Re: Etch install on A1200. Floating point exception
Stephen R Marenka wrote:
On Wed, Jun 04, 2008 at 12:14:13AM +0100, Nigel wrote:
Hello,
I thought I'd give Debian a go on my Amiga A1200 (with 68030 CPU/MMU,
68882 FPU & 32MB RAM).
I downloaded a kernel (kernels/vmlinuz-2.6.24-1-amiga) and initrd
(hd-media/initrd.gz) from
http://people.debian.org/~smarenka/d-i/images-m68k/daily/
Transferred (the decompressed) kernel and initrd to my A1200 and booted
with:
amiboot -d -k vmlinuz-2.6.24-1-amiga -r initrd.gz root=/dev/ram video=pal
Interesting, I thought fb=false was also required for amiga. I suppose
that got fixed.
And my A1200 booted into the installer. (I am impressed!)
The installer proceeds in low memory mode, I select region then country
and reach the "Detecting hardware to find drives" screen. Once the
progress bar completes I get a blank blue screen, and after a minute or
so the message "Floating point exception" is displayed
/sbin/debian-installer exits, init restarts it and the "Detecting
hardware to find drives" is repeated. This loops around and around and
blocks the install.
I guess we're going to need to figure out that FPE.
I suspect my problem is caused by my choice of kernel and initrd coming
from a "daily" directory. So my question is:
Q) Am I using the correct kernel+initrd / install procedure to install
Etch on my A1200?
I would say yes. The etch release media wasn't tested (by me) after the
etch-m68k release (where the distribution name got changed). The daily
installer with suite etch-m68k is the best tested combination of d-i
since before etch released. (I just haven't done it lately.)
You need to add suite=etch-m68k and modules=etch-support on the kernel
command line to get etch, see also
<http://wiki.debian.org/DebianInstaller/M68k#issues>.
(Okay, I haven't verified that modules=etch-support is the correct
solution yet, but it sounds right.)
Unfortunately, none of this gets your past the FPE.
I haven't figured out what package generates the "Detecting hardware to
find drives" screen. I'd like to see what it's doing when the FPE gets
called.
I've discovered the FPE is coming from db_progress.
The line
db_progress START 0 $DEV_COUNT iso/scan/progress_title
as invoked from the script /var/lib/dpkg/info/iso-scan.postinst causes
the FPE.
And on my A1200 DEV_COUNT=0 (a divide by zero)?
So it seems there are 2 problems here
1 - iso-scan.postinst and db_progress do not handle the situation of no
devices found
2 - vmlinuz-2.6.24-1-amiga does not see my A1200's IDE hard disk
Regarding the second issue, when booting with the vmlinuz-2.6.24-1-amiga
kernel the boot messages relating to the ide interface are as follows:
Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
ide: Assuming 50MHz system bus speed for PIO modes; override with idebus=xx
ide0: Gayle IDE interface (A1200 style)
Driver 'sd' needs updating - please use bus_type methods
mice: PS/2 mouse device common for all mice
As you can see no hda is detected by 2.6.24 kernel, which I suspect
accounts for the zero causing the FPE later on.
Interestingly booting an older vmlinuz-2.6.18-4-amiga kernel does
discover hda.
Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
ide: Assuming 50MHz system bus speed for PIO modes; override with idebus=xx
ide0: Gayle IDE interface (A1200 style)
hda: ST9235AG, ATA DISK drive
ide0 at 0x80da0000 on irq 2
hda: max request size: 128KiB
hda: 409760 sectors (209 MB) w/64KiB Cache, CHS=985/13/32
<snip>
RDSK (512) hda1 (DOS^A)(res 2 spb 1) hda2 (DOS^A)(res 2 spb 1) hda3
(DOS^A)(res 2 spb 1) hda4 (DOS^A)(res 2 spb 1) hda5 (DOS^A)(res 2 spb 1)
hda6 (DOS^A)(res 2 spb 1) hda7 (DOS^A)(res 2 spb 1)
mice: PS/2 mouse device common for all mice
However I'm unable to mount initrd using vmlinuz-2.6.18-4-amiga so using
this older kernel for installation is not an option.
Hope this information helps.
Regards,
Nigel
Reply to: