Re: PCI hardware detection [RFC]
On Thu, Jul 19, 2001 at 07:21:36AM -0700, David Kimdon wrote:
> > [First we need to be able to reach _one_ supplementary source of data :
> > Network in this case, but this can be local disks too]
> > - GRUB as a bootloader
> Neat idea, using grub to choose modules.
As Glenn stressed, the problem is that bootloaders are different (even if
there are some projects/ideas to try to create a kind of micro-OS, putting
some stuff like limited shell, fs handling etc in common for different arch)
but here my main target is x86, so the bootstrapping procedure involving
basic detection by the bootloader is a possible way (note: FreeBSD
bootloader at least is planned to do a part of the hardware detection too).
> When we did pci hardware detection for d-i we used straight libdetect. It
> is missing some of the space optimizations that you have in here, but
> avoids the 're-inventing the wheel' problem. What is you feeling on ISA
> hardware? support it/don't?
Yes, support it too. But we all know that the P&P is tricky ;)
> I'm starting to work on d-i again, but all I really want is a next
> generation installer for Debian, and if it works for other OS's as well
> that would be a big bonus. I hope we can combine efforts, maybe even
> This idea for d-i is that the first disk (i386 only at the moment) does
> hardware detection of ethernet modules. After that the rest of the
> installer is downloaded over the ethernet and the installer can proceed
> with partitioning, base install, etc.
Same here. In fact the memory limits are avoided after that via a rootfs
(using initrd just to load the correct module for the NIC in order to access
supplementary resources via the device --- network fs). The trick is in fact
to try to have the smallest bootstrapping procedure and there seems to be
already a large part of the work done.
The last time I checked for libdetect it was such a huge beast and decided
to find another way to do the stuff.
I don't want to do something on my own, and do want to work with others. The
only problem is that I have some deadlines at work, and that I try to both
develop for a general purpose (d-i), and to address more peculiar and
limited needs at work ;)
Thierry LARONDE, Centre de Ressources Informatiques, Archamps - France