Skip to contentSkip to conversation

at91: add support for CDU9G25 board

PATCH Superseded Jiri Prchal 2 participants 6 messages Sep 13, 2013 – Dec 9, 2013 Arm (32-bit)

lore
Share & tools
lore
#0 lore raw

PATCH v5 1/1 at91: add support for CDU9G25 board

This patch adds support for our companies board CDU9G25 with Atmel AT91SAM9G25, 128MB DDR2, 256MB NAND.

---
v.2
Fixed static IP and MAC addr cofiguration by random MAC and DHCP. Added entry to MAINTAINERS.
Fixed line lenght, trailing spaces and other cosmetics. v.3
Removed static eth_gadget addresses.
MAC address set from serial number of F-RAM.
v.4
Changed debug to printf if used random MAC addr, print that addr. Keep MAC addr set in enviroment.
v.5
Changes to Bo Shen comments.

---
 arch/arm/include/asm/mach-types.h |    1 +
 board/AKsignal/cdu9g25/Makefile   |   34 ++++++
 board/AKsignal/cdu9g25/cdu9g25.c  |  202 +++++++++++++++++++++++++++++++++++
 boards.cfg                        |    1 +
 include/configs/cdu9g25.h         |  210 +++++++++++++++++++++++++++++++++++++
 5 files changed, 448 insertions(+)
 create mode 100644 board/AKsignal/cdu9g25/Makefile
 create mode 100644 board/AKsignal/cdu9g25/cdu9g25.c
 create mode 100644 include/configs/cdu9g25.h
arch/arm/include/asm/mach-types.h +1 -0
index 440b041..9b274ba 100644
@@ -986,6 +986,7 @@ extern unsigned int __machine_arch_type;
 #define MACH_TYPE_VIT_IBOX             3371
 #define MACH_TYPE_DM6441_ESP           3372
 #define MACH_TYPE_AT91SAM9X5EK         3373
+#define MACH_TYPE_CDU9G25              3373
 #define MACH_TYPE_LIBRA                3374
 #define MACH_TYPE_EASYCRRH             3375
 #define MACH_TYPE_TRIPEL               3376
board/AKsignal/cdu9g25/Makefile new file +34 -0
new file mode 100644
index 0000000..b644a83
@@ -0,0 +1,34 @@
+#
+# U-boot - Makefile
+#
+# (C) Copyright 2013
+# Jiri Prchal <[email protected]>
+# AK signal <www.aksignal.cz>
+#
+# (C) Copyright 2000-2006
+# Wolfgang Denk, DENX Software Engineering, wd at denx.de.
+#
+# SPDX-License-Identifier:	GPL-2.0+
+#
+
+include $(TOPDIR)/config.mk
+
+LIB	= $(obj)lib$(BOARD).o
+
+COBJS-y	+= cdu9g25.o
+
+SRCS	:= $(SOBJS:.o=.S) $(COBJS-y:.o=.c)
+OBJS	:= $(addprefix $(obj),$(COBJS-y))
+SOBJS	:= $(addprefix $(obj),$(SOBJS))
+
+$(LIB):	$(obj).depend $(OBJS) $(SOBJS)
+	$(call cmd_link_o_target, $(OBJS) $(SOBJS))
+
+#########################################################################
+
+# defines $(obj).depend target
+include $(SRCTREE)/rules.mk
+
+sinclude $(obj).depend
+
+#########################################################################
board/AKsignal/cdu9g25/cdu9g25.c new file +202 -0

202 lines not on this page · raw message

boards.cfg +1 -0
index dbd8479..7a7d39e 100644
@@ -77,6 +77,7 @@ Active  arm         arm926ejs      -           armltd          integrator
 Active  arm         arm926ejs      armada100   Marvell         -                   aspenite                             -                                                                                                                                 Prafulla Wadaskar <[email protected]>
 Active  arm         arm926ejs      armada100   Marvell         -                   gplugd                               -                                                                                                                                 Ajay Bhargav <[email protected]>
 Active  arm         arm926ejs      at91        -               -                   afeb9260                             -                                                                                                                                 Sergey Lapin <[email protected]>
+Active  arm         arm926ejs      at91        AKsignal        cdu9g25             cdu9g25                              -                                                                                                                                 Jiri Prchal <[email protected]>
 Active  arm         arm926ejs      at91        atmel           at91sam9260ek       at91sam9260ek_dataflash_cs0          at91sam9260ek:AT91SAM9260,SYS_USE_DATAFLASH_CS0                                                                                   Stelian Pop <[email protected]>
 Active  arm         arm926ejs      at91        atmel           at91sam9260ek       at91sam9260ek_dataflash_cs1          at91sam9260ek:AT91SAM9260,SYS_USE_DATAFLASH_CS1                                                                                   Stelian Pop <[email protected]>
 Active  arm         arm926ejs      at91        atmel           at91sam9260ek       at91sam9260ek_nandflash              at91sam9260ek:AT91SAM9260,SYS_USE_NANDFLASH                                                                                       Stelian Pop <[email protected]>
include/configs/cdu9g25.h new file +210 -0

210 lines not on this page · raw message

--
1.7.9.5

#1 lore raw

PATCH v5 1/1 at91: add support for CDU9G25 board

Dear Jiri Prchal,

Quoted 34 lines · Jiri Prchal

This patch adds support for our companies board CDU9G25 with Atmel AT91SAM9G25, 128MB DDR2, 256MB NAND.

---
v.2
Fixed static IP and MAC addr cofiguration by random MAC and DHCP. Added entry to MAINTAINERS.
Fixed line lenght, trailing spaces and other cosmetics. v.3
Removed static eth_gadget addresses.
MAC address set from serial number of F-RAM.
v.4
Changed debug to printf if used random MAC addr, print that addr. Keep MAC addr set in enviroment.
v.5
Changes to Bo Shen comments.

---
 arch/arm/include/asm/mach-types.h |    1 +
 board/AKsignal/cdu9g25/Makefile   |   34 ++++++
 board/AKsignal/cdu9g25/cdu9g25.c  |  202 +++++++++++++++++++++++++++++++++++
 boards.cfg                        |    1 +
 include/configs/cdu9g25.h         |  210 +++++++++++++++++++++++++++++++++++++
 5 files changed, 448 insertions(+)
 create mode 100644 board/AKsignal/cdu9g25/Makefile
 create mode 100644 board/AKsignal/cdu9g25/cdu9g25.c
 create mode 100644 include/configs/cdu9g25.h
arch/arm/include/asm/mach-types.h +0 -0
index 440b041..9b274ba 100644

arch/arm/include/asm/mach-types.h extern unsigned int __machine_arch_type; · line 989

Quoted 4 lines
arch/arm/include/asm/mach-types.h +1 -0
@@ -986,6 +986,7 @@ extern unsigned int __machine_arch_type;
 #define MACH_TYPE_VIT_IBOX             3371
 #define MACH_TYPE_DM6441_ESP           3372
 #define MACH_TYPE_AT91SAM9X5EK         3373
+#define MACH_TYPE_CDU9G25              3373

NAK, please obtain a mach type:
http://www.arm.linux.org.uk/developer/machines/?action=new

Quoted 182 lines
arch/arm/include/asm/mach-types.h +0 -0
 #define MACH_TYPE_LIBRA                3374
 #define MACH_TYPE_EASYCRRH             3375
 #define MACH_TYPE_TRIPEL               3376
board/AKsignal/cdu9g25/Makefile new file +34 -0
new file mode 100644
index 0000000..b644a83
@@ -0,0 +1,34 @@
+#
+# U-boot - Makefile
+#
+# (C) Copyright 2013
+# Jiri Prchal <[email protected]>
+# AK signal <www.aksignal.cz>
+#
+# (C) Copyright 2000-2006
+# Wolfgang Denk, DENX Software Engineering, wd at denx.de.
+#
+# SPDX-License-Identifier:	GPL-2.0+
+#
+
+include $(TOPDIR)/config.mk
+
+LIB	= $(obj)lib$(BOARD).o
+
+COBJS-y	+= cdu9g25.o
+
+SRCS	:= $(SOBJS:.o=.S) $(COBJS-y:.o=.c)
+OBJS	:= $(addprefix $(obj),$(COBJS-y))
+SOBJS	:= $(addprefix $(obj),$(SOBJS))
+
+$(LIB):	$(obj).depend $(OBJS) $(SOBJS)
+	$(call cmd_link_o_target, $(OBJS) $(SOBJS))
+
+#########################################################################
+
+# defines $(obj).depend target
+include $(SRCTREE)/rules.mk
+
+sinclude $(obj).depend
+
+#########################################################################
board/AKsignal/cdu9g25/cdu9g25.c new file +133 -0
new file mode 100644
index 0000000..7321a02
@@ -0,0 +1,202 @@
+/*
+ * (C) Copyright 2013
+ * Jiri Prchal <[email protected]>
+ * AK signal <www.aksignal.cz>
+ *
+ * SPDX-License-Identifier:	GPL-2.0+
+ */
+
+#include <common.h>
+#include <malloc.h>
+#include <asm/io.h>
+#include <asm/arch/at91sam9x5_matrix.h>
+#include <asm/arch/at91sam9_smc.h>
+#include <asm/arch/at91_common.h>
+#include <asm/arch/at91_pmc.h>
+#include <asm/arch/at91_rstc.h>
+#include <asm/arch/at91_spi.h>
+#include <asm/arch/gpio.h>
+#include <asm/arch/clk.h>
+#include <atmel_hlcdc.h>
+#include <atmel_mci.h>
+#include <spi.h>
+#include <net.h>
+#include <netdev.h>
+
+DECLARE_GLOBAL_DATA_PTR;
+
+static void cdu9g25_nand_hw_init(void)
+{
+	struct at91_smc *smc = (struct at91_smc *)ATMEL_BASE_SMC;
+	struct at91_matrix *matrix = (struct at91_matrix *)ATMEL_BASE_MATRIX;
+	struct at91_pmc *pmc = (struct at91_pmc *)ATMEL_BASE_PMC;
+	unsigned long csa;
+
+	/* Enable CS3 */
+	csa = readl(&matrix->ebicsa);
+	csa |= AT91_MATRIX_EBI_CS3A_SMC_SMARTMEDIA;
+	writel(csa, &matrix->ebicsa);
+
+	/* Configure SMC CS3 for NAND/SmartMedia */
+	writel(AT91_SMC_SETUP_NWE(1) | AT91_SMC_SETUP_NCS_WR(0) |
+		AT91_SMC_SETUP_NRD(1) | AT91_SMC_SETUP_NCS_RD(0),
+		&smc->cs[3].setup);
+	writel(AT91_SMC_PULSE_NWE(3) | AT91_SMC_PULSE_NCS_WR(5) |
+	       AT91_SMC_PULSE_NRD(4) | AT91_SMC_PULSE_NCS_RD(6),
+	       &smc->cs[3].pulse);
+	writel(AT91_SMC_CYCLE_NWE(5) | AT91_SMC_CYCLE_NRD(6),
+	       &smc->cs[3].cycle);
+	writel(AT91_SMC_MODE_RM_NRD | AT91_SMC_MODE_WM_NWE |
+		AT91_SMC_MODE_EXNW_DISABLE |
+		AT91_SMC_MODE_DBW_8 |
+		AT91_SMC_MODE_TDF_CYCLE(1),
+		&smc->cs[3].mode);
+
+	writel(1 << ATMEL_ID_PIOCD, &pmc->pcer);
+
+	/* Configure RDY/BSY */
+	at91_set_gpio_input(CONFIG_SYS_NAND_READY_PIN, 1);
+
+	/* Enable NandFlash */
+	at91_set_gpio_output(CONFIG_SYS_NAND_ENABLE_PIN, 1);
+
+	at91_set_a_periph(AT91_PIO_PORTD, 0, 1);	/* NAND OE */
+	at91_set_a_periph(AT91_PIO_PORTD, 1, 1);	/* NAND WE */
+	at91_set_a_periph(AT91_PIO_PORTD, 2, 1);	/* NAND ALE */
+	at91_set_a_periph(AT91_PIO_PORTD, 3, 1);	/* NAND CLE */
+}
+
+int spi_cs_is_valid(unsigned int bus, unsigned int cs)
+{
+	debug("spi_cs_is_valid: bus=%u cs=%u\n", bus, cs);
+	if (bus < 2 && cs < 4)
+		return 1;
+
+	return 0;
+}
+
+void spi_cs_activate(struct spi_slave *slave)
+{
+	debug("spi_cs_activate: bus=%u cs=%u\n", slave->bus, slave->cs);
+	at91_set_pio_output(AT91_PIO_PORTA, 23, 0);
+}
+
+void spi_cs_deactivate(struct spi_slave *slave)
+{
+	debug("spi_cs_deactivate: bus=%u cs=%u\n", slave->bus, slave->cs);
+	at91_set_pio_output(AT91_PIO_PORTA, 23, 1);
+}
+
+void cdu9g25_spi0_hw_init(void)
+{
+	at91_pmc_t *pmc = (at91_pmc_t *)ATMEL_BASE_PMC;
+
+	at91_set_a_periph(AT91_PIO_PORTA, 11, 0);	/* SPI0_MISO */
+	at91_set_a_periph(AT91_PIO_PORTA, 12, 0);	/* SPI0_MOSI */
+	at91_set_a_periph(AT91_PIO_PORTA, 13, 0);	/* SPI0_SPCK */
+
+	/* Enable clock */
+	writel(1 << ATMEL_ID_SPI0, &pmc->pcer);
+
+	at91_set_pio_output(AT91_PIO_PORTA, 23, 1);
+}
+
+void cdu9g25_init_enetaddr(void)
+{
+	struct spi_slave *slave;
+	uchar sernum[9];
+	int ret;
+
+	if (eth_getenv_enetaddr("ethaddr", sernum))
+		return;
+
+	slave = spi_setup_slave(0, 3, 1000000, SPI_MODE_0);
+	if (!slave) {
+		debug("Failed to set up FRAM slave\n");
+		goto err_spi_setup_slave;
+	}
+
+	ret = spi_claim_bus(slave);
+	if (ret) {
+		debug("Failed to claim SPI bus: %d\n", ret);
+		goto err_spi_claim_bus;
+	}
+
+	*sernum = 0xc3; /* opcode to read serial number */
+	ret = spi_xfer(slave, sizeof(sernum) * 8, sernum, sernum,
+		       SPI_XFER_BEGIN | SPI_XFER_END);
+	if (ret) {
+		debug("Failed to read serial#: %d\n", ret);
+		goto err_spi_xfer;
+	}
+
+	memmove(sernum + 1, sernum + 3, 6);

board/AKsignal/cdu9g25/cdu9g25.c line 143

Quoted 10 lines
board/AKsignal/cdu9g25/cdu9g25.c new file +10 -0
+	*sernum = 0x02;
+
+err_spi_xfer:
+	spi_release_bus(slave);
+err_spi_claim_bus:
+	spi_free_slave(slave);
+err_spi_setup_slave:
+	if (!is_valid_ether_addr(sernum)) {
+		eth_random_enetaddr(sernum);
+		printf("Using random MAC address %pM\n", sernum);

is that %p formating intended? Shouldn't you print the ethaddr here in hex rather than the pointer to the memory location?

Quoted 104 lines
board/AKsignal/cdu9g25/cdu9g25.c new file +59 -0
+	}
+	eth_setenv_enetaddr("ethaddr", sernum);
+}
+
+int board_eth_init(bd_t *bis)
+{
+	int rc = 0;
+
+	cdu9g25_init_enetaddr();
+	rc = macb_eth_initialize(0, (void *)ATMEL_BASE_EMAC0, 0x00);
+
+	return rc;
+}
+
+static void cdu9g25_led_init(void)
+{
+	at91_set_gpio_output(CONFIG_LED_B, 1);
+	at91_set_gpio_value(CONFIG_LED_B, 0);
+}
+
+int board_early_init_f(void)
+{
+	at91_seriald_hw_init();
+
+	return 0;
+}
+
+int board_init(void)
+{
+	gd->bd->bi_arch_number = MACH_TYPE_CDU9G25;
+
+	/* adress of boot parameters */
+	gd->bd->bi_boot_params = CONFIG_SYS_SDRAM_BASE + 0x100;
+
+	cdu9g25_led_init();
+
+	cdu9g25_nand_hw_init();
+
+	cdu9g25_spi0_hw_init();
+
+	at91_macb_hw_init();
+
+	return 0;
+}
+
+int dram_init(void)
+{
+	gd->ram_size = get_ram_size((void *)CONFIG_SYS_SDRAM_BASE,
+		CONFIG_SYS_SDRAM_SIZE);
+
+	return 0;
+}
+
+#ifdef CONFIG_RESET_PHY_R
+void reset_phy(void)
+{
+}
+#endif
+
boards.cfg +1 -0
index dbd8479..7a7d39e 100644
@@ -77,6 +77,7 @@ Active  arm         arm926ejs      -           armltd          integrator
 Active  arm         arm926ejs      armada100   Marvell         -                   aspenite                             -                                                                                                                                 Prafulla Wadaskar <[email protected]>
 Active  arm         arm926ejs      armada100   Marvell         -                   gplugd                               -                                                                                                                                 Ajay Bhargav <[email protected]>
 Active  arm         arm926ejs      at91        -               -                   afeb9260                             -                                                                                                                                 Sergey Lapin <[email protected]>
+Active  arm         arm926ejs      at91        AKsignal        cdu9g25             cdu9g25                              -                                                                                                                                 Jiri Prchal <[email protected]>
 Active  arm         arm926ejs      at91        atmel           at91sam9260ek       at91sam9260ek_dataflash_cs0          at91sam9260ek:AT91SAM9260,SYS_USE_DATAFLASH_CS0                                                                                   Stelian Pop <[email protected]>
 Active  arm         arm926ejs      at91        atmel           at91sam9260ek       at91sam9260ek_dataflash_cs1          at91sam9260ek:AT91SAM9260,SYS_USE_DATAFLASH_CS1                                                                                   Stelian Pop <[email protected]>
 Active  arm         arm926ejs      at91        atmel           at91sam9260ek       at91sam9260ek_nandflash              at91sam9260ek:AT91SAM9260,SYS_USE_NANDFLASH                                                                                       Stelian Pop <[email protected]>
include/configs/cdu9g25.h new file +27 -0
new file mode 100644
index 0000000..a4189cc
@@ -0,0 +1,210 @@
+/*
+ * (C) Copyright 2013
+ * Jiri Prchal <[email protected]>
+ * AK signal <www.aksignal.cz>
+ *
+ * SPDX-License-Identifier:	GPL-2.0+
+ */
+
+#ifndef __CONFIG_H__
+#define __CONFIG_H__
+
+#define CONFIG_AT91SAM9X5
+#include <asm/hardware.h>
+
+/* ARM asynchronous clock */
+#define CONFIG_SYS_AT91_SLOW_CLOCK	32768		/* slow clock xtal */
+#define CONFIG_SYS_AT91_MAIN_CLOCK	12000000	/* main clock xtal */
+#define CONFIG_SYS_HZ			1000
+
+/* Define actual evaluation board type from used processor type */
+#define CONFIG_CDU9G25
+#define CONFIG_AT91FAMILY
+#define CONFIG_IDENT_STRING	"_cdu9g25"
+
+/* Misc CPU related */
+#define CONFIG_CMDLINE_TAG		/* enable passing of ATAGs */
+#define CONFIG_SETUP_MEMORY_TAGS

include/configs/cdu9g25.h line 37

Quoted 10 lines
include/configs/cdu9g25.h new file +10 -0
+#define CONFIG_INITRD_TAG
+#define CONFIG_SKIP_LOWLEVEL_INIT
+#define CONFIG_BOARD_EARLY_INIT_F
+#define CONFIG_DISPLAY_CPUINFO
+
+#define CONFIG_CMD_BOOTZ
+#define CONFIG_OF_LIBFDT
+
+/* general purpose I/O */
+#define CONFIG_ATMEL_LEGACY		/* required until (g)pio is fixed */

I doubt you need this for gpio. Could you please check, if it is really required?

Quoted 19 lines
include/configs/cdu9g25.h new file +19 -0
+#define CONFIG_AT91_GPIO
+#define CONFIG_AT91_GPIO_PULLUP	1	/* keep pullups on peripheral pins */
+
+/* LEDs */
+#define	CONFIG_LED_R		AT91_PIN_PB18
+#define	CONFIG_LED_G		AT91_PIN_PB16
+#define	CONFIG_LED_Y		AT91_PIN_PB14
+#define	CONFIG_LED_B		AT91_PIN_PB12
+
+/* serial console */
+#define CONFIG_ATMEL_USART
+#define CONFIG_USART_BASE		ATMEL_BASE_DBGU
+#define CONFIG_USART_ID			ATMEL_ID_SYS
+#define CONFIG_BAUDRATE			115200
+
+#define CONFIG_BOOTDELAY	3
+#define CONFIG_RESET_TO_RETRY
+#define CONFIG_BOOT_RETRY_TIME	600
+

include/configs/cdu9g25.h line 66

Quoted 10 lines
include/configs/cdu9g25.h new file +10 -0
+#define CONFIG_HW_WATCHDOG
+#define CONFIG_AT91SAM9_WATCHDOG
+
+/* SPI */
+#define CONFIG_ATMEL_SPI
+#define CONFIG_CMD_SPI
+
+/* RAMTRON FRAM */
+#define CONFIG_CMD_SF
+#define CONFIG_ATMEL_SPI0		/* SPI used for FRAM is SPI0 */

This define (CONFIG_ATMEL_SPI0) is not required nor is it somewhere declared. It is just used in the top9000 board where this define is used to switch on/off by config usage of spi0.

Quoted 44 lines
include/configs/cdu9g25.h new file +44 -0
+#define CONFIG_SPI_FLASH		/* RAMTRON FRAM on SPI bus */
+#define CONFIG_SPI_FRAM_RAMTRON
+
+/*
+ * BOOTP options
+ */
+#define CONFIG_BOOTP_BOOTFILESIZE
+#define CONFIG_BOOTP_BOOTPATH
+#define CONFIG_BOOTP_GATEWAY
+#define CONFIG_BOOTP_HOSTNAME
+
+/*
+ * Command line configuration.
+ */
+#include <config_cmd_default.h>
+#undef CONFIG_CMD_FPGA
+#undef CONFIG_CMD_IMI
+#undef CONFIG_CMD_IMLS
+#undef CONFIG_CMD_LOADS
+
+#define CONFIG_CMD_PING
+#define CONFIG_CMD_DHCP
+#define CONFIG_CMD_NAND
+#define CONFIG_CMD_MMC
+#define CONFIG_CMD_FAT
+
+/*
+ * SDRAM: 1 bank, 128 MB
+ * Initialized before u-boot gets started.
+ */
+#define CONFIG_NR_DRAM_BANKS		1
+#define CONFIG_SYS_SDRAM_BASE		0x20000000 /* ATMEL_BASE_CS1 */
+#define CONFIG_SYS_SDRAM_SIZE		0x08000000
+#define CONFIG_SYS_TEXT_BASE		0x27f00000
+
+/*
+ * Initial stack pointer: 4k - GENERATED_GBL_DATA_SIZE in internal SRAM,
+ * leaving the correct space for initial global data structure above
+ * that address while providing maximum stack area below.
+ */
+#define CONFIG_SYS_INIT_SP_ADDR	(CONFIG_SYS_SDRAM_BASE + 4 * 1024\
+				- GENERATED_GBL_DATA_SIZE)
+
+/* no NOR flash */

include/configs/cdu9g25.h line 120

Quoted 10 lines
include/configs/cdu9g25.h new file +10 -0
+#define CONFIG_SYS_NO_FLASH
+
+/* NAND flash */
+#define CONFIG_NAND_ATMEL
+#define CONFIG_SYS_MAX_NAND_DEVICE	1
+#define CONFIG_SYS_NAND_BASE		0x40000000
+#define CONFIG_SYS_NAND_MASK_ALE	(1 << 21)
+#define CONFIG_SYS_NAND_MASK_CLE	(1 << 22)
+#define CONFIG_SYS_NAND_ENABLE_PIN	AT91_PIN_PD4
+#define CONFIG_SYS_NAND_READY_PIN	AT91_PIN_PD5

Damn .. you need the legacy stuff here. We really should convert it soon!

Quoted 10 lines
include/configs/cdu9g25.h new file +10 -0
+
+/* PMECC & PMERRLOC */
+#define CONFIG_ATMEL_NAND_HWECC
+#define CONFIG_ATMEL_NAND_HW_PMECC
+#define CONFIG_PMECC_CAP		8
+#define CONFIG_PMECC_SECTOR_SIZE	512
+
+#define CONFIG_MTD_DEVICE
+#define CONFIG_CMD_MTDPARTS
+#define CONFIG_MTD_PARTITIONS

include/configs/cdu9g25.h line 140

Quoted 10 lines
include/configs/cdu9g25.h new file +10 -0
+#define CONFIG_RBTREE
+#define CONFIG_LZO
+#define CONFIG_CMD_UBI
+#define CONFIG_CMD_UBIFS
+#define CONFIG_CMD_NAND_TRIMFFS
+#define MTDIDS_DEFAULT			"nand0=nand"
+#define MTDPARTS_DEFAULT		"mtdparts=nand:256k(bootstrap),"\
+					"768k(uboot),256k(ubootenv),"\
+					"4864k(kernel),"\
+					"-(root)"

just to mention it:
a) a secondary env is sometimes useful
b) 256k is you page size, isn't it? how about bad blocks?

Quoted 37 lines
include/configs/cdu9g25.h new file +37 -0
+
+/* MMC */
+#ifdef CONFIG_CMD_MMC
+#define CONFIG_MMC
+#define CONFIG_GENERIC_MMC
+#define CONFIG_GENERIC_ATMEL_MCI
+#endif
+
+/* FAT */
+#ifdef CONFIG_CMD_FAT
+#define CONFIG_DOS_PARTITION
+#endif
+
+/* Ethernet */
+#define CONFIG_MACB
+#define CONFIG_RMII
+#define CONFIG_NET_RETRY_COUNT		20
+#define CONFIG_MACB_SEARCH_PHY
+#define CONFIG_RANDOM_MACADDR
+
+/* USB */
+#ifdef CONFIG_CMD_USB
+#ifdef CONFIG_USB_EHCI
+#define CONFIG_USB_EHCI_ATMEL
+#define CONFIG_SYS_USB_EHCI_MAX_ROOT_PORTS	2
+#else
+#define CONFIG_USB_OHCI_NEW
+#define CONFIG_SYS_USB_OHCI_CPU_INIT
+#define CONFIG_SYS_USB_OHCI_REGS_BASE		ATMEL_BASE_OHCI
+#define CONFIG_SYS_USB_OHCI_SLOT_NAME		"at91sam9x5"
+#define CONFIG_SYS_USB_OHCI_MAX_ROOT_PORTS	3
+#endif
+#define CONFIG_USB_ATMEL
+#define CONFIG_USB_STORAGE
+#endif
+
+#define CONFIG_SYS_LOAD_ADDR		0x21000000 /* load address */

include/configs/cdu9g25.h line 187

Quoted 10 lines
include/configs/cdu9g25.h new file +10 -0
+
+#define CONFIG_SYS_MEMTEST_START	CONFIG_SYS_SDRAM_BASE
+#define CONFIG_SYS_MEMTEST_END		CONFIG_SYS_TEXT_BASE
+
+/* CONFIG_SYS_USE_NANDFLASH */
+/* bootstrap + u-boot + env + linux in nandflash */
+#define CONFIG_ENV_IS_IN_NAND
+#define CONFIG_ENV_OFFSET	0x100000
+#define CONFIG_ENV_SIZE		0x20000 /* 1 sector = 128kB */
+#define CONFIG_ENV_RANGE	0x40000 /* 2 sectors for bad sector handling */

ah, got it. so you have a chance to have one of the sectors working.

include/configs/cdu9g25.h line 193

Quoted 6 lines
include/configs/cdu9g25.h new file +6 -0
+#define CONFIG_BOOTCOMMAND	"nand read 21000000 kernel; bootm"
+#define CONFIG_BOOTARGS		"console=ttyS0,115200 ubi.mtd=root "\
+				"root=ubi0:root rootfstype=ubifs rw"
+#define CONFIG_SERVERIP		10.0.1.1
+#define CONFIG_BOOTFILE		"kernel_cdu9g25"
+#define CONFIG_PREBOOT		"mtdparts default" /* for partitions */

Would you really like to always reset a new mtd partitioning on every boot to the default one?

Quoted 5 lines
include/configs/cdu9g25.h new file +5 -0
+
+#define CONFIG_SYS_PROMPT	"U-Boot> "
+#define CONFIG_SYS_CBSIZE	256
+#define CONFIG_SYS_MAXARGS	16
+#define CONFIG_SYS_PBSIZE	(CONFIG_SYS_CBSIZE + sizeof(CONFIG_SYS_PROMPT)\

include/configs/cdu9g25.h line 208

Quoted 10 lines
include/configs/cdu9g25.h new file +10 -0
+				+ 16)
+#define CONFIG_SYS_LONGHELP
+#define CONFIG_CMDLINE_EDITING
+#define CONFIG_AUTO_COMPLETE
+#define CONFIG_SYS_HUSH_PARSER
+
+/*
+ * Size of malloc() pool
+ */
+#define CONFIG_SYS_MALLOC_LEN		(512 * 1024 + CONFIG_ENV_SIZE)

512 KiB may be too small for recovering UBI partitions ... 1-4 MiB is more likely.

Quoted 5 lines
include/configs/cdu9g25.h new file +2 -1
+
+#endif
--

1.7.9.5

Best regards

Andreas Bie?mann

#2 lore raw

PATCH v5 1/1 at91: add support for CDU9G25 board

Hi Andreas,

Dne 13.9.2013 16:10, Andreas Bie?mann napsal(a):

Quoted 32 lines

Dear Jiri Prchal,

Quoted 29 lines · Jiri Prchal

This patch adds support for our companies board CDU9G25 with Atmel AT91SAM9G25, 128MB DDR2, 256MB NAND.

---
v.2
Fixed static IP and MAC addr cofiguration by random MAC and DHCP. Added entry to MAINTAINERS.
Fixed line lenght, trailing spaces and other cosmetics. v.3
Removed static eth_gadget addresses.
MAC address set from serial number of F-RAM.
v.4
Changed debug to printf if used random MAC addr, print that addr. Keep MAC addr set in enviroment.
v.5
Changes to Bo Shen comments.

---
  arch/arm/include/asm/mach-types.h |    1 +
  board/AKsignal/cdu9g25/Makefile   |   34 ++++++
  board/AKsignal/cdu9g25/cdu9g25.c  |  202 +++++++++++++++++++++++++++++++++++
  boards.cfg                        |    1 +
  include/configs/cdu9g25.h         |  210 +++++++++++++++++++++++++++++++++++++
  5 files changed, 448 insertions(+)

create mode 100644 board/AKsignal/cdu9g25/Makefile create mode 100644 board/AKsignal/cdu9g25/cdu9g25.c create mode 100644 include/configs/cdu9g25.h

Quoted 12 lines

arch/arm/include/asm/mach-types.h extern unsigned int __machine_arch_type; · line 989

Quoted 9 lines
arch/arm/include/asm/mach-types.h +1 -0
index 440b041..9b274ba 100644
@@ -986,6 +986,7 @@ extern unsigned int __machine_arch_type;
  #define MACH_TYPE_VIT_IBOX             3371
  #define MACH_TYPE_DM6441_ESP           3372
  #define MACH_TYPE_AT91SAM9X5EK         3373
+#define MACH_TYPE_CDU9G25              3373

NAK, please obtain a mach type:
http://www.arm.linux.org.uk/developer/machines/?action=new

Should I register machine? I develop DT only:
"NOTE 1:If you are developing a DT-only platform, you do not need to register a machine type for it. Please do not register a machine type. Thanks."
Do I need this MACH_TYPE_* at all?

Quoted 188 lines
Quoted 188 lines
arch/arm/include/asm/mach-types.h +0 -0
  #define MACH_TYPE_LIBRA                3374
  #define MACH_TYPE_EASYCRRH             3375
  #define MACH_TYPE_TRIPEL               3376
board/AKsignal/cdu9g25/Makefile new file +34 -0
new file mode 100644
index 0000000..b644a83
@@ -0,0 +1,34 @@
+#
+# U-boot - Makefile
+#
+# (C) Copyright 2013
+# Jiri Prchal <[email protected]>
+# AK signal <www.aksignal.cz>
+#
+# (C) Copyright 2000-2006
+# Wolfgang Denk, DENX Software Engineering, wd at denx.de.
+#
+# SPDX-License-Identifier:	GPL-2.0+
+#
+
+include $(TOPDIR)/config.mk
+
+LIB	= $(obj)lib$(BOARD).o
+
+COBJS-y	+= cdu9g25.o
+
+SRCS	:= $(SOBJS:.o=.S) $(COBJS-y:.o=.c)
+OBJS	:= $(addprefix $(obj),$(COBJS-y))
+SOBJS	:= $(addprefix $(obj),$(SOBJS))
+
+$(LIB):	$(obj).depend $(OBJS) $(SOBJS)
+	$(call cmd_link_o_target, $(OBJS) $(SOBJS))
+
+#########################################################################
+
+# defines $(obj).depend target
+include $(SRCTREE)/rules.mk
+
+sinclude $(obj).depend
+
+#########################################################################
board/AKsignal/cdu9g25/cdu9g25.c new file +138 -0
new file mode 100644
index 0000000..7321a02
@@ -0,0 +1,202 @@
+/*
+ * (C) Copyright 2013
+ * Jiri Prchal <[email protected]>
+ * AK signal <www.aksignal.cz>
+ *
+ * SPDX-License-Identifier:	GPL-2.0+
+ */
+
+#include <common.h>
+#include <malloc.h>
+#include <asm/io.h>
+#include <asm/arch/at91sam9x5_matrix.h>
+#include <asm/arch/at91sam9_smc.h>
+#include <asm/arch/at91_common.h>
+#include <asm/arch/at91_pmc.h>
+#include <asm/arch/at91_rstc.h>
+#include <asm/arch/at91_spi.h>
+#include <asm/arch/gpio.h>
+#include <asm/arch/clk.h>
+#include <atmel_hlcdc.h>
+#include <atmel_mci.h>
+#include <spi.h>
+#include <net.h>
+#include <netdev.h>
+
+DECLARE_GLOBAL_DATA_PTR;
+
+static void cdu9g25_nand_hw_init(void)
+{
+	struct at91_smc *smc = (struct at91_smc *)ATMEL_BASE_SMC;
+	struct at91_matrix *matrix = (struct at91_matrix *)ATMEL_BASE_MATRIX;
+	struct at91_pmc *pmc = (struct at91_pmc *)ATMEL_BASE_PMC;
+	unsigned long csa;
+
+	/* Enable CS3 */
+	csa = readl(&matrix->ebicsa);
+	csa |= AT91_MATRIX_EBI_CS3A_SMC_SMARTMEDIA;
+	writel(csa, &matrix->ebicsa);
+
+	/* Configure SMC CS3 for NAND/SmartMedia */
+	writel(AT91_SMC_SETUP_NWE(1) | AT91_SMC_SETUP_NCS_WR(0) |
+		AT91_SMC_SETUP_NRD(1) | AT91_SMC_SETUP_NCS_RD(0),
+		&smc->cs[3].setup);
+	writel(AT91_SMC_PULSE_NWE(3) | AT91_SMC_PULSE_NCS_WR(5) |
+	       AT91_SMC_PULSE_NRD(4) | AT91_SMC_PULSE_NCS_RD(6),
+	       &smc->cs[3].pulse);
+	writel(AT91_SMC_CYCLE_NWE(5) | AT91_SMC_CYCLE_NRD(6),
+	       &smc->cs[3].cycle);
+	writel(AT91_SMC_MODE_RM_NRD | AT91_SMC_MODE_WM_NWE |
+		AT91_SMC_MODE_EXNW_DISABLE |
+		AT91_SMC_MODE_DBW_8 |
+		AT91_SMC_MODE_TDF_CYCLE(1),
+		&smc->cs[3].mode);
+
+	writel(1 << ATMEL_ID_PIOCD, &pmc->pcer);
+
+	/* Configure RDY/BSY */
+	at91_set_gpio_input(CONFIG_SYS_NAND_READY_PIN, 1);
+
+	/* Enable NandFlash */
+	at91_set_gpio_output(CONFIG_SYS_NAND_ENABLE_PIN, 1);
+
+	at91_set_a_periph(AT91_PIO_PORTD, 0, 1);	/* NAND OE */
+	at91_set_a_periph(AT91_PIO_PORTD, 1, 1);	/* NAND WE */
+	at91_set_a_periph(AT91_PIO_PORTD, 2, 1);	/* NAND ALE */
+	at91_set_a_periph(AT91_PIO_PORTD, 3, 1);	/* NAND CLE */
+}
+
+int spi_cs_is_valid(unsigned int bus, unsigned int cs)
+{
+	debug("spi_cs_is_valid: bus=%u cs=%u\n", bus, cs);
+	if (bus < 2 && cs < 4)
+		return 1;
+
+	return 0;
+}
+
+void spi_cs_activate(struct spi_slave *slave)
+{
+	debug("spi_cs_activate: bus=%u cs=%u\n", slave->bus, slave->cs);
+	at91_set_pio_output(AT91_PIO_PORTA, 23, 0);
+}
+
+void spi_cs_deactivate(struct spi_slave *slave)
+{
+	debug("spi_cs_deactivate: bus=%u cs=%u\n", slave->bus, slave->cs);
+	at91_set_pio_output(AT91_PIO_PORTA, 23, 1);
+}
+
+void cdu9g25_spi0_hw_init(void)
+{
+	at91_pmc_t *pmc = (at91_pmc_t *)ATMEL_BASE_PMC;
+
+	at91_set_a_periph(AT91_PIO_PORTA, 11, 0);	/* SPI0_MISO */
+	at91_set_a_periph(AT91_PIO_PORTA, 12, 0);	/* SPI0_MOSI */
+	at91_set_a_periph(AT91_PIO_PORTA, 13, 0);	/* SPI0_SPCK */
+
+	/* Enable clock */
+	writel(1 << ATMEL_ID_SPI0, &pmc->pcer);
+
+	at91_set_pio_output(AT91_PIO_PORTA, 23, 1);
+}
+
+void cdu9g25_init_enetaddr(void)
+{
+	struct spi_slave *slave;
+	uchar sernum[9];
+	int ret;
+
+	if (eth_getenv_enetaddr("ethaddr", sernum))
+		return;
+
+	slave = spi_setup_slave(0, 3, 1000000, SPI_MODE_0);
+	if (!slave) {
+		debug("Failed to set up FRAM slave\n");
+		goto err_spi_setup_slave;
+	}
+
+	ret = spi_claim_bus(slave);
+	if (ret) {
+		debug("Failed to claim SPI bus: %d\n", ret);
+		goto err_spi_claim_bus;
+	}
+
+	*sernum = 0xc3; /* opcode to read serial number */
+	ret = spi_xfer(slave, sizeof(sernum) * 8, sernum, sernum,
+		       SPI_XFER_BEGIN | SPI_XFER_END);
+	if (ret) {
+		debug("Failed to read serial#: %d\n", ret);
+		goto err_spi_xfer;
+	}
+
+	memmove(sernum + 1, sernum + 3, 6);
+	*sernum = 0x02;
+
+err_spi_xfer:
+	spi_release_bus(slave);
+err_spi_claim_bus:
Quoted 8 lines

board/AKsignal/cdu9g25/cdu9g25.c line 143

Quoted 5 lines
board/AKsignal/cdu9g25/cdu9g25.c new file +5 -0
+	spi_free_slave(slave);
+err_spi_setup_slave:
+	if (!is_valid_ether_addr(sernum)) {
+		eth_random_enetaddr(sernum);
+		printf("Using random MAC address %pM\n", sernum);

is that %p formating intended? Shouldn't you print the ethaddr here in hex rather than the pointer to the memory location?

No, this is no pointer modifier, it's together %pM and it prints ethaddr like this: 02:22:23:15:86:a5.

Quoted 110 lines
Quoted 110 lines
board/AKsignal/cdu9g25/cdu9g25.c new file +59 -0
+	}
+	eth_setenv_enetaddr("ethaddr", sernum);
+}
+
+int board_eth_init(bd_t *bis)
+{
+	int rc = 0;
+
+	cdu9g25_init_enetaddr();
+	rc = macb_eth_initialize(0, (void *)ATMEL_BASE_EMAC0, 0x00);
+
+	return rc;
+}
+
+static void cdu9g25_led_init(void)
+{
+	at91_set_gpio_output(CONFIG_LED_B, 1);
+	at91_set_gpio_value(CONFIG_LED_B, 0);
+}
+
+int board_early_init_f(void)
+{
+	at91_seriald_hw_init();
+
+	return 0;
+}
+
+int board_init(void)
+{
+	gd->bd->bi_arch_number = MACH_TYPE_CDU9G25;
+
+	/* adress of boot parameters */
+	gd->bd->bi_boot_params = CONFIG_SYS_SDRAM_BASE + 0x100;
+
+	cdu9g25_led_init();
+
+	cdu9g25_nand_hw_init();
+
+	cdu9g25_spi0_hw_init();
+
+	at91_macb_hw_init();
+
+	return 0;
+}
+
+int dram_init(void)
+{
+	gd->ram_size = get_ram_size((void *)CONFIG_SYS_SDRAM_BASE,
+		CONFIG_SYS_SDRAM_SIZE);
+
+	return 0;
+}
+
+#ifdef CONFIG_RESET_PHY_R
+void reset_phy(void)
+{
+}
+#endif
+
boards.cfg +1 -0
index dbd8479..7a7d39e 100644
@@ -77,6 +77,7 @@ Active  arm         arm926ejs      -           armltd          integrator
  Active  arm         arm926ejs      armada100   Marvell         -                   aspenite                             -                                                                                                                                 Prafulla Wadaskar <[email protected]>
  Active  arm         arm926ejs      armada100   Marvell         -                   gplugd                               -                                                                                                                                 Ajay Bhargav <[email protected]>
  Active  arm         arm926ejs      at91        -               -                   afeb9260                             -                                                                                                                                 Sergey Lapin <[email protected]>
+Active  arm         arm926ejs      at91        AKsignal        cdu9g25             cdu9g25                              -                                                                                                                                 Jiri Prchal <[email protected]>
  Active  arm         arm926ejs      at91        atmel           at91sam9260ek       at91sam9260ek_dataflash_cs0          at91sam9260ek:AT91SAM9260,SYS_USE_DATAFLASH_CS0                                                                                   Stelian Pop <[email protected]>
  Active  arm         arm926ejs      at91        atmel           at91sam9260ek       at91sam9260ek_dataflash_cs1          at91sam9260ek:AT91SAM9260,SYS_USE_DATAFLASH_CS1                                                                                   Stelian Pop <[email protected]>
  Active  arm         arm926ejs      at91        atmel           at91sam9260ek       at91sam9260ek_nandflash              at91sam9260ek:AT91SAM9260,SYS_USE_NANDFLASH                                                                                       Stelian Pop <[email protected]>
include/configs/cdu9g25.h new file +32 -0
new file mode 100644
index 0000000..a4189cc
@@ -0,0 +1,210 @@
+/*
+ * (C) Copyright 2013
+ * Jiri Prchal <[email protected]>
+ * AK signal <www.aksignal.cz>
+ *
+ * SPDX-License-Identifier:	GPL-2.0+
+ */
+
+#ifndef __CONFIG_H__
+#define __CONFIG_H__
+
+#define CONFIG_AT91SAM9X5
+#include <asm/hardware.h>
+
+/* ARM asynchronous clock */
+#define CONFIG_SYS_AT91_SLOW_CLOCK	32768		/* slow clock xtal */
+#define CONFIG_SYS_AT91_MAIN_CLOCK	12000000	/* main clock xtal */
+#define CONFIG_SYS_HZ			1000
+
+/* Define actual evaluation board type from used processor type */
+#define CONFIG_CDU9G25
+#define CONFIG_AT91FAMILY
+#define CONFIG_IDENT_STRING	"_cdu9g25"
+
+/* Misc CPU related */
+#define CONFIG_CMDLINE_TAG		/* enable passing of ATAGs */
+#define CONFIG_SETUP_MEMORY_TAGS
+#define CONFIG_INITRD_TAG
+#define CONFIG_SKIP_LOWLEVEL_INIT
+#define CONFIG_BOARD_EARLY_INIT_F
+#define CONFIG_DISPLAY_CPUINFO
+
Quoted 8 lines

include/configs/cdu9g25.h line 37

Quoted 5 lines
include/configs/cdu9g25.h new file +5 -0
+#define CONFIG_CMD_BOOTZ
+#define CONFIG_OF_LIBFDT
+
+/* general purpose I/O */
+#define CONFIG_ATMEL_LEGACY		/* required until (g)pio is fixed */

I doubt you need this for gpio. Could you please check, if it is really required?

Yes, I knew that, but I have looked many other board files and they use both GPIO and PIO in one file. If is necessary I'll re-base it to PIO.

Quoted 107 lines
Quoted 20 lines
include/configs/cdu9g25.h new file +19 -0
+#define CONFIG_AT91_GPIO
+#define CONFIG_AT91_GPIO_PULLUP	1	/* keep pullups on peripheral pins */
+
+/* LEDs */
+#define	CONFIG_LED_R		AT91_PIN_PB18
+#define	CONFIG_LED_G		AT91_PIN_PB16
+#define	CONFIG_LED_Y		AT91_PIN_PB14
+#define	CONFIG_LED_B		AT91_PIN_PB12
+
+/* serial console */
+#define CONFIG_ATMEL_USART
+#define CONFIG_USART_BASE		ATMEL_BASE_DBGU
+#define CONFIG_USART_ID			ATMEL_ID_SYS
+#define CONFIG_BAUDRATE			115200
+
+#define CONFIG_BOOTDELAY	3
+#define CONFIG_RESET_TO_RETRY
+#define CONFIG_BOOT_RETRY_TIME	600
+

include/configs/cdu9g25.h line 66

Quoted 10 lines
include/configs/cdu9g25.h new file +10 -0
+#define CONFIG_HW_WATCHDOG
+#define CONFIG_AT91SAM9_WATCHDOG
+
+/* SPI */
+#define CONFIG_ATMEL_SPI
+#define CONFIG_CMD_SPI
+
+/* RAMTRON FRAM */
+#define CONFIG_CMD_SF
+#define CONFIG_ATMEL_SPI0		/* SPI used for FRAM is SPI0 */

This define (CONFIG_ATMEL_SPI0) is not required nor is it somewhere declared. It is just used in the top9000 board where this define is used to switch on/off by config usage of spi0.

Quoted 44 lines
include/configs/cdu9g25.h new file +44 -0
+#define CONFIG_SPI_FLASH		/* RAMTRON FRAM on SPI bus */
+#define CONFIG_SPI_FRAM_RAMTRON
+
+/*
+ * BOOTP options
+ */
+#define CONFIG_BOOTP_BOOTFILESIZE
+#define CONFIG_BOOTP_BOOTPATH
+#define CONFIG_BOOTP_GATEWAY
+#define CONFIG_BOOTP_HOSTNAME
+
+/*
+ * Command line configuration.
+ */
+#include <config_cmd_default.h>
+#undef CONFIG_CMD_FPGA
+#undef CONFIG_CMD_IMI
+#undef CONFIG_CMD_IMLS
+#undef CONFIG_CMD_LOADS
+
+#define CONFIG_CMD_PING
+#define CONFIG_CMD_DHCP
+#define CONFIG_CMD_NAND
+#define CONFIG_CMD_MMC
+#define CONFIG_CMD_FAT
+
+/*
+ * SDRAM: 1 bank, 128 MB
+ * Initialized before u-boot gets started.
+ */
+#define CONFIG_NR_DRAM_BANKS		1
+#define CONFIG_SYS_SDRAM_BASE		0x20000000 /* ATMEL_BASE_CS1 */
+#define CONFIG_SYS_SDRAM_SIZE		0x08000000
+#define CONFIG_SYS_TEXT_BASE		0x27f00000
+
+/*
+ * Initial stack pointer: 4k - GENERATED_GBL_DATA_SIZE in internal SRAM,
+ * leaving the correct space for initial global data structure above
+ * that address while providing maximum stack area below.
+ */
+#define CONFIG_SYS_INIT_SP_ADDR	(CONFIG_SYS_SDRAM_BASE + 4 * 1024\
+				- GENERATED_GBL_DATA_SIZE)
+
+/* no NOR flash */

include/configs/cdu9g25.h line 120

Quoted 10 lines
include/configs/cdu9g25.h new file +10 -0
+#define CONFIG_SYS_NO_FLASH
+
+/* NAND flash */
+#define CONFIG_NAND_ATMEL
+#define CONFIG_SYS_MAX_NAND_DEVICE	1
+#define CONFIG_SYS_NAND_BASE		0x40000000
+#define CONFIG_SYS_NAND_MASK_ALE	(1 << 21)
+#define CONFIG_SYS_NAND_MASK_CLE	(1 << 22)
+#define CONFIG_SYS_NAND_ENABLE_PIN	AT91_PIN_PD4
+#define CONFIG_SYS_NAND_READY_PIN	AT91_PIN_PD5

Damn .. you need the legacy stuff here. We really should convert it soon!

Quoted 15 lines
include/configs/cdu9g25.h new file +15 -0
+
+/* PMECC & PMERRLOC */
+#define CONFIG_ATMEL_NAND_HWECC
+#define CONFIG_ATMEL_NAND_HW_PMECC
+#define CONFIG_PMECC_CAP		8
+#define CONFIG_PMECC_SECTOR_SIZE	512
+
+#define CONFIG_MTD_DEVICE
+#define CONFIG_CMD_MTDPARTS
+#define CONFIG_MTD_PARTITIONS
+#define CONFIG_RBTREE
+#define CONFIG_LZO
+#define CONFIG_CMD_UBI
+#define CONFIG_CMD_UBIFS
+#define CONFIG_CMD_NAND_TRIMFFS
Quoted 8 lines

include/configs/cdu9g25.h line 140

Quoted 5 lines
include/configs/cdu9g25.h new file +5 -0
+#define MTDIDS_DEFAULT			"nand0=nand"
+#define MTDPARTS_DEFAULT		"mtdparts=nand:256k(bootstrap),"\
+					"768k(uboot),256k(ubootenv),"\
+					"4864k(kernel),"\
+					"-(root)"

just to mention it:
a) a secondary env is sometimes useful

Is that true? I thought use one copy env, if not then use default.

Quoted 52 lines

b) 256k is you page size, isn't it? how about bad blocks?

Quoted 37 lines
include/configs/cdu9g25.h new file +37 -0
+
+/* MMC */
+#ifdef CONFIG_CMD_MMC
+#define CONFIG_MMC
+#define CONFIG_GENERIC_MMC
+#define CONFIG_GENERIC_ATMEL_MCI
+#endif
+
+/* FAT */
+#ifdef CONFIG_CMD_FAT
+#define CONFIG_DOS_PARTITION
+#endif
+
+/* Ethernet */
+#define CONFIG_MACB
+#define CONFIG_RMII
+#define CONFIG_NET_RETRY_COUNT		20
+#define CONFIG_MACB_SEARCH_PHY
+#define CONFIG_RANDOM_MACADDR
+
+/* USB */
+#ifdef CONFIG_CMD_USB
+#ifdef CONFIG_USB_EHCI
+#define CONFIG_USB_EHCI_ATMEL
+#define CONFIG_SYS_USB_EHCI_MAX_ROOT_PORTS	2
+#else
+#define CONFIG_USB_OHCI_NEW
+#define CONFIG_SYS_USB_OHCI_CPU_INIT
+#define CONFIG_SYS_USB_OHCI_REGS_BASE		ATMEL_BASE_OHCI
+#define CONFIG_SYS_USB_OHCI_SLOT_NAME		"at91sam9x5"
+#define CONFIG_SYS_USB_OHCI_MAX_ROOT_PORTS	3
+#endif
+#define CONFIG_USB_ATMEL
+#define CONFIG_USB_STORAGE
+#endif
+
+#define CONFIG_SYS_LOAD_ADDR		0x21000000 /* load address */

include/configs/cdu9g25.h line 187

Quoted 10 lines
include/configs/cdu9g25.h new file +10 -0
+
+#define CONFIG_SYS_MEMTEST_START	CONFIG_SYS_SDRAM_BASE
+#define CONFIG_SYS_MEMTEST_END		CONFIG_SYS_TEXT_BASE
+
+/* CONFIG_SYS_USE_NANDFLASH */
+/* bootstrap + u-boot + env + linux in nandflash */
+#define CONFIG_ENV_IS_IN_NAND
+#define CONFIG_ENV_OFFSET	0x100000
+#define CONFIG_ENV_SIZE		0x20000 /* 1 sector = 128kB */
+#define CONFIG_ENV_RANGE	0x40000 /* 2 sectors for bad sector handling */

ah, got it. so you have a chance to have one of the sectors working.

Quoted 9 lines

include/configs/cdu9g25.h line 193

Quoted 6 lines
include/configs/cdu9g25.h new file +6 -0
+#define CONFIG_BOOTCOMMAND	"nand read 21000000 kernel; bootm"
+#define CONFIG_BOOTARGS		"console=ttyS0,115200 ubi.mtd=root "\
+				"root=ubi0:root rootfstype=ubifs rw"
+#define CONFIG_SERVERIP		10.0.1.1
+#define CONFIG_BOOTFILE		"kernel_cdu9g25"
+#define CONFIG_PREBOOT		"mtdparts default" /* for partitions */

Would you really like to always reset a new mtd partitioning on every boot to the default one?

Does it mean that I don't need run this command if I like use partitions in nand commands?

Quoted 28 lines
Quoted 6 lines
include/configs/cdu9g25.h new file +5 -0
+
+#define CONFIG_SYS_PROMPT	"U-Boot> "
+#define CONFIG_SYS_CBSIZE	256
+#define CONFIG_SYS_MAXARGS	16
+#define CONFIG_SYS_PBSIZE	(CONFIG_SYS_CBSIZE + sizeof(CONFIG_SYS_PROMPT)\

include/configs/cdu9g25.h line 208

Quoted 10 lines
include/configs/cdu9g25.h new file +10 -0
+				+ 16)
+#define CONFIG_SYS_LONGHELP
+#define CONFIG_CMDLINE_EDITING
+#define CONFIG_AUTO_COMPLETE
+#define CONFIG_SYS_HUSH_PARSER
+
+/*
+ * Size of malloc() pool
+ */
+#define CONFIG_SYS_MALLOC_LEN		(512 * 1024 + CONFIG_ENV_SIZE)

512 KiB may be too small for recovering UBI partitions ... 1-4 MiB is more likely.

Quoted 5 lines
include/configs/cdu9g25.h new file +2 -1
+
+#endif
--

1.7.9.5

Best regards

Quoted 3 lines

Andreas Bie?mann

With many thanks
Jiri

#3 lore raw

PATCH v5 1/1 at91: add support for CDU9G25 board

Dear Ji?? Prchal,

Quoted 2 lines · Ji?? Prchal

Dne 13.9.2013 16:10, Andreas Bie?mann napsal(a):

Quoted 1 line · Jiri Prchal

<snip>

Quoted 14 lines
Quoted 5 lines
Quoted 5 lines
arch/arm/include/asm/mach-types.h +0 -0
index 440b041..9b274ba 100644
Quoted 8 lines

arch/arm/include/asm/mach-types.h extern unsigned int __machine_arch_type; · line 989

Quoted 5 lines
arch/arm/include/asm/mach-types.h +1 -0
@@ -986,6 +986,7 @@ extern unsigned int __machine_arch_type;
  #define MACH_TYPE_VIT_IBOX             3371
  #define MACH_TYPE_DM6441_ESP           3372
  #define MACH_TYPE_AT91SAM9X5EK         3373
+#define MACH_TYPE_CDU9G25              3373

NAK, please obtain a mach type:
http://www.arm.linux.org.uk/developer/machines/?action=new

Quoted 5 lines

Should I register machine? I develop DT only:
"NOTE 1:If you are developing a DT-only platform, you do not need to register a machine type for it.
Please do not register a machine type. Thanks."
Do I need this MACH_TYPE_* at all?

I'm not really familiar with FDT only boards, but I think it is sane to just use a zero machid then. Could you please check this and if working just omit the machid setting?

<snip>

Quoted 12 lines
Quoted 11 lines

arch/arm/include/asm/mach-types.h extern unsigned int __machine_arch_type;

Quoted 8 lines
arch/arm/include/asm/mach-types.h +8 -0
+err_spi_xfer:
+    spi_release_bus(slave);
+err_spi_claim_bus:
+    spi_free_slave(slave);
+err_spi_setup_slave:
+    if (!is_valid_ether_addr(sernum)) {
+        eth_random_enetaddr(sernum);
+        printf("Using random MAC address %pM\n", sernum);

is that %p formating intended? Shouldn't you print the ethaddr here in hex rather than the pointer to the memory location?

Quoted 2 lines

No, this is no pointer modifier, it's together %pM and it prints ethaddr like this: 02:22:23:15:86:a5.

Sorry, my fault.

<snip>

Quoted 9 lines
Quoted 8 lines

arch/arm/include/asm/mach-types.h

Quoted 5 lines
arch/arm/include/asm/mach-types.h +5 -0
+#define CONFIG_CMD_BOOTZ
+#define CONFIG_OF_LIBFDT
+
+/* general purpose I/O */
+#define CONFIG_ATMEL_LEGACY        /* required until (g)pio is fixed */

I doubt you need this for gpio. Could you please check, if it is really required?

Quoted 3 lines

Yes, I knew that, but I have looked many other board files and they use both GPIO and PIO in one file.
If is necessary I'll re-base it to PIO.

No, nand is not yet prepared for the generic gpio framework. I'll try to post patches next weeks (for 2014.01), but I'm quite busy right now. Since your board will also end up in 2014.01 this should go together. If you have the time I would be grateful if you could prepare patches for that.

<snip>

Quoted 17 lines
Quoted 8 lines
Quoted 8 lines
arch/arm/include/asm/mach-types.h +8 -0
+#define CONFIG_MTD_DEVICE
+#define CONFIG_CMD_MTDPARTS
+#define CONFIG_MTD_PARTITIONS
+#define CONFIG_RBTREE
+#define CONFIG_LZO
+#define CONFIG_CMD_UBI
+#define CONFIG_CMD_UBIFS
+#define CONFIG_CMD_NAND_TRIMFFS
Quoted 8 lines

arch/arm/include/asm/mach-types.h

Quoted 5 lines
arch/arm/include/asm/mach-types.h +5 -0
+#define MTDIDS_DEFAULT            "nand0=nand"
+#define MTDPARTS_DEFAULT        "mtdparts=nand:256k(bootstrap),"\
+                    "768k(uboot),256k(ubootenv),"\
+                    "4864k(kernel),"\
+                    "-(root)"

just to mention it:
a) a secondary env is sometimes useful

Quoted 1 line

Is that true? I thought use one copy env, if not then use default.

Well, your point is true, but how about heavily changed env in the field? If something goes wrong (two bad blocks at worst place, wrong crc due to too much bit errors, ...) it will fall back to the compiled in env, which may be completely wrong then.
But it is also true that an enabled secondary env will be read on every startup and compared to the first env. Also you always need to write two locations. It depends on your use case, but for me it is careless to have just one env on NAND.

<snip>

Quoted 11 lines
Quoted 9 lines

arch/arm/include/asm/mach-types.h

Quoted 6 lines
arch/arm/include/asm/mach-types.h +6 -0
+#define CONFIG_BOOTCOMMAND    "nand read 21000000 kernel; bootm"
+#define CONFIG_BOOTARGS        "console=ttyS0,115200 ubi.mtd=root "\
+                "root=ubi0:root rootfstype=ubifs rw"
+#define CONFIG_SERVERIP        10.0.1.1
+#define CONFIG_BOOTFILE        "kernel_cdu9g25"
+#define CONFIG_PREBOOT        "mtdparts default" /* for partitions */

Would you really like to always reset a new mtd partitioning on every boot to the default one?

Does it mean that I don't need run this command if I like use partitions in nand commands?

No, not fully. You'll need the environment parameter 'mtdparts' with contend 'mdtparts=...' and respective mtdids env. This could be set by the command 'mtdparts default', it will set the compiled in values of MTDIDS_DEFAULT and MTDPARTS_DEFAULT.
If you like to change the nand partitioning in any way you need to change the mtdparts env. But when you call 'mtdparts default' afterwards you will reset this to the default value ...
To avoid this we have in our default environment following setting:

---8<---
#define CONFIG_EXTRA_ENV_SETTINGS \
"mtdparts=" MTDPARTS_DEFAULT "\0" \
"mtdids=" MTDIDS_DEFAULT "\0" \
...
--->8---

Additionally we avoid to use the 'mtdparts default' command. With this setup one could change the mtdparts environment any time and it would take effect.

(Just my 2?, no need to do it that way)

Best regards

Andreas Bie?mann

#4 lore raw

PATCH v5 1/1 at91: add support for CDU9G25 board

Dear Andreas,

Dne 16.9.2013 10:27, Andreas Bie?mann napsal(a):

Quoted 28 lines

Dear Ji?? Prchal,

Quoted 2 lines · Ji?? Prchal

Dne 13.9.2013 16:10, Andreas Bie?mann napsal(a):

Quoted 1 line · Jiri Prchal

<snip>

Quoted 14 lines
Quoted 5 lines
Quoted 5 lines
arch/arm/include/asm/mach-types.h +0 -0
index 440b041..9b274ba 100644
Quoted 8 lines

arch/arm/include/asm/mach-types.h extern unsigned int __machine_arch_type; · line 989

Quoted 5 lines
arch/arm/include/asm/mach-types.h +1 -0
@@ -986,6 +986,7 @@ extern unsigned int __machine_arch_type;
   #define MACH_TYPE_VIT_IBOX             3371
   #define MACH_TYPE_DM6441_ESP           3372
   #define MACH_TYPE_AT91SAM9X5EK         3373
+#define MACH_TYPE_CDU9G25              3373

NAK, please obtain a mach type:
http://www.arm.linux.org.uk/developer/machines/?action=new

Quoted 5 lines

Should I register machine? I develop DT only:
"NOTE 1:If you are developing a DT-only platform, you do not need to register a machine type for it.
Please do not register a machine type. Thanks."
Do I need this MACH_TYPE_* at all?

Quoted 3 lines

I'm not really familiar with FDT only boards, but I think it is sane to just use a zero machid then. Could you please check this and if working just omit the machid setting?

I'm not really familiar with FDT too, but I tested with 0 and it works.

Quoted 12 lines
Quoted 12 lines
Quoted 11 lines

arch/arm/include/asm/mach-types.h extern unsigned int __machine_arch_type;

Quoted 8 lines
arch/arm/include/asm/mach-types.h +8 -0
+err_spi_xfer:
+    spi_release_bus(slave);
+err_spi_claim_bus:
+    spi_free_slave(slave);
+err_spi_setup_slave:
+    if (!is_valid_ether_addr(sernum)) {
+        eth_random_enetaddr(sernum);
+        printf("Using random MAC address %pM\n", sernum);

is that %p formating intended? Shouldn't you print the ethaddr here in hex rather than the pointer to the memory location?

Quoted 15 lines
Quoted 2 lines

No, this is no pointer modifier, it's together %pM and it prints ethaddr like this: 02:22:23:15:86:a5.

Sorry, my fault.

<snip>

Quoted 8 lines
Quoted 8 lines

arch/arm/include/asm/mach-types.h

Quoted 5 lines
arch/arm/include/asm/mach-types.h +5 -0
+#define CONFIG_CMD_BOOTZ
+#define CONFIG_OF_LIBFDT
+
+/* general purpose I/O */
+#define CONFIG_ATMEL_LEGACY        /* required until (g)pio is fixed */

I doubt you need this for gpio. Could you please check, if it is really required?

Yes, it's required.

Quoted 7 lines
Quoted 3 lines

Yes, I knew that, but I have looked many other board files and they use both GPIO and PIO in one file.
If is necessary I'll re-base it to PIO.

No, nand is not yet prepared for the generic gpio framework. I'll try to post patches next weeks (for 2014.01), but I'm quite busy right now. Since your board will also end up in 2014.01 this should go together.

I'll wait for that.

Quoted 2 lines

If you have the time I would be grateful if you could prepare patches for that.

Sorry, but I have look in atmel_nand.c and didn't find any pio. Seems to me it's prepared to gpio. So I leave it up to you.

Quoted 22 lines

<snip>

Quoted 17 lines
Quoted 8 lines
Quoted 8 lines
arch/arm/include/asm/mach-types.h +8 -0
+#define CONFIG_MTD_DEVICE
+#define CONFIG_CMD_MTDPARTS
+#define CONFIG_MTD_PARTITIONS
+#define CONFIG_RBTREE
+#define CONFIG_LZO
+#define CONFIG_CMD_UBI
+#define CONFIG_CMD_UBIFS
+#define CONFIG_CMD_NAND_TRIMFFS
Quoted 8 lines

arch/arm/include/asm/mach-types.h

Quoted 5 lines
arch/arm/include/asm/mach-types.h +5 -0
+#define MTDIDS_DEFAULT            "nand0=nand"
+#define MTDPARTS_DEFAULT        "mtdparts=nand:256k(bootstrap),"\
+                    "768k(uboot),256k(ubootenv),"\
+                    "4864k(kernel),"\
+                    "-(root)"

just to mention it:
a) a secondary env is sometimes useful

Quoted 1 line

Is that true? I thought use one copy env, if not then use default.

Quoted 9 lines

Well, your point is true, but how about heavily changed env in the field? If something goes wrong (two bad blocks at worst place, wrong crc due to too much bit errors, ...) it will fall back to the compiled in env, which may be completely wrong then.
But it is also true that an enabled secondary env will be read on every startup and compared to the first env. Also you always need to write two locations. It depends on your use case, but for me it is careless to have just one env on NAND.

We use in 99% compiled in env and even if not there is no need to heavily change env.

Quoted 27 lines

<snip>

Quoted 11 lines
Quoted 9 lines

arch/arm/include/asm/mach-types.h

Quoted 6 lines
arch/arm/include/asm/mach-types.h +6 -0
+#define CONFIG_BOOTCOMMAND    "nand read 21000000 kernel; bootm"
+#define CONFIG_BOOTARGS        "console=ttyS0,115200 ubi.mtd=root "\
+                "root=ubi0:root rootfstype=ubifs rw"
+#define CONFIG_SERVERIP        10.0.1.1
+#define CONFIG_BOOTFILE        "kernel_cdu9g25"
+#define CONFIG_PREBOOT        "mtdparts default" /* for partitions */

Would you really like to always reset a new mtd partitioning on every boot to the default one?

Does it mean that I don't need run this command if I like use partitions in nand commands?

No, not fully. You'll need the environment parameter 'mtdparts' with contend 'mdtparts=...' and respective mtdids env. This could be set by the command 'mtdparts default', it will set the compiled in values of MTDIDS_DEFAULT and MTDPARTS_DEFAULT.
If you like to change the nand partitioning in any way you need to change the mtdparts env. But when you call 'mtdparts default' afterwards you will reset this to the default value ...
To avoid this we have in our default environment following setting:

---8<---

#define CONFIG_EXTRA_ENV_SETTINGS \
   "mtdparts=" MTDPARTS_DEFAULT "\0" \
   "mtdids=" MTDIDS_DEFAULT "\0" \
Quoted 6 lines

...
--->8---

Additionally we avoid to use the 'mtdparts default' command. With this setup one could change the mtdparts environment any time and it would take effect.

I've changed it this way, you're right. I'm still learning new things.

Quoted 7 lines

(Just my 2?, no need to do it that way)

Best regards

Andreas Bie?mann

Thanks
Jiri

#5 lore raw

PATCH v5 1/1 at91: add support for CDU9G25 board

Dear Ji?? Prchal,

Quoted 27 lines · Ji?? Prchal

Dne 16.9.2013 10:27, Andreas Bie?mann napsal(a):

Quoted 26 lines · Ji?? Prchal
Quoted 2 lines

Dne 13.9.2013 16:10, Andreas Bie?mann napsal(a):

Quoted 1 line · Jiri Prchal

<snip>

Quoted 19 lines
Quoted 5 lines
Quoted 5 lines
arch/arm/include/asm/mach-types.h +0 -0
index 440b041..9b274ba 100644
Quoted 8 lines

arch/arm/include/asm/mach-types.h extern unsigned int __machine_arch_type; · line 989

Quoted 5 lines
arch/arm/include/asm/mach-types.h +1 -0
@@ -986,6 +986,7 @@ extern unsigned int __machine_arch_type;
   #define MACH_TYPE_VIT_IBOX             3371
   #define MACH_TYPE_DM6441_ESP           3372
   #define MACH_TYPE_AT91SAM9X5EK         3373
+#define MACH_TYPE_CDU9G25              3373

NAK, please obtain a mach type:
http://www.arm.linux.org.uk/developer/machines/?action=new

Should I register machine? I develop DT only:
"NOTE 1:If you are developing a DT-only platform, you do not need to register a machine type for it.
Please do not register a machine type. Thanks."
Do I need this MACH_TYPE_* at all?

Quoted 5 lines
Quoted 3 lines

I'm not really familiar with FDT only boards, but I think it is sane to just use a zero machid then. Could you please check this and if working just omit the machid setting?

I'm not really familiar with FDT too, but I tested with 0 and it works.

so eliminate the MACH id in your patch then.

Quoted 16 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 6 lines
arch/arm/include/asm/mach-types.h +5 -0
+#define CONFIG_CMD_BOOTZ
+#define CONFIG_OF_LIBFDT
+
+/* general purpose I/O */
+#define CONFIG_ATMEL_LEGACY        /* required until (g)pio is

fixed */

I doubt you need this for gpio. Could you please check, if it is really required?

Yes, it's required.

Quoted 4 lines
Quoted 3 lines

Yes, I knew that, but I have looked many other board files and they use both GPIO and PIO in one file.
If is necessary I'll re-base it to PIO.

Quoted 5 lines
Quoted 3 lines

No, nand is not yet prepared for the generic gpio framework. I'll try to post patches next weeks (for 2014.01), but I'm quite busy right now. Since your board will also end up in 2014.01 this should go together.

I'll wait for that.

My changes are in u-boot-atmel/master now. If you send a rebased version of your patch I'll apply it for 2014.01 release.

Best regards

Andreas Bie?mann