When new timing values are calculated in marvell_nfc_setup_interface()
ensure that they will be applied in marvell_nfc_select_target() by
clearing the selected_chip pointer.

Fixes: b25251414f6e ("mtd: rawnand: marvell: Stop implementing ->select_chip()")
Suggested-by: Miquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz>
---

Notes:
    This at least gets me to a point where I can illustrated the problem
    reported to me. It appears that despite the chip correctly reporting
    support for SDR timing modes up to 4 the observed tWC is 20ns. I've not
    seen any actual problem running in this state the only complaint is that
    the datasheet says the minimum tWC is 25ns.
    
    If I make a change to my bootloader such that the NAND Clock Frequency
    Select bit (0xF2440700:0) to 1 before booting the kernel _and_ I remove
    the extra factor of 2 from the period_ns calculation I observe tWC of
    about 60ns. If I don't remove the factor of 2 the NAND interface doesn't
    work (can't write BBT).
    
    Changes in v2:
    - reword comment per suggestion from Miquel, add fixes tag

 drivers/mtd/nand/raw/marvell_nand.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/mtd/nand/raw/marvell_nand.c b/drivers/mtd/nand/raw/marvell_nand.c
index afb424579f0b..f1fcf136ad03 100644
--- a/drivers/mtd/nand/raw/marvell_nand.c
+++ b/drivers/mtd/nand/raw/marvell_nand.c
@@ -2457,6 +2457,12 @@ static int marvell_nfc_setup_interface(struct nand_chip *chip, int chipnr,
 			NDTR1_WAIT_MODE;
 	}
 
+	/*
+	 * Reset nfc->selected_chip so the next command will cause the timing
+	 * registers to be updated in marvell_nfc_select_target().
+	 */
+	nfc->selected_chip = NULL;
+
 	return 0;
 }
 
-- 
2.40.1

On 24/05/23 12:05, Chris Packham wrote:
> When new timing values are calculated in marvell_nfc_setup_interface()
> ensure that they will be applied in marvell_nfc_select_target() by
> clearing the selected_chip pointer.
>
> Fixes: b25251414f6e ("mtd: rawnand: marvell: Stop implementing ->select_chip()")
> Suggested-by: Miquel Raynal <miquel.raynal@bootlin.com>
> Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz>
> ---
>
> Notes:
>      This at least gets me to a point where I can illustrated the problem
>      reported to me. It appears that despite the chip correctly reporting
>      support for SDR timing modes up to 4 the observed tWC is 20ns. I've not
>      seen any actual problem running in this state the only complaint is that
>      the datasheet says the minimum tWC is 25ns.
>      
>      If I make a change to my bootloader such that the NAND Clock Frequency
>      Select bit (0xF2440700:0) to 1 before booting the kernel _and_ I remove
>      the extra factor of 2 from the period_ns calculation I observe tWC of
>      about 60ns. If I don't remove the factor of 2 the NAND interface doesn't
>      work (can't write BBT).
>      
>      Changes in v2:
>      - reword comment per suggestion from Miquel, add fixes tag

Actually thinking about this I should probably also include a change to 
drop the unconditional NAND Clock Frequency select. Some marvell_nand 
users probably aren't getting the best timings without this change but 
those with a system controller will probably end up with timings faster 
than intended (which may work depending on design tolerances).

>
>   drivers/mtd/nand/raw/marvell_nand.c | 6 ++++++
>   1 file changed, 6 insertions(+)
>
> diff --git a/drivers/mtd/nand/raw/marvell_nand.c b/drivers/mtd/nand/raw/marvell_nand.c
> index afb424579f0b..f1fcf136ad03 100644
> --- a/drivers/mtd/nand/raw/marvell_nand.c
> +++ b/drivers/mtd/nand/raw/marvell_nand.c
> @@ -2457,6 +2457,12 @@ static int marvell_nfc_setup_interface(struct nand_chip *chip, int chipnr,
>   			NDTR1_WAIT_MODE;
>   	}
>   
> +	/*
> +	 * Reset nfc->selected_chip so the next command will cause the timing
> +	 * registers to be updated in marvell_nfc_select_target().
> +	 */
> +	nfc->selected_chip = NULL;
> +
>   	return 0;
>   }
>