Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 12 Apr 2008 14:28:54 -0400
From:      Coleman Kane <cokane@FreeBSD.org>
To:        Jeremy Messenger <mezz7@cox.net>
Cc:        gnome@freebsd.org
Subject:   Re: Seahorse issues
Message-ID:  <1208024934.1327.9.camel@localhost>
In-Reply-To: <op.t9ifvbe79aq2h7@mezz.mezzweb.com>
References:  <47FD09AC.2020907@FreeBSD.org> <1207776230.61729.28.camel@shumai.marcuscom.com> <47FD34E8.2000005@FreeBSD.org> <1207807915.61729.40.camel@shumai.marcuscom.com> <op.t9id5keu9aq2h7@mezz.mezzweb.com> <1208022563.82222.22.camel@shumai.marcuscom.com> <op.t9ifvbe79aq2h7@mezz.mezzweb.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 2008-04-12 at 13:19 -0500, Jeremy Messenger wrote:
> On Sat, 12 Apr 2008 12:49:23 -0500, Joe Marcus Clarke  
> <marcus@marcuscom.com> wrote:
> 
> > On Sat, 2008-04-12 at 12:42 -0500, Jeremy Messenger wrote:
> >> On Thu, 10 Apr 2008 01:11:55 -0500, Joe Marcus Clarke
> >> <marcus@marcuscom.com> wrote:
> >>
> >> <snip>
> >> > The problem is the fact that FreeBSD's mlock() requires setuid
> >> > privileges, and thus seahorse cannot allocate secure memory.  The
> >>
> >> Yesterday, I have found archives about mlock() in freebsd-arch@.
> >>
> >> http://lists.freebsd.org/pipermail/freebsd-arch/2006-July/005496.html
> >
> > Yes, this thread talks about the problem exactly.  The patch I just sent
> > out attempts to address this concern using a user-settable sysctl.
> > Peter is suggesting this be handled automatically by setting a
> > reasonable default limit on RLIMIT_MEMLOCK.
> 
> Yeah and even rwatson liked his suggest. I like automatically better, but  
> tweak in sysctl is fine with me too. Some hardcore probably prefer sysctl  
> than automatically one.
> 
> >> It leads to:
> >>
> >> http://people.freebsd.org/~kib/overcommit/index.html
> >>
> >> I am not sure if it's useful for this issue.
> >
> > This doesn't look like it will help this issue.  This is dealing with
> > overcommitting swap.
> 
> It's what I though so.
> 
> Cheers,
> Mezz
> 
> > Joe
> 

If we could turn this into a per-user, system-wide value, I think that
would be the best approach. However, this probably ends up violating the
canonical meaning of RLIMIT_MEMLOCK (if there is even a standard
set-in-stone meaning for it).

I am curious if we could use something like the MAC facility to provide
a method for preventing mlock() access to some non-root users, while
providing it to others.

--
Coleman Kane




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1208024934.1327.9.camel>