Talk:Running Debian on Freecom FSG-3

From lxadm | Linux administration tips, tutorials, HOWTOs and articles
Jump to: navigation, search

If you have any questions, please subscribe to the mailing list on

Problems found by tsr[edit]

Ok, maybe the problems found here, and my lack of skills to fix them shows that I shouldn't even try, I don't know. Either way I am going too and I will appriciate any help I can get.


Running an up-to-date Ubuntu 7.04 I get the following error:

root#: tar xjf fsg-3-deploy-stick.tar.bz2 
tar: media/disk/dev/pts: orimligt gammal tidsstämpel 1970-01-01 01:00:00

Where "orimligt gammal tidsstämpel" means something like 'unreasonably old timestamp'

md5sum gives the correct hash so not a matter of corrupt download.

I decided to ignore this, is that ok?

Yes, ignore it. It just means that the tar archive was made when the system clock was not up-to-date.

strings is not found[edit]

I don't seem to have the command 'strings' in my newly restored FSG-3 (firmware 3.3.14), so I can't check where to flash the new kernel.

Ok, so instead I plugged in the USB and tried to use that 'USB-stick/usr/bin/strings' but it gave the following error:

1: Syntax error: word unexpected (expecting ")")

Original firmware is big-endian (BE); Debian is little-endian (LE). You can't run LE binaries on BE, and BE on LE.
So: either find or compile a big-endian "strings", or copy the partition to a system which has "strings" binary (dd if=/dev/mtd3 of=/home/mtd3.fsg, then copy that file to some other system).
Simply make a symbolic link from /usr/bin/strings to /bin/busybox. -- 15:56, 13 July 2007 (CEST)

Dual Booting[edit]

Is it possible to get a dual boot loader which loads debian from the USB-stick and if this is not inserted loads the Freecom image from flash? -- 22:39, 19 May 2007 (CEST)


Theoretically, yes, but in practice, it would be lots of work. Why would you want to do that? (tch)

In times where i can play around and test, i could install and configure Debian, while in times where i have to work (often remote over the internet) i could go back to a running installation. It would make testing much less risky. -- 13:51, 20 May 2007 (CEST)
Original FSG-3 firmware has only a subset of features Debian offers. In other words, if you want to do it, well, nothing keeps you :)
Also, you can make one Debian on HDD, and second on a USB-stick (to play with). (tch)
That's the intention. But until i have a Debian version which is perfectly configured, i would like to have a fallback to Freecom. If a running Debian installation is available, i would prefer to have this running and a testing version of Debian. Because there are two MTD devices i would asume that it's possible to install Debian from one and run Freecom by using kexec from the other one. Is that correct or is there a major problem in doing so? -- 17:58, 20 May 2007 (CEST)
Technically, you should be able to do it. There's one thing I'm not sure of: can little endian kernel with kexec start a big endian kernel? I've no idea. (tch)
As RedBoot is able to start an other endian binary, i would assume that this works. But who knows? Hope that i have some time next month to play around a little bit. -- 12:02, 26 May 2007 (CEST)

Error in root/deploy/redboot-config/redboot-config[edit]

My original redboot-config (mtd3) contains the string

exec -c "console=ttyS0,115200 root=/dev/hda2 mem=64M@0x00000000"

while yours from fsg-3-deploy-stick.tar.bz2 contains

exec ^@c "console=ttyS0,115200 root=/dev/hda1 mem=64M@0x00000000"

where ^@ is a binary 0. Think that this should be a -. -- 09:56, 21 May 2007 (CEST)

The "exec" line doesn't really matter - it's not used.
It is really a "leftover" from the previous RedBoot partition (just like you remove a file from a disk, but data is still on the disk).
What is important, is the "fis load kern2" line - it tells to read our new kernel. This new kernel has all bootup logic builtin into initramfs (it boots either from /dev/sdb2, or /dev/sda2, depending on what environment the kernel detects). (tch)
According to this side the "fis load" command only loads the kernel but does not execute it. Thus (in my understanding) it has to be followed by either a "go" or an "exec" command. Or is the only image loaded executed if no more commands follow? -- 14:02, 22 May 2007 (CEST)
Oh - stupid. We have an exec command, but now without additional kernel options. OK. Thinking a little bit, before writing could help ;) -- 14:04, 22 May 2007 (CEST)

Running without Firmware[edit]

If i understand right the firmware itself is a Linux kernel which is started from mtd1 or mtd2. Is it possible to advise RedBoot to start the kernel directly from disk? This would prevent us from using the extra step using flash memory and would be much easier to maintain for people who know Linux. -- 15:47, 13 July 2007 (CEST)

No - RedBoot has no storage/disk drivers, so it can't boot a kernel from disk.
Later however, you can use kexec to boot new kernels, so you don't need to flash them anymore. The original FSG-3 kernel doesn't have kexec functionality (tch).
But i am right, that it's a normal kernel which is flushed to mtd? Or does it need something more? -- 16:23, 13 July 2007 (CEST)
To give you a little bit a background for my questions: Rod Whitby has posted in openfsg forum that OpenWRT works on FSG. If i can change RedBoot to boot from USB-Stick by changing the arguments in mtd3 and if i simply can flush the OpenWRT-kernel to mtd1 or mtd2 i don't need to get serial line access to my FSG. -- 17:12, 13 July 2007 (CEST)
It's a normal kernel (almost - it has bytes swapped: RedBoot is big-endian, and kernel is little-endian)
You don't need a serial cable to flash the kernel (tch).
Thanks. The OpenWRT build is bigendian, thus is should be no problem :) -- 21:25, 13 July 2007 (CEST)

New Firmware installed, FSG does not start and recovery is impossible[edit]

Have just installed a new (own compiled) firmware. After that the FSG does not start any longer and it too does not react on the recovery procedure. After resetting it sends a BOOTP request, which is answered by FSGA with a BOOTP response and after that nothing more happens. Do you have any idea what to do (eg how to get a serial cable and hoe to work with it)? -- 16:16, 15 July 2007 (CEST)

Own compiled firmware, or own compiled kernel? How did you do it?
Anyway - you should have used kexec first, just to see if it boots properly.
Right now, probably your new kernel panics, or RedBoot doesn't boot what it should.
Get a serial cable, and see where it stops - you have some instructions in FSG-3 and NSLU2 wikis (tch).
Do you know is somewhere a fitting serial cable is available? -- 21:59, 15 July 2007 (CEST)
AFAIK, you can't buy a "proper" serial cable for FSG-3. You have to make your own, according to these wikis (it's not that hard) (tch).

Problems(?) found by jpc[edit]

Or maybe the problem is me. Anyway, here they are in case anyone wants to know or has some idea

Boots perfectly from flash, but unable to see internal disk for installation[edit]

After booting from flash the internal disk is not seen(it is the original 500GB disk that came with the FSG, connection is IDE)

fdisk -l, /proc/scsi, dmesg, ... nowhere could I see a trace of the disk.

After doing recovery and using the FSG firmware (4.3.8 beta) I discarded a hardware failure. The disk is seen again. The relevant lines from dmesg are:

<7>libata version 2.00 loaded.
<7>sata_via 0000:00:0c.0: version 2.0
<4>PCI: enabling device 0000:00:0c.0 (0000 -> 0001)
<6>sata_via 0000:00:0c.0: routed to hard irq line 6
<6>ata1: PATA max UDMA/133 cmd 0x1420 ctl 0x142A bmdma 0x1400 irq 22
<6>ata2: PATA max UDMA/133 cmd 0x1440 ctl 0x144A bmdma 0x1410 irq 22
<6>ata3: PATA max UDMA/133 cmd 0x1430 ctl 0x143A bmdma 0x1408 irq 22
<6>scsi0 : sata_via
<4>ATA: abnormal status 0x7F on port 0x1427
<6>scsi1 : sata_via
<6>ata2.00: ATA-7, max UDMA/133, 976773168 sectors: LBA48
<6>ata2.00: ata2: dev 0 multi count 0
<6>ata2.00: configured for UDMA/133
<6>scsi2 : sata_via
<4>ATA: abnormal status 0x7F on port 0x1437
<5>  Vendor: ATA       Model: Hitachi HDT72505  Rev: V56O
<5>  Type:   Direct-Access                      ANSI SCSI revision: 05
<5>SCSI device sda: 976773168 512-byte hdwr sectors (500108 MB)
<5>sda: Write Protect is off
<7>sda: Mode Sense: 00 3a 00 00
<5>SCSI device sda: drive cache: write back
<5>SCSI device sda: 976773168 512-byte hdwr sectors (500108 MB)
<5>sda: Write Protect is off
<7>sda: Mode Sense: 00 3a 00 00
<5>SCSI device sda: drive cache: write back
<6> sda: sda1 sda2
<5>sd 1:0:0:0: Attached scsi disk sda

From /proc/config.gz, the part about the disk is:

# CONFIG_IDE is not set

# SCSI device support
# CONFIG_RAID_ATTRS is not set

# SCSI support type (disk, tape, CD-ROM)
# CONFIG_CHR_DEV_ST is not set
# CONFIG_CHR_DEV_OSST is not set
# CONFIG_CHR_DEV_SG is not set
# CONFIG_CHR_DEV_SCH is not set
# SCSI low-level drivers
# (JPC comment, all of them disabled except the following ones)

So my guess is that maybe some module/configuration is missing in the kernel provided with fsg-3-deploy-stick.tar.bz2.

Could anyone who has managed to install Debian on the hard disk please post here the results of fdisk -l, dmesg, ... ? Thanks

If you want to reach me, I'm jpc, and the part after @ is the two letters h and p followed by .com

I have the same issue and it seems timing related. I'm trying to solve it using a rescan of the IDE bus (echo "- - -" >/sys/class/scsi_host/host2/scan). This always work. After this I mount by label, so it doesn't matter if the partition I need is on sda or sdb.

Copy Request[edit]

May I have permission to copy the contents of this page to the FSG-3 Wiki?

What URL does the page have? Sure, you can copy the contents, but it would be nice to link to this wiki on as well, for reference, or in case if there are any updates here.