Arlindo Sierra Neto said...
marc wrote:
Arlindo Sierra Neto said...
I followed this steps
http://forums.gentoo.org/viewtopic-t-270763.html
My cdrom worked fine.
The problem is that I have to install debian with kernel 2.4 and then compile an 2.6 kernel with the modification in libata.h
It workerd for me.
Excellent. Could you tell me which kernel you compiled, please. I'm just
downloading 2.6.12-1 to compile, since that is the latest in
testing/etch and hence has a condig file available.
Would it be possible to 'borrow' your config file, please? (Plse email
if it's okay.)
This is my first kernel compile, so I just wanted to check that I'm
starting with the right source. Thanks.
Marc,
When you install a Debian kernel package, there is a config file in
/boot/config-2.6.xx
If you don't find one I'll send you mine. I'm running on kernel 2.6.13.1
Okay, my latest update.
I'm compiling 2.6.12.1 downloaded from www.kernel.org. I have not
applied the suggested patch to ATA_ENABLE_ATAPI yet, because I want to
first ensure that the kernel I compile falls over in the same way as the
deb version, before continuing.
For my first try, I took /boot/config-2.6.12-1-686 into menuconfig,
changed nothing, but saved the result as the default (/usr/src/linux-
2.6.12.1/.config).
# first question: Is menuconfig the appropriate/best tool for this job?
This kernel compiled okay. I deb'd it, and installed. This locked up on
boot as expected. Excellent!
I applied acpi=off to the grub line, rebooted. This tried to boot, but
the error I received was different to the deb 2.6.12.1 error. I didn't
write it down, but it was failing to mount root - which is kinda
terminal.
Suspecting the configs, I noticed that the original (from /boot/) and
the "supposedly" unchanged version produced by menuconfig were different
sizes. A compare confirmed this. Only about seven or eight differences,
but one is enough.
Should this happen? It's easy to repeat.
Onward.
Next, I simply copied /boot/config-2.6.12-1-686 to /usr/src/linux-
2.6.12.1/.config. I then ran
# make-kpkg clean
# fakeroot make-kpkg --append_to_version marc kernel_image
This stops with:
PNP EIDE support (BLK_DEV_IDEPNP) [N/y/?] (NEW)
awaiting input.
Two questions:
- Shouldn't this compile?
- What do you suggest I answer in response to the idling script?
I'm compiling this under 2.4.27-2 on stable.
Is stable an issue when compiling this kernel, which is in testing?
I carried on running the script for a bit, entering essentially random
responses. There seem to be an awful lot of interruptions. So, now I'm
puzzled. Whatever menuconfig did to the original config file avoided
these halts to the compile. That's a good thing. But then, why doesn't
the "clean" config file compile without the interruptions?
I could ask a few more things, but hopefully that's enough for someone
to slap me for a stupid mistake and get me rolling... or at least
staggering forward.