This is on Slackware current which ran fine on the (from cdrom) the 2.4
adaptec.s kernel using aic7xxx

/boot is on /dev/sda1 and is ext2

/ is on /dev/sda3 and is reiserfs

(kernel panic it locked up with 2.6 kernel)

I had read and followed readme.initrd

In /testing/packages/linux-2.6.16.9 I had installed all of the *.tgz except
for I did not install: kernel-headers-2.6.16.9-i386-1.tgz

In /boot I ran mkinitrd, thus:

mkinitrd -c -k 2.6.16.9 -m reiserfs:aic7xxx -f reiserfs -r /dev/sda3

I edited lilo.conf as per what readme.initrd said to do. I ran /sbin/lilo

Lastly I found a new rc.modules file in /etc/rc.d and I uncommented the line

/sbin/modprobe aic7xxx

I realize I'm using aic7xxx twice -- once from the initrd and once from
/etc/rc.d/rc.modules

Which is I don't know if I'm supposed to do this (aic7xxx twice) since I
don't know if initrd is just initial and then goes away or if initrd
stays or exists long term.

When panic/locked the error had something to do with aic7xxx that it was
unable to find a correct entry in aic7xxx

invalid entry (in aic7xxx) it said.

So, it appeared that it searched aic7xxx and then reported as I
mentioned above.

Could my problem be due to my mentioned the using aic7xxx twice?

Or, what else? This is my very first time at 2.6 kernel.

I've been running Slack 10, 10.1, 10.2 with 2.4 kernels for over a year
and a half with no problem. Thanks.

--
abc alan_cu b c n u

Alan_C <mtbr0228AT@sbcglobaldot.net> wrote:
> Lastly I found a new rc.modules file in /etc/rc.d and I uncommented the line
>
> /sbin/modprobe aic7xxx

That one must be in the initrd too and loaded there.
/etc/rc.d/rc.modules is only done after the disk is accessable and then,
of course, both aic7xxx as well as reiserfs HAVE already been loaded,
so "won't do" anything anymore in the rc.modules.

> Which is I don't know if I'm supposed to do this (aic7xxx twice) since I
> don't know if initrd is just initial and then goes away or if initrd
> stays or exists long term.

initrd is temporary (and replaced by the "real root disk"), the loaded kernel,
INCLUDING all loaded modules, stays in memory, so no need to install modules
already loaded FROM the initrd setting.

> invalid entry (in aic7xxx) it said.

You may need more of the SCSI support for the testing kernel.

Or, better, build your own 2.6 kernel with SCSI support, aic7xxx and reiserfs
BUILT-in, so that you won't need the initrd at all anymore (this is basically
what the adaptec.d (2.4) kernel _is_, a kernel with SCSI support and all
Adaptec adaptor support built-in).

> Could my problem be due to my mentioned the using aic7xxx twice?

No, the 2nd time it shouldn't do anything as the module is already there
(try removing it in rc.modules and looking with the "lsmod" command,
you should see something like:
aic7xxx 146644 0
scsi_transport_spi 18208 1 aic7xxx

That shows that aic7xxx also needs a module scsi_transport_spi, so that
may have to be in the initrd too (this - of course - is a 2.6.16 kernel,
so it MAY be different in the testing one, which is older).
That is one of the problems with the initrd, you need to determine
"automatically loaded, because of dependancies" modules too because
they TOO have to be in the initrd.
I also see, in my system, a scsi_transport_fc module, of which I don't
know when it is needed.
Of course we're using our own build, currently of 2.6.16.7, so the list
of modules will be different.

As I only need the aic7xxx module for the tape drive (DDS4) or external disks,
I don't need it built-in, so it is just a module here.
Ext3 and ReiserFS support, though, has been built-in here.
--
************************************************** ******************
** Eef Hartman, Delft University of Technology, dept. EWI/TW **
** e-mail: E.J.M.Hartman@math.tudelft.nl, fax: +31-15-278 7295 **
** snail-mail: P.O. Box 5031, 2600 GA Delft, The Netherlands **
************************************************** ******************

Eef Hartman <E.J.M.Hartman@math.tudelft.nl> writes:

> Alan_C <mtbr0228AT@sbcglobaldot.net> wrote:
> >
[ snip ]
> initrd is temporary (and replaced by the "real root disk"), the loaded kernel,
> INCLUDING all loaded modules, stays in memory, so no need to install modules
> already loaded FROM the initrd setting.

Ok. So, the via initrd loaded module stays loaded, even once initrd has gone.
Therefore, there's no need to once again load the same module from
/etc/rc.d/rc.modules
>
> > invalid entry (in aic7xxx) it said.
>
> You may need more of the SCSI support for the testing kernel.

Bingo!!! the Thunder! now emits at low volume from the 18GB scsi HD whilst
/testing/packages/2.6.16.9_kernel powered Slackware Current boots/loads.
Runs super! I'm using it now to write this.

All I did so far is boot from install cd disk #1 and at boot prompt enter:

boot: adaptec.s root=/dev/sda3 noinitrd ro

Once booted up, I did root in a terminal and redid my mkinitrd command then ran
/sbin/lilo and then copied back the unmodified /etc/rc.d/rc.modules file
(so as to not load the aic7xxx from there).

Formerly (didn't work, kernel panic):
mkinitrd -c -k 2.6.16.9 -m reiserfs:aic7xxx -f reiserfs -r /dev/sda3

Now, my redid (which works):
mkinitrd -c -k 2.6.16.9 -m reiserfs:scsi_transport_spi:aic7xxx -f reiserfs
-r /dev/sda3

>
> Or, better, build your own 2.6 kernel with SCSI support, aic7xxx and reiserfs
> BUILT-in, so that you won't need the initrd at all anymore (this is basically
> what the adaptec.d (2.4) kernel _is_, a kernel with SCSI support and all
> Adaptec adaptor support built-in).

Will do. I've several identical hardware. And have disk image (from before
experimented with the kernel). Will clone onto the other hardware and try it.

But for now, I'm up and running.

Actually I started off by build my own kernel. But in the make menu there were
SO many choices. And it's my first time ever. So I bailed, opting to try the
Slack testing kernel.
>
> > Could my problem be due to my mentioned the using aic7xxx twice?
>
> No, the 2nd time it shouldn't do anything as the module is already there
> (try removing it in rc.modules and looking with the "lsmod" command,
> you should see something like:
> aic7xxx 146644 0
> scsi_transport_spi 18208 1 aic7xxx
>
> That shows that aic7xxx also needs a module scsi_transport_spi, so that
> may have to be in the initrd too

Yep. After redoing my mkinitrd as per enclosed above and fixing rc.modules
so it does not load aic7xxx what I have is:

root@AB60R:~# lsmod
Module Size Used by
[ snipped many modules ]
aic7xxx 146612 3
scsi_transport_spi 18944 1 aic7xxx
reiserfs 238064 1

Those three loaded by the initrd.

> That is one of the problems with the initrd, you need to determine
> "automatically loaded, because of dependancies" modules too because
> they TOO have to be in the initrd.

Now I see that aic7xxx and scsi_transport_spi are two modules that form a combo,
or are applicable to this, as you mentioned, "dependencies" issue.
--

keywords: 2.6.16.9 kernel testing kernel 2.6 installpkg kernel 2.6 initrd
mkinitrd module dependencies

--
Alan C a.k.a abc alan_cu b c n u