UFS2 mount in FreeBSD

hi ,

afaik ufs2 is the "home" filesystem of the freebsd OS.

I have:

A 1TB 2,5" HDD that ist formatted on a Panasonic Plasma. The TV does the following with the disk:

gpart show (under freebsd 9.0 64 bit)
Code:
=>        34     1953525101    da1   GPT   (931G)
          34             54          - free - (27k)
          88     1953520065      1   freebsd  (931G)
  1953520153           4982          - free - (2.4M)

# mount -t ufs /dev/da1s1 /mnt/ufs
Code:
mount : /dev/da1s1 : Invalid Argument
Why?

With a Knoppix 7.05 Live Cd I get rw access to the partition. Problem is under Linux I cannot copy files greater 2 GB onto the partition.

Has anyone ideas why I cannot mount under FreeBSD??

Thanks.
 
I assume because you called it /dev/da1s1 when it should be /dev/da1p1.

edit: Now I see what SirDice must have meant. The partition type is "freebsd" not "freebsd-ufs".
 
It's a slice, not a partition. There are partitions inside it. You need to mount those.
 
OK I tried also

- /dev/da1p1
- /dev/da1p1a
- /dev/da1s1a

-> no such file or directory

Why can I mount it with ubuntu, knoppix but not with FreeBSD?

Any other ideas?
Do you need more informations?

THX
 
Code:
cd /dev
ls -lac | grep p   # may show other stuff
ls -Gacd /*        # look in each subdir, maybe mountable via label
mount .... /mnt    # mount at the base not at a subdirectory
Stuff to try?
 
Sorry that I answer so late. I tried all of your stuff.

The reason why is, I think, it's a slice formatted on a Big endian system. Now I am trying to get an environment where I can test that. Testing around with qemu. Is anyone familiar with emulating e.g. a PowerPC an installing an FreeBSD on it?

Thanks.

Chris
 
hi community,

I thought it would be more trivial to solve that. I need help. The partition is created and filled with a big endian system. The partition can be mounted in other distros. It's also possible to write on it. BUT only for files until 2GB. larger files are not supported. Is there any way to pick out the driver out of the FreeBSD and compile it into a e.g. ubuntu distro?

I tried to set up an qemu with powerpc or sparc with FreeBSD, but I didn't make it.

Any hints?
Thanks very much
Chris
 
I don't think it's a UFS filesystem (Endianness has nothing to do with it).
 
That should not be a problem of endianess, it could be much easier.
There is UFS and UFS2 and FreeBSD can use both. UFS(1) might have some of the limits you see (2GB), but I am currently not able to check that up.
 
The UFS explorer in windows says it is a big endian partition. So in an other forum they told me the UFS driver in linux is limited to 2GB. I don't understand why this is so. But why it isn't possible to mount it with FreeBSD? Why mounting in ubuntu works?

More ideas?

Thanks.
 
No I tried to copy large files with knoppix. After 2GB the system stops copying.

More stuff:

# fsck_ufs /dev/da0s1
Code:
cannot find file system superblock
ioctl ( GCINFO ) Inappropriate ioctl for device
fsck_ufs : /dev/da0s1: can't read disk label

Problem with GPT??
 
This thread makes my head hurt. When did the drive change from /dev/da1p1 to /dev/da0s1?

da0s1 is MBR. GPT partitions appear as p1, p2, and so on. MBR slices are s1, s2.

So no, not a problem with GPT.

Are there FreeBSD partitions inside that slice?
$ gpart show da0s1
 
# gpart show da0s1

Code:
 gpart: No such geom: /dev/da0s1

# ls /dev/da0

Code:
/dev/da0      /dev/da0s1

da0p1
doesn't exist.

At present the HDD is connected through an USB-Sata Bridge. I will try also connect it to my desktop computer and boot a FreeBSD there. I don't want to hurt you, but I hope you help me further to fix that problem. What's about the idea porting the ufs driver to ubuntu? Isn't that possible?

Thank you very much.
Chris
 
It's possible that Panasonic is not using a standard filesystem or any filesystem. Let's try this:
$ fdisk da0
# file -s /dev/da0s1
 
the output is too long, i cannot copy it out of my vm.

I attached Screenshots.

i can only repeat: Mounting in Ubuntu, Debian, Knoppix ...no Problem.

the

# file

output looks interesting.

its your turn again.

THX
Chris
 

Attachments

  • fdisk.jpg
    fdisk.jpg
    94.6 KB · Views: 1,645
  • file.jpg
    file.jpg
    84.3 KB · Views: 1,381
I think you get da0s1 if you use GPT partioning but create a bsdlabel(8) on the first GPT partition. Such scheme is not recommended of course but it should work.

Does this produce anything ?

# bsdlabel /dev/da0s1
 
That... doesn't sound right. A bsdlabel on any GPT partition can't change the partition scheme, it would just be da0p1a. It's really, really odd that gpart(8) shows a GPT partition scheme but the device shows up in /dev as MBR.

LaChris, have you built a custom kernel?
 
wblock@ said:
That... doesn't sound right. A bsdlabel on any GPT partition can't change the partition scheme, it would just be da0p1a. It's really, really odd that gpart(8) shows a GPT partition scheme but the device shows up in /dev as MBR.

LaChris, have you built a custom kernel?


Yes you're right, bsdlabel on GPT partition creates da0p1a and so on.

I'm pretty sure any recent FreeBSD version creates little-endian filesystems even on big-endian systems so that the filesystems are usable between architectures with different endianess. The filesystem in question might not from FreeBSD after all as suggested above.
 
It could be NetBSD in that TV, but I am currently not aware of the differences in file system implementations between NetBSD and FreeBSD. I know there are some, but to what degree I do not know right now. But maybe there is some confusion on the disc. According to the AllKnowingLandfill there is a MBR on the disc along the GPT. My hunch would be that this MBR is not the "protective" MBR but tries to map the same area as the GPT in an attempt to make the disc usable for MBR-only and GPT-only systems. The fdisk output you supplied suggests that this is the case, but my experience with this is limited.
 
Hi,

No ideas left? Is there no chance to use parts of the distro? How can I run that kernel in a virtual machine? Please don't let me alone with that.

Regards and thanks,

Chris
 
Back
Top