[PATCH 00/19] Ensure good ECC settings

[PATCH 00/19] Ensure good ECC settings

[Miquel Raynal]

Hello,

As part of a previous work (~70 patches), the DT parsing operation has
been moved in the core in order to be generic to all NAND devices when
applicable. Depending on the organization of the driver, this could
have an impact, or not.

The problem here is (mainly) the way the ECC provider is
selected. Many host controller drivers simply set a default, others
only support a specific value. Both used the same code for these two
purposes.

This series should have been merged before the previously mentioned
series, but I simply overlooked that aspect and now I get various bug
reports, all related. So I plan to merge this series as part of a
fixes PR to be sure they get merged for v5.10.

The idea in all the following patches is to move the ECC
initialization in the ->attach_chip() hook, which gets called during
nand_scan(), after NAND discovery. As this stage, we know the NAND
chip memory layout, features and constraints. It is the moment when
the core requests the NAND controller driver to finalize its
initialization (ie. the ECC bits).

When the controller does not embed a hardware ECC engine, software
engines are used, we usually don't have the choice, so we keep
enforcing the only values supported by the controller. Otherwise, we
make a distinction between the choice of the hardware ECC engine or
any other engine.

Regards,
Miquèl


Miquel Raynal (19):
  mtd: rawnand: gpio: Move the ECC initialization to ->attach_chip()
  mtd: rawnand: xway: Move the ECC initialization to ->attach_chip()
  mtd: rawnand: ams-delta: Move the ECC initialization to
    ->attach_chip()
  mtd: rawnand: au1550: Move the ECC initialization to ->attach_chip()
  mtd: rawnand: cs553x: Move the ECC initialization to ->attach_chip()
  mtd: rawnand: davinci: Move the ECC initialization to ->attach_chip()
  mtd: rawnand: diskonchip: Move the ECC initialization to
    ->attach_chip()
  mtd: rawnand: fsmc: Move the ECC initialization to ->attach_chip()
  mtd: rawnand: lpc32xx_mlc: Move the ECC initialization to
    ->attach_chip()
  mtd: rawnand: lpc32xx_slc: Move the ECC initialization to
    ->attach_chip()
  mtd: rawnand: mpc5121: Move the ECC initialization to ->attach_chip()
  mtd: rawnand: orion: Move the ECC initialization to ->attach_chip()
  mtd: rawnand: txx9ndfmc: Move the ECC initialization to
    ->attach_chip()
  mtd: rawnand: tmio: Move the ECC initialization to ->attach_chip()
  mtd: rawnand: pasemi: Move the ECC initialization to ->attach_chip()
  mtd: rawnand: plat_nand: Move the ECC initialization to
    ->attach_chip()
  mtd: rawnand: r852: Move the ECC initialization to ->attach_chip()
  mtd: rawnand: sharpsl: Move the ECC initialization to ->attach_chip()
  mtd: rawnand: socrates: Move the ECC initialization to ->attach_chip()

 drivers/mtd/nand/raw/ams-delta.c     | 12 ++++++---
 drivers/mtd/nand/raw/au1550nd.c      | 11 ++++++--
 drivers/mtd/nand/raw/cs553x_nand.c   | 24 +++++++++++------
 drivers/mtd/nand/raw/davinci_nand.c  |  8 +++---
 drivers/mtd/nand/raw/diskonchip.c    | 29 +++++++++++++-------
 drivers/mtd/nand/raw/fsmc_nand.c     | 30 ++++++++++-----------
 drivers/mtd/nand/raw/gpio.c          | 11 ++++++--
 drivers/mtd/nand/raw/lpc32xx_mlc.c   | 23 +++++++++-------
 drivers/mtd/nand/raw/lpc32xx_slc.c   | 26 +++++++++---------
 drivers/mtd/nand/raw/mpc5121_nfc.c   | 19 +++++++++++--
 drivers/mtd/nand/raw/orion_nand.c    | 19 +++++++++++--
 drivers/mtd/nand/raw/pasemi_nand.c   | 19 +++++++++++--
 drivers/mtd/nand/raw/plat_nand.c     | 20 +++++++++++---
 drivers/mtd/nand/raw/r852.c          | 40 +++++++++++++++++++---------
 drivers/mtd/nand/raw/r852.h          |  1 +
 drivers/mtd/nand/raw/sharpsl.c       | 32 ++++++++++++++++------
 drivers/mtd/nand/raw/socrates_nand.c | 21 ++++++++++++---
 drivers/mtd/nand/raw/tmio_nand.c     | 33 ++++++++++++++++-------
 drivers/mtd/nand/raw/txx9ndfmc.c     | 14 ++++++----
 drivers/mtd/nand/raw/xway_nand.c     | 18 +++++++++++--
 20 files changed, 294 insertions(+), 116 deletions(-)

-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 01/19] mtd: rawnand: gpio: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip(), a NAND controller
hook.

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Reported-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Tested-by: Christophe Leroy <christophe.leroy@csgroup.eu>
---
 drivers/mtd/nand/raw/gpio.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/drivers/mtd/nand/raw/gpio.c b/drivers/mtd/nand/raw/gpio.c
index 4ec0a1e10867..eb03b8cea1cb 100644
--- a/drivers/mtd/nand/raw/gpio.c
+++ b/drivers/mtd/nand/raw/gpio.c
@@ -161,8 +161,17 @@ static int gpio_nand_exec_op(struct nand_chip *chip,
 	return ret;
 }
 
+static int gpio_nand_attach_chip(struct nand_chip *chip)
+{
+	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
+	chip->ecc.algo = NAND_ECC_ALGO_HAMMING;
+
+	return 0;
+}
+
 static const struct nand_controller_ops gpio_nand_ops = {
 	.exec_op = gpio_nand_exec_op,
+	.attach_chip = gpio_nand_attach_chip,
 };
 
 #ifdef CONFIG_OF
@@ -342,8 +351,6 @@ static int gpio_nand_probe(struct platform_device *pdev)
 	gpiomtd->base.ops = &gpio_nand_ops;
 
 	nand_set_flash_node(chip, pdev->dev.of_node);
-	chip->ecc.engine_type	= NAND_ECC_ENGINE_TYPE_SOFT;
-	chip->ecc.algo		= NAND_ECC_ALGO_HAMMING;
 	chip->options		= gpiomtd->plat.options;
 	chip->controller	= &gpiomtd->base;
 
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 01/19] mtd: rawnand: gpio: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:06 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip(), a NAND controller
> hook.
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Reported-by: Christophe Leroy <christophe.leroy@csgroup.eu>
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
> Tested-by: Christophe Leroy <christophe.leroy@csgroup.eu>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 02/19] mtd: rawnand: xway: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip(), a NAND controller
hook.

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/xway_nand.c | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/drivers/mtd/nand/raw/xway_nand.c b/drivers/mtd/nand/raw/xway_nand.c
index f2dbd63a5c1f..efc5bf5434e0 100644
--- a/drivers/mtd/nand/raw/xway_nand.c
+++ b/drivers/mtd/nand/raw/xway_nand.c
@@ -62,6 +62,7 @@
 #define NAND_CON_NANDM		1
 
 struct xway_nand_data {
+	struct nand_controller	controller;
 	struct nand_chip	chip;
 	unsigned long		csflags;
 	void __iomem		*nandaddr;
@@ -145,6 +146,18 @@ static void xway_write_buf(struct nand_chip *chip, const u_char *buf, int len)
 		xway_writeb(nand_to_mtd(chip), NAND_WRITE_DATA, buf[i]);
 }
 
+static int xway_attach_chip(struct nand_chip *chip)
+{
+	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
+	chip->ecc.algo = NAND_ECC_ALGO_HAMMING;
+
+	return 0;
+}
+
+static const struct nand_controller_ops xway_nand_ops = {
+	.attach_chip = xway_attach_chip,
+};
+
 /*
  * Probe for the NAND device.
  */
@@ -180,8 +193,9 @@ static int xway_nand_probe(struct platform_device *pdev)
 	data->chip.legacy.read_byte = xway_read_byte;
 	data->chip.legacy.chip_delay = 30;
 
-	data->chip.ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
-	data->chip.ecc.algo = NAND_ECC_ALGO_HAMMING;
+	nand_controller_init(&data->controller);
+	data->controller.ops = &xway_nand_ops;
+	data->chip.controller = &data->controller;
 
 	platform_set_drvdata(pdev, data);
 	nand_set_controller_data(&data->chip, data);
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 02/19] mtd: rawnand: xway: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:07 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip(), a NAND controller
> hook.
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 03/19] mtd: rawnand: ams-delta: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/ams-delta.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/drivers/mtd/nand/raw/ams-delta.c b/drivers/mtd/nand/raw/ams-delta.c
index d3c5cc513c8f..0c352b39ad4b 100644
--- a/drivers/mtd/nand/raw/ams-delta.c
+++ b/drivers/mtd/nand/raw/ams-delta.c
@@ -215,8 +215,17 @@ static int gpio_nand_setup_interface(struct nand_chip *this, int csline,
 	return 0;
 }
 
+static int gpio_nand_attach_chip(struct nand_chip *chip)
+{
+	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
+	chip->ecc.algo = NAND_ECC_ALGO_HAMMING;
+
+	return 0;
+}
+
 static const struct nand_controller_ops gpio_nand_ops = {
 	.exec_op = gpio_nand_exec_op,
+	.attach_chip = gpio_nand_attach_chip,
 	.setup_interface = gpio_nand_setup_interface,
 };
 
@@ -260,9 +269,6 @@ static int gpio_nand_probe(struct platform_device *pdev)
 		return err;
 	}
 
-	this->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
-	this->ecc.algo = NAND_ECC_ALGO_HAMMING;
-
 	platform_set_drvdata(pdev, priv);
 
 	/* Set chip enabled but write protected */
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 03/19] mtd: rawnand: ams-delta: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:08 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 04/19] mtd: rawnand: au1550: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/au1550nd.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/drivers/mtd/nand/raw/au1550nd.c b/drivers/mtd/nand/raw/au1550nd.c
index 79b057400fe9..7892022bd6dd 100644
--- a/drivers/mtd/nand/raw/au1550nd.c
+++ b/drivers/mtd/nand/raw/au1550nd.c
@@ -236,8 +236,17 @@ static int au1550nd_exec_op(struct nand_chip *this,
 	return ret;
 }
 
+static int au1550nd_attach_chip(struct nand_chip *chip)
+{
+	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
+	chip->ecc.algo = NAND_ECC_ALGO_HAMMING;
+
+	return 0;
+}
+
 static const struct nand_controller_ops au1550nd_ops = {
 	.exec_op = au1550nd_exec_op,
+	.attach_chip = au1550nd_attach_chip,
 };
 
 static int au1550nd_probe(struct platform_device *pdev)
@@ -294,8 +303,6 @@ static int au1550nd_probe(struct platform_device *pdev)
 	nand_controller_init(&ctx->controller);
 	ctx->controller.ops = &au1550nd_ops;
 	this->controller = &ctx->controller;
-	this->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
-	this->ecc.algo = NAND_ECC_ALGO_HAMMING;
 
 	if (pd->devwidth)
 		this->options |= NAND_BUSWIDTH_16;
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 04/19] mtd: rawnand: au1550: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:09 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 05/19] mtd: rawnand: cs553x: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/cs553x_nand.c | 24 ++++++++++++++++--------
 1 file changed, 16 insertions(+), 8 deletions(-)

diff --git a/drivers/mtd/nand/raw/cs553x_nand.c b/drivers/mtd/nand/raw/cs553x_nand.c
index b066dcae31ab..6edf78c16fc8 100644
--- a/drivers/mtd/nand/raw/cs553x_nand.c
+++ b/drivers/mtd/nand/raw/cs553x_nand.c
@@ -242,8 +242,24 @@ static int cs_calculate_ecc(struct nand_chip *this, const u_char *dat,
 
 static struct cs553x_nand_controller *controllers[4];
 
+static int cs553x_attach_chip(struct nand_chip *chip)
+{
+	if (chip->ecc.engine_type != NAND_ECC_ENGINE_TYPE_ON_HOST)
+		return 0;
+
+	chip->ecc.size = 256;
+	chip->ecc.bytes = 3;
+	chip->ecc.hwctl  = cs_enable_hwecc;
+	chip->ecc.calculate = cs_calculate_ecc;
+	chip->ecc.correct  = rawnand_sw_hamming_correct;
+	chip->ecc.strength = 1;
+
+	return 0;
+}
+
 static const struct nand_controller_ops cs553x_nand_controller_ops = {
 	.exec_op = cs553x_exec_op,
+	.attach_chip = cs553x_attach_chip,
 };
 
 static int __init cs553x_init_one(int cs, int mmio, unsigned long adr)
@@ -285,14 +301,6 @@ static int __init cs553x_init_one(int cs, int mmio, unsigned long adr)
 		goto out_mtd;
 	}
 
-	this->ecc.engine_type = NAND_ECC_ENGINE_TYPE_ON_HOST;
-	this->ecc.size = 256;
-	this->ecc.bytes = 3;
-	this->ecc.hwctl  = cs_enable_hwecc;
-	this->ecc.calculate = cs_calculate_ecc;
-	this->ecc.correct  = rawnand_sw_hamming_correct;
-	this->ecc.strength = 1;
-
 	/* Enable the following for a flash based bad block table */
 	this->bbt_options = NAND_BBT_USE_FLASH;
 
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 06/19] mtd: rawnand: davinci: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Cc: Bartosz Golaszewski <bgolaszewski@baylibre.com>
---
 drivers/mtd/nand/raw/davinci_nand.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/mtd/nand/raw/davinci_nand.c b/drivers/mtd/nand/raw/davinci_nand.c
index 427f320fb79b..f8c36d19ab47 100644
--- a/drivers/mtd/nand/raw/davinci_nand.c
+++ b/drivers/mtd/nand/raw/davinci_nand.c
@@ -585,6 +585,10 @@ static int davinci_nand_attach_chip(struct nand_chip *chip)
 	if (IS_ERR(pdata))
 		return PTR_ERR(pdata);
 
+	/* Use board-specific ECC config */
+	info->chip.ecc.engine_type = pdata->engine_type;
+	info->chip.ecc.placement = pdata->ecc_placement;
+
 	switch (info->chip.ecc.engine_type) {
 	case NAND_ECC_ENGINE_TYPE_NONE:
 		pdata->ecc_bits = 0;
@@ -850,10 +854,6 @@ static int nand_davinci_probe(struct platform_device *pdev)
 	info->mask_ale		= pdata->mask_ale ? : MASK_ALE;
 	info->mask_cle		= pdata->mask_cle ? : MASK_CLE;
 
-	/* Use board-specific ECC config */
-	info->chip.ecc.engine_type = pdata->engine_type;
-	info->chip.ecc.placement = pdata->ecc_placement;
-
 	spin_lock_irq(&davinci_nand_lock);
 
 	/* put CSxNAND into NAND mode */
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 06/19] mtd: rawnand: davinci: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:11 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
> Cc: Bartosz Golaszewski <bgolaszewski@baylibre.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 07/19] mtd: rawnand: diskonchip: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/diskonchip.c | 29 +++++++++++++++++++----------
 1 file changed, 19 insertions(+), 10 deletions(-)

diff --git a/drivers/mtd/nand/raw/diskonchip.c b/drivers/mtd/nand/raw/diskonchip.c
index 94432a453e5e..26b265e4384a 100644
--- a/drivers/mtd/nand/raw/diskonchip.c
+++ b/drivers/mtd/nand/raw/diskonchip.c
@@ -1269,12 +1269,31 @@ static inline int __init doc2001plus_init(struct mtd_info *mtd)
 	return 1;
 }
 
+static int doc200x_attach_chip(struct nand_chip *chip)
+{
+	if (chip->ecc.engine_type != NAND_ECC_ENGINE_TYPE_ON_HOST)
+		return 0;
+
+	chip->ecc.placement = NAND_ECC_PLACEMENT_INTERLEAVED;
+	chip->ecc.size = 512;
+	chip->ecc.bytes = 6;
+	chip->ecc.strength = 2;
+	chip->ecc.options = NAND_ECC_GENERIC_ERASED_CHECK;
+	chip->ecc.hwctl = doc200x_enable_hwecc;
+	chip->ecc.calculate = doc200x_calculate_ecc;
+	chip->ecc.correct = doc200x_correct_data;
+
+	return 0;
+}
+
 static const struct nand_controller_ops doc200x_ops = {
 	.exec_op = doc200x_exec_op,
+	.attach_chip = doc200x_attach_chip,
 };
 
 static const struct nand_controller_ops doc2001plus_ops = {
 	.exec_op = doc2001plus_exec_op,
+	.attach_chip = doc200x_attach_chip,
 };
 
 static int __init doc_probe(unsigned long physadr)
@@ -1452,16 +1471,6 @@ static int __init doc_probe(unsigned long physadr)
 
 	nand->controller	= &doc->base;
 	nand_set_controller_data(nand, doc);
-	nand->ecc.hwctl		= doc200x_enable_hwecc;
-	nand->ecc.calculate	= doc200x_calculate_ecc;
-	nand->ecc.correct	= doc200x_correct_data;
-
-	nand->ecc.engine_type	= NAND_ECC_ENGINE_TYPE_ON_HOST;
-	nand->ecc.placement	= NAND_ECC_PLACEMENT_INTERLEAVED;
-	nand->ecc.size		= 512;
-	nand->ecc.bytes		= 6;
-	nand->ecc.strength	= 2;
-	nand->ecc.options	= NAND_ECC_GENERIC_ERASED_CHECK;
 	nand->bbt_options	= NAND_BBT_USE_FLASH;
 	/* Skip the automatic BBT scan so we can run it manually */
 	nand->options		|= NAND_SKIP_BBTSCAN | NAND_NO_BBM_QUIRK;
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 07/19] mtd: rawnand: diskonchip: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:12 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 08/19] mtd: rawnand: fsmc: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/fsmc_nand.c | 30 +++++++++++++++---------------
 1 file changed, 15 insertions(+), 15 deletions(-)

diff --git a/drivers/mtd/nand/raw/fsmc_nand.c b/drivers/mtd/nand/raw/fsmc_nand.c
index d0855d4f5627..0101c0fab50a 100644
--- a/drivers/mtd/nand/raw/fsmc_nand.c
+++ b/drivers/mtd/nand/raw/fsmc_nand.c
@@ -879,6 +879,20 @@ static int fsmc_nand_attach_chip(struct nand_chip *nand)
 	struct mtd_info *mtd = nand_to_mtd(nand);
 	struct fsmc_nand_data *host = nand_to_fsmc(nand);
 
+	if (nand->ecc.engine_type == NAND_ECC_ENGINE_TYPE_INVALID)
+		nand->ecc.engine_type = NAND_ECC_ENGINE_TYPE_ON_HOST;
+
+	if (!nand->ecc.size)
+		nand->ecc.size = 512;
+
+	if (AMBA_REV_BITS(host->pid) >= 8) {
+		nand->ecc.read_page = fsmc_read_page_hwecc;
+		nand->ecc.calculate = fsmc_read_hwecc_ecc4;
+		nand->ecc.correct = fsmc_bch8_correct_data;
+		nand->ecc.bytes = 13;
+		nand->ecc.strength = 8;
+	}
+
 	if (AMBA_REV_BITS(host->pid) >= 8) {
 		switch (mtd->oobsize) {
 		case 16:
@@ -904,6 +918,7 @@ static int fsmc_nand_attach_chip(struct nand_chip *nand)
 		dev_info(host->dev, "Using 1-bit HW ECC scheme\n");
 		nand->ecc.calculate = fsmc_read_hwecc_ecc1;
 		nand->ecc.correct = rawnand_sw_hamming_correct;
+		nand->ecc.hwctl = fsmc_enable_hwecc;
 		nand->ecc.bytes = 3;
 		nand->ecc.strength = 1;
 		nand->ecc.options |= NAND_ECC_SOFT_HAMMING_SM_ORDER;
@@ -1054,13 +1069,6 @@ static int __init fsmc_nand_probe(struct platform_device *pdev)
 
 	mtd->dev.parent = &pdev->dev;
 
-	/*
-	 * Setup default ECC mode. nand_dt_init() called from nand_scan_ident()
-	 * can overwrite this value if the DT provides a different value.
-	 */
-	nand->ecc.engine_type = NAND_ECC_ENGINE_TYPE_ON_HOST;
-	nand->ecc.hwctl = fsmc_enable_hwecc;
-	nand->ecc.size = 512;
 	nand->badblockbits = 7;
 
 	if (host->mode == USE_DMA_ACCESS) {
@@ -1083,14 +1091,6 @@ static int __init fsmc_nand_probe(struct platform_device *pdev)
 		nand->options |= NAND_KEEP_TIMINGS;
 	}
 
-	if (AMBA_REV_BITS(host->pid) >= 8) {
-		nand->ecc.read_page = fsmc_read_page_hwecc;
-		nand->ecc.calculate = fsmc_read_hwecc_ecc4;
-		nand->ecc.correct = fsmc_bch8_correct_data;
-		nand->ecc.bytes = 13;
-		nand->ecc.strength = 8;
-	}
-
 	nand_controller_init(&host->base);
 	host->base.ops = &fsmc_nand_controller_ops;
 	nand->controller = &host->base;
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 09/19] mtd: rawnand: lpc32xx_mlc: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Cc: Vladimir Zapolskiy <vz@mleia.com>
Cc: Sylvain Lemieux <slemieux.tyco@gmail.com>
---
 drivers/mtd/nand/raw/lpc32xx_mlc.c | 23 +++++++++++++----------
 1 file changed, 13 insertions(+), 10 deletions(-)

diff --git a/drivers/mtd/nand/raw/lpc32xx_mlc.c b/drivers/mtd/nand/raw/lpc32xx_mlc.c
index bd854e044898..452ecaf7775a 100644
--- a/drivers/mtd/nand/raw/lpc32xx_mlc.c
+++ b/drivers/mtd/nand/raw/lpc32xx_mlc.c
@@ -647,6 +647,9 @@ static int lpc32xx_nand_attach_chip(struct nand_chip *chip)
 	struct lpc32xx_nand_host *host = nand_get_controller_data(chip);
 	struct device *dev = &host->pdev->dev;
 
+	if (chip->ecc.engine_type != NAND_ECC_ENGINE_TYPE_ON_HOST)
+		return 0;
+
 	host->dma_buf = devm_kzalloc(dev, mtd->writesize, GFP_KERNEL);
 	if (!host->dma_buf)
 		return -ENOMEM;
@@ -655,8 +658,17 @@ static int lpc32xx_nand_attach_chip(struct nand_chip *chip)
 	if (!host->dummy_buf)
 		return -ENOMEM;
 
-	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_ON_HOST;
 	chip->ecc.size = 512;
+	chip->ecc.hwctl = lpc32xx_ecc_enable;
+	chip->ecc.read_page_raw = lpc32xx_read_page;
+	chip->ecc.read_page = lpc32xx_read_page;
+	chip->ecc.write_page_raw = lpc32xx_write_page_lowlevel;
+	chip->ecc.write_page = lpc32xx_write_page_lowlevel;
+	chip->ecc.write_oob = lpc32xx_write_oob;
+	chip->ecc.read_oob = lpc32xx_read_oob;
+	chip->ecc.strength = 4;
+	chip->ecc.bytes = 10;
+
 	mtd_set_ooblayout(mtd, &lpc32xx_ooblayout_ops);
 	host->mlcsubpages = mtd->writesize / 512;
 
@@ -740,15 +752,6 @@ static int lpc32xx_nand_probe(struct platform_device *pdev)
 	platform_set_drvdata(pdev, host);
 
 	/* Initialize function pointers */
-	nand_chip->ecc.hwctl = lpc32xx_ecc_enable;
-	nand_chip->ecc.read_page_raw = lpc32xx_read_page;
-	nand_chip->ecc.read_page = lpc32xx_read_page;
-	nand_chip->ecc.write_page_raw = lpc32xx_write_page_lowlevel;
-	nand_chip->ecc.write_page = lpc32xx_write_page_lowlevel;
-	nand_chip->ecc.write_oob = lpc32xx_write_oob;
-	nand_chip->ecc.read_oob = lpc32xx_read_oob;
-	nand_chip->ecc.strength = 4;
-	nand_chip->ecc.bytes = 10;
 	nand_chip->legacy.waitfunc = lpc32xx_waitfunc;
 
 	nand_chip->options = NAND_NO_SUBPAGE_WRITE;
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 09/19] mtd: rawnand: lpc32xx_mlc: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:14 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
> Cc: Vladimir Zapolskiy <vz@mleia.com>
> Cc: Sylvain Lemieux <slemieux.tyco@gmail.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 10/19] mtd: rawnand: lpc32xx_slc: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Cc: Vladimir Zapolskiy <vz@mleia.com>
Cc: Sylvain Lemieux <slemieux.tyco@gmail.com>
---
 drivers/mtd/nand/raw/lpc32xx_slc.c | 26 ++++++++++++++------------
 1 file changed, 14 insertions(+), 12 deletions(-)

diff --git a/drivers/mtd/nand/raw/lpc32xx_slc.c b/drivers/mtd/nand/raw/lpc32xx_slc.c
index e98ab22f9047..6b7269cfb7d8 100644
--- a/drivers/mtd/nand/raw/lpc32xx_slc.c
+++ b/drivers/mtd/nand/raw/lpc32xx_slc.c
@@ -774,6 +774,9 @@ static int lpc32xx_nand_attach_chip(struct nand_chip *chip)
 	struct mtd_info *mtd = nand_to_mtd(chip);
 	struct lpc32xx_nand_host *host = nand_get_controller_data(chip);
 
+	if (chip->ecc.engine_type != NAND_ECC_ENGINE_TYPE_ON_HOST)
+		return 0;
+
 	/* OOB and ECC CPU and DMA work areas */
 	host->ecc_buf = (uint32_t *)(host->data_buf + LPC32XX_DMA_DATA_SIZE);
 
@@ -785,11 +788,22 @@ static int lpc32xx_nand_attach_chip(struct nand_chip *chip)
 	if (mtd->writesize <= 512)
 		mtd_set_ooblayout(mtd, &lpc32xx_ooblayout_ops);
 
+	chip->ecc.placement = NAND_ECC_PLACEMENT_INTERLEAVED;
 	/* These sizes remain the same regardless of page size */
 	chip->ecc.size = 256;
+	chip->ecc.strength = 1;
 	chip->ecc.bytes = LPC32XX_SLC_DEV_ECC_BYTES;
 	chip->ecc.prepad = 0;
 	chip->ecc.postpad = 0;
+	chip->ecc.read_page_raw = lpc32xx_nand_read_page_raw_syndrome;
+	chip->ecc.read_page = lpc32xx_nand_read_page_syndrome;
+	chip->ecc.write_page_raw = lpc32xx_nand_write_page_raw_syndrome;
+	chip->ecc.write_page = lpc32xx_nand_write_page_syndrome;
+	chip->ecc.write_oob = lpc32xx_nand_write_oob_syndrome;
+	chip->ecc.read_oob = lpc32xx_nand_read_oob_syndrome;
+	chip->ecc.calculate = lpc32xx_nand_ecc_calculate;
+	chip->ecc.correct = rawnand_sw_hamming_correct;
+	chip->ecc.hwctl = lpc32xx_nand_ecc_enable;
 
 	/*
 	 * Use a custom BBT marker setup for small page FLASH that
@@ -880,21 +894,9 @@ static int lpc32xx_nand_probe(struct platform_device *pdev)
 	platform_set_drvdata(pdev, host);
 
 	/* NAND callbacks for LPC32xx SLC hardware */
-	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_ON_HOST;
-	chip->ecc.placement = NAND_ECC_PLACEMENT_INTERLEAVED;
 	chip->legacy.read_byte = lpc32xx_nand_read_byte;
 	chip->legacy.read_buf = lpc32xx_nand_read_buf;
 	chip->legacy.write_buf = lpc32xx_nand_write_buf;
-	chip->ecc.read_page_raw = lpc32xx_nand_read_page_raw_syndrome;
-	chip->ecc.read_page = lpc32xx_nand_read_page_syndrome;
-	chip->ecc.write_page_raw = lpc32xx_nand_write_page_raw_syndrome;
-	chip->ecc.write_page = lpc32xx_nand_write_page_syndrome;
-	chip->ecc.write_oob = lpc32xx_nand_write_oob_syndrome;
-	chip->ecc.read_oob = lpc32xx_nand_read_oob_syndrome;
-	chip->ecc.calculate = lpc32xx_nand_ecc_calculate;
-	chip->ecc.correct = rawnand_sw_hamming_correct;
-	chip->ecc.strength = 1;
-	chip->ecc.hwctl = lpc32xx_nand_ecc_enable;
 
 	/*
 	 * Allocate a large enough buffer for a single huge page plus
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 11/19] mtd: rawnand: mpc5121: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/mpc5121_nfc.c | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/drivers/mtd/nand/raw/mpc5121_nfc.c b/drivers/mtd/nand/raw/mpc5121_nfc.c
index dfd0d3ed5ed0..fb4c0b11689f 100644
--- a/drivers/mtd/nand/raw/mpc5121_nfc.c
+++ b/drivers/mtd/nand/raw/mpc5121_nfc.c
@@ -104,6 +104,7 @@
 #define NFC_TIMEOUT		(HZ / 10)	/* 1/10 s */
 
 struct mpc5121_nfc_prv {
+	struct nand_controller	controller;
 	struct nand_chip	chip;
 	int			irq;
 	void __iomem		*regs;
@@ -602,6 +603,18 @@ static void mpc5121_nfc_free(struct device *dev, struct mtd_info *mtd)
 		iounmap(prv->csreg);
 }
 
+static int mpc5121_nfc_attach_chip(struct nand_chip *chip)
+{
+	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
+	chip->ecc.algo = NAND_ECC_ALGO_HAMMING;
+
+	return 0;
+}
+
+static const struct nand_controller_ops mpc5121_nfc_ops = {
+	.attach_chip = mpc5121_nfc_attach_chip,
+};
+
 static int mpc5121_nfc_probe(struct platform_device *op)
 {
 	struct device_node *dn = op->dev.of_node;
@@ -634,6 +647,10 @@ static int mpc5121_nfc_probe(struct platform_device *op)
 	chip = &prv->chip;
 	mtd = nand_to_mtd(chip);
 
+	nand_controller_init(&prv->controller);
+	prv->controller.ops = &mpc5121_nfc_ops;
+	chip->controller = &prv->controller;
+
 	mtd->dev.parent = dev;
 	nand_set_controller_data(chip, prv);
 	nand_set_flash_node(chip, dn);
@@ -688,8 +705,6 @@ static int mpc5121_nfc_probe(struct platform_device *op)
 	chip->legacy.set_features = nand_get_set_features_notsupp;
 	chip->legacy.get_features = nand_get_set_features_notsupp;
 	chip->bbt_options = NAND_BBT_USE_FLASH;
-	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
-	chip->ecc.algo = NAND_ECC_ALGO_HAMMING;
 
 	/* Support external chip-select logic on ADS5121 board */
 	if (of_machine_is_compatible("fsl,mpc5121ads")) {
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 11/19] mtd: rawnand: mpc5121: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:16 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 12/19] mtd: rawnand: orion: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/orion_nand.c | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/drivers/mtd/nand/raw/orion_nand.c b/drivers/mtd/nand/raw/orion_nand.c
index df9c0f8e4b4e..e3bb65fd3ab2 100644
--- a/drivers/mtd/nand/raw/orion_nand.c
+++ b/drivers/mtd/nand/raw/orion_nand.c
@@ -22,6 +22,7 @@
 #include <linux/platform_data/mtd-orion_nand.h>
 
 struct orion_nand_info {
+	struct nand_controller controller;
 	struct nand_chip chip;
 	struct clk *clk;
 };
@@ -82,6 +83,18 @@ static void orion_nand_read_buf(struct nand_chip *chip, uint8_t *buf, int len)
 		buf[i++] = readb(io_base);
 }
 
+static int orion_nand_attach_chip(struct nand_chip *chip)
+{
+	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
+	chip->ecc.algo = NAND_ECC_ALGO_HAMMING;
+
+	return 0;
+}
+
+static const struct nand_controller_ops orion_nand_ops = {
+	.attach_chip = orion_nand_attach_chip,
+};
+
 static int __init orion_nand_probe(struct platform_device *pdev)
 {
 	struct orion_nand_info *info;
@@ -101,6 +114,10 @@ static int __init orion_nand_probe(struct platform_device *pdev)
 	nc = &info->chip;
 	mtd = nand_to_mtd(nc);
 
+	nand_controller_init(&info->controller);
+	info->controller.ops = &orion_nand_ops;
+	nc->controller = &info->controller;
+
 	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
 	io_base = devm_ioremap_resource(&pdev->dev, res);
 
@@ -139,8 +156,6 @@ static int __init orion_nand_probe(struct platform_device *pdev)
 	nc->legacy.IO_ADDR_R = nc->legacy.IO_ADDR_W = io_base;
 	nc->legacy.cmd_ctrl = orion_nand_cmd_ctrl;
 	nc->legacy.read_buf = orion_nand_read_buf;
-	nc->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
-	nc->ecc.algo = NAND_ECC_ALGO_HAMMING;
 
 	if (board->chip_delay)
 		nc->legacy.chip_delay = board->chip_delay;
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 12/19] mtd: rawnand: orion: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:17 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 13/19] mtd: rawnand: txx9ndfmc: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/txx9ndfmc.c | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)

diff --git a/drivers/mtd/nand/raw/txx9ndfmc.c b/drivers/mtd/nand/raw/txx9ndfmc.c
index 5f3dbff60d47..1a9449e53bf9 100644
--- a/drivers/mtd/nand/raw/txx9ndfmc.c
+++ b/drivers/mtd/nand/raw/txx9ndfmc.c
@@ -252,6 +252,11 @@ static int txx9ndfmc_attach_chip(struct nand_chip *chip)
 {
 	struct mtd_info *mtd = nand_to_mtd(chip);
 
+	if (chip->ecc.engine_type != NAND_ECC_ENGINE_TYPE_ON_HOST)
+		return 0;
+
+	chip->ecc.strength = 1;
+
 	if (mtd->writesize >= 512) {
 		chip->ecc.size = 512;
 		chip->ecc.bytes = 6;
@@ -260,6 +265,10 @@ static int txx9ndfmc_attach_chip(struct nand_chip *chip)
 		chip->ecc.bytes = 3;
 	}
 
+	chip->ecc.calculate = txx9ndfmc_calculate_ecc;
+	chip->ecc.correct = txx9ndfmc_correct_data;
+	chip->ecc.hwctl = txx9ndfmc_enable_hwecc;
+
 	return 0;
 }
 
@@ -325,11 +334,6 @@ static int __init txx9ndfmc_probe(struct platform_device *dev)
 		chip->legacy.write_buf = txx9ndfmc_write_buf;
 		chip->legacy.cmd_ctrl = txx9ndfmc_cmd_ctrl;
 		chip->legacy.dev_ready = txx9ndfmc_dev_ready;
-		chip->ecc.calculate = txx9ndfmc_calculate_ecc;
-		chip->ecc.correct = txx9ndfmc_correct_data;
-		chip->ecc.hwctl = txx9ndfmc_enable_hwecc;
-		chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_ON_HOST;
-		chip->ecc.strength = 1;
 		chip->legacy.chip_delay = 100;
 		chip->controller = &drvdata->controller;
 
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 13/19] mtd: rawnand: txx9ndfmc: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:18 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 14/19] mtd: rawnand: tmio: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/tmio_nand.c | 33 +++++++++++++++++++++++---------
 1 file changed, 24 insertions(+), 9 deletions(-)

diff --git a/drivers/mtd/nand/raw/tmio_nand.c b/drivers/mtd/nand/raw/tmio_nand.c
index 31e5e50e77e2..de8e919d0ebe 100644
--- a/drivers/mtd/nand/raw/tmio_nand.c
+++ b/drivers/mtd/nand/raw/tmio_nand.c
@@ -102,6 +102,7 @@
 /*--------------------------------------------------------------------------*/
 
 struct tmio_nand {
+	struct nand_controller controller;
 	struct nand_chip chip;
 	struct completion comp;
 
@@ -354,6 +355,25 @@ static void tmio_hw_stop(struct platform_device *dev, struct tmio_nand *tmio)
 		cell->disable(dev);
 }
 
+static int tmio_attach_chip(struct nand_chip *chip)
+{
+	if (chip->ecc.engine_type != NAND_ECC_ENGINE_TYPE_ON_HOST)
+		return 0;
+
+	chip->ecc.size = 512;
+	chip->ecc.bytes = 6;
+	chip->ecc.strength = 2;
+	chip->ecc.hwctl = tmio_nand_enable_hwecc;
+	chip->ecc.calculate = tmio_nand_calculate_ecc;
+	chip->ecc.correct = tmio_nand_correct_data;
+
+	return 0;
+}
+
+static const struct nand_controller_ops tmio_ops = {
+	.attach_chip = tmio_attach_chip,
+};
+
 static int tmio_probe(struct platform_device *dev)
 {
 	struct tmio_nand_data *data = dev_get_platdata(&dev->dev);
@@ -384,6 +404,10 @@ static int tmio_probe(struct platform_device *dev)
 	mtd->name = "tmio-nand";
 	mtd->dev.parent = &dev->dev;
 
+	nand_controller_init(&tmio->controller);
+	tmio->controller.ops = &tmio_ops;
+	nand_chip->controller = &tmio->controller;
+
 	tmio->ccr = devm_ioremap(&dev->dev, ccr->start, resource_size(ccr));
 	if (!tmio->ccr)
 		return -EIO;
@@ -408,15 +432,6 @@ static int tmio_probe(struct platform_device *dev)
 	nand_chip->legacy.write_buf = tmio_nand_write_buf;
 	nand_chip->legacy.read_buf = tmio_nand_read_buf;
 
-	/* set eccmode using hardware ECC */
-	nand_chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_ON_HOST;
-	nand_chip->ecc.size = 512;
-	nand_chip->ecc.bytes = 6;
-	nand_chip->ecc.strength = 2;
-	nand_chip->ecc.hwctl = tmio_nand_enable_hwecc;
-	nand_chip->ecc.calculate = tmio_nand_calculate_ecc;
-	nand_chip->ecc.correct = tmio_nand_correct_data;
-
 	if (data)
 		nand_chip->badblock_pattern = data->badblock_pattern;
 
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 14/19] mtd: rawnand: tmio: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:19 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 15/19] mtd: rawnand: pasemi: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/pasemi_nand.c | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/drivers/mtd/nand/raw/pasemi_nand.c b/drivers/mtd/nand/raw/pasemi_nand.c
index 7e9d711d5a8a..5741887d7a24 100644
--- a/drivers/mtd/nand/raw/pasemi_nand.c
+++ b/drivers/mtd/nand/raw/pasemi_nand.c
@@ -28,6 +28,7 @@
 
 static unsigned int lpcctl;
 static struct mtd_info *pasemi_nand_mtd;
+static struct nand_controller controller;
 static const char driver_name[] = "pasemi-nand";
 
 static void pasemi_read_buf(struct nand_chip *chip, u_char *buf, int len)
@@ -72,6 +73,18 @@ static int pasemi_device_ready(struct nand_chip *chip)
 	return !!(inl(lpcctl) & LBICTRL_LPCCTL_NR);
 }
 
+static int pasemi_attach_chip(struct nand_chip *chip)
+{
+	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
+	chip->ecc.algo = NAND_ECC_ALGO_HAMMING;
+
+	return 0;
+}
+
+static const struct nand_controller_ops pasemi_ops = {
+	.attach_chip = pasemi_attach_chip,
+};
+
 static int pasemi_nand_probe(struct platform_device *ofdev)
 {
 	struct device *dev = &ofdev->dev;
@@ -99,6 +112,10 @@ static int pasemi_nand_probe(struct platform_device *ofdev)
 		goto out;
 	}
 
+	controller.ops = &pasemi_ops;
+	nand_controller_init(&controller);
+	chip->controller = &controller;
+
 	pasemi_nand_mtd = nand_to_mtd(chip);
 
 	/* Link the private data with the MTD structure */
@@ -131,8 +148,6 @@ static int pasemi_nand_probe(struct platform_device *ofdev)
 	chip->legacy.read_buf = pasemi_read_buf;
 	chip->legacy.write_buf = pasemi_write_buf;
 	chip->legacy.chip_delay = 0;
-	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
-	chip->ecc.algo = NAND_ECC_ALGO_HAMMING;
 
 	/* Enable the following for a flash based bad block table */
 	chip->bbt_options = NAND_BBT_USE_FLASH;
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 15/19] mtd: rawnand: pasemi: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:20 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 16/19] mtd: rawnand: plat_nand: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/plat_nand.c | 20 +++++++++++++++++---
 1 file changed, 17 insertions(+), 3 deletions(-)

diff --git a/drivers/mtd/nand/raw/plat_nand.c b/drivers/mtd/nand/raw/plat_nand.c
index b98c0d5c413f..93d9f1694dc1 100644
--- a/drivers/mtd/nand/raw/plat_nand.c
+++ b/drivers/mtd/nand/raw/plat_nand.c
@@ -14,10 +14,23 @@
 #include <linux/mtd/platnand.h>
 
 struct plat_nand_data {
+	struct nand_controller	controller;
 	struct nand_chip	chip;
 	void __iomem		*io_base;
 };
 
+static int plat_nand_attach_chip(struct nand_chip *chip)
+{
+	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
+	chip->ecc.algo = NAND_ECC_ALGO_HAMMING;
+
+	return 0;
+}
+
+static const struct nand_controller_ops plat_nand_ops = {
+	.attach_chip = plat_nand_attach_chip,
+};
+
 /*
  * Probe for the NAND device.
  */
@@ -46,6 +59,10 @@ static int plat_nand_probe(struct platform_device *pdev)
 	if (!data)
 		return -ENOMEM;
 
+	data->controller.ops = &plat_nand_ops;
+	nand_controller_init(&data->controller);
+	data->chip.controller = &data->controller;
+
 	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
 	data->io_base = devm_ioremap_resource(&pdev->dev, res);
 	if (IS_ERR(data->io_base))
@@ -66,9 +83,6 @@ static int plat_nand_probe(struct platform_device *pdev)
 	data->chip.options |= pdata->chip.options;
 	data->chip.bbt_options |= pdata->chip.bbt_options;
 
-	data->chip.ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
-	data->chip.ecc.algo = NAND_ECC_ALGO_HAMMING;
-
 	platform_set_drvdata(pdev, data);
 
 	/* Handle any platform specific setup */
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 16/19] mtd: rawnand: plat_nand: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:21 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 17/19] mtd: rawnand: r852: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Cc: Maxim Levitsky <maximlevitsky@gmail.com>
---
 drivers/mtd/nand/raw/r852.c | 40 +++++++++++++++++++++++++------------
 drivers/mtd/nand/raw/r852.h |  1 +
 2 files changed, 28 insertions(+), 13 deletions(-)

diff --git a/drivers/mtd/nand/raw/r852.c b/drivers/mtd/nand/raw/r852.c
index 6b7addd2c420..620708135d3d 100644
--- a/drivers/mtd/nand/raw/r852.c
+++ b/drivers/mtd/nand/raw/r852.c
@@ -817,6 +817,29 @@ static irqreturn_t r852_irq(int irq, void *data)
 	return ret;
 }
 
+static int r852_attach_chip(struct nand_chip *chip)
+{
+	if (chip->ecc.engine_type != NAND_ECC_ENGINE_TYPE_ON_HOST)
+		return 0;
+
+	chip->ecc.placement = NAND_ECC_PLACEMENT_INTERLEAVED;
+	chip->ecc.size = R852_DMA_LEN;
+	chip->ecc.bytes = SM_OOB_SIZE;
+	chip->ecc.strength = 2;
+	chip->ecc.hwctl = r852_ecc_hwctl;
+	chip->ecc.calculate = r852_ecc_calculate;
+	chip->ecc.correct = r852_ecc_correct;
+
+	/* TODO: hack */
+	chip->ecc.read_oob = r852_read_oob;
+
+	return 0;
+}
+
+static const struct nand_controller_ops r852_ops = {
+	.attach_chip = r852_attach_chip,
+};
+
 static int  r852_probe(struct pci_dev *pci_dev, const struct pci_device_id *id)
 {
 	int error;
@@ -848,6 +871,10 @@ static int  r852_probe(struct pci_dev *pci_dev, const struct pci_device_id *id)
 	if (!chip)
 		goto error4;
 
+	nand_controller_init(&dev->controller);
+	dev->controller.ops = &r852_ops;
+	chip->controller = &dev->controller;
+
 	/* commands */
 	chip->legacy.cmd_ctrl = r852_cmdctl;
 	chip->legacy.waitfunc = r852_wait;
@@ -858,19 +885,6 @@ static int  r852_probe(struct pci_dev *pci_dev, const struct pci_device_id *id)
 	chip->legacy.read_buf = r852_read_buf;
 	chip->legacy.write_buf = r852_write_buf;
 
-	/* ecc */
-	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_ON_HOST;
-	chip->ecc.placement = NAND_ECC_PLACEMENT_INTERLEAVED;
-	chip->ecc.size = R852_DMA_LEN;
-	chip->ecc.bytes = SM_OOB_SIZE;
-	chip->ecc.strength = 2;
-	chip->ecc.hwctl = r852_ecc_hwctl;
-	chip->ecc.calculate = r852_ecc_calculate;
-	chip->ecc.correct = r852_ecc_correct;
-
-	/* TODO: hack */
-	chip->ecc.read_oob = r852_read_oob;
-
 	/* init our device structure */
 	dev = kzalloc(sizeof(struct r852_device), GFP_KERNEL);
 
diff --git a/drivers/mtd/nand/raw/r852.h b/drivers/mtd/nand/raw/r852.h
index e9ce299c499d..96fe301d15da 100644
--- a/drivers/mtd/nand/raw/r852.h
+++ b/drivers/mtd/nand/raw/r852.h
@@ -104,6 +104,7 @@
 #define DMA_MEMORY	1
 
 struct r852_device {
+	struct nand_controller		controller;
 	void __iomem *mmio;		/* mmio */
 	struct nand_chip *chip;		/* nand chip backpointer */
 	struct pci_dev *pci_dev;	/* pci backpointer */
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 17/19] mtd: rawnand: r852: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:22 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
> Cc: Maxim Levitsky <maximlevitsky@gmail.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 17/19] mtd: rawnand: r852: Move the ECC initialization to ->attach_chip()

[Nathan Chancellor]

On Fri, Nov 13, 2020 at 01:34:22PM +0100, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
> Cc: Maxim Levitsky <maximlevitsky@gmail.com>

Just as a heads up, this patch was applied to mtd/fixes as
commit 7ef969a04228 ("mtd: rawnand: r852: Move the ECC
initialization to ->attach_chip()") but it appears to also be applied to
nand/next as commit 0c2b69f55132 ("mtd: rawnand: r852: Move the ECC
initialization to ->attach_chip()"), where due to its mismerged nature,
it causes a clang warning:

../drivers/mtd/nand/raw/r852.c:874:24: warning: variable 'dev' is uninitialized when used here [-Wuninitialized]
        nand_controller_init(&dev->controller);
                              ^~~
../drivers/mtd/nand/raw/r852.c:847:25: note: initialize the variable 'dev' to silence this warning
        struct r852_device *dev;
                               ^
                                = NULL
1 warning generated.

Cheers,
Nathan Chancellor

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 17/19] mtd: rawnand: r852: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

Hi Nathan,

Nathan Chancellor <natechancellor@gmail.com> wrote on Sun, 29 Nov 2020
18:07:55 -0700:

> On Fri, Nov 13, 2020 at 01:34:22PM +0100, Miquel Raynal wrote:
> > The probe function is only supposed to initialize the controller
> > hardware but not the ECC engine. Indeed, we don't know anything about
> > the NAND chip(s) at this stage. Let's move the logic initializing the
> > ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> > called during nand_scan() routine, after the NAND chip discovery. As
> > the previously mentioned logic is supposed to parse the DT for us, it
> > is likely that the chip->ecc.* entries be overwritten. So let's avoid
> > this by moving these lines to ->attach_chip().
> > 
> > Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> > Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
> > Cc: Maxim Levitsky <maximlevitsky@gmail.com>  
> 
> Just as a heads up, this patch was applied to mtd/fixes as
> commit 7ef969a04228 ("mtd: rawnand: r852: Move the ECC
> initialization to ->attach_chip()") but it appears to also be applied to
> nand/next as commit 0c2b69f55132 ("mtd: rawnand: r852: Move the ECC
> initialization to ->attach_chip()"), where due to its mismerged nature,
> it causes a clang warning:
> 
> ../drivers/mtd/nand/raw/r852.c:874:24: warning: variable 'dev' is uninitialized when used here [-Wuninitialized]
>         nand_controller_init(&dev->controller);
>                               ^~~
> ../drivers/mtd/nand/raw/r852.c:847:25: note: initialize the variable 'dev' to silence this warning
>         struct r852_device *dev;
>                                ^
>                                 = NULL
> 1 warning generated.
> 
> Cheers,
> Nathan Chancellor

Absolutely, this fix came late, nand/next was already set up and there
was conflicts between the two. I applied these same patches at the root
of nand/next to be sure the branch would still be valid until the next
-rc would come out (today).

Thanks,
Miquèl

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 18/19] mtd: rawnand: sharpsl: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/sharpsl.c | 32 ++++++++++++++++++++++++--------
 1 file changed, 24 insertions(+), 8 deletions(-)

diff --git a/drivers/mtd/nand/raw/sharpsl.c b/drivers/mtd/nand/raw/sharpsl.c
index 6ceb961d2c9f..5612ee628425 100644
--- a/drivers/mtd/nand/raw/sharpsl.c
+++ b/drivers/mtd/nand/raw/sharpsl.c
@@ -19,6 +19,7 @@
 #include <linux/io.h>
 
 struct sharpsl_nand {
+	struct nand_controller	controller;
 	struct nand_chip	chip;
 
 	void __iomem		*io;
@@ -95,6 +96,25 @@ static int sharpsl_nand_calculate_ecc(struct nand_chip *chip,
 	return readb(sharpsl->io + ECCCNTR) != 0;
 }
 
+static int sharpsl_attach_chip(struct nand_chip *chip)
+{
+	if (chip->ecc.engine_type != NAND_ECC_ENGINE_TYPE_ON_HOST)
+		return 0;
+
+	chip->ecc.size = 256;
+	chip->ecc.bytes = 3;
+	chip->ecc.strength = 1;
+	chip->ecc.hwctl = sharpsl_nand_enable_hwecc;
+	chip->ecc.calculate = sharpsl_nand_calculate_ecc;
+	chip->ecc.correct = rawnand_sw_hamming_correct;
+
+	return 0;
+}
+
+static const struct nand_controller_ops sharpsl_ops = {
+	.attach_chip = sharpsl_attach_chip,
+};
+
 /*
  * Main initialization routine
  */
@@ -135,6 +155,10 @@ static int sharpsl_nand_probe(struct platform_device *pdev)
 	/* Get pointer to private data */
 	this = (struct nand_chip *)(&sharpsl->chip);
 
+	nand_controller_init(&sharpsl->controller);
+	sharpsl->controller.ops = &sharpsl_ops;
+	this->controller = &sharpsl->controller;
+
 	/* Link the private data with the MTD structure */
 	mtd = nand_to_mtd(this);
 	mtd->dev.parent = &pdev->dev;
@@ -155,15 +179,7 @@ static int sharpsl_nand_probe(struct platform_device *pdev)
 	this->legacy.dev_ready = sharpsl_nand_dev_ready;
 	/* 15 us command delay time */
 	this->legacy.chip_delay = 15;
-	/* set eccmode using hardware ECC */
-	this->ecc.engine_type = NAND_ECC_ENGINE_TYPE_ON_HOST;
-	this->ecc.size = 256;
-	this->ecc.bytes = 3;
-	this->ecc.strength = 1;
 	this->badblock_pattern = data->badblock_pattern;
-	this->ecc.hwctl = sharpsl_nand_enable_hwecc;
-	this->ecc.calculate = sharpsl_nand_calculate_ecc;
-	this->ecc.correct = rawnand_sw_hamming_correct;
 
 	/* Scan to find existence of the device */
 	err = nand_scan(this, 1);
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

[PATCH 19/19] mtd: rawnand: socrates: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

The probe function is only supposed to initialize the controller
hardware but not the ECC engine. Indeed, we don't know anything about
the NAND chip(s) at this stage. Let's move the logic initializing the
ECC engine, even pretty simple, to the ->attach_chip() hook which gets
called during nand_scan() routine, after the NAND chip discovery. As
the previously mentioned logic is supposed to parse the DT for us, it
is likely that the chip->ecc.* entries be overwritten. So let's avoid
this by moving these lines to ->attach_chip().

Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
---
 drivers/mtd/nand/raw/socrates_nand.c | 21 +++++++++++++++++----
 1 file changed, 17 insertions(+), 4 deletions(-)

diff --git a/drivers/mtd/nand/raw/socrates_nand.c b/drivers/mtd/nand/raw/socrates_nand.c
index 0f63ff6f7fe7..107208311987 100644
--- a/drivers/mtd/nand/raw/socrates_nand.c
+++ b/drivers/mtd/nand/raw/socrates_nand.c
@@ -22,6 +22,7 @@
 #define FPGA_NAND_DATA_SHIFT		16
 
 struct socrates_nand_host {
+	struct nand_controller	controller;
 	struct nand_chip	nand_chip;
 	void __iomem		*io_base;
 	struct device		*dev;
@@ -116,6 +117,18 @@ static int socrates_nand_device_ready(struct nand_chip *nand_chip)
 	return 1;
 }
 
+static int socrates_attach_chip(struct nand_chip *chip)
+{
+	chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
+	chip->ecc.algo = NAND_ECC_ALGO_HAMMING;
+
+	return 0;
+}
+
+static const struct nand_controller_ops socrates_ops = {
+	.attach_chip = socrates_attach_chip,
+};
+
 /*
  * Probe for the NAND device.
  */
@@ -141,6 +154,10 @@ static int socrates_nand_probe(struct platform_device *ofdev)
 	mtd = nand_to_mtd(nand_chip);
 	host->dev = &ofdev->dev;
 
+	nand_controller_init(&host->controller);
+	host->controller.ops = &socrates_ops;
+	nand_chip->controller = &host->controller;
+
 	/* link the private data structures */
 	nand_set_controller_data(nand_chip, host);
 	nand_set_flash_node(nand_chip, ofdev->dev.of_node);
@@ -153,10 +170,6 @@ static int socrates_nand_probe(struct platform_device *ofdev)
 	nand_chip->legacy.read_buf = socrates_nand_read_buf;
 	nand_chip->legacy.dev_ready = socrates_nand_device_ready;
 
-	/* enable ECC */
-	nand_chip->ecc.engine_type = NAND_ECC_ENGINE_TYPE_SOFT;
-	nand_chip->ecc.algo = NAND_ECC_ALGO_HAMMING;
-
 	/* TODO: I have no idea what real delay is. */
 	nand_chip->legacy.chip_delay = 20;	/* 20us command delay time */
 
-- 
2.20.1


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: [PATCH 19/19] mtd: rawnand: socrates: Move the ECC initialization to ->attach_chip()

[Miquel Raynal]

On Fri, 2020-11-13 at 12:34:24 UTC, Miquel Raynal wrote:
> The probe function is only supposed to initialize the controller
> hardware but not the ECC engine. Indeed, we don't know anything about
> the NAND chip(s) at this stage. Let's move the logic initializing the
> ECC engine, even pretty simple, to the ->attach_chip() hook which gets
> called during nand_scan() routine, after the NAND chip discovery. As
> the previously mentioned logic is supposed to parse the DT for us, it
> is likely that the chip->ecc.* entries be overwritten. So let's avoid
> this by moving these lines to ->attach_chip().
> 
> Fixes: d7157ff49a5b ("mtd: rawnand: Use the ECC framework user input parsing bits")
> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/fixes.

Miquel

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/