On 8/15/07, Bryan Kadzban <[EMAIL PROTECTED]> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: RIPEMD160
>
> Dan Nicholson wrote:
> > On 8/15/07, Kevin Day <[EMAIL PROTECTED]> wrote:
> >> All this patch does is change the in kernel default to
> >> /sbin/udevtrigger from /sbin/hotplug.
> >> This way there should be no need to echo '/sbin/udevtrigger' >>
> >> /proc/sys/kernel/hotplug
> >
> > But we `echo "" > /proc/sys/kernel/hotplug' so the kernel won't
> > execute anything on uevents. It would be a bad idea to execute
> > udevtrigger for every event processed. It would be better if you just
> > changed it to be "". Unless I'm missing something.
That explains a lot, I haven't updated udev successfully until
recently so I was not as up-to-date on the current LFS practices.
With that in mind, I've tweaked the patch accordingly

(It seems I was the one missing something)

>
> Nope, you're not missing anything -- the current LFS udev script echos
> nothing into /proc/sys/kernel/hotplug before starting udevd; see line 77
> of lfs/init.d/udev in the trunk/bootscripts repo.
>
> (It could also echo the empty string into /sys/kernel/uevent_helper,
> since that's the newer file to use.  The other may end up going away in
> a future kernel.)
>
> And udevtrigger is *completely* the wrong program to be using for that
> anyway -- udevtrigger creates fake uevents for *all* kobjects, not just
> the "current" kobject.  You'll end up with O(N^2) uevents, instead of
> plain old O(N), because for every uevent, you'll re-run the events for
> all the old kobjects also.  Bad, bad idea...  ;-)
Yep, thats probably why I was never able to ugprade to newer udev's for so long.

>
> I would have no problem with setting it to "", though.  Well, as long as
> nobody uses the hotplug program in their initramfs (and nobody should
> be, now that udevtrigger exists, but I don't know for sure).

How's this version?
I renamed the patch accordingly, but if you can come up with a better
patch name, be my guest.

-- 
Kevin Day

Attachment: linux-2.6.22.2-no_default_uevent_helper-2.patch
Description: Binary data

-- 
http://linuxfromscratch.org/mailman/listinfo/patches
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Reply via email to