2021-03-23 14:33:37

by Michael Walle

[permalink] [raw]
Subject: [PATCH v2 1/2] mtd: spi-nor: sfdp: save a copy of the SFDP data

Due to possible mode switching to 8D-8D-8D, it might not be possible to
read the SFDP after the initial probe. To be able to dump the SFDP via
sysfs afterwards, make a complete copy of it.

Signed-off-by: Michael Walle <[email protected]>
---
drivers/mtd/spi-nor/core.h | 10 +++++++
drivers/mtd/spi-nor/sfdp.c | 58 +++++++++++++++++++++++++++++++++++++
include/linux/mtd/spi-nor.h | 2 ++
3 files changed, 70 insertions(+)

diff --git a/drivers/mtd/spi-nor/core.h b/drivers/mtd/spi-nor/core.h
index db07832ee66c..08d2469837da 100644
--- a/drivers/mtd/spi-nor/core.h
+++ b/drivers/mtd/spi-nor/core.h
@@ -406,6 +406,16 @@ struct spi_nor_manufacturer {
const struct spi_nor_fixups *fixups;
};

+/**
+ * struct sfdp - SFDP data
+ * @num_dwords: number of entries in the dwords array
+ * @dwords: array of double words of the SFDP data
+ */
+struct sfdp {
+ size_t num_dwords;
+ u32 *dwords;
+};
+
/* Manufacturer drivers. */
extern const struct spi_nor_manufacturer spi_nor_atmel;
extern const struct spi_nor_manufacturer spi_nor_catalyst;
diff --git a/drivers/mtd/spi-nor/sfdp.c b/drivers/mtd/spi-nor/sfdp.c
index 23c28e91f698..c500c2118a5d 100644
--- a/drivers/mtd/spi-nor/sfdp.c
+++ b/drivers/mtd/spi-nor/sfdp.c
@@ -16,6 +16,7 @@
(((p)->parameter_table_pointer[2] << 16) | \
((p)->parameter_table_pointer[1] << 8) | \
((p)->parameter_table_pointer[0] << 0))
+#define SFDP_PARAM_HEADER_PARAM_LEN(p) ((p)->length * 4)

#define SFDP_BFPT_ID 0xff00 /* Basic Flash Parameter Table */
#define SFDP_SECTOR_MAP_ID 0xff81 /* Sector Map Table */
@@ -1245,6 +1246,8 @@ int spi_nor_parse_sfdp(struct spi_nor *nor)
struct sfdp_parameter_header *param_headers = NULL;
struct sfdp_header header;
struct device *dev = nor->dev;
+ struct sfdp *sfdp;
+ size_t sfdp_size;
size_t psize;
int i, err;

@@ -1267,6 +1270,9 @@ int spi_nor_parse_sfdp(struct spi_nor *nor)
bfpt_header->major != SFDP_JESD216_MAJOR)
return -EINVAL;

+ sfdp_size = SFDP_PARAM_HEADER_PTP(bfpt_header) +
+ SFDP_PARAM_HEADER_PARAM_LEN(bfpt_header);
+
/*
* Allocate memory then read all parameter headers with a single
* Read SFDP command. These parameter headers will actually be parsed
@@ -1293,6 +1299,58 @@ int spi_nor_parse_sfdp(struct spi_nor *nor)
}
}

+ /*
+ * Cache the complete SFDP data. It is not (easily) possible to fetch
+ * SFDP after probe time and we need it for the sysfs access.
+ */
+ for (i = 0; i < header.nph; i++) {
+ param_header = &param_headers[i];
+ sfdp_size = max_t(size_t, sfdp_size,
+ SFDP_PARAM_HEADER_PTP(param_header) +
+ SFDP_PARAM_HEADER_PARAM_LEN(param_header));
+ }
+
+ /*
+ * Limit the total size to a reasonable value to avoid allocating too
+ * much memory just of because the flash returned some insane values.
+ */
+ if (sfdp_size > PAGE_SIZE) {
+ dev_dbg(dev, "SFDP data (%zu) too big, truncating\n",
+ sfdp_size);
+ sfdp_size = PAGE_SIZE;
+ }
+
+ sfdp = devm_kzalloc(dev, sizeof(*sfdp), GFP_KERNEL);
+ if (!sfdp) {
+ err = -ENOMEM;
+ goto exit;
+ }
+
+ /*
+ * The SFDP is organized in chunks of DWORDs. Thus, in theory, the
+ * sfdp_size should be a multiple of DWORDs. But in case a flash
+ * is not spec compliant, make sure that we have enough space to store
+ * the complete SFDP data.
+ */
+ sfdp->num_dwords = DIV_ROUND_UP(sfdp_size, sizeof(*sfdp->dwords));
+ sfdp->dwords = devm_kcalloc(dev, sfdp->num_dwords,
+ sizeof(*sfdp->dwords), GFP_KERNEL);
+ if (!sfdp->dwords) {
+ err = -ENOMEM;
+ devm_kfree(dev, sfdp);
+ goto exit;
+ }
+
+ err = spi_nor_read_sfdp(nor, 0, sfdp_size, sfdp->dwords);
+ if (err < 0) {
+ dev_dbg(dev, "failed to read SFDP data\n");
+ devm_kfree(dev, sfdp->dwords);
+ devm_kfree(dev, sfdp);
+ goto exit;
+ }
+
+ nor->sfdp = sfdp;
+
/*
* Check other parameter headers to get the latest revision of
* the basic flash parameter table.
diff --git a/include/linux/mtd/spi-nor.h b/include/linux/mtd/spi-nor.h
index a0d572855444..2215e3565422 100644
--- a/include/linux/mtd/spi-nor.h
+++ b/include/linux/mtd/spi-nor.h
@@ -375,6 +375,7 @@ struct spi_nor_flash_parameter;
* @read_proto: the SPI protocol for read operations
* @write_proto: the SPI protocol for write operations
* @reg_proto: the SPI protocol for read_reg/write_reg/erase operations
+ * @sfdp: the SFDP data of the flash
* @controller_ops: SPI NOR controller driver specific operations.
* @params: [FLASH-SPECIFIC] SPI NOR flash parameters and settings.
* The structure includes legacy flash parameters and
@@ -404,6 +405,7 @@ struct spi_nor {
bool sst_write_second;
u32 flags;
enum spi_nor_cmd_ext cmd_ext_type;
+ struct sfdp *sfdp;

const struct spi_nor_controller_ops *controller_ops;

--
2.20.1


2021-03-23 17:39:27

by Pratyush Yadav

[permalink] [raw]
Subject: Re: [PATCH v2 1/2] mtd: spi-nor: sfdp: save a copy of the SFDP data

On 23/03/21 03:31PM, Michael Walle wrote:
> Due to possible mode switching to 8D-8D-8D, it might not be possible to
> read the SFDP after the initial probe. To be able to dump the SFDP via
> sysfs afterwards, make a complete copy of it.
>
> Signed-off-by: Michael Walle <[email protected]>

Reviewed-by: Pratyush Yadav <[email protected]>

--
Regards,
Pratyush Yadav
Texas Instruments Inc.

2021-03-23 18:47:33

by Michael Walle

[permalink] [raw]
Subject: Re: [PATCH v2 1/2] mtd: spi-nor: sfdp: save a copy of the SFDP data

Am 23. März 2021 18:37:17 MEZ schrieb Pratyush Yadav <[email protected]>:
>On 23/03/21 03:31PM, Michael Walle wrote:
>> Due to possible mode switching to 8D-8D-8D, it might not be possible
>to
>> read the SFDP after the initial probe. To be able to dump the SFDP
>via
>> sysfs afterwards, make a complete copy of it.
>>
>> Signed-off-by: Michael Walle <[email protected]>
>
>Reviewed-by: Pratyush Yadav <[email protected]>

thanks for reviewing!

-michael

2021-03-23 19:05:50

by Heiko Thiery

[permalink] [raw]
Subject: Re: [PATCH v2 1/2] mtd: spi-nor: sfdp: save a copy of the SFDP data

Hi Michael,

Am Di., 23. März 2021 um 19:45 Uhr schrieb Michael Walle <[email protected]>:
>
> Am 23. März 2021 18:37:17 MEZ schrieb Pratyush Yadav <[email protected]>:
> >On 23/03/21 03:31PM, Michael Walle wrote:
> >> Due to possible mode switching to 8D-8D-8D, it might not be possible
> >to
> >> read the SFDP after the initial probe. To be able to dump the SFDP
> >via
> >> sysfs afterwards, make a complete copy of it.
> >>
> >> Signed-off-by: Michael Walle <[email protected]>
> >
> >Reviewed-by: Pratyush Yadav <[email protected]>

Tested-by: Heiko Thiery <[email protected]>

>
> thanks for reviewing!
>
> -michael
>
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/