ubifs wear leveling across entire device

twebb taliaferro62 at gmail.com
Mon Dec 7 10:41:08 EST 2009


>>
>> I'm confused about how wear leveling occurs across the entire device
>> in the following example.  Could someone help me understand what I'm
>> missing?
>>
>> 4 GiB flash device (nand MLC):
>> 4 MTD partitions defined:
>> 0 - "p1" - 0x00000000 - 0x00800000
>> 1 - "p2" - 0x00800000 - 0x10000000
>> 2 - "p3" - 0x10000000 - 0x80000000
>> 3 - "p4" - 0x80000000 - 0x100000000
>>
>> kernel args specify 3 ubi volumes:
>> ubi.mtd=p1 ubi.mtd=p2 ubi.mtd=p3
>> (p4 is reserved and no volumes or filesystems sit on top of it)
>
> That is 3 UBI devices not 3 UBI volumes
>

Excuse me.  I'm new to using UBI and UBIFS and am not completely
familiar with the terminology.  I think what I described in the
example is 3 UBI devices, with a single volume associated with each
device.  I had neglected to show that each UBI device had a single
volume.

Maybe there's a better way to word my question.  Does UBI wear
leveling occur independently across each MTD device (p1, p2, and p3
only in example above) or does it somehow occur over the entire
physical device (p1-p4 inclusive)?

I'm trying to understand this so I can determine how to partition the
physical device to provide the best wear leveling, while still
allowing a portion of the physical device (p4) to exist outside of any
UBI device or volume.

Thanks,
twebb



More information about the linux-mtd mailing list