On Thu, 2018-07-26 at 08:49 +0200, Miquel Raynal wrote:

Hi xiaolei,

xiaolei li [off-list ref] wrote on Thu, 26 Jul 2018 14:46:29
+0800:

quoted

On Thu, 2018-07-26 at 08:14 +0200, Boris Brezillon wrote:

quoted

On Thu, 26 Jul 2018 14:06:41 +0800
xiaolei li [off-list ref] wrote:
  

quoted

On Sat, 2018-07-21 at 19:10 +0200, Boris Brezillon wrote:  

quoted

On Fri, 20 Jul 2018 17:15:06 +0200
Miquel Raynal [off-list ref] wrote:
    

quoted

Two helpers have been added to the core to make ECC-related
configuration between the detection phase and the final NAND scan. Use
these hooks and convert the driver to just use nand_scan() instead of
both nand_scan_ident() and nand_scan_tail().

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/mtk_nand.c | 75 ++++++++++++++++++++++++-----------------
 1 file changed, 44 insertions(+), 31 deletions(-)

diff --git a/drivers/mtd/nand/raw/mtk_nand.c b/drivers/mtd/nand/raw/mtk_nand.c
index 7bc6be3f6ec0..967418f945ea 100644
--- a/drivers/mtd/nand/raw/mtk_nand.c
+++ b/drivers/mtd/nand/raw/mtk_nand.c

@@ -1250,13 +1250,54 @@ static int mtk_nfc_ecc_init(struct device *dev, struct mtd_info *mtd)
 	return 0;
 }
 
+static int mtk_nfc_attach_chip(struct nand_chip *chip)
+{
+	struct mtd_info *mtd = nand_to_mtd(chip);
+	struct device *dev = mtd->dev.parent;
+	struct mtk_nfc *nfc = nand_get_controller_data(chip);
+	struct mtk_nfc_nand_chip *mtk_nand = to_mtk_nand(chip);
+	int len;
+	int ret;
+
+	if (chip->options & NAND_BUSWIDTH_16) {
+		dev_err(dev, "16bits buswidth not supported");
+		return -EINVAL;
+	}
+
+	/* store bbt magic in page, cause OOB is not protected */
+	if (chip->bbt_options & NAND_BBT_USE_FLASH)
+		chip->bbt_options |= NAND_BBT_NO_OOB;
+
+	ret = mtk_nfc_ecc_init(dev, mtd);
+	if (ret)
+		return ret;
+
+	ret = mtk_nfc_set_spare_per_sector(&mtk_nand->spare_per_sector, mtd);
+	if (ret)
+		return ret;
+
+	mtk_nfc_set_fdm(&mtk_nand->fdm, mtd);
+	mtk_nfc_set_bad_mark_ctl(&mtk_nand->bad_mark, mtd);
+
+	len = mtd->writesize + mtd->oobsize;
+	nfc->buffer = devm_kzalloc(dev, len, GFP_KERNEL);
+	if (!nfc->buffer)
+		return  -ENOMEM;
+
+	return 0;
+}
+
+static const struct nand_controller_ops mtk_nfc_controller_ops = {
+	.attach_chip = mtk_nfc_attach_chip,
+};
+
 static int mtk_nfc_nand_chip_init(struct device *dev, struct mtk_nfc *nfc,
 				  struct device_node *np)
 {
 	struct mtk_nfc_nand_chip *chip;
 	struct nand_chip *nand;
 	struct mtd_info *mtd;
-	int nsels, len;
+	int nsels;
 	u32 tmp;
 	int ret;
 	int i;

@@ -1287,6 +1328,7 @@ static int mtk_nfc_nand_chip_init(struct device *dev, struct mtk_nfc *nfc,
 
 	nand = &chip->nand;
 	nand->controller = &nfc->controller;
+	nand->controller->ops = &mtk_nfc_controller_ops;    

Just like for the marvell driver, this assignment should be moved here
[1].     

Agree to this.
 

quoted

quoted

Also, it looks like this driver is open-coding
nand_controller_init(), probably something we should fix (in a separate
patch).    

May I ask if you mean driver should use nand_hw_control_init helper to
do controller_init?  

Well, now it's named nand_controller_init(), but yes, that's what I
meant.  

OK. I will fix it base on this patch series later.

Nice!

Then you can just wait for the next -rc1, this series should be
available.

OK. Thanks for your improvement!

Assuming ->controller->ops assignment will be fixed,

Acked-by: Xiaolei Li <redacted>

Thanks,
Miqu?l