Re: mtd_oobtest fails with GPMI-NAND

[Huang Shijie <b32955@freescale.com>]

于 2013年01月29日 01:04, Vikram Narayanan 写道:
> On 1/28/2013 8:50 AM, Huang Shijie wrote:
>> 于 2013年01月28日 10:39, Vikram Narayanan 写道:
>>> Hello Huang,
>>>
>>> On 1/21/2013 7:42 AM, Huang Shijie wrote:
>>>> 于 2013年01月19日 00:52, Vikram Narayanan 写道:
>>>>> Hi,
>>>>>
>>>>> When I try to run the mtd_oobtest on an i.Mx6Q board, it results in a
Which mx6q' board are you using? the mx6q-arm2 or mx6q-ard?

The kernel only supports the mx6q-arm2 now.

>>>>> failure with the following error.
>>>>> I'm using 3.5.7 Kernel.
>>>>>
>>>>> root@freescale:/# insmod mtd_oobtest.ko dev=6
>>>>> [ 7534.508880]
>>>>> [ 7534.511228] =================================================
>>>>> [ 7534.518460] mtd_oobtest: MTD device: 6
>>>>> [ 7534.523000] mtd_oobtest: MTD device size 304087040, eraseblock 
>>>>> size
>>>>> 262144, page size 4096, count of eraseblocks 1160, pages per
>>>>> eraseblock 64, OOB size 128
>>>>> [ 7534.539301] mtd_oobtest: scanning for bad eraseblocks
>>>>> [ 7534.545345] mtd_oobtest: scanned 1160 eraseblocks, 0 are bad
>>>>> [ 7534.552482] mtd_oobtest: test 1 of 5
>>>>> [ 7534.556797] mtd_oobtest: erasing whole device
>>>>> [ 7537.523540] mtd_oobtest: erased 1160 eraseblocks
>>>>> [ 7537.528888] mtd_oobtest: writing OOBs of whole device
>>>>> [ 7537.534704] mtd_oobtest: error: writeoob failed at 0x0
>>>>> [ 7537.540562] mtd_oobtest: error: use_len 0, use_offset 0
>>>>> [ 7537.546565] mtd_oobtest: error -22 occurred
>>>>> [ 7537.551456] =================================================
>>>>>
>>>>> This boils down to the fake "struct nand_ecclayout" defined in
>>>>> <drivers/mtd/nand/gpmi-nand/gpmi-nand.c>
>>>>>
>>>>> Is there a way to run this test successfully?
>>>> The gpmi-nand may use all the oob. So the oobtest may fails.
>>>
>>> I'm in receipt of the error mentioned in [1]. The FAQ also suggests to
>>> run mtd_tests. mtd_oobtest might give more information on whether or
>>> not the NAND driver is buggy.
>>>
>>> Since I couldn't run this test on gpmi-nand due to the driver design,
>>> any ideas on how do I resolve the "ubi_io_read: error -74 (ECC error)"
>>> while mounting my UBIFS?
>>>
>> [1] what's the kernel's version?
>
> The above log is from 3.5.7, but the results are same for the latest 
> kernel (3.8-rc) too.
>
>> [2] what's type of the nand chips?
>>      could you show me the nand chip's geometry?
>
> Toshiba, 4 Gbit, (4096 +224) bytes ×64 pages ×2048blocks

The Toshiba's nand is not well supported by the kernel.
Are you sure the kernel parse out the correct geometry for the toshiba's 
nand?
Please recheck the page size and oob size with "mtdinfo /dev/mtd0".

thanks
Huang Shijie
> writesize - 4KiB
> oobsize - 224 bytes
>
> Regards,
> Vikram
>