Author Topic: Results of two tests  (Read 23414 times)

0 Members and 1 Guest are viewing this topic.

Offline A543

  • Senior Member
  • Veteran.
  • *****
  • Posts: 214
Results of two tests
« on: March 14, 2005, 02:53:45 am »
Here are the results of two tests I've run so far.

Test #1:
In order to determine if the magic key testing routines are part of the code for the first-run processor (for lack of a better term) I added an empty file named "forceupdate" in the root of the PHTSYS partition.  I did this on my non-Phatnoise Toshiba drive - the same model as my Kenwood DMS drive, but with a different firmware revision I've since found out.  Anyway, after a raw sector by sector copy of my original DMS to this non DMS drive the Keg wouldn't recognize the drive. So by adding the forceupdate file, and seeing the drive get rejected in exactly the same manner as without this file, it appears the key checking code is indeed part of the first-run processors routines, and runs before the code that updates the firmware.
Interestingly, the garbage (key) stored between the partition table and the start of the first partition is about 130k in size.  I'm wondering if this isn't actually part of the first-run processors code!
What I would like to do now is to rename a file or two on the DMS just to be absolutely certain that the Keg doesn't load anything off of the drive, after mounting it, in order to check the key.  What I need is for someone to tell me the order that the files are loaded.  Is the file linux loaded first of all, and then royallin.ux?  What are these two versions anyway?  I assume one of them has to be loaded before rc.sh is called, right? Please let me know.

Test #2:
I wanted to see if the partition sizes can be changed.  Having a small 10GB DMS and seeing 230+MB of space wasted on the PHTSYS partition prompted this.  In a nutshell, it worked!
Here are the gory details:
I first tried to reduce the partition down to about 31MB and use FAT16.  Barf.  The keg only recognizes FAT32. Of interest is the fact that when I plugged the DMS into the Keg with the FAT16 PHTSYS partition, it passed the key check and started a firmware update! (The drive was initialized due to the reformat, hence the forceupdate file on the DMS. )  The firmware update didn't complete in over 30 minutes so I knew it wasn't working, it couldn't read the format of the drive. I had no choice but to pull the DMS out while the lights were blinking, and they didn't stop blinking!  I had to unplug the Keg to HU cable to kill the lights.
I then reformated the partition to the smallest size (around 33MB) that FAT32 would allow and reformated.  Unfortunately, I couldn't force the 1k cluster size I wanted, it defaulted to 1 sector per cluster.  I was hoping to reduce the FAT size during this experiment as well. I figured it couldn't hurt.  By the way, the original cluster size was 8k, so there was a bit of slack waste, but not anymore, and the FAT is roughly the same size it was before. I wonder if the Keg caches the FAT. I'm guessing not so a smaller FAT would mean faster file access.  I think the format utility that comes with XP allows for forcing the cluster size, so I might play with that next for some more fun.  Anyway, after reformatting and reinitializing, the DMS worked fine!
So I gained around 230MB, which is roughly 4 or 5 more CDs.  And I also feel this proves that there is no partition table data in the key as well.  AND as a side effect of this I corrected Phatnoises somewhat wacky partition table.  They use a Linux utility to do the partitioning and to the best of my knowledge it doesn't break any rules, it just does things in an unusual way.  It starts the second partition in the very next sector after the extended partition tables sector.  Usually a head (side) is skipped.  This is one of the reasons that programs like Partition Magic won't operate on the partition tables of the DMS.  Nortons Disk Doctor also complains about problems with the partitions.  At least mine did.  Phatnoises partitioning program also incorrectly sets up the extended partition as just that, an extended partition, not an extendedx partition, which is really what is should be, seeing as how the partition crosses the 1024 track boundary.  Maybe this doesn't matter with Windows, but it does with DOS.  Anyway, my partition table are cleaned up and Norton and Partition Magic and DOS and Windows and the Keg are all happy with it this way.
I do want to say that I wouldn't recommend that anyone attempt to repartition their DMS unless they are absolutely sure they know what they are doing. Most partition utitlities WILL erase the magic key.  If there is enough interest I will post a guide on how one can change their partition sizes, but even then it will require some understanding of partition tables and also the tools to manipulate them.

So here are some questions, maybe they might be better in their own thread. I'll leave it up to the answerer to start a new one.
What is the boot order of the DMS files?
Are there two different versions of Linux on the DMS? Royallin.ux, and Linux?
Does anyone know what the first-run processor is?

Offline AndyMan

  • Getting the hang of things.
  • **
  • Posts: 75
Re: Results of two tests
« Reply #1 on: March 14, 2005, 04:23:35 am »
I also did a sector copy using "Acronis True Image", good dms and "exact" non dms 10gb drive

No joy...

I think that royallin.ux is the master self-extracting linux os that loads into memory... I'm looking into it


Offline slowwagon

  • A few posts under my belt.
  • *
  • Posts: 23
Re: Results of two tests
« Reply #2 on: March 14, 2005, 06:36:14 am »
Quote
If there is enough interest I will post a guide on how one can change their partition sizes, but even then it will require some understanding of partition tables and also the tools to manipulate them.


I am interested if you get a chance to post it.

Thanks,
Sean

Offline A543

  • Senior Member
  • Veteran.
  • *****
  • Posts: 214
Re: Results of two tests
« Reply #3 on: March 14, 2005, 01:16:58 pm »
Quote
I also did a sector copy using "Acronis True Image", good dms and "exact" non dms 10gb drive


Can Acronis copy sectors that fall outside of partitions?  For instance Norton Ghost only works with partitions and cannot copy the magic key, but DiskEdit will copy every and all sectors making complete bit for bit copies.
And are the firmware versions of your two 10GB drives the same?  I'm just curious to know if you have a more exact match than I do, my model is the same but with a different firmware revision.

Offline AndyMan

  • Getting the hang of things.
  • **
  • Posts: 75
Re: Results of two tests
« Reply #4 on: March 14, 2005, 02:00:54 pm »
How do I find the firmware revision, thru code, or physically written on the disc?

Offline Firefox

  • Needs to get outside.
  • ***
  • Posts: 124
  • Kenwood 710 Keg 80GB ;-)
Re: Results of two tests
« Reply #5 on: March 14, 2005, 05:35:01 pm »
Quote
How do I find the firmware revision, thru code, or physically written on the disc?


You can try this - it's not ideal as it is needs to be installed. Also not sure if it picks up all drives or not as I can't test vs. a DMS at work.

http://www.browsedatabase.com/hdd101.zip

Sample output...

HDD Firmware Serial Number & Other Details:
===========================================================

Controller = 0, Master Drive
HDD Firmware Serial Number:     CLP429F4HGEW3A
HDD Model Number:               IC25N040ATCS05-0
HDD Controller Revision Number: CS4OA61A
===========================================================

*** Please note this is NOT a serial number for a DMS !!!! ***

If anyone has a better solution please share...

PS I'm pretty confident PN must be using this HDD info and then encrypt it using a Private Key and save it to the DMS when they "bless" the drive before selling it.
The Phatbox then decrypts the encrypted string it reads from the DMS using the corresponding Public Key and compares it to the actual value(s) it gets using hdparm.
If they match, you're in, otherwise flashing lights ad infinitum......

That also means without the Private Key we are scr*wed unless the whole check can be bypassed somehow.
« Last Edit: March 14, 2005, 06:27:11 pm by Firefox »

Offline A543

  • Senior Member
  • Veteran.
  • *****
  • Posts: 214
Re: Results of two tests
« Reply #6 on: March 14, 2005, 09:29:11 pm »
Quote
How do I find the firmware revision, thru code, or physically written on the disc?


I used an older version of IBMs drive fitness test.

http://www.tacktech.com/display.cfm?ttid=287



Quote
The Phatbox then decrypts the encrypted string it reads from the DMS using the corresponding Public Key and compares it to the actual value(s) it gets using hdparm.  

I'm guessing that the key is checked using only the first-boot processors code, or that the first boot-processors code is the key itself, or part of it.  I need to get a handle on the order of the boot files so I can start renaming some of them to see if it interferes with the key check.  I suspect it won't.
Remember, it seems that the first-boot processor is responsible for performing the firmware updates and the key check appears to occur before the firmware update routines.  It's still just speculation though.
Maybe that's why there appears to be two kernels on the DMS, one for the first-boot processor and one for the main processor.  We need to find out.
We also could use a kindly Red box owner who wouldn't mind doing a few tests such as renaming some files to see if they are used on the Red box.

Offline Firefox

  • Needs to get outside.
  • ***
  • Posts: 124
  • Kenwood 710 Keg 80GB ;-)
Re: Results of two tests
« Reply #7 on: March 14, 2005, 11:01:53 pm »
Quote
...Maybe that's why there appears to be two kernels on the DMS, one for the first-boot processor and one for the main processor.  We need to find out.


Maybe. Maybe not.

Royallin.ux (Royal Linux) is for embedded devices...
http://wwww.embeddedsoftwaregroup.com/royallinux.php

Probably for the main ARM processor...
http://www.cirrus.com/en/press/releases/P161.html

But this link implies PN did their own port...
http://linuxdevices.com/articles/AT6469667521.html

(Those links are a bit old and probably refer to the Red Phatbox design).

All a bit confusing....

What we need is a disgruntled PN employee to leak us a copy of the signing utility ;)
Any takes? They're bound to be lurking here just for laughs......  ;D

According to press kit they are all very happy though  :( (but some good PN history in here)...
http://www.phatnoise.com/downloads/PhatNoise_Press_Kit_10.2004.pdf

Offline AndyMan

  • Getting the hang of things.
  • **
  • Posts: 75
Re: Results of two tests
« Reply #8 on: March 15, 2005, 03:06:30 am »
Has anyone tries aprocaching

Vincent Busam... he left Phatnoise in February



« Last Edit: March 15, 2005, 03:06:46 am by AndyMan »

Offline MarkH

  • Newbie
  • Posts: 12
Re: Results of two tests
« Reply #9 on: March 15, 2005, 10:07:31 am »
from vinces website:
------
December 2000 - Februrary 2005 PhatNoise, Inc.
Los Angeles, CA
Advanced Products Development
Helped develop car-based MP3 player, controlled by CD interface to existing head units.
Led design and implementation of client/server networking product.
Designed and implemented other leading-edge features for product, including Palm interface, TCP/IP networking, video playback, and game emulation.
Resposible for all of the "user level" programs running on the PhatBox in the car to manage the system and interface with kernel drivers, audio decoders, song database, and playback control mechanisms. Most projects involved embedded Linux development in C.
------

can't find the video playback or game emulation on my music keg ?  ;D

Offline Firefox

  • Needs to get outside.
  • ***
  • Posts: 124
  • Kenwood 710 Keg 80GB ;-)
Re: Results of two tests
« Reply #10 on: March 15, 2005, 10:44:29 am »
also from sixpak...

http://boxster.sixpak.org/mp3/

i'd never explored that part of the site before - very cool story!

Offline sbingner

  • Administrator
  • Veteran.
  • *****
  • Posts: 1301
Re: Results of two tests
« Reply #11 on: March 18, 2005, 06:14:30 pm »
As far as I know, you can make the two partitions any size you like so long as you don't overwrite the first two sectors in the process...

Offline A543

  • Senior Member
  • Veteran.
  • *****
  • Posts: 214
Re: Results of two tests
« Reply #12 on: March 18, 2005, 06:29:24 pm »
The first partition is limited to being no smaller than 33MB, this is the smallest partition that FAT32 recognizes.  There was some speculation that the partition table or table information was involved in the key, but by changing my partitions I basically negated that. So other than the above restriction, it does appear that any sized partitions should work.
Also the first sector has to be changed, that's where the partition table is.  The second sector is empty and is probably safe to overwrite.  It's sectors 1792-2047 that you should avoid overwriting.

Offline JohanDC

  • Newbie
  • Posts: 1
Re: Results of two tests
« Reply #13 on: March 19, 2005, 12:57:49 pm »
What about the Master Boot Record?

Maybe there is something special in there, maybe the PB will mount the DMS and the MBR will start the entire boot process?

Anyone care to take a look?

Offline A543

  • Senior Member
  • Veteran.
  • *****
  • Posts: 214
Re: Results of two tests
« Reply #14 on: March 19, 2005, 05:33:27 pm »
There is no MBR on the DMS.  I'm guessing booting handled from ROM or firmware.

Offline para

  • Senior Member
  • Veteran.
  • *****
  • Posts: 181
Re: Results of two tests
« Reply #15 on: March 19, 2005, 08:24:05 pm »
I'm interested in one thing which I doubt anyone will try:

If one backups his DMS completely by a raw sector copy (maybe using dd), would it possible to restore that image to the SAME drive and have it running again? This way one could at least save some money if he ruins the DMS layout by accident :-/

Para

Offline sbingner

  • Administrator
  • Veteran.
  • *****
  • Posts: 1301
Re: Results of two tests
« Reply #16 on: March 22, 2005, 10:13:36 am »
I don't see how it would be possible for that NOT to work... so yes that should be safe if you're going to try something strange with your DMS... part of why I have a full image of mine

Offline AndyMan

  • Getting the hang of things.
  • **
  • Posts: 75
Re: Results of two tests
« Reply #17 on: March 22, 2005, 02:34:11 pm »
I've wiped and reformatted my 10gb dms without any issues, the only thing  you have to do is file copy (BUT Include the invisible files and directories)

Offline para

  • Senior Member
  • Veteran.
  • *****
  • Posts: 181
Re: Results of two tests
« Reply #18 on: March 22, 2005, 05:09:40 pm »
Huh? And what about the "blessing" sectors holding the key...?

Offline sbingner

  • Administrator
  • Veteran.
  • *****
  • Posts: 1301
Re: Results of two tests
« Reply #19 on: March 22, 2005, 08:00:22 pm »
Quote
I've wiped and reformatted my 10gb dms without any issues, the only thing  you have to do is file copy (BUT Include the invisible files and directories)



I'm guessing you mean just reformat, not repartition?  The repartitioning is where problems could arise....