Re: [PATCH 2/3] mtd: spi-nor: core code for the Altera Quadspi Flash Controller v2

From: Cyrille Pitchen
Date: Mon Jul 03 2017 - 20:40:37 EST


Hi Matthew,


Le 26/06/2017 Ã 18:13, matthew.gerlach@xxxxxxxxxxxxxxx a Ãcrit :
> From: Matthew Gerlach <matthew.gerlach@xxxxxxxxxxxxxxx>
>
> Signed-off-by: Matthew Gerlach <matthew.gerlach@xxxxxxxxxxxxxxx>
> ---
> MAINTAINERS | 7 +
> drivers/mtd/spi-nor/Kconfig | 5 +
> drivers/mtd/spi-nor/Makefile | 4 +-
> drivers/mtd/spi-nor/altera-quadspi.c | 676 +++++++++++++++++++++++++++++++++++
> include/linux/mtd/altera-quadspi.h | 28 ++
> 5 files changed, 719 insertions(+), 1 deletion(-)
> create mode 100644 drivers/mtd/spi-nor/altera-quadspi.c
> create mode 100644 include/linux/mtd/altera-quadspi.h
>
> diff --git a/MAINTAINERS b/MAINTAINERS
> index 6b4395c..ae33fa6 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -648,6 +648,13 @@ L: linux-gpio@xxxxxxxxxxxxxxx
> S: Maintained
> F: drivers/gpio/gpio-altera.c
>
> +ALTERA QUADSPI FLASH DRIVER
> +M: Matthew Gerlach <matthew.gerlach@xxxxxxxxxxxxxxx>
> +L: linux-mtd@xxxxxxxxxxxxxxxxxxx
> +S: Maintained
> +F: drivers/mtd/spi-nor/altera-quadspi.c
> +F: inclulde/linux/mtd/altera-quadspi.h
> +
> ALTERA SYSTEM RESOURCE DRIVER FOR ARRIA10 DEVKIT
> M: Thor Thayer <thor.thayer@xxxxxxxxxxxxxxx>
> S: Maintained
> diff --git a/drivers/mtd/spi-nor/Kconfig b/drivers/mtd/spi-nor/Kconfig
> index 293c8a4..89fe425 100644
> --- a/drivers/mtd/spi-nor/Kconfig
> +++ b/drivers/mtd/spi-nor/Kconfig
> @@ -113,4 +113,9 @@ config SPI_STM32_QUADSPI
> This enables support for the STM32 Quad SPI controller.
> We only connect the NOR to this controller.
>
> +config SPI_ALTERA_QUADSPI
> + tristate "Altera Quad SPI Flash Controller II"
> + help
> + Enable support for version 2 of Altera Quad SPI Flash Controller.
> +
> endif # MTD_SPI_NOR
> diff --git a/drivers/mtd/spi-nor/Makefile b/drivers/mtd/spi-nor/Makefile
> index 285aab8..024c6ac 100644
> --- a/drivers/mtd/spi-nor/Makefile
> +++ b/drivers/mtd/spi-nor/Makefile
> @@ -8,4 +8,6 @@ obj-$(CONFIG_MTD_MT81xx_NOR) += mtk-quadspi.o
> obj-$(CONFIG_SPI_NXP_SPIFI) += nxp-spifi.o
> obj-$(CONFIG_SPI_INTEL_SPI) += intel-spi.o
> obj-$(CONFIG_SPI_INTEL_SPI_PLATFORM) += intel-spi-platform.o
> -obj-$(CONFIG_SPI_STM32_QUADSPI) += stm32-quadspi.o
> \ No newline at end of file
> +obj-$(CONFIG_SPI_STM32_QUADSPI) += stm32-quadspi.o
> +obj-$(CONFIG_SPI_ALTERA_QUADSPI) += altera-quadspi.o
> +
> diff --git a/drivers/mtd/spi-nor/altera-quadspi.c b/drivers/mtd/spi-nor/altera-quadspi.c
> new file mode 100644
> index 0000000..de65453
> --- /dev/null
> +++ b/drivers/mtd/spi-nor/altera-quadspi.c
> @@ -0,0 +1,676 @@
> +/*
> + * Copyright (C) 2014 Altera Corporation. All rights reserved.
> + * Copyright (C) 2017 Intel Corporation. All rights reserved.
> + *
> + * This program is free software; you can redistribute it and/or modify it
> + * under the terms and conditions of the GNU General Public License,
> + * version 2, as published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope it will be useful, but WITHOUT
> + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
> + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
> + * more details.
> + *
> + * You should have received a copy of the GNU General Public License along with
> + * this program. If not, see <http://www.gnu.org/licenses/>.
> + */
> +
> +
> +#include <linux/bitrev.h>
> +#include <linux/module.h>
> +#include <linux/mtd/altera-quadspi.h>
> +#include <linux/mtd/mtd.h>
> +#include <linux/mtd/spi-nor.h>
> +
> +#define ALTERA_QUADSPI_RESOURCE_NAME "altera_quadspi"
> +
> +#define EPCS_OPCODE_ID 1
> +#define NON_EPCS_OPCODE_ID 2
> +
> +#define WRITE_CHECK 1
> +#define ERASE_CHECK 0
> +
> +#define QUADSPI_SR_REG 0x0
> +#define QUADSPI_SR_MASK 0x0000000F
> +
> +/* defines for device id register */
> +#define QUADSPI_SID_REG 0x4
> +#define QUADSPI_RDID_REG 0x8
> +#define QUADSPI_ID_MASK 0x000000FF
> +
> +/*
> + * QUADSPI_MEM_OP register offset
> + *
> + * The QUADSPI_MEM_OP register is used to do memory protect and erase operations
> + *
> + */
> +#define QUADSPI_MEM_OP_REG 0xC
> +
> +#define QUADSPI_MEM_OP_CMD_MASK 0x00000003
> +#define QUADSPI_MEM_OP_BULK_ERASE_CMD 0x00000001
> +#define QUADSPI_MEM_OP_SECTOR_ERASE_CMD 0x00000002
> +#define QUADSPI_MEM_OP_SECTOR_PROTECT_CMD 0x00000003
> +#define QUADSPI_MEM_OP_SECTOR_WRITE_ENABLE_CMD 0x00000004
> +#define QUADSPI_MEM_OP_SECTOR_VALUE_MASK 0x0003FF00
> +
> +#define QUADSPI_MEM_OP_SECTOR_PROTECT_SHIFT 8
> +#define QUADSPI_MEM_OP_SECTOR_PROTECT_VALUE_MASK 0x00001F00
> +/*
> + * QUADSPI_ISR register offset
> + *
> + * The QUADSPI_ISR register is used to determine whether an invalid write or
> + * erase operation trigerred an interrupt
> + *
> + */
> +#define QUADSPI_ISR_REG 0x10
> +
> +#define QUADSPI_ISR_ILLEGAL_ERASE_MASK 0x00000001
> +#define QUADSPI_ISR_ILLEGAL_WRITE_MASK 0x00000002
> +
> +/*
> + * QUADSPI_IMR register offset
> + *
> + * The QUADSPI_IMR register is used to mask the invalid erase or the invalid
> + * write interrupts.
> + *
> + */
> +#define QUADSPI_IMR_REG 0x14
> +#define QUADSPI_IMR_ILLEGAL_ERASE_MASK 0x00000001
> +
> +#define QUADSPI_IMR_ILLEGAL_WRITE_MASK 0x00000002
> +
> +#define QUADSPI_CHIP_SELECT_REG 0x18
> +#define QUADSPI_CHIP_SELECT_MASK 0x00000007
> +#define QUADSPI_CHIP_SELECT_0 0x00000001
> +#define QUADSPI_CHIP_SELECT_1 0x00000002
> +#define QUADSPI_CHIP_SELECT_2 0x00000004
> +
> +#define QUADSPI_FLAG_STATUS_REG 0x1C
> +#define QUADSPI_DEV_ID_DATA_0 0x20
> +#define QUADSPI_DEV_ID_DATA_1 0x24
> +#define QUADSPI_DEV_ID_DATA_2 0x28
> +#define QUADSPI_DEV_ID_DATA_3 0x2C
> +#define QUADSPI_DEV_ID_DATA_4 0x30
> +
> +#define QUADSPI_WIN_OCC_REG 0x4
> +#define QUADSPI_WIN_OCC_SFT 24
> +
> +#define QUADSPI_WIN_SEL_REG 0x8
> +
> +struct altera_quadspi {
> + u32 opcode_id;
> + void __iomem *csr_base;
> + void __iomem *data_base;
> + void __iomem *window_base;
> + size_t window_size;
> + u32 num_flashes;
> + u32 flags;
> + struct device *dev;
> + struct altera_quadspi_flash *flash[ALTERA_QUADSPI_MAX_NUM_FLASH_CHIP];
> + struct device_node *np[ALTERA_QUADSPI_MAX_NUM_FLASH_CHIP];
> +};
> +
> +struct altera_quadspi_flash {
> + struct spi_nor nor;
> + struct altera_quadspi *q;
> + u32 bank;
> +};
> +
> +struct flash_device {
> + char *name;
> + u32 opcode_id;
> + u32 device_id;
> +};
> +
> +#ifdef DEBUG
> +static inline u32 alt_qspi_readl(void __iomem *base, off_t offset)
> +{
> + u32 val = readl(base + offset);
> +
> + pr_info("%s 0x%x from offset 0x%lx\n", __func__, val, offset);
> + return val;
> +}
> +static inline void alt_qspi_writel(u32 val, void __iomem *base, off_t offset)
> +{
> + writel(val, base + offset);
> + pr_info("%s 0x%x to offset 0x%lx\n", __func__, val, offset);
> +}
> +#else
> +#define alt_qspi_readl(base, offset) readl(base+offset)
> +#define alt_qspi_writel(val, base, offset) writel(val, base + offset)
> +#endif
> +
> +static void altera_quadspi_chip_select(struct altera_quadspi *q, u32 bank)
> +{
> + u32 val = 0;
> +
> + switch (bank) {
> + case 0:
> + val = QUADSPI_CHIP_SELECT_0;
> + break;
> + case 1:
> + val = QUADSPI_CHIP_SELECT_1;
> + break;
> + case 2:
> + val = QUADSPI_CHIP_SELECT_2;
> + break;
> + default:
> + dev_err(q->dev, "invalid bank\n");
> + return;
> + }
> + alt_qspi_writel(val, q->csr_base, QUADSPI_CHIP_SELECT_REG);
> +}
> +
> +static int altera_quadspi_write_reg(struct spi_nor *nor, u8 opcode, u8 *val,
> + int len)
> +{
> + struct altera_quadspi_flash *flash = nor->priv;
> + struct altera_quadspi *q = flash->q;
> +
> + altera_quadspi_chip_select(q, flash->bank);
> +
> + switch (opcode) {
> + case SPINOR_OP_WREN:
> + dev_dbg(q->dev, "%s enabling write\n", __func__);
> + alt_qspi_writel(QUADSPI_MEM_OP_SECTOR_WRITE_ENABLE_CMD,
> + q->csr_base, QUADSPI_MEM_OP_REG);
> + break;
> +
> + case SPINOR_OP_CHIP_ERASE:
> + alt_qspi_writel(QUADSPI_MEM_OP_BULK_ERASE_CMD,
> + q->csr_base, QUADSPI_MEM_OP_REG);
> + break;
> +
> + default:
> + dev_dbg(q->dev, "%s UNHANDLED write_reg 0x%x\n",
> + __func__, opcode);

Looking at the code I assume, the hardware can only send
predefined/hard-coded SPI op codes but not those actually chosen by
spi-nor.c so what will happen when spi-nor introduces new op codes?
Does it mean that this driver will immediately be broken?

> +
> + }
> +
> + return 0;
> +}
> +
> +static int altera_quadspi_read_reg(struct spi_nor *nor, u8 opcode, u8 *val,
> + int len)
> +{
> + struct altera_quadspi_flash *flash = nor->priv;
> + struct altera_quadspi *q = flash->q;
> + u32 data = 0;
> +
> + memset(val, 0, len);
> +
> + altera_quadspi_chip_select(q, flash->bank);
> +
> + switch (opcode) {
> + case SPINOR_OP_RDSR:
> + data = alt_qspi_readl(q->csr_base, QUADSPI_SR_REG);
> + dev_dbg(q->dev, "%s RDSR 0x%x\n", __func__, data);
> + *val = (u8)data & QUADSPI_SR_MASK;
> + break;
> + case SPINOR_OP_RDID:
> + if (q->opcode_id == EPCS_OPCODE_ID)

Where do you initialize opcode_id ?

> + data = alt_qspi_readl(q->csr_base, QUADSPI_SID_REG);
> + else
> + data = alt_qspi_readl(q->csr_base, QUADSPI_RDID_REG);

Why 2 differents registers for the JEDEC ID? Is data actually the result
of some READ JEDEC ID (9Fh) command or some hard-coded value?

> +
> + *((u32 *)val) = data;
> + break;
> + case SPINOR_OP_RDFSR:
> + data = alt_qspi_readl(q->csr_base, QUADSPI_FLAG_STATUS_REG);
> + dev_dbg(q->dev, "%s RDFSR 0x%x\n", __func__, data);
> + *val = (u8)(data & 0xff);
> + break;

This is a "Micron only" register, other chunks of this patch let me
think that the Altera controller can only work this Micron/EPCS
memories, am I wrong?

OK, then let assume a Micron memory: what if one day spi-nor.c needs to
read the VCR and/or EVCR registers of Micron memories to check some
specific bits? This driver is very likely to fail.

> + default:
> + dev_dbg(q->dev, "%s UNHANDLED read_reg 0x%x\n",
> + __func__, opcode);
> + *val = 0;
> + break;
> + }
> + return 0;
> +}
> +
> +static int altera_quadspi_write_erase_check(struct spi_nor *nor,
> + bool write_erase)
> +{
> + struct altera_quadspi_flash *flash = nor->priv;
> + struct altera_quadspi *q = flash->q;
> + u32 val;
> + u32 mask;
> +
> + if (write_erase)
> + mask = QUADSPI_ISR_ILLEGAL_WRITE_MASK;
> + else
> + mask = QUADSPI_ISR_ILLEGAL_ERASE_MASK;
> +
> + val = alt_qspi_readl(q->csr_base, QUADSPI_ISR_REG);
> +
> + if (val & mask) {
> + dev_err(nor->dev,
> + "write/erase failed, sector might be protected\n");
> + alt_qspi_writel(0, q->csr_base, QUADSPI_FLAG_STATUS_REG);
> +
> + return -EIO;
> + }
> +
> + return 0;
> +}
> +
> +static int altera_quadspi_addr_to_sector(struct mtd_info *mtd, uint64_t offset)
> +{
> + if (mtd->erasesize_shift)
> + return offset >> mtd->erasesize_shift;
> + do_div(offset, mtd->erasesize);
> + return offset;
> +}
> +
> +static int altera_quadspi_erase(struct spi_nor *nor, loff_t offset)
> +{
> + struct altera_quadspi_flash *flash = nor->priv;
> + struct altera_quadspi *q = flash->q;
> + struct mtd_info *mtd = &nor->mtd;
> + u32 val;
> + int sector_value;
> +
> + altera_quadspi_chip_select(q, flash->bank);
> +
> + sector_value = altera_quadspi_addr_to_sector(mtd, offset);

Here the driver uses nor->mtd.erasesize to do some conversion of the
offset into some kind of sector index.
However this function doesn't care about nor->erase_opcode. Then I guess
that the hardware can only use a hard-coded op code for Block/Sector
Erase operations.

I don't know which erase op code is actually used by the hardware but
whatever... How do you know that you've computed the right sector index
before asking the hardware to execute some erase operation since this
index is based on nor->mtd.erasesize ?

nor->mtd.erasesize depends on the chosen nor->erase_opcode.
For instance, depending on whether CONFIG_MTD_SPI_NOR_USE_4K_SECTORS is
defined or not, there will be at least one case where nor->erase_opcode
doesn't match the hard-coded op code used by the Altera controller:
different op codes, different sector sizes, different sector indexes...

buggy nor->erase() function


> +
> + dev_dbg(q->dev, "%s sector %d\n", __func__, sector_value);
> +
> + if (sector_value < 0)
> + return -EINVAL;
> +
> + val = (sector_value << 8) & QUADSPI_MEM_OP_SECTOR_VALUE_MASK;
> +
> + val |= QUADSPI_MEM_OP_SECTOR_ERASE_CMD;
> +
> + alt_qspi_writel(val, q->csr_base, QUADSPI_MEM_OP_REG);
> +
> + dev_dbg(q->dev, "%s SR=0x%x FSR=0x%x\n", __func__,
> + alt_qspi_readl(q->csr_base, QUADSPI_SR_REG),
> + alt_qspi_readl(q->csr_base, QUADSPI_FLAG_STATUS_REG));
> +
> + return altera_quadspi_write_erase_check(nor, ERASE_CHECK);
> +}
> +
> +#define WINDOW_ALIGN 4
> +#define WINDOW_MASK (WINDOW_ALIGN - 1)
> +
> +static ssize_t altera_quadspi_windowed_read(struct altera_quadspi *q,
> + loff_t from,
> + size_t len, u_char *buf)
> +{
> + size_t bytes_left = len;
> + size_t bytes_to_read, i;
> + loff_t next_window_off;
> + u64 start_window;
> + u32 window;
> + u32 *dst;
> +
> + if ((from & WINDOW_MASK) || (len & WINDOW_MASK) ||
> + !IS_ALIGNED((unsigned long)buf, WINDOW_ALIGN)) {
> + dev_err(q->dev, "%s only 32 bit aligned accesses allowed\n",
> + __func__);
> + return 0;
> + }
> +
> + start_window = from;
> + do_div(start_window, q->window_size);
> + window = (u32)(start_window & 0xffffffff);
> +
> + next_window_off = (window + 1) * q->window_size;
> +
> + while (bytes_left > 0) {
> +
> + writel(window, q->window_base + QUADSPI_WIN_SEL_REG);
> +
> + bytes_to_read = min((size_t)bytes_left,
> + (size_t)(next_window_off - from));
> +
> + dev_dbg(q->dev,
> + "window%u fr0x%llx next0x%llx left%zu num0x%zx\n",
> + window, from, next_window_off, bytes_left,
> + bytes_to_read);
> +
> + dst = (u32 *)buf;
> + for (i = 0; i < bytes_to_read; i += 4, dst++)
> + *dst = readl(q->data_base +
> + (from & (q->window_size - 1)) + i);
> +
> + bytes_left -= bytes_to_read;
> + buf += bytes_to_read;
> + from += bytes_to_read;
> + window++;
> + next_window_off += q->window_size;
> + }
> +
> + return len;
> +}
> +static ssize_t altera_quadspi_windowed_write(struct altera_quadspi *q,
> + loff_t to, size_t len,
> + const u_char *buf)
> +{
> + size_t bytes_left = len;
> + u32 window_mask = q->window_size - 1;
> + u32 read_back;
> + size_t bytes_to_write, i;
> + loff_t next_window_off;
> + u64 start_window;
> + u32 window;
> + const u32 *src;
> + u32 words_can_write;
> +
> + if ((to & WINDOW_MASK) || (len & WINDOW_MASK) ||
> + !IS_ALIGNED((unsigned long)buf, WINDOW_ALIGN)) {
> + dev_err(q->dev, "%s only 32 bit aligned accesses allowed\n",
> + __func__);
> + return 0;
> + }
> +
> + start_window = to;
> + do_div(start_window, q->window_size);
> + window = (u32)(start_window & 0xffffffff);
> +
> + next_window_off = (window + 1) * q->window_size;
> +
> + while (bytes_left > 0) {
> +
> + writel(window, q->window_base + QUADSPI_WIN_SEL_REG);
> +
> + bytes_to_write = min((size_t)bytes_left,
> + (size_t)(next_window_off - to));
> +
> + dev_dbg(q->dev,
> + "window%u to0x%llx next0x%llx left%zu num0x%zx\n",
> + window, to, next_window_off, bytes_left,
> + bytes_to_write);
> +
> + src = (u32 *)buf;
> + for (i = 0; i < bytes_to_write;) {
> + words_can_write =
> + readl(q->window_base + QUADSPI_WIN_OCC_REG) >>
> + QUADSPI_WIN_OCC_SFT;
> + dev_dbg(q->dev, "can write 0x%x\n", words_can_write);
> +
> + for (; words_can_write > 0; words_can_write--) {
> + writel(*src,
> + q->data_base +
> + (to & window_mask) + i);
> + read_back = readl(q->data_base +
> + (to & window_mask) + i);
> + if (*src != read_back) {
> + dev_err(q->dev, "%s 0x%x != 0x%x\n",
> + __func__, *src, read_back);
> + return (len - bytes_left);
> + }
> + i += 4;
> + src++;
> + }
> + }
> +
> + bytes_left -= bytes_to_write;
> + buf += bytes_to_write;
> + to += bytes_to_write;
> + window++;
> + next_window_off += q->window_size;
> + }
> +
> + return len;
> +}
> +
> +static ssize_t altera_quadspi_read(struct spi_nor *nor, loff_t from, size_t len,
> + u_char *buf)
> +{
> + struct altera_quadspi_flash *flash = nor->priv;
> + struct altera_quadspi *q = flash->q;
> + size_t i;
> +
> + altera_quadspi_chip_select(q, flash->bank);
> +
> + if (q->window_size)
> + altera_quadspi_windowed_read(q, from, len, buf);
> + else
> + memcpy_fromio(buf, q->data_base + from, len);
> +
> + if (q->flags & ALTERA_QUADSPI_FL_BITREV_READ) {
> + for (i = 0; i < len; i++, buf++)
> + *buf = bitrev8(*buf);
> + }
> +
> + return len;
> +}
> +
> +static ssize_t altera_quadspi_write(struct spi_nor *nor, loff_t to,
> + size_t len, const u_char *buf)
> +{
> + struct altera_quadspi_flash *flash = nor->priv;
> + struct altera_quadspi *q = flash->q;
> + u_char *bitrev_buf = NULL;
> + const u_char *src;
> + u_char *dst;
> + size_t i;
> + int ret = 0;
> +
> + altera_quadspi_chip_select(q, flash->bank);
> +
> + if (q->flags & ALTERA_QUADSPI_FL_BITREV_WRITE) {
> + bitrev_buf = devm_kzalloc(q->dev, len, GFP_KERNEL);
> + if (!bitrev_buf)
> + return 0;
> +
> + src = buf;
> + dst = bitrev_buf;
> + for (i = 0; i < len; i++, src++, dst++)
> + *dst = bitrev8(*src);
> +
> + buf = bitrev_buf;
> + }
> +
> + if (q->window_size)
> + altera_quadspi_windowed_write(q, to, len, buf);
> + else
> + memcpy_toio(q->data_base + to, buf, len);
> +
> +
> + if (bitrev_buf)
> + devm_kfree(q->dev, bitrev_buf);
> +
> + ret = altera_quadspi_write_erase_check(nor, WRITE_CHECK);
> +
> + return len;
> +
> +}
> +
> +static int altera_quadspi_lock(struct spi_nor *nor, loff_t ofs, uint64_t len)
> +{
> + struct altera_quadspi_flash *flash = nor->priv;
> + struct altera_quadspi *q = flash->q;
> + struct mtd_info *mtd = &nor->mtd;
> + uint32_t offset = ofs;
> + u32 sector_start, sector_end;
> + uint64_t num_sectors;
> + u32 mem_op;
> + u32 sr_bp;
> + u32 sr_tb;
> +
> + altera_quadspi_chip_select(q, flash->bank);
> +
> + sector_start = offset;
> + sector_end = altera_quadspi_addr_to_sector(mtd, offset + len);
> + num_sectors = mtd->size;
> + do_div(num_sectors, mtd->erasesize);
> +
> + dev_dbg(nor->dev, "%s: sector start is %u,sector end is %u\n",
> + __func__, sector_start, sector_end);
> +
> + if (sector_start >= num_sectors / 2) {
> + sr_bp = fls(num_sectors - 1 - sector_start) + 1;
> + sr_tb = 0;
> + } else if ((sector_end < num_sectors / 2) &&
> + (q->opcode_id != EPCS_OPCODE_ID)) {
> + sr_bp = fls(sector_end) + 1;
> + sr_tb = 1;
> + } else {
> + sr_bp = 16;
> + sr_tb = 0;
> + }
> +
> + mem_op = (sr_tb << 12) | (sr_bp << 8);
> + mem_op &= QUADSPI_MEM_OP_SECTOR_PROTECT_VALUE_MASK;
> + mem_op |= QUADSPI_MEM_OP_SECTOR_PROTECT_CMD;
> +
> + alt_qspi_writel(mem_op, q->csr_base, QUADSPI_MEM_OP_REG);
> +
> + return 0;
> +}
> +
> +static int altera_quadspi_unlock(struct spi_nor *nor, loff_t ofs, uint64_t len)
> +{
> + struct altera_quadspi_flash *flash = nor->priv;
> + struct altera_quadspi *q = flash->q;
> + u32 mem_op;
> +
> + dev_dbg(nor->dev, "Unlock all protected area\n");
> +
> + altera_quadspi_chip_select(q, flash->bank);
> +
> + mem_op = QUADSPI_MEM_OP_SECTOR_PROTECT_CMD;
> + alt_qspi_writel(mem_op, q->csr_base, QUADSPI_MEM_OP_REG);
> +
> + return 0;
> +}
> +
> +static int altera_quadspi_setup_banks(struct device *dev,
> + u32 bank, struct device_node *np)
> +{
> + struct altera_quadspi *q = dev_get_drvdata(dev);
> + struct altera_quadspi_flash *flash;
> + struct spi_nor *nor;
> + int ret = 0;
> + char modalias[40] = {0};
> + struct spi_nor_hwcaps hwcaps = {
> + .mask = SNOR_HWCAPS_READ |
> + SNOR_HWCAPS_READ_FAST |
> + SNOR_HWCAPS_READ_1_1_2 |
> + SNOR_HWCAPS_READ_1_1_4 |
> + SNOR_HWCAPS_PP,
> + };

since aletera_quadspi_{read|erase} just don't care about
nor->read_opcode, nor->program_opcode and so on and anyway override all
settings chosen by spi-nor.c, it means they will use Dual or Quad SPI
controllers as they want, whether SNOR_HWCAPS_READ_1_1_{2|4} are set or not.
Then I think it's risky to declare the READ_1_1_2 and READ_1_1_4 hwcaps
because it may trigger additionnal calls of nor->read_reg() /
nor->write_reg() from spi_nor_scan() with op codes not supported by
altera_quadspi_{read|write}_reg().

> +
> + if (bank > q->num_flashes - 1)
> + return -EINVAL;
> +
> + altera_quadspi_chip_select(q, bank);
> +
> + flash = devm_kzalloc(q->dev, sizeof(*flash), GFP_KERNEL);
> + if (!flash)
> + return -ENOMEM;
> +
> + q->flash[bank] = flash;
> + nor = &flash->nor;
> + nor->dev = dev;
> + nor->priv = flash;
> + nor->mtd.priv = nor;
> + flash->q = q;
> + flash->bank = bank;
> + spi_nor_set_flash_node(nor, np);
> +
> + /* spi nor framework*/
> + nor->read_reg = altera_quadspi_read_reg;
> + nor->write_reg = altera_quadspi_write_reg;
> + nor->read = altera_quadspi_read;
> + nor->write = altera_quadspi_write;
> + nor->erase = altera_quadspi_erase;
> + nor->flash_lock = altera_quadspi_lock;
> + nor->flash_unlock = altera_quadspi_unlock;

nor->flash_lock and nor->flash_unlock are described as "FLASH SPECIFIC"
in include/linux/mtd/spi-nor.h as opposed to "DRIVER SPECIFIC" functions
like nor->read, nor->read_reg, ...

It means the actual implementations should be provided by the spi-nor
sub-system but not by each SPI controller driver.



For me, it really sounds like a bad idea that this driver tries so much
to mystify the spi-nor sub-system.

I can understand that you have to cope with the hardware design and its
limitations but clearly it looks the spi-nor API is not suited to this
hardware. This driver ignores and by-passes any settings selected by
spi_nor_scan().
Duplicating code is generally a bad idea but in this case, I don't know
if trying to reuse spi_nor_read() / spi_nor_write() and spi_nor_erase()
from spi-nor.c is that helpful.

Why not directly plug your driver into the above mtd layer implementing
you own version of mtd->_read(), mtd->_write() and mtd->_erase() then
registering the mtd device? It may be not the way to go but at least we
should study this alternative.

*IF* I have understood the hardware constraints of the Altera Quad SPI
controller and what this driver tries to discover by calling
spi_nor_scan(), I think you actually don't need spi_nor_scan() but only
to have access to the spi_nor_ids[] array to initialize mtd->erasesize
with info->sector_size. Am I right?

Then why not just exporting spi_nor_ids[] so your driver could directly
search into this table to get the pieces of information it needs?

I'm not confortable with the idea of a driver pretending to be compliant
with the API of spi-nor when obviously it is not.

Even if the Altera hardware looks really limited, I think we can still
try to help you finding a solution to add its support into mainline but
I would like to find a solution with less "hacks" because I'm pretty
sure that if we let pass a driver like this one, it would often be
broken and it would be a real pain to maintain.

Besides, I generally claim that SPI controller drivers should not use
SPINOR_OP_* macros because when they do so, they almost always try to
implement some kind of hack/by-pass of the spi-nor sub-system and the
result is maintainance issues when we update or add new features in
spi-nor.{c|h}.

IMHO, this driver goes in the wrong direction. Marek, do have an idea of
a better/cleaner solution to add support to Altera hardware in mainline?

Best regards,

Cyrille

> +
> + /* scanning flash and checking dev id */
> +#ifdef CONFIG_OF
> + if (np && (of_modalias_node(np, modalias, sizeof(modalias)) < 0))
> + return -EINVAL;
> +#endif
> +
> + ret = spi_nor_scan(nor, modalias, &hwcaps);
> + if (ret) {
> + dev_err(nor->dev, "flash not found\n");
> + return ret;
> + }
> +
> + ret = mtd_device_register(&nor->mtd, NULL, 0);
> +
> + altera_quadspi_unlock(nor, 0, 0);
> +
> + return ret;
> +}
> +
> +int altera_quadspi_create(struct device *dev, void __iomem *csr_base,
> + void __iomem *data_base, void __iomem *window_base,
> + size_t window_size, u32 flags)
> +{
> + struct altera_quadspi *q;
> +
> + q = devm_kzalloc(dev, sizeof(*q), GFP_KERNEL);
> + if (!q)
> + return -ENOMEM;
> +
> + q->dev = dev;
> + q->csr_base = csr_base;
> + q->data_base = data_base;
> + q->window_base = window_base;
> + q->window_size = window_size;
> +
> + q->flags = flags;
> +
> + dev_set_drvdata(dev, q);
> +
> + dev_dbg(dev, "%s SR=0x%x FSR=0x%x\n", __func__,
> + alt_qspi_readl(q->csr_base, QUADSPI_SR_REG),
> + alt_qspi_readl(q->csr_base, QUADSPI_FLAG_STATUS_REG));
> +
> + return 0;
> +}
> +EXPORT_SYMBOL_GPL(altera_quadspi_create);
> +
> +int altera_qspi_add_bank(struct device *dev,
> + u32 bank, struct device_node *np)
> +{
> + struct altera_quadspi *q = dev_get_drvdata(dev);
> +
> + if (q->num_flashes >= ALTERA_QUADSPI_MAX_NUM_FLASH_CHIP)
> + return -ENOMEM;
> +
> + q->num_flashes++;
> +
> + return altera_quadspi_setup_banks(dev, bank, np);
> +}
> +EXPORT_SYMBOL_GPL(altera_qspi_add_bank);
> +
> +int altera_quadspi_remove_banks(struct device *dev)
> +{
> + struct altera_quadspi *q = dev_get_drvdata(dev);
> + struct altera_quadspi_flash *flash;
> + int i;
> + int ret = 0;
> +
> + /* clean up for all nor flash */
> + for (i = 0; i < q->num_flashes; i++) {
> + flash = q->flash[i];
> + if (!flash)
> + continue;
> +
> + /* clean up mtd stuff */
> + ret = mtd_device_unregister(&flash->nor.mtd);
> + if (ret) {
> + dev_err(dev, "error removing mtd\n");
> + return ret;
> + }
> + }
> +
> + return 0;
> +}
> +EXPORT_SYMBOL_GPL(altera_quadspi_remove_banks);
> +
> +MODULE_AUTHOR("Viet Nga Dao <vndao@xxxxxxxxxx>");
> +MODULE_AUTHOR("Yong Sern Lau <lau.yong.sern@xxxxxxxxx>");
> +MODULE_AUTHOR("Matthew Gerlach <matthew.gerlach@xxxxxxxxxxxxxxx>");
> +MODULE_DESCRIPTION("Altera QuadSPI Version 2 Driver");
> +MODULE_LICENSE("GPL v2");
> diff --git a/include/linux/mtd/altera-quadspi.h b/include/linux/mtd/altera-quadspi.h
> new file mode 100644
> index 0000000..58f31ee
> --- /dev/null
> +++ b/include/linux/mtd/altera-quadspi.h
> @@ -0,0 +1,28 @@
> +/*
> + *
> + * Copyright 2017 Intel Corporation, Inc.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> + */
> +#ifndef __ALTERA_QUADSPI_H
> +#define __ALTERA_QUADSPI_H
> +
> +#include <linux/device.h>
> +
> +#define ALTERA_QUADSPI_FL_BITREV_READ BIT(0)
> +#define ALTERA_QUADSPI_FL_BITREV_WRITE BIT(1)
> +
> +#define ALTERA_QUADSPI_MAX_NUM_FLASH_CHIP 3
> +
> +int altera_quadspi_create(struct device *dev, void __iomem *csr_base,
> + void __iomem *data_base, void __iomem *window_reg,
> + size_t window_size, u32 flags);
> +
> +int altera_qspi_add_bank(struct device *dev,
> + u32 bank, struct device_node *np);
> +
> +int altera_quadspi_remove_banks(struct device *dev);
> +#endif
>