[PATCH 0/2] mtd: spi-nor: add driver for Atmel QSPI controller

[PATCH 0/2] mtd: spi-nor: add driver for Atmel QSPI controller

[Cyrille Pitchen]

Hi all,

this series of patches adds support to the Atmel QSPI controller.
It was tested on a sama5d2 xplained board + Macronix mx25l25673g QSPI
memory.

The driver limits its support of Quad SPI memories to the SPI 1-1-4
protocol for Fast Read operations and SPI 1-1-1 protocol for all other
commands.

The series is based on next-20160609

Best regards,

Cyrille

Cyrille Pitchen (2):
  Documentation: atmel-quadspi: add binding file for Atmel QSPI driver
  mtd: atmel-quadspi: add driver for Atmel QSPI controller

 .../devicetree/bindings/mtd/atmel-quadspi.txt      |  32 +
 drivers/mtd/spi-nor/Kconfig                        |   9 +
 drivers/mtd/spi-nor/Makefile                       |   1 +
 drivers/mtd/spi-nor/atmel-quadspi.c                | 741 +++++++++++++++++++++
 4 files changed, 783 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/mtd/atmel-quadspi.txt
 create mode 100644 drivers/mtd/spi-nor/atmel-quadspi.c

-- 
1.8.2.2

[PATCH 1/2] Documentation: atmel-quadspi: add binding file for Atmel QSPI driver

[Cyrille Pitchen]

This patch documents the DT bindings for the driver of the Atmel QSPI
controller embedded inside sama5d2x SoCs.

Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com>
Acked-by: Rob Herring <robh@kernel.org>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
 .../devicetree/bindings/mtd/atmel-quadspi.txt      | 32 ++++++++++++++++++++++
 1 file changed, 32 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/mtd/atmel-quadspi.txt

diff --git a/Documentation/devicetree/bindings/mtd/atmel-quadspi.txt b/Documentation/devicetree/bindings/mtd/atmel-quadspi.txt
new file mode 100644
index 000000000000..489807005eda
--- /dev/null
+++ b/Documentation/devicetree/bindings/mtd/atmel-quadspi.txt
@@ -0,0 +1,32 @@
+* Atmel Quad Serial Peripheral Interface (QSPI)
+
+Required properties:
+- compatible:     Should be "atmel,sama5d2-qspi".
+- reg:            Should contain the locations and lengths of the base registers
+                  and the mapped memory.
+- reg-names:      Should contain the resource reg names:
+                  - qspi_base: configuration register address space
+                  - qspi_mmap: memory mapped address space
+- interrupts:     Should contain the interrupt for the device.
+- clocks:         The phandle of the clock needed by the QSPI controller.
+- #address-cells: Should be <1>.
+- #size-cells:    Should be <0>.
+
+Example:
+
+spi@f0020000 {
+	compatible = "atmel,sama5d2-qspi";
+	reg = <0xf0020000 0x100>, <0xd0000000 0x8000000>;
+	reg-names = "qspi_base", "qspi_mmap";
+	interrupts = <52 IRQ_TYPE_LEVEL_HIGH 7>;
+	clocks = <&spi0_clk>;
+	#address-cells = <1>;
+	#size-cells = <0>;
+	pinctrl-names = "default";
+	pinctrl-0 = <&pinctrl_spi0_default>;
+	status = "okay";
+
+	m25p80@0 {
+		...
+	};
+};
-- 
1.8.2.2

Re: [PATCH 1/2] Documentation: atmel-quadspi: add binding file for Atmel QSPI driver

[Brian Norris]

On Mon, Jun 13, 2016 at 05:10:25PM +0200, Cyrille Pitchen wrote:
> This patch documents the DT bindings for the driver of the Atmel QSPI
> controller embedded inside sama5d2x SoCs.
> 
> Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com>
> Acked-by: Rob Herring <robh@kernel.org>
> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>

Pushed this patch to l2-mtd.git. Comments on the other.

[PATCH 2/2] mtd: atmel-quadspi: add driver for Atmel QSPI controller

[Cyrille Pitchen]

This driver add support to the new Atmel QSPI controller embedded into
sama5d2x SoCs. It expects a NOR memory to be connected to the QSPI
controller.

Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
 drivers/mtd/spi-nor/Kconfig         |   9 +
 drivers/mtd/spi-nor/Makefile        |   1 +
 drivers/mtd/spi-nor/atmel-quadspi.c | 741 ++++++++++++++++++++++++++++++++++++
 3 files changed, 751 insertions(+)
 create mode 100644 drivers/mtd/spi-nor/atmel-quadspi.c

diff --git a/drivers/mtd/spi-nor/Kconfig b/drivers/mtd/spi-nor/Kconfig
index d42c98e1f581..c546efd1357b 100644
--- a/drivers/mtd/spi-nor/Kconfig
+++ b/drivers/mtd/spi-nor/Kconfig
@@ -38,6 +38,15 @@ config SPI_FSL_QUADSPI
 	  This controller does not support generic SPI. It only supports
 	  SPI NOR.
 
+config SPI_ATMEL_QUADSPI
+	tristate "Atmel Quad SPI Controller"
+	depends on ARCH_AT91 || (ARM && COMPILE_TEST)
+	depends on OF && HAS_IOMEM
+	help
+	  This enables support for the Quad SPI controller in master mode.
+	  This driver does not support generic SPI. The implementation only
+	  supports SPI NOR.
+
 config SPI_NXP_SPIFI
 	tristate "NXP SPI Flash Interface (SPIFI)"
 	depends on OF && (ARCH_LPC18XX || COMPILE_TEST)
diff --git a/drivers/mtd/spi-nor/Makefile b/drivers/mtd/spi-nor/Makefile
index 0bf3a7f81675..1525913698ad 100644
--- a/drivers/mtd/spi-nor/Makefile
+++ b/drivers/mtd/spi-nor/Makefile
@@ -2,3 +2,4 @@ obj-$(CONFIG_MTD_SPI_NOR)	+= spi-nor.o
 obj-$(CONFIG_SPI_FSL_QUADSPI)	+= fsl-quadspi.o
 obj-$(CONFIG_MTD_MT81xx_NOR)    += mtk-quadspi.o
 obj-$(CONFIG_SPI_NXP_SPIFI)	+= nxp-spifi.o
+obj-$(CONFIG_SPI_ATMEL_QUADSPI)	+= atmel-quadspi.o
diff --git a/drivers/mtd/spi-nor/atmel-quadspi.c b/drivers/mtd/spi-nor/atmel-quadspi.c
new file mode 100644
index 000000000000..06d1bf276dd0
--- /dev/null
+++ b/drivers/mtd/spi-nor/atmel-quadspi.c
@@ -0,0 +1,741 @@
+/*
+ * Driver for Atmel QSPI Controller
+ *
+ * Copyright (C) 2015 Atmel Corporation
+ *
+ * Author: Cyrille Pitchen <cyrille.pitchen@atmel.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that 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/>.
+ *
+ * This driver is based on drivers/mtd/spi-nor/fsl-quadspi.c from Freescale.
+ */
+
+#include <linux/kernel.h>
+#include <linux/clk.h>
+#include <linux/module.h>
+#include <linux/platform_device.h>
+#include <linux/delay.h>
+#include <linux/err.h>
+#include <linux/interrupt.h>
+#include <linux/mtd/mtd.h>
+#include <linux/mtd/partitions.h>
+#include <linux/mtd/spi-nor.h>
+#include <linux/platform_data/atmel.h>
+#include <linux/of.h>
+
+#include <linux/io.h>
+#include <linux/gpio.h>
+#include <linux/pinctrl/consumer.h>
+
+/* QSPI register offsets */
+#define QSPI_CR      0x0000  /* Control Register */
+#define QSPI_MR      0x0004  /* Mode Register */
+#define QSPI_RD      0x0008  /* Receive Data Register */
+#define QSPI_TD      0x000c  /* Transmit Data Register */
+#define QSPI_SR      0x0010  /* Status Register */
+#define QSPI_IER     0x0014  /* Interrupt Enable Register */
+#define QSPI_IDR     0x0018  /* Interrupt Disable Register */
+#define QSPI_IMR     0x001c  /* Interrupt Mask Register */
+#define QSPI_SCR     0x0020  /* Serial Clock Register */
+
+#define QSPI_IAR     0x0030  /* Instruction Address Register */
+#define QSPI_ICR     0x0034  /* Instruction Code Register */
+#define QSPI_IFR     0x0038  /* Instruction Frame Register */
+
+#define QSPI_SMR     0x0040  /* Scrambling Mode Register */
+#define QSPI_SKR     0x0044  /* Scrambling Key Register */
+
+#define QSPI_WPMR    0x00E4  /* Write Protection Mode Register */
+#define QSPI_WPSR    0x00E8  /* Write Protection Status Register */
+
+#define QSPI_VERSION 0x00FC  /* Version Register */
+
+
+/* Bitfields in QSPI_CR (Control Register) */
+#define QSPI_CR_QSPIEN                  BIT(0)
+#define QSPI_CR_QSPIDIS                 BIT(1)
+#define QSPI_CR_SWRST                   BIT(7)
+#define QSPI_CR_LASTXFER                BIT(24)
+
+/* Bitfields in QSPI_MR (Mode Register) */
+#define QSPI_MR_SSM                     BIT(0)
+#define QSPI_MR_LLB                     BIT(1)
+#define QSPI_MR_WDRBT                   BIT(2)
+#define QSPI_MR_SMRM                    BIT(3)
+#define QSPI_MR_CSMODE_MASK             GENMASK(5, 4)
+#define QSPI_MR_CSMODE_NOT_RELOADED     (0 << 4)
+#define QSPI_MR_CSMODE_LASTXFER         (1 << 4)
+#define QSPI_MR_CSMODE_SYSTEMATICALLY   (2 << 4)
+#define QSPI_MR_NBBITS_MASK             GENMASK(11, 8)
+#define QSPI_MR_NBBITS(n)               ((((n) - 8) << 8) & QSPI_MR_NBBITS_MASK)
+#define QSPI_MR_DLYBCT_MASK             GENMASK(23, 16)
+#define QSPI_MR_DLYBCT(n)               (((n) << 16) & QSPI_MR_DLYBCT_MASK)
+#define QSPI_MR_DLYCS_MASK              GENMASK(31, 24)
+#define QSPI_MR_DLYCS(n)                (((n) << 24) & QSPI_MR_DLYCS_MASK)
+
+/* Bitfields in QSPI_SR/QSPI_IER/QSPI_IDR/QSPI_IMR  */
+#define QSPI_SR_RDRF                    BIT(0)
+#define QSPI_SR_TDRE                    BIT(1)
+#define QSPI_SR_TXEMPTY                 BIT(2)
+#define QSPI_SR_OVRES                   BIT(3)
+#define QSPI_SR_CSR                     BIT(8)
+#define QSPI_SR_CSS                     BIT(9)
+#define QSPI_SR_INSTRE                  BIT(10)
+#define QSPI_SR_QSPIENS                 BIT(24)
+
+#define QSPI_SR_CMD_COMPLETED	(QSPI_SR_INSTRE | QSPI_SR_CSR)
+
+/* Bitfields in QSPI_SCR (Serial Clock Register) */
+#define QSPI_SCR_CPOL                   BIT(0)
+#define QSPI_SCR_CPHA                   BIT(1)
+#define QSPI_SCR_SCBR_MASK              GENMASK(15, 8)
+#define QSPI_SCR_SCBR(n)                (((n) << 8) & QSPI_SCR_SCBR_MASK)
+#define QSPI_SCR_DLYBS_MASK             GENMASK(23, 16)
+#define QSPI_SCR_DLYBS(n)               (((n) << 16) & QSPI_SCR_DLYBS_MASK)
+
+/* Bitfields in QSPI_ICR (Instruction Code Register) */
+#define QSPI_ICR_INST_MASK              GENMASK(7, 0)
+#define QSPI_ICR_INST(inst)             (((inst) << 0) & QSPI_ICR_INST_MASK)
+#define QSPI_ICR_OPT_MASK               GENMASK(23, 16)
+#define QSPI_ICR_OPT(opt)               (((opt) << 16) & QSPI_ICR_OPT_MASK)
+
+/* Bitfields in QSPI_IFR (Instruction Frame Register) */
+#define QSPI_IFR_WIDTH_MASK             GENMASK(2, 0)
+#define QSPI_IFR_WIDTH_SINGLE_BIT_SPI   (0 << 0)
+#define QSPI_IFR_WIDTH_DUAL_OUTPUT      (1 << 0)
+#define QSPI_IFR_WIDTH_QUAD_OUTPUT      (2 << 0)
+#define QSPI_IFR_WIDTH_DUAL_IO          (3 << 0)
+#define QSPI_IFR_WIDTH_QUAD_IO          (4 << 0)
+#define QSPI_IFR_WIDTH_DUAL_CMD         (5 << 0)
+#define QSPI_IFR_WIDTH_QUAD_CMD         (6 << 0)
+#define QSPI_IFR_INSTEN                 BIT(4)
+#define QSPI_IFR_ADDREN                 BIT(5)
+#define QSPI_IFR_OPTEN                  BIT(6)
+#define QSPI_IFR_DATAEN                 BIT(7)
+#define QSPI_IFR_OPTL_MASK              GENMASK(9, 8)
+#define QSPI_IFR_OPTL_1BIT              (0 << 8)
+#define QSPI_IFR_OPTL_2BIT              (1 << 8)
+#define QSPI_IFR_OPTL_4BIT              (2 << 8)
+#define QSPI_IFR_OPTL_8BIT              (3 << 8)
+#define QSPI_IFR_ADDRL                  BIT(10)
+#define QSPI_IFR_TFRTYP_MASK            GENMASK(13, 12)
+#define QSPI_IFR_TFRTYP_TRSFR_READ      (0 << 12)
+#define QSPI_IFR_TFRTYP_TRSFR_READ_MEM  (1 << 12)
+#define QSPI_IFR_TFRTYP_TRSFR_WRITE     (2 << 12)
+#define QSPI_IFR_TFRTYP_TRSFR_WRITE_MEM (3 << 13)
+#define QSPI_IFR_CRM                    BIT(14)
+#define QSPI_IFR_NBDUM_MASK             GENMASK(20, 16)
+#define QSPI_IFR_NBDUM(n)               (((n) << 16) & QSPI_IFR_NBDUM_MASK)
+
+/* Bitfields in QSPI_SMR (Scrambling Mode Register) */
+#define QSPI_SMR_SCREN                  BIT(0)
+#define QSPI_SMR_RVDIS                  BIT(1)
+
+/* Bitfields in QSPI_WPMR (Write Protection Mode Register) */
+#define QSPI_WPMR_WPEN                  BIT(0)
+#define QSPI_WPMR_WPKEY_MASK            GENMASK(31, 8)
+#define QSPI_WPMR_WPKEY(wpkey)          (((wpkey) << 8) & QSPI_WPMR_WPKEY_MASK)
+
+/* Bitfields in QSPI_WPSR (Write Protection Status Register) */
+#define QSPI_WPSR_WPVS                  BIT(0)
+#define QSPI_WPSR_WPVSRC_MASK           GENMASK(15, 8)
+#define QSPI_WPSR_WPVSRC(src)           (((src) << 8) & QSPI_WPSR_WPVSRC)
+
+
+struct atmel_qspi {
+	void __iomem		*regs;
+	void __iomem		*mem;
+	struct clk		*clk;
+	struct platform_device	*pdev;
+	u32			pending;
+
+	struct spi_nor		nor;
+	u32			clk_rate;
+	struct completion	cmd_completion;
+};
+
+struct atmel_qspi_command {
+	union {
+		struct {
+			u32	instruction:1;
+			u32	address:3;
+			u32	mode:1;
+			u32	dummy:1;
+			u32	data:1;
+			u32	reserved:25;
+		}		bits;
+		u32	word;
+	}	enable;
+	u8	instruction;
+	u8	mode;
+	u8	num_mode_cycles;
+	u8	num_dummy_cycles;
+	u32	address;
+
+	size_t		buf_len;
+	const void	*tx_buf;
+	void		*rx_buf;
+};
+
+/* Register access functions */
+static inline u32 qspi_readl(struct atmel_qspi *aq, u32 reg)
+{
+	return readl_relaxed(aq->regs + reg);
+}
+
+static inline void qspi_writel(struct atmel_qspi *aq, u32 reg, u32 value)
+{
+	writel_relaxed(value, aq->regs + reg);
+}
+
+static int atmel_qspi_run_transfer(struct atmel_qspi *aq,
+				   const struct atmel_qspi_command *cmd)
+{
+	void __iomem *ahb_mem;
+
+	/* Then fallback to a PIO transfer (memcpy() DOES NOT work!) */
+	ahb_mem = aq->mem;
+	if (cmd->enable.bits.address)
+		ahb_mem += cmd->address;
+	if (cmd->tx_buf)
+		_memcpy_toio(ahb_mem, cmd->tx_buf, cmd->buf_len);
+	else
+		_memcpy_fromio(cmd->rx_buf, ahb_mem, cmd->buf_len);
+
+	return 0;
+}
+
+#ifdef DEBUG
+static void atmel_qspi_debug_command(struct atmel_qspi *aq,
+				     const struct atmel_qspi_command *cmd,
+				     u32 ifr)
+{
+	u8 cmd_buf[SPI_NOR_MAX_CMD_SIZE];
+	size_t len = 0;
+	int i;
+
+	if (cmd->enable.bits.instruction)
+		cmd_buf[len++] = cmd->instruction;
+
+	for (i = cmd->enable.bits.address-1; i >= 0; --i)
+		cmd_buf[len++] = (cmd->address >> (i << 3)) & 0xff;
+
+	if (cmd->enable.bits.mode)
+		cmd_buf[len++] = cmd->mode;
+
+	if (cmd->enable.bits.dummy) {
+		int num = cmd->num_dummy_cycles;
+
+		switch (ifr & QSPI_IFR_WIDTH_MASK) {
+		case QSPI_IFR_WIDTH_SINGLE_BIT_SPI:
+		case QSPI_IFR_WIDTH_DUAL_OUTPUT:
+		case QSPI_IFR_WIDTH_QUAD_OUTPUT:
+			num >>= 3;
+			break;
+		case QSPI_IFR_WIDTH_DUAL_IO:
+		case QSPI_IFR_WIDTH_DUAL_CMD:
+			num >>= 2;
+			break;
+		case QSPI_IFR_WIDTH_QUAD_IO:
+		case QSPI_IFR_WIDTH_QUAD_CMD:
+			num >>= 1;
+			break;
+		default:
+			return;
+		}
+
+		for (i = 0; i < num; ++i)
+			cmd_buf[len++] = 0;
+	}
+
+	/* Dump the SPI command */
+	print_hex_dump(KERN_DEBUG, "qspi cmd: ", DUMP_PREFIX_NONE,
+		       32, 1, cmd_buf, len, false);
+
+#ifdef VERBOSE_DEBUG
+	/* If verbose debug is enabled, also dump the TX data */
+	if (cmd->enable.bits.data && cmd->tx_buf)
+		print_hex_dump(KERN_DEBUG, "qspi tx : ", DUMP_PREFIX_NONE,
+			       32, 1, cmd->tx_buf, cmd->buf_len, false);
+#endif
+}
+#else
+#define atmel_qspi_debug_command(aq, cmd, ifr)
+#endif
+
+static int atmel_qspi_run_command(struct atmel_qspi *aq,
+				  const struct atmel_qspi_command *cmd,
+				  u32 ifr_tfrtyp, u32 ifr_width)
+{
+	u32 iar, icr, ifr, sr;
+	int err = 0;
+
+	iar = 0;
+	icr = 0;
+	ifr = ifr_tfrtyp | ifr_width;
+
+	/* Compute instruction parameters */
+	if (cmd->enable.bits.instruction) {
+		icr |= QSPI_ICR_INST(cmd->instruction);
+		ifr |= QSPI_IFR_INSTEN;
+	}
+
+	/* Compute address parameters */
+	switch (cmd->enable.bits.address) {
+	case 4:
+		ifr |= QSPI_IFR_ADDRL;
+		/* fall through to the 24bit (3 byte) address case. */
+	case 3:
+		iar = (cmd->enable.bits.data) ? 0 : cmd->address;
+		ifr |= QSPI_IFR_ADDREN;
+		break;
+	case 0:
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	/* Compute option parameters */
+	if (cmd->enable.bits.mode && cmd->num_mode_cycles) {
+		u32 mode_cycle_bits, mode_bits;
+
+		icr |= QSPI_ICR_OPT(cmd->mode);
+		ifr |= QSPI_IFR_OPTEN;
+
+		switch (ifr & QSPI_IFR_WIDTH_MASK) {
+		case QSPI_IFR_WIDTH_SINGLE_BIT_SPI:
+		case QSPI_IFR_WIDTH_DUAL_OUTPUT:
+		case QSPI_IFR_WIDTH_QUAD_OUTPUT:
+			mode_cycle_bits = 1;
+			break;
+		case QSPI_IFR_WIDTH_DUAL_IO:
+		case QSPI_IFR_WIDTH_DUAL_CMD:
+			mode_cycle_bits = 2;
+			break;
+		case QSPI_IFR_WIDTH_QUAD_IO:
+		case QSPI_IFR_WIDTH_QUAD_CMD:
+			mode_cycle_bits = 4;
+			break;
+		default:
+			return -EINVAL;
+		}
+
+		mode_bits = cmd->num_mode_cycles * mode_cycle_bits;
+		switch (mode_bits) {
+		case 1:
+			ifr |= QSPI_IFR_OPTL_1BIT;
+			break;
+
+		case 2:
+			ifr |= QSPI_IFR_OPTL_2BIT;
+			break;
+
+		case 4:
+			ifr |= QSPI_IFR_OPTL_4BIT;
+			break;
+
+		case 8:
+			ifr |= QSPI_IFR_OPTL_8BIT;
+			break;
+
+		default:
+			return -EINVAL;
+		}
+	}
+
+	/* Set number of dummy cycles */
+	if (cmd->enable.bits.dummy)
+		ifr |= QSPI_IFR_NBDUM(cmd->num_dummy_cycles);
+
+	/* Set data enable */
+	if (cmd->enable.bits.data) {
+		ifr |= QSPI_IFR_DATAEN;
+
+		/* Special case for Continuous Read Mode */
+		if (!cmd->tx_buf && !cmd->rx_buf)
+			ifr |= QSPI_IFR_CRM;
+	}
+
+	/* Clear pending interrupts */
+	(void)qspi_readl(aq, QSPI_SR);
+
+	/* Set QSPI Instruction Frame registers */
+	atmel_qspi_debug_command(aq, cmd, ifr);
+	qspi_writel(aq, QSPI_IAR, iar);
+	qspi_writel(aq, QSPI_ICR, icr);
+	qspi_writel(aq, QSPI_IFR, ifr);
+
+	/* Skip to the final steps if there is no data */
+	if (!cmd->enable.bits.data)
+		goto no_data;
+
+	/* Dummy read of QSPI_IFR to synchronize APB and AHB accesses */
+	(void)qspi_readl(aq, QSPI_IFR);
+
+	/* Stop here for continuous read */
+	if (!cmd->tx_buf && !cmd->rx_buf)
+		return 0;
+	/* Send/Receive data */
+	err = atmel_qspi_run_transfer(aq, cmd);
+
+	/* Release the chip-select */
+	qspi_writel(aq, QSPI_CR, QSPI_CR_LASTXFER);
+
+	if (err)
+		return err;
+
+#if defined(DEBUG) && defined(VERBOSE_DEBUG)
+	/*
+	 * If verbose debug is enabled, also dump the RX data in addition to
+	 * the SPI command previously dumped by atmel_qspi_debug_command()
+	 */
+	if (cmd->rx_buf)
+		print_hex_dump(KERN_DEBUG, "qspi rx : ", DUMP_PREFIX_NONE,
+			       32, 1, cmd->rx_buf, cmd->buf_len, false);
+#endif
+no_data:
+	/* Poll INSTRuction End status */
+	sr = qspi_readl(aq, QSPI_SR);
+	if ((sr & QSPI_SR_CMD_COMPLETED) == QSPI_SR_CMD_COMPLETED)
+		return err;
+
+	/* Wait for INSTRuction End interrupt */
+	reinit_completion(&aq->cmd_completion);
+	aq->pending = sr & QSPI_SR_CMD_COMPLETED;
+	qspi_writel(aq, QSPI_IER, QSPI_SR_CMD_COMPLETED);
+	if (!wait_for_completion_timeout(&aq->cmd_completion,
+					 msecs_to_jiffies(1000)))
+		err = -ETIMEDOUT;
+	qspi_writel(aq, QSPI_IDR, QSPI_SR_CMD_COMPLETED);
+
+	return err;
+}
+
+static int atmel_qspi_read_reg(struct spi_nor *nor, u8 opcode,
+			       u8 *buf, int len)
+{
+	struct atmel_qspi *aq = nor->priv;
+	struct atmel_qspi_command cmd;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.enable.bits.instruction = 1;
+	cmd.enable.bits.data = 1;
+	cmd.instruction = opcode;
+	cmd.rx_buf = buf;
+	cmd.buf_len = len;
+	return atmel_qspi_run_command(aq, &cmd, QSPI_IFR_TFRTYP_TRSFR_READ,
+				      QSPI_IFR_WIDTH_SINGLE_BIT_SPI);
+}
+
+static int atmel_qspi_write_reg(struct spi_nor *nor, u8 opcode,
+				u8 *buf, int len)
+{
+	struct atmel_qspi *aq = nor->priv;
+	struct atmel_qspi_command cmd;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.enable.bits.instruction = 1;
+	cmd.enable.bits.data = (buf != NULL && len > 0);
+	cmd.instruction = opcode;
+	cmd.tx_buf = buf;
+	cmd.buf_len = len;
+	return atmel_qspi_run_command(aq, &cmd, QSPI_IFR_TFRTYP_TRSFR_WRITE,
+				      QSPI_IFR_WIDTH_SINGLE_BIT_SPI);
+}
+
+static ssize_t atmel_qspi_write(struct spi_nor *nor, loff_t to, size_t len,
+				const u_char *write_buf)
+{
+	struct atmel_qspi *aq = nor->priv;
+	struct atmel_qspi_command cmd;
+	ssize_t ret;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.enable.bits.instruction = 1;
+	cmd.enable.bits.address = nor->addr_width;
+	cmd.enable.bits.data = 1;
+	cmd.instruction = nor->program_opcode;
+	cmd.address = (u32)to;
+	cmd.tx_buf = write_buf;
+	cmd.buf_len = len;
+	ret = atmel_qspi_run_command(aq, &cmd, QSPI_IFR_TFRTYP_TRSFR_WRITE_MEM,
+				     QSPI_IFR_WIDTH_SINGLE_BIT_SPI);
+	return (ret < 0) ? ret : len;
+}
+
+static int atmel_qspi_erase(struct spi_nor *nor, loff_t offs)
+{
+	struct atmel_qspi *aq = nor->priv;
+	struct atmel_qspi_command cmd;
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.enable.bits.instruction = 1;
+	cmd.enable.bits.address = nor->addr_width;
+	cmd.instruction = nor->erase_opcode;
+	cmd.address = (u32)offs;
+	return atmel_qspi_run_command(aq, &cmd, QSPI_IFR_TFRTYP_TRSFR_WRITE,
+				      QSPI_IFR_WIDTH_SINGLE_BIT_SPI);
+}
+
+static ssize_t atmel_qspi_read(struct spi_nor *nor, loff_t from, size_t len,
+			       u_char *read_buf)
+{
+	struct atmel_qspi *aq = nor->priv;
+	struct atmel_qspi_command cmd;
+	u8 num_mode_cycles, num_dummy_cycles;
+	u32 ifr_width;
+	ssize_t ret;
+
+	switch (nor->flash_read) {
+	case SPI_NOR_NORMAL:
+	case SPI_NOR_FAST:
+		ifr_width = QSPI_IFR_WIDTH_SINGLE_BIT_SPI;
+		break;
+
+	case SPI_NOR_DUAL:
+		ifr_width = QSPI_IFR_WIDTH_DUAL_OUTPUT;
+		break;
+
+	case SPI_NOR_QUAD:
+		ifr_width = QSPI_IFR_WIDTH_QUAD_OUTPUT;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	if (nor->read_dummy >= 2) {
+		num_mode_cycles = 2;
+		num_dummy_cycles = nor->read_dummy - 2;
+	} else {
+		num_mode_cycles = nor->read_dummy;
+		num_dummy_cycles = 0;
+	}
+
+	memset(&cmd, 0, sizeof(cmd));
+	cmd.enable.bits.instruction = 1;
+	cmd.enable.bits.address = nor->addr_width;
+	cmd.enable.bits.mode = (num_mode_cycles > 0);
+	cmd.enable.bits.dummy = (num_dummy_cycles > 0);
+	cmd.enable.bits.data = 1;
+	cmd.instruction = nor->read_opcode;
+	cmd.address = (u32)from;
+	cmd.mode = 0xff; /* This value prevents from entering the 0-4-4 mode */
+	cmd.num_mode_cycles = num_mode_cycles;
+	cmd.num_dummy_cycles = num_dummy_cycles;
+	cmd.rx_buf = read_buf;
+	cmd.buf_len = len;
+	ret = atmel_qspi_run_command(aq, &cmd, QSPI_IFR_TFRTYP_TRSFR_READ_MEM,
+				     ifr_width);
+	return (ret < 0) ? ret : len;
+}
+
+static int atmel_qspi_init(struct atmel_qspi *aq)
+{
+	unsigned long src_rate;
+	u32 mr, scr, scbr;
+
+	/* Reset the QSPI controller */
+	qspi_writel(aq, QSPI_CR, QSPI_CR_SWRST);
+
+	/* Set the QSPI controller in Serial Memory Mode */
+	mr = QSPI_MR_NBBITS(8) | QSPI_MR_SSM;
+	qspi_writel(aq, QSPI_MR, mr);
+
+	src_rate = clk_get_rate(aq->clk);
+	if (!src_rate)
+		return -EINVAL;
+
+	/* Compute the QSPI baudrate */
+	scbr = DIV_ROUND_UP(src_rate, aq->clk_rate);
+	if (scbr > 0)
+		scbr--;
+	scr = QSPI_SCR_SCBR(scbr);
+	qspi_writel(aq, QSPI_SCR, scr);
+
+	/* Enable the QSPI controller */
+	qspi_writel(aq, QSPI_CR, QSPI_CR_QSPIEN);
+
+	return 0;
+}
+
+static irqreturn_t atmel_qspi_interrupt(int irq, void *dev_id)
+{
+	struct atmel_qspi *aq = (struct atmel_qspi *)dev_id;
+	u32 status, mask, pending;
+
+	status = qspi_readl(aq, QSPI_SR);
+	mask = qspi_readl(aq, QSPI_IMR);
+	pending = status & mask;
+
+	if (!pending)
+		return IRQ_NONE;
+
+	aq->pending |= pending;
+	if ((aq->pending & QSPI_SR_CMD_COMPLETED) == QSPI_SR_CMD_COMPLETED)
+		complete(&aq->cmd_completion);
+
+	return IRQ_HANDLED;
+}
+
+static int atmel_qspi_probe(struct platform_device *pdev)
+{
+	struct device_node *child, *np = pdev->dev.of_node;
+	char modalias[SPI_NAME_SIZE];
+	const char *name = NULL;
+	struct atmel_qspi *aq;
+	struct resource *res;
+	struct spi_nor *nor;
+	struct mtd_info *mtd;
+	int irq, err = 0;
+
+	if (of_get_child_count(np) != 1)
+		return -ENODEV;
+	child = of_get_next_child(np, NULL);
+
+	aq = devm_kzalloc(&pdev->dev, sizeof(*aq), GFP_KERNEL);
+	if (!aq) {
+		err = -ENOMEM;
+		goto exit;
+	}
+
+	platform_set_drvdata(pdev, aq);
+	init_completion(&aq->cmd_completion);
+	aq->pdev = pdev;
+
+	/* Map the registers */
+	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "qspi_base");
+	aq->regs = devm_ioremap_resource(&pdev->dev, res);
+	if (IS_ERR(aq->regs)) {
+		dev_err(&pdev->dev, "missing registers\n");
+		err = PTR_ERR(aq->regs);
+		goto exit;
+	}
+
+	/* Map the AHB memory */
+	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "qspi_mmap");
+	aq->mem = devm_ioremap_resource(&pdev->dev, res);
+	if (IS_ERR(aq->mem)) {
+		dev_err(&pdev->dev, "missing AHB memory\n");
+		err = PTR_ERR(aq->mem);
+		goto exit;
+	}
+
+	/* Get the peripheral clock */
+	aq->clk = devm_clk_get(&pdev->dev, NULL);
+	if (IS_ERR(aq->clk)) {
+		dev_err(&pdev->dev, "missing peripheral clock\n");
+		err = PTR_ERR(aq->clk);
+		goto exit;
+	}
+
+	/* Enable the peripheral clock */
+	err = clk_prepare_enable(aq->clk);
+	if (err) {
+		dev_err(&pdev->dev, "failed to enable the peripheral clock\n");
+		goto exit;
+	}
+
+	/* Request the IRQ */
+	irq = platform_get_irq(pdev, 0);
+	if (irq < 0) {
+		dev_err(&pdev->dev, "missing IRQ\n");
+		err = irq;
+		goto disable_clk;
+	}
+	err = devm_request_irq(&pdev->dev, irq, atmel_qspi_interrupt,
+			       0, dev_name(&pdev->dev), aq);
+	if (err)
+		goto disable_clk;
+
+	/* Setup the spi-nor */
+	nor = &aq->nor;
+	mtd = &nor->mtd;
+
+	nor->dev = &pdev->dev;
+	spi_nor_set_flash_node(nor, child);
+	nor->priv = aq;
+	mtd->priv = nor;
+
+	nor->read_reg = atmel_qspi_read_reg;
+	nor->write_reg = atmel_qspi_write_reg;
+	nor->read = atmel_qspi_read;
+	nor->write = atmel_qspi_write;
+	nor->erase = atmel_qspi_erase;
+
+	err = of_modalias_node(child, modalias, sizeof(modalias));
+	if (err < 0)
+		goto disable_clk;
+
+	if (strcmp(modalias, "spi-nor"))
+		name = modalias;
+
+	err = of_property_read_u32(child, "spi-max-frequency", &aq->clk_rate);
+	if (err < 0)
+		goto disable_clk;
+
+	err = atmel_qspi_init(aq);
+	if (err)
+		goto disable_clk;
+
+	err = spi_nor_scan(nor, name, SPI_NOR_QUAD);
+	if (err)
+		goto disable_clk;
+
+	err = mtd_device_register(mtd, NULL, 0);
+	if (err)
+		goto disable_clk;
+
+	of_node_put(child);
+
+	return 0;
+
+disable_clk:
+	clk_disable_unprepare(aq->clk);
+exit:
+	of_node_put(child);
+
+	return err;
+}
+
+static int atmel_qspi_remove(struct platform_device *pdev)
+{
+	struct atmel_qspi *aq = platform_get_drvdata(pdev);
+
+	mtd_device_unregister(&aq->nor.mtd);
+	qspi_writel(aq, QSPI_CR, QSPI_CR_QSPIDIS);
+	clk_disable_unprepare(aq->clk);
+	return 0;
+}
+
+
+static const struct of_device_id atmel_qspi_dt_ids[] = {
+	{ .compatible = "atmel,sama5d2-qspi" },
+	{ /* sentinel */ }
+};
+
+MODULE_DEVICE_TABLE(of, atmel_qspi_dt_ids);
+
+static struct platform_driver atmel_qspi_driver = {
+	.driver = {
+		.name	= "atmel_qspi",
+		.of_match_table	= atmel_qspi_dt_ids,
+	},
+	.probe		= atmel_qspi_probe,
+	.remove		= atmel_qspi_remove,
+};
+module_platform_driver(atmel_qspi_driver);
+
+MODULE_AUTHOR("Cyrille Pitchen <cyrille.pitchen@atmel.com>");
+MODULE_DESCRIPTION("Atmel QSPI Controller driver");
+MODULE_LICENSE("GPL v2");
-- 
1.8.2.2

Re: [PATCH 2/2] mtd: atmel-quadspi: add driver for Atmel QSPI controller

[Brian Norris]

Hi,

On Mon, Jun 13, 2016 at 05:10:26PM +0200, Cyrille Pitchen wrote:
> This driver add support to the new Atmel QSPI controller embedded into
> sama5d2x SoCs. It expects a NOR memory to be connected to the QSPI
> controller.
> 
> Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com>
> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
> ---
>  drivers/mtd/spi-nor/Kconfig         |   9 +
>  drivers/mtd/spi-nor/Makefile        |   1 +
>  drivers/mtd/spi-nor/atmel-quadspi.c | 741 ++++++++++++++++++++++++++++++++++++
>  3 files changed, 751 insertions(+)
>  create mode 100644 drivers/mtd/spi-nor/atmel-quadspi.c
> 
> diff --git a/drivers/mtd/spi-nor/Kconfig b/drivers/mtd/spi-nor/Kconfig
> index d42c98e1f581..c546efd1357b 100644
> --- a/drivers/mtd/spi-nor/Kconfig
> +++ b/drivers/mtd/spi-nor/Kconfig
> @@ -38,6 +38,15 @@ config SPI_FSL_QUADSPI
>  	  This controller does not support generic SPI. It only supports
>  	  SPI NOR.
>  
> +config SPI_ATMEL_QUADSPI
> +	tristate "Atmel Quad SPI Controller"
> +	depends on ARCH_AT91 || (ARM && COMPILE_TEST)
> +	depends on OF && HAS_IOMEM
> +	help
> +	  This enables support for the Quad SPI controller in master mode.
> +	  This driver does not support generic SPI. The implementation only
> +	  supports SPI NOR.
> +
>  config SPI_NXP_SPIFI
>  	tristate "NXP SPI Flash Interface (SPIFI)"
>  	depends on OF && (ARCH_LPC18XX || COMPILE_TEST)
> diff --git a/drivers/mtd/spi-nor/Makefile b/drivers/mtd/spi-nor/Makefile
> index 0bf3a7f81675..1525913698ad 100644
> --- a/drivers/mtd/spi-nor/Makefile
> +++ b/drivers/mtd/spi-nor/Makefile
> @@ -2,3 +2,4 @@ obj-$(CONFIG_MTD_SPI_NOR)	+= spi-nor.o
>  obj-$(CONFIG_SPI_FSL_QUADSPI)	+= fsl-quadspi.o
>  obj-$(CONFIG_MTD_MT81xx_NOR)    += mtk-quadspi.o
>  obj-$(CONFIG_SPI_NXP_SPIFI)	+= nxp-spifi.o
> +obj-$(CONFIG_SPI_ATMEL_QUADSPI)	+= atmel-quadspi.o
> diff --git a/drivers/mtd/spi-nor/atmel-quadspi.c b/drivers/mtd/spi-nor/atmel-quadspi.c
> new file mode 100644
> index 000000000000..06d1bf276dd0
> --- /dev/null
> +++ b/drivers/mtd/spi-nor/atmel-quadspi.c
> @@ -0,0 +1,741 @@

[...]

> +struct atmel_qspi_command {
> +	union {
> +		struct {
> +			u32	instruction:1;
> +			u32	address:3;
> +			u32	mode:1;
> +			u32	dummy:1;
> +			u32	data:1;
> +			u32	reserved:25;
> +		}		bits;

Are you sure this bitfield is going to do what you want, all the time?
What about on big-endian architectures? And are you guaranteed it'll
pack properly, with no padding? IIRC, the answer to the first 2 is "no",
and I have no idea about the 3rd. Honestly, I've been scared away from
using bitfields on anything where the ordering mattered, and I thought
that's because I was hurt by the lack of guarantee once. But I easily
could be misguided.

> +		u32	word;
> +	}	enable;
> +	u8	instruction;
> +	u8	mode;
> +	u8	num_mode_cycles;
> +	u8	num_dummy_cycles;
> +	u32	address;
> +
> +	size_t		buf_len;
> +	const void	*tx_buf;
> +	void		*rx_buf;
> +};
> +

[...]

> +static int atmel_qspi_probe(struct platform_device *pdev)
> +{
> +	struct device_node *child, *np = pdev->dev.of_node;
> +	char modalias[SPI_NAME_SIZE];
> +	const char *name = NULL;
> +	struct atmel_qspi *aq;
> +	struct resource *res;
> +	struct spi_nor *nor;
> +	struct mtd_info *mtd;
> +	int irq, err = 0;
> +
> +	if (of_get_child_count(np) != 1)
> +		return -ENODEV;
> +	child = of_get_next_child(np, NULL);
> +
> +	aq = devm_kzalloc(&pdev->dev, sizeof(*aq), GFP_KERNEL);
> +	if (!aq) {
> +		err = -ENOMEM;
> +		goto exit;
> +	}
> +
> +	platform_set_drvdata(pdev, aq);
> +	init_completion(&aq->cmd_completion);
> +	aq->pdev = pdev;
> +
> +	/* Map the registers */
> +	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "qspi_base");
> +	aq->regs = devm_ioremap_resource(&pdev->dev, res);
> +	if (IS_ERR(aq->regs)) {
> +		dev_err(&pdev->dev, "missing registers\n");
> +		err = PTR_ERR(aq->regs);
> +		goto exit;
> +	}
> +
> +	/* Map the AHB memory */
> +	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "qspi_mmap");
> +	aq->mem = devm_ioremap_resource(&pdev->dev, res);
> +	if (IS_ERR(aq->mem)) {
> +		dev_err(&pdev->dev, "missing AHB memory\n");
> +		err = PTR_ERR(aq->mem);
> +		goto exit;
> +	}
> +
> +	/* Get the peripheral clock */
> +	aq->clk = devm_clk_get(&pdev->dev, NULL);
> +	if (IS_ERR(aq->clk)) {
> +		dev_err(&pdev->dev, "missing peripheral clock\n");
> +		err = PTR_ERR(aq->clk);
> +		goto exit;
> +	}
> +
> +	/* Enable the peripheral clock */
> +	err = clk_prepare_enable(aq->clk);
> +	if (err) {
> +		dev_err(&pdev->dev, "failed to enable the peripheral clock\n");
> +		goto exit;
> +	}
> +
> +	/* Request the IRQ */
> +	irq = platform_get_irq(pdev, 0);
> +	if (irq < 0) {
> +		dev_err(&pdev->dev, "missing IRQ\n");
> +		err = irq;
> +		goto disable_clk;
> +	}
> +	err = devm_request_irq(&pdev->dev, irq, atmel_qspi_interrupt,
> +			       0, dev_name(&pdev->dev), aq);
> +	if (err)
> +		goto disable_clk;
> +
> +	/* Setup the spi-nor */
> +	nor = &aq->nor;
> +	mtd = &nor->mtd;
> +
> +	nor->dev = &pdev->dev;
> +	spi_nor_set_flash_node(nor, child);
> +	nor->priv = aq;
> +	mtd->priv = nor;
> +
> +	nor->read_reg = atmel_qspi_read_reg;
> +	nor->write_reg = atmel_qspi_write_reg;
> +	nor->read = atmel_qspi_read;
> +	nor->write = atmel_qspi_write;
> +	nor->erase = atmel_qspi_erase;
> +
> +	err = of_modalias_node(child, modalias, sizeof(modalias));
> +	if (err < 0)
> +		goto disable_clk;
> +
> +	if (strcmp(modalias, "spi-nor"))
> +		name = modalias;

What is this modalias handling for? Are you trying to support passing in
a flash-specific string, like m25p80 does? And you're enforcing that to
be only the first entry in the 'compatible' list? Seems fragile; m25p80
is a special case (and it's already fragile) that I'd rather not
imitate.

I understand that we discussed extending the use of compatible for
spi-nor drivers, but I still don't think that issue is resolved. And if
we are going to extend the use of compatible, then I think we should
fixup all the spi-nor drivers in the same way.

> +
> +	err = of_property_read_u32(child, "spi-max-frequency", &aq->clk_rate);
> +	if (err < 0)
> +		goto disable_clk;
> +
> +	err = atmel_qspi_init(aq);
> +	if (err)
> +		goto disable_clk;
> +
> +	err = spi_nor_scan(nor, name, SPI_NOR_QUAD);
> +	if (err)
> +		goto disable_clk;
> +
> +	err = mtd_device_register(mtd, NULL, 0);
> +	if (err)
> +		goto disable_clk;
> +
> +	of_node_put(child);
> +
> +	return 0;
> +
> +disable_clk:
> +	clk_disable_unprepare(aq->clk);
> +exit:
> +	of_node_put(child);
> +
> +	return err;
> +}
> +
> +static int atmel_qspi_remove(struct platform_device *pdev)
> +{
> +	struct atmel_qspi *aq = platform_get_drvdata(pdev);
> +
> +	mtd_device_unregister(&aq->nor.mtd);
> +	qspi_writel(aq, QSPI_CR, QSPI_CR_QSPIDIS);
> +	clk_disable_unprepare(aq->clk);
> +	return 0;
> +}
> +
> +
> +static const struct of_device_id atmel_qspi_dt_ids[] = {
> +	{ .compatible = "atmel,sama5d2-qspi" },
> +	{ /* sentinel */ }
> +};
> +
> +MODULE_DEVICE_TABLE(of, atmel_qspi_dt_ids);
> +
> +static struct platform_driver atmel_qspi_driver = {
> +	.driver = {
> +		.name	= "atmel_qspi",
> +		.of_match_table	= atmel_qspi_dt_ids,
> +	},
> +	.probe		= atmel_qspi_probe,
> +	.remove		= atmel_qspi_remove,
> +};
> +module_platform_driver(atmel_qspi_driver);
> +
> +MODULE_AUTHOR("Cyrille Pitchen <cyrille.pitchen@atmel.com>");
> +MODULE_DESCRIPTION("Atmel QSPI Controller driver");
> +MODULE_LICENSE("GPL v2");

I'm likely to apply this driver, with the following diff (the bitfield
issue (if there is one) is less of a maintenance issue), barring any
major objections.

diff --git a/drivers/mtd/spi-nor/atmel-quadspi.c b/drivers/mtd/spi-nor/atmel-quadspi.c
index 06d1bf276dd0..47937d9beec6 100644
--- a/drivers/mtd/spi-nor/atmel-quadspi.c
+++ b/drivers/mtd/spi-nor/atmel-quadspi.c
@@ -591,8 +591,6 @@ static irqreturn_t atmel_qspi_interrupt(int irq, void *dev_id)
 static int atmel_qspi_probe(struct platform_device *pdev)
 {
 	struct device_node *child, *np = pdev->dev.of_node;
-	char modalias[SPI_NAME_SIZE];
-	const char *name = NULL;
 	struct atmel_qspi *aq;
 	struct resource *res;
 	struct spi_nor *nor;
@@ -673,13 +671,6 @@ static int atmel_qspi_probe(struct platform_device *pdev)
 	nor->write = atmel_qspi_write;
 	nor->erase = atmel_qspi_erase;
 
-	err = of_modalias_node(child, modalias, sizeof(modalias));
-	if (err < 0)
-		goto disable_clk;
-
-	if (strcmp(modalias, "spi-nor"))
-		name = modalias;
-
 	err = of_property_read_u32(child, "spi-max-frequency", &aq->clk_rate);
 	if (err < 0)
 		goto disable_clk;
@@ -688,7 +679,7 @@ static int atmel_qspi_probe(struct platform_device *pdev)
 	if (err)
 		goto disable_clk;
 
-	err = spi_nor_scan(nor, name, SPI_NOR_QUAD);
+	err = spi_nor_scan(nor, NULL, SPI_NOR_QUAD);
 	if (err)
 		goto disable_clk;
 

Re: [PATCH 2/2] mtd: atmel-quadspi: add driver for Atmel QSPI controller

[Brian Norris]

Hi Cyrille,

On Wed, Jul 13, 2016 at 06:32:42PM -0700, Brian Norris wrote:
> On Mon, Jun 13, 2016 at 05:10:26PM +0200, Cyrille Pitchen wrote:
> > This driver add support to the new Atmel QSPI controller embedded into
> > sama5d2x SoCs. It expects a NOR memory to be connected to the QSPI
> > controller.
> > 
> > Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com>
> > Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
> > ---
> >  drivers/mtd/spi-nor/Kconfig         |   9 +
> >  drivers/mtd/spi-nor/Makefile        |   1 +
> >  drivers/mtd/spi-nor/atmel-quadspi.c | 741 ++++++++++++++++++++++++++++++++++++
> >  3 files changed, 751 insertions(+)
> >  create mode 100644 drivers/mtd/spi-nor/atmel-quadspi.c
> > 
> > diff --git a/drivers/mtd/spi-nor/Kconfig b/drivers/mtd/spi-nor/Kconfig
> > index d42c98e1f581..c546efd1357b 100644
> > --- a/drivers/mtd/spi-nor/Kconfig
> > +++ b/drivers/mtd/spi-nor/Kconfig
> > @@ -38,6 +38,15 @@ config SPI_FSL_QUADSPI
> >  	  This controller does not support generic SPI. It only supports
> >  	  SPI NOR.
> >  
> > +config SPI_ATMEL_QUADSPI
> > +	tristate "Atmel Quad SPI Controller"
> > +	depends on ARCH_AT91 || (ARM && COMPILE_TEST)
> > +	depends on OF && HAS_IOMEM
> > +	help
> > +	  This enables support for the Quad SPI controller in master mode.
> > +	  This driver does not support generic SPI. The implementation only
> > +	  supports SPI NOR.
> > +
> >  config SPI_NXP_SPIFI
> >  	tristate "NXP SPI Flash Interface (SPIFI)"
> >  	depends on OF && (ARCH_LPC18XX || COMPILE_TEST)
> > diff --git a/drivers/mtd/spi-nor/Makefile b/drivers/mtd/spi-nor/Makefile
> > index 0bf3a7f81675..1525913698ad 100644
> > --- a/drivers/mtd/spi-nor/Makefile
> > +++ b/drivers/mtd/spi-nor/Makefile
> > @@ -2,3 +2,4 @@ obj-$(CONFIG_MTD_SPI_NOR)	+= spi-nor.o
> >  obj-$(CONFIG_SPI_FSL_QUADSPI)	+= fsl-quadspi.o
> >  obj-$(CONFIG_MTD_MT81xx_NOR)    += mtk-quadspi.o
> >  obj-$(CONFIG_SPI_NXP_SPIFI)	+= nxp-spifi.o
> > +obj-$(CONFIG_SPI_ATMEL_QUADSPI)	+= atmel-quadspi.o
> > diff --git a/drivers/mtd/spi-nor/atmel-quadspi.c b/drivers/mtd/spi-nor/atmel-quadspi.c
> > new file mode 100644
> > index 000000000000..06d1bf276dd0
> > --- /dev/null
> > +++ b/drivers/mtd/spi-nor/atmel-quadspi.c
> > @@ -0,0 +1,741 @@
> 
> [...]
> 
> > +struct atmel_qspi_command {
> > +	union {
> > +		struct {
> > +			u32	instruction:1;
> > +			u32	address:3;
> > +			u32	mode:1;
> > +			u32	dummy:1;
> > +			u32	data:1;
> > +			u32	reserved:25;
> > +		}		bits;
> 
> Are you sure this bitfield is going to do what you want, all the time?
> What about on big-endian architectures? And are you guaranteed it'll
> pack properly, with no padding? IIRC, the answer to the first 2 is "no",
> and I have no idea about the 3rd. Honestly, I've been scared away from
> using bitfields on anything where the ordering mattered, and I thought
> that's because I was hurt by the lack of guarantee once. But I easily
> could be misguided.
> 
> > +		u32	word;
> > +	}	enable;
> > +	u8	instruction;
> > +	u8	mode;
> > +	u8	num_mode_cycles;
> > +	u8	num_dummy_cycles;
> > +	u32	address;
> > +
> > +	size_t		buf_len;
> > +	const void	*tx_buf;
> > +	void		*rx_buf;
> > +};
> > +
> 
> [...]
> 
> > +static int atmel_qspi_probe(struct platform_device *pdev)
> > +{
> > +	struct device_node *child, *np = pdev->dev.of_node;
> > +	char modalias[SPI_NAME_SIZE];
> > +	const char *name = NULL;
> > +	struct atmel_qspi *aq;
> > +	struct resource *res;
> > +	struct spi_nor *nor;
> > +	struct mtd_info *mtd;
> > +	int irq, err = 0;
> > +
> > +	if (of_get_child_count(np) != 1)
> > +		return -ENODEV;
> > +	child = of_get_next_child(np, NULL);
> > +
> > +	aq = devm_kzalloc(&pdev->dev, sizeof(*aq), GFP_KERNEL);
> > +	if (!aq) {
> > +		err = -ENOMEM;
> > +		goto exit;
> > +	}
> > +
> > +	platform_set_drvdata(pdev, aq);
> > +	init_completion(&aq->cmd_completion);
> > +	aq->pdev = pdev;
> > +
> > +	/* Map the registers */
> > +	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "qspi_base");
> > +	aq->regs = devm_ioremap_resource(&pdev->dev, res);
> > +	if (IS_ERR(aq->regs)) {
> > +		dev_err(&pdev->dev, "missing registers\n");
> > +		err = PTR_ERR(aq->regs);
> > +		goto exit;
> > +	}
> > +
> > +	/* Map the AHB memory */
> > +	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "qspi_mmap");
> > +	aq->mem = devm_ioremap_resource(&pdev->dev, res);
> > +	if (IS_ERR(aq->mem)) {
> > +		dev_err(&pdev->dev, "missing AHB memory\n");
> > +		err = PTR_ERR(aq->mem);
> > +		goto exit;
> > +	}
> > +
> > +	/* Get the peripheral clock */
> > +	aq->clk = devm_clk_get(&pdev->dev, NULL);
> > +	if (IS_ERR(aq->clk)) {
> > +		dev_err(&pdev->dev, "missing peripheral clock\n");
> > +		err = PTR_ERR(aq->clk);
> > +		goto exit;
> > +	}
> > +
> > +	/* Enable the peripheral clock */
> > +	err = clk_prepare_enable(aq->clk);
> > +	if (err) {
> > +		dev_err(&pdev->dev, "failed to enable the peripheral clock\n");
> > +		goto exit;
> > +	}
> > +
> > +	/* Request the IRQ */
> > +	irq = platform_get_irq(pdev, 0);
> > +	if (irq < 0) {
> > +		dev_err(&pdev->dev, "missing IRQ\n");
> > +		err = irq;
> > +		goto disable_clk;
> > +	}
> > +	err = devm_request_irq(&pdev->dev, irq, atmel_qspi_interrupt,
> > +			       0, dev_name(&pdev->dev), aq);
> > +	if (err)
> > +		goto disable_clk;
> > +
> > +	/* Setup the spi-nor */
> > +	nor = &aq->nor;
> > +	mtd = &nor->mtd;
> > +
> > +	nor->dev = &pdev->dev;
> > +	spi_nor_set_flash_node(nor, child);
> > +	nor->priv = aq;
> > +	mtd->priv = nor;
> > +
> > +	nor->read_reg = atmel_qspi_read_reg;
> > +	nor->write_reg = atmel_qspi_write_reg;
> > +	nor->read = atmel_qspi_read;
> > +	nor->write = atmel_qspi_write;
> > +	nor->erase = atmel_qspi_erase;
> > +
> > +	err = of_modalias_node(child, modalias, sizeof(modalias));
> > +	if (err < 0)
> > +		goto disable_clk;
> > +
> > +	if (strcmp(modalias, "spi-nor"))
> > +		name = modalias;
> 
> What is this modalias handling for? Are you trying to support passing in
> a flash-specific string, like m25p80 does? And you're enforcing that to
> be only the first entry in the 'compatible' list? Seems fragile; m25p80
> is a special case (and it's already fragile) that I'd rather not
> imitate.
> 
> I understand that we discussed extending the use of compatible for
> spi-nor drivers, but I still don't think that issue is resolved. And if
> we are going to extend the use of compatible, then I think we should
> fixup all the spi-nor drivers in the same way.
> 
> > +
> > +	err = of_property_read_u32(child, "spi-max-frequency", &aq->clk_rate);
> > +	if (err < 0)
> > +		goto disable_clk;
> > +
> > +	err = atmel_qspi_init(aq);
> > +	if (err)
> > +		goto disable_clk;
> > +
> > +	err = spi_nor_scan(nor, name, SPI_NOR_QUAD);
> > +	if (err)
> > +		goto disable_clk;
> > +
> > +	err = mtd_device_register(mtd, NULL, 0);
> > +	if (err)
> > +		goto disable_clk;
> > +
> > +	of_node_put(child);
> > +
> > +	return 0;
> > +
> > +disable_clk:
> > +	clk_disable_unprepare(aq->clk);
> > +exit:
> > +	of_node_put(child);
> > +
> > +	return err;
> > +}
> > +
> > +static int atmel_qspi_remove(struct platform_device *pdev)
> > +{
> > +	struct atmel_qspi *aq = platform_get_drvdata(pdev);
> > +
> > +	mtd_device_unregister(&aq->nor.mtd);
> > +	qspi_writel(aq, QSPI_CR, QSPI_CR_QSPIDIS);
> > +	clk_disable_unprepare(aq->clk);
> > +	return 0;
> > +}
> > +
> > +
> > +static const struct of_device_id atmel_qspi_dt_ids[] = {
> > +	{ .compatible = "atmel,sama5d2-qspi" },
> > +	{ /* sentinel */ }
> > +};
> > +
> > +MODULE_DEVICE_TABLE(of, atmel_qspi_dt_ids);
> > +
> > +static struct platform_driver atmel_qspi_driver = {
> > +	.driver = {
> > +		.name	= "atmel_qspi",
> > +		.of_match_table	= atmel_qspi_dt_ids,
> > +	},
> > +	.probe		= atmel_qspi_probe,
> > +	.remove		= atmel_qspi_remove,
> > +};
> > +module_platform_driver(atmel_qspi_driver);
> > +
> > +MODULE_AUTHOR("Cyrille Pitchen <cyrille.pitchen@atmel.com>");
> > +MODULE_DESCRIPTION("Atmel QSPI Controller driver");
> > +MODULE_LICENSE("GPL v2");
> 
> I'm likely to apply this driver, with the following diff (the bitfield
> issue (if there is one) is less of a maintenance issue), barring any
> major objections.
> 
> diff --git a/drivers/mtd/spi-nor/atmel-quadspi.c b/drivers/mtd/spi-nor/atmel-quadspi.c
> index 06d1bf276dd0..47937d9beec6 100644
> --- a/drivers/mtd/spi-nor/atmel-quadspi.c
> +++ b/drivers/mtd/spi-nor/atmel-quadspi.c
> @@ -591,8 +591,6 @@ static irqreturn_t atmel_qspi_interrupt(int irq, void *dev_id)
>  static int atmel_qspi_probe(struct platform_device *pdev)
>  {
>  	struct device_node *child, *np = pdev->dev.of_node;
> -	char modalias[SPI_NAME_SIZE];
> -	const char *name = NULL;
>  	struct atmel_qspi *aq;
>  	struct resource *res;
>  	struct spi_nor *nor;
> @@ -673,13 +671,6 @@ static int atmel_qspi_probe(struct platform_device *pdev)
>  	nor->write = atmel_qspi_write;
>  	nor->erase = atmel_qspi_erase;
>  
> -	err = of_modalias_node(child, modalias, sizeof(modalias));
> -	if (err < 0)
> -		goto disable_clk;
> -
> -	if (strcmp(modalias, "spi-nor"))
> -		name = modalias;
> -
>  	err = of_property_read_u32(child, "spi-max-frequency", &aq->clk_rate);
>  	if (err < 0)
>  		goto disable_clk;
> @@ -688,7 +679,7 @@ static int atmel_qspi_probe(struct platform_device *pdev)
>  	if (err)
>  		goto disable_clk;
>  
> -	err = spi_nor_scan(nor, name, SPI_NOR_QUAD);
> +	err = spi_nor_scan(nor, NULL, SPI_NOR_QUAD);
>  	if (err)
>  		goto disable_clk;
>  

Applied to l2-mtd.git with that fixup.

Re: [PATCH 2/2] mtd: atmel-quadspi: add driver for Atmel QSPI controller

[Arnd Bergmann]

On Friday, July 15, 2016 5:45:07 PM CEST Brian Norris wrote:
> Applied to l2-mtd.git with that fixup.

I'm getting this build error now on a randconfig build:

drivers/mtd/built-in.o: In function `atmel_qspi_run_command':
:(.text+0x1ee3c): undefined reference to `_memcpy_toio'
:(.text+0x1ee48): undefined reference to `_memcpy_fromio'

On ARCH_EBSA, which doesn't build the file that contains the two
functions. I don't see any other driver on ARM using those two
functions directly. What is the specific reason for using them
here? Do you require byte-wise accesses, or could you use
the normal memcpy_toio/memcpy_fromio that turn into aligned
32-bit word accesses instead? If you have to use the non-portable
functions, maybe we can just make the driver depend on !ARCH_EBSA?

	Arnd

Re: [PATCH 2/2] mtd: atmel-quadspi: add driver for Atmel QSPI controller

[Brian Norris]

On Mon, Jul 18, 2016 at 09:35:39PM +0200, Arnd Bergmann wrote:
> On Friday, July 15, 2016 5:45:07 PM CEST Brian Norris wrote:
> > Applied to l2-mtd.git with that fixup.
> 
> I'm getting this build error now on a randconfig build:
> 
> drivers/mtd/built-in.o: In function `atmel_qspi_run_command':
> :(.text+0x1ee3c): undefined reference to `_memcpy_toio'
> :(.text+0x1ee48): undefined reference to `_memcpy_fromio'

Whoops, I noticed those during review, but I don't know why I forgot to
mention them nor fix them up before applying.

> On ARCH_EBSA, which doesn't build the file that contains the two
> functions. I don't see any other driver on ARM using those two
> functions directly. What is the specific reason for using them
> here? Do you require byte-wise accesses, or could you use
> the normal memcpy_toio/memcpy_fromio that turn into aligned
> 32-bit word accesses instead?

Good questions. I would suspect that aligned 32-bit accesses are what
they're looking for, but I'm not absolutely sure.

> If you have to use the non-portable
> functions, maybe we can just make the driver depend on !ARCH_EBSA?

I don't see an ARCH_EBSA. Did you mean ARCH_EBSA110?

Or we could just drop the '|| (ARM && COMPILE_TEST)' clause for now:

	depends on ARCH_AT91 || (ARM && COMPILE_TEST)

Brian

Re: [PATCH 2/2] mtd: atmel-quadspi: add driver for Atmel QSPI controller

[Arnd Bergmann]

On Monday, July 18, 2016 12:55:11 PM CEST Brian Norris wrote:
> On Mon, Jul 18, 2016 at 09:35:39PM +0200, Arnd Bergmann wrote:
> > On Friday, July 15, 2016 5:45:07 PM CEST Brian Norris wrote:
> > > Applied to l2-mtd.git with that fixup.
> > 
> > I'm getting this build error now on a randconfig build:
> > 
> > drivers/mtd/built-in.o: In function `atmel_qspi_run_command':
> > :(.text+0x1ee3c): undefined reference to `_memcpy_toio'
> > :(.text+0x1ee48): undefined reference to `_memcpy_fromio'
> 
> Whoops, I noticed those during review, but I don't know why I forgot to
> mention them nor fix them up before applying.
> 
> > On ARCH_EBSA, which doesn't build the file that contains the two
> > functions. I don't see any other driver on ARM using those two
> > functions directly. What is the specific reason for using them
> > here? Do you require byte-wise accesses, or could you use
> > the normal memcpy_toio/memcpy_fromio that turn into aligned
> > 32-bit word accesses instead?
> 
> Good questions. I would suspect that aligned 32-bit accesses are what
> they're looking for, but I'm not absolutely sure.

Ok, so we should look at that first. If the driver supports 32-bit
access, using the regular accessors will also make the transfers
much faster.

> > If you have to use the non-portable
> > functions, maybe we can just make the driver depend on !ARCH_EBSA?
> 
> I don't see an ARCH_EBSA. Did you mean ARCH_EBSA110?

Yes, sorry for the typo.

> Or we could just drop the '|| (ARM && COMPILE_TEST)' clause for now:
> 
>         depends on ARCH_AT91 || (ARM && COMPILE_TEST)

I'd prefer to keep the COMPILE_TEST option, after all it's how
I found the problem. On a related note, what is the ARM dependency
for? Is that just for the _memcpy_toio/_memcpy_fromio? Maybe we
can drop too if we find the right architecture-independent
replacement for those two calls.

	Arnd

Re: [PATCH 2/2] mtd: atmel-quadspi: add driver for Atmel QSPI controller

[Cyrille Pitchen]

Hi all,

Le 18/07/2016 à 21:59, Arnd Bergmann a écrit :
> On Monday, July 18, 2016 12:55:11 PM CEST Brian Norris wrote:
>> On Mon, Jul 18, 2016 at 09:35:39PM +0200, Arnd Bergmann wrote:
>>> On Friday, July 15, 2016 5:45:07 PM CEST Brian Norris wrote:
>>>> Applied to l2-mtd.git with that fixup.
>>>
>>> I'm getting this build error now on a randconfig build:
>>>
>>> drivers/mtd/built-in.o: In function `atmel_qspi_run_command':
>>> :(.text+0x1ee3c): undefined reference to `_memcpy_toio'
>>> :(.text+0x1ee48): undefined reference to `_memcpy_fromio'
>>
>> Whoops, I noticed those during review, but I don't know why I forgot to
>> mention them nor fix them up before applying.
>>
>>> On ARCH_EBSA, which doesn't build the file that contains the two
>>> functions. I don't see any other driver on ARM using those two
>>> functions directly. What is the specific reason for using them
>>> here? Do you require byte-wise accesses, or could you use
>>> the normal memcpy_toio/memcpy_fromio that turn into aligned
>>> 32-bit word accesses instead?
>>
>> Good questions. I would suspect that aligned 32-bit accesses are what
>> they're looking for, but I'm not absolutely sure.
> 
> Ok, so we should look at that first. If the driver supports 32-bit
> access, using the regular accessors will also make the transfers
> much faster.
> 
>>> If you have to use the non-portable
>>> functions, maybe we can just make the driver depend on !ARCH_EBSA?
>>
>> I don't see an ARCH_EBSA. Did you mean ARCH_EBSA110?
> 
> Yes, sorry for the typo.
> 
>> Or we could just drop the '|| (ARM && COMPILE_TEST)' clause for now:
>>
>>         depends on ARCH_AT91 || (ARM && COMPILE_TEST)
> 
> I'd prefer to keep the COMPILE_TEST option, after all it's how
> I found the problem. On a related note, what is the ARM dependency
> for? Is that just for the _memcpy_toio/_memcpy_fromio? Maybe we
> can drop too if we find the right architecture-independent
> replacement for those two calls.
> 
> 	Arnd
> 


Indeed I added the ARM dependency for the COMPILE_TEST case only for
_memcpy_toio() and _memcpy_fromio(). I thought it would be enough.

Also, I use _memcpy_toio() and _memcpy_fromio() on purpose as opposed to
memcpy_toio() and mempcy_fromio(). I've tested the two latest functions quite
a long time ago and it didn't work. If I remember, on our architecture,
memcpy_toio() and memcpy_fromio() are simply implemented with the regular
memcpy().

The additional memory barriers inserted by readb() and writeb() seems to be
needed to guarantee the IO accesses are not reordered on the system bus.

In the atmel-quadspi.c driver, there is a comment on the use of memcpy():
/* Then fallback to a PIO transfer (memcpy() DOES NOT work!) */


Best regards,

Cyrille

Re: [PATCH 2/2] mtd: atmel-quadspi: add driver for Atmel QSPI controller

[Arnd Bergmann]

On Tuesday, July 19, 2016 12:03:11 PM CEST Cyrille Pitchen wrote:
> 
> Indeed I added the ARM dependency for the COMPILE_TEST case only for
> _memcpy_toio() and _memcpy_fromio(). I thought it would be enough.
> 
> Also, I use _memcpy_toio() and _memcpy_fromio() on purpose as opposed to
> memcpy_toio() and mempcy_fromio(). I've tested the two latest functions quite
> a long time ago and it didn't work. If I remember, on our architecture,
> memcpy_toio() and memcpy_fromio() are simply implemented with the regular
> memcpy().

They are implemented using "mmiocpy" when CONFIG_BIG_ENDIAN is not set.
mmiocpy is an alias for the out-of-line memcpy() implementation, but it's
different from calling memcpy() as the compiler can decide to replace
small memcpy with a pointer dereference, and that can be unaligned on
ARMv7 for non-MMIO pointers.

> The additional memory barriers inserted by readb() and writeb() seems to be
> needed to guarantee the IO accesses are not reordered on the system bus.

Those barriers should not be needed at all, I've never seen an I/O bus
that reorders consecutive reads to the same address range.
 
> In the atmel-quadspi.c driver, there is a comment on the use of memcpy():
> /* Then fallback to a PIO transfer (memcpy() DOES NOT work!) */

Please try again, I think this was fixed with

commit 1bd46782d08b01b73df0085b51ea1021b19b44fd
Author: Russell King <rmk+kernel@arm.linux.org.uk>
Date:   Fri Jul 3 15:22:54 2015 +0100

    ARM: avoid unwanted GCC memset()/memcpy() optimisations for IO variants
    
    We don't want GCC optimising our memset_io(), memcpy_fromio() or
    memcpy_toio() variants, so we must not call one of the standard
    functions.  Provide a separate name for our assembly memcpy() and
    memset() functions, and use that instead, thereby bypassing GCC's
    ability to optimise these operations.
    
    GCCs optimisation may introduce unaligned accesses which are invalid
    for device mappings.
    
    Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

	Arnd

Re: [PATCH 2/2] mtd: atmel-quadspi: add driver for Atmel QSPI controller

[Cyrille Pitchen]

Hi Brian,


Le 16/07/2016 à 02:45, Brian Norris a écrit :
> Hi Cyrille,
> 
> On Wed, Jul 13, 2016 at 06:32:42PM -0700, Brian Norris wrote:
>> On Mon, Jun 13, 2016 at 05:10:26PM +0200, Cyrille Pitchen wrote:
>>> This driver add support to the new Atmel QSPI controller embedded into
>>> sama5d2x SoCs. It expects a NOR memory to be connected to the QSPI
>>> controller.
>>>
>>> Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com>
>>> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
>>> ---
>>>  drivers/mtd/spi-nor/Kconfig         |   9 +
>>>  drivers/mtd/spi-nor/Makefile        |   1 +
>>>  drivers/mtd/spi-nor/atmel-quadspi.c | 741 ++++++++++++++++++++++++++++++++++++
>>>  3 files changed, 751 insertions(+)
>>>  create mode 100644 drivers/mtd/spi-nor/atmel-quadspi.c
>>>
>>> diff --git a/drivers/mtd/spi-nor/Kconfig b/drivers/mtd/spi-nor/Kconfig
>>> index d42c98e1f581..c546efd1357b 100644
>>> --- a/drivers/mtd/spi-nor/Kconfig
>>> +++ b/drivers/mtd/spi-nor/Kconfig
>>> @@ -38,6 +38,15 @@ config SPI_FSL_QUADSPI
>>>  	  This controller does not support generic SPI. It only supports
>>>  	  SPI NOR.
>>>  
>>> +config SPI_ATMEL_QUADSPI
>>> +	tristate "Atmel Quad SPI Controller"
>>> +	depends on ARCH_AT91 || (ARM && COMPILE_TEST)
>>> +	depends on OF && HAS_IOMEM
>>> +	help
>>> +	  This enables support for the Quad SPI controller in master mode.
>>> +	  This driver does not support generic SPI. The implementation only
>>> +	  supports SPI NOR.
>>> +
>>>  config SPI_NXP_SPIFI
>>>  	tristate "NXP SPI Flash Interface (SPIFI)"
>>>  	depends on OF && (ARCH_LPC18XX || COMPILE_TEST)
>>> diff --git a/drivers/mtd/spi-nor/Makefile b/drivers/mtd/spi-nor/Makefile
>>> index 0bf3a7f81675..1525913698ad 100644
>>> --- a/drivers/mtd/spi-nor/Makefile
>>> +++ b/drivers/mtd/spi-nor/Makefile
>>> @@ -2,3 +2,4 @@ obj-$(CONFIG_MTD_SPI_NOR)	+= spi-nor.o
>>>  obj-$(CONFIG_SPI_FSL_QUADSPI)	+= fsl-quadspi.o
>>>  obj-$(CONFIG_MTD_MT81xx_NOR)    += mtk-quadspi.o
>>>  obj-$(CONFIG_SPI_NXP_SPIFI)	+= nxp-spifi.o
>>> +obj-$(CONFIG_SPI_ATMEL_QUADSPI)	+= atmel-quadspi.o
>>> diff --git a/drivers/mtd/spi-nor/atmel-quadspi.c b/drivers/mtd/spi-nor/atmel-quadspi.c
>>> new file mode 100644
>>> index 000000000000..06d1bf276dd0
>>> --- /dev/null
>>> +++ b/drivers/mtd/spi-nor/atmel-quadspi.c
>>> @@ -0,0 +1,741 @@
>>
>> [...]
>>
>>> +struct atmel_qspi_command {
>>> +	union {
>>> +		struct {
>>> +			u32	instruction:1;
>>> +			u32	address:3;
>>> +			u32	mode:1;
>>> +			u32	dummy:1;
>>> +			u32	data:1;
>>> +			u32	reserved:25;
>>> +		}		bits;
>>
>> Are you sure this bitfield is going to do what you want, all the time?
>> What about on big-endian architectures? And are you guaranteed it'll
>> pack properly, with no padding? IIRC, the answer to the first 2 is "no",
>> and I have no idea about the 3rd. Honestly, I've been scared away from
>> using bitfields on anything where the ordering mattered, and I thought
>> that's because I was hurt by the lack of guarantee once. But I easily
>> could be misguided.
>>

The struct atmel_qspi_command is used by the driver only for internal purpose,
especially the data are not sent to the hardware, they are only used by the
CPU. Hence the actual order of the fields does not matter.

On a little-endian architecture, instruction will be bit0, then address
bits[3:1] and so on up to 31 whereas on a big-endian architecture, instruction
would be bit31 then address bits[30:28] down to 0.
No field crosses the byte boundary so I guess there should be no issue even on
a big-endian architecture.

I could have also use u8 type like this:
u8	instruction
u8	address
u8	mode
u8	dummy
u8	data

However I've just wanted to reduce the size of the structure (not a big deal
here). Otherwise I could have use bitmask but I've just thought bitfields
would have been a little bit more easy to read.

For instance:

#define ATMEL_QSPI_ENABLE_CMD		BIT(0)
#define ATMEL_QSPI_ENABLE_ADDR_MSK	GENMASK(3, 1)
#define ATMEL_QSPI_ENABLE_ADDR(addr) \
	(((addr) << 1) & ATMEL_QSPI_ENABLE_ADDR_MSK)
#define ATMEL_QSPI_ENABLE_MODE		BIT(4)
#define ATMEL_QSPI_ENABLE_DUMMY		BIT(5)
#define ATMEL_QSPI_ENABLE_DATA		BIT(6)

u32	enable;

.enable = (ATMEL_QSPI_ENABLE_CMD |
	   ATMEL_QSPI_ENABLE_ADDR(3) |
	   ATMEL_QSPI_ENABLE_DUMMY |
	   ATMEL_QSPI_ENABLE_DATA);

Maybe a little bit more optimal considering the generated code and endianness
independent but a little less easy to read?

>>> +		u32	word;
>>> +	}	enable;
>>> +	u8	instruction;
>>> +	u8	mode;
>>> +	u8	num_mode_cycles;
>>> +	u8	num_dummy_cycles;
>>> +	u32	address;
>>> +
>>> +	size_t		buf_len;
>>> +	const void	*tx_buf;
>>> +	void		*rx_buf;
>>> +};
>>> +
>>
>> [...]
>>
>>> +static int atmel_qspi_probe(struct platform_device *pdev)
>>> +{
>>> +	struct device_node *child, *np = pdev->dev.of_node;
>>> +	char modalias[SPI_NAME_SIZE];
>>> +	const char *name = NULL;
>>> +	struct atmel_qspi *aq;
>>> +	struct resource *res;
>>> +	struct spi_nor *nor;
>>> +	struct mtd_info *mtd;
>>> +	int irq, err = 0;
>>> +
>>> +	if (of_get_child_count(np) != 1)
>>> +		return -ENODEV;
>>> +	child = of_get_next_child(np, NULL);
>>> +
>>> +	aq = devm_kzalloc(&pdev->dev, sizeof(*aq), GFP_KERNEL);
>>> +	if (!aq) {
>>> +		err = -ENOMEM;
>>> +		goto exit;
>>> +	}
>>> +
>>> +	platform_set_drvdata(pdev, aq);
>>> +	init_completion(&aq->cmd_completion);
>>> +	aq->pdev = pdev;
>>> +
>>> +	/* Map the registers */
>>> +	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "qspi_base");
>>> +	aq->regs = devm_ioremap_resource(&pdev->dev, res);
>>> +	if (IS_ERR(aq->regs)) {
>>> +		dev_err(&pdev->dev, "missing registers\n");
>>> +		err = PTR_ERR(aq->regs);
>>> +		goto exit;
>>> +	}
>>> +
>>> +	/* Map the AHB memory */
>>> +	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "qspi_mmap");
>>> +	aq->mem = devm_ioremap_resource(&pdev->dev, res);
>>> +	if (IS_ERR(aq->mem)) {
>>> +		dev_err(&pdev->dev, "missing AHB memory\n");
>>> +		err = PTR_ERR(aq->mem);
>>> +		goto exit;
>>> +	}
>>> +
>>> +	/* Get the peripheral clock */
>>> +	aq->clk = devm_clk_get(&pdev->dev, NULL);
>>> +	if (IS_ERR(aq->clk)) {
>>> +		dev_err(&pdev->dev, "missing peripheral clock\n");
>>> +		err = PTR_ERR(aq->clk);
>>> +		goto exit;
>>> +	}
>>> +
>>> +	/* Enable the peripheral clock */
>>> +	err = clk_prepare_enable(aq->clk);
>>> +	if (err) {
>>> +		dev_err(&pdev->dev, "failed to enable the peripheral clock\n");
>>> +		goto exit;
>>> +	}
>>> +
>>> +	/* Request the IRQ */
>>> +	irq = platform_get_irq(pdev, 0);
>>> +	if (irq < 0) {
>>> +		dev_err(&pdev->dev, "missing IRQ\n");
>>> +		err = irq;
>>> +		goto disable_clk;
>>> +	}
>>> +	err = devm_request_irq(&pdev->dev, irq, atmel_qspi_interrupt,
>>> +			       0, dev_name(&pdev->dev), aq);
>>> +	if (err)
>>> +		goto disable_clk;
>>> +
>>> +	/* Setup the spi-nor */
>>> +	nor = &aq->nor;
>>> +	mtd = &nor->mtd;
>>> +
>>> +	nor->dev = &pdev->dev;
>>> +	spi_nor_set_flash_node(nor, child);
>>> +	nor->priv = aq;
>>> +	mtd->priv = nor;
>>> +
>>> +	nor->read_reg = atmel_qspi_read_reg;
>>> +	nor->write_reg = atmel_qspi_write_reg;
>>> +	nor->read = atmel_qspi_read;
>>> +	nor->write = atmel_qspi_write;
>>> +	nor->erase = atmel_qspi_erase;
>>> +
>>> +	err = of_modalias_node(child, modalias, sizeof(modalias));
>>> +	if (err < 0)
>>> +		goto disable_clk;
>>> +
>>> +	if (strcmp(modalias, "spi-nor"))
>>> +		name = modalias;
>>
>> What is this modalias handling for? Are you trying to support passing in
>> a flash-specific string, like m25p80 does? And you're enforcing that to
>> be only the first entry in the 'compatible' list? Seems fragile; m25p80
>> is a special case (and it's already fragile) that I'd rather not
>> imitate.
>>
>> I understand that we discussed extending the use of compatible for
>> spi-nor drivers, but I still don't think that issue is resolved. And if
>> we are going to extend the use of compatible, then I think we should
>> fixup all the spi-nor drivers in the same way.
>>

Yes you're right this part of the code was introduce at the driver side when
I had to use the Macronix mx25l25673g memory, which shares the very same JEDEC
ID as other Macronix memories but with other hardware features.
I no longer use it since I use my series about SFDP parsing.

So I'm perfectly fine with removing this modalias handling from the driver.
Now I guess I will focus on the SFDP solution.

Also this qspi driver is ported to mailine from our linux-at91 kernels since we
need to provide users a way to support QSPI memories in Linux with our QSPI
controller.

>>> +
>>> +	err = of_property_read_u32(child, "spi-max-frequency", &aq->clk_rate);
>>> +	if (err < 0)
>>> +		goto disable_clk;
>>> +
>>> +	err = atmel_qspi_init(aq);
>>> +	if (err)
>>> +		goto disable_clk;
>>> +
>>> +	err = spi_nor_scan(nor, name, SPI_NOR_QUAD);
>>> +	if (err)
>>> +		goto disable_clk;
>>> +
>>> +	err = mtd_device_register(mtd, NULL, 0);
>>> +	if (err)
>>> +		goto disable_clk;
>>> +
>>> +	of_node_put(child);
>>> +
>>> +	return 0;
>>> +
>>> +disable_clk:
>>> +	clk_disable_unprepare(aq->clk);
>>> +exit:
>>> +	of_node_put(child);
>>> +
>>> +	return err;
>>> +}
>>> +
>>> +static int atmel_qspi_remove(struct platform_device *pdev)
>>> +{
>>> +	struct atmel_qspi *aq = platform_get_drvdata(pdev);
>>> +
>>> +	mtd_device_unregister(&aq->nor.mtd);
>>> +	qspi_writel(aq, QSPI_CR, QSPI_CR_QSPIDIS);
>>> +	clk_disable_unprepare(aq->clk);
>>> +	return 0;
>>> +}
>>> +
>>> +
>>> +static const struct of_device_id atmel_qspi_dt_ids[] = {
>>> +	{ .compatible = "atmel,sama5d2-qspi" },
>>> +	{ /* sentinel */ }
>>> +};
>>> +
>>> +MODULE_DEVICE_TABLE(of, atmel_qspi_dt_ids);
>>> +
>>> +static struct platform_driver atmel_qspi_driver = {
>>> +	.driver = {
>>> +		.name	= "atmel_qspi",
>>> +		.of_match_table	= atmel_qspi_dt_ids,
>>> +	},
>>> +	.probe		= atmel_qspi_probe,
>>> +	.remove		= atmel_qspi_remove,
>>> +};
>>> +module_platform_driver(atmel_qspi_driver);
>>> +
>>> +MODULE_AUTHOR("Cyrille Pitchen <cyrille.pitchen@atmel.com>");
>>> +MODULE_DESCRIPTION("Atmel QSPI Controller driver");
>>> +MODULE_LICENSE("GPL v2");
>>
>> I'm likely to apply this driver, with the following diff (the bitfield
>> issue (if there is one) is less of a maintenance issue), barring any
>> major objections.
>>
>> diff --git a/drivers/mtd/spi-nor/atmel-quadspi.c b/drivers/mtd/spi-nor/atmel-quadspi.c
>> index 06d1bf276dd0..47937d9beec6 100644
>> --- a/drivers/mtd/spi-nor/atmel-quadspi.c
>> +++ b/drivers/mtd/spi-nor/atmel-quadspi.c
>> @@ -591,8 +591,6 @@ static irqreturn_t atmel_qspi_interrupt(int irq, void *dev_id)
>>  static int atmel_qspi_probe(struct platform_device *pdev)
>>  {
>>  	struct device_node *child, *np = pdev->dev.of_node;
>> -	char modalias[SPI_NAME_SIZE];
>> -	const char *name = NULL;
>>  	struct atmel_qspi *aq;
>>  	struct resource *res;
>>  	struct spi_nor *nor;
>> @@ -673,13 +671,6 @@ static int atmel_qspi_probe(struct platform_device *pdev)
>>  	nor->write = atmel_qspi_write;
>>  	nor->erase = atmel_qspi_erase;
>>  
>> -	err = of_modalias_node(child, modalias, sizeof(modalias));
>> -	if (err < 0)
>> -		goto disable_clk;
>> -
>> -	if (strcmp(modalias, "spi-nor"))
>> -		name = modalias;
>> -
>>  	err = of_property_read_u32(child, "spi-max-frequency", &aq->clk_rate);
>>  	if (err < 0)
>>  		goto disable_clk;
>> @@ -688,7 +679,7 @@ static int atmel_qspi_probe(struct platform_device *pdev)
>>  	if (err)
>>  		goto disable_clk;
>>  
>> -	err = spi_nor_scan(nor, name, SPI_NOR_QUAD);
>> +	err = spi_nor_scan(nor, NULL, SPI_NOR_QUAD);
>>  	if (err)
>>  		goto disable_clk;
>>  
> 
> Applied to l2-mtd.git with that fixup.
> 

It's OK for me, thanks! :)

Best regards,

Cyrille

Re: [PATCH 0/2] mtd: spi-nor: add driver for Atmel QSPI controller

[Cyrille Pitchen]

Hi Brian,

Almost a year has passed since the first publication of the driver for the
Atmel Quad SPI controller:
https://patchwork.kernel.org/patch/6808311/

I totally understand the previous versions needed to be discussed as they were
based on a deep rework of the spi-nor framework to enhance the support of Quad
SPI protocols but this latest version does not require any modification of the
spi-nor framework: now the driver only supports SPI 1-1-1 and 1-1-4 protocols
like other spi-nor controller drivers already accepted.

So I guess this time there is no reason not to integrate this version.

I've also included this very same patch inside my latest series for Quad SPI
protocols support and SFDP table parsing.
Honestly the Atmel QSPI controller driver itself has nothing to do in this
series dedicated to support of Quad SPI memories but the QSPI controller of
the sama5d2 SoC is the only hardware I can use to test my patches.

Also, since patch 5/9 "mtd: spi-nor: add support of SPI protocols like
SPI 1-2-2 and SPI 1-4-4" updates all drivers calling spi_nor_scan(), including
atmel-quadspi.c, I needed to include the Atmel QSPI controller driver first
in the series before actually doing the core modifications of spi-nor.c.
https://patchwork.ozlabs.org/patch/638138/

I'm not trying to by-pass or short-cut the review process but I feel that
there is currently no discussion or other feedbacks suggesting that the
QSPI controller driver itself needs further modifications.

Best regards,

Cyrille


Le 13/06/2016 17:10, Cyrille Pitchen a écrit :
> Hi all,
> 
> this series of patches adds support to the Atmel QSPI controller.
> It was tested on a sama5d2 xplained board + Macronix mx25l25673g QSPI
> memory.
> 
> The driver limits its support of Quad SPI memories to the SPI 1-1-4
> protocol for Fast Read operations and SPI 1-1-1 protocol for all other
> commands.
> 
> The series is based on next-20160609
> 
> Best regards,
> 
> Cyrille
> 
> Cyrille Pitchen (2):
>   Documentation: atmel-quadspi: add binding file for Atmel QSPI driver
>   mtd: atmel-quadspi: add driver for Atmel QSPI controller
> 
>  .../devicetree/bindings/mtd/atmel-quadspi.txt      |  32 +
>  drivers/mtd/spi-nor/Kconfig                        |   9 +
>  drivers/mtd/spi-nor/Makefile                       |   1 +
>  drivers/mtd/spi-nor/atmel-quadspi.c                | 741 +++++++++++++++++++++
>  4 files changed, 783 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/mtd/atmel-quadspi.txt
>  create mode 100644 drivers/mtd/spi-nor/atmel-quadspi.c
> 

[PATCH 0/2] mtd: spi-nor: add driver for Atmel QSPI controller

[Cyrille Pitchen]

Hi all,

this series of patches add support to the Atmel QSPI controller.
It was tested on a sama5d2 xplained board + Macronix mx25l25673g QSPI
memory.

The series is based on Michal and Brian series:
"[PATCH v7 00/10] Add error checking to spi-nor read and write"

Best regards,

Cyrille

Cyrille Pitchen (2):
  Documentation: atmel-quadspi: add binding file for Atmel QSPI driver
  mtd: atmel-quadspi: add driver for Atmel QSPI controller

 .../devicetree/bindings/mtd/atmel-quadspi.txt      |  32 +
 drivers/mtd/spi-nor/Kconfig                        |   9 +
 drivers/mtd/spi-nor/Makefile                       |   1 +
 drivers/mtd/spi-nor/atmel-quadspi.c                | 741 +++++++++++++++++++++
 4 files changed, 783 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/mtd/atmel-quadspi.txt
 create mode 100644 drivers/mtd/spi-nor/atmel-quadspi.c

-- 
1.8.2.2

[PATCH 1/2] Documentation: atmel-quadspi: add binding file for Atmel QSPI driver

[Cyrille Pitchen]

This patch documents the DT bindings for the driver of the Atmel QSPI
controller embedded inside sama5d2x SoCs.

Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com>
Acked-by: Rob Herring <robh@kernel.org>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
 .../devicetree/bindings/mtd/atmel-quadspi.txt      | 32 ++++++++++++++++++++++
 1 file changed, 32 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/mtd/atmel-quadspi.txt

diff --git a/Documentation/devicetree/bindings/mtd/atmel-quadspi.txt b/Documentation/devicetree/bindings/mtd/atmel-quadspi.txt
new file mode 100644
index 000000000000..489807005eda
--- /dev/null
+++ b/Documentation/devicetree/bindings/mtd/atmel-quadspi.txt
@@ -0,0 +1,32 @@
+* Atmel Quad Serial Peripheral Interface (QSPI)
+
+Required properties:
+- compatible:     Should be "atmel,sama5d2-qspi".
+- reg:            Should contain the locations and lengths of the base registers
+                  and the mapped memory.
+- reg-names:      Should contain the resource reg names:
+                  - qspi_base: configuration register address space
+                  - qspi_mmap: memory mapped address space
+- interrupts:     Should contain the interrupt for the device.
+- clocks:         The phandle of the clock needed by the QSPI controller.
+- #address-cells: Should be <1>.
+- #size-cells:    Should be <0>.
+
+Example:
+
+spi@f0020000 {
+	compatible = "atmel,sama5d2-qspi";
+	reg = <0xf0020000 0x100>, <0xd0000000 0x8000000>;
+	reg-names = "qspi_base", "qspi_mmap";
+	interrupts = <52 IRQ_TYPE_LEVEL_HIGH 7>;
+	clocks = <&spi0_clk>;
+	#address-cells = <1>;
+	#size-cells = <0>;
+	pinctrl-names = "default";
+	pinctrl-0 = <&pinctrl_spi0_default>;
+	status = "okay";
+
+	m25p80@0 {
+		...
+	};
+};
-- 
1.8.2.2