[JFFS2] Request help on EBS option in Linux-2.6.21.5

[zheng shi]

On Wed, Oct 8, 2008 at 11:36 AM, zheng shi <neversetsun at gmail.com> wrote:
> I got a little confused about making an empty file system.
> I think "flash_eraseall /dev/mtdX" will do, but I get messages like:
>
> ...
> JFFS2: Erase block at 0x03f40000 is not formatted. It will be erased
> JFFS2: Erase block at 0x03f80000 is not formatted. It will be erased
> JFFS2: Erase block at 0x03fc0000 is not formatted. It will be erased
> Cowardly refusing to erase blocks on filesystem with no valid JFFS2 nodes
> empty_blocks 0, bad_blocks 0, c->nr_blocks 256
>
> So I changed my way:
> 1. mkdir empty
> 2. mkfs.jffs2 -e 0x40000 -s 0x800 --pad=0x40000 -n -d empty/ -o img/emtpy.img
> (-p 0x40000 seems not to work, it just generates an image with zero
> size, only long option --pad generates an image with 256*1024 byte)

sorry for missing:
2.5  flash_eraseall  /dev/mtd3

> 3. nandwrite -a -p -m /dev/mtd3 img/empty.img
>
> Have I mistaken anything?
>
> BTW,
> My NAND flash's parameter:
> page size: 2KB
> oob size: 64B
> Block size: 256KB
>
> On Fri, Oct 3, 2008 at 3:03 PM, David Woodhouse <dwmw2 at infradead.org> wrote:
>> On Thu, 2008-10-02 at 10:00 +0800, zheng shi wrote:
>>>
>>>
>>> I'm using JFFS2 in linux-2.6.21.5.
>>>
>>> When I enabled JFFS2 summary support (EXPERIMENTAL) option, the EBS
>>> mechanism did promote mount time.
>>> But unfortunately files (size>10k) couldn't be written on flash
>>> permanently, i.e., though cp was done, but when remount the file just
>>> copied disappeared.
>>
>> That's strange. Can you reproduce on an empty file system? Just mount it
>> empty, write the large file, then unmount. Take a snapshot of the
>> contents of the flash. Then mount it, confirm that the file is missing,
>> and show me the contents of the flash.
>>
>> Also, build your kernel with CONFIG_JFFS2_FS_DEBUG=1 and capture all the
>> debug output (over a serial port) when you mount the offending file
>> system.
>>
>> The 2.6.21.5 kernel isn't particularly interesting though -- please try
>> something more recent, first.
>>
>>> I have tried -o sync with mount option.
>>
>> That should make no difference.
>>
>> --
>> David Woodhouse                            Open Source Technology Centre
>> David.Woodhouse at intel.com                              Intel Corporation
>>
>>
>
>
>
> --
> Regards, neversetsun
>



-- 
Regards, neversetsun