Hi Cyrille,
On Tue, Jan 24, 2017 at 11:11 AM, Cyrille Pitchen
<cyrille.pitchen@xxxxxxxxx> wrote:
> Le 24/01/2017 à 08:43, Geert Uytterhoeven a écrit :
>> CC linux-mtd
>>
>> On Tue, Jan 24, 2017 at 7:34 AM, DongCV <cv-dong@xxxxxxxxxxx> wrote:
>>> I've tested the linux v4.10-rc2 for Gen2 Lager and found an issues when
>>> writing data to SPI-FLASH with jffs2 format
>>>
>>> "root@linaro-naro:~# cp test.dat /mnt/media
>>> [ 223.874078] ------------[ cut here ]------------
>>> [ 223.888017] WARNING: CPU: 2 PID: 2008 at
>>> drivers/mtd/spi-nor/spi-nor.c:1182 spi_nor_write+0x84/0x160
>>> [ 223.915507] Writing at offset 12 into a NOR page. Writing partial pages
>>> may decrease reliability and increase wear of NOR flash.
>>> [ 223.915543] CPU: 2 PID: 2008 Comm: cp Not tainted 4.10.0-rc2-dirty #166
>>> [ 223.970037] Hardware name: Generic R8A7790 (Flattened Device Tree)
>>> [ 223.988563] Backtrace:
>>> [ 223.995934] [<c010a8a0>] (dump_backtrace) from [<c010aad8>]
>>> (show_stack+0x18/0x1c)
>>> [ 224.018639] r6:c04786f0 r5:00000000 r4:600f0013 r3:00404000
>>> [ 224.035626] [<c010aac0>] (show_stack) from [<c033b154>]
>>> (dump_stack+0x80/0xa0)
>>> [ 224.057294] [<c033b0d4>] (dump_stack) from [<c011e49c>]
>>> (__warn+0xd4/0x104)
>>> [ 224.078169] r5:00000000 r4:ef203ba8
>>> [ 224.088894] [<c011e3c8>] (__warn) from [<c011e50c>]
>>> (warn_slowpath_fmt+0x40/0x48)
>>> [ 224.111337] r9:c0a2a6cb r8:ef203c8c r7:0000000c r6:00000000 r5:00000000
>>> r4:eebfa418
>>> [ 224.134564] [<c011e4d0>] (warn_slowpath_fmt) from [<c04786f0>]
>>> (spi_nor_write+0x84/0x160)
>>> ..."
>>>
>>> Please have a look at the attached testlog file for detail.
>>
>> I didn't keep the attachment, the test just does
>>
>> # flash_erase -j /dev/mtd2 0 0
>> # mount -t jffs2 /dev/mtdblock2 /mnt/media
>> # dd if=/dev/mtdblock2 of=test.dat bs=512 count=1
>> # cp test.dat /mnt/media
>>
>>> And I found the patch which I think it is the cause of this issues.
>>> The commit name is 'e5d05cb mtd: spi-nor: simplify write loop'
>>> If I reverted this patch and retested it on H2 board, the issue will be
>>> gone!
>
> This is a known issue fixed by the patch "mtd: spi-nor: remove WARN_ONCE()
> message in spi_nor_write()", which is planned to be mainlined in 4.11.
>
> This patch only removes the WARN_ONCE() introduced by commit e5d05cb ("mtd:
> spi-nor: simplify write loop"). All other modifications of commit e5d05cb
> are fine and can be kept. It's only the WARN_ONCE() message that is
> annoying and not so relevant.
>
> So don't worry about this (false) warning and sorry for the noise!
Thanks for explaining!
As this is a user-visible warning, isn't that patch suitable for v4.10?
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
