Sheevaplug installer - version 1.0

[rooster]

NOTE for ionics Sheeva Plug users:

The current installer does not support this new plug since the ftdi vendor/product differs from the GlobalScale plug, to make the installer work from Linux host you need to:
1) change the runme.php line that modprobes the ftdi device driver to:

exec("modprobe ftdi_sio vendor=0x1c0c product=0x0102", $out, $rc);

2) change the uboot/openocd/config/interface/sheevaplug.cfg to look like this:

interface ft2232
ft2232_layout sheevaplug
ft2232_vid_pid 0x1c0c 0x0102
#ft2232_device_desc "SheevaPlug JTAGKey FT2232D B"
jtag_khz 2000

In any case users should apply

modprobe ftdi_sio vendor=0x1c0c product=0x0102

to use the console.

A new installer version that supports both plugs will be uploaded...

[mail4asim]

Whenever I follow your instructions, I am able to run  runme.exe nand and it goes through a bunch of stuff. But in the end I get stuck at

****   Burning uboot and environment variables ... This will take few minutes ...
Open On-Chip Debugger 0.2.0 (2009-07-26-14:56) Release
$URL: http://svn.berlios.de/svnroot/repos/openocd/tags/openocd-0.2.0/src/openocd.c $
For bug reports, read http://svn.berlios.de/svnroot/repos/openocd/trunk/BUGS
2000 kHz
jtag_nsrst_delay: 200
jtag_ntrst_delay: 200
dcc downloads are enabled



At this point  I am unable to do anything. I have tried different USB drives. I have let it run for over an hour, but the computer just locks up and I have to reboot.

Any idea to what I am doing wrong ?

-------------------

I finally got my plug working. I had poured over these forums and the plug wiki for hours.................... This is for use with the version 1 installer.

For instillation to nand with Windows:

1: Install the proper drivers. You will find all the information you need to do so here: http://plugcomputer.org/plugwiki/index.php/Serial_terminal_program
  1a: The drivers can be found on your CD.

2: Install libusb-win32-filter-bin-0.1.12.1.
  2a: http://sourceforge.net/projects/libusb-win32/files/libusb-win32-releases/0.1.12.1/libusb-win32-filter-bin-0.1.12.1.exe/download

3: Update u-boot. You will find everything you need here: http://www.cyrius.com/debian/kirkwood/sheevaplug/uboot-upgrade.html
  3a: TFTP server I used: http://tftpd32.jounin.net/download/Tftpd32-3.35-setup.exe

4: Prepare the uboot-nand-custom.txt.
  4a: Replace the mac address in uboot/uboot-env/uboot-nand-custom.txt with the mac address on the back of your plug.

5: Copy all files in the the installer directory to a usb thumb drive(fat32 format).

6: Insert usb drive into your sheevaplug, connect console cable, connect power cable.

8: Click Start -> Run -> cmd -> <path to files> -> runme.exe nand

9: Wait for it to say "wrote file uboot.bin to nand". Remove console cable.

10: You can now log onto it through putty with user: root, pass: nosoup4u.


This is what worked for me. Hope it helps someone.

[kirkwood]

Just want to share my experience with previously mentioned error message.

Code:

Error: unable to open ftdi device: device not found
Runtime error, file "command.c", line 469:
     ****       openocd FAILED
 ****   Is the mini USB cable connected?
 ****   Try powering down, then replugging the Sheevaplug

I did receive the same on my desktop although I got a serial terminal working flawlessly on different usb ports. I tested all USB ports. No success.
Then I tried the same on my laptop, 2 USB ports. On one port the installer works fail-save on the other port I get above mentioned error message. Serial Terminal can be run through both ports. Both PCs come with XP installed.

I did

- install the WindowsTeraTermUSBDriver.zip (actually 3 different drivers) from the Installer CD

- the libusb-win32-filter-bin-0.1.12.1.exe from sourceforge.net

- edit the uboot-nand-custom.txt within the installer directory

- copy all files as explained in the readme

- plug the USB drive into the plug

- connect the miniusb to the laptop

-  run runme.exe nand

So if you get this message you might want to test all available usb ports and/or try a different machine.

At least that worked for me.

[Tobias.Kornmayer]

Hi Pros,
two days ago I started messing around with the sheeva. I compiled the oreon kernel and installed it. The first problem was that the kernel didnt recognize the rootfs.
Okay, so I had to flash again via ftdi. Well up to my 10th attempt uboot survived, now its gone also. heres what I got with the installer (even with working uboot):

 ****   Burning uboot and environment variables ... This will take few minutes ...
Open On-Chip Debugger 0.3.0 (2009-11-05-22:57)
$URL$
For bug reports, read
        http://openocd.berlios.de/doc/doxygen/bugs.html
2000 kHz
trst_and_srst separate srst_gates_jtag trst_push_pull srst_open_drain
jtag_nsrst_delay: 200
jtag_ntrst_delay: 200
dcc downloads are enabled
Warn : use 'feroceon.cpu' as target identifier, not '0'
Info : clock speed 2000 kHz
Info : JTAG tap: feroceon.cpu tap/device found: 0x20a023d3 (mfg: 0x1e9, part: 0x0a02, ver: 0x2)
Info : Embedded ICE version 0
target state: halted
target halted in ARM state due to debug-request, current mode: Supervisor
cpsr: 0x000000d3 pc: 0xffff0000
MMU: disabled, D-Cache: disabled, I-Cache: disabled
0 0 1 0: 00052078
NAND flash device 'NAND 512MiB 3,3V 8-bit' found
erased blocks 5 to 7 on NAND flash device #0 'NAND 512MiB 3,3V 8-bit'
Error: No working memory available. Specify -work-area-virt to target.
Error: No working memory available. Specify -work-area-virt to target.
Error: No working memory available. Specify -work-area-virt to target.
Error: No working memory available. Specify -work-area-virt to target.
Error: No working memory available. Specify -work-area-virt to target.

...

and so on. has anybody an idea whats going wrong? is any information missing?
I tried really everything, read each readme and compiled everything to run on 64bit.

What could I do? I think I am getting a few lines further than mail4asim, maybe
its a similar problem.

[badcam]

I've tried this:

If you are using the installer do the following:
change the following sheevaplug-installer-v1.0/uboot/uboot-env/uboot-nand-custom.txt
1)
from: mtdpartitions mtdparts=orion_nand:0x400000@0x100000(uImage),0x1fb00000@0x500000(rootfs)
to:    mtdpartitions mtdparts=orion_nand:0x400000@0x200000(uImage),0x1fb00000@0x600000(rootfs)
2)
from: recover3 run recover4; nand erase clean 0x00100000 0x00400000; nand write.e 0x00800000 0x00100000 0x00400000
to:    recover3 run recover4; nand erase clean 0x00200000 0x00400000; nand write.e 0x00800000 0x00200000 0x00400000
3)
from: real_bootcmd setenv bootargs $(bootargs_console) $(mtdpartitions) $(bootargs_root); nand read.e 0x00800000 0x00100000 0x00400000; bootm 0x00800000
to:    real_bootcmd setenv bootargs $(bootargs_console) $(mtdpartitions) $(bootargs_root); nand read.e 0x00800000 0x00200000 0x00400000; bootm 0x00800000

These changes should skip the first mega of nand flash nd burn the uImage in offset 2MB and also burn the correct uboot parameters for uboot to boot from.

But that didn't make a difference.

Then I tried this:

If you are using the installer do the following:
change the following sheevaplug-installer-v1.0/uboot/uboot-env/uboot-nand-custom.txt
1)
from: mtdpartitions mtdparts=orion_nand:0x400000@0x100000(uImage),0x1fb00000@0x500000(rootfs)
to:    mtdpartitions mtdparts=orion_nand:0x400000@0x200000(uImage),0x1fb00000@0x600000(rootfs)
2)
from: recover3 run recover4; nand erase clean 0x00100000 0x00400000; nand write.e 0x00800000 0x00100000 0x00400000
to:    recover3 run recover4; nand erase clean 0x00200000 0x00400000; nand write.e 0x00800000 0x00200000 0x00400000
3)
from: real_bootcmd setenv bootargs $(bootargs_console) $(mtdpartitions) $(bootargs_root); nand read.e 0x00800000 0x00100000 0x00400000; bootm 0x00800000
to:    real_bootcmd setenv bootargs $(bootargs_console) $(mtdpartitions) $(bootargs_root); nand read.e 0x00800000 0x00200000 0x00400000; bootm 0x00800000

These changes should skip the first mega of nand flash nd burn the uImage in offset 2MB and also burn the correct uboot parameters for uboot to boot from.

Didn't actually seem to work, so i changed them back, reflashed, and it works

Thanks rooster!

...and that didn't make any difference.

So now, I still have a useless bricked Sheevaplug. Is there anything else I can try?

[rooster]

I tried really everything, read each readme and compiled everything to run on 64bit

Do you have a 32bit PC? I would try it out.
I never seen this kinda error from the installer, you can aloso try to build openocd version 2.0 (which was used with the installer).

[ckaos]

Hello everyone! i recieved my SheevaPlug a few days ago, and i just spent a few hours trying to get the installer v1.0 to work.
After a lot of experimentation and fruitless attempts i managed to go through the process but now i am getting the following error at boot:

"invalid partition type "" (expect "U-boot")
Booting image at 00800000
Bad magic number

I'm using the mmc option (8gig class 6 Patriot SD Card), any thoughts on how to fix this ?

Just an information that might be usefull for those people running into "bad magic number" problems when the sheevaplug is trying to read the data from the usb DOK : i kept getting those errors until i reformated the usb DOK on my host machine using the following sudo mkfs.vfat -F 16 /dev/sdb1 -n NameOfUsbDok and copied the files to it again

Thanks in advance for any way to solve the boot problem, i am pretty much stuck right now:)

[badcam]

I've tried this:

If you are using the installer do the following:
change the following sheevaplug-installer-v1.0/uboot/uboot-env/uboot-nand-custom.txt
1)
from: mtdpartitions mtdparts=orion_nand:0x400000@0x100000(uImage),0x1fb00000@0x500000(rootfs)
to:    mtdpartitions mtdparts=orion_nand:0x400000@0x200000(uImage),0x1fb00000@0x600000(rootfs)
2)
from: recover3 run recover4; nand erase clean 0x00100000 0x00400000; nand write.e 0x00800000 0x00100000 0x00400000
to:    recover3 run recover4; nand erase clean 0x00200000 0x00400000; nand write.e 0x00800000 0x00200000 0x00400000
3)
from: real_bootcmd setenv bootargs $(bootargs_console) $(mtdpartitions) $(bootargs_root); nand read.e 0x00800000 0x00100000 0x00400000; bootm 0x00800000
to:    real_bootcmd setenv bootargs $(bootargs_console) $(mtdpartitions) $(bootargs_root); nand read.e 0x00800000 0x00200000 0x00400000; bootm 0x00800000

These changes should skip the first mega of nand flash nd burn the uImage in offset 2MB and also burn the correct uboot parameters for uboot to boot from.

But that didn't make a difference.

Then I tried this:

If you are using the installer do the following:
change the following sheevaplug-installer-v1.0/uboot/uboot-env/uboot-nand-custom.txt
1)
from: mtdpartitions mtdparts=orion_nand:0x400000@0x100000(uImage),0x1fb00000@0x500000(rootfs)
to:    mtdpartitions mtdparts=orion_nand:0x400000@0x200000(uImage),0x1fb00000@0x600000(rootfs)
2)
from: recover3 run recover4; nand erase clean 0x00100000 0x00400000; nand write.e 0x00800000 0x00100000 0x00400000
to:    recover3 run recover4; nand erase clean 0x00200000 0x00400000; nand write.e 0x00800000 0x00200000 0x00400000
3)
from: real_bootcmd setenv bootargs $(bootargs_console) $(mtdpartitions) $(bootargs_root); nand read.e 0x00800000 0x00100000 0x00400000; bootm 0x00800000
to:    real_bootcmd setenv bootargs $(bootargs_console) $(mtdpartitions) $(bootargs_root); nand read.e 0x00800000 0x00200000 0x00400000; bootm 0x00800000

These changes should skip the first mega of nand flash nd burn the uImage in offset 2MB and also burn the correct uboot parameters for uboot to boot from.

Didn't actually seem to work, so i changed them back, reflashed, and it works

Thanks rooster!

...and that didn't make any difference.

So now, I still have a useless bricked Sheevaplug. Is there anything else I can try?

Rooster:

Please. I'm desperate for a solution here. Can I try stepping up the data ranges (Sorry, I don't know the terminology)? In your example you suggested changing 1) for instance:

from: mtdpartitions mtdparts=orion_nand:0x400000@0x100000(uImage),0x1fb00000@0x500000(rootfs)
to:    mtdpartitions mtdparts=orion_nand:0x400000@0x200000(uImage),0x1fb00000@0x600000(rootfs)

Now, I've tried changing this to:

to:    mtdpartitions mtdparts=orion_nand:0x400000@0x300000(uImage),0x1fb00000@0x600000(rootfs)

and similar for 2) & 3).

Can I step this up until I find a range where the Nand will flash and no longer give me the CRC errors?

I'm at a loss here, so if I can't resolve this, then I'm going to have to send the thing back. Please. If you could offer some suggestions.

[rooster]

Hi badcam,

I just looked at your post and noticed that uImage was not found on the DOK:

reading uImage

** Unable to read "uImage" from usb 0:1 **

So your first problem is first to get the uboot to read the uImage from the USB DOK and flash it to the NAND first, can you try other DOK if the one you used fails.

Format it to FAT32 or FAT16 and make sure the uImage from the installer is in the root directory

[Tobias.Kornmayer]

Hi rooster,

any idea what "Error: No working memory available. Specify -work-area-virt to target.
" could mean? I even messed up uBoot after several tries of reflashing. May it be possible that the nand was damaged? How many write cycles are possible? From my arduino i know that it shouldnt be that dangerous to rewrite it hundreds of times, but does this apply to the Sheeva too?

thanks,
Tobi

[rooster]

Hi Tobi,

I have yet used openOcd v0.3.0 and hae never seen such an error, to my understanding it is an error comming from the host PC and you might need to add the -work-area-virt openOcd configure options.

Try editing the sheevaplug-installer-v1.0/uboot/openocd/config/board/sheevaplug.cfg file:
from

$_TARGETNAME configure -work-area-phys 0x10000000 -work-area-size 65536 -work-area-backup 0

to 

$_TARGETNAME configure -work-area-virt 0 -work-area-phys 0x10000000 -work-area-size 65536 -work-area-backup 0

[epapa]

To the few people that were looking into installing from a Mac OS X host: any luck ?
what are the linux modules needed by the SheevaPlug Installer 1.0 ?

many thanks,
eli

[rooster]

I confirmed that version 2.6.30.9 is the last working kernel version I managed to use the installer with, all other (2.6.31-2.6.31.n) failed with the same error.

I did notice some changes in the kernel config (2.6.30.9 vs 2.6.31) but nothing stood out that I could say that it is the root cause for the RAMDISK failure.

This one needs more debugging...

I got the installer to burn latest kernel/modules to my plug, you need to modify the default sheevaplug-installer-v1.0\uboot\uboot-env\uboot-nand-custom.txt (for nand installation, the same can be done for mmc install)
change the following line:
from

recover2 run recover3; setenv bootcmd $(real_bootcmd); saveenv; setenv bootargs $(bootargs_console) $(mtdpartitions) root=/dev/ram0 rw ramdisk=0x01100000,8M install_type=nand; bootm 0x00800000 0x01100000

to

recover2 run recover3; setenv bootcmd $(real_bootcmd); saveenv; setenv bootargs $(bootargs_console) $(mtdpartitions) root=/dev/ram0 rw ramdisk=0x01100000,8M install_type=nand; fatload usb 0 0x00800000 uImage; bootm 0x00800000 0x01100000

and another line
from

recover4 usb start; fatload usb 0 0x00800000 uImage; fatload usb 0 0x01100000 initrd

to

recover4 usb start; fatload usb 0 0x00800000 sheeva-2.6.31.6-uImage; fatload usb 0 0x01100000 initrd

Now copy the new kernel image to the DOK (I used sheeva-2.6.31.6-uImage), so now you should have two uImages on the DOK, one used by the ramdisk (uImage) and the other which is burned to the nand/mmc.

Copy also the all kernel modules to the DOK and make sure the tarball name is modules.tar.gz

From here on, the same installation procedure: runme.php/exe nand/mmc

[alexschomb]

hey there,

thanks for your great work. i received my own sheevaplug on thursday and till then i really spend much time in getting it to work properly. as i found out about your installer and the benefits of it (like solving the common problems) i decided to give it a try. sadly this wasn't that easy as posted, as most of the terms were very new to me and i didn't even understand those tutorials, so i googled like everything and after some hours of work i found out about serial consoles, tftp (doesn't work for me ... sheeva always get octet timeout 5, no solution yet!), and all of this stuff. now i am able to run the installer properly and receive no errors anymore in the host environment, but it still doesn't work out in the serial console. i tried about 10 diffrent usb-sticks (mine + friends) which all, except my first one, don't even get recognized as usb storage device (running "usb start" in marvell prompt). sadly the only recognized one all time gets errors at install, as it is a very buggy old usb stick, that doesn't run safely on my desktop too. so i'm asking myself if there is a way installing via your installer directly from mmc instead of usb. am i right thinking that i only need to change those lines concerning "fatload mmc *" to "fatload usb *"?
what about the mmc installation? can i make up another partition for the mmc serving the installer files and install ubuntu on mmc at the same time?

secondly i'm interested in using the installer combined with the actual kernel, uImage, ... is it as easy as just replacing the files out of the archive with the new ones before copying it to the usb (in my case mmc)? i recognized that your last post is about excactly this issue, but it is not quite clear to my whether this is only for those guys having a running environment on their sheevaplug or not. also i'd like to know if i need to replace the initrd with a new one. honestly i never found out what this file is for. also i don't know how to use the patches of http://sheeva.with-linux.com/sheeva

i appreciate your help, if you don't understand some of my sentences feel free to ask me anytime, as english is not my primary language.

[alexschomb]

Okay, after some more recherche and trying i found about u-boot problems. so i tried to compile my own by patching the originial 1.1.4 with the given patch on this website, but i weren't able to compile as i never found out how to configure MAKEFILE to direct to my local gcc. there has to be a very big .bin - file from sourcergy-g++ that has to do something with this too, but as i stubled upon the page of cyrus (http://www.cyrius.com/debian/kirkwood/sheevaplug/uboot-upgrade.html) i quit my work on uboot and used his precompiled uboot for sheevaplug.

all i had to do is replace the given uboot out of the sheevaplug installer package uboot directory with the new uboot.bin for further installations. as all of my usb sticks won't be recognized by the old uboot i need to upgrade uboot at first on my own. the procedure was similar to the usb-way on cyrus tutorial, but for all of you interested that don't want to mess with all those uboot commands (expecting you formatted your mmc/sd-card with fat32 and put your new uboot.bin on it!):

Code:

mmcinit
fatload mmc 0:1 0x0800000 uboot.bin
nand erase 0x0 0xa0000
nand write 0x0800000 0x0 0xa0000
reset

(you have to run this command in your serial console connected to the right serial port for your usb cable. you can find it out by installing "usbutils" and running "lsusb". in most cases the port should be /dev/ttyUSB0 or /dev/ttyUSB1. i prefer using gtkterm as it tells you what ports are active in its port-configuration)

so for now i'm able to install to the built-in nand flashdrive of sheevaplug. sadly mmc installation quits while trying to format the mmc, here a screenshot:

i already tried adding another "mmc init" to the /uboot/uboot-env/uboot-mmc-custom.txt, but this didn't help out. interpreting the output on my own i think the installer found the mmc card and it was able to format for the first partition (ext2) but can't format the second partition (ext3). i also mounted the mmc to my desktop and found out that the first partition is correctly setup, but the second one is corrupted. any help on this? i can confirm that the sdhc card i'm using has a partition table as i can create and mount the two partitions (ext2 & ext3) using my desktop.