From: Marek Vasut <[email protected]>
The SPARE_AREA_SKIP_BYTES register is reset when the controller reset
signal is toggled. Yet, this register must be configured to match the
content of the NAND OOB area. The current default value is always set
to 8 and is programmed into the hardware in case the hardware was not
programmed before (e.g. in a bootloader) with a different value. This
however does not work when the block is reset properly by Linux.
On Altera SoCFPGA CycloneV, ArriaV and Arria10, which are the SoCFPGA
platforms which support booting from NAND, the SPARE_AREA_SKIP_BYTES
value must be set to 2. On Socionext Uniphier, the value is 8. This
patch adds support for preconfiguring the default value and handles
the special SoCFPGA case by setting the default to 2 on all SoCFPGA
platforms, while retaining the original behavior and default value of
8 on all the other platforms.
Signed-off-by: Marek Vasut <[email protected]>
Cc: Miquel Raynal <[email protected]>
Cc: Richard Weinberger <[email protected]>
Cc: Vignesh Raghavendra <[email protected]>
To: [email protected]
Reviewed-by: Tudor Ambarus <[email protected]>
Acked-by: Masahiro Yamada <[email protected]>
---
Changes in v3:
[Masahiro]
- Rebase and give my Acked-by
Changes in v2:
- V2: Move denali->oob_skip_bytes = data->oob_skip_bytes; right after
of_device_get_match_data()
drivers/mtd/nand/raw/denali.c | 13 ++++++++++---
drivers/mtd/nand/raw/denali_dt.c | 5 +++++
2 files changed, 15 insertions(+), 3 deletions(-)
diff --git a/drivers/mtd/nand/raw/denali.c b/drivers/mtd/nand/raw/denali.c
index 3102ddbd8abd..b6c463d02167 100644
--- a/drivers/mtd/nand/raw/denali.c
+++ b/drivers/mtd/nand/raw/denali.c
@@ -1302,14 +1302,21 @@ int denali_init(struct denali_controller *denali)
/*
* Set how many bytes should be skipped before writing data in OOB.
+ * If a non-zero value has already been configured, update it in HW.
* If a non-zero value has already been set (by firmware or something),
* just use it. Otherwise, set the driver's default.
*/
- denali->oob_skip_bytes = ioread32(denali->reg + SPARE_AREA_SKIP_BYTES);
- if (!denali->oob_skip_bytes) {
- denali->oob_skip_bytes = DENALI_DEFAULT_OOB_SKIP_BYTES;
+ if (denali->oob_skip_bytes) {
iowrite32(denali->oob_skip_bytes,
denali->reg + SPARE_AREA_SKIP_BYTES);
+ } else {
+ denali->oob_skip_bytes =
+ ioread32(denali->reg + SPARE_AREA_SKIP_BYTES);
+ if (!denali->oob_skip_bytes) {
+ denali->oob_skip_bytes = DENALI_DEFAULT_OOB_SKIP_BYTES;
+ iowrite32(denali->oob_skip_bytes,
+ denali->reg + SPARE_AREA_SKIP_BYTES);
+ }
}
iowrite32(0, denali->reg + TRANSFER_SPARE_REG);
diff --git a/drivers/mtd/nand/raw/denali_dt.c b/drivers/mtd/nand/raw/denali_dt.c
index 276187939689..699255fb2dd8 100644
--- a/drivers/mtd/nand/raw/denali_dt.c
+++ b/drivers/mtd/nand/raw/denali_dt.c
@@ -27,6 +27,7 @@ struct denali_dt {
struct denali_dt_data {
unsigned int revision;
unsigned int caps;
+ unsigned int oob_skip_bytes;
const struct nand_ecc_caps *ecc_caps;
};
@@ -34,6 +35,7 @@ NAND_ECC_CAPS_SINGLE(denali_socfpga_ecc_caps, denali_calc_ecc_bytes,
512, 8, 15);
static const struct denali_dt_data denali_socfpga_data = {
.caps = DENALI_CAP_HW_ECC_FIXUP,
+ .oob_skip_bytes = 2,
.ecc_caps = &denali_socfpga_ecc_caps,
};
@@ -42,6 +44,7 @@ NAND_ECC_CAPS_SINGLE(denali_uniphier_v5a_ecc_caps, denali_calc_ecc_bytes,
static const struct denali_dt_data denali_uniphier_v5a_data = {
.caps = DENALI_CAP_HW_ECC_FIXUP |
DENALI_CAP_DMA_64BIT,
+ .oob_skip_bytes = 8,
.ecc_caps = &denali_uniphier_v5a_ecc_caps,
};
@@ -51,6 +54,7 @@ static const struct denali_dt_data denali_uniphier_v5b_data = {
.revision = 0x0501,
.caps = DENALI_CAP_HW_ECC_FIXUP |
DENALI_CAP_DMA_64BIT,
+ .oob_skip_bytes = 8,
.ecc_caps = &denali_uniphier_v5b_ecc_caps,
};
@@ -123,6 +127,7 @@ static int denali_dt_probe(struct platform_device *pdev)
denali->revision = data->revision;
denali->caps = data->caps;
+ denali->oob_skip_bytes = data->oob_skip_bytes;
denali->ecc_caps = data->ecc_caps;
denali->dev = dev;
--
2.17.1
On Fri, 2019-12-20 at 11:31:52 UTC, Masahiro Yamada wrote:
> From: Marek Vasut <[email protected]>
>
> The SPARE_AREA_SKIP_BYTES register is reset when the controller reset
> signal is toggled. Yet, this register must be configured to match the
> content of the NAND OOB area. The current default value is always set
> to 8 and is programmed into the hardware in case the hardware was not
> programmed before (e.g. in a bootloader) with a different value. This
> however does not work when the block is reset properly by Linux.
>
> On Altera SoCFPGA CycloneV, ArriaV and Arria10, which are the SoCFPGA
> platforms which support booting from NAND, the SPARE_AREA_SKIP_BYTES
> value must be set to 2. On Socionext Uniphier, the value is 8. This
> patch adds support for preconfiguring the default value and handles
> the special SoCFPGA case by setting the default to 2 on all SoCFPGA
> platforms, while retaining the original behavior and default value of
> 8 on all the other platforms.
>
> Signed-off-by: Marek Vasut <[email protected]>
> Cc: Miquel Raynal <[email protected]>
> Cc: Richard Weinberger <[email protected]>
> Cc: Vignesh Raghavendra <[email protected]>
> To: [email protected]
> Reviewed-by: Tudor Ambarus <[email protected]>
> Acked-by: Masahiro Yamada <[email protected]>
Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git nand/next, thanks.
Miquel