Re: [UBIFS] issue with ubiattach

[Artem Bityutskiy <dedekind1@gmail.com>]

Hi,

On Wed, 2009-12-16 at 14:25 +0100, Bosi Daniele wrote:
> Hello,
> I'm working on a linux 2.6.24.6 for powerpc MPC5121 with latest ubi patches installed.
> With the latest patch from the official repository I run in the following error when I try to prepare my UBI partitions on a NOR flash

Veeeery old kernel .... We do not support it any longer.

> I did this steps:
> 
> 1. flash_eraseall /dev/mtd2;
> 2. ubiformat /dev/mtd2;

If you use ubiformat, you should not use flash_eraseall, because it
kills erase markers.

> 3. ubiattach /dev/ubi_ctrl -m 2;
> 4. ubimkvol /dev/ubi0 -N fredubifsCFG -s 4MiB;
> 5. ubimkvol /dev/ubi0 -N fredubifsRW -m;
> 
> The output of step 3 is:
> 
> [   33.815099] UBI: attaching mtd2 to ubi0
> [   33.819451] UBI: physical eraseblock size:   131072 bytes (128 KiB)
> [   33.826203] UBI: logical eraseblock size:    130944 bytes
> [   33.832033] UBI: smallest flash I/O unit:    1
> [   33.836912] UBI: VID header offset:          64 (aligned 64)
> [   33.843018] UBI: data offset:                128
> [   33.867213] UBI: attached mtd2 to ubi0
> [   33.874851] UBI: MTD device name:            "filesystemNOR"
> [   33.882135] UBI: MTD device size:            47 MiB
> [   33.888384] UBI: number of good PEBs:        376
> [   33.893514] UBI: number of bad PEBs:         0
> [   33.898388] UBI: max. allowed volumes:       128
> [   33.903474] UBI: wear-leveling threshold:    4096
> [   33.908597] UBI: number of internal volumes: 1
> [   33.913458] UBI: number of user volumes:     0
> [   33.918342] UBI: available PEBs:             372
> [   33.923381] UBI: total number of reserved PEBs: 4
> [   33.928503] UBI: number of PEBs reserved for bad PEB handling: 0
> [   33.934976] UBI: max/mean erase counter: 3/1
> [   33.939663] UBI: image sequence number: 0
> [   33.944101] UBI: background thread "ubi_bgt0d" started, PID 424
> UBI device number 0, total 376 LEBs (49234944 bytes, 47.0 MiB), available 372 LEBs (48711168 bytes, 46.5 MiB), LEB size 130944 bytes (127.9 KiB)
> 
> But after step 3 I found under the directory "/dev" a device node like this:
> 
> root@devece:~# ls -l /dev/ubi*
> brw-rw----    1 root     root     251,   0 Dec 15 17:48 /dev/ubi0
> crw-rw----    1 root     root      10,  63 Jan  1  1970 /dev/ubi_ctrl

Please, check /proc/devices - it lists all character devices. I believe
UBI is creating a character device. The nodes are created by udev or
similar stuff, which is in user-space, and that is the place you should
dig.

> 
> As you can see there's an error on "/dev/ubi0" that should be a char device instead of a block device.
> Then I get stucked on step 4 with the following output:
> 
> libubi: error!: "/dev/ubi0" is not a character device
> ubimkvol: error!: error while probing "/dev/ubi0"
>           error 22 (Invalid argument)
> 
> BTW the mtd partitions on the device are like this:
> 
> root@device:~# cat /proc/mtd
> dev:    size   erasesize  name
> mtd0: 00040000 00020000 "protected"
> mtd1: 00c00000 00020000 "rootfsRO"
> mtd2: 02f00000 00020000 "filesystemNOR"
> mtd3: 00380000 00020000 "kernel"
> mtd4: 00040000 00020000 "device-tree"
> mtd5: 00100000 00020000 "u-boot"
> mtd6: 04000000 00020000 "flash1"
> mtd7: 04000000 00020000 "flash2"
> mtd8: 04000000 00020000 "flash3"
> mtd9: 40000000 00020000 "nand"
> 
> So my questions are:
> Do somebody knows why the device node is wrong after the ubiattach?

Probably an error in udev rules or whatever you use instead of it.

> Who's the device node creator (because I've seen that the device is created after the ubiattach operation)?

The character device is created inside the kernel when it attaches the
MTD device. Then user-space creates a device node (/dev/ubi0), and this
is where you should search the error.

> Is it mdev / udev or the ubiattach (I didn't find any evidence of this in the ubiattach srcs)?

The former.

-- 
Best Regards,
Artem Bityutskiy (Артём Битюцкий)