UltraPenguin 1.1.9 install problem

UltraPenguin 1.1.9 install problem

[Craig Ruff]

I grabbed up-1.1.9.iso from ultra.penguin.cz this weekend to give a try.
I burned a CD from the iso image.  I tried installing on my Ultra-1
today, and ran into a problem.  The kernel boots ok, but there appears
to be a problem in the initial ram disk image.  The kernel finds all of
the hardware on the system, then emits this error:

/linuxrc: 51:syntax error: "done" unexpected (expecting "fi")
sr0: disc change detected
kernel panic: VFS: unable to mount root fs on 0b:00

I don't have my Linux system available at the moment to poke around in the 
ram disk image.  Anyone else seen it?

-- 

Craig Ruff      	NCAR			cruff@ucar.edu
(303) 497-1211  	P.O. Box 3000
			Boulder, CO  80307	Amateur Call KI0NO

Re: UltraPenguin 1.1.9 install problem

[Jakub Jelinek]

> 
> I grabbed up-1.1.9.iso from ultra.penguin.cz this weekend to give a try.
> I burned a CD from the iso image.  I tried installing on my Ultra-1
> today, and ran into a problem.  The kernel boots ok, but there appears
> to be a problem in the initial ram disk image.  The kernel finds all of
> the hardware on the system, then emits this error:
> 
> /linuxrc: 51:syntax error: "done" unexpected (expecting "fi")
> sr0: disc change detected
> kernel panic: VFS: unable to mount root fs on 0b:00
> 
> I don't have my Linux system available at the moment to poke around in the 
> ram disk image.  Anyone else seen it?

Thanks for report, I have fixed it in my copy and will put it up tomorrow
with other changes. In the mean time, you might want to boot net with the
included tftpboot.img and choose Local CD-ROM, that should work.

Cheers,
    Jakub
___________________________________________________________________
Jakub Jelinek | jj@sunsite.mff.cuni.cz | http://sunsite.mff.cuni.cz
Administrator of SunSITE Czech Republic, MFF, Charles University
___________________________________________________________________
Ultralinux - first 64bit OS to take full power of the UltraSparc
Linux version 2.1.130 on a sparc64 machine (3958.37 BogoMips)
___________________________________________________________________

Re: UltraPenguin 1.1.9 install problem

[Roger J. Allen]

First, I want to thank everyone who worked on this.  It must have taken
a huge amount of time, and I appreciate it greatly.

On Mon, 14 Dec 1998, Craig Ruff wrote:

> I grabbed up-1.1.9.iso from ultra.penguin.cz this weekend to give a try.
> I burned a CD from the iso image.  I tried installing on my Ultra-1
> today, and ran into a problem.  The kernel boots ok, but there appears
> to be a problem in the initial ram disk image.  The kernel finds all of
> the hardware on the system, then emits this error:
> 
> /linuxrc: 51:syntax error: "done" unexpected (expecting "fi")
> sr0: disc change detected
> kernel panic: VFS: unable to mount root fs on 0b:00
> 
> I don't have my Linux system available at the moment to poke around in the 
> ram disk image.  Anyone else seen it?
> 

I got the same thing trying to boot a CD on an Ultra 1 Creator 3D Model
170E, a SparcStation 2, and a SparcStation IPC.  The Ultra1 is running
UltraPenguin 1.1 and Solaris 2.6.  The SS2 is running RedHat 5.2 and
Solaris 2.6.  The IPC is running RedHat 5.2.

I did find a way to install it on the Ultra1, but not the SS2 or IPC.

I was able to mount the CD on the Ultra1 with Solaris 2.6.  That let me
copy the RedHat directory to a UFS partition.  Then, I could boot the
tftpboot.img from either the net or from an Exabyte 8505XLE 8mm tape
drive and do a hard disk install from the UFS partition.  There were
a few error messages on one of the virtual consoles during the mount
about the type of UFS partition defaulting to "old" instead of "sun",
but it did work.  If anyone wants the exact error messages, then I can
re-install and take better notes.  It took a little less than 400MB on
the UFS partition to hold the RedHat directory.  I did not test the
install very much.  The disks had already been partitioned, so I did not
do much with disk druid/fdisk.  There was some problem installing the
urw-fonts package.  This is from the install.log:

Installing urw-fonts.
execution of script failed
	cannot open RPM file urw-fonts-1.0-4.noarch.rpm: Success

That rpm is in the RedHat/RPMS directory.  I did not install any of the
latin fonts, so that might be the problem.

The bootable diskette that it tried to make during the install does
not boot, either.  It just prints "SI".

When running PPP for a few minutes, it will hang with a skput error, or
something like that.  Also, the scsi tape still has problems.  I'll get
more info on this later.  I was spending more time testing different
ways to install.

The tftprescue.img did not boot any way that I tried.  With a "boot net"
on the Ultra1, once, it failed mounting the root with (I think that I
scribbled this):

request_module[block-major-226]: Root fs not mounted
VFS: Cannot open root device e2:2f
Kernel Panic: VFS: unable to mount root fs on e2:2f
Press L1-A to return to the boot prom

The rest of the time, it would fail right after the TILO with:

339400
TILO

Fast Data Access MMU Miss

ok

That "Fast Data Access MMU Miss" error always showed up with a "boot
tape" for the tftprescue.img.

For the SS2, "boot net" of the tftpboot.img displayed:

2b3e00 receive failed
TILO

and ran until it got to the:

RAMDISK: compressed image found at block 0

Then the lights went out (the green one on the SS2, and DTR on the break
out box on the serial port), and it was dead.  It took about 7 minutes
of waiting after the 2b3e00 showed up until the "receive failed"
appeared.

I also tried booting from floppy, but none of the three machines would
boot with the diskettes that I made under Solaris 2.6.  I forget which
device I used, but I think that it was something like:

dd if=boot.img of=/dev/diskette bs\x1440k 

Trying to boot the tftpboot.img from floppy just printed "SI".
Sometimes, it would also print "SIProgram terminated" and go to the OK
prompt.

Booting the rescue.img from floppy produced:

Bad magic number in disk label
Can't open disk label package

-- 
Roger J. Allen                Rush-Presbyterian-St. Luke's Medical Center
System Administrator                                   Chicago,  IL   USA
Surgical Information Systems                       Voice:  (312)-942-4825
Internet: rja@sis.rpslmc.edu                         FAX:  (312)-733-6921

Re: UltraPenguin 1.1.9 install problem

[Jakub Jelinek]

> I did find a way to install it on the Ultra1, but not the SS2 or IPC.
> 
> I was able to mount the CD on the Ultra1 with Solaris 2.6.  That let me
> copy the RedHat directory to a UFS partition.  Then, I could boot the

You could just choose Local CD-ROM and install the packages from CD
directly.

> tftpboot.img from either the net or from an Exabyte 8505XLE 8mm tape
> drive and do a hard disk install from the UFS partition.  There were
> a few error messages on one of the virtual consoles during the mount
> about the type of UFS partition defaulting to "old" instead of "sun",
> but it did work.  If anyone wants the exact error messages, then I can
> re-install and take better notes.  

No, that should be fixed in my copy already since yesterday.

> It took a little less than 400MB on
> the UFS partition to hold the RedHat directory.  I did not test the
> install very much.  The disks had already been partitioned, so I did not
> do much with disk druid/fdisk.  There was some problem installing the
> urw-fonts package.  This is from the install.log:
> 
> Installing urw-fonts.
> execution of script failed
> 	cannot open RPM file urw-fonts-1.0-4.noarch.rpm: Success

It is on my TODO list, I got the same bugreport.
> 
> That rpm is in the RedHat/RPMS directory.  I did not install any of the
> latin fonts, so that might be the problem.
> 
> The bootable diskette that it tried to make during the install does
> not boot, either.  It just prints "SI".

Ouch, one further thing to check.
> 
> When running PPP for a few minutes, it will hang with a skput error, or
> something like that.  Also, the scsi tape still has problems.  I'll get
> more info on this later.  I was spending more time testing different
> ways to install.
> 
> The tftprescue.img did not boot any way that I tried.  With a "boot net"
> on the Ultra1, once, it failed mounting the root with (I think that I
> scribbled this):
> 
> request_module[block-major-226]: Root fs not mounted
> VFS: Cannot open root device e2:2f
> Kernel Panic: VFS: unable to mount root fs on e2:2f
> Press L1-A to return to the boot prom

Was this from tftprescue? That's strange. I think tftprescue with its
current bug should just tell you Fast TLB data miss and die off.
I have fixed it in the kernel already, just need to test it.
But the problem is on sparc64 only, 32bit boxes provided they have enough
RAM should be just fine (correct me if I'm wrong).

> For the SS2, "boot net" of the tftpboot.img displayed:
> 
> 2b3e00 receive failed
> TILO
> 
> and ran until it got to the:
> 
> RAMDISK: compressed image found at block 0
> 
> Then the lights went out (the green one on the SS2, and DTR on the break
> out box on the serial port), and it was dead.  It took about 7 minutes
> of waiting after the 2b3e00 showed up until the "receive failed"
> appeared.

How much memory does it have?
> 
> I also tried booting from floppy, but none of the three machines would
> boot with the diskettes that I made under Solaris 2.6.  I forget which
> device I used, but I think that it was something like:
> 
> dd if=boot.img of=/dev/diskette bs\x1440k 
> 
> Trying to boot the tftpboot.img from floppy just printed "SI".
> Sometimes, it would also print "SIProgram terminated" and go to the OK
> prompt.
> 
> Booting the rescue.img from floppy produced:
> 
> Bad magic number in disk label
> Can't open disk label package

rescue.img is just a second floppy, so this is expected. It has to be booted
with the floppy made either during installation or afterwards (in that case
using mkbootdisk).

Cheers,
    Jakub
___________________________________________________________________
Jakub Jelinek | jj@sunsite.mff.cuni.cz | http://sunsite.mff.cuni.cz
Administrator of SunSITE Czech Republic, MFF, Charles University
___________________________________________________________________
Ultralinux - first 64bit OS to take full power of the UltraSparc
Linux version 2.1.130 on a sparc64 machine (3958.37 BogoMips)
___________________________________________________________________

Re: UltraPenguin 1.1.9 install problem

[Jakub Jelinek]

> I was able to get tftpboot.img to boot and the install procedure appeared
> to run ok.  However, I could not boot from my Linux drive (sd@2,0).

The strange thing is that SILO did not install itself properly.
BTW: Where does the unexpected message in 3 comes from? When it was printed?
I haven't seen it ever, nor I know about SILO/kernel to ever print something
like that.

> I got an "unexpected message in 3" error.  tftprescue.img did not work
> either.

This is known bug and I have it fixed in my copy.

>   Rebooting the install and switching to the shell VC, I poked
> around.  I mounted the Linux file systems (they look fine), I ran
> silo manually:
> 
> 	silo -r /mnt -i /boot/ultra.b -u
> 
> resulted in a complaint that it could not open the super block on
> /dev/sdc1.  The device entry does exist in /mnt/dev, and seems to
> be correct.  fdisk sees the proper partition table, and sdc1 starts
> at cylinder 0.

This is a known bug and will be fixed in SILO 0.8.5.
The issue is that SILO 0.8.4 does not like stale symlinks in /dev, so if you
want to do it with 0.8.4, you should from the second VC poke like this:

chroot /mnt
mount -t proc /proc /proc
silo -f

> 
> Any clues?  Should I try your new CD-ROM image?
> 

I haven't put it there yet, hopefully today.

Cheers,
    Jakub
___________________________________________________________________
Jakub Jelinek | jj@sunsite.mff.cuni.cz | http://sunsite.mff.cuni.cz
Administrator of SunSITE Czech Republic, MFF, Charles University
___________________________________________________________________
UltraLinux  |  http://ultra.linux.cz/  |  http://ultra.penguin.cz/
Linux version 2.1.130 on a sparc64 machine (3958.37 BogoMips)
___________________________________________________________________

Re: UltraPenguin 1.1.9 install problem

[Jakub Jelinek]

> 
> I tried several things to get running, but it looks like the boot
> process doesn't like the Seagate ST15230N drive (SCSI ID 2) that
> I first installed UltraPenguin on.  So I switched to a Fujitsu
> internal drive (SCSI ID 1), and Linux now boots ok.  I'm running it
> right now.

BTW: Have you been booting the right device, ie. probably 
disk2 in your case?

Cheers,
    Jakub
___________________________________________________________________
Jakub Jelinek | jj@sunsite.mff.cuni.cz | http://sunsite.mff.cuni.cz
Administrator of SunSITE Czech Republic, MFF, Charles University
___________________________________________________________________
UltraLinux  |  http://ultra.linux.cz/  |  http://ultra.penguin.cz/
Linux version 2.1.130 on a sparc64 machine (3958.37 BogoMips)
___________________________________________________________________

Re: UltraPenguin 1.1.9 install problem

[Roger J. Allen]

> > I did find a way to install it on the Ultra1, but not the SS2 or IPC.
> > 
> 
> > For the SS2, "boot net" of the tftpboot.img displayed:
> > 
> > 2b3e00 receive failed
> > TILO
> > 
> > and ran until it got to the:
> > 
> > RAMDISK: compressed image found at block 0
> > 
> > Then the lights went out (the green one on the SS2, and DTR on the break
> > out box on the serial port), and it was dead.  It took about 7 minutes
> > of waiting after the 2b3e00 showed up until the "receive failed"
> > appeared.
> 

The new tftpboot.img fixes that problem and now lets me boot the Sparc
Station 2.

Thanks for the quick work.

-- 
Roger J. Allen                Rush-Presbyterian-St. Luke's Medical Center
System Administrator                                   Chicago,  IL   USA
Surgical Information Systems                       Voice:  (312)-942-4825
Internet: rja@sis.rpslmc.edu                         FAX:  (312)-733-6921

Re: UltraPenguin 1.1.9 install problem

[Jakub Jelinek]

> 
> The XSun server seems to ignore the -ar1 and -ar2 key repeat settings.
> The keyboard is too speedy for my taste.

That's why I'm adding now limited support for XF86Config to Xsun* (keyboard
and files sections), so that one can easily configure xkb keyboards, set
repeat setting and configure fonts (well, most is doable by xkbcomp or xset
from xinitrc &co. scripts, but it is not that easy for an unexperienced
user).
> 
> All my other testing seems to be ok.  I noticed the hme0 interface
> failed to autoconfig against our Cisco 5000 Etherswitch, but Solaris
> and Irix fail to do so also.  Are there module arguments to force
> the speed/duplex settings?

David Miller is going to write an eth-tool utility, which will allow setting
stuff like that.

Cheers,
    Jakub
___________________________________________________________________
Jakub Jelinek | jj@sunsite.mff.cuni.cz | http://sunsite.mff.cuni.cz
Administrator of SunSITE Czech Republic, MFF, Charles University
___________________________________________________________________
UltraLinux  |  http://ultra.linux.cz/  |  http://ultra.penguin.cz/
Linux version 2.1.130 on a sparc64 machine (3958.37 BogoMips)
___________________________________________________________________