bad prefetch case on ppc64) and, to boot, never used, since the
comparison inside the loop always succeeded and immediately returned
the first entry.

A new set of results should hopefully be coming next week.


This archive was generated by hypermail 2.1.3 : Thu Mar 31 2005 - 19:35:33 PST