- Add driver for NXP FlexSPI host controller
FlexSPI is a flexsible SPI host controller [1], Chapter 30 page 1475,
which supports two SPI channels and up to 4 external devices.
Each channel supports Single/Dual/Quad/Octal mode data transfer (1/2/4/8 bidirectional data lines)
i.e. FlexSPI acts as an interface to external devices, maximum 4, each with up to 8
bidirectional data lines.
- Tested this driver with mtd_debug(Erase/Write/Read) utility and JFFS2
filesystem mounting and booting on NXP LX2160ARDB[2] and LX2160AQDS targets.
LX2160ARDB is having two NOR slave device connected on single bus A
i.e. A0 and A1 (CS0 and CS1).
LX2160AQDS is having two NOR slave device connected on separate buses
one flash on A0 and second on B1 i.e. (CS0 and CS3).
Verified this driver on following SPI NOR flashes:
Micron, mt35xu512aba[3], [Read - 1 bit mode]
Cypress, s25fl512s, [Read - 1/2/4 bit mode]
Support for octal flash commands and other Octal data transfer related framework
changes work is in progress.
[1] https://www.nxp.com/docs/en/reference-manual/IMXRT1050RM.pdf
[2] https://patchwork.kernel.org/project/linux-arm-kernel/list/?series=9721
[3] https://patchwork.ozlabs.org/project/linux-mtd/list/?series=63445
Yogesh Gaur (5):
spi: spi-mem: Add driver for NXP FlexSPI controller
dt-bindings: spi: add binding file for NXP FlexSPI controller
arm64: dts: lx2160a: add FlexSPI node property
arm64: defconfig: enable NXP FlexSPI driver
MAINTAINERS: add maintainers for the NXP FlexSPI driver
Changes for v3:
- Incorporated review comments for
patch 'spi: spi-mem: Add driver for NXP FlexSPI controller'.
Changes for v2:
- Incorporated Boris review comments and drop below patches as per the comments.
- Patch 'spi: add slave device size in spi_device struct'
- Patch 'spi: add flags for octal I/O data transfer'
- Incorporated DTS and Binding file review comments of Shawn Guo and Rob Herring.
.../devicetree/bindings/spi/spi-nxp-fspi.txt | 42 +
MAINTAINERS | 6 +
arch/arm64/boot/dts/freescale/fsl-lx2160a-rdb.dts | 22 +
arch/arm64/boot/dts/freescale/fsl-lx2160a.dtsi | 12 +
arch/arm64/configs/defconfig | 1 +
drivers/spi/Kconfig | 10 +
drivers/spi/Makefile | 1 +
drivers/spi/spi-nxp-fspi.c | 1231 ++++++++++++++++++++
8 files changed, 1324 insertions(+)
create mode 100644 Documentation/devicetree/bindings/spi/spi-nxp-fspi.txt
create mode 100644 drivers/spi/spi-nxp-fspi.c
--
2.7.4
- Add driver for NXP FlexSPI host controller
(0) What is the FlexSPI controller?
FlexSPI is a flexsible SPI host controller which supports two SPI
channels and up to 4 external devices. Each channel supports
Single/Dual/Quad/Octal mode data transfer (1/2/4/8 bidirectional
data lines) i.e. FlexSPI acts as an interface to external devices,
maximum 4, each with up to 8 bidirectional data lines.
It uses new SPI memory interface of the SPI framework to issue
flash memory operations to up to four connected flash
devices (2 buses with 2 CS each).
(1) Tested this driver with the mtd_debug and JFFS2 filesystem utility
on NXP LX2160ARDB and LX2160AQDS targets.
LX2160ARDB is having two NOR slave device connected on single bus A
i.e. A0 and A1 (CS0 and CS1).
LX2160AQDS is having two NOR slave device connected on separate buses
one flash on A0 and second on B1 i.e. (CS0 and CS3).
Verified this driver on following SPI NOR flashes:
Micron, mt35xu512ab, [Read - 1 bit mode]
Cypress, s25fl512s, [Read - 1/2/4 bit mode]
Signed-off-by: Yogesh Gaur <[email protected]>
---
Changes for v3:
- Added endianness flag in platform specific structure instead of DTS.
- Modified nxp_fspi_read_ahb(), removed remapping code.
Changes for v2:
- Incorporated Boris review comments.
- Remove dependency of driver over connected flash device size.
- Modified the logic to select requested CS.
- Remove SPI-Octal Macros.
drivers/spi/Kconfig | 10 +
drivers/spi/Makefile | 1 +
drivers/spi/spi-nxp-fspi.c | 1231 ++++++++++++++++++++++++++++++++++++++++++++
3 files changed, 1241 insertions(+)
create mode 100644 drivers/spi/spi-nxp-fspi.c
diff --git a/drivers/spi/Kconfig b/drivers/spi/Kconfig
index ad5d68e..68da874 100644
--- a/drivers/spi/Kconfig
+++ b/drivers/spi/Kconfig
@@ -251,6 +251,16 @@ config SPI_FSL_LPSPI
help
This enables Freescale i.MX LPSPI controllers in master mode.
+config SPI_NXP_FLEXSPI
+ tristate "NXP Flex SPI controller"
+ depends on ARCH_LAYERSCAPE || HAS_IOMEM
+ help
+ This enables support for the Flex SPI controller in master mode.
+ Up to four slave devices can be connected on two buses with two
+ chipselects each.
+ This controller does not support generic SPI messages and only
+ supports the high-level SPI memory interface.
+
config SPI_GPIO
tristate "GPIO-based bitbanging SPI Master"
depends on GPIOLIB || COMPILE_TEST
diff --git a/drivers/spi/Makefile b/drivers/spi/Makefile
index cb1f437..52c9f18 100644
--- a/drivers/spi/Makefile
+++ b/drivers/spi/Makefile
@@ -59,6 +59,7 @@ obj-$(CONFIG_SPI_MPC52xx) += spi-mpc52xx.o
obj-$(CONFIG_SPI_MT65XX) += spi-mt65xx.o
obj-$(CONFIG_SPI_MXS) += spi-mxs.o
obj-$(CONFIG_SPI_NUC900) += spi-nuc900.o
+obj-$(CONFIG_SPI_NXP_FLEXSPI) += spi-nxp-fspi.o
obj-$(CONFIG_SPI_OC_TINY) += spi-oc-tiny.o
spi-octeon-objs := spi-cavium.o spi-cavium-octeon.o
obj-$(CONFIG_SPI_OCTEON) += spi-octeon.o
diff --git a/drivers/spi/spi-nxp-fspi.c b/drivers/spi/spi-nxp-fspi.c
new file mode 100644
index 0000000..0967651
--- /dev/null
+++ b/drivers/spi/spi-nxp-fspi.c
@@ -0,0 +1,1230 @@
+// SPDX-License-Identifier: GPL-2.0+
+
+/*
+ * NXP FlexSPI(FSPI) controller driver.
+ *
+ * Copyright 2018 NXP.
+ *
+ * FlexSPI is a flexsible SPI host controller which supports two SPI
+ * channels and up to 4 external devices. Each channel supports
+ * Single/Dual/Quad/Octal mode data transfer (1/2/4/8 bidirectional
+ * data lines).
+ *
+ * FlexSPI controller is driven by the LUT(Look-up Table) registers
+ * LUT registers are a look-up-table for sequences of instructions.
+ * A valid sequence consists of four LUT registers.
+ * Maximum 32 LUT sequences can be programmed simultaneously.
+ *
+ * LUTs are being created at run-time based on the commands passed
+ * from the spi-mem framework, thus using single LUT index.
+ *
+ * Software triggered Flash read/write access by IP Bus.
+ *
+ * Memory mapped read access by AHB Bus.
+ *
+ * Based on SPI MEM interface and spi-fsl-qspi.c driver.
+ *
+ * Author:
+ * Yogesh Gaur <[email protected]>
+ * Boris Brezillion <[email protected]>
+ * Frieder Schrempf <[email protected]>
+ */
+
+#include <linux/bitops.h>
+#include <linux/clk.h>
+#include <linux/completion.h>
+#include <linux/delay.h>
+#include <linux/err.h>
+#include <linux/errno.h>
+#include <linux/interrupt.h>
+#include <linux/io.h>
+#include <linux/jiffies.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+#include <linux/platform_device.h>
+#include <linux/pm_qos.h>
+#include <linux/sizes.h>
+
+#include <linux/spi/spi.h>
+#include <linux/spi/spi-mem.h>
+
+/*
+ * The driver only uses one single LUT entry, that is updated on
+ * each call of exec_op(). Index 0 is preset at boot with a basic
+ * read operation, so let's use the last entry (31).
+ */
+#define SEQID_LUT 31
+
+/* Registers used by the driver */
+#define FSPI_MCR0 0x00
+#define FSPI_MCR0_AHB_TIMEOUT_SHIFT 24
+#define FSPI_MCR0_AHB_TIMEOUT_MASK (0xFF << FSPI_MCR0_AHB_TIMEOUT_SHIFT)
+#define FSPI_MCR0_IP_TIMEOUT_SHIFT 16
+#define FSPI_MCR0_IP_TIMEOUT_MASK (0xFF << FSPI_MCR0_IP_TIMEOUT_SHIFT)
+#define FSPI_MCR0_LEARN_EN_SHIFT 15
+#define FSPI_MCR0_LEARN_EN_MASK (1 << FSPI_MCR0_LEARN_EN_SHIFT)
+#define FSPI_MCR0_SCRFRUN_EN_SHIFT 14
+#define FSPI_MCR0_SCRFRUN_EN_MASK (1 << FSPI_MCR0_SCRFRUN_EN_SHIFT)
+#define FSPI_MCR0_OCTCOMB_EN_SHIFT 13
+#define FSPI_MCR0_OCTCOMB_EN_MASK (1 << FSPI_MCR0_OCTCOMB_EN_SHIFT)
+#define FSPI_MCR0_DOZE_EN_SHIFT 12
+#define FSPI_MCR0_DOZE_EN_MASK (1 << FSPI_MCR0_DOZE_EN_SHIFT)
+#define FSPI_MCR0_HSEN_SHIFT 11
+#define FSPI_MCR0_HSEN_MASK (1 << FSPI_MCR0_HSEN_SHIFT)
+#define FSPI_MCR0_SERCLKDIV_SHIFT 8
+#define FSPI_MCR0_SERCLKDIV_MASK (7 << FSPI_MCR0_SERCLKDIV_SHIFT)
+#define FSPI_MCR0_ATDF_EN_SHIFT 7
+#define FSPI_MCR0_ATDF_EN_MASK (1 << FSPI_MCR0_ATDF_EN_SHIFT)
+#define FSPI_MCR0_ARDF_EN_SHIFT 6
+#define FSPI_MCR0_ARDF_EN_MASK (1 << FSPI_MCR0_ARDF_EN_SHIFT)
+#define FSPI_MCR0_RXCLKSRC_SHIFT 4
+#define FSPI_MCR0_RXCLKSRC_MASK (3 << FSPI_MCR0_RXCLKSRC_SHIFT)
+#define FSPI_MCR0_END_CFG_SHIFT 2
+#define FSPI_MCR0_END_CFG_MASK (3 << FSPI_MCR0_END_CFG_SHIFT)
+#define FSPI_MCR0_MDIS_SHIFT 1
+#define FSPI_MCR0_MDIS_MASK (1 << FSPI_MCR0_MDIS_SHIFT)
+#define FSPI_MCR0_SWRST_SHIFT 0
+#define FSPI_MCR0_SWRST_MASK (1 << FSPI_MCR0_SWRST_SHIFT)
+
+#define FSPI_MCR1 0x04
+#define FSPI_MCR1_SEQ_TIMEOUT_SHIFT 16
+#define FSPI_MCR1_SEQ_TIMEOUT_MASK (0xFFFF << FSPI_MCR1_SEQ_TIMEOUT_SHIFT)
+#define FSPI_MCR1_AHB_TIMEOUT_SHIFT 0
+#define FSPI_MCR1_AHB_TIMEOUT_MASK (0xFFFF << FSPI_MCR1_AHB_TIMEOUT_SHIFT)
+
+#define FSPI_MCR2 0x08
+#define FSPI_MCR2_IDLE_WAIT_SHIFT 24
+#define FSPI_MCR2_IDLE_WAIT_MASK (0xFF << FSPI_MCR2_IDLE_WAIT_SHIFT)
+#define FSPI_MCR2_SAMEDEVICEEN_SHIFT 15
+#define FSPI_MCR2_SAMEDEVICEEN_MASK (1 << FSPI_MCR2_SAMEDEVICEEN_SHIFT)
+#define FSPI_MCR2_CLRLRPHS_SHIFT 14
+#define FSPI_MCR2_CLRLRPHS_MASK (1 << FSPI_MCR2_CLRLRPHS_SHIFT)
+#define FSPI_MCR2_ABRDATSZ_SHIFT 8
+#define FSPI_MCR2_ABRDATSZ_MASK (1 << FSPI_MCR2_ABRDATSZ_SHIFT)
+#define FSPI_MCR2_ABRLEARN_SHIFT 7
+#define FSPI_MCR2_ABRLEARN_MASK (1 << FSPI_MCR2_ABRLEARN_SHIFT)
+#define FSPI_MCR2_ABR_READ_SHIFT 6
+#define FSPI_MCR2_ABR_READ_MASK (1 << FSPI_MCR2_ABR_READ_SHIFT)
+#define FSPI_MCR2_ABRWRITE_SHIFT 5
+#define FSPI_MCR2_ABRWRITE_MASK (1 << FSPI_MCR2_ABRWRITE_SHIFT)
+#define FSPI_MCR2_ABRDUMMY_SHIFT 4
+#define FSPI_MCR2_ABRDUMMY_MASK (1 << FSPI_MCR2_ABRDUMMY_SHIFT)
+#define FSPI_MCR2_ABR_MODE_SHIFT 3
+#define FSPI_MCR2_ABR_MODE_MASK (1 << FSPI_MCR2_ABR_MODE_SHIFT)
+#define FSPI_MCR2_ABRCADDR_SHIFT 2
+#define FSPI_MCR2_ABRCADDR_MASK (1 << FSPI_MCR2_ABRCADDR_SHIFT)
+#define FSPI_MCR2_ABRRADDR_SHIFT 1
+#define FSPI_MCR2_ABRRADDR_MASK (1 << FSPI_MCR2_ABRRADDR_SHIFT)
+#define FSPI_MCR2_ABR_CMD_SHIFT 0
+#define FSPI_MCR2_ABR_CMD_MASK (1 << FSPI_MCR2_ABR_CMD_SHIFT)
+
+#define FSPI_AHBCR 0x0c
+#define FSPI_AHBCR_RDADDROPT_SHIFT 6
+#define FSPI_AHBCR_RDADDROPT_MASK (1 << FSPI_AHBCR_RDADDROPT_SHIFT)
+#define FSPI_AHBCR_PREF_EN_SHIFT 5
+#define FSPI_AHBCR_PREF_EN_MASK (1 << FSPI_AHBCR_PREF_EN_SHIFT)
+#define FSPI_AHBCR_BUFF_EN_SHIFT 4
+#define FSPI_AHBCR_BUFF_EN_MASK (1 << FSPI_AHBCR_BUFF_EN_SHIFT)
+#define FSPI_AHBCR_CACH_EN_SHIFT 3
+#define FSPI_AHBCR_CACH_EN_MASK (1 << FSPI_AHBCR_CACH_EN_SHIFT)
+#define FSPI_AHBCR_CLRTXBUF_SHIFT 2
+#define FSPI_AHBCR_CLRTXBUF_MASK (1 << FSPI_AHBCR_CLRTXBUF_SHIFT)
+#define FSPI_AHBCR_CLRRXBUF_SHIFT 1
+#define FSPI_AHBCR_CLRRXBUF_MASK (1 << FSPI_AHBCR_CLRRXBUF_SHIFT)
+#define FSPI_AHBCR_PAR_EN_SHIFT 0
+#define FSPI_AHBCR_PAR_EN_MASK (1 << FSPI_AHBCR_PAR_EN_SHIFT)
+
+#define FSPI_INTEN 0x10
+#define FSPI_INTEN_SCLKSBWR_SHIFT 9
+#define FSPI_INTEN_SCLKSBWR_MASK (1 << FSPI_INTEN_SCLKSBWR_SHIFT)
+#define FSPI_INTEN_SCLKSBRD_SHIFT 8
+#define FSPI_INTEN_SCLKSBRD_MASK (1 << FSPI_INTEN_SCLKSBRD_SHIFT)
+#define FSPI_INTEN_DATALRNFL_SHIFT 7
+#define FSPI_INTEN_DATALRNFL_MASK (1 << FSPI_INTEN_DATALRNFL_SHIFT)
+#define FSPI_INTEN_IPTXWE_SHIFT 6
+#define FSPI_INTEN_IPTXWE_MASK (1 << FSPI_INTEN_IPTXWE_SHIFT)
+#define FSPI_INTEN_IPRXWA_SHIFT 5
+#define FSPI_INTEN_IPRXWA_MASK (1 << FSPI_INTEN_IPRXWA_SHIFT)
+#define FSPI_INTEN_AHBCMDERR_SHIFT 4
+#define FSPI_INTEN_AHBCMDERR_MASK (1 << FSPI_INTEN_AHBCMDERR_SHIFT)
+#define FSPI_INTEN_IPCMDERR_SHIFT 3
+#define FSPI_INTEN_IPCMDERR_MASK (1 << FSPI_INTEN_IPCMDERR_SHIFT)
+#define FSPI_INTEN_AHBCMDGE_SHIFT 2
+#define FSPI_INTEN_AHBCMDGE_MASK (1 << FSPI_INTEN_AHBCMDGE_SHIFT)
+#define FSPI_INTEN_IPCMDGE_SHIFT 1
+#define FSPI_INTEN_IPCMDGE_MASK (1 << FSPI_INTEN_IPCMDGE_SHIFT)
+#define FSPI_INTEN_IPCMDDONE_SHIFT 0
+#define FSPI_INTEN_IPCMDDONE_MASK (1 << FSPI_INTEN_IPCMDDONE_SHIFT)
+
+#define FSPI_INTR 0x14
+#define FSPI_INTR_SCLKSBWR_SHIFT 9
+#define FSPI_INTR_SCLKSBWR_MASK (1 << FSPI_INTR_SCLKSBWR_SHIFT)
+#define FSPI_INTR_SCLKSBRD_SHIFT 8
+#define FSPI_INTR_SCLKSBRD_MASK (1 << FSPI_INTR_SCLKSBRD_SHIFT)
+#define FSPI_INTR_DATALRNFL_SHIFT 7
+#define FSPI_INTR_DATALRNFL_MASK (1 << FSPI_INTR_DATALRNFL_SHIFT)
+#define FSPI_INTR_IPTXWE_SHIFT 6
+#define FSPI_INTR_IPTXWE_MASK (1 << FSPI_INTR_IPTXWE_SHIFT)
+#define FSPI_INTR_IPRXWA_SHIFT 5
+#define FSPI_INTR_IPRXWA_MASK (1 << FSPI_INTR_IPRXWA_SHIFT)
+#define FSPI_INTR_AHBCMDERR_SHIFT 4
+#define FSPI_INTR_AHBCMDERR_MASK (1 << FSPI_INTR_AHBCMDERR_SHIFT)
+#define FSPI_INTR_IPCMDERR_SHIFT 3
+#define FSPI_INTR_IPCMDERR_MASK (1 << FSPI_INTR_IPCMDERR_SHIFT)
+#define FSPI_INTR_AHBCMDGE_SHIFT 2
+#define FSPI_INTR_AHBCMDGE_MASK (1 << FSPI_INTR_AHBCMDGE_SHIFT)
+#define FSPI_INTR_IPCMDGE_SHIFT 1
+#define FSPI_INTR_IPCMDGE_MASK (1 << FSPI_INTR_IPCMDGE_SHIFT)
+#define FSPI_INTR_IPCMDDONE_SHIFT 0
+#define FSPI_INTR_IPCMDDONE_MASK (1 << FSPI_INTR_IPCMDDONE_SHIFT)
+
+#define FSPI_LUTKEY 0x18
+#define FSPI_LUTKEY_VALUE 0x5AF05AF0
+
+#define FSPI_LCKCR 0x1C
+
+#define FSPI_LCKER_LOCK 0x1
+#define FSPI_LCKER_UNLOCK 0x2
+
+#define FSPI_BUFXCR_INVALID_MSTRID 0xE
+#define FSPI_AHBRX_BUF0CR0 0x20
+#define FSPI_AHBRX_BUF1CR0 0x24
+#define FSPI_AHBRX_BUF2CR0 0x28
+#define FSPI_AHBRX_BUF3CR0 0x2C
+#define FSPI_AHBRX_BUF4CR0 0x30
+#define FSPI_AHBRX_BUF5CR0 0x34
+#define FSPI_AHBRX_BUF6CR0 0x38
+#define FSPI_AHBRX_BUF7CR0 0x3C
+#define FSPI_AHBRXBUF0CR7_PREF_SHIFT 31
+#define FSPI_AHBRXBUF0CR7_PREF_MASK (1 << FSPI_AHBRXBUF0CR7_PREF_SHIFT)
+
+#define FSPI_AHBRX_BUF0CR1 0x40
+#define FSPI_AHBRX_BUF1CR1 0x44
+#define FSPI_AHBRX_BUF2CR1 0x48
+#define FSPI_AHBRX_BUF3CR1 0x4C
+#define FSPI_AHBRX_BUF4CR1 0x50
+#define FSPI_AHBRX_BUF5CR1 0x54
+#define FSPI_AHBRX_BUF6CR1 0x58
+#define FSPI_AHBRX_BUF7CR1 0x5C
+#define FSPI_BUFXCR1_MSID_SHIFT 0x0
+#define FSPI_BUFXCR1_MSID_MASK (0xF << FSPI_BUFXCR1_MSID_SHIFT)
+#define FSPI_BUFXCR1_PRIO_SHIFT 0x8
+#define FSPI_BUFXCR1_PRIO_MASK (0x7 << FSPI_BUFXCR1_PRIO_SHIFT)
+
+#define FSPI_FLSHA1CR0 0x60
+#define FSPI_FLSHA2CR0 0x64
+#define FSPI_FLSHB1CR0 0x68
+#define FSPI_FLSHB2CR0 0x6C
+#define FSPI_FLSHXCR0_SZ_SHIFT 10
+#define FSPI_FLSHXCR0_SZ_MASK (0x3FFFFF << FSPI_FLSHXCR0_SZ_SHIFT)
+
+#define FSPI_FLSHA1CR1 0x70
+#define FSPI_FLSHA2CR1 0x74
+#define FSPI_FLSHB1CR1 0x78
+#define FSPI_FLSHB2CR1 0x7C
+#define FSPI_FLSHXCR1_CSINTR_SHIFT 16
+#define FSPI_FLSHXCR1_CSINTR_MASK \
+ (0xFFFF << FSPI_FLSHXCR1_CSINTR_SHIFT)
+#define FSPI_FLSHXCR1_CAS_SHIFT 11
+#define FSPI_FLSHXCR1_CAS_MASK (0xF << FSPI_FLSHXCR1_CAS_SHIFT)
+#define FSPI_FLSHXCR1_WA_SHIFT 10
+#define FSPI_FLSHXCR1_WA_MASK (1 << FSPI_FLSHXCR1_WA_SHIFT)
+#define FSPI_FLSHXCR1_TCSH_SHIFT 5
+#define FSPI_FLSHXCR1_TCSH_MASK (0x1F << FSPI_FLSHXCR1_TCSH_SHIFT)
+#define FSPI_FLSHXCR1_TCSS_SHIFT 0
+#define FSPI_FLSHXCR1_TCSS_MASK (0x1F << FSPI_FLSHXCR1_TCSS_SHIFT)
+
+#define FSPI_FLSHA1CR2 0x80
+#define FSPI_FLSHA2CR2 0x84
+#define FSPI_FLSHB1CR2 0x88
+#define FSPI_FLSHB2CR2 0x8C
+#define FSPI_FLSHXCR2_CLRINSP_SHIFT 24
+#define FSPI_FLSHXCR2_CLRINSP_MASK (1 << FSPI_FLSHXCR2_CLRINSP_SHIFT)
+#define FSPI_FLSHXCR2_AWRWAIT_SHIFT 16
+#define FSPI_FLSHXCR2_AWRWAIT_MASK (0xFF << FSPI_FLSHXCR2_AWRWAIT_SHIFT)
+#define FSPI_FLSHXCR2_AWRSEQN_SHIFT 13
+#define FSPI_FLSHXCR2_AWRSEQN_MASK (0x7 << FSPI_FLSHXCR2_AWRSEQN_SHIFT)
+#define FSPI_FLSHXCR2_AWRSEQI_SHIFT 8
+#define FSPI_FLSHXCR2_AWRSEQI_MASK (0xF << FSPI_FLSHXCR2_AWRSEQI_SHIFT)
+#define FSPI_FLSHXCR2_ARDSEQN_SHIFT 5
+#define FSPI_FLSHXCR2_ARDSEQN_MASK (0x7 << FSPI_FLSHXCR2_ARDSEQN_SHIFT)
+#define FSPI_FLSHXCR2_ARDSEQI_SHIFT 0
+#define FSPI_FLSHXCR2_ARDSEQI_MASK (0xF << FSPI_FLSHXCR2_ARDSEQI_SHIFT)
+
+#define FSPI_IPCR0 0xA0
+
+#define FSPI_IPCR1 0xA4
+#define FSPI_IPCR1_IPAREN_SHIFT 31
+#define FSPI_IPCR1_IPAREN_MASK (1 << FSPI_IPCR1_IPAREN_SHIFT)
+#define FSPI_IPCR1_SEQNUM_SHIFT 24
+#define FSPI_IPCR1_SEQNUM_MASK (0xF << FSPI_IPCR1_SEQNUM_SHIFT)
+#define FSPI_IPCR1_SEQID_SHIFT 16
+#define FSPI_IPCR1_SEQID_MASK (0xF << FSPI_IPCR1_SEQID_SHIFT)
+#define FSPI_IPCR1_IDATSZ_SHIFT 0
+#define FSPI_IPCR1_IDATSZ_MASK (0xFFFF << FSPI_IPCR1_IDATSZ_SHIFT)
+
+#define FSPI_IPCMD 0xB0
+#define FSPI_IPCMD_TRG_SHIFT 0
+#define FSPI_IPCMD_TRG_MASK (1 << FSPI_IPCMD_TRG_SHIFT)
+
+#define FSPI_DLPR 0xB4
+
+#define FSPI_IPRXFCR 0xB8
+#define FSPI_IPRXFCR_CLR_SHIFT 0
+#define FSPI_IPRXFCR_CLR_MASK (1 << FSPI_IPRXFCR_CLR_SHIFT)
+#define FSPI_IPRXFCR_DMA_EN_SHIFT 1
+#define FSPI_IPRXFCR_DMA_EN_MASK (1 << FSPI_IPRXFCR_DMA_EN_SHIFT)
+#define FSPI_IPRXFCR_WMRK_SHIFT 2
+#define FSPI_IPRXFCR_WMRK_MASK (0x1F << FSPI_IPRXFCR_WMRK_SHIFT)
+
+#define FSPI_IPTXFCR 0xBC
+#define FSPI_IPTXFCR_CLR_SHIFT 0
+#define FSPI_IPTXFCR_CLR_MASK (1 << FSPI_IPTXFCR_CLR_SHIFT)
+#define FSPI_IPTXFCR_DMA_EN_SHIFT 1
+#define FSPI_IPTXFCR_DMA_EN_MASK (1 << FSPI_IPTXFCR_DMA_EN_SHIFT)
+#define FSPI_IPTXFCR_WMRK_SHIFT 2
+#define FSPI_IPTXFCR_WMRK_MASK (0x1F << FSPI_IPTXFCR_WMRK_SHIFT)
+
+#define FSPI_DLLACR 0xC0
+#define FSPI_DLLACR_OVRDEN_SHIFT 8
+#define FSPI_DLLACR_OVRDEN_MASK (1 << FSPI_DLLACR_OVRDEN_SHIFT)
+
+#define FSPI_DLLBCR 0xC4
+#define FSPI_DLLBCR_OVRDEN_SHIFT 8
+#define FSPI_DLLBCR_OVRDEN_MASK (1 << FSPI_DLLBCR_OVRDEN_SHIFT)
+
+#define FSPI_STS0 0xE0
+#define FSPI_STS0_DLPHA_SHIFT 9
+#define FSPI_STS0_DLPHA_MASK (0x1F << FSPI_STS0_DLPHA_SHIFT)
+#define FSPI_STS0_DLPHB_SHIFT 4
+#define FSPI_STS0_DLPHB_MASK (0x1F << FSPI_STS0_DLPHB_SHIFT)
+#define FSPI_STS0_CMD_SRC_SHIFT 2
+#define FSPI_STS0_CMD_SRC_MASK (3 << FSPI_STS0_CMD_SRC_SHIFT)
+#define FSPI_STS0_ARB_IDLE_SHIFT 1
+#define FSPI_STS0_ARB_IDLE_MASK (1 << FSPI_STS0_ARB_IDLE_SHIFT)
+#define FSPI_STS0_SEQ_IDLE_SHIFT 0
+#define FSPI_STS0_SEQ_IDLE_MASK (1 << FSPI_STS0_SEQ_IDLE_SHIFT)
+
+#define FSPI_STS1 0xE4
+#define FSPI_STS1_IP_ERRCD_SHIFT 24
+#define FSPI_STS1_IP_ERRCD_MASK (0xF << FSPI_STS1_IP_ERRCD_SHIFT)
+#define FSPI_STS1_IP_ERRID_SHIFT 16
+#define FSPI_STS1_IP_ERRID_MASK (0xF << FSPI_STS1_IP_ERRID_SHIFT)
+#define FSPI_STS1_AHB_ERRCD_SHIFT 8
+#define FSPI_STS1_AHB_ERRCD_MASK (0xF << FSPI_STS1_AHB_ERRCD_SHIFT)
+#define FSPI_STS1_AHB_ERRID_SHIFT 0
+#define FSPI_STS1_AHB_ERRID_MASK (0xF << FSPI_STS1_AHB_ERRID_SHIFT)
+
+#define FSPI_AHBSPNST 0xEC
+#define FSPI_AHBSPNST_DATLFT_SHIFT 16
+#define FSPI_AHBSPNST_DATLFT_MASK \
+ (0xFFFF << FSPI_AHBSPNST_DATLFT_SHIFT)
+#define FSPI_AHBSPNST_BUFID_SHIFT 1
+#define FSPI_AHBSPNST_BUFID_MASK (7 << FSPI_AHBSPNST_BUFID_SHIFT)
+#define FSPI_AHBSPNST_ACTIVE_SHIFT 0
+#define FSPI_AHBSPNST_ACTIVE_MASK (1 << FSPI_AHBSPNST_ACTIVE_SHIFT)
+
+#define FSPI_IPRXFSTS 0xF0
+#define FSPI_IPRXFSTS_RDCNTR_SHIFT 16
+#define FSPI_IPRXFSTS_RDCNTR_MASK \
+ (0xFFFF << FSPI_IPRXFSTS_RDCNTR_SHIFT)
+#define FSPI_IPRXFSTS_FILL_SHIFT 0
+#define FSPI_IPRXFSTS_FILL_MASK (0xFF << FSPI_IPRXFSTS_FILL_SHIFT)
+
+#define FSPI_IPTXFSTS 0xF4
+#define FSPI_IPTXFSTS_WRCNTR_SHIFT 16
+#define FSPI_IPTXFSTS_WRCNTR_MASK \
+ (0xFFFF << FSPI_IPTXFSTS_WRCNTR_SHIFT)
+#define FSPI_IPTXFSTS_FILL_SHIFT 0
+#define FSPI_IPTXFSTS_FILL_MASK (0xFF << FSPI_IPTXFSTS_FILL_SHIFT)
+
+#define FSPI_RFDR 0x100
+#define FSPI_TFDR 0x180
+
+#define FSPI_LUT_BASE 0x200
+#define FSPI_LUT_OFFSET (SEQID_LUT * 4 * 4)
+#define FSPI_LUT_REG(idx) \
+ (FSPI_LUT_BASE + FSPI_LUT_OFFSET + (idx) * 4)
+
+/* register map end */
+
+/* Instruction set for the LUT register. */
+#define LUT_STOP 0x00
+#define LUT_CMD 0x01
+#define LUT_ADDR 0x02
+#define LUT_CADDR_SDR 0x03
+#define LUT_MODE 0x04
+#define LUT_MODE2 0x05
+#define LUT_MODE4 0x06
+#define LUT_MODE8 0x07
+#define LUT_NXP_WRITE 0x08
+#define LUT_NXP_READ 0x09
+#define LUT_LEARN_SDR 0x0A
+#define LUT_DATSZ_SDR 0x0B
+#define LUT_DUMMY 0x0C
+#define LUT_DUMMY_RWDS_SDR 0x0D
+#define LUT_JMP_ON_CS 0x1F
+#define LUT_CMD_DDR 0x21
+#define LUT_ADDR_DDR 0x22
+#define LUT_CADDR_DDR 0x23
+#define LUT_MODE_DDR 0x24
+#define LUT_MODE2_DDR 0x25
+#define LUT_MODE4_DDR 0x26
+#define LUT_MODE8_DDR 0x27
+#define LUT_WRITE_DDR 0x28
+#define LUT_READ_DDR 0x29
+#define LUT_LEARN_DDR 0x2A
+#define LUT_DATSZ_DDR 0x2B
+#define LUT_DUMMY_DDR 0x2C
+#define LUT_DUMMY_RWDS_DDR 0x2D
+
+/*
+ * Calculate number of required PAD bits for LUT register.
+ *
+ * The pad stands for the number of IO lines [0:7].
+ * For example, the octal read needs eight IO lines,
+ * so you should use LUT_PAD(8). This macro
+ * returns 3 i.e. use eight (2^3) IP lines for read.
+ */
+#define LUT_PAD(x) (fls(x) - 1)
+
+/*
+ * Macro for constructing the LUT entries with the following
+ * register layout:
+ *
+ * ---------------------------------------------------
+ * | INSTR1 | PAD1 | OPRND1 | INSTR0 | PAD0 | OPRND0 |
+ * ---------------------------------------------------
+ */
+#define PAD_SHIFT 8
+#define INSTR_SHIFT 10
+#define OPRND_SHIFT 16
+
+/* Macros for constructing the LUT register. */
+#define LUT_DEF(idx, ins, pad, opr) \
+ ((((ins) << INSTR_SHIFT) | ((pad) << PAD_SHIFT) | \
+ (opr)) << (((idx) % 2) * OPRND_SHIFT))
+
+/* Oprands for the LUT register. */
+#define ADDR8BIT 0x08
+#define ADDR16BIT 0x10
+#define ADDR24BIT 0x18
+#define ADDR32BIT 0x20
+
+#define L_ENDIAN 0
+#define B_ENDIAN 1
+
+enum nxp_fspi_devtype {
+ NXP_FSPI_LX2160A,
+};
+
+struct nxp_fspi_devtype_data {
+ enum nxp_fspi_devtype devtype;
+ unsigned int rxfifo;
+ unsigned int txfifo;
+ unsigned int ahb_buf_size;
+ unsigned int quirks;
+ bool endianness;
+};
+
+static const struct nxp_fspi_devtype_data lx2160a_data = {
+ .devtype = NXP_FSPI_LX2160A,
+ .rxfifo = SZ_512, /* (64 * 64 bits) */
+ .txfifo = SZ_1K, /* (128 * 64 bits) */
+ .ahb_buf_size = SZ_2K, /* (256 * 64 bits) */
+ .quirks = 0,
+ .endianness = L_ENDIAN,
+};
+
+#define NXP_FSPI_MAX_CHIPSELECT 4
+
+struct nxp_fspi {
+ void __iomem *iobase;
+ void __iomem *ahb_addr;
+ u32 memmap_phy;
+ u32 memmap_phy_size;
+ struct clk *clk, *clk_en;
+ struct device *dev;
+ struct completion c;
+ const struct nxp_fspi_devtype_data *devtype_data;
+ struct mutex lock;
+ struct pm_qos_request pm_qos_req;
+ int selected;
+ void (*write)(u32 val, void __iomem *addr);
+ u32 (*read)(void __iomem *addr);
+};
+
+static void fspi_writel_be(u32 val, void __iomem *addr)
+{
+ iowrite32be(val, addr);
+}
+
+static void fspi_writel(u32 val, void __iomem *addr)
+{
+ iowrite32(val, addr);
+}
+
+static u32 fspi_readl_be(void __iomem *addr)
+{
+ return ioread32be(addr);
+}
+
+static u32 fspi_readl(void __iomem *addr)
+{
+ return ioread32(addr);
+}
+
+static irqreturn_t nxp_fspi_irq_handler(int irq, void *dev_id)
+{
+ struct nxp_fspi *f = dev_id;
+ u32 reg;
+
+ /* clear interrupt */
+ reg = f->read(f->iobase + FSPI_INTR);
+ f->write(FSPI_INTR_IPCMDDONE_MASK, f->iobase + FSPI_INTR);
+
+ if (reg & FSPI_INTR_IPCMDDONE_MASK)
+ complete(&f->c);
+
+ return IRQ_HANDLED;
+}
+
+static int nxp_fspi_check_buswidth(struct nxp_fspi *f, u8 width)
+{
+ switch (width) {
+ case 1:
+ case 2:
+ case 4:
+ case 8:
+ return 0;
+ }
+
+ return -ENOTSUPP;
+}
+
+static bool nxp_fspi_supports_op(struct spi_mem *mem,
+ const struct spi_mem_op *op)
+{
+ struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->master);
+ int ret;
+
+ ret = nxp_fspi_check_buswidth(f, op->cmd.buswidth);
+
+ if (op->addr.nbytes)
+ ret |= nxp_fspi_check_buswidth(f, op->addr.buswidth);
+
+ if (op->dummy.nbytes)
+ ret |= nxp_fspi_check_buswidth(f, op->dummy.buswidth);
+
+ if (op->data.nbytes)
+ ret |= nxp_fspi_check_buswidth(f, op->data.buswidth);
+
+ if (ret)
+ return false;
+
+ /*
+ * The number of instructions needed for the op, needs
+ * to fit into a single LUT entry.
+ */
+ if (op->addr.nbytes +
+ (op->dummy.nbytes ? 1:0) +
+ (op->data.nbytes ? 1:0) > 6)
+ return false;
+
+ /* Max 64 dummy clock cycles supported */
+ if (op->dummy.buswidth &&
+ (op->dummy.nbytes * 8 / op->dummy.buswidth > 64))
+ return false;
+
+ /* Max data length, check controller limits and alignment */
+ if (op->data.dir == SPI_MEM_DATA_IN &&
+ (op->data.nbytes > f->devtype_data->ahb_buf_size ||
+ (op->data.nbytes > f->devtype_data->rxfifo - 4 &&
+ !IS_ALIGNED(op->data.nbytes, 8))))
+ return false;
+
+ if (op->data.dir == SPI_MEM_DATA_OUT &&
+ op->data.nbytes > f->devtype_data->txfifo)
+ return false;
+
+ return true;
+}
+
+/*
+ * If the slave device content being changed by Write/Erase, need to
+ * invalidate the AHB buffer. This can be achieved by doing the reset
+ * of controller after setting MCR0[SWRESET] bit.
+ */
+static inline void nxp_fspi_invalid(struct nxp_fspi *f)
+{
+ u32 reg;
+
+ reg = f->read(f->iobase + FSPI_MCR0);
+ f->write(reg | FSPI_MCR0_SWRST_MASK, f->iobase + FSPI_MCR0);
+
+ while (f->read(f->iobase + FSPI_MCR0) & FSPI_MCR0_SWRST_MASK)
+ ;
+}
+
+static void nxp_fspi_prepare_lut(struct nxp_fspi *f,
+ const struct spi_mem_op *op)
+{
+ void __iomem *base = f->iobase;
+ u32 lutval[4] = {};
+ int lutidx = 1, i;
+
+ /* cmd */
+ lutval[0] |= LUT_DEF(0, LUT_CMD, LUT_PAD(op->cmd.buswidth),
+ op->cmd.opcode);
+
+ /* addr bus width */
+ if (op->addr.nbytes) {
+ u32 addrlen = 0;
+
+ if (op->addr.nbytes == 3)
+ addrlen = ADDR24BIT;
+ else if (op->addr.nbytes == 4)
+ addrlen = ADDR32BIT;
+
+ lutval[lutidx / 2] |= LUT_DEF(lutidx, LUT_ADDR,
+ LUT_PAD(op->addr.buswidth),
+ addrlen);
+ lutidx++;
+ }
+
+ /* dummy bytes, if needed */
+ if (op->dummy.nbytes) {
+ lutval[lutidx / 2] |= LUT_DEF(lutidx, LUT_DUMMY,
+ LUT_PAD(op->dummy.buswidth),
+ op->dummy.nbytes * 8 /
+ op->dummy.buswidth);
+ lutidx++;
+ }
+
+ /* read/write data bytes */
+ if (op->data.nbytes) {
+ lutval[lutidx / 2] |= LUT_DEF(lutidx,
+ op->data.dir == SPI_MEM_DATA_IN ?
+ LUT_NXP_READ : LUT_NXP_WRITE,
+ LUT_PAD(op->data.buswidth),
+ 0);
+ lutidx++;
+ }
+
+ /* stop condition. */
+ lutval[lutidx / 2] |= LUT_DEF(lutidx, LUT_STOP, 0, 0);
+
+ /* unlock LUT */
+ f->write(FSPI_LUTKEY_VALUE, f->iobase + FSPI_LUTKEY);
+ f->write(FSPI_LCKER_UNLOCK, f->iobase + FSPI_LCKCR);
+
+ /* fill LUT */
+ for (i = 0; i < ARRAY_SIZE(lutval); i++)
+ f->write(lutval[i], base + FSPI_LUT_REG(i));
+
+ dev_dbg(f->dev, "CMD[%x] lutval[0:%x \t 1:%x \t 2:%x \t 3:%x]\n",
+ op->cmd.opcode, lutval[0], lutval[1], lutval[2], lutval[3]);
+
+ /* lock LUT */
+ f->write(FSPI_LUTKEY_VALUE, f->iobase + FSPI_LUTKEY);
+ f->write(FSPI_LCKER_LOCK, f->iobase + FSPI_LCKCR);
+}
+
+static int nxp_fspi_clk_prep_enable(struct nxp_fspi *f)
+{
+ int ret;
+
+ ret = clk_prepare_enable(f->clk_en);
+ if (ret)
+ return ret;
+
+ ret = clk_prepare_enable(f->clk);
+ if (ret) {
+ clk_disable_unprepare(f->clk_en);
+ return ret;
+ }
+
+ return 0;
+}
+
+static void nxp_fspi_clk_disable_unprep(struct nxp_fspi *f)
+{
+ clk_disable_unprepare(f->clk);
+ clk_disable_unprepare(f->clk_en);
+}
+
+/*
+ * In FlexSPI controller, flash access is based on value of FSPI_FLSHXXCR0
+ * register and start base address of the slave device.
+ *
+ * (Higher address)
+ * -------- <-- FLSHB2CR0
+ * | B2 |
+ * | |
+ * B2 start address --> -------- <-- FLSHB1CR0
+ * | B1 |
+ * | |
+ * B1 start address --> -------- <-- FLSHA2CR0
+ * | A2 |
+ * | |
+ * A2 start address --> -------- <-- FLSHA1CR0
+ * | A1 |
+ * | |
+ * A1 start address --> -------- (Lower address)
+ *
+ *
+ * Start base address defines the starting address range for given CS and
+ * FSPI_FLSHXXCR0 defines the size of the slave device connected at given CS.
+ *
+ * But, different targets are having different combinations of number of CS,
+ * some targets only have single CS or two CS covering controller's full
+ * memory mapped space area.
+ * Thus, implementation is being done as independent of the size and number
+ * of the connected slave device.
+ * Assign controller memory mapped space size as the size to the connected
+ * slave device.
+ * Mark FLSHxxCR0 as zero initially and then assign value only to the selected
+ * chip-select Flash configuration register.
+ *
+ * For e.g. to access CS2 (B1), FLSHB1CR0 register would be equal to the
+ * memory mapped size of the controller.
+ * Value for rest of the CS FLSHxxCR0 register would be zero.
+ *
+ */
+static void nxp_fspi_select_mem(struct nxp_fspi *f, struct spi_device *spi)
+{
+ unsigned long rate = spi->max_speed_hz;
+ int ret;
+ uint64_t size_kb;
+
+ /*
+ * Return, if previously selected slave device is same as current
+ * requested slave device.
+ */
+ if (f->selected == spi->chip_select)
+ return;
+
+ /* Reset FLSHxxCR0 registers */
+ f->write(0, f->iobase + FSPI_FLSHA1CR0);
+ f->write(0, f->iobase + FSPI_FLSHA2CR0);
+ f->write(0, f->iobase + FSPI_FLSHB1CR0);
+ f->write(0, f->iobase + FSPI_FLSHB2CR0);
+
+ /* Assign controller memory mapped space, as size of the flash. */
+ size_kb = f->memmap_phy_size >> FSPI_FLSHXCR0_SZ_SHIFT;
+
+ switch (spi->chip_select) {
+ case 0:
+ f->write(size_kb, f->iobase + FSPI_FLSHA1CR0);
+ break;
+ case 1:
+ f->write(size_kb, f->iobase + FSPI_FLSHA2CR0);
+ break;
+ case 2:
+ f->write(size_kb, f->iobase + FSPI_FLSHB1CR0);
+ break;
+ case 3:
+ f->write(size_kb, f->iobase + FSPI_FLSHB2CR0);
+ break;
+ default:
+ dev_err(f->dev, "In-correct CS provided\n");
+ return;
+ }
+
+ dev_dbg(f->dev, "Slave device [CS:%x] selected\n", spi->chip_select);
+
+ nxp_fspi_clk_disable_unprep(f);
+
+ ret = clk_set_rate(f->clk, rate);
+ if (ret)
+ return;
+
+ ret = nxp_fspi_clk_prep_enable(f);
+ if (ret)
+ return;
+ f->selected = spi->chip_select;
+}
+
+static void nxp_fspi_read_ahb(struct nxp_fspi *f, const struct spi_mem_op *op)
+{
+ u32 len = op->data.nbytes;
+
+ /* Read out the data directly from the AHB buffer. */
+ memcpy_fromio(op->data.buf.in, (f->ahb_addr + op->addr.val), len);
+}
+
+static void nxp_fspi_fill_txfifo(struct nxp_fspi *f,
+ const struct spi_mem_op *op)
+{
+ void __iomem *base = f->iobase;
+ int i, j;
+ int size, tmp_size, wm_size;
+ u32 data = 0;
+ u32 *txbuf = (u32 *) op->data.buf.out;
+
+ /* clear the TX FIFO. */
+ f->write(FSPI_IPTXFCR_CLR_MASK, base + FSPI_IPTXFCR);
+
+ /* Default value of water mark level is 8 bytes. */
+ wm_size = 8;
+ size = op->data.nbytes / wm_size;
+ for (i = 0; i < size; i++) {
+ /* Wait for TXFIFO empty */
+ while (!(f->read(base + FSPI_INTR) & FSPI_INTR_IPTXWE_MASK))
+ ;
+ j = 0;
+ tmp_size = wm_size;
+ while (tmp_size > 0) {
+ data = 0;
+ memcpy(&data, txbuf, 4);
+ f->write(data, base + FSPI_TFDR + j * 4);
+ tmp_size -= 4;
+ j++;
+ txbuf += 1;
+ }
+ f->write(FSPI_INTR_IPTXWE_MASK, base + FSPI_INTR);
+ }
+
+ size = op->data.nbytes % wm_size;
+ if (size) {
+ /* Wait for TXFIFO empty */
+ while (!(f->read(base + FSPI_INTR) & FSPI_INTR_IPTXWE_MASK))
+ ;
+ j = 0;
+ tmp_size = 0;
+ while (size > 0) {
+ data = 0;
+ tmp_size = (size < 4) ? size : 4;
+ memcpy(&data, txbuf, tmp_size);
+ f->write(data, base + FSPI_TFDR + j * 4);
+ size -= tmp_size;
+ j++;
+ txbuf += 1;
+ }
+ f->write(FSPI_INTR_IPTXWE_MASK, base + FSPI_INTR);
+ }
+}
+
+static void nxp_fspi_read_rxfifo(struct nxp_fspi *f,
+ const struct spi_mem_op *op)
+{
+ void __iomem *base = f->iobase;
+ int i, j;
+ int size, tmp_size, wm_size;
+ u32 tmp = 0;
+ u8 *buf = op->data.buf.in;
+ u32 len = op->data.nbytes;
+
+ /* Default value of water mark level is 8 bytes. */
+ wm_size = 8;
+
+ while (len > 0) {
+ size = len / wm_size;
+
+ for (i = 0; i < size; i++) {
+ /* Wait for RXFIFO available */
+ while (!(f->read(base + FSPI_INTR)
+ & FSPI_INTR_IPRXWA_MASK))
+ ;
+ j = 0;
+ tmp_size = wm_size;
+ while (tmp_size > 0) {
+ tmp = 0;
+ tmp = f->read(base + FSPI_RFDR + j * 4);
+ memcpy(buf, &tmp, 4);
+ tmp_size -= 4;
+ j++;
+ buf += 4;
+ }
+ /* move the FIFO pointer */
+ f->write(FSPI_INTR_IPRXWA_MASK, base + FSPI_INTR);
+ len -= wm_size;
+ }
+
+ size = len % wm_size;
+
+ j = 0;
+ if (size) {
+ /* Wait for RXFIFO available */
+ while (!(f->read(base + FSPI_INTR)
+ & FSPI_INTR_IPRXWA_MASK))
+ ;
+ while (len > 0) {
+ tmp = 0;
+ size = (len < 4) ? len : 4;
+ tmp = f->read(base + FSPI_RFDR + j * 4);
+ memcpy(buf, &tmp, size);
+ len -= size;
+ j++;
+ buf += size;
+ }
+ }
+
+ /* invalid the RXFIFO */
+ f->write(FSPI_IPRXFCR_CLR_MASK, base + FSPI_IPRXFCR);
+ /* move the FIFO pointer */
+ f->write(FSPI_INTR_IPRXWA_MASK, base + FSPI_INTR);
+ }
+}
+
+static int nxp_fspi_do_op(struct nxp_fspi *f, const struct spi_mem_op *op)
+{
+ void __iomem *base = f->iobase;
+ int seqnum = 0;
+ int err = 0;
+ u32 reg;
+
+ reg = f->read(base + FSPI_IPRXFCR);
+ /* invalid RXFIFO first */
+ reg &= ~FSPI_IPRXFCR_DMA_EN_MASK;
+ reg = reg | FSPI_IPRXFCR_CLR_MASK;
+ f->write(reg, base + FSPI_IPRXFCR);
+
+ init_completion(&f->c);
+
+ f->write(op->addr.val, base + FSPI_IPCR0);
+ /*
+ * Always start the sequence at the same index since we update
+ * the LUT at each exec_op() call. And also specify the DATA
+ * length, since it's has not been specified in the LUT.
+ */
+ f->write(op->data.nbytes |
+ (SEQID_LUT << FSPI_IPCR1_SEQID_SHIFT) |
+ (seqnum << FSPI_IPCR1_SEQNUM_SHIFT),
+ base + FSPI_IPCR1);
+
+ /* Trigger the LUT now. */
+ f->write(FSPI_IPCMD_TRG_MASK, base + FSPI_IPCMD);
+
+ /* Wait for the interrupt. */
+ if (!wait_for_completion_timeout(&f->c, msecs_to_jiffies(1000)))
+ err = -ETIMEDOUT;
+
+ /* Invoke IP data read, if request is of data read. */
+ if (!err && op->data.nbytes && op->data.dir == SPI_MEM_DATA_IN)
+ nxp_fspi_read_rxfifo(f, op);
+
+ return err;
+}
+
+static int nxp_fspi_exec_op(struct spi_mem *mem, const struct spi_mem_op *op)
+{
+ struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->master);
+ void __iomem *base = f->iobase;
+ int err = 0;
+ unsigned int timeout = 1000;
+
+ mutex_lock(&f->lock);
+
+ /* wait for the controller being ready */
+ do {
+ u32 status;
+
+ status = f->read(base + FSPI_STS0);
+ if ((status & FSPI_STS0_ARB_IDLE_MASK) &&
+ (status & FSPI_STS0_SEQ_IDLE_MASK))
+ break;
+ udelay(1);
+ dev_dbg(f->dev, "The controller is busy, 0x%x\n", status);
+ } while (--timeout);
+
+ nxp_fspi_select_mem(f, mem->spi);
+
+ nxp_fspi_prepare_lut(f, op);
+ /*
+ * If we have large chunks of data, we read them through the AHB bus
+ * by accessing the mapped memory. In all other cases we use
+ * IP commands to access the flash.
+ */
+ if (op->data.nbytes > (f->devtype_data->rxfifo - 4) &&
+ op->data.dir == SPI_MEM_DATA_IN) {
+ nxp_fspi_read_ahb(f, op);
+ } else {
+ if (op->data.nbytes && op->data.dir == SPI_MEM_DATA_OUT)
+ nxp_fspi_fill_txfifo(f, op);
+
+ err = nxp_fspi_do_op(f, op);
+
+ /* Invalidate the data in the AHB buffer. */
+ if (op->data.nbytes && op->data.dir == SPI_MEM_DATA_OUT)
+ nxp_fspi_invalid(f);
+ }
+
+ mutex_unlock(&f->lock);
+
+ return err;
+}
+
+static int nxp_fspi_adjust_op_size(struct spi_mem *mem, struct spi_mem_op *op)
+{
+ struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->master);
+
+ if (op->data.dir == SPI_MEM_DATA_OUT) {
+ if (op->data.nbytes > f->devtype_data->txfifo)
+ op->data.nbytes = f->devtype_data->txfifo;
+ } else {
+ if (op->data.nbytes > f->devtype_data->ahb_buf_size)
+ op->data.nbytes = f->devtype_data->ahb_buf_size;
+ else if (op->data.nbytes > (f->devtype_data->rxfifo - 4))
+ op->data.nbytes = ALIGN_DOWN(op->data.nbytes, 8);
+ }
+
+ return 0;
+}
+
+static int nxp_fspi_default_setup(struct nxp_fspi *f)
+{
+ void __iomem *base = f->iobase;
+ int ret, i;
+ u32 reg;
+
+ /* disable and unprepare clock to avoid glitch pass to controller */
+ nxp_fspi_clk_disable_unprep(f);
+
+ /* the default frequency, we will change it later if necessary. */
+ ret = clk_set_rate(f->clk, 20000000);
+ if (ret)
+ return ret;
+
+ ret = nxp_fspi_clk_prep_enable(f);
+ if (ret)
+ return ret;
+
+ /* Reset the module */
+ f->write(FSPI_MCR0_SWRST_MASK, base + FSPI_MCR0);
+ do {
+ udelay(1);
+ } while (f->read(f->iobase + FSPI_MCR0) & FSPI_MCR0_SWRST_MASK);
+
+ /* Disable the module */
+ f->write(FSPI_MCR0_MDIS_MASK, base + FSPI_MCR0);
+
+ /* Reset the DLL register to default value */
+ f->write(FSPI_DLLACR_OVRDEN_MASK, base + FSPI_DLLACR);
+ f->write(FSPI_DLLBCR_OVRDEN_MASK, base + FSPI_DLLBCR);
+
+ /* enable module */
+ f->write(FSPI_MCR0_AHB_TIMEOUT_MASK | FSPI_MCR0_IP_TIMEOUT_MASK,
+ base + FSPI_MCR0);
+
+ /*
+ * Disable same device enable bit and configure all slave devices
+ * independently.
+ */
+ reg = f->read(f->iobase + FSPI_MCR2);
+ reg = reg & ~(1 << (FSPI_MCR2_SAMEDEVICEEN_SHIFT));
+ f->write(reg, base + FSPI_MCR2);
+
+ /* AHB configuration for access buffer 0~7. */
+ for (i = 0; i < 7; i++)
+ f->write(0, base + FSPI_AHBRX_BUF0CR0 + 4 * i);
+
+ /*
+ * Set ADATSZ with the maximum AHB buffer size to improve the read
+ * performance.
+ */
+ f->write((f->devtype_data->ahb_buf_size / 8 |
+ FSPI_AHBRXBUF0CR7_PREF_MASK), base + FSPI_AHBRX_BUF7CR0);
+
+ /* prefetch and no start address alignment limitation */
+ f->write(FSPI_AHBCR_PREF_EN_MASK | FSPI_AHBCR_RDADDROPT_MASK,
+ base + FSPI_AHBCR);
+
+ /* AHB Read - Set lut sequence ID for all CS. */
+ f->write(SEQID_LUT, base + FSPI_FLSHA1CR2);
+ f->write(SEQID_LUT, base + FSPI_FLSHA2CR2);
+ f->write(SEQID_LUT, base + FSPI_FLSHB1CR2);
+ f->write(SEQID_LUT, base + FSPI_FLSHB2CR2);
+
+ f->selected = -1;
+
+ /* enable the interrupt */
+ f->write(FSPI_INTEN_IPCMDDONE_MASK, base + FSPI_INTEN);
+
+ return 0;
+}
+
+static const struct spi_controller_mem_ops nxp_fspi_mem_ops = {
+ .adjust_op_size = nxp_fspi_adjust_op_size,
+ .supports_op = nxp_fspi_supports_op,
+ .exec_op = nxp_fspi_exec_op,
+};
+
+static int nxp_fspi_probe(struct platform_device *pdev)
+{
+ struct spi_controller *ctlr;
+ struct device *dev = &pdev->dev;
+ struct device_node *np = dev->of_node;
+ struct resource *res;
+ struct nxp_fspi *f;
+ int ret;
+
+ ctlr = spi_alloc_master(&pdev->dev, sizeof(*f));
+ if (!ctlr)
+ return -ENOMEM;
+
+ ctlr->mode_bits = SPI_RX_DUAL | SPI_RX_QUAD |
+ SPI_TX_DUAL | SPI_TX_QUAD;
+
+ f = spi_controller_get_devdata(ctlr);
+ f->dev = dev;
+ f->devtype_data = of_device_get_match_data(dev);
+ if (!f->devtype_data) {
+ ret = -ENODEV;
+ goto err_put_ctrl;
+ }
+
+ platform_set_drvdata(pdev, f);
+
+ /* find the resources - configuration register address space */
+ res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "fspi_base");
+ f->iobase = devm_ioremap_resource(dev, res);
+ if (IS_ERR(f->iobase)) {
+ ret = PTR_ERR(f->iobase);
+ goto err_put_ctrl;
+ }
+
+ /* find the resources - controller memory mapped space */
+ res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "fspi_mmap");
+ f->ahb_addr = devm_ioremap_resource(dev, res);
+ if (IS_ERR(f->ahb_addr)) {
+ ret = PTR_ERR(f->ahb_addr);
+ goto err_put_ctrl;
+ }
+
+ /* assign memory mapped starting address and mapped size. */
+ f->memmap_phy = res->start;
+ f->memmap_phy_size = resource_size(res);
+
+ /* find the clocks */
+ f->clk_en = devm_clk_get(dev, "fspi_en");
+ if (IS_ERR(f->clk_en)) {
+ ret = PTR_ERR(f->clk_en);
+ goto err_put_ctrl;
+ }
+
+ f->clk = devm_clk_get(dev, "fspi");
+ if (IS_ERR(f->clk)) {
+ ret = PTR_ERR(f->clk);
+ goto err_put_ctrl;
+ }
+
+ /*
+ * R/W functions for big- or little-endian registers:
+ * The FSPI controller's endianness is independent of
+ * the CPU core's endianness. So far, although the CPU
+ * core is little-endian the FSPI controller can use
+ * big-endian or little-endian.
+ */
+ if (f->devtype_data->endianness == L_ENDIAN) {
+ f->write = fspi_writel;
+ f->read = fspi_readl;
+ } else {
+ f->write = fspi_writel_be;
+ f->read = fspi_readl_be;
+ }
+
+ ret = nxp_fspi_clk_prep_enable(f);
+ if (ret) {
+ dev_err(dev, "can not enable the clock\n");
+ goto err_put_ctrl;
+ }
+
+ /* find the irq */
+ ret = platform_get_irq(pdev, 0);
+ if (ret < 0) {
+ dev_err(dev, "failed to get the irq: %d\n", ret);
+ goto err_disable_clk;
+ }
+
+ ret = devm_request_irq(dev, ret,
+ nxp_fspi_irq_handler, 0, pdev->name, f);
+ if (ret) {
+ dev_err(dev, "failed to request irq: %d\n", ret);
+ goto err_disable_clk;
+ }
+
+ mutex_init(&f->lock);
+
+ ctlr->bus_num = -1;
+ ctlr->num_chipselect = NXP_FSPI_MAX_CHIPSELECT;
+ ctlr->mem_ops = &nxp_fspi_mem_ops;
+
+ nxp_fspi_default_setup(f);
+
+ ctlr->dev.of_node = np;
+
+ ret = spi_register_controller(ctlr);
+ if (ret)
+ goto err_destroy_mutex;
+
+ return 0;
+
+err_destroy_mutex:
+ mutex_destroy(&f->lock);
+
+err_disable_clk:
+ nxp_fspi_clk_disable_unprep(f);
+
+err_put_ctrl:
+ spi_controller_put(ctlr);
+
+ dev_err(dev, "NXP FSPI probe failed\n");
+ return ret;
+}
+
+static int nxp_fspi_remove(struct platform_device *pdev)
+{
+ struct nxp_fspi *f = platform_get_drvdata(pdev);
+
+ /* disable the hardware */
+ f->write(FSPI_MCR0_MDIS_MASK, f->iobase + FSPI_MCR0);
+
+ nxp_fspi_clk_disable_unprep(f);
+
+ mutex_destroy(&f->lock);
+
+ return 0;
+}
+
+static int nxp_fspi_suspend(struct device *dev)
+{
+ return 0;
+}
+
+static int nxp_fspi_resume(struct device *dev)
+{
+ struct nxp_fspi *f = dev_get_drvdata(dev);
+
+ nxp_fspi_default_setup(f);
+
+ return 0;
+}
+
+static const struct of_device_id nxp_fspi_dt_ids[] = {
+ { .compatible = "nxp,lx2160a-fspi", .data = (void *)&lx2160a_data, },
+ { /* sentinel */ }
+};
+MODULE_DEVICE_TABLE(of, nxp_fspi_dt_ids);
+
+static const struct dev_pm_ops nxp_fspi_pm_ops = {
+ .suspend = nxp_fspi_suspend,
+ .resume = nxp_fspi_resume,
+};
+
+static struct platform_driver nxp_fspi_driver = {
+ .driver = {
+ .name = "nxp-fspi",
+ .of_match_table = nxp_fspi_dt_ids,
+ .pm = &nxp_fspi_pm_ops,
+ },
+ .probe = nxp_fspi_probe,
+ .remove = nxp_fspi_remove,
+};
+module_platform_driver(nxp_fspi_driver);
+
+MODULE_DESCRIPTION("NXP FSPI Controller Driver");
+MODULE_AUTHOR("NXP Semiconductor");
+MODULE_LICENSE("GPL v2");
--
2.7.4
Add fspi node property for LX2160A SoC for FlexSPI driver.
Property added for the FlexSPI controller and for the connected
slave device for the LX2160ARDB target.
This is having two SPI-NOR flash device, mt35xu512aba, connected
at CS0 and CS1.
Signed-off-by: Yogesh Gaur <[email protected]>
---
Changes for v3:
- None
Changes for v2:
- Incorporated Shawn review comments.
arch/arm64/boot/dts/freescale/fsl-lx2160a-rdb.dts | 22 ++++++++++++++++++++++
arch/arm64/boot/dts/freescale/fsl-lx2160a.dtsi | 12 ++++++++++++
2 files changed, 34 insertions(+)
diff --git a/arch/arm64/boot/dts/freescale/fsl-lx2160a-rdb.dts b/arch/arm64/boot/dts/freescale/fsl-lx2160a-rdb.dts
index 70fad20..901ca346 100644
--- a/arch/arm64/boot/dts/freescale/fsl-lx2160a-rdb.dts
+++ b/arch/arm64/boot/dts/freescale/fsl-lx2160a-rdb.dts
@@ -32,6 +32,28 @@
status = "okay";
};
+&fspi {
+ status = "okay";
+
+ mt35xu512aba0: flash@0 {
+ #address-cells = <1>;
+ #size-cells = <1>;
+ compatible = "spansion,m25p80";
+ m25p,fast-read;
+ spi-max-frequency = <20000000>;
+ reg = <0>;
+ };
+
+ mt35xu512aba1: flash@1 {
+ #address-cells = <1>;
+ #size-cells = <1>;
+ compatible = "spansion,m25p80";
+ m25p,fast-read;
+ spi-max-frequency = <20000000>;
+ reg = <1>;
+ };
+};
+
&i2c0 {
status = "okay";
pca9547@77 {
diff --git a/arch/arm64/boot/dts/freescale/fsl-lx2160a.dtsi b/arch/arm64/boot/dts/freescale/fsl-lx2160a.dtsi
index e35e494..bd2d606 100644
--- a/arch/arm64/boot/dts/freescale/fsl-lx2160a.dtsi
+++ b/arch/arm64/boot/dts/freescale/fsl-lx2160a.dtsi
@@ -568,5 +568,17 @@
timeout-sec = <30>;
};
+ fspi: flexspi@20c0000 {
+ compatible = "nxp,lx2160a-fspi";
+ #address-cells = <1>;
+ #size-cells = <0>;
+ reg = <0x0 0x20c0000 0x0 0x10000>,
+ <0x0 0x20000000 0x0 0x10000000>;
+ reg-names = "fspi_base", "fspi_mmap";
+ interrupts = <0 25 0x4>; /* Level high type */
+ clocks = <&clockgen 4 3>, <&clockgen 4 3>;
+ clock-names = "fspi_en", "fspi";
+ status = "disabled";
+ };
};
};
--
2.7.4
Enable driver support of NXP FlexSPI controller.
Signed-off-by: Yogesh Gaur <[email protected]>
---
Changes for v3:
- None
Changes for v2:
- None
arch/arm64/configs/defconfig | 1 +
1 file changed, 1 insertion(+)
diff --git a/arch/arm64/configs/defconfig b/arch/arm64/configs/defconfig
index 3cfa8ca..75ceddf 100644
--- a/arch/arm64/configs/defconfig
+++ b/arch/arm64/configs/defconfig
@@ -335,6 +335,7 @@ CONFIG_SPI_QUP=y
CONFIG_SPI_ROCKCHIP=y
CONFIG_SPI_S3C64XX=y
CONFIG_SPI_SPIDEV=m
+CONFIG_SPI_NXP_FLEXSPI=y
CONFIG_SPMI=y
CONFIG_PINCTRL_IPQ8074=y
CONFIG_PINCTRL_SINGLE=y
--
2.7.4
Add maintainers for the NXP FlexSPI driver
Signed-off-by: Yogesh Gaur <[email protected]>
---
Changes for v3:
- None
Changes for v2:
- None
MAINTAINERS | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index 9d5eeff..2696898 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -10228,6 +10228,12 @@ L: [email protected] (moderated for non-subscribers)
S: Supported
F: drivers/nfc/nxp-nci
+NXP FSPI DRIVER
+M: Yogesh Gaur <[email protected]>
+L: [email protected]
+S: Maintained
+F: drivers/spi/spi-nxp-fspi.c
+
OBJTOOL
M: Josh Poimboeuf <[email protected]>
M: Peter Zijlstra <[email protected]>
--
2.7.4
Add binding file for NXP FlexSPI controller
Signed-off-by: Yogesh Gaur <[email protected]>
---
Changes for v3:
- None
Changes for v2:
- Incorporated Rob review comments.
.../devicetree/bindings/spi/spi-nxp-fspi.txt | 42 ++++++++++++++++++++++
1 file changed, 42 insertions(+)
create mode 100644 Documentation/devicetree/bindings/spi/spi-nxp-fspi.txt
diff --git a/Documentation/devicetree/bindings/spi/spi-nxp-fspi.txt b/Documentation/devicetree/bindings/spi/spi-nxp-fspi.txt
new file mode 100644
index 0000000..94b5203
--- /dev/null
+++ b/Documentation/devicetree/bindings/spi/spi-nxp-fspi.txt
@@ -0,0 +1,42 @@
+* NXP Flex Serial Peripheral Interface (FSPI)
+
+Required properties:
+ - compatible : Should be "nxp,lx2160a-fspi"
+ - reg : First contains the register location and length,
+ Second contains the memory mapping address and length
+ - reg-names : Should contain the resource reg names:
+ - fspi_base: configuration register address space
+ - fspi_mmap: memory mapped address space
+ - interrupts : Should contain the interrupt for the device
+
+Optional properties:
+ - big-endian : See common-properties.txt.
+
+Required SPI slave node properties:
+ - reg : There are two buses (A and B) with two chip selects each.
+ This encodes to which bus and CS the flash is connected:
+ - <0>: Bus A, CS 0
+ - <1>: Bus A, CS 1
+ - <2>: Bus B, CS 0
+ - <3>: Bus B, CS 1
+
+Example showing the usage of two SPI NOR slave devices on bus A:
+
+fspi@0: flexspi@20c0000 {
+ compatible = "nxp,lx2160a-fspi";
+ reg = <0x0 0x20c0000 0x0 0x10000>, <0x0 0x20000000 0x0 0x10000000>;
+ reg-names = "fspi_base", "fspi_mmap";
+ interrupts = <0 25 0x4>; /* Level high type */
+ clocks = <&clockgen 4 3>, <&clockgen 4 3>;
+ clock-names = "fspi_en", "fspi";
+
+ flash@0: mt35xu512aba@0 {
+ reg = <0>;
+ ....
+ };
+
+ flash@1: mt35xu512aba@1 {
+ reg = <1>;
+ ....
+ };
+};
--
2.7.4
On Fri, Sep 21, 2018 at 03:52:00PM +0530, Yogesh Gaur wrote:
> Add binding file for NXP FlexSPI controller
>
> Signed-off-by: Yogesh Gaur <[email protected]>
> ---
> Changes for v3:
> - None
> Changes for v2:
> - Incorporated Rob review comments.
>
> .../devicetree/bindings/spi/spi-nxp-fspi.txt | 42 ++++++++++++++++++++++
> 1 file changed, 42 insertions(+)
> create mode 100644 Documentation/devicetree/bindings/spi/spi-nxp-fspi.txt
>
> diff --git a/Documentation/devicetree/bindings/spi/spi-nxp-fspi.txt b/Documentation/devicetree/bindings/spi/spi-nxp-fspi.txt
> new file mode 100644
> index 0000000..94b5203
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/spi/spi-nxp-fspi.txt
> @@ -0,0 +1,42 @@
> +* NXP Flex Serial Peripheral Interface (FSPI)
> +
> +Required properties:
> + - compatible : Should be "nxp,lx2160a-fspi"
> + - reg : First contains the register location and length,
> + Second contains the memory mapping address and length
> + - reg-names : Should contain the resource reg names:
> + - fspi_base: configuration register address space
> + - fspi_mmap: memory mapped address space
> + - interrupts : Should contain the interrupt for the device
> +
> +Optional properties:
> + - big-endian : See common-properties.txt.
> +
> +Required SPI slave node properties:
> + - reg : There are two buses (A and B) with two chip selects each.
> + This encodes to which bus and CS the flash is connected:
> + - <0>: Bus A, CS 0
> + - <1>: Bus A, CS 1
> + - <2>: Bus B, CS 0
> + - <3>: Bus B, CS 1
> +
> +Example showing the usage of two SPI NOR slave devices on bus A:
> +
> +fspi@0: flexspi@20c0000 {
Use standard node names.
spi@...
> + compatible = "nxp,lx2160a-fspi";
> + reg = <0x0 0x20c0000 0x0 0x10000>, <0x0 0x20000000 0x0 0x10000000>;
> + reg-names = "fspi_base", "fspi_mmap";
> + interrupts = <0 25 0x4>; /* Level high type */
> + clocks = <&clockgen 4 3>, <&clockgen 4 3>;
> + clock-names = "fspi_en", "fspi";
> +
> + flash@0: mt35xu512aba@0 {
You've got the label and node name reversed. Drop the '@' in the label.
That should be reserved for node names.
> + reg = <0>;
> + ....
> + };
> +
> + flash@1: mt35xu512aba@1 {
> + reg = <1>;
> + ....
> + };
> +};
> --
> 2.7.4
>
Hi Yogesh,
On Fri, 21 Sep 2018 15:51:59 +0530
Yogesh Gaur <[email protected]> wrote:
> +/* Registers used by the driver */
> +#define FSPI_MCR0 0x00
> +#define FSPI_MCR0_AHB_TIMEOUT_SHIFT 24
> +#define FSPI_MCR0_AHB_TIMEOUT_MASK (0xFF << FSPI_MCR0_AHB_TIMEOUT_SHIFT)
> +#define FSPI_MCR0_IP_TIMEOUT_SHIFT 16
> +#define FSPI_MCR0_IP_TIMEOUT_MASK (0xFF << FSPI_MCR0_IP_TIMEOUT_SHIFT)
> +#define FSPI_MCR0_LEARN_EN_SHIFT 15
> +#define FSPI_MCR0_LEARN_EN_MASK (1 << FSPI_MCR0_LEARN_EN_SHIFT)
> +#define FSPI_MCR0_SCRFRUN_EN_SHIFT 14
> +#define FSPI_MCR0_SCRFRUN_EN_MASK (1 << FSPI_MCR0_SCRFRUN_EN_SHIFT)
> +#define FSPI_MCR0_OCTCOMB_EN_SHIFT 13
> +#define FSPI_MCR0_OCTCOMB_EN_MASK (1 << FSPI_MCR0_OCTCOMB_EN_SHIFT)
> +#define FSPI_MCR0_DOZE_EN_SHIFT 12
> +#define FSPI_MCR0_DOZE_EN_MASK (1 << FSPI_MCR0_DOZE_EN_SHIFT)
> +#define FSPI_MCR0_HSEN_SHIFT 11
> +#define FSPI_MCR0_HSEN_MASK (1 << FSPI_MCR0_HSEN_SHIFT)
> +#define FSPI_MCR0_SERCLKDIV_SHIFT 8
> +#define FSPI_MCR0_SERCLKDIV_MASK (7 << FSPI_MCR0_SERCLKDIV_SHIFT)
> +#define FSPI_MCR0_ATDF_EN_SHIFT 7
> +#define FSPI_MCR0_ATDF_EN_MASK (1 << FSPI_MCR0_ATDF_EN_SHIFT)
> +#define FSPI_MCR0_ARDF_EN_SHIFT 6
> +#define FSPI_MCR0_ARDF_EN_MASK (1 << FSPI_MCR0_ARDF_EN_SHIFT)
> +#define FSPI_MCR0_RXCLKSRC_SHIFT 4
> +#define FSPI_MCR0_RXCLKSRC_MASK (3 << FSPI_MCR0_RXCLKSRC_SHIFT)
> +#define FSPI_MCR0_END_CFG_SHIFT 2
> +#define FSPI_MCR0_END_CFG_MASK (3 << FSPI_MCR0_END_CFG_SHIFT)
> +#define FSPI_MCR0_MDIS_SHIFT 1
> +#define FSPI_MCR0_MDIS_MASK (1 << FSPI_MCR0_MDIS_SHIFT)
> +#define FSPI_MCR0_SWRST_SHIFT 0
> +#define FSPI_MCR0_SWRST_MASK (1 << FSPI_MCR0_SWRST_SHIFT)
Do we really need all those _SHIFT/_MASK defs? I mean
#define FSPI_MCR0_SWRST BIT(0)
or
#define FSPI_MCR0_AHB_TIMEOUT(x) ((x) << 24)
#define FSPI_MCR0_AHB_TIMEOUT_MASK GENMASK(31, 24)
are just fine.
> +
> +enum nxp_fspi_devtype {
> + NXP_FSPI_LX2160A,
> +};
I'm pretty sure you don't need this enum if you describe all dev caps
in the nxp_fspi_devtype_data struct.
> +
> +struct nxp_fspi_devtype_data {
> + enum nxp_fspi_devtype devtype;
> + unsigned int rxfifo;
> + unsigned int txfifo;
> + unsigned int ahb_buf_size;
> + unsigned int quirks;
> + bool endianness;
How about renaming this variable big_endian and dropping the
{L,B}_ENDIAN macros?
> +};
[...]
> +struct nxp_fspi {
> + void __iomem *iobase;
> + void __iomem *ahb_addr;
> + u32 memmap_phy;
> + u32 memmap_phy_size;
> + struct clk *clk, *clk_en;
> + struct device *dev;
> + struct completion c;
> + const struct nxp_fspi_devtype_data *devtype_data;
> + struct mutex lock;
> + struct pm_qos_request pm_qos_req;
> + int selected;
> + void (*write)(u32 val, void __iomem *addr);
> + u32 (*read)(void __iomem *addr);
> +};
> +
> +static void fspi_writel_be(u32 val, void __iomem *addr)
> +{
> + iowrite32be(val, addr);
> +}
> +
> +static void fspi_writel(u32 val, void __iomem *addr)
> +{
> + iowrite32(val, addr);
> +}
> +
> +static u32 fspi_readl_be(void __iomem *addr)
> +{
> + return ioread32be(addr);
> +}
> +
> +static u32 fspi_readl(void __iomem *addr)
> +{
> + return ioread32(addr);
> +}
Hm, I'd recommend dropping the ->read/write() hooks and providing the
following functions:
static void fspi_writel(struct nxp_fspi *f, u32 val, void __iomem *addr)
{
if (f->big_endian)
iowrite32be(val, addr);
else
iowrite32(val, addr);
}
static u32 fspi_readl(struct nxp_fspi *f, void __iomem *addr)
{
if (f->big_endian)
return ioread32be(addr);
else
return ioread32(addr);
}
> +
> +static irqreturn_t nxp_fspi_irq_handler(int irq, void *dev_id)
> +{
> + struct nxp_fspi *f = dev_id;
> + u32 reg;
> +
> + /* clear interrupt */
> + reg = f->read(f->iobase + FSPI_INTR);
> + f->write(FSPI_INTR_IPCMDDONE_MASK, f->iobase + FSPI_INTR);
> +
> + if (reg & FSPI_INTR_IPCMDDONE_MASK)
> + complete(&f->c);
> +
> + return IRQ_HANDLED;
> +}
[...]
> +/*
> + * If the slave device content being changed by Write/Erase, need to
> + * invalidate the AHB buffer. This can be achieved by doing the reset
> + * of controller after setting MCR0[SWRESET] bit.
> + */
> +static inline void nxp_fspi_invalid(struct nxp_fspi *f)
> +{
> + u32 reg;
> +
> + reg = f->read(f->iobase + FSPI_MCR0);
> + f->write(reg | FSPI_MCR0_SWRST_MASK, f->iobase + FSPI_MCR0);
> +
> + while (f->read(f->iobase + FSPI_MCR0) & FSPI_MCR0_SWRST_MASK)
> + ;
Did you consider using readl_poll_timeout[_atomic]()?
if (f->big_endian)
mask = (u32)cpu_to_be32(FSPI_MCR0_SWRST_MASK);
else
mask = (u32)cpu_to_be32(FSPI_MCR0_SWRST_MASK);
ret = readl_poll_timeout(f->iobase + FSPI_MCR0, reg,
reg & mask, 0, FSPI_SWRST_TIMEOUT);
WARN_ON(ret);
> +}
[...]
> +static void nxp_fspi_read_ahb(struct nxp_fspi *f, const struct spi_mem_op *op)
> +{
> + u32 len = op->data.nbytes;
> +
> + /* Read out the data directly from the AHB buffer. */
> + memcpy_fromio(op->data.buf.in, (f->ahb_addr + op->addr.val), len);
Don't know if it's supported, but if it is, I recommend using DMA to do
this copy, because otherwise you might stall the CPU for quite a long
time if the flash is operating in a low-speed mode, and RT maintainers
will complain about that at some point ;-).
> +}
> +
> +static void nxp_fspi_fill_txfifo(struct nxp_fspi *f,
> + const struct spi_mem_op *op)
> +{
> + void __iomem *base = f->iobase;
> + int i, j;
> + int size, tmp_size, wm_size;
> + u32 data = 0;
> + u32 *txbuf = (u32 *) op->data.buf.out;
> +
> + /* clear the TX FIFO. */
> + f->write(FSPI_IPTXFCR_CLR_MASK, base + FSPI_IPTXFCR);
> +
> + /* Default value of water mark level is 8 bytes. */
> + wm_size = 8;
> + size = op->data.nbytes / wm_size;
> + for (i = 0; i < size; i++) {
> + /* Wait for TXFIFO empty */
> + while (!(f->read(base + FSPI_INTR) & FSPI_INTR_IPTXWE_MASK))
> + ;
Use readl_poll_timeout(), or even better, provide an helper
(fspi_readl_poll_timeout()?) that hides the BE/LE stuff, so that you
can reuse it when this pattern occurs.
[...]
> +static int nxp_fspi_exec_op(struct spi_mem *mem, const struct spi_mem_op *op)
> +{
> + struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->master);
> + void __iomem *base = f->iobase;
> + int err = 0;
> + unsigned int timeout = 1000;
> +
> + mutex_lock(&f->lock);
> +
> + /* wait for the controller being ready */
> + do {
> + u32 status;
> +
> + status = f->read(base + FSPI_STS0);
> + if ((status & FSPI_STS0_ARB_IDLE_MASK) &&
> + (status & FSPI_STS0_SEQ_IDLE_MASK))
> + break;
> + udelay(1);
> + dev_dbg(f->dev, "The controller is busy, 0x%x\n", status);
Same here.
Note that I didn't spend time looking at how the IP works, which
explains why I focus on tiny details here. Unfortunately, I won't have
time to review the driver in more details, so I'll leave that to
someone else, or let Mark decides if he's happy enough with the current
version.
Regards,
Boris
Hi Boris,
On 29.09.2018 17:40, Boris Brezillon wrote:
> Hi Yogesh,
>
> On Fri, 21 Sep 2018 15:51:59 +0530
> Yogesh Gaur <[email protected]> wrote:
>
>> +/* Registers used by the driver */
>> +#define FSPI_MCR0 0x00
>> +#define FSPI_MCR0_AHB_TIMEOUT_SHIFT 24
>> +#define FSPI_MCR0_AHB_TIMEOUT_MASK (0xFF << FSPI_MCR0_AHB_TIMEOUT_SHIFT)
>> +#define FSPI_MCR0_IP_TIMEOUT_SHIFT 16
>> +#define FSPI_MCR0_IP_TIMEOUT_MASK (0xFF << FSPI_MCR0_IP_TIMEOUT_SHIFT)
>> +#define FSPI_MCR0_LEARN_EN_SHIFT 15
>> +#define FSPI_MCR0_LEARN_EN_MASK (1 << FSPI_MCR0_LEARN_EN_SHIFT)
>> +#define FSPI_MCR0_SCRFRUN_EN_SHIFT 14
>> +#define FSPI_MCR0_SCRFRUN_EN_MASK (1 << FSPI_MCR0_SCRFRUN_EN_SHIFT)
>> +#define FSPI_MCR0_OCTCOMB_EN_SHIFT 13
>> +#define FSPI_MCR0_OCTCOMB_EN_MASK (1 << FSPI_MCR0_OCTCOMB_EN_SHIFT)
>> +#define FSPI_MCR0_DOZE_EN_SHIFT 12
>> +#define FSPI_MCR0_DOZE_EN_MASK (1 << FSPI_MCR0_DOZE_EN_SHIFT)
>> +#define FSPI_MCR0_HSEN_SHIFT 11
>> +#define FSPI_MCR0_HSEN_MASK (1 << FSPI_MCR0_HSEN_SHIFT)
>> +#define FSPI_MCR0_SERCLKDIV_SHIFT 8
>> +#define FSPI_MCR0_SERCLKDIV_MASK (7 << FSPI_MCR0_SERCLKDIV_SHIFT)
>> +#define FSPI_MCR0_ATDF_EN_SHIFT 7
>> +#define FSPI_MCR0_ATDF_EN_MASK (1 << FSPI_MCR0_ATDF_EN_SHIFT)
>> +#define FSPI_MCR0_ARDF_EN_SHIFT 6
>> +#define FSPI_MCR0_ARDF_EN_MASK (1 << FSPI_MCR0_ARDF_EN_SHIFT)
>> +#define FSPI_MCR0_RXCLKSRC_SHIFT 4
>> +#define FSPI_MCR0_RXCLKSRC_MASK (3 << FSPI_MCR0_RXCLKSRC_SHIFT)
>> +#define FSPI_MCR0_END_CFG_SHIFT 2
>> +#define FSPI_MCR0_END_CFG_MASK (3 << FSPI_MCR0_END_CFG_SHIFT)
>> +#define FSPI_MCR0_MDIS_SHIFT 1
>> +#define FSPI_MCR0_MDIS_MASK (1 << FSPI_MCR0_MDIS_SHIFT)
>> +#define FSPI_MCR0_SWRST_SHIFT 0
>> +#define FSPI_MCR0_SWRST_MASK (1 << FSPI_MCR0_SWRST_SHIFT)
>
> Do we really need all those _SHIFT/_MASK defs? I mean
>
> #define FSPI_MCR0_SWRST BIT(0)
>
> or
>
> #define FSPI_MCR0_AHB_TIMEOUT(x) ((x) << 24)
> #define FSPI_MCR0_AHB_TIMEOUT_MASK GENMASK(31, 24)
>
> are just fine.
>
>> +
>> +enum nxp_fspi_devtype {
>> + NXP_FSPI_LX2160A,
>> +};
>
> I'm pretty sure you don't need this enum if you describe all dev caps
> in the nxp_fspi_devtype_data struct.
>
>> +
>> +struct nxp_fspi_devtype_data {
>> + enum nxp_fspi_devtype devtype;
>> + unsigned int rxfifo;
>> + unsigned int txfifo;
>> + unsigned int ahb_buf_size;
>> + unsigned int quirks;
>> + bool endianness;
>
> How about renaming this variable big_endian and dropping the
> {L,B}_ENDIAN macros?
>
>> +};
>
> [...]
>
>> +struct nxp_fspi {
>> + void __iomem *iobase;
>> + void __iomem *ahb_addr;
>> + u32 memmap_phy;
>> + u32 memmap_phy_size;
>> + struct clk *clk, *clk_en;
>> + struct device *dev;
>> + struct completion c;
>> + const struct nxp_fspi_devtype_data *devtype_data;
>> + struct mutex lock;
>> + struct pm_qos_request pm_qos_req;
>> + int selected;
>> + void (*write)(u32 val, void __iomem *addr);
>> + u32 (*read)(void __iomem *addr);
>> +};
>> +
>> +static void fspi_writel_be(u32 val, void __iomem *addr)
>> +{
>> + iowrite32be(val, addr);
>> +}
>> +
>> +static void fspi_writel(u32 val, void __iomem *addr)
>> +{
>> + iowrite32(val, addr);
>> +}
>> +
>> +static u32 fspi_readl_be(void __iomem *addr)
>> +{
>> + return ioread32be(addr);
>> +}
>> +
>> +static u32 fspi_readl(void __iomem *addr)
>> +{
>> + return ioread32(addr);
>> +}
>
> Hm, I'd recommend dropping the ->read/write() hooks and providing the
> following functions:
>
> static void fspi_writel(struct nxp_fspi *f, u32 val, void __iomem *addr)
> {
> if (f->big_endian)
> iowrite32be(val, addr);
> else
> iowrite32(val, addr);
> }
>
>
> static u32 fspi_readl(struct nxp_fspi *f, void __iomem *addr)
> {
> if (f->big_endian)
> return ioread32be(addr);
> else
> return ioread32(addr);
> }
I introduced the ->read/write() hooks in the QSPI driver because I was
told to remove the conditional in the read/write path, but I can't
really tell if this really makes any difference.
Regards,
Frieder
>
>> +
>> +static irqreturn_t nxp_fspi_irq_handler(int irq, void *dev_id)
>> +{
>> + struct nxp_fspi *f = dev_id;
>> + u32 reg;
>> +
>> + /* clear interrupt */
>> + reg = f->read(f->iobase + FSPI_INTR);
>> + f->write(FSPI_INTR_IPCMDDONE_MASK, f->iobase + FSPI_INTR);
>> +
>> + if (reg & FSPI_INTR_IPCMDDONE_MASK)
>> + complete(&f->c);
>> +
>> + return IRQ_HANDLED;
>> +}
>
> [...]
>
>> +/*
>> + * If the slave device content being changed by Write/Erase, need to
>> + * invalidate the AHB buffer. This can be achieved by doing the reset
>> + * of controller after setting MCR0[SWRESET] bit.
>> + */
>> +static inline void nxp_fspi_invalid(struct nxp_fspi *f)
>> +{
>> + u32 reg;
>> +
>> + reg = f->read(f->iobase + FSPI_MCR0);
>> + f->write(reg | FSPI_MCR0_SWRST_MASK, f->iobase + FSPI_MCR0);
>> +
>> + while (f->read(f->iobase + FSPI_MCR0) & FSPI_MCR0_SWRST_MASK)
>> + ;
>
> Did you consider using readl_poll_timeout[_atomic]()?
>
> if (f->big_endian)
> mask = (u32)cpu_to_be32(FSPI_MCR0_SWRST_MASK);
> else
> mask = (u32)cpu_to_be32(FSPI_MCR0_SWRST_MASK);
>
> ret = readl_poll_timeout(f->iobase + FSPI_MCR0, reg,
> reg & mask, 0, FSPI_SWRST_TIMEOUT);
> WARN_ON(ret);
>
>> +}
>
> [...]
>
>> +static void nxp_fspi_read_ahb(struct nxp_fspi *f, const struct spi_mem_op *op)
>> +{
>> + u32 len = op->data.nbytes;
>> +
>> + /* Read out the data directly from the AHB buffer. */
>> + memcpy_fromio(op->data.buf.in, (f->ahb_addr + op->addr.val), len);
>
> Don't know if it's supported, but if it is, I recommend using DMA to do
> this copy, because otherwise you might stall the CPU for quite a long
> time if the flash is operating in a low-speed mode, and RT maintainers
> will complain about that at some point ;-).
>
>> +}
>> +
>> +static void nxp_fspi_fill_txfifo(struct nxp_fspi *f,
>> + const struct spi_mem_op *op)
>> +{
>> + void __iomem *base = f->iobase;
>> + int i, j;
>> + int size, tmp_size, wm_size;
>> + u32 data = 0;
>> + u32 *txbuf = (u32 *) op->data.buf.out;
>> +
>> + /* clear the TX FIFO. */
>> + f->write(FSPI_IPTXFCR_CLR_MASK, base + FSPI_IPTXFCR);
>> +
>> + /* Default value of water mark level is 8 bytes. */
>> + wm_size = 8;
>> + size = op->data.nbytes / wm_size;
>> + for (i = 0; i < size; i++) {
>> + /* Wait for TXFIFO empty */
>> + while (!(f->read(base + FSPI_INTR) & FSPI_INTR_IPTXWE_MASK))
>> + ;
>
> Use readl_poll_timeout(), or even better, provide an helper
> (fspi_readl_poll_timeout()?) that hides the BE/LE stuff, so that you
> can reuse it when this pattern occurs.
>
> [...]
>
>> +static int nxp_fspi_exec_op(struct spi_mem *mem, const struct spi_mem_op *op)
>> +{
>> + struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->master);
>> + void __iomem *base = f->iobase;
>> + int err = 0;
>> + unsigned int timeout = 1000;
>> +
>> + mutex_lock(&f->lock);
>> +
>> + /* wait for the controller being ready */
>> + do {
>> + u32 status;
>> +
>> + status = f->read(base + FSPI_STS0);
>> + if ((status & FSPI_STS0_ARB_IDLE_MASK) &&
>> + (status & FSPI_STS0_SEQ_IDLE_MASK))
>> + break;
>> + udelay(1);
>> + dev_dbg(f->dev, "The controller is busy, 0x%x\n", status);
>
> Same here.
>
> Note that I didn't spend time looking at how the IP works, which
> explains why I focus on tiny details here. Unfortunately, I won't have
> time to review the driver in more details, so I'll leave that to
> someone else, or let Mark decides if he's happy enough with the current
> version.
>
> Regards,
>
> Boris
>
Hi Boris,
> -----Original Message-----
> From: Frieder Schrempf [mailto:[email protected]]
> Sent: Monday, October 1, 2018 11:48 AM
> To: Boris Brezillon <[email protected]>; Yogesh Narayan Gaur
> <[email protected]>
> Cc: [email protected]; [email protected]; linux-
> [email protected]; [email protected]; [email protected];
> [email protected]; [email protected]; linux-arm-
> [email protected]; [email protected]; linux-
> [email protected]
> Subject: Re: [PATCH v3 1/5] spi: spi-mem: Add driver for NXP FlexSPI controller
>
> Hi Boris,
>
> On 29.09.2018 17:40, Boris Brezillon wrote:
> > Hi Yogesh,
> >
> > On Fri, 21 Sep 2018 15:51:59 +0530
> > Yogesh Gaur <[email protected]> wrote:
> >
> >> +/* Registers used by the driver */
> >> +#define FSPI_MCR0 0x00
> >> +#define FSPI_MCR0_AHB_TIMEOUT_SHIFT 24
> >> +#define FSPI_MCR0_AHB_TIMEOUT_MASK (0xFF <<
> FSPI_MCR0_AHB_TIMEOUT_SHIFT)
> >> +#define FSPI_MCR0_IP_TIMEOUT_SHIFT 16
> >> +#define FSPI_MCR0_IP_TIMEOUT_MASK (0xFF <<
> FSPI_MCR0_IP_TIMEOUT_SHIFT)
> >> +#define FSPI_MCR0_LEARN_EN_SHIFT 15
> >> +#define FSPI_MCR0_LEARN_EN_MASK (1 <<
> FSPI_MCR0_LEARN_EN_SHIFT)
> >> +#define FSPI_MCR0_SCRFRUN_EN_SHIFT 14
> >> +#define FSPI_MCR0_SCRFRUN_EN_MASK (1 <<
> FSPI_MCR0_SCRFRUN_EN_SHIFT)
> >> +#define FSPI_MCR0_OCTCOMB_EN_SHIFT 13
> >> +#define FSPI_MCR0_OCTCOMB_EN_MASK (1 <<
> FSPI_MCR0_OCTCOMB_EN_SHIFT)
> >> +#define FSPI_MCR0_DOZE_EN_SHIFT 12
> >> +#define FSPI_MCR0_DOZE_EN_MASK (1 <<
> FSPI_MCR0_DOZE_EN_SHIFT)
> >> +#define FSPI_MCR0_HSEN_SHIFT 11
> >> +#define FSPI_MCR0_HSEN_MASK (1 << FSPI_MCR0_HSEN_SHIFT)
> >> +#define FSPI_MCR0_SERCLKDIV_SHIFT 8
> >> +#define FSPI_MCR0_SERCLKDIV_MASK (7 <<
> FSPI_MCR0_SERCLKDIV_SHIFT)
> >> +#define FSPI_MCR0_ATDF_EN_SHIFT 7
> >> +#define FSPI_MCR0_ATDF_EN_MASK (1 <<
> FSPI_MCR0_ATDF_EN_SHIFT)
> >> +#define FSPI_MCR0_ARDF_EN_SHIFT 6
> >> +#define FSPI_MCR0_ARDF_EN_MASK (1 <<
> FSPI_MCR0_ARDF_EN_SHIFT)
> >> +#define FSPI_MCR0_RXCLKSRC_SHIFT 4
> >> +#define FSPI_MCR0_RXCLKSRC_MASK (3 <<
> FSPI_MCR0_RXCLKSRC_SHIFT)
> >> +#define FSPI_MCR0_END_CFG_SHIFT 2
> >> +#define FSPI_MCR0_END_CFG_MASK (3 <<
> FSPI_MCR0_END_CFG_SHIFT)
> >> +#define FSPI_MCR0_MDIS_SHIFT 1
> >> +#define FSPI_MCR0_MDIS_MASK (1 << FSPI_MCR0_MDIS_SHIFT)
> >> +#define FSPI_MCR0_SWRST_SHIFT 0
> >> +#define FSPI_MCR0_SWRST_MASK (1 <<
> FSPI_MCR0_SWRST_SHIFT)
> >
> > Do we really need all those _SHIFT/_MASK defs? I mean
> >
> > #define FSPI_MCR0_SWRST BIT(0)
> >
> > or
> >
> > #define FSPI_MCR0_AHB_TIMEOUT(x) ((x) << 24)
> > #define FSPI_MCR0_AHB_TIMEOUT_MASK GENMASK(31, 24)
> >
> > are just fine.
> >
> >> +
> >> +enum nxp_fspi_devtype {
> >> + NXP_FSPI_LX2160A,
> >> +};
> >
> > I'm pretty sure you don't need this enum if you describe all dev caps
> > in the nxp_fspi_devtype_data struct.
> >
> >> +
> >> +struct nxp_fspi_devtype_data {
> >> + enum nxp_fspi_devtype devtype;
> >> + unsigned int rxfifo;
> >> + unsigned int txfifo;
> >> + unsigned int ahb_buf_size;
> >> + unsigned int quirks;
> >> + bool endianness;
> >
> > How about renaming this variable big_endian and dropping the
> > {L,B}_ENDIAN macros?
> >
> >> +};
> >
> > [...]
> >
> >> +struct nxp_fspi {
> >> + void __iomem *iobase;
> >> + void __iomem *ahb_addr;
> >> + u32 memmap_phy;
> >> + u32 memmap_phy_size;
> >> + struct clk *clk, *clk_en;
> >> + struct device *dev;
> >> + struct completion c;
> >> + const struct nxp_fspi_devtype_data *devtype_data;
> >> + struct mutex lock;
> >> + struct pm_qos_request pm_qos_req;
> >> + int selected;
> >> + void (*write)(u32 val, void __iomem *addr);
> >> + u32 (*read)(void __iomem *addr);
> >> +};
> >> +
> >> +static void fspi_writel_be(u32 val, void __iomem *addr) {
> >> + iowrite32be(val, addr);
> >> +}
> >> +
> >> +static void fspi_writel(u32 val, void __iomem *addr) {
> >> + iowrite32(val, addr);
> >> +}
> >> +
> >> +static u32 fspi_readl_be(void __iomem *addr) {
> >> + return ioread32be(addr);
> >> +}
> >> +
> >> +static u32 fspi_readl(void __iomem *addr) {
> >> + return ioread32(addr);
> >> +}
> >
> > Hm, I'd recommend dropping the ->read/write() hooks and providing the
> > following functions:
> >
> > static void fspi_writel(struct nxp_fspi *f, u32 val, void __iomem
> > *addr) {
> > if (f->big_endian)
> > iowrite32be(val, addr);
> > else
> > iowrite32(val, addr);
> > }
> >
> >
> > static u32 fspi_readl(struct nxp_fspi *f, void __iomem *addr) {
> > if (f->big_endian)
> > return ioread32be(addr);
> > else
> > return ioread32(addr);
> > }
>
> I introduced the ->read/write() hooks in the QSPI driver because I was told to
> remove the conditional in the read/write path, but I can't really tell if this really
> makes any difference.
>
Yes, I have taken these hooks by looking into the comments received for Frieder's QSPI patch series.
For me this looks more clean and can be decided in the controller initialization sequence which hook would going to be invoked.
> Regards,
> Frieder
>
> >
> >> +
> >> +static irqreturn_t nxp_fspi_irq_handler(int irq, void *dev_id) {
> >> + struct nxp_fspi *f = dev_id;
> >> + u32 reg;
> >> +
> >> + /* clear interrupt */
> >> + reg = f->read(f->iobase + FSPI_INTR);
> >> + f->write(FSPI_INTR_IPCMDDONE_MASK, f->iobase + FSPI_INTR);
> >> +
> >> + if (reg & FSPI_INTR_IPCMDDONE_MASK)
> >> + complete(&f->c);
> >> +
> >> + return IRQ_HANDLED;
> >> +}
> >
> > [...]
> >
> >> +/*
> >> + * If the slave device content being changed by Write/Erase, need to
> >> + * invalidate the AHB buffer. This can be achieved by doing the
> >> +reset
> >> + * of controller after setting MCR0[SWRESET] bit.
> >> + */
> >> +static inline void nxp_fspi_invalid(struct nxp_fspi *f) {
> >> + u32 reg;
> >> +
> >> + reg = f->read(f->iobase + FSPI_MCR0);
> >> + f->write(reg | FSPI_MCR0_SWRST_MASK, f->iobase + FSPI_MCR0);
> >> +
> >> + while (f->read(f->iobase + FSPI_MCR0) & FSPI_MCR0_SWRST_MASK)
> >> + ;
> >
> > Did you consider using readl_poll_timeout[_atomic]()?
> >
> > if (f->big_endian)
> > mask = (u32)cpu_to_be32(FSPI_MCR0_SWRST_MASK);
> > else
> > mask = (u32)cpu_to_be32(FSPI_MCR0_SWRST_MASK);
> >
> > ret = readl_poll_timeout(f->iobase + FSPI_MCR0, reg,
> > reg & mask, 0, FSPI_SWRST_TIMEOUT);
> > WARN_ON(ret);
> >
> >> +}
Ok, would check usage of readl_poll_timeout_atomic() and modify current implementation of doing busy waiting using while() loop.
--
Regards
Yogesh Gaur.
> >
> > [...]
> >
> >> +static void nxp_fspi_read_ahb(struct nxp_fspi *f, const struct
> >> +spi_mem_op *op) {
> >> + u32 len = op->data.nbytes;
> >> +
> >> + /* Read out the data directly from the AHB buffer. */
> >> + memcpy_fromio(op->data.buf.in, (f->ahb_addr + op->addr.val), len);
> >
> > Don't know if it's supported, but if it is, I recommend using DMA to
> > do this copy, because otherwise you might stall the CPU for quite a
> > long time if the flash is operating in a low-speed mode, and RT
> > maintainers will complain about that at some point ;-).
> >
> >> +}
> >> +
> >> +static void nxp_fspi_fill_txfifo(struct nxp_fspi *f,
> >> + const struct spi_mem_op *op)
> >> +{
> >> + void __iomem *base = f->iobase;
> >> + int i, j;
> >> + int size, tmp_size, wm_size;
> >> + u32 data = 0;
> >> + u32 *txbuf = (u32 *) op->data.buf.out;
> >> +
> >> + /* clear the TX FIFO. */
> >> + f->write(FSPI_IPTXFCR_CLR_MASK, base + FSPI_IPTXFCR);
> >> +
> >> + /* Default value of water mark level is 8 bytes. */
> >> + wm_size = 8;
> >> + size = op->data.nbytes / wm_size;
> >> + for (i = 0; i < size; i++) {
> >> + /* Wait for TXFIFO empty */
> >> + while (!(f->read(base + FSPI_INTR) & FSPI_INTR_IPTXWE_MASK))
> >> + ;
> >
> > Use readl_poll_timeout(), or even better, provide an helper
> > (fspi_readl_poll_timeout()?) that hides the BE/LE stuff, so that you
> > can reuse it when this pattern occurs.
> >
> > [...]
> >
> >> +static int nxp_fspi_exec_op(struct spi_mem *mem, const struct
> >> +spi_mem_op *op) {
> >> + struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->master);
> >> + void __iomem *base = f->iobase;
> >> + int err = 0;
> >> + unsigned int timeout = 1000;
> >> +
> >> + mutex_lock(&f->lock);
> >> +
> >> + /* wait for the controller being ready */
> >> + do {
> >> + u32 status;
> >> +
> >> + status = f->read(base + FSPI_STS0);
> >> + if ((status & FSPI_STS0_ARB_IDLE_MASK) &&
> >> + (status & FSPI_STS0_SEQ_IDLE_MASK))
> >> + break;
> >> + udelay(1);
> >> + dev_dbg(f->dev, "The controller is busy, 0x%x\n", status);
> >
> > Same here.
> >
> > Note that I didn't spend time looking at how the IP works, which
> > explains why I focus on tiny details here. Unfortunately, I won't have
> > time to review the driver in more details, so I'll leave that to
> > someone else, or let Mark decides if he's happy enough with the
> > current version.
> >
> > Regards,
> >
> > Boris
> >
On Mon, 1 Oct 2018 09:02:32 +0000
Yogesh Narayan Gaur <[email protected]> wrote:
> > > static void fspi_writel(struct nxp_fspi *f, u32 val, void __iomem
> > > *addr) {
> > > if (f->big_endian)
> > > iowrite32be(val, addr);
> > > else
> > > iowrite32(val, addr);
> > > }
> > >
> > >
> > > static u32 fspi_readl(struct nxp_fspi *f, void __iomem *addr) {
> > > if (f->big_endian)
> > > return ioread32be(addr);
> > > else
> > > return ioread32(addr);
> > > }
> >
> > I introduced the ->read/write() hooks in the QSPI driver because I
> > was told to remove the conditional in the read/write path, but I
> > can't really tell if this really makes any difference.
> >
> Yes, I have taken these hooks by looking into the comments received
> for Frieder's QSPI patch series. For me this looks more clean and can
> be decided in the controller initialization sequence which hook would
> going to be invoked.
Well, depending on your CPU, the indirect branch (caused by the
function pointer call) might have an higher cost than the conditional
branch. Not sure we care about such micro-optimizations though.
Regarding the readability aspect, I still prefer my version, but I
guess that's a matter of taste.
Hi Boris,
> -----Original Message-----
> From: Boris Brezillon [mailto:[email protected]]
> Sent: Saturday, September 29, 2018 9:10 PM
> To: Yogesh Narayan Gaur <[email protected]>
> Cc: [email protected]; [email protected]; linux-
> [email protected]; [email protected]; [email protected];
> [email protected]; [email protected]; linux-arm-
> [email protected]; [email protected];
> [email protected]; [email protected]
> Subject: Re: [PATCH v3 1/5] spi: spi-mem: Add driver for NXP FlexSPI controller
>
> Hi Yogesh,
>
> On Fri, 21 Sep 2018 15:51:59 +0530
> Yogesh Gaur <[email protected]> wrote:
>
> > +/* Registers used by the driver */
> > +#define FSPI_MCR0 0x00
> > +#define FSPI_MCR0_AHB_TIMEOUT_SHIFT 24
> > +#define FSPI_MCR0_AHB_TIMEOUT_MASK (0xFF <<
> FSPI_MCR0_AHB_TIMEOUT_SHIFT)
> > +#define FSPI_MCR0_IP_TIMEOUT_SHIFT 16
> > +#define FSPI_MCR0_IP_TIMEOUT_MASK (0xFF <<
> FSPI_MCR0_IP_TIMEOUT_SHIFT)
> > +#define FSPI_MCR0_LEARN_EN_SHIFT 15
> > +#define FSPI_MCR0_LEARN_EN_MASK (1 <<
> FSPI_MCR0_LEARN_EN_SHIFT)
> > +#define FSPI_MCR0_SCRFRUN_EN_SHIFT 14
> > +#define FSPI_MCR0_SCRFRUN_EN_MASK (1 <<
> FSPI_MCR0_SCRFRUN_EN_SHIFT)
> > +#define FSPI_MCR0_OCTCOMB_EN_SHIFT 13
> > +#define FSPI_MCR0_OCTCOMB_EN_MASK (1 <<
> FSPI_MCR0_OCTCOMB_EN_SHIFT)
> > +#define FSPI_MCR0_DOZE_EN_SHIFT 12
> > +#define FSPI_MCR0_DOZE_EN_MASK (1 <<
> FSPI_MCR0_DOZE_EN_SHIFT)
> > +#define FSPI_MCR0_HSEN_SHIFT 11
> > +#define FSPI_MCR0_HSEN_MASK (1 << FSPI_MCR0_HSEN_SHIFT)
> > +#define FSPI_MCR0_SERCLKDIV_SHIFT 8
> > +#define FSPI_MCR0_SERCLKDIV_MASK (7 <<
> FSPI_MCR0_SERCLKDIV_SHIFT)
> > +#define FSPI_MCR0_ATDF_EN_SHIFT 7
> > +#define FSPI_MCR0_ATDF_EN_MASK (1 <<
> FSPI_MCR0_ATDF_EN_SHIFT)
> > +#define FSPI_MCR0_ARDF_EN_SHIFT 6
> > +#define FSPI_MCR0_ARDF_EN_MASK (1 <<
> FSPI_MCR0_ARDF_EN_SHIFT)
> > +#define FSPI_MCR0_RXCLKSRC_SHIFT 4
> > +#define FSPI_MCR0_RXCLKSRC_MASK (3 <<
> FSPI_MCR0_RXCLKSRC_SHIFT)
> > +#define FSPI_MCR0_END_CFG_SHIFT 2
> > +#define FSPI_MCR0_END_CFG_MASK (3 <<
> FSPI_MCR0_END_CFG_SHIFT)
> > +#define FSPI_MCR0_MDIS_SHIFT 1
> > +#define FSPI_MCR0_MDIS_MASK (1 << FSPI_MCR0_MDIS_SHIFT)
> > +#define FSPI_MCR0_SWRST_SHIFT 0
> > +#define FSPI_MCR0_SWRST_MASK (1 <<
> FSPI_MCR0_SWRST_SHIFT)
>
> Do we really need all those _SHIFT/_MASK defs? I mean
>
Changes done in next version of the patch series, v4 [1].
> #define FSPI_MCR0_SWRST BIT(0)
>
> or
>
> #define FSPI_MCR0_AHB_TIMEOUT(x) ((x) << 24)
> #define FSPI_MCR0_AHB_TIMEOUT_MASK GENMASK(31, 24)
>
> are just fine.
>
> > +
> > +enum nxp_fspi_devtype {
> > + NXP_FSPI_LX2160A,
> > +};
>
> I'm pretty sure you don't need this enum if you describe all dev caps in the
> nxp_fspi_devtype_data struct.
Done, in v4.
>
> > +
> > +struct nxp_fspi_devtype_data {
> > + enum nxp_fspi_devtype devtype;
> > + unsigned int rxfifo;
> > + unsigned int txfifo;
> > + unsigned int ahb_buf_size;
> > + unsigned int quirks;
> > + bool endianness;
>
> How about renaming this variable big_endian and dropping the {L,B}_ENDIAN
> macros?
>
Done in v4, default make as little_endian to reduce indirect branch checking.
> > +};
>
> [...]
>
> > +struct nxp_fspi {
> > + void __iomem *iobase;
> > + void __iomem *ahb_addr;
> > + u32 memmap_phy;
> > + u32 memmap_phy_size;
> > + struct clk *clk, *clk_en;
> > + struct device *dev;
> > + struct completion c;
> > + const struct nxp_fspi_devtype_data *devtype_data;
> > + struct mutex lock;
> > + struct pm_qos_request pm_qos_req;
> > + int selected;
> > + void (*write)(u32 val, void __iomem *addr);
> > + u32 (*read)(void __iomem *addr);
> > +};
> > +
> > +static void fspi_writel_be(u32 val, void __iomem *addr) {
> > + iowrite32be(val, addr);
> > +}
> > +
> > +static void fspi_writel(u32 val, void __iomem *addr) {
> > + iowrite32(val, addr);
> > +}
> > +
> > +static u32 fspi_readl_be(void __iomem *addr) {
> > + return ioread32be(addr);
> > +}
> > +
> > +static u32 fspi_readl(void __iomem *addr) {
> > + return ioread32(addr);
> > +}
>
> Hm, I'd recommend dropping the ->read/write() hooks and providing the
> following functions:
>
> static void fspi_writel(struct nxp_fspi *f, u32 val, void __iomem *addr) {
> if (f->big_endian)
> iowrite32be(val, addr);
> else
> iowrite32(val, addr);
> }
>
>
> static u32 fspi_readl(struct nxp_fspi *f, void __iomem *addr) {
> if (f->big_endian)
> return ioread32be(addr);
> else
> return ioread32(addr);
> }
>
> > +
> > +static irqreturn_t nxp_fspi_irq_handler(int irq, void *dev_id) {
> > + struct nxp_fspi *f = dev_id;
> > + u32 reg;
> > +
> > + /* clear interrupt */
> > + reg = f->read(f->iobase + FSPI_INTR);
> > + f->write(FSPI_INTR_IPCMDDONE_MASK, f->iobase + FSPI_INTR);
> > +
> > + if (reg & FSPI_INTR_IPCMDDONE_MASK)
> > + complete(&f->c);
> > +
> > + return IRQ_HANDLED;
> > +}
>
> [...]
>
> > +/*
> > + * If the slave device content being changed by Write/Erase, need to
> > + * invalidate the AHB buffer. This can be achieved by doing the reset
> > + * of controller after setting MCR0[SWRESET] bit.
> > + */
> > +static inline void nxp_fspi_invalid(struct nxp_fspi *f) {
> > + u32 reg;
> > +
> > + reg = f->read(f->iobase + FSPI_MCR0);
> > + f->write(reg | FSPI_MCR0_SWRST_MASK, f->iobase + FSPI_MCR0);
> > +
> > + while (f->read(f->iobase + FSPI_MCR0) & FSPI_MCR0_SWRST_MASK)
> > + ;
>
> Did you consider using readl_poll_timeout[_atomic]()?
>
> if (f->big_endian)
> mask = (u32)cpu_to_be32(FSPI_MCR0_SWRST_MASK);
> else
> mask = (u32)cpu_to_be32(FSPI_MCR0_SWRST_MASK);
>
> ret = readl_poll_timeout(f->iobase + FSPI_MCR0, reg,
> reg & mask, 0, FSPI_SWRST_TIMEOUT);
> WARN_ON(ret);
>
> > +}
>
Thanks, added readl_poll_timeout() functionality instead of busy looping.
> [...]
>
> > +static void nxp_fspi_read_ahb(struct nxp_fspi *f, const struct
> > +spi_mem_op *op) {
> > + u32 len = op->data.nbytes;
> > +
> > + /* Read out the data directly from the AHB buffer. */
> > + memcpy_fromio(op->data.buf.in, (f->ahb_addr + op->addr.val), len);
>
> Don't know if it's supported, but if it is, I recommend using DMA to do this copy,
> because otherwise you might stall the CPU for quite a long time if the flash is
> operating in a low-speed mode, and RT maintainers will complain about that at
> some point ;-).
>
Read using DMA is not supported by the controller in AHB mode, only supported in IP mode.
Have to use memcpy_fromio() calls. Maximum data size can be read in single call is 0x800 using AHB read.
--
Regards
Yogesh Gaur.
[1] https://patchwork.ozlabs.org/project/linux-mtd/list/?series=69535
> > +}
> > +
> > +static void nxp_fspi_fill_txfifo(struct nxp_fspi *f,
> > + const struct spi_mem_op *op)
> > +{
> > + void __iomem *base = f->iobase;
> > + int i, j;
> > + int size, tmp_size, wm_size;
> > + u32 data = 0;
> > + u32 *txbuf = (u32 *) op->data.buf.out;
> > +
> > + /* clear the TX FIFO. */
> > + f->write(FSPI_IPTXFCR_CLR_MASK, base + FSPI_IPTXFCR);
> > +
> > + /* Default value of water mark level is 8 bytes. */
> > + wm_size = 8;
> > + size = op->data.nbytes / wm_size;
> > + for (i = 0; i < size; i++) {
> > + /* Wait for TXFIFO empty */
> > + while (!(f->read(base + FSPI_INTR) & FSPI_INTR_IPTXWE_MASK))
> > + ;
>
> Use readl_poll_timeout(), or even better, provide an helper
> (fspi_readl_poll_timeout()?) that hides the BE/LE stuff, so that you can reuse it
> when this pattern occurs.
>
> [...]
>
> > +static int nxp_fspi_exec_op(struct spi_mem *mem, const struct
> > +spi_mem_op *op) {
> > + struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->master);
> > + void __iomem *base = f->iobase;
> > + int err = 0;
> > + unsigned int timeout = 1000;
> > +
> > + mutex_lock(&f->lock);
> > +
> > + /* wait for the controller being ready */
> > + do {
> > + u32 status;
> > +
> > + status = f->read(base + FSPI_STS0);
> > + if ((status & FSPI_STS0_ARB_IDLE_MASK) &&
> > + (status & FSPI_STS0_SEQ_IDLE_MASK))
> > + break;
> > + udelay(1);
> > + dev_dbg(f->dev, "The controller is busy, 0x%x\n", status);
>
> Same here.
>
> Note that I didn't spend time looking at how the IP works, which explains why I
> focus on tiny details here. Unfortunately, I won't have time to review the driver
> in more details, so I'll leave that to someone else, or let Mark decides if he's
> happy enough with the current version.
>
> Regards,
>
> Boris
On Mon, 8 Oct 2018 11:21:13 +0000
Yogesh Narayan Gaur <[email protected]> wrote:
> > > +static void nxp_fspi_read_ahb(struct nxp_fspi *f, const struct
> > > +spi_mem_op *op) {
> > > + u32 len = op->data.nbytes;
> > > +
> > > + /* Read out the data directly from the AHB buffer. */
> > > + memcpy_fromio(op->data.buf.in, (f->ahb_addr + op->addr.val), len);
> >
> > Don't know if it's supported, but if it is, I recommend using DMA to do this copy,
> > because otherwise you might stall the CPU for quite a long time if the flash is
> > operating in a low-speed mode, and RT maintainers will complain about that at
> > some point ;-).
> >
> Read using DMA is not supported by the controller in AHB mode, only supported in IP mode.
> Have to use memcpy_fromio() calls. Maximum data size can be read in single call is 0x800 using AHB read.
>
Still, blocking the CPU until the SPI controller has read 0x800 bytes
is enough to make you miss a real-time deadline. Don't you have a
controller that supports mem2mem transfers (I'm pretty sure a mem2mem
DMA transfer would do the trick here)?
Regards,
Boris