Can't find correct configuration

[Zhihao Cheng]

在 2023/9/6 14:57, Leon Pollak 写道:
> Thousands of thanks, Zhihao Cheng!!!
> It worked exactly as you wrote.
> The only question remained is:
> My partition size is 248MiB.
> I thought that leaving 8MiB "spare' will be used just for these
> reserved PEBs, but now I understand this was incorrect.
> So, may I utilize these 8MiB back or will they still be used by UBI
> for something?
> 

I believe you have two mtd devices, 248M(mtd0 for ubi) and 8M(reserved, 
won't be accessed by UBI) now. Please don't split the 256M into two mtd 
devices if you only need 1 UBI device, just format the whole flash(256M) 
into UBI, and UBI itself will handle its reservation without user 
intervening.

> On Wed, 6 Sept 2023 at 05:50, Zhihao Cheng <chengzhihao1 at huawei.com> wrote:
>>
>> 在 2023/9/6 10:42, Zhihao Cheng 写道:
>>> 在 2023/9/5 22:45, Leon Pollak 写道:
>>> Hi
>>>> Hello, all.
>>>> I will be very very thankful for the help to find the correct
>>>> configuration for my NAND.
>>>> I passed through a lot of mails in the lists and google and ChatGPT -
>>>> no success.
>>>> I have 256MB NAND and 248MB partition for UBIFS. I run:
>>>> mkfs.ubifs -v -r rootfs -F -m 2048 -e 126976 -c 1981 -o ./ubifs.img
>>>> ubinize -v -o rootfs.img -m 2048 -p 128KiB -s 512 -O 2048 ubinize.cfg
>>>> -----
>>>> and ubinize.cfg looks like:
>>>> [ubifs]
>>>> mode=ubi
>>>> image=./ubifs.img
>>>> vol_id=0
>>>> vol_size=240MiB
>>>> vol_type=dynamic
>>>> vol_name=ubi_rootfs
>>>> vol_alignment=1
>>>> vol_flags=autoresize
>>>> ------
>>>> Kernel configuration  was with CONFIG_MTD_UBI_BEB_LIMIT=20.
>>>> and here is the kernel log:
>>>> [    2.695659] ubi0: attaching mtd4
>>>> [    3.372343] ubi0: scanning is finished
>>>> [    3.383066] ubi0 warning: ubi_eba_init: cannot reserve enough PEBs
>>>> for bad PEB handling, reserved 4, need 40
>>>> [    3.404320] ubi0: volume 0 ("ubi_rootfs") re-sized from 1982 to
>>>> 1982 LEBs
>>>> [    3.411667] ubi0: attached mtd4 (name "FS", size 248 MiB)
>>>> [    3.417095] ubi0: PEB size: 131072 bytes (128 KiB), LEB size:
>>>> 126976 bytes
>>>> [    3.424006] ubi0: min./max. I/O unit sizes: 2048/2048, sub-page
>>>> size 512
>>>> [    3.430740] ubi0: VID header offset: 2048 (aligned 2048), data
>>>> offset: 4096
>>>> [    3.437734] ubi0: good PEBs: 1990, bad PEBs: 0, corrupted PEBs: 0
>>>> [    3.443860] ubi0: user volume: 1, internal volumes: 1, max. volumes
>>>> count: 128
>>>> [    3.451117] ubi0: max/mean erase counter: 1/0, WL threshold: 4096,
>>>> image sequence number: 1590570073
>>>> [    3.460295] ubi0: available PEBs: 0, total reserved PEBs: 1990,
>>>> PEBs reserved for bad PEB handling: 4
>>>> [    3.469575] ubi0: background thread "ubi_bgt0d" started, PID 131
>>>> [    4.648718] UBIFS (ubi0:0): background thread "ubifs_bgt0_0"
>>>> started, PID 136
>>>> [    4.676640] UBIFS (ubi0:0): start fixing up free space
>>>> [   10.250099] UBIFS (ubi0:0): free space fixup complete
>>>> [   10.269046] UBIFS (ubi0:0): UBIFS: mounted UBI device 0, volume 0,
>>>> name "ubi_rootfs"
>>>> [   10.276830] UBIFS (ubi0:0): LEB size: 126976 bytes (124 KiB),
>>>> min./max. I/O unit sizes: 2048 bytes/2048 bytes
>>>> [   10.286794] UBIFS (ubi0:0): FS size: 250142720 bytes (238 MiB, 1970
>>>> LEBs), journal size 9023488 bytes (8 MiB, 72 LEBs)
>>>> [   10.297540] UBIFS (ubi0:0): reserved for root: 0 bytes (0 KiB)
>>>> [   10.303406] UBIFS (ubi0:0): media format: w4/r0 (latest is w5/r0),
>>>> UUID A14D586A-32E2-45D7-A07E-BDC472FDF31D, small LPT model
>>>>
>>>> My question is: why does this warning about reserved 4, needed 40
>>>> appear? How can I solve this?
>>>>
>>>
>>> Solution: Modify ubinize.cfg,  vol_size=240MiB -> vol_size=235MiB
>>>
>>> Why does this warning about reserved 4, needed 40 appear?
>>> The count of bad PEBs reserved by UBI is: total_pebs(nand chip) *
>>> CONFIG_MTD_UBI_BEB_LIMIT/1024 = 2048 *  20 / 1024 = 40
>>> When fastmap is not enabled, the count of internal PEBs reserved by UBI
>>> is: UBI_LAYOUT_VOLUME_EBS[2] + WL_RESERVED_PEBS[1] +
>>> EBA_RESERVED_PEBS[1] = 4
>>>
>>> Total PEBs for UBI is 1990(good PEBs: 1990), the max count of user
>>> volume PEBs is 1990 - 40 - 4 = 1946
>>>
>>> In ubinize.cfg, vol_size=240MiB, the count of user volume PEBs is(see
>>> ubinize.c in mtduitls): vi_size + (leb_size - 1) / leb_size
>>> vi_size = 240M = 240 * 1024 * 1024
>>> leb_size = 128K - 2K(ec) - 2K(vid) = 124 * 1024
>>> result = 1982
>>>
>>> So there are 1990 - 1982 - 4 = 4 PEBs reserved for bad PEB handling,
>>> which is less than 40.
>>>
>>> When we set  vol_size=235MiB, the count of user volume PEBs is: (235 *
>>> 1024 * 1024 + 124 * 1024 - 1) / (124 * 1024) = 1941
>>> 1990 - 4 - 40 = 1946, finally maybe there is message from kernel like
>>> "("ubi_rootfs") re-sized from 1941 to 1946 LEBs".
>>
>> Acutally, the 'vol_size' in ubinize.cfg means the size of data on
>> volume(exclude vid/ec), not the size of volume(include vid/ec).
>>
> 
> .
>