uboot源码分析六 uboot启动流程三 lowlevel_init 函数 s_init 函数 _main 函数

lowlevel_init 函数

函数 lowlevel_init 在文件 arch/arm/cpu/armv7/lowlevel_init.S 中定义，

/*
 * A lowlevel_init function that sets up the stack to call a C function to
 * perform further init.
 *
 * (C) Copyright 2010
 * Texas Instruments, <www.ti.com>
 *
 * Author :
 *	Aneesh V	<aneesh@ti.com>
 *
 * SPDX-License-Identifier:	GPL-2.0+
 */

#include <asm-offsets.h>
#include <config.h>
#include <linux/linkage.h>

ENTRY(lowlevel_init)
	/*
	 * Setup a temporary stack. Global data is not available yet.
	 */
	ldr	sp, =CONFIG_SYS_INIT_SP_ADDR
	bic	sp, sp, #7 /* 8-byte alignment for ABI compliance */
#ifdef CONFIG_SPL_DM
	mov	r9, #0
#else
	/*
	 * Set up global data for boards that still need it. This will be
	 * removed soon.
	 */
#ifdef CONFIG_SPL_BUILD
	ldr	r9, =gdata
#else
	sub	sp, sp, #GD_SIZE
	bic	sp, sp, #7
	mov	r9, sp
#endif
#endif
	/*
	 * Save the old lr(passed in ip) and the current lr to stack
	 */
	push	{ip, lr}

	/*
	 * Call the very early init function. This should do only the
	 * absolute bare minimum to get started. It should not:
	 *
	 * - set up DRAM
	 * - use global_data
	 * - clear BSS
	 * - try to start a console
	 *
	 * For boards with SPL this should be empty since SPL can do all of
	 * this init in the SPL board_init_f() function which is called
	 * immediately after this.
	 */
	bl	s_init
	pop	{ip, pc}
ENDPROC(lowlevel_init)

第 22 行设置 sp 指向 CONFIG_SYS_INIT_SP_ADDR， CONFIG_SYS_INIT_SP_ADDR 在include/configs/mx6ullevk.h 文件中

IRAM_BASE_ADDR 和 IRAM_SIZE 在 文 件arch/arm/include/asm/arch-mx6/imx-regs.h 中有定义


如果 408 行的条件成立的话 IRAM_SIZE=0X40000，当定义了 CONFIG_MX6SX、CONFIG_MX6U、 CONFIG_MX6SLL 和 CONFIG_MX6SL 中的任意一个的话条件就不成立，在.config 中定义了 CONFIG_MX6UL，所以条件不成立，因此 IRAM_SIZE=0X20000=128KB。

可以得到如下值：

CONFIG_SYS_INIT_RAM_ADDR = IRAM_BASE_ADDR = 0x00900000。
CONFIG_SYS_INIT_RAM_SIZE = 0x00020000 =128KB。

还需要知道GENERATED_GBL_DATA_SIZE的值，在文件include/generated/generic-asm-offsets.h中有定义

GENERATED_GBL_DATA_SIZE=256， GENERATED_GBL_DATA_SIZE 的含义为

(sizeof(struct global_data) + 15) & ~15 。

综上所述， CONFIG_SYS_INIT_SP_ADDR 值如下：

CONFIG_SYS_INIT_SP_OFFSET = 0x00020000 – 256 = 0x1FF00。
CONFIG_SYS_INIT_SP_ADDR = 0x00900000 + 0X1FF00 = 0X0091FF00，

此时 sp 指向 0X91FF00，这属于 IMX6UL/IMX6ULL 的内部 ram。继续回到文件 lowlevel_init.S，第 23 行对 sp 指针做 8 字节对齐处理！
第 34 行， sp 指针减去 GD_SIZE， GD_SIZE 同样在 generic-asm-offsets.h 中定了，大小为248
第 35 行对 sp 做 8 字节对齐，此时 sp 的地址为 0X0091FF00-248=0X0091FE08，此时 sp 位置如图所示：

第 36 行将 sp 地址保存在 r9 寄存器中。
第 42 行将 ip 和 lr 压栈
第 57 行调用函数 s_init
第 58 行将第 36 行入栈的 ip 和 lr 进行出栈，并将 lr 赋给 pc。

s_init 函数

s_init 函数定义在文件arch/arm/cpu/armv7/mx6/soc.c 中

在第 816 行会判断当前 CPU 类型，如果 CPU 为 MX6SX、 MX6UL、 MX6ULL 或 MX6SLL中 的 任 意 一 种 ， 那 么 就 会 直 接 返 回 ， 相 当 于 s_init 函 数 什 么 都 没 做 。 所 以 对 于I.MX6UL/I.MX6ULL 来说， s_init 就是个空函数。从 s_init 函数退出以后进入函数 lowlevel_init，但是 lowlevel_init 函数也执行完成了，返回到了函数 cpu_init_crit，函数 cpu_init_crit 也执行完成了，最终返回到 save_boot_params_ret

接下来要执行的是 save_boot_params_ret 中的_main 函数，接下来分析_main 函数。

_main 函数

_main 函数定义在文件 arch/arm/lib/crt0.S 中

/*
 *  crt0 - C-runtime startup Code for ARM U-Boot
 *
 *  Copyright (c) 2012  Albert ARIBAUD <albert.u.boot@aribaud.net>
 *
 * SPDX-License-Identifier:	GPL-2.0+
 */

#include <config.h>
#include <asm-offsets.h>
#include <linux/linkage.h>
#ifdef CONFIG_CPU_V7M
#include <asm/armv7m.h>
#endif

/*
 * This file handles the target-independent stages of the U-Boot
 * start-up where a C runtime environment is needed. Its entry point
 * is _main and is branched into from the target's start.S file.
 *
 * _main execution sequence is:
 *
 * 1. Set up initial environment for calling board_init_f().
 *    This environment only provides a stack and a place to store
 *    the GD ('global data') structure, both located in some readily
 *    available RAM (SRAM, locked cache...). In this context, VARIABLE
 *    global data, initialized or not (BSS), are UNAVAILABLE; only
 *    CONSTANT initialized data are available. GD should be zeroed
 *    before board_init_f() is called.
 *
 * 2. Call board_init_f(). This function prepares the hardware for
 *    execution from system RAM (DRAM, DDR...) As system RAM may not
 *    be available yet, , board_init_f() must use the current GD to
 *    store any data which must be passed on to later stages. These
 *    data include the relocation destination, the future stack, and
 *    the future GD location.
 *
 * 3. Set up intermediate environment where the stack and GD are the
 *    ones allocated by board_init_f() in system RAM, but BSS and
 *    initialized non-const data are still not available.
 *
 * 4a.For U-Boot proper (not SPL), call relocate_code(). This function
 *    relocates U-Boot from its current location into the relocation
 *    destination computed by board_init_f().
 *
 * 4b.For SPL, board_init_f() just returns (to crt0). There is no
 *    code relocation in SPL.
 *
 * 5. Set up final environment for calling board_init_r(). This
 *    environment has BSS (initialized to 0), initialized non-const
 *    data (initialized to their intended value), and stack in system
 *    RAM (for SPL moving the stack and GD into RAM is optional - see
 *    CONFIG_SPL_STACK_R). GD has retained values set by board_init_f().
 *
 * 6. For U-Boot proper (not SPL), some CPUs have some work left to do
 *    at this point regarding memory, so call c_runtime_cpu_setup.
 *
 * 7. Branch to board_init_r().
 *
 * For more information see 'Board Initialisation Flow in README.
 */

/*
 * entry point of crt0 sequence
 */

ENTRY(_main)

/*
 * Set up initial C runtime environment and call board_init_f(0).
 */

#if defined(CONFIG_SPL_BUILD) && defined(CONFIG_SPL_STACK)
	ldr	sp, =(CONFIG_SPL_STACK)
#else
	ldr	sp, =(CONFIG_SYS_INIT_SP_ADDR)
#endif
#if defined(CONFIG_CPU_V7M)	/* v7M forbids using SP as BIC destination */
	mov	r3, sp
	bic	r3, r3, #7
	mov	sp, r3
#else
	bic	sp, sp, #7	/* 8-byte alignment for ABI compliance */
#endif
	mov	r0, sp
	bl	board_init_f_alloc_reserve
	mov	sp, r0
	/* set up gd here, outside any C code */
	mov	r9, r0
	bl	board_init_f_init_reserve

	mov	r0, #0
	bl	board_init_f

#if ! defined(CONFIG_SPL_BUILD)

/*
 * Set up intermediate environment (new sp and gd) and call
 * relocate_code(addr_moni). Trick here is that we'll return
 * 'here' but relocated.
 */

	ldr	sp, [r9, #GD_START_ADDR_SP]	/* sp = gd->start_addr_sp */
#if defined(CONFIG_CPU_V7M)	/* v7M forbids using SP as BIC destination */
	mov	r3, sp
	bic	r3, r3, #7
	mov	sp, r3
#else
	bic	sp, sp, #7	/* 8-byte alignment for ABI compliance */
#endif
	ldr	r9, [r9, #GD_BD]		/* r9 = gd->bd */
	sub	r9, r9, #GD_SIZE		/* new GD is below bd */

	adr	lr, here
	ldr	r0, [r9, #GD_RELOC_OFF]		/* r0 = gd->reloc_off */
	add	lr, lr, r0
#if defined(CONFIG_CPU_V7M)
	orr	lr, #1				/* As required by Thumb-only */
#endif
	ldr	r0, [r9, #GD_RELOCADDR]		/* r0 = gd->relocaddr */
	b	relocate_code
here:
/*
 * now relocate vectors
 */

	bl	relocate_vectors

/* Set up final (full) environment */

	bl	c_runtime_cpu_setup	/* we still call old routine here */
#endif
#if !defined(CONFIG_SPL_BUILD) || defined(CONFIG_SPL_FRAMEWORK)
# ifdef CONFIG_SPL_BUILD
	/* Use a DRAM stack for the rest of SPL, if requested */
	bl	spl_relocate_stack_gd
	cmp	r0, #0
	movne	sp, r0
	movne	r9, r0
# endif
	ldr	r0, =__bss_start	/* this is auto-relocated! */

#ifdef CONFIG_USE_ARCH_MEMSET
	ldr	r3, =__bss_end		/* this is auto-relocated! */
	mov	r1, #0x00000000		/* prepare zero to clear BSS */

	subs	r2, r3, r0		/* r2 = memset len */
	bl	memset
#else
	ldr	r1, =__bss_end		/* this is auto-relocated! */
	mov	r2, #0x00000000		/* prepare zero to clear BSS */

clbss_l:cmp	r0, r1			/* while not at end of BSS */
#if defined(CONFIG_CPU_V7M)
	itt	lo
#endif
	strlo	r2, [r0]		/* clear 32-bit BSS word */
	addlo	r0, r0, #4		/* move to next */
	blo	clbss_l
#endif

#if ! defined(CONFIG_SPL_BUILD)
	bl coloured_LED_init
	bl red_led_on
#endif
	/* call board_init_r(gd_t *id, ulong dest_addr) */
	mov     r0, r9                  /* gd_t */
	ldr	r1, [r9, #GD_RELOCADDR]	/* dest_addr */
	/* call board_init_r */
#if defined(CONFIG_SYS_THUMB_BUILD)
	ldr	lr, =board_init_r	/* this is auto-relocated! */
	bx	lr
#else
	ldr	pc, =board_init_r	/* this is auto-relocated! */
#endif
	/* we should not return here. */
#endif

ENDPROC(_main)

第 76 行，设置 sp 指针为 CONFIG_SYS_INIT_SP_ADDR，也就是 sp 指向 0X0091FF00。
第 83 行， sp 做 8 字节对齐。
第 85 行，读取 sp 到寄存器 r0 里面，此时 r0=0X0091FF00。
第 86 行，调用函数 board_init_f_alloc_reserve，此函数有一个参数，参数为 r0 中的值，也就是 0X0091FF00，此函数定义在文件 common/init/board_init.c 中

函数 board_init_f_alloc_reserve 主要是留出早期的 malloc 内存区域和 gd 内存区域，其中CONFIG_SYS_MALLOC_F_LEN=0X400( 在 文 件 include/generated/autoconf.h 中 定 义 ) ，sizeof(struct global_data)=248(GD_SIZE 值)

函数 board_init_f_alloc_reserve 是有返回值的，返回值为新的 top 值，从图 32.2.4.1 可知，此时 top=0X0091FA00。
第 87 行，将 r0 写入到 sp 里面， r0 保存着函数board_init_f_alloc_reserve 的返回值，所以这一句也就是设置 sp=0X0091FA00。
第 89 行，将 r0 寄存器的值写到寄存器 r9 里面，因为 r9 寄存器存放着全局变量 gd 的地址，在文件 arch/arm/include/asm/global_data.h 中有如图所示宏定义

uboot 中定义了一个指向 gd_t 的指针 gd， gd 存放在寄存器 r9 里面的，因此 gd 是个全局变量。 gd_t 是个结构体，在 include/asm-generic/global_data.h 里面有定义，

*
 * Copyright (c) 2012 The Chromium OS Authors.
 * (C) Copyright 2002-2010
 * Wolfgang Denk, DENX Software Engineering, wd@denx.de.
 *
 * SPDX-License-Identifier:	GPL-2.0+
 */

#ifndef __ASM_GENERIC_GBL_DATA_H
#define __ASM_GENERIC_GBL_DATA_H
/*
 * The following data structure is placed in some memory which is
 * available very early after boot (like DPRAM on MPC8xx/MPC82xx, or
 * some locked parts of the data cache) to allow for a minimum set of
 * global variables during system initialization (until we have set
 * up the memory controller so that we can use RAM).
 *
 * Keep it *SMALL* and remember to set GENERATED_GBL_DATA_SIZE > sizeof(gd_t)
 *
 * Each architecture has its own private fields. For now all are private
 */

#ifndef __ASSEMBLY__
#include <membuff.h>
#include <linux/list.h>

typedef struct global_data {
	bd_t *bd;
	unsigned long flags;
	unsigned int baudrate;
	unsigned long cpu_clk;	/* CPU clock in Hz!		*/
	unsigned long bus_clk;
	/* We cannot bracket this with CONFIG_PCI due to mpc5xxx */
	unsigned long pci_clk;
	unsigned long mem_clk;
#if defined(CONFIG_LCD) || defined(CONFIG_VIDEO)
	unsigned long fb_base;	/* Base address of framebuffer mem */
#endif
#if defined(CONFIG_POST) || defined(CONFIG_LOGBUFFER)
	unsigned long post_log_word;  /* Record POST activities */
	unsigned long post_log_res; /* success of POST test */
	unsigned long post_init_f_time;  /* When post_init_f started */
#endif
#ifdef CONFIG_BOARD_TYPES
	unsigned long board_type;
#endif
	unsigned long have_console;	/* serial_init() was called */
#ifdef CONFIG_PRE_CONSOLE_BUFFER
	unsigned long precon_buf_idx;	/* Pre-Console buffer index */
#endif
	unsigned long env_addr;	/* Address  of Environment struct */
	unsigned long env_valid;	/* Checksum of Environment valid? */

	unsigned long ram_top;	/* Top address of RAM used by U-Boot */

	unsigned long relocaddr;	/* Start address of U-Boot in RAM */
	phys_size_t ram_size;	/* RAM size */
#ifdef CONFIG_SYS_MEM_RESERVE_SECURE
#define MEM_RESERVE_SECURE_SECURED	0x1
#define MEM_RESERVE_SECURE_MAINTAINED	0x2
#define MEM_RESERVE_SECURE_ADDR_MASK	(~0x3)
	/*
	 * Secure memory addr
	 * This variable needs maintenance if the RAM base is not zero,
	 * or if RAM splits into non-consecutive banks. It also has a
	 * flag indicating the secure memory is marked as secure by MMU.
	 * Flags used: 0x1 secured
	 *             0x2 maintained
	 */
	phys_addr_t secure_ram;
#endif
	unsigned long mon_len;	/* monitor len */
	unsigned long irq_sp;		/* irq stack pointer */
	unsigned long start_addr_sp;	/* start_addr_stackpointer */
	unsigned long reloc_off;
	struct global_data *new_gd;	/* relocated global data */

#ifdef CONFIG_DM
	struct udevice	*dm_root;	/* Root instance for Driver Model */
	struct udevice	*dm_root_f;	/* Pre-relocation root instance */
	struct list_head uclass_root;	/* Head of core tree */
#endif
#ifdef CONFIG_TIMER
	struct udevice	*timer;	/* Timer instance for Driver Model */
#endif

	const void *fdt_blob;	/* Our device tree, NULL if none */
	void *new_fdt;		/* Relocated FDT */
	unsigned long fdt_size;	/* Space reserved for relocated FDT */
	struct jt_funcs *jt;		/* jump table */
	char env_buf[32];	/* buffer for getenv() before reloc. */
#ifdef CONFIG_TRACE
	void		*trace_buff;	/* The trace buffer */
#endif
#if defined(CONFIG_SYS_I2C)
	int		cur_i2c_bus;	/* current used i2c bus */
#endif
#ifdef CONFIG_SYS_I2C_MXC
	void *srdata[10];
#endif
	unsigned long timebase_h;
	unsigned long timebase_l;
#ifdef CONFIG_SYS_MALLOC_F_LEN
	unsigned long malloc_base;	/* base address of early malloc() */
	unsigned long malloc_limit;	/* limit address */
	unsigned long malloc_ptr;	/* current address */
#endif
#ifdef CONFIG_PCI
	struct pci_controller *hose;	/* PCI hose for early use */
	phys_addr_t pci_ram_top;	/* top of region accessible to PCI */
#endif
#ifdef CONFIG_PCI_BOOTDELAY
	int pcidelay_done;
#endif
	struct udevice *cur_serial_dev;	/* current serial device */
	struct arch_global_data arch;	/* architecture-specific data */
#ifdef CONFIG_CONSOLE_RECORD
	struct membuff console_out;	/* console output */
	struct membuff console_in;	/* console input */
#endif
#ifdef CONFIG_DM_VIDEO
	ulong video_top;		/* Top of video frame buffer area */
	ulong video_bottom;		/* Bottom of video frame buffer area */
#endif
} gd_t;
#endif

/*
 * Global Data Flags - the top 16 bits are reserved for arch-specific flags
 */
#define GD_FLG_RELOC		0x00001	/* Code was relocated to RAM	   */
#define GD_FLG_DEVINIT		0x00002	/* Devices have been initialized   */
#define GD_FLG_SILENT		0x00004	/* Silent mode			   */
#define GD_FLG_POSTFAIL		0x00008	/* Critical POST test failed	   */
#define GD_FLG_POSTSTOP		0x00010	/* POST seqeunce aborted	   */
#define GD_FLG_LOGINIT		0x00020	/* Log Buffer has been initialized */
#define GD_FLG_DISABLE_CONSOLE	0x00040	/* Disable console (in & out)	   */
#define GD_FLG_ENV_READY	0x00080	/* Env. imported into hash table   */
#define GD_FLG_SERIAL_READY	0x00100	/* Pre-reloc serial console ready  */
#define GD_FLG_FULL_MALLOC_INIT	0x00200	/* Full malloc() is ready	   */
#define GD_FLG_SPL_INIT		0x00400	/* spl_init() has been called	   */
#define GD_FLG_SKIP_RELOC	0x00800	/* Don't relocate */
#define GD_FLG_RECORD		0x01000	/* Record console */

#endif /* __ASM_GENERIC_GBL_DATA_H */

这一行代码就是设置 gd 所指向的位置，也就是 gd 指向 0X0091FA00。

第 90 行调用函数 board_init_f_init_reserve，此函数在文件common/init/board_init.c 中有定义

/*
 * Code shared between SPL and U-Boot proper
 *
 * Copyright (c) 2015 Google, Inc
 * Written by Simon Glass <sjg@chromium.org>
 *
 * SPDX-License-Identifier:	GPL-2.0+
 */

#include <common.h>

DECLARE_GLOBAL_DATA_PTR;

/*
 * It isn't trivial to figure out whether memcpy() exists. The arch-specific
 * memcpy() is not normally available in SPL due to code size.
 */
#if !defined(CONFIG_SPL_BUILD) || \
		(defined(CONFIG_SPL_LIBGENERIC_SUPPORT) && \
		!defined(CONFIG_USE_ARCH_MEMSET))
#define _USE_MEMCPY
#endif

/* Unfortunately x86 or ARM can't compile this code as gd cannot be assigned */
#if !defined(CONFIG_X86) && !defined(CONFIG_ARM)
__weak void arch_setup_gd(struct global_data *gd_ptr)
{
	gd = gd_ptr;
}
#endif /* !CONFIG_X86 && !CONFIG_ARM */

/*
 * Allocate reserved space for use as 'globals' from 'top' address and
 * return 'bottom' address of allocated space
 *
 * Notes:
 *
 * Actual reservation cannot be done from within this function as
 * it requires altering the C stack pointer, so this will be done by
 * the caller upon return from this function.
 *
 * IMPORTANT:
 *
 * Alignment constraints may differ for each 'chunk' allocated. For now:
 *
 * - GD is aligned down on a 16-byte boundary
 *
 *  - the early malloc arena is not aligned, therefore it follows the stack
 *   alignment constraint of the architecture for which we are bulding.
 *
 *  - GD is allocated last, so that the return value of this functions is
 *   both the bottom of the reserved area and the address of GD, should
 *   the calling context need it.
 */

ulong board_init_f_alloc_reserve(ulong top)
{
	/* Reserve early malloc arena */
#if defined(CONFIG_SYS_MALLOC_F)
	top -= CONFIG_SYS_MALLOC_F_LEN;
#endif
	/* LAST : reserve GD (rounded up to a multiple of 16 bytes) */
	top = rounddown(top-sizeof(struct global_data), 16);

	return top;
}

/*
 * Initialize reserved space (which has been safely allocated on the C
 * stack from the C runtime environment handling code).
 *
 * Notes:
 *
 * Actual reservation was done by the caller; the locations from base
 * to base+size-1 (where 'size' is the value returned by the allocation
 * function above) can be accessed freely without risk of corrupting the
 * C runtime environment.
 *
 * IMPORTANT:
 *
 * Upon return from the allocation function above, on some architectures
 * the caller will set gd to the lowest reserved location. Therefore, in
 * this initialization function, the global data MUST be placed at base.
 *
 * ALSO IMPORTANT:
 *
 * On some architectures, gd will already be good when entering this
 * function. On others, it will only be good once arch_setup_gd() returns.
 * Therefore, global data accesses must be done:
 *
 * - through gd_ptr if before the call to arch_setup_gd();
 *
 * - through gd once arch_setup_gd() has been called.
 *
 * Do not use 'gd->' until arch_setup_gd() has been called!
 *
 * IMPORTANT TOO:
 *
 * Initialization for each "chunk" (GD, early malloc arena...) ends with
 * an incrementation line of the form 'base += <some size>'. The last of
 * these incrementations seems useless, as base will not be used any
 * more after this incrementation; but if/when a new "chunk" is appended,
 * this increment will be essential as it will give base right value for
 * this new chunk (which will have to end with its own incrementation
 * statement). Besides, the compiler's optimizer will silently detect
 * and remove the last base incrementation, therefore leaving that last
 * (seemingly useless) incrementation causes no code increase.
 */

void board_init_f_init_reserve(ulong base)
{
	struct global_data *gd_ptr;
#ifndef _USE_MEMCPY
	int *ptr;
#endif

	/*
	 * clear GD entirely and set it up.
	 * Use gd_ptr, as gd may not be properly set yet.
	 */

	gd_ptr = (struct global_data *)base;
	/* zero the area */
#ifdef _USE_MEMCPY
	memset(gd_ptr, '\0', sizeof(*gd));
#else
	for (ptr = (int *)gd_ptr; ptr < (int *)(gd_ptr + 1); )
		*ptr++ = 0;
#endif
	/* set GD unless architecture did it already */
#if !defined(CONFIG_ARM)
	arch_setup_gd(gd_ptr);
#endif
	/* next alloc will be higher by one GD plus 16-byte alignment */
	base += roundup(sizeof(struct global_data), 16);

	/*
	 * record early malloc arena start.
	 * Use gd as it is now properly set for all architectures.
	 */

#if defined(CONFIG_SYS_MALLOC_F)
	/* go down one 'early malloc arena' */
	gd->malloc_base = base;
	/* next alloc will be higher by one 'early malloc arena' size */
	base += CONFIG_SYS_MALLOC_F_LEN;
#endif
}

可以看出，此函数用于初始化 gd，其实就是清零处理。 另外，此函数还设置了gd->malloc_base 为 gd 基地址+gd 大小=0X0091FA00+248=0X0091FAF8，在做 16 字节对齐，最终 gd->malloc_base=0X0091FB00，这个也就是 early malloc 的起始地址。

继续回到代码main 中，第 92 行设置 R0 为 0。
第 93 行，调用 board_init_f 函数，此函数定义在文件 common/board_f.c 中！主要用来初始化 DDR，定时器，完成代码拷贝等等，此函数我们后面在详细的分析。
第 103 行，重新设置环境(sp 和 gd)、获取 gd->start_addr_sp 的值赋给 sp，在函数 board_init_f中会初始化 gd 的所有成员变量，其中 gd->start_addr_sp=0X9EF44E90， 所以这里相当于设置sp=gd->start_addr_sp=0X9EF44E90。 0X9EF44E90 是 DDR 中的地址，说明新的 sp 和 gd 将会存放到 DDR 中，而不是内部的 RAM 了。 GD_START_ADDR_SP=64，参考示例代码 32.2.2.4。
第 109 行， sp 做 8 字节对齐。
第 111 行，获取 gd->bd 的地址赋给 r9，此时 r9 存放的是老的 gd，这里通过获取 gd->bd 的地址来计算出新的 gd 的位置。 GD_BD=0，参考示例代码 32.2.2.4。
第 112 行，新的 gd 在 bd 下面，所以 r9 减去 gd 的大小就是新的 gd 的位置，获取到新的 gd的位置以后赋值给 r9。
第 114 行，设置 lr 寄存器为 here，这样后面执行其他函数返回的时候就返回到了第 122 行的 here 位置处。
第 115，读取 gd->reloc_off 的值复制给 r0 寄存器， GD_RELOC_OFF=68
第 116 行， lr 寄存器的值加上 r0 寄存器的值，重新赋值给 lr 寄存器。因为接下来要重定位代码，也就是把代码拷贝到新的地方去(现在的 uboot 存放的起始地址为 0X87800000，下面要将 uboot 拷贝到 DDR 最后面的地址空间出，将 0X87800000 开始的内存空出来)，其中就包括here，因此 lr 中的 here 要使用重定位后的位置。
第 120 行，读取 gd->relocaddr 的值赋给 r0 寄存器，此时 r0 寄存器就保存着 uboot 要拷贝的目的地址，为 0X9FF47000。GD_RELOCADDR=48，参考示例代码 32.2.2.4。
第 121 行，调用函数 relocate_code，也就是代码重定位函数，此函数负责将 uboot 拷贝到新的地方去，此函数定义在文件 arch/arm/lib/relocate.S 中稍后会详细分析此函数。
第 127 行，调用函数 relocate_vectors，对中断向量表做重定位，此函数定义在文件arch/arm/lib/relocate.S 中，稍后会详细分析此函数。
第 131 行，调用函数 c_runtime_cpu_setup，此函数定义在文件 arch/arm/cpu/armv7/start.S 中，
函数内容如下：

79 * If I-cache is enabled invalidate it
80 */
81 #ifndef CONFIG_SYS_ICACHE_OFF
82 mcr p15, 0, r0, c7, c5, 0 @ invalidate icache
83 mcr p15, 0, r0, c7, c10, 4 @ DSB
84 mcr p15, 0, r0, c7, c5, 4 @ ISB
85 #endif
86
87 bx lr
88
89 ENDPROC(c_runtime_cpu_setup)

第 141~159 行，清除 BSS 段。
第 167 行，设置函数 board_init_r 的两个参数，函数 board_init_r 声明如下：board_init_r(gd_t *id, ulong dest_addr) 第一个参数是 gd，因此读取 r9 保存到 r0 里面。
第 168 行，设置函数 board_init_r 的第二个参数是目的地址，因此 r1= gd->relocaddr。
第 174 行、调用函数 board_init_r，此函数定义在文件 common/board_r.c 中，稍后会详细的分析此函数。
这个就是_main 函数的运行流程，在_main 函数里面调用了 board_init_f、 relocate_code、relocate_vectors 和 board_init_r 这 4 个函数