Ooops, of course I should have detailed my hardware, i.e. what I mean by "fairly ancient notebook" :
Toshiba Satellite Pro 410CDT dated from 1996 =:o)
Pentium 90 MHz
40 MB RAM (max. equipped ;-)
60 GB E-IDE HD (70x the capacity compared to the beginning)
no CD-ROM ... but fortunately a 16-bit PCMCIA Fast-EtherNet-Card
BIOS: PnP 1.0a, NOT supporting Int 13h Extensions (LBA)
(... which means that the BIOS (and "legacy" GRUB) is only happy if the
HD has a /dev/hda1 of only 16MB inhabiting the /boot partition
... once having booted Linux and Win98 feel fine with additional /root and /home partitions
... but that's another quite lengthy story ...)
The minimum memory required during boot had risen from 32MB with TCL2.0 to 48MB with TCL2.1, which breaks my above mentioned RAM limit :-(( .
So I concentrate on TCL2.0 at first ... being aware that for all following TCL releases I would anyway have to fix the initramfs tinycore.gz .
Meanwhile I've tried to dig even deeper :
To get a thorough _comparison_ between
TCL2.0 booting badly to a kernel panic
and
bzImage-2.0 + tinycore-1.4.3.gz booting nicely in this combination
I implemented a _serial_ terminal connection between the "culprit" and my Ubuntu desktop pc
(1) connecting a Null-Modem-Cable between both serial ports
(btw. 3-wired was sufficient compared to lots of oversized proposed Null-Modem-Cables www...)
and
(2) setting up an appropriate GRUB's menu.lst on the "culprit" and employing minicom on Ubuntu desktop pc
(It just took a couple of hours till working smoothly ... incl. restoring my old soldering iron to life again ... ;-)
The entries in my grub's menu.lst give rise to the highest possible level of debugging :
title TinyCore bzImage-2.0 + tinycore-2.0.gz + debugging to SERIAL CONSOLE
kernel (hd0,0)/boot/bzImage-2.0 noscsi nousb noagp nofirewire acpi=off pci=off S debug loglevel=7 initcall_debug console=tty0 console=ttyS0
initrd (hd0,0)/boot/tinycore-2.0.gz
title TinyCore bzImage-2.0 + tinycore-1.4.3.gz + debugging to SERIAL CONSOLE
kernel (hd0,0)/boot/bzImage-2.0 noscsi nousb noagp nofirewire acpi=off pci=off S debug loglevel=7 initcall_debug console=tty0 console=ttyS0
initrd (hd0,0)/boot/tinycore-1.4.3.gz
Note: I just renamed the kernel and initramfs files with version suffixes to keep track of all the combinations I already tested ...
Attached is the comparison generated by
diff -yW 200 minicom-output-file-of-bzImage-2.0+tinycore-1.4.3.gz-booting minicom-output-file-of-TCL2.0-booting
Result :
Aside their different execution times all the initcalls are identical - even if their return code is _NOT_ 0 !
The combination of bzImage-2.0+tinycore-1.4.3.gz is even successful despite of FATAL errors due to missing an appropriate modules.dep file !
Immediately after the line "Freeing unused kernel memory: 344k freed" BusyBox' init process starts on the "exotic" combination side while a Kernel panic occurs on "standard" TCL2.0 side :-((
calling 0xc0499c01 @ 1 calling 0xc0499c01 @ 1
initcall 0xc0499c01 returned 0 after 22 usecs initcall 0xc0499c01 returned 0 after 22 usecs
async_waiting @ 1 async_waiting @ 1
async_continuing @ 1 after 17 usec async_continuing @ 1 after 17 usec
Freeing unused kernel memory: 344k freed Freeing unused kernel memory: 344k freed
init started: BusyBox v1.13.3 (2009-03-18 12:08:31 PDT) | Kernel panic - not syncing: Attempted to kill init!
starting pid 40, tty '': '/etc/init.d/rcS' <
Booting tinycore_1.4.3 <
Running Linux Kernel 2.6.29.1-tinycore. <
Starting udev daemon for hotplug support...input: PS/2 Generic Mouse <
Done. <
Boot options checked. <
As you may imagine I'm somewhat helpless now ... =o-/
Any comments, hints, advice are welcome !
[attachment deleted by admin]