Choosing uboot/uImage for Guru Server

[crunch]

Hi plug people, I am having trouble finding the proper combination of uboot and kernel uImage for my Guruplug Server (NOT Plus). I don't care about bleeding edge as much as having it semi-functional. I flashed the uImage and rootfs from http://openplug.org/index.php/us/resources/downloads , but the unit hangs on kernel boot (initrd issue perhaps?). Here is some relevant info:

Code:

uboot version AFTER reflash (could be wrong version?):
U-Boot 2009.11-rc1-00602-gff67d57 (Feb 08 2010 - 03:53:16)  / Marvell-Plug2L

arch_number = 0x00000A29 (2601 decimal)

/proc/mtd BEFORE reformat:
dev:    size   erasesize  name
mtd0: 00100000 00020000 "u-boot"
mtd1: 00400000 00020000 "uImage"
mtd2: 1fb00000 00020000 "root"

So what I'm wondering is:
* Is my uboot recent enough to boot a custom kernel?
* Where would I find a working uImage and/or rootfs image for the unit?
* Are there any environment variables, entry points, or boot options I'll need to set?

Thanks.

[pingtoo]

Your u-boot version looks OK, there are few more enhancement after your version on the ARM code but does not make it any better if all you want just boot linux from NAND.

there are some parameters is necessary to boot linux on a plug, like console=ttyS0,xxxx and root=??? you must setup or else linux will not come up. there may be something else depend on linux distro you are using. All these parameters are set in u-boot environment.

if you post output of "printenv" we can start from there.

Good luck

[ppmt]

Also for some reason unexplained to me sometime the plug will just not pass booting the kernel until you unplugged from the wall and plug it back

this is my experience and I know other have had the same issue.

[crunch]

I used the default environment from the Guruplug installer, except ipaddr and serverip:

Code:

Marvell>> printenv
bootdelay=3
baudrate=115200
ethaddr=00:50:43:67:64:06
eth1addr=00:50:43:67:64:07
x_bootcmd_ethernet1=set ethact egiga1
x_bootcmd_ethernet=ping 192.168.2.1
bootargs=console=ttyS0,115200 ubi.mtd=2 root=ubi0:rootfs rootfstype=ubifs
ethact=egiga0
filesize=29064C
fileaddr=6400000
ipaddr=10.10.1.3
serverip=10.10.1.4
x_bootcmd_usb=usb start
x_bootcmd_kernel=nand read.e 0x6400000 0x100000 0x400000
x_bootargs_root=ubi.mtd=2 root=ubi0:rootfs rootfstype=ubifs
x_bootargs=console=ttyS0,115200
x_bootcmd=$(x_bootcmd_usb); $(x_bootcmd_kernel); setenv bootargs $(x_bootargs) $(x_bootargs_root) ;bootm 0x6400000;
bootcmd=run x_bootcmd
stdin=serial
stdout=serial
stderr=serial
Environment size: 664/131068 bytes

I was unsure if I should set arcnumber and/or mainlinelinux variables so I didn't. Let me know if anything looks bad there... thanks.

[pingtoo]

What is boot error? Your u-boot environment looks OK. The ${x_cmd_usb} at beginning of  x_bootcmd is unnecessary but it doesn't hurt either. have you try to use kernel from http://sheeva.with-linux.com/sheeva/?

Your boot command argument setup to use NAND as root file system is this your intention?

You don't need to setup mainLineLinux or archNumber, those are for special patched u-boot for sheevaplug. Your u-boot is mainline das u-boot it does use those variables.

[crunch]

There is no boot error, the kernel hangs after
Uncompressing Linux... done, booting the kernel.

I am intending to boot from NAND because I have no SD slot and I figure that USB booting will be more difficult than a stock setup (though I am open to trying it and I have a FAT16 USB drive ready).

I tried the 2.6.38 uImage from sheeva.with-linux.com (didn't know about that site, thanks), but I got the same result.

Here's how I am flashing my uImages, maybe I am doing something wrong here?

Code:

tftp 0x6400000 uImage
nand erase 0x100000 0x400000
nand write.e 0x6400000 0x100000 0x400000

[ppmt]

sorry to repeat myself but have you try to remove the power from the plug and plug it again?

[pingtoo]

This is very strange, I don't understand why you kernel boot does not display any messages, did you try to load kernel from tftp and boot?

After you try ppmt suggestion and if you already did tftp boot still not success, than we have to start from beginning, we need to get in sync. first we need to verify your u-boot version, then u-boot env. (in case typo somewhere we can see, please try cut&paste result) also bdinfo output.

[crunch]

Sorry I should've said awhile ago that I unplug/replug the device all the time to reboot. Others have said the reset command is not always a hard reset, so I've adopted this procedure. Here's my output from a tftpboot attempt:

Code:

Marvell>> tftpboot 0x2000000 sheeva-2.6.38-uImage
Using egiga0 device
TFTP from server 10.10.1.2; our IP address is 10.10.1.3
Filename 'sheeva-2.6.38-uImage'.
Load address: 0x2000000
Loading: #################################################################
         #################################################################
         #############################################################
done
Bytes transferred = 2798200 (2ab278 hex)

Marvell>> bootm 0x2000000
## Booting kernel from Legacy Image at 02000000 ...
   Image Name:   Linux-2.6.38
   Image Type:   ARM Linux Kernel Image (uncompressed)
   Data Size:    2798136 Bytes =  2.7 MB
   Load Address: 00008000
   Entry Point:  00008000
   Verifying Checksum ... OK
   Loading Kernel Image ... OK
OK

Starting kernel ...

Uncompressing Linux... done, booting the kernel.

I waited a good 10 minutes in case something needed to timeout. Also, I tried pinging the plug network interface constantly and got no reply packets. However, I did see activity on the interface lights - bursts of 3 or 4 flashes on both lights, then a delay. I turned off my software firewall.


I am definitely mystified by this problem as well. Thanks for all the help with this.

[ppmt]

as pingtoo says we need to start from scratch.

When this happened to me ( and the hard reset didn't help!) I ended reflashing the all plug with the uboot, uImage and kernel from the
globalscale page as they are known to work...

Then I updated the uboot (the one from flip-flip) and I installed the latest debian.

[pingtoo]

I see two options, one we continue debug. start from beginning. the other option replace your u-boot to more recent version and start from there. I am currently develop my dreamplug system. if you interest I can produce a u-boot and/or kernel too see if it will help. if you choose second option you will need to be patient, I can produce u-boot right away but the kernel is where I am develop at the moment, I am building my initramfs to support multiple root fs option.

[crunch]

Reflashing my box with stock Guruplug images is really what I'm trying to do - the OS images hosted with the installer don't work for me. What I suspect is that my plug is from an "in-between lot" which was built in a slightly different way from the normal gurus. For example, I noticed that when I tried to update my plug, the apt repo was on an unreachable local IP subnet, probably somewhere deep in globalscale-land. There were a few other things laying around that suggested the box was built hastily and/or in a one-off manner - backup files, extra wireless firmware, and some sloppy init scripts. This doesn't bother me particularly, but it does indicate that my box might've been built in a custom manner. I can't really think of anything that would cause this behavior however. I did notice that when I loaded Debian uImages (from their nightly kirkwood build repository), I got an "unsupported device" error message.

Perhaps I should nand erase the entire flash, reinstall uBoot, and build up the plug from nothing? I might be able to set up debugging via openocd/gdb, which could help.

pingtoo, I'd be happy to try out some of your uboot and/or kernel builds. I'm fine with learning more about these devices while I'm busy unbricking... Eventually I should probably set up USB booting so I don't wear out the NAND.

[pingtoo]

OK, let me try to make u-boot so we can test. I only have guruplug-plus. it make take me a few try before I can get openocd sync to test my u-boot.

[newitjames]

Reflashing my box with stock Guruplug images is really what I'm trying to do - the OS images hosted with the installer don't work for me.

Hi Crunch,

Here is a howto guide for restoring the Guruplug with the Globalscale images

http://www.newit.co.uk/forum/index.php/topic,1810.0.html

[pingtoo]

Reflashing my box with stock Guruplug images is really what I'm trying to do - the OS images hosted with the installer don't work for me. What I suspect is that my plug is from an "in-between lot" which was built in a slightly different way from the normal gurus. For example, I noticed that when I tried to update my plug, the apt repo was on an unreachable local IP subnet, probably somewhere deep in globalscale-land. There were a few other things laying around that suggested the box was built hastily and/or in a one-off manner - backup files, extra wireless firmware, and some sloppy init scripts. This doesn't bother me particularly, but it does indicate that my box might've been built in a custom manner. I can't really think of anything that would cause this behavior however. I did notice that when I loaded Debian uImages (from their nightly kirkwood build repository), I got an "unsupported device" error message.

Perhaps I should nand erase the entire flash, reinstall uBoot, and build up the plug from nothing? I might be able to set up debugging via openocd/gdb, which could help.

pingtoo, I'd be happy to try out some of your uboot and/or kernel builds. I'm fine with learning more about these devices while I'm busy unbricking... Eventually I should probably set up USB booting so I don't wear out the NAND.

Are you comfortable loading u-boot from openocd? I am having problem setup u-boot from tftp for guruplug+.