[PATCH v3 3/3] mtd: cfi_cmdset_0002: Use chip_ready() for write on S29GL064N

[Miquel Raynal]

Hi Tokunori,

ikegami.t at gmail.com wrote on Wed, 16 Mar 2022 01:56:07 +0900:

> As pointed out by this bug report [1], the buffered write is now broken on

                                       , buffered writes are now broken

> S29GL064N. The reason is that changed the buffered write to use chip_good
> instead of chip_ready.

"This issue comes from a rework which switched from using chip_good()
to chip_ready(), because <explain the difference here>."

[please note I am just trying to understand what the root cause is,
please rephrase if I'm wrong].

> One way to solve the issue is to revert the change
> partially to use chip_ready for S29GL064N since the way of least surprise.

s/since the way of least surprise//


> 
> [1] https://lore.kernel.org/r/b687c259-6413-26c9-d4c9-b3afa69ea124@pengutronix.de/
> 
> Fixes: dfeae1073583("mtd: cfi_cmdset_0002: Change write buffer to check correct value")
> Signed-off-by: Tokunori Ikegami <ikegami.t at gmail.com>
> Tested-by: Ahmad Fatoum <a.fatoum at pengutronix.de>
> Cc: Miquel Raynal <miquel.raynal at bootlin.com>
> Cc: Richard Weinberger <richard at nod.at>
> Cc: Vignesh Raghavendra <vigneshr at ti.com>
> Cc: linux-mtd at lists.infradead.org

I think you can get rid of all the above Cc: tags and just copy all 3
of us + the mailing list when sending your v4.

> Cc: stable at vger.kernel.org
> ---

Please also include a Fixes/stable tag in the patch before (2/3) to explain
that both patches are required in order to fix the issue and the current patch alone won't apply.

You should mention that with a nice comment below the three dashes ("---") in patch 2/3 as well.

>  drivers/mtd/chips/cfi_cmdset_0002.c | 10 ++++++++++
>  1 file changed, 10 insertions(+)
> 
> diff --git a/drivers/mtd/chips/cfi_cmdset_0002.c b/drivers/mtd/chips/cfi_cmdset_0002.c
> index 8f3f0309dc03..fa11db066c99 100644
> --- a/drivers/mtd/chips/cfi_cmdset_0002.c
> +++ b/drivers/mtd/chips/cfi_cmdset_0002.c
> @@ -867,10 +867,20 @@ static int __xipram chip_good(struct map_info *map, struct flchip *chip,
>  	return chip_check(map, chip, addr, &expected);
>  }
>  
> +static bool __xipram cfi_use_chip_ready_for_write(struct map_info *map)
> +{
> +	struct cfi_private *cfi = map->fldrv_priv;
> +
> +	return cfi->mfr == CFI_MFR_AMD && cfi->id == S29GL064N_MN12;
> +}
> +
>  static int __xipram chip_good_for_write(struct map_info *map,
>  					struct flchip *chip, unsigned long addr,
>  					map_word expected)
>  {
> +	if (cfi_use_chip_ready_for_write(map))
> +		return chip_ready(map, chip, addr);
> +
>  	return chip_good(map, chip, addr, expected);
>  }
>  

This is much more understandable.

Vignesh, perhaps it would be better to provide a way for manufacturers
to overload certain callbacks instead of applying quirks like this in
the code. But that will come in a second time of course.


Thanks,
Miquèl