UBI ECC errors on kernel 3.16.2

Mon Oct 6 13:13:02 PDT 2014

Hello Angelo,

On Monday 06 October 2014 02:32 PM, Angelo Dureghello wrote:
> Dear,
>
> many thakns for your support. Hope i can contribute to help you also in the
> future, i will try to post some useful patch someday.
>
> I tried to run mtd tests, but for some other issues encoutered on
> running this
> kernel version (PHY link not detected) i couldn't mount the NFS for
> testing,
> and couldn't execute them.So i had to fight inside the mtd and mtd/nand
> folders to find out the reason of this.
>
>
> The following temporary fix into nand_base.c seems to solve my issue.
>
>           if (!ecc->write_oob)
>               ecc->write_oob = nand_write_oob_std;
>           /* Angelo
>            * temporary patch, no subpage wrote allowed for ipam390,
>            * must comment them out, or ECC layout get broken.
>            */
> !        //if (!ecc->read_subpage)
> !        //    ecc->read_subpage = nand_read_subpage;
> !        //if (!ecc->write_subpage)
> !        //    ecc->write_subpage = nand_write_subpage_hwecc;
>
> It is connected to the fact that the driver want to use "subpage" write
> from
> some kernel versions, since this binding was not there into 3.5.1.
>
> Maybe you can now clarify some mistery i still cannot understand:
> 1) My flash does not support subpages write (page 2048, no subpages
> support from
> the datasheet), so i set subpage and max/io size to 2048. But sill UBI
> attach
> shows the 512 supbage as available. Upon this nand flash limit, is also the
> davinci-nand (aemif HW) controller able to support subpages write ? If
> yes, is it
> also the driver for it able to support subpages write ? From my
> understanding, it
> was not, and now it should.
>
I'm not sure how mtd-utils use the arguments but following combination 
worked for me to mount a UBI image without sub-pages.
(1) mkfs.ubifs  -m (min-io-size=2048)   [-s (subpage-size=2048)]
(2) ubinize     -m (min-io-size=2048)
(3) ubiformat   -O 2048  (indicates that offset of volume-ID header from 
start of block)

> 2) So how does it works this subpages-write support if the nand flash
> does not
> supports subpages write ?
>
You can disable sub-page write by adding following in your driver probe.
+	chip->options |= NAND_NO_SUBPAGE_WRITE

You should also be able to specify this via DT bindings if driver 
supports one.

with regards, pekon

> Thanks
>
> Regards,
> Angelo
>
>
> On 06/10/2014 06:49, Pekon Gupta wrote:
>> Hi Angelo,
>>
>>
>> On 02-10-2014, Angelo Dureghello wrote:
>>> Hi all,
>>>
>>> still some updated on these -74 EBADMSG i am receiving.
>>>
>>> well, seems the first ecc error is detected as soon as the kernel
>>> driver starts to
>>> read the ubifs (so file system) data part of the rootfs.ubi image.
>>>
>>> Before reading the file system data, so attaching, there is no ecc
>>> error detected at all.
>>> I added traces on some kernel file as nand_base.c.
>>>
>>>
>>> Ubi scanning / attaching  ...
>>>
>>> nand_read_page_hwecc_oob_first page    :3659
>>> nand_read_page_hwecc_oob_first correct p:c883d800 p[0]:p[1] 00:00 i:0
>>> eccpos[i]:06 ecc_code[i]:0b;
>>> nand_read_page_hwecc_oob_first correct p:c883da00 p[0]:p[1] 00:00
>>> i:10 eccpos[i]:16 ecc_code[i]:58;
>>> nand_read_page_hwecc_oob_first correct p:c883dc00 p[0]:p[1] 00:00
>>> i:20 eccpos[i]:26 ecc_code[i]:cf;
>>> nand_read_page_hwecc_oob_first correct p:c883de00 p[0]:p[1] 00:00
>>> i:30 eccpos[i]:36 ecc_code[i]:8b;
>>> nand_read_page_hwecc_oob_first page    :3660
>>> nand_read_page_hwecc_oob_first correct p:c883e000 p[0]:p[1] 00:00 i:0
>>> eccpos[i]:06 ecc_code[i]:9b;
>>> nand_read_page_hwecc_oob_first correct p:c883e200 p[0]:p[1] 00:00
>>> i:10 eccpos[i]:16 ecc_code[i]:f1;
>>> nand_read_page_hwecc_oob_first correct p:c883e400 p[0]:p[1] 00:00
>>> i:20 eccpos[i]:26 ecc_code[i]:26;
>>> nand_read_page_hwecc_oob_first correct p:c883e600 p[0]:p[1] ff:ff
>>> i:30 eccpos[i]:36 ecc_code[i]:3f;
>>> UBI: volume 0 ("rootfs") re-sized from 205 to 456 LEBs
>>> UBI: attached mtd6 (name "rootfs", size 60 MiB) to ubi0
>>> UBI: PEB size: 131072 bytes (128 KiB), LEB size: 126976 bytes
>>> UBI: min./max. I/O unit sizes: 2048/2048, sub-page size 512
>>> UBI: VID header offset: 2048 (aligned 2048), data offset: 4096
>>> UBI: good PEBs: 480, bad PEBs: 0, corrupted PEBs: 0
>>> UBI: user volume: 1, internal volumes: 1, max. volumes count: 128
>>> UBI: max/mean erase counter: 1/0, WL threshold: 4096, image sequence
>>> number: 272604537
>>> UBI: available PEBs: 0, total reserved PEBs: 480, PEBs reserved for
>>> bad PEB handling: 20
>>> UBI: background thread "ubi_bgt0d" started, PID 995
>>> gpio-keys gpio-keys.0: Failed to request GPIO 126, error -517
>>> platform gpio-keys.0: Driver gpio-keys requests probe deferral
>>> omap_rtc da830-rtc: setting system clock to 2014-10-02 15:59:28 UTC
>>> (1412265568)
>>> ALSA device list:
>>>   No soundcards found.
>>>
>>> *** reading the file system here ***
>>>
>>> At page 3712 there is the first of the file system blocks
>>> 3712        3713           3714         3715
>>> EC HEADER  |  VID HEADER  |  fs data   |   fs data   etc
>>>                            ^
>>>                            ^
>>>
>>> nand_read_page_hwecc_oob_first page    :3714
>>> nand_read_page_hwecc_oob_first error   p:c7906000 p[0]:p[1] 31:18 i:0
>>> eccpos[i]:06 ecc_code[i]:1f;    <<< ERROR
>>> nand_read_page_hwecc_oob_first correct p:c7906200 p[0]:p[1] 00:00
>>> i:10 eccpos[i]:16 ecc_code[i]:00;
>>> nand_read_page_hwecc_oob_first correct p:c7906400 p[0]:p[1] 00:00
>>> i:20 eccpos[i]:26 ecc_code[i]:00;
>>> nand_read_page_hwecc_oob_first correct p:c7906600 p[0]:p[1] 00:00
>>> i:30 eccpos[i]:36 ecc_code[i]:00;
>>> ecc_failed !!
>>> nand_read_page_hwecc_oob_first page    :3715
>>> nand_read_page_hwecc_oob_first correct p:c7906800 p[0]:p[1] 00:00 i:0
>>> eccpos[i]:06 ecc_code[i]:00;
>>> nand_read_page_hwecc_oob_first correct p:c7906a00 p[0]:p[1] 00:00
>>> i:10 eccpos[i]:16 ecc_code[i]:00;
>>> nand_read_page_hwecc_oob_first correct p:c7906c00 p[0]:p[1] 00:00
>>> i:20 eccpos[i]:26 ecc_code[i]:00;
>>> nand_read_page_hwecc_oob_first correct p:c7906e00 p[0]:p[1] 00:00
>>> i:30 eccpos[i]:36 ecc_code[i]:00;
>>> UBI warning: ubi_io_read: error -74 (ECC error) while reading 4096
>>> bytes from PEB 2:4096, read only 4096 bytes, retry
>>>
>>>
>>> I am tracing the first 2 bytes only of each 512B eccblock.
>>> I verified, first 2 bytes with errors (0x31, 0x18) are sane, as in
>>> the rootfs.ubi file.
>>>
>>> So, i am supposing these errors are caused from a misalignment from
>>> u-boot and kernel davinci / nand drivers
>>> that calculate the ecc values.
>>>
>>> U-Boot 2014.07-03397-gab92542 (Oct 02 2014 - 16:14:43)
>>> Kernel is 3.16.2
>>>
>>> What do you think ?
>>>
>>> Regards angelo
>>>
>> Yes, this seems to be mis-alignment in ecc-scheme (or layout)
>>
>> + rogerq at ti.com
>>
>> Also following User-Guides for *OMAP* processors (not davinci)
>> may help you narrow down the cause.
>>
>>
>> with regards, pekon
>>
>> [1]
>> http://processors.wiki.ti.com/index.php/Linux_Core_NAND_User%27s_Guide#ECC_schemes_support
>>
>> [2] http://processors.wiki.ti.com/index.php/UBIFS_Support
>>
>>
>> ------------------------
>> Powered by BigRock.com
>>
>

------------------------
Powered by BigRock.com