I leave the project

mail1234 at koevis.demon.nl mail1234 at koevis.demon.nl
Wed Jul 7 22:39:46 PDT 2004



> Matthew Burgess wrote:
>
> > Then please accept my apologies.  I read between the lines wrongly!
>
> No problem.
>
> > As far as hotplug goes, I'm quite happy for it to stay in unstable.  It
> > vastly simplifies module loading under a udev setup, but from what I
> > could discern, which admittedly may not be correct, is that it was being
> > developed into a full on auto-hardware detection and support mechanism,
> > which I don't believe is suitable for LFS.  hotplug, if used solely for
> > loading modules for hotpluggable devices, I don't see a problem with
> > (other than the technical problems currently associated with it).
>
> Unfortunately if you put hotplug on the system you are going to have to
> deal with some of those other issues. Granted, it's not necessary to
> have fully hotplug-driven networking, or hotplug-driven filesystem
> mounting. Those are outside the scope of LFS as it is defined today.
>
> However, there are issues raised that don't even concern the hotplug
> package; some of them are just between module loading (either manually
> or with Zack's "modules" script) and accessing the devices created by
> those modules. Those issues will exist without the hotplug package even
> being installed, because they are timing issues between the module load,
> the hotplug event being sent to userspace and udev processing. Anyone
> who expects that every device supported by the modules they load with
> the "modules" script to appear and be ready for use before the next
> bootscript runs has the potential to be rudely surprised :-)
>
> The only permanent solution to that problem is to move to hotplug-based
> system startup, which means that no attempts are made to use any devices
> until after the system has identified them and created the relevant
> device nodes. This is where the hotplug package comes back into the
> picture. Personally I will be interested to see what happens when
> regular users start building testing with 2.6/udev and without any
> warning about what can happen if they use modules for >their devices.

I am not very linux savvy , so I have a question. Whats all that hotplugging
about? Isnt hotplugging implemented in 2.6 by default?
could anyone please awnser my question? thanks in advance!

Xanni




More information about the lfs-chat mailing list