Re: Hardware-specific drivers/kernels archive (WAS: Re: FWD: RMS and Debian on his Toshiba)
Enrique Zanardi wrote:
> You mean just copying that kernel to a "standard" rescue disk? Or just
> making a boot floppy with that kernel and no syslinux bootloader?
> I ask because I suspect you were bitten by a syslinux bug, not a kernel
> one. (And the pcmcia modules on the drivers disk won't work with a custom
> kernel, you have to rebuild those too).
I copied the new kernel to a standard rescue disk, and did all the
rdev'ing outlined in the docs. Yes, the drivers disk didn't work ...
but then I compiled in everything I needed, so it wasn't that big a
deal.
> > Maybe nothing. It just rang a big bell in the back of my head when he
> > mentioned problems booting a Thinkpad 600 -- I use a custom kernel here,
> > and so I haven't had occasion to even try the Debian boot floppies on a
> > Thinkpad since about August or so. If you'd like, I'll test them out --
> > but only to the point of booting. (Don't have the space to try a
> > complete install ... :)
>
> Any help/test-report is welcome!
Okay, I'll try out the latest version and let you know what happens.
> I guess we can put a few custom compiled kernels under
> disks-i386/current/special-kernels . Two questions: which configs? and,
> who will mantain them?
Hmmm, this sounds like as good an excuse as any to apply as a developer.
:) I'll get right on it.
As for which configs -- could we have some input from everyone on the
list as to what are the most common reasons for using a custom kernel
(aside from getting new kernels, etc.)? I think a good start would be
some common laptop options --- such as a kernel with APM compiled in and
the APM_NOINTS definition fixed to support Thinkpads. A kernel with
those two features (and compiled zImage, I guess) would obviate -my-
need for custom kernels. I can't really speak for anyone else ... and
except for keeping up with dev kernels (another interesting idea for
kernel packages, maybe?) I have no other need for custom kernels.
PS Oh, I remember another obscure one -- some drivers' probing
interferes with SCSI card setups unless you give the "exclude
0xXXX-0xXXX" options, and then you can't use your SCSI. :) The VLB
Ultrastor 34F (used to have one) is one such card. I never did figure
out which probe caused the Ultrastor to get in a huff, but if there are
such known drivers, maybe a kernel or two without those drivers compiled
in would be an option. (And a non-SCSI kernel, to reduce size, might be
a desired option.) Let's not go overboard here, but what few options
would really make people's lives easier?
--
Anderson MacKay <mackay@rice.edu>
Reply to: