Subject: Re: de-modularising for the win! Message-ID: 20080920105355.GA6663@nostromo.devel.redhat.com References: 20080918191355.GA9683@nostromo.devel.redhat.com 48D2D94E.2020902@redhat.com 20080918231301.GA4734@nostromo.devel.redhat.com 1221790735.15361.152.camel@aglarond.local MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: 1221790735.15361.152.camel@aglarond.local User-Agent: Mutt/1.5.18 (2008-05-17)
Jeremy Katz (katzj@redhat.com) said:
On Thu, 2008-09-18 at 16:13 -0700, Bill Nottingham wrote:
- killing the initrd for that general 90% case can be a big win
Ermm, the general 90% (or some large-ish generalizing percentage) are set up to use LVM. Which then requires an initrd.
Yes, but ... LVM is overkill, in general. (That's another discussion.)
On further consideration, though, the biggest issue with kicking out the initrd is getting the policy lodaed.
Bill
On Sat, Sep 20, 2008 at 06:53:55AM -0400, Bill Nottingham wrote:
Subject: Re: de-modularising for the win! Message-ID: 20080920105355.GA6663@nostromo.devel.redhat.com References: 20080918191355.GA9683@nostromo.devel.redhat.com 48D2D94E.2020902@redhat.com 20080918231301.GA4734@nostromo.devel.redhat.com 1221790735.15361.152.camel@aglarond.local MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: 1221790735.15361.152.camel@aglarond.local User-Agent: Mutt/1.5.18 (2008-05-17)
Jeremy Katz (katzj@redhat.com) said:
On Thu, 2008-09-18 at 16:13 -0700, Bill Nottingham wrote:
- killing the initrd for that general 90% case can be a big win
Ermm, the general 90% (or some large-ish generalizing percentage) are set up to use LVM. Which then requires an initrd.
Yes, but ... LVM is overkill, in general. (That's another discussion.)
On further consideration, though, the biggest issue with kicking out the initrd is getting the policy lodaed.
But, won't we all be using btrfs instead of lvm in F11? ;-)
On Sat, 20 Sep 2008 13:13:38 -0400 Kyle McMartin kyle@redhat.com wrote:
On further consideration, though, the biggest issue with kicking out the initrd is getting the policy lodaed.
/me wonders about the option of having selinux using request_firmware() to get its policy
On Saturday 20 September 2008 1:22:31 pm Arjan van de Ven wrote:
On Sat, 20 Sep 2008 13:13:38 -0400
Kyle McMartin kyle@redhat.com wrote:
On further consideration, though, the biggest issue with kicking out the initrd is getting the policy lodaed.
/me wonders about the option of having selinux using request_firmware() to get its policy
I've started ignoring most of the firmware loading threads a while ago so I've kinda lost most of the plot there ... a few questions:
* Is the firmware request asynchronous?
* Is there currently a way, or at least nothing in the way that would prevent a mechanism from bring created, to pause the boot process until the policy/firmware is loaded? I ask because we would want to make sure that SELinux policy was loaded before any services are started.
Regardless, we would still need to keep the current policy loading mechanism in place since we can't break userspace. You should post this idea on the SELinux list to get some further thought on this ...
Kyle McMartin wrote:
On Sat, Sep 20, 2008 at 06:53:55AM -0400, Bill Nottingham wrote:
Subject: Re: de-modularising for the win! Message-ID: 20080920105355.GA6663@nostromo.devel.redhat.com References: 20080918191355.GA9683@nostromo.devel.redhat.com 48D2D94E.2020902@redhat.com 20080918231301.GA4734@nostromo.devel.redhat.com 1221790735.15361.152.camel@aglarond.local MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: 1221790735.15361.152.camel@aglarond.local User-Agent: Mutt/1.5.18 (2008-05-17)
Jeremy Katz (katzj@redhat.com) said:
On Thu, 2008-09-18 at 16:13 -0700, Bill Nottingham wrote:
- killing the initrd for that general 90% case can be a big win
Ermm, the general 90% (or some large-ish generalizing percentage) are set up to use LVM. Which then requires an initrd.
Yes, but ... LVM is overkill, in general. (That's another discussion.)
On further consideration, though, the biggest issue with kicking out the initrd is getting the policy lodaed.
But, won't we all be using btrfs instead of lvm in F11? ;-)
Aw crap, you mean F10 was my only fleeting chance to make ext4 the default? :)
-Eric
kernel@lists.fedoraproject.org
-
Arjan van de Ven -
Bill Nottingham -
Eric Sandeen -
Kyle McMartin -
Paul Moore