Re: [PATCH v4] mtd: spi-nor: Add support for S3AN spi-nor devices

From: Cyrille Pitchen
Date: Fri Sep 16 2016 - 12:33:39 EST

Next message: Mike Galbraith: "Re: [tip regression] efi: Allow drivers to reserve boot services forever == toxic"
Previous message: Mark Brown: "Re: [PATCH v4 4/4] regulator: Prevent falling too fast"
In reply to: Ricardo Ribalda Delgado: "Re: [PATCH v4] mtd: spi-nor: Add support for S3AN spi-nor devices"
Next in thread: Ricardo Ribalda Delgado: "Re: [PATCH v4] mtd: spi-nor: Add support for S3AN spi-nor devices"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Hi Ricardo,

Le 14/07/2016 à 22:18, Ricardo Ribalda Delgado a écrit :
> Xilinx Spartan-3AN FPGAs contain an In-System Flash where they keep
> their configuration data and (optionally) some user data.
>
> The protocol of this flash follows most of the spi-nor standard. With
> the following differences:
>
> - Page size might not be a power of two.
> - The address calculation (default addressing mode).
> - The spi nor commands used.
>
> Protocol is described on Xilinx User Guide UG333
>
> Signed-off-by: Ricardo Ribalda Delgado <ricardo.ribalda@xxxxxxxxx>
> ---
> v4:
> -Rebase on top of l2-mtd/master
>
> v3:
> -Rebase on top of mtd-next
> -Rename ADDR_NATIVE to ADDR_DEFAULT to follow UG333 naming
> -Fix bug on probe
>
> v2: Suggested by Brian Norris <computersforpeace@xxxxxxxxx>
>
> -Remove inline qualifier
> -Improve documentation of Default Addressing Mode
> -Convert function callbacks into SNOR_F_
> -Fix missmatch braces
> -Improve documentation of SPI_S3AN flag
>
> drivers/mtd/devices/m25p80.c | 3 ++
> drivers/mtd/spi-nor/spi-nor.c | 110 ++++++++++++++++++++++++++++++++++++++++--
> include/linux/mtd/spi-nor.h | 26 ++++++++++
> 3 files changed, 136 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/mtd/devices/m25p80.c b/drivers/mtd/devices/m25p80.c
> index 9cf7fcd28034..a44465f2b73d 100644
> --- a/drivers/mtd/devices/m25p80.c
> +++ b/drivers/mtd/devices/m25p80.c
> @@ -49,6 +49,9 @@ static int m25p80_read_reg(struct spi_nor *nor, u8 code, u8 *val, int len)
>
> static void m25p_addr2cmd(struct spi_nor *nor, unsigned int addr, u8 *cmd)
> {
> + if (nor->flags & SNOR_F_S3AN_ADDR_DEFAULT)
> + addr = spi_nor_s3an_addr_convert(nor, addr);
> +

The address translation is done in spi-nor.c for Page Program and Sector Erase
operations. Why not be consistent and also do that translation in
spi_nor_read() in the very same file for (Fast) Read operations?

No need to modify m25p80.c or to export the spi_nor_s3an_addr_convert()
function.

> /* opcode is in cmd[0] */
> cmd[1] = addr >> (nor->addr_width * 8 - 8);
> cmd[2] = addr >> (nor->addr_width * 8 - 16);
> diff --git a/drivers/mtd/spi-nor/spi-nor.c b/drivers/mtd/spi-nor/spi-nor.c
> index d0fc165d7d66..dce0a2f156e1 100644
> --- a/drivers/mtd/spi-nor/spi-nor.c
> +++ b/drivers/mtd/spi-nor/spi-nor.c
> @@ -75,6 +75,12 @@ struct flash_info {
> * bit. Must be used with
> * SPI_NOR_HAS_LOCK.
> */
> +#define SPI_S3AN BIT(10) /*
> + * Xilinx Spartan 3AN In-System Flash
> + * (MFR cannot be used for probing
> + * because it has the same value as
> + * ATMEL flashes)
> + */
> };
>
> #define JEDEC_MFR(info) ((info)->id[0])
> @@ -217,6 +223,21 @@ static inline int set_4byte(struct spi_nor *nor, const struct flash_info *info,
> return nor->write_reg(nor, SPINOR_OP_BRWR, nor->cmd_buf, 1);
> }
> }
> +
> +static int s3an_sr_ready(struct spi_nor *nor)
> +{
> + int ret;
> + u8 val;
> +
> + ret = nor->read_reg(nor, SPINOR_OP_XRDSR, &val, 1);
> + if (ret < 0) {
> + pr_err("error %d reading XRDSR\n", (int) ret);
Maybe substitute "pr_err" with "dev_err(nor->dev, "
It would be more consistent with other error outputs in spi-nor.c.

> + return ret;
> + }
> +
> + return !!(val & XSR_RDY);
> +}
> +
> static inline int spi_nor_sr_ready(struct spi_nor *nor)
> {
> int sr = read_sr(nor);
> @@ -238,7 +259,9 @@ static inline int spi_nor_fsr_ready(struct spi_nor *nor)
> static int spi_nor_ready(struct spi_nor *nor)
> {
> int sr, fsr;
> - sr = spi_nor_sr_ready(nor);
> +
> + sr = nor->flags & SNOR_F_READY_XSR_RDY ? s3an_sr_ready(nor) :
> + spi_nor_sr_ready(nor);
> if (sr < 0)
> return sr;
> fsr = nor->flags & SNOR_F_USE_FSR ? spi_nor_fsr_ready(nor) : 1;
> @@ -320,6 +343,23 @@ static void spi_nor_unlock_and_unprep(struct spi_nor *nor, enum spi_nor_ops ops)
> }
>
> /*
> + * This code converts an address to the Default Address Mode, that has non
> + * power of two page sizes. We must support this mode because it is the default
> + * mode supported by Xilinx tools, it can access the whole flash area and
> + * changing over to the Power-of-two mode is irreversible and corrupts the
> + * original data.
> + */
> +unsigned int spi_nor_s3an_addr_convert(struct spi_nor *nor, unsigned int addr)
> +{
> + unsigned int offset;
> +
> + offset = (nor->page_size == 264) ? (addr % 264) : (addr % 528);
> +
> + return ((addr - offset) << 1) | offset;
> +}
> +EXPORT_SYMBOL_GPL(spi_nor_s3an_addr_convert);
> +
> +/*
> * Initiate the erasure of a single sector
> */
> static int spi_nor_erase_sector(struct spi_nor *nor, u32 addr)
> @@ -330,6 +370,9 @@ static int spi_nor_erase_sector(struct spi_nor *nor, u32 addr)
> if (nor->erase)
> return nor->erase(nor, addr);
>
> + if (nor->flags & SNOR_F_S3AN_ADDR_DEFAULT)
> + addr = spi_nor_s3an_addr_convert(nor, addr);
> +

Why not put the address translation before the if (nor->erase) test?

Actually it might be not blocking as I expect m25p80 to be the only driver
that can support S3AN memories and m25p80 no longer implement the .erase() hook
but relies on the .write_reg() hook instead.

Due to the address translation, I guess S3AN memories cannot be used with
system memory mapped SPI controllers.
> /*
> * Default implementation, if driver doesn't have a specialized HW
> * control
> @@ -368,7 +411,7 @@ static int spi_nor_erase(struct mtd_info *mtd, struct erase_info *instr)
> return ret;
>
> /* whole-chip erase? */
> - if (len == mtd->size) {
> + if (len == mtd->size && !(nor->flags & SNOR_F_NO_OP_CHIP_ERASE)) {
> unsigned long timeout;
>
> write_enable(nor);
> @@ -782,6 +825,19 @@ static int spi_nor_is_locked(struct mtd_info *mtd, loff_t ofs, uint64_t len)
> .addr_width = (_addr_width), \
> .flags = (_flags),
>
> +#define S3AN_INFO(_jedec_id, _n_sectors, _page_size) \
> + .id = { \
> + ((_jedec_id) >> 16) & 0xff, \
> + ((_jedec_id) >> 8) & 0xff, \
> + (_jedec_id) & 0xff \
> + }, \
> + .id_len = 3, \
> + .sector_size = (8*_page_size), \
> + .n_sectors = (_n_sectors), \
> + .page_size = _page_size, \
> + .addr_width = 3, \
> + .flags = SPI_NOR_NO_FR | SPI_S3AN,
> +
> /* NOTE: double check command sets and memory organization when you add
> * more nor chips. This current list focusses on newer chips, which
> * have been converging on command sets which including JEDEC ID.
> @@ -1009,6 +1065,13 @@ static const struct flash_info spi_nor_ids[] = {
> { "cat25c09", CAT25_INFO( 128, 8, 32, 2, SPI_NOR_NO_ERASE | SPI_NOR_NO_FR) },
> { "cat25c17", CAT25_INFO( 256, 8, 32, 2, SPI_NOR_NO_ERASE | SPI_NOR_NO_FR) },
> { "cat25128", CAT25_INFO(2048, 8, 64, 2, SPI_NOR_NO_ERASE | SPI_NOR_NO_FR) },
> +
> + /* Xilinx S3AN Internal Flash */
> + { "3S50AN", S3AN_INFO(0x1f2200, 64, 264) },
> + { "3S200AN", S3AN_INFO(0x1f2400, 256, 264) },
> + { "3S400AN", S3AN_INFO(0x1f2400, 256, 264) },
> + { "3S700AN", S3AN_INFO(0x1f2500, 512, 264) },
> + { "3S1400AN", S3AN_INFO(0x1f2600, 512, 528) },
> { },
> };
>
> @@ -1171,7 +1234,13 @@ static int spi_nor_write(struct mtd_info *mtd, loff_t to, size_t len,
> for (i = 0; i < len; ) {
> ssize_t written;
>
> - page_offset = (to + i) & (nor->page_size - 1);
> + if (hweight32(nor->page_size) == 1)
> + page_offset = to & (nor->page_size - 1);

The original formula was (to + i) & (nor->page_size - 1) whereas the new
formula is now to & (nor->page_size - 1).

It should be (to + i), shouldn't it? No change should be done for power of 2
page size.

> + else {
> + uint64_t aux = to;
what about i ?
> +
> + page_offset = do_div(aux, nor->page_size);
> + }
> WARN_ONCE(page_offset,
> "Writing at offset %zu into a NOR page. Writing partial pages may decrease reliability and increase wear of NOR flash.",
> page_offset);
> @@ -1300,6 +1369,35 @@ static int spi_nor_check(struct spi_nor *nor)
> return 0;
> }
>
> +static int s3an_nor_scan(const struct flash_info *info, struct spi_nor *nor)
> +{
> + int ret;
> + u8 val;
> +
> + ret = nor->read_reg(nor, SPINOR_OP_XRDSR, &val, 1);
> + if (ret < 0) {
> + pr_err("error %d reading XRDSR\n", (int) ret);
> + return ret;
> + }
> +
> + nor->erase_opcode = SPINOR_OP_XSE;
> + nor->program_opcode = SPINOR_OP_XPP;
> + nor->read_opcode = SPINOR_OP_READ;
> + nor->flags |= SNOR_F_NO_OP_CHIP_ERASE | SNOR_F_READY_XSR_RDY;
> +
> + /* Flash in Power of 2 mode */
> + if (val & XSR_PAGESIZE) {
> + nor->page_size = (nor->page_size == 264) ? 256 : 512;
> + nor->mtd.writebufsize = nor->page_size;
> + nor->mtd.size = 8 * nor->page_size * info->n_sectors;
> + nor->mtd.erasesize = 8 * nor->page_size;
> + } else {
> + nor->flags |= SNOR_F_S3AN_ADDR_DEFAULT;
> + }
> +
> + return 0;
> +}
> +
> int spi_nor_scan(struct spi_nor *nor, const char *name, enum read_mode mode)
> {
> const struct flash_info *info = NULL;
> @@ -1505,6 +1603,12 @@ int spi_nor_scan(struct spi_nor *nor, const char *name, enum read_mode mode)
>
> nor->read_dummy = spi_nor_read_dummy_cycles(nor);
>
> + if (info->flags & SPI_S3AN) {
> + ret = s3an_nor_scan(info, nor);
> + if (ret)
> + return ret;
> + }
> +
> dev_info(dev, "%s (%lld Kbytes)\n", info->name,
> (long long)mtd->size >> 10);
>
> diff --git a/include/linux/mtd/spi-nor.h b/include/linux/mtd/spi-nor.h
> index c425c7b4c2a0..6378de414bca 100644
> --- a/include/linux/mtd/spi-nor.h
> +++ b/include/linux/mtd/spi-nor.h
> @@ -68,6 +68,15 @@
> #define SPINOR_OP_WRDI 0x04 /* Write disable */
> #define SPINOR_OP_AAI_WP 0xad /* Auto address increment word program */
>
> +/* Used for S3AN flashes only */
> +#define SPINOR_OP_XSE 0x50 /* Sector erase */
> +#define SPINOR_OP_XPP 0x82 /* Page program */
> +#define SPINOR_OP_XRDSR 0xd7 /* Read status register */
> +
> +#define XSR_PAGESIZE BIT(0) /* Page size in Po2 or Linear */
> +#define XSR_RDY BIT(7) /* Ready */
> +
> +
> /* Used for Macronix and Winbond flashes. */
> #define SPINOR_OP_EN4B 0xb7 /* Enter 4-byte mode */
> #define SPINOR_OP_EX4B 0xe9 /* Exit 4-byte mode */
> @@ -119,6 +128,9 @@ enum spi_nor_ops {
> enum spi_nor_option_flags {
> SNOR_F_USE_FSR = BIT(0),
> SNOR_F_HAS_SR_TB = BIT(1),
> + SNOR_F_NO_OP_CHIP_ERASE = BIT(2),
> + SNOR_F_S3AN_ADDR_DEFAULT = BIT(3),
> + SNOR_F_READY_XSR_RDY = BIT(4),
> };
>
> /**
> @@ -197,6 +209,20 @@ static inline struct device_node *spi_nor_get_flash_node(struct spi_nor *nor)
> return mtd_get_of_node(&nor->mtd);
> }
>
> +
> +/**
> + * spi_nor_s3an_addr_convert() - convert an address to Default Address Mode
> + * @nor: the spi_nor structure
> + * @addr: the contiguous address value
> + *
> + * Spartan-3AN In-System Flash requires an special address mode to access its
> + * whole Flash area. This address mode, named Default Address Mode by Xilinx,
> + * has non power of two page sizes.
> + *
> + * Return: Address in Default Address Mode
> + */
> +unsigned int spi_nor_s3an_addr_convert(struct spi_nor *nor, unsigned int addr);
> +
> /**
> * spi_nor_scan() - scan the SPI NOR
> * @nor: the spi_nor structure
>

Next message: Mike Galbraith: "Re: [tip regression] efi: Allow drivers to reserve boot services forever == toxic"
Previous message: Mark Brown: "Re: [PATCH v4 4/4] regulator: Prevent falling too fast"
In reply to: Ricardo Ribalda Delgado: "Re: [PATCH v4] mtd: spi-nor: Add support for S3AN spi-nor devices"
Next in thread: Ricardo Ribalda Delgado: "Re: [PATCH v4] mtd: spi-nor: Add support for S3AN spi-nor devices"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]