A few notes on testing

Andrew Benton andy at benton987.fsnet.co.uk
Tue Jul 13 15:53:49 PDT 2004



Matthew Burgess wrote:
> On Tue, 13 Jul 2004 10:30:03 +0100
> Ian Molton <spyro at f2s.com> wrote:
> 
> 
>>On Mon, 12 Jul 2004 12:20:54 -0500
>>James Robertson <jwrober at linuxfromscratch.org> wrote:
>>
>>
>>>That is good to know.  Was not aware of that one.  Is proc going 
>>>entirely away in favor of that new sysfs thing?
>>
>>No. its just going to be for process information, as it always should
>>have been.
> 
> 
> Then I wonder how hotplug will do it's coldplugging then?  I mean the
> *.rc scripts currently use /proc/[bus] (where bus is USB, PCI, etc.)
> to determine what hardware is present on boot.  Maybe it'll force
> hotplug to only do hotplugging (which arguably should be the case now
> anyway).  I still wonder though what happens if one compiles an
> "always-present" hardware device driver as a module though in such a
> situation.  The module won't be loaded until access is attempted, but
> access can't be attempted because the device isn't present yet
> (assuming udev here).  Maybe they're expecting us to make manual calls
> to 'mknod' within our bootscripts?  Or maybe they're trying to force
> people to do the Right Thing with regard to modules - i.e. only use them
> for hotpluggable devices.  Sucks to have an NVidia or ATI graphics card
> then eh? :)
> 
> Regards,
> 
> Matt.
The new Nvidia module is working well with hotplug and udev. It loads on 
boot. Hotplug also has no trouble synthesising a hotplug event for my 
speedtouch modem even though I compiled the driver into the kernel. It 
loads the firmware on boot and dials out.



More information about the lfs-dev mailing list