Create, Flash and Mount UBIFS Images

Hello,

Could anybody explain me how to build an ubi.img that I could flash inside my system ?

NAND Flash: Micron 256 MByte
Physical Eraseblock Size = 131072
Logical Eraseblock Size = 129024
Min I/O Size = 2048
SubPage = 512

The Partition /dev/mtd6 is 32MByte.

I did the following steps:
On my host I created the file through:

mkfs.ubifs -d /exports/ -m 2048 -e 129024 -o ubifs.img -x "none" -c 2047
ubinize -p 128KiB -m 2048 -s 512 -O 512 -o ubi.img ubi.ini

with ubi.ini
[TEST]
mode=ubi
image=ubifs.img
vol_id=0
vol_size=3MiB
vol_type=dynamic
vol_alignment=1
vol_flags=autoresize

Then I copy this creted ubi.img to my embedded system.

On my embedded system I did:
flash_eraseall /dev/mtd6
ubiformat /dev/mtd6 -s 512 -f ubi.img
ubiattach /dev/ubi_ctrl -m6 -d6 -O512
mount -t ubifs ubi6_0 /mnt/ubi6

Then at the mount command I got the error:
validate_sb: bad superblock

So what´s wrong ?

Thank you

Hi,

Could you please try to enable debugging, as it is asked here:

http://www.linux-mtd.infradead.org/doc/ubifs.html#L_how_send_bugreport

And I do not understand why you did not provide all UBIFS output. The print is:

ubifs_err("bad superblock, error %d", err);

and you did not provide error code number.

Best regards,
Artem Bityutskiy (Битюцкий Артём)

Hello, sorry for the previous post without informations...

Here are the last two steps I did embedded:
[Attach the /dev/mtd6 to ubifs and try to mount it] [Before I did a "flash_eraseall /dev/mtd6" and a "ubiformat /dev/mtd6 -s512 -f ubi.img"]

/usr/sbin # ubiattach /dev/ubi_ctrl -m6 -d6 -O512
UBI: attaching mtd6 to ubi6
UBI: physical eraseblock size: 131072 bytes (128 KiB)
UBI: logical eraseblock size: 129024 bytes
UBI: smallest flash I/O unit: 2048
UBI: sub-page size: 512
UBI: VID header offset: 512 (aligned 512)
UBI: data offset: 2048
UBI: attached mtd6 to ubi6
UBI: MTD device name: "Partition 7 - UPDATE : 32 MByte"
UBI: MTD device size: 32 MiB
UBI: number of good PEBs: 256
UBI: number of bad PEBs: 0
UBI: max. allowed volumes: 128
UBI: wear-leveling threshold: 4096
UBI: number of internal volumes: 1
UBI: number of user volumes: 1
UBI: available PEBs: 233
UBI: total number of reserved PEBs: 23
UBI: number of PEBs reserved for bad PEB handling: 2
UBI: max/mean erase counter: 0/0
UBI: background thread "ubi_bgt6d" started, PID 1336
UBI device number 6, total 256 LEBs (33030144 bytes, 31.5 MiB), available 233 LEBs (30062592 bytes, 28.7 MiB), LEB size 129024 bytes (126.0 KiB)
/usr/sbin # mount -t ubifs ubi

ubiattach     ubidetach     ubimirror     ubinize       ubiupdatevol
ubicrc32      ubiformat     ubimkvol      ubirename
ubicrc32.pl   ubigen        ubinfo        ubirmvol

/usr/sbin # mount -t ubifs ubi6_0 /mnt/OBIDfs_ubi/
UBIFS: static UBI volume - read-only mode
UBIFS error (pid 1391): mount_ubifs: cannot mount read-write - read-only media
UBIFS: static UBI volume - read-only mode
UBIFS error (pid 1391): validate_sb: bad superblock, error 4
mount: mounting ubi6_0 on /mnt/OBIDfs_ubi/ failed: Invalid argument

Any idea why this seems to fail ?

By the way if I use:
"flash_eraseall /dev/mtd6"
"ubiformat /dev/mtd6 -s512"
"ubiattach /dev/ubi_ctrl -d6 -m6 -O512"
"ubimkvol /dev/ubi6 -m -N "TEST""
"mount -t ubifs ubi6_0 /mnt/ubi/"

this works ??

Thank you very much

Hi,

Could you please try to enable debugging, as it is asked here:

http://www.linux-mtd.infradead.org/doc/ubifs.html#L_how_send_bugreport

And I do not understand why you did not provide all UBIFS output. The print is:

ubifs_err("bad superblock, error %d", err);

and you did not provide error code number.

Best regards,
Artem Bityutskiy (Битюцкий Артём)

This ubi6 is weird. Do you have 6 UBI devices, or this is just because of "-d6"? Could you please try without "-d6"?

In the previous e-mail it was said that the volume is dynamic. Why it is static now? hmm..

Could you please enable UBIFS debugging ?
And then send what dmesg says you, not only what you see on the console, please.

Best regards,
Artem Bityutskiy (Битюцкий Артём)

Are you sure -c 2047 is a sane value for a 32MiB volume?

Best regards,
Artem Bityutskiy (Битюцкий Артём)

Hello, thank you for your help.
I got something working, but don´t know really what I´ve done:

OK, let me gibve you some more information about my sytem:

My 256MB NAND Flash is divided into 7 MTD partitions:
mtd0: 128KiB (First Level Bootloader)
mtd1: 896KiB (UBoot)
mtd2: 4MiB (Kernel)
mtd3: 40MiB (RFS)
mtd4: 20MiB (Private Partition)
mtd5: 20 MiB (Private Partition2)
mtd6: 32Mib (Backup Partition)

Now I at first I need to build an ubi image with mkfs.ubifs and with ubinze on my host. I tried it with this settings:
mkfs.ubifs -d /exports/ -m 2048 -e 129024 -c 326 -o RFS.ubifs

There is my first question:
----------------------------------------
What about the pramater -c ?
In the FAQ: -c 2047: specifies maximum file-system size in logical eraseblocks; this means that it will be possible to use the resulting file-system on volumes up to this size (less or equivalent); so in this particular case, the resulting FS may be put on volumes up to about 251MiB (129024 multiplied by 2047); See this section for more details.

So my MTD Partition is 40MiB; Logical Ersablocksize is 129024 Bytes;
--> (40MiB / 129024)+1 = c --> c=326
Is this correct ?

Now I go on with ubinize:
ubinize -p128KiB -m 2048 -s 512 -O 512 -o RFS.ubi config.ini
with config.ini:
[RFS]
mode=ubi
image=RFS.ubifs
vol_size=30MiB
vol_id=0
vol_type=dynamic
vol_name=RFS
vol_alignment=1

There is my second question:
-----------------------------------------
Did I miss something in the ini File ?
What about the line vol_flags=autoresize - Do I need this one ?

There is my third question:
-----------------------------------------
vol_size=30MiB -> The mtd partition is 40MiB, but if I use vol_size=40MiB there appears an error concerning a table ? Which value should I use ?

There is my fourth question:
-----------------------------------------
Now the ubi image is created -> I flashed it into the nand an put the kernel paramter mtd.ubi in the kernel command line.
After the ubi images are attached and mounted, there seems to less memory ?
There should by 30MiB instead of 28 MiB ?
What about the journal size ?
Is my user Memory = UBIFS: file system size - UBIFS: journal size ?

UBIFS: recovery needed
UBIFS: recovery completed
UBIFS: mounted UBI device 0, volume 0, name "RFS"
UBIFS: file system size: 30191616 bytes (29484 KiB, 28 MiB, 234
LEBs)
UBIFS: journal size: 5548032 bytes (5418 KiB, 5 MiB, 43 LEBs)
UBIFS: media format: 4 (latest is 4)
UBIFS: default compressor: LZO
UBIFS: reserved for root: 0 bytes (0 KiB)
VFS: Mounted root (ubifs filesystem).

df -h returns
Filesystem 1K-blocks Used Available Use%
Mounted on
ubi0_0 27256 17788 9468 65%
/

Thank you very much
Manuel

Are you sure -c 2047 is a sane value for a 32MiB volume?

Best regards,
Artem Bityutskiy (Битюцкий Артём)

Hi,

Well, you have so many MTD partitions which is in general bad design. The whole idea of UBI spreading wear evenly across the chip is just not used. But you probably know what you do...

Is the description not understandable? If yes, ask specific question please. And also, try this "See this section for more details" link.

Sounds right.

I do not know if you need it or not. First of all try to understand what is autoresize, this is documented at the MTD web site. If you cannot understand, ask specific questions please.

Volume size is the amount of free space in the volume. There is overhead, which should be taken into account. Volume of size M needs partition of size F, and F > M. The overhead calculation is described on the MTD web site.

If you want a volume to take whole partition, the easiest way to do so is to use the "autoresize" flag for this volume, and make the volume 30MiB in this case. Please, read docs to realize why.

You mean "flash memory", i guess.

It is also not clear if you refer UBI volume size or amount of free space in UBIFS. UBIFS sits _inside_ UBI volume. It is encapsulation. It does not affect UBI volume size.

No, not really. There are many sources of overhead, and in general UBIFS cannot tell how much space you have. Please, take a look at UBIFS docs:
http://www.linux-mtd.infradead.org/doc/ubifs.html#L_spaceacc

The available part is usually not accurate.

Best regards,
Artem Bityutskiy (Битюцкий Артём)

Hello,

concerning the autoresize flag:

As far as I understand:
If a partition is 40 MiB, I have to set the vol_size < Partition size, and set the autoresize flag to use the whole Partition size of 40 MiB. right ?

Is it correct that the resize depends on the value -c of the mkfs.ubifs program ?
--> (40MiB / 129024)+1 = c --> c=326
With c=326 the autoresize feature should make the ubifs use the whole partitoon sizue of 40 MiB - right ?

concerning the kernel command parameter:
Is it possible to define the ubi device number trough the kernel command line argument, like it is possible through the ubiattach program ?
Example: ubi.mtd=4 is actually attached to /dev/ubi1 and /dev/ubi1_0. I want to have /dev/ubi4 and /dev/ubi4_0

possible ?

Thank you very much.
Manuel

Hi,

Well, you have so many MTD partitions which is in general bad design. The whole idea of UBI spreading wear evenly across the chip is just not
used. But you probably know what you do...

resulting

this

details.

Is the description not understandable? If yes, ask specific question please. And also, try this "See this section for more details" link.

Sounds right.

I do not know if you need it or not. First of all try to understand what
is autoresize, this is documented at the MTD web site. If you cannot understand, ask specific questions please.

Volume size is the amount of free space in the volume. There is overhead, which should be taken into account. Volume of size M needs partition of size F, and F > M. The overhead calculation is described on the MTD web site.

If you want a volume to take whole partition, the easiest way to do so is to use the "autoresize" flag for this volume, and make the volume 30MiB in this case. Please, read docs to realize why.

the

You mean "flash memory", i guess.

It is also not clear if you refer UBI volume size or amount of free space in UBIFS. UBIFS sits _inside_ UBI volume. It is encapsulation. It does not affect UBI volume size.

No, not really. There are many sources of overhead, and in general UBIFS cannot tell how much space you have. Please, take a look at UBIFS docs:
http://www.linux-mtd.infradead.org/doc/ubifs.html#L_spaceacc

The available part is usually not accurate.

Best regards,
Artem Bityutskiy (Битюцкий Артём)

Yes.

The autoresize is UBI thing. -c is UBIFS thing. Yes, -c defines maximum UBIFS size in LEBs. Make it larger, e.g. -c 511 if you are concerned.

Not exactly. But kind of. You should distinguish between UBI and UBIFS - they are different thing. UBIFS does not know anuthing about MTD partition, it knows only about UBI space, which is less than partition space.

But just use -c511, it is safe. UBIFS overhead increases considerably when this -c value goes over a power-of-2 boundary. So there is not that much difference between -c 326 and -c 512.

Hmm, not really. We did not think about this. May be you may add this feature - should not be too difficult.

Best regards,
Artem Bityutskiy (Битюцкий Артём)