Troubleshooting failure of configuration.

[peasthope]

Andr, Sven & others,

http://www.ocp.inf.ethz.ch/wiki/Documentation/Installation
didn't ring any new bells.

This is the configuration on the HP Pavilion.  TraceMode="1"
produces nothing beyond "OBER"; but perhaps someone will
spot a flaw. 

Partitions.SetConfig IDE0#03
  BootVol1="AOS AosFS IDE0#03"
  AosFS="AosDiskVolumes.New AosDiskFS.NewFS"
  Boot="AosDisplayLinear.Install"
  Init="105"
  Boot1="AosKeyboard;AosMousePS2"
  Boot2="WindowManager.Install;MainMenu.Open"
  Boot3="AosLoopback.Install"
  TraceMode="1"
  CacheSize="1000"
  APM="0"
  ATADetect="legacy"
  ExtMemSize="54"
~

Thanks,          ... Peter E.

[peasthope]

Previously ...
TraceMode="1"
produces nothing beyond "OBER";

Oops; forgot Init="".  Will try that Tuesday afternoon.

              ... Peter E.

[andre]

Mike,

That will not change a iota. The 386 AMD K6-2 based Pavilion must perform as you expect
but it comes to a grinding halt after having done practically nothing. To recap:
  1 - BIOS takes control
  2 - BIOS fetches the MBR on IDE0 and passes control to it
  3 - The MBR knows which partition to access (aka the PBR): one of them is marked "active"
      else a boot manager does the job
      IDE0#3 is what you expect to reach, but you have other partitions.
      The MBR transfers control to a PBR - In which one it is skidding? A landmines area ....
  4 - The PBR loads 3 further 512 bytes block (the boot loader) and must fetch 4 blocks of
      configuration data and the kernel. Now things already turn sour -> HALT.

      Please explore the different partitions with Partitions.ShowBlocks IDE0#x 4 4 ~
      (alternatively with the WMPartitions GUI)
      At offset:
      0   you should see 07h. If you see 0FFh it's all wrong
      1Ch you should see 08h. If you see 0FFh it's all wrong.
      Further on you should recognize the config strings. . .
      The rest of these 4 blocks contains 0FFh.

      An example:
      0000 07 00 00 00 1C 00 00 00 00 10 00 00 01 00 1B 00
      0010 00 10 00 00 00 00 00 00 DA 02 00 00 08 00 00 00
      0020 08 01 00 00 24 6F 6F 74 56 6F 6C 31 00 41 4F 53
                                B  o   o   t   V  o   l   1       A  O   S

Bootloader will only succeeed when encountering that kind of data structure.

This should clarify your matter. Nothing to trace yet, as said.
Let's hope, that puts you on track.
Andr

[peasthope]

Andr,

a> Please explore the different partitions with Partitions.ShowBlocks IDE0#x 4 4 ~
   ...
a> An example:
    0000 07 00 00 00 1C 00 00 00 00 10 00 00 01 00 1B 00
    0010 00 10 00 00 00 00 00 00 DA 02 00 00 08 00 00 00
    0020 08 01 00 00 24 6F 6F 74 56 6F 6C 31 00 41 4F 53
                  B  o  o  t  V  o  l  1     A  O  S

This is what Peter gets.
Partitions.ShowBlocks IDE0#3 4 4 ~
00000000 08 00 00 00 EC 00 00 00 42 6F 6F 74 56 6F 6C 31 ........BootVol1
00000010 00 41 4F 53 20 41 6F 73 46 53 20 49 44 45 30 23 .AOS AosFS IDE0#
00000020 30 33 00 41 6F 73 46 53 00 41 6F 73 44 69 73 6B 03.AosFS.AosDisk

"All wrong" for sure!

So I conclude that the configuration data structure
of recent Bluebottles is different from the structure
in Native Oberon, Alpha, 05.01.2003.  Which contradicts
this sentence from AosInst.Tool.
"Execute the following on Native Oberon 12.12.00 or later or any Bluebottle release."

OK; news to me.

a> (alternatively with the WMPartitions GUI)

Seeing that neither target machine can boot from USB
I really do need to burn a CD!  Or possibly a Zip
disk for the HP 6360.

Thanks,         ... Peter E.

[staubesv]

Dear Peter,

The excerpt of your block 4 on device IDE0#3 looks fine. There are two different kind of entries store in the configuration section. One (type 7) contains information about the boot image and the other (type 8 ) contains the actual config strings.
 
Andre has installed the boot image when he has formatted the partition and afterwards set the config strings. That's why his data starts with 07H (information about the boot image). At offset 01CH the config strings start.

Your data indicates that the boot image has not been written to the partition when formatting the disk. If you used the UpdateBootFile command after that, it is possible the the type 7 entry follows later in the part of data we don't see here.

I could reproduce the "OBER" output: It happens when trying to boot a AosFS formatted partition that has no boot image installed.

-> Most likely, you don't have a boot image on IDE0#3

If you post all the data of block 4-7, I can check this.

[peasthope]

At July 9 Sven wrote,
"Most likely, you don't have a boot image on IDE0#3"

Correct.  Just 1s after the config strings.

ss> "If you post all the data of block 4-7, I can check this."
 
Now visible here.
  http://carnot.yi.org/IDE0.3.4..7.txt

I'll try to find why the boot image wasn't written.

Thanks,             ... Peter E.

[staubesv]

The first line means that there is one entry of type 08H (configuration strings) and size ECH bytes.

Code:

08 00 00 00 EC 00 00 00 42 6F 6F 74 56 6F 6C 31

The entry of type 07H (boot image information) is missing. This indicates that there is no boot image on this AosFS volume and this will cause the boot process to hang after displaying the string "OBER".

You should use the command UpdateBootImage (Either in Oberon: Partitions.UpdateBootImage or A2: Partition Tool -> Bluebottle -> UpdateBootfile). Be careful when reading the output of the command. After that, the type 07H entry should be found in block 4 of IDE0#3 (most likely at offset 0ECH).

[peasthope]

Sven,

ss> Oberon: Partitions.UpdateBootImage[?] ... Be careful when reading the output of the command.

Partitions.UpdateBootBootFile works with no complaint.

After that, the type 07H entry should be found in block 4 of IDE0#3 (most likely at offset 0ECH).

The configuration still begins at block 4 and is
followed by 1s to the end of block 7.
The IDE.Bin code begins at block 8.  The first
two characters are 60 BE.  The 07 isn't obvious.

Pressing on ...

My volume/filesystem alias was wrong.
AosFS=AosDiskVolumes.New AosDiskFS.NewFS
needed correcting to
AosFS=DiskVolumes.New DiskFS.NewFS

Then I got a black screen.  TraceMode=1 gave these
messages.

DiskVolumes: IDE0#3 wrong file system version.
BootConsole: Mount failed on BootVol1

Is there any chance that the filesystem was revised
so that my old NO format is not compatible with Aos?

Thanks,          ... Peter E.

[staubesv]

If the string "OBERON loading" appeared, the config enty of type 07H has been located successfully (otherwise, the boot process would stop after displaying the string "OBER").

DiskVolumes: IDE0#3 wrong file system version.
BootConsole: Mount failed on BootVol1

When we added the prefix "Oberon." to all filenames of Oberon modules, some filenames became longer than 32 characters (=maximum filename length of AosFS). Therefore, AosFS has been changed to support filenames up to 128 characters (incl. 0X terminator).
The error message means that you tried to mount the old AosFS (32 character filenames) with the new AosFS driver (128 character filenames).
You can mount the "old" AosFS by using the file system alias OldAosFS, e.g. FSTools.Mount Prefix OldAosFS Device#Partition ~

If you installed a new boot image on a volume that has been formatted using the OldAosFS, things get a bit more complicated:

1. You need a boot image that supports the old AosFS. This can be generated using the command

  Linker.Link \PAOS: \.Obx AOS:IDE.Bin 0100000H 1000H Kernel Traps
  ATADisks OldDiskVolumes OldDiskFS BootConsole ~

2. Write the boot image (IDE.Bin) to your AosFS volume using Partitions.UpdateBootImage

3. Adapt the config string AosFS to AosFS=OldDiskVolumes.New OldDiskFS.New

[peasthope]

Andre & Sven,

afi> (alternatively with the WMPartitions GUI)

Right oh!  Boot CD, A2 2008-08-27, Revision 1560.
The display is visible here.

http://carnot.yi.org/StartA2CD.jpg

NO works with 1024 x 768 x 8 but Init=105 still
yields a black screen.  So "Invalid Init" might be
misleading.

ss> " ... a new boot image on a volume that has been formatted using the OldAosFS, ... "

Will return to that method later.  For now, I am
interested to make the CD work.

Thanks,             ... Peter E.

[peasthope]

More details.

The old Toshiba 4000CDS starts from the CD and the
GUI based installer appears to be OK!  Credits to
Toshiba for building standards conformant hardware!
The windows open bigger than the old screen.  I can
drag the window narrower but fail to see how to reach
the bottom of the window.  A2 is only for a BIG display? 
Ouch!

Putatively, I can take the hdd from any troublesome
target machine, connect it to the Toshiba via an ATA-USB
adapter and install A2 that way.  Will the target machine
boot?

Two machines yield the black screen and "Invalid Init"
when config string editing is invoked.  The IBM
NetVista 6578-RAU which burned the installer CD
and the HP Pavilion 6360.

Regards,                   ... Peter E.

[peasthope]

Experimentation With the CD Based Installer.

The HDD from a Toshiba 205CDS was connected
to the Toshiba 4000CDS via a USB-ATA adapter. 
Starting the 4000 from the installer CD yielded
the Kernel.log faintly visible here.
Desktops.OpenDoc 
  "http://carnot.yi.org/Installer.Kernel.log.jpg"
 
These lines appeared and of course the external
drive was not mounted.
UsbHubDriver: Warning: Connected high speed capable device to low-/full-speed controller.
UsbHubDriver: Root Hub USBHC0 (UHCI USB Host Controller): GetStatus request failed.

Is mass storage via a low-/full-speed controller
excluded?

Regards,                 ... Peter E.

[staubesv]

UsbHubDriver: Warning: Connected high speed capable device to low-/full-speed controller.
UsbHubDriver: Root Hub USBHC0 (UHCI USB Host Controller): GetStatus request failed.
Is mass storage via a low-/full-speed controller excluded?

No. The warning just indicates that you have attached a USB highspeed device to a controller that only supports low-/fullspeed. USB highspeed devices are required to also support fullspeed operation - it is just slower.
In the screen shot, there is a message ".... is now accessible.". This means, that the USB mass storage device driver has been successfully bound to the device an that you can "access" the device using its name (in your example, USB0).  It is not a guarantee that the device works without problems, however.

These lines appeared and of course the external drive was not mounted.

Have you tried to mount it? (accessible does not mean mounted)

Two machines yield the black screen and "Invalid Init" when config string editing is invoked.

I will ask Andre what the exact meaning of "invalid init" is. Most likely, the corresponding mode is simply not supported by your graphics hardware. To find out which modes are supported, you can use Andre`s Boot Manager.

Putatively, I can take the hdd from any troublesome
target machine, connect it to the Toshiba via an ATA-USB
adapter and install A2 that way.  Will the target machine
boot?

There is no guarantee, but it principle, it is possible to install A2 that way. I cannot remember what CPU your Toshiba devices have, but as far as I know, A2 requires at least a Pentium. It should be possible to make it work on 486 CPU (not 486SX) without much effort. To make it work on older CPUs would be quite a lot of effort because we don't have a floating point emulation for Intel IA-32 (does NO have one?).

[peasthope]

...  "access" the device using its name (in your example, USB0). ... Have you tried to mount it? (accessible does not mean mounted)

In this photo, http://carnot.yi.org/PartitionTool.jpg , the drive is at
the lower right of the screen and the partition tool shows Error: 0
for USB0(Toshiba 2659).   Yet it has several partitions including
Native Oberon which works in the 205CDS.  Shouldn't the partitions
appear as for the HDD?

Regards,              ... Peter E.

[staubesv]

Hmm... this looks like USB does not work correctly on your machine. Are there any USB-related error messages on the Kernel Log when you try to mount the device? Do other USB devices work on that computer, e.g. USB flash memory?