研究内核源码和内核运行原理的时候,很总要的一点是要了解内核的初始情况,也就是要了解内核启动过程。我在研究内核的内存管理的时候,想知道内核启动后的页表的放置,页表的初始化等信息,这促使我这次仔细地研究内核的启动代码。
CPU在bootloader的帮助下将内核载入到了内存中,并开始执行。当然,bootloader必须为zImage做好必要的准备:
1. CPU 寄存器的设置: |
R0=0; R1=Machine ID(即Machine Type Number,定义在linux/arch/arm/tools/mach-types); R2=内核启动参数在 RAM 中起始基地址; |
2. CPU 模式: |
必须禁止中断(IRQs和FIQs); CPU 必须 SVC 模式; |
3. Cache 和 MMU 的设置: |
MMU 必须关闭; 指令 Cache 可以打开也可以关闭; 数据 Cache 必须关闭; |
知道内核zImage生成的朋友一定知道:真正的内核执行映像其实是在编译时生成arch/$(ARCH)/boot/文件夹中的Image文件(bin文件),而zImage其实是将这个可执行文件作为数据段包含在了自身中,而zImage的代码功能就是将这个数据(Image)正确地解压到编译时确定的位置中去,并跳到Image中运行。所以实现bootloader引导的压缩映像zImage的入口是由arch/arm /boot/compressed/vmlinux.lds决定的(这个文件是由vmlinux.lds.in生成的)。所以从vmlinux.lds.in中可以看出压缩映像的入口在哪:
- ......
- OUTPUT_ARCH(arm)
- ENTRY(_start)
- SECTIONS
- {
- /DISCARD/ : {
- *(.ARM.exidx*)
- *(.ARM.extab*)
- /*
- * Discard any r/w data - this produces a link error if we have any,
- * which is required for PIC decompression. Local data generates
- * GOTOFF relocations, which prevents it being relocated independently
- * of the text/got segments.
- */
- *(.data)
- }
- . = TEXT_START;
- _text = .;
- .text : {
- _start = .;
- *(.start)
- *(.text)
- ......
我们可以在arch/arm/boot/compressed/head.S找到这个start入口,这样就可以从这里开始用代码分析的方法研究bootloader跳转到压缩内核映像后的自解压启动过程:
再看到MMU设置的时候,我只研究了armv7的指令。看这些代码,必须对ARM的MMU有一定的了解,建议参考ARMv7的构架手册和网上的一份PDF《ARM MMU中文详解》(就是ARM手册中MMU部分的翻译)
- /*
- * linux/arch/arm/boot/compressed/head.S
- *
- * Copyright (C) 1996-2002 Russell King
- * Copyright (C) 2004 Hyok S. Choi (MPU support)
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 as
- * published by the Free Software Foundation.
- */
- #include
- /*
- * 调试宏
- *
- * 注意:这些宏必须不包含那些非100%可重定位的代码
- * 任何试图这样做的结果是导致程序崩溃
- * 当打开调试时请选择以下一个使用
- */
- #ifdef DEBUG /* 调试宏-中间层 */
- #if defined(CONFIG_DEBUG_ICEDCC) /* 使用内部调试协处理器CP14 */
- #if defined(CONFIG_CPU_V6) || defined(CONFIG_CPU_V6K) || defined(CONFIG_CPU_V7)
- .macro loadsp, rb, tmp
- .endm
- .macro writeb, ch, rb
- mcr p14, 0, \ch, c0, c5, 0
- .endm
- #elif defined(CONFIG_CPU_XSCALE)
- .macro loadsp, rb, tmp
- .endm
- .macro writeb, ch, rb
- mcr p14, 0, \ch, c8, c0, 0
- .endm
- #else
- .macro loadsp, rb, tmp
- .endm
- .macro writeb, ch, rb
- mcr p14, 0, \ch, c1, c0, 0
- .endm
- #endif
- #else /* 使用串口作为调试通道 */
- #include /* 包含构架相关的的调试宏的汇编文件 调试宏-底层 */
- .macro writeb, ch, rb
- senduart \ch, \rb
- .endm
- #if defined(CONFIG_ARCH_SA1100)
- .macro loadsp, rb, tmp
- mov \rb, #0x80000000 @ physical base address
- #ifdef CONFIG_DEBUG_LL_SER3
- add \rb, \rb, #0x00050000 @ Ser3
- #else
- add \rb, \rb, #0x00010000 @ Ser1
- #endif
- .endm
- #elif defined(CONFIG_ARCH_S3C2410)
- .macro loadsp, rb, tmp
- mov \rb, #0x50000000
- add \rb, \rb, #0x4000 * CONFIG_S3C_LOWLEVEL_UART_PORT
- .endm
- #else
- .macro loadsp, rb, tmp
- addruart \rb, \tmp
- .endm
- #endif
- #endif
- #endif /* DEBUG */
- /* 调试宏-上层 */
- .macro kputc,val /* 打印字符 */
- mov r0, \val
- bl putc
- .endm
- .macro kphex,val,len /* 打印十六进制数 */
- mov r0, \val
- mov r1, #\len
- bl phex
- .endm
- .macro debug_reloc_start /* 重定位内核调试宏-开始 */
- #ifdef DEBUG
- kputc #‘\n‘
- kphex r6, 8 /* 处理器 id */
- kputc #‘:‘
- kphex r7, 8 /* 构架 id */
- #ifdef CONFIG_CPU_CP15
- kputc #‘:‘
- mrc p15, 0, r0, c1, c0
- kphex r0, 8 /* 控制寄存器 */
- #endif
- kputc #‘\n‘
- kphex r5, 8 /* 解压后的内核起始地址 */
- kputc #‘-‘
- kphex r9, 8 /* 解压后的内核结束地址 */
- kputc #‘>‘
- kphex r4, 8 /* 内核执行地址 */
- kputc #‘\n‘
- #endif
- .endm
- .macro debug_reloc_end /* 重定位内核调试宏-结束 */
- #ifdef DEBUG
- kphex r5, 8 /* 内核结束地址 */
- kputc #‘\n‘
- mov r0, r4
- bl memdump /* 打印内核起始处 256 字节 */
- #endif
- .endm
- .section ".start", #alloc, #execinstr
- /*
- * 清理不同的调用约定
- */
- .align
- .arm @ 启动总是进入ARM状态
- start:
- .type start,#function
- .rept 7
- mov r0, r0
- .endr
- ARM( mov r0, r0 )
- ARM( b 1f )
- THUMB( adr r12, BSYM(1f) )
- THUMB( bx r12 )
- .word 0x016f2818 @ 用于boot loader的魔数
- .word start @ 加载/运行zImage的绝对地址(编译时确定)
- .word _edata @ zImage结束地址
- THUMB( .thumb )
- 1: mov r7, r1 @ 保存构架ID到r7(此前由bootloader放入r1)
- mov r8, r2 @ 保存内核启动参数地址到r8(此前由bootloader放入r2)
- #ifndef __ARM_ARCH_2__
- /*
- * 通过Angel调试器启动 - 必须进入 SVC模式且关闭FIQs/IRQs
- * (numeric definitions from angel arm.h source).
- * 如果进入时在user模式下,我们只需要做这些
- */
- mrs r2, cpsr @ 获取当前模式
- tst r2, #3 @ 判断是否是user模式
- bne not_angel
- mov r0, #0x17 @ angel_SWIreason_EnterSVC
- ARM( swi 0x123456 ) @ angel_SWI_ARM
- THUMB( svc 0xab ) @ angel_SWI_THUMB
- not_angel:
- mrs r2, cpsr @ 关闭中断
- orr r2, r2, #0xc0 @ 以保护调试器的运作
- msr cpsr_c, r2
- #else
- teqp pc, #0x0c000003 @ 关闭中断(此外bootloader已设置模式为SVC)
- #endif
- /*
- * 注意一些缓存的刷新和其他事务可能需要在这里完成
- * - is there an Angel SWI call for this?
- */
- /*
- * 一些构架的特定代码可以在这里被连接器插入,
- * 但是不应使用 r7(保存构架ID), r8(保存内核启动参数地址), and r9.
- */
- .text
- /*
- * 此处确定解压后的内核映像的绝对地址(物理地址),保存于r4
- * 由于配置的不同可能有的结果
- * (1)定义了CONFIG_AUTO_ZRELADDR
- * ZRELADDR是已解压内核最终存放的物理地址
- * 如果AUTO_ZRELADDR被选择了, 这个地址将会在运行是确定:
- * 将当pc值和0xf8000000做与操作,
- * 并加上TEXT_OFFSET(内核最终存放的物理地址与内存起始的偏移)
- * 这里假定zImage被放在内存开始的128MB内
- * (2)没有定义CONFIG_AUTO_ZRELADDR
- * 直接使用zreladdr(此值位于arch/arm/mach-xxx/Makefile.boot文件确定)
- */
- #ifdef CONFIG_AUTO_ZRELADDR
- @ 确定内核映像地址
- mov r4, pc
- and r4, r4, #0xf8000000
- add r4, r4, #TEXT_OFFSET
- #else
- ldr r4, =zreladdr
- #endif
- bl cache_on /* 开启缓存(以及MMU) */
- restart: adr r0, LC0
- ldmia r0, {r1, r2, r3, r6, r10, r11, r12}
- ldr sp, [r0, #28]
- /*
- * 我们可能运行在一个与编译时定义的不同地址上,
- * 所以我们必须修正变量指针
- */
- sub r0, r0, r1 @ 计算偏移量
- add r6, r6, r0 @ 重新计算_edata
- add r10, r10, r0 @ 重新获得压缩后的内核大小数据位置
- /*
- * 内核编译系统将解压后的内核大小数据
- * 以小端格式
- * 附加在压缩数据的后面(其实是“gzip -f -9”命令的结果)
- * 下面代码的作用是将解压后的内核大小数据正确地放入r9中(避免了大小端问题)
- */
- ldrb r9, [r10, #0]
- ldrb lr, [r10, #1]
- orr r9, r9, lr, lsl #8
- ldrb lr, [r10, #2]
- ldrb r10, [r10, #3]
- orr r9, r9, lr, lsl #16
- orr r9, r9, r10, lsl #24
- /*
- * 下面代码的作用是将正确的当前执行映像的结束地址放入r10
- */
- #ifndef CONFIG_ZBOOT_ROM
- /* malloc 获取的内存空间位于重定向的栈指针之上 (64k max) */
- add sp, sp, r0
- add r10, sp, #0x10000
- #else
- /*
- * 如果定义了 ZBOOT_ROM, bss/stack 是非可重定位的,
- * 但有些人依然可以将其放在RAM中运行,
- * 这时我们可以参考 _edata.
- */
- mov r10, r6
- #endif
- /*
- * 检测我们是否会发生自我覆盖的问题
- * r4 = 解压后的内核起始地址(最终执行位置)
- * r9 = 解压后内核的大小
- * r10 = 当前执行映像的结束地址, 包含了 bss/stack/malloc 空间(假设是非XIP执行的)
- * 我们的基本需求是:
- * (若最终执行位置r4在当前映像之后)r4 - 16k 页目录 >= r10 -> OK
- * (若最终执行位置r4在当前映像之前)r4 + 解压后的内核大小 <= 当前位置 (pc) -> OK
- * 如果上面的条件不满足,就会自我覆盖,必须先搬运当前映像
- */
- add r10, r10, #16384
- cmp r4, r10 @ 假设最终执行位置r4在当前映像之后
- bhs wont_overwrite
- add r10, r4, r9 @ 假设最终执行位置r4在当前映像之前
- ARM( cmp r10, pc ) @ r10 = 解压后的内核结束地址
- THUMB( mov lr, pc )
- THUMB( cmp r10, lr )
- bls wont_overwrite
- /*
- * 将当前的映像重定向到解压后的内核之后(会发生自我覆盖时才执行,否则就被跳过)
- * r6 = _edata(已校正)
- * r10 = 解压后的内核结束地址
- * 因为我们要把当前映像向后移动, 所以我们必须由后往前复制代码,
- * 以防原数据和目标数据的重叠
- */
- /*
- * 将解压后的内核结束地址r10扩展(reloc_code_end - restart),
- * 并对齐到下一个256B边界。
- * 这样避免了当搬运的偏移较小时的自我覆盖
- */
- add r10, r10, #((reloc_code_end - restart + 256) & ~255)
- bic r10, r10, #255
- /* 获取需要搬运的当前映像的起始位置r5,并向下做32B对齐. */
- adr r5, restart
- bic r5, r5, #31
- sub r9, r6, r5 @ _edata - restart(已向下对齐)= 需要搬运的大小
- add r9, r9, #31
- bic r9, r9, #31 @ 做32B对齐 ,r9 = 需要搬运的大小
- add r6, r9, r5 @ r6 = 当前映像需要搬运的结束地址
- add r9, r9, r10 @ r9 = 当前映像搬运的目的地的结束地址
- /* 搬运当前执行映像,不包含 bss/stack/malloc 空间*/
- 1: ldmdb r6!, {r0 - r3, r10 - r12, lr}
- cmp r6, r5
- stmdb r9!, {r0 - r3, r10 - r12, lr}
- bhi 1b
- /* 保存偏移量,用来修改sp和实现代码跳转 */
- sub r6, r9, r6
- #ifndef CONFIG_ZBOOT_ROM
- /* cache_clean_flush 可能会使用栈,所以重定向sp指针 */
- add sp, sp, r6
- #endif
- bl cache_clean_flush @ 刷新缓存
- /* 通过搬运的偏移和当前的实际 restart 地址来实现代码跳转*/
- adr r0, BSYM(restart)
- add r0, r0, r6
- mov pc, r0
- /* 在上面的跳转之后,程序又从restart开始。
- * 但这次在检查自我覆盖的时候,新的执行位置必然满足
- * 最终执行位置r4在当前映像之前,r4 + 压缩后的内核大小 <= 当前位置 (pc)
- * 所以必然直接跳到了下面的wont_overwrite执行
- */
- wont_overwrite:
- /*
- * 如果delta(当前映像地址与编译时的地址偏移)为0, 我们运行的地址就是编译时确定的地址.
- * r0 = delta
- * r2 = BSS start(编译值)
- * r3 = BSS end(编译值)
- * r4 = 内核最终运行的物理地址
- * r7 = 构架ID(bootlodaer传递值)
- * r8 = 内核启动参数指针(bootlodaer传递值)
- * r11 = GOT start(编译值)
- * r12 = GOT end(编译值)
- * sp = stack pointer(修正值)
- */
- teq r0, #0 @测试delta值
- beq not_relocated @如果delta为0,无须对GOT表项和BSS进行重定位
- add r11, r11, r0 @重定位GOT start
- add r12, r12, r0 @重定位GOT end
- #ifndef CONFIG_ZBOOT_ROM
- /*
- * 如果内核配置 CONFIG_ZBOOT_ROM = n,
- * 我们必须修正BSS段的指针
- * 注意:sp已经被修正
- */
- add r2, r2, r0 @重定位BSS start
- add r3, r3, r0 @重定位BSS end
- /*
- * 重定位所有GOT表的入口项
- */
- 1: ldr r1, [r11, #0] @ 重定位GOT表的入口项
- add r1, r1, r0 @ 这个修正了 C 引用
- str r1, [r11], #4
- cmp r11, r12
- blo 1b
- #else
- /*
- * 重定位所有GOT表的入口项.
- * 我们只重定向在(已重定向后)BSS段外的入口
- */
- 1: ldr r1, [r11, #0] @ 重定位GOT表的入口项
- cmp r1, r2 @ entry < bss_start ||
- cmphs r3, r1 @ _end < entry table
- addlo r1, r1, r0 @ 这个修正了 C 引用
- str r1, [r11], #4
- cmp r11, r12
- blo 1b
- #endif
- /*
- * 至此当前映像的搬运和调整已经完成
- * 可以开始真正的工作的
- */
- not_relocated: mov r0, #0
- 1: str r0, [r2], #4 @ 清零 bss(初始化BSS段)
- str r0, [r2], #4
- str r0, [r2], #4
- str r0, [r2], #4
- cmp r2, r3
- blo 1b
- /*
- * C运行时环境已经充分建立.
- * 设置一些指针就可以解压内核了.
- * r4 = 内核最终运行的物理地址
- * r7 = 构架ID
- * r8 = 内核启动参数指针
- *
- * 下面对r0~r3的配置是decompress_kernel函数对应参数
- * r0 = 解压后的输出位置首地址
- * r1 = 可用RAM空间首地址
- * r2 = 可用RAM空间结束地址
- * r3 = 构架ID
- * 就是这个decompress_kernel(C函数)输出了"Uncompressing Linux..."
- * 以及" done, booting the kernel.\n"
- */
- mov r0, r4
- mov r1, sp @ malloc 获取的内存空间位于栈指针之上
- add r2, sp, #0x10000 @ 64k max
- mov r3, r7
- bl decompress_kernel
- /*
- * decompress_kernel(misc.c)--调用-->
- * do_decompress(decompress.c)--调用-->
- * decompress(../../../../lib/decompress_xxxx.c根据压缩方式的配置而不同)
- */
- /*
- * 以下是为跳入解压后的内核,再次做准备(恢复解压前的状态)
- */
- bl cache_clean_flush
- bl cache_off @ 数据缓存必须关闭(内核的要求)
- mov r0, #0 @ r0必须为0
- mov r1, r7 @ 恢复构架ID到r1
- mov r2, r8 @ 恢复内核启动参数指针到r2
- mov pc, r4 @ 跳入解压后的内核映像(Image)入口(arch/arm/kernel/head.S)
- /*
- * 以下是为了确定当前运行时的地址和编译时确定的地址偏差,
- * 而将编译时确定的映像数据保存如下,用于检测对比
- */
- .align 2
- .type LC0, #object
- LC0: .word LC0 @ r1
- .word __bss_start @ r2
- .word _end @ r3
- .word _edata @ r6
- .word input_data_end - 4 @ r10 (inflated size location)
- .word _got_start @ r11
- .word _got_end @ ip
- .word .L_user_stack_end @ sp
- .size LC0, . - LC0
- #ifdef CONFIG_ARCH_RPC
- .globl params
- params: ldr r0, =0x10000100 @ params_phys for RPC
- mov pc, lr
- .ltorg
- .align
- #endif
- /*
- * 开启缓存.
- * 我们必须创建页表(并开启MMU)才可以开启数据和指令缓存。
- * 我们把页表(节描述符)放在内核执行地址前16k(0x4000)的空间中,
- * 且我们希望没人会去用这段地址空间.
- * 如果我们使用了,可能会出问题的!
- *
- * 进入时,
- * r4 = 内核最终运行的物理地址
- * r7 = 构架ID
- * r8 = 内核启动参数指针
- * 退出时,
- * r0, r1, r2, r3, r9, r10, r12 被修改
- * 此例程必须保护:
- * r4, r7, r8
- */
- .align 5
- cache_on: mov r3, #8 @ 调用cache_on 函数
- b call_cache_fn
- /*
- * Initialize the highest priority protection region, PR7
- * to cover all 32bit address and cacheable and bufferable.
- */
- __armv4_mpu_cache_on:
- mov r0, #0x3f @ 4G, the whole
- mcr p15, 0, r0, c6, c7, 0 @ PR7 Area Setting
- mcr p15, 0, r0, c6, c7, 1
- mov r0, #0x80 @ PR7
- mcr p15, 0, r0, c2, c0, 0 @ D-cache on
- mcr p15, 0, r0, c2, c0, 1 @ I-cache on
- mcr p15, 0, r0, c3, c0, 0 @ write-buffer on
- mov r0, #0xc000
- mcr p15, 0, r0, c5, c0, 1 @ I-access permission
- mcr p15, 0, r0, c5, c0, 0 @ D-access permission
- mov r0, #0
- mcr p15, 0, r0, c7, c10, 4 @ drain write buffer
- mcr p15, 0, r0, c7, c5, 0 @ flush(inval) I-Cache
- mcr p15, 0, r0, c7, c6, 0 @ flush(inval) D-Cache
- mrc p15, 0, r0, c1, c0, 0 @ read control reg
- @ ...I .... ..D. WC.M
- orr r0, r0, #0x002d @ .... .... ..1. 11.1
- orr r0, r0, #0x1000 @ ...1 .... .... ....
- mcr p15, 0, r0, c1, c0, 0 @ write control reg
- mov r0, #0
- mcr p15, 0, r0, c7, c5, 0 @ flush(inval) I-Cache
- mcr p15, 0, r0, c7, c6, 0 @ flush(inval) D-Cache
- mov pc, lr
- __armv3_mpu_cache_on:
- mov r0, #0x3f @ 4G, the whole
- mcr p15, 0, r0, c6, c7, 0 @ PR7 Area Setting
- mov r0, #0x80 @ PR7
- mcr p15, 0, r0, c2, c0, 0 @ cache on
- mcr p15, 0, r0, c3, c0, 0 @ write-buffer on
- mov r0, #0xc000
- mcr p15, 0, r0, c5, c0, 0 @ access permission
- mov r0, #0
- mcr p15, 0, r0, c7, c0, 0 @ invalidate whole cache v3
- /*
- * ?? ARMv3 MMU does not allow reading the control register,
- * does this really work on ARMv3 MPU?
- */
- mrc p15, 0, r0, c1, c0, 0 @ read control reg
- @ .... .... .... WC.M
- orr r0, r0, #0x000d @ .... .... .... 11.1
- /* ?? this overwrites the value constructed above? */
- mov r0, #0
- mcr p15, 0, r0, c1, c0, 0 @ write control reg
- /* ?? invalidate for the second time? */
- mcr p15, 0, r0, c7, c0, 0 @ invalidate whole cache v3
- mov pc, lr
- /*
- * 初始化MMU页表
- * 内核最终运行的物理地址向下16K的空间
- * 存放可以寻址4G空间节描述符
- * (16KB/4B=4K个描述符,每个描述符映射1MB空间,4K*1MB = 4GB)
- * 进入时,
- * r4 = 内核最终运行的物理地址
- * r7 = 构架ID
- * r8 = 内核启动参数指针
- * 退出时,
- * r0, r1, r2, r3, r9, r10 被修改
- * 此例程必须保护:
- * r4, r7, r8
- */
- __setup_mmu: sub r3, r4, #16384 @ 页目录大小为16K
- bic r3, r3, #0xff @ 页目录指针向下对齐
- bic r3, r3, #0x3f00 @ 对齐方式-16KB
- /*
- * 对于这个对齐,是MMU硬件的要求
- * 转换表基址寄存器(CP15的寄存器2)保存着第一级转换表基址的物理地址。
- * 只有bits[31:14]有效,bits[13:0]应该是零(SBZ)。
- * 所以第一级表必须16KB对齐。
- */
- /*
- * 初始化页表, 仅针对RAM(最大到256MB)开启
- * 缓存(cacheable)和缓冲(bufferable)位
- * r3 = 页目录基址(内核最终运行的物理地址向下16K的位置)
- */
- mov r0, r3 @ 页目录指针给r0
- mov r9, r0, lsr #18
- mov r9, r9, lsl #18 @ 通过移位清零低18bit,得到RAM基地址(推测值,r9)
- add r10, r9, #0x10000000 @ 加一个合理的RAM大小(猜测值) = RAM结束地址(猜测值,r10)
- mov r1, #0x12
- orr r1, r1, #3 << 10 @ 初始化节描述符r1 = 0b110000010010(完全访问:0域:XN:节)
- add r2, r3, #16384 @ r2 = 内核最终运行的物理地址(可能)
- 1: cmp r1, r9 @ if virt > start of RAM(针对RAM开启缓存和缓冲)
- #ifdef CONFIG_CPU_DCACHE_WRITETHROUGH
- orrhs r1, r1, #0x08 @ 设置 cacheable
- #else
- orrhs r1, r1, #0x0c @ 设置 cacheable, bufferable
- #endif
- cmp r1, r10 @ if virt > end of RAM
- bichs r1, r1, #0x0c @ 清除 cacheable, bufferable
- str r1, [r0], #4 @ 设置节描述符-1:1 映射(虚拟地址 == 物理地址)
- add r1, r1, #1048576 @ r1 + 1MB(每节管理的地址长度)下一个节描述符
- teq r0, r2
- bne 1b
- /*
- * 如果我们在flash中运行, 那么我们一定要为我们当前的代码开启缓存。
- * 我们映射2MB的代码,
- * 所以对于多达1MB压缩的内核没有映射重叠的问题??
- * 如果我们在RAM中运行, 那么我们只需要完成上面的工作即可,下面重复了.
- */
- mov r1, #0x1e
- orr r1, r1, #3 << 10 @ 初始化节描述符r1 = 0b110000011110(完全访问:0域:XN:cacheable:bufferable:节)
- mov r2, pc
- mov r2, r2, lsr #20 @ 当前执行地址的节基址
- orr r1, r1, r2, lsl #20 @ 生成节描述符
- add r0, r3, r2, lsl #2 @ 获得页目录中相应的入口
- str r1, [r0], #4 @ 设置节描述符-1:1 映射(虚拟地址 == 物理地址)
- add r1, r1, #1048576 @ r1 + 1MB(每节管理的地址长度)下一个节描述符
- str r1, [r0] @ 设置节描述符(只做2MB映射)
- mov pc, lr
- ENDPROC(__setup_mmu)
- __arm926ejs_mmu_cache_on:
- #ifdef CONFIG_CPU_DCACHE_WRITETHROUGH
- mov r0, #4 @ put dcache in WT mode
- mcr p15, 7, r0, c15, c0, 0
- #endif
- __armv4_mmu_cache_on:
- mov r12, lr
- #ifdef CONFIG_MMU
- bl __setup_mmu
- mov r0, #0
- mcr p15, 0, r0, c7, c10, 4 @ drain write buffer
- mcr p15, 0, r0, c8, c7, 0 @ flush I,D TLBs
- mrc p15, 0, r0, c1, c0, 0 @ read control reg
- orr r0, r0, #0x5000 @ I-cache enable, RR cache replacement
- orr r0, r0, #0x0030
- #ifdef CONFIG_CPU_ENDIAN_BE8
- orr r0, r0, #1 << 25 @ big-endian page tables
- #endif
- bl __common_mmu_cache_on
- mov r0, #0
- mcr p15, 0, r0, c8, c7, 0 @ flush I,D TLBs
- #endif
- mov pc, r12
- __armv7_mmu_cache_on:
- mov r12, lr @保存lr到r12
- #ifdef CONFIG_MMU
- mrc p15, 0, r11, c0, c1, 4 @ 读取CP15的ID_MMFR0(内存模块特性)寄存器
- tst r11, #0xf @ 测试VMSA(虚拟内存系统构架)A8 = 0x3
- blne __setup_mmu @ 如果VMSA不是0xf,就进入mmu页表初始化(节模式)
- mov r0, #0
- mcr p15, 0, r0, c7, c10, 4 @ 数据内存屏障(保证上面的写操作完成才继续)
- tst r11, #0xf @ 测试VMSA(虚拟内存系统构架)A8 = 0x3
- mcrne p15, 0, r0, c8, c7, 0 @ flush I,D TLBs缓存
- #endif
- mrc p15, 0, r0, c1, c0, 0 @ 读系统控制寄存器
- orr r0, r0, #0x5000 @ I-cache 使能, RR cache replacement
- orr r0, r0, #0x003c @ write buffer
- #ifdef CONFIG_MMU
- #ifdef CONFIG_CPU_ENDIAN_BE8
- orr r0, r0, #1 << 25 @ 大端模式页表
- #endif
- orrne r0, r0, #1 @ 设置MMU 开启位
- movne r1, #-1
- mcrne p15, 0, r3, c2, c0, 0 @ 载入页表基址到TTBR0
- mcrne p15, 0, r1, c3, c0, 0 @ 载入域访问控制数据到DACR(所有域都是Manager,所以XN会被忽略)
- #endif
- mcr p15, 0, r0, c1, c0, 0 @ 写系统控制寄存器
- mrc p15, 0, r0, c1, c0, 0 @ 回读系统控制寄存器
- mov r0, #0
- mcr p15, 0, r0, c7, c5, 4 @ 指令同步屏障(确保上面指令完成才返回)
- mov pc, r12 @ 此处返回(此时MMU已启用,RAM缓存已开启)
- __fa526_cache_on:
- mov r12, lr
- bl __setup_mmu
- mov r0, #0
- mcr p15, 0, r0, c7, c7, 0 @ Invalidate whole cache
- mcr p15, 0, r0, c7, c10, 4 @ drain write buffer
- mcr p15, 0, r0, c8, c7, 0 @ flush UTLB
- mrc p15, 0, r0, c1, c0, 0 @ read control reg
- orr r0, r0, #0x1000 @ I-cache enable
- bl __common_mmu_cache_on
- mov r0, #0
- mcr p15, 0, r0, c8, c7, 0 @ flush UTLB
- mov pc, r12
- __arm6_mmu_cache_on:
- mov r12, lr
- bl __setup_mmu
- mov r0, #0
- mcr p15, 0, r0, c7, c0, 0 @ invalidate whole cache v3
- mcr p15, 0, r0, c5, c0, 0 @ invalidate whole TLB v3
- mov r0, #0x30
- bl __common_mmu_cache_on
- mov r0, #0
- mcr p15, 0, r0, c5, c0, 0 @ invalidate whole TLB v3
- mov pc, r12
- __common_mmu_cache_on:
- #ifndef CONFIG_THUMB2_KERNEL
- #ifndef DEBUG
- orr r0, r0, #0x000d @ Write buffer, mmu
- #endif
- mov r1, #-1
- mcr p15, 0, r3, c2, c0, 0 @ load page table pointer
- mcr p15, 0, r1, c3, c0, 0 @ load domain access control
- b 1f
- .align 5 @ cache line aligned
- 1: mcr p15, 0, r0, c1, c0, 0 @ load control register
- mrc p15, 0, r0, c1, c0, 0 @ and read it back to
- sub pc, lr, r0, lsr #32 @ properly flush pipeline
- #endif
- #define PROC_ENTRY_SIZE (4*5)
- /*
- * 这里是为不同的处理器提供遵循可重定向缓存支持的函数
- * 这是一个通用的为 定位入口 和 跳入一个(从块起始处到)特定偏移的指令 的钩子函数。
- * 请注意这是一个位置无关代码。
- *
- * r1 = 被修改
- * r2 = 被修改
- * r3 = 相对每个入口的功能函数位置偏移(on:#08|off:#12|flush:#16)
- * r9 = 被修改
- * r12 = 被修改
- */
- call_cache_fn: adr r12, proc_types
- #ifdef CONFIG_CPU_CP15
- mrc p15, 0, r9, c0, c0 @ 动态获取处理器ID
- #else
- ldr r9, =CONFIG_PROCESSOR_ID @ 使用预编译的处理器ID
- #endif
- 1: ldr r1, [r12, #0] @ 获取ID值
- ldr r2, [r12, #4] @ 获取对应的掩码
- eor r1, r1, r9 @ (real ^ match) 检测是否匹配
- tst r1, r2 @ & mask 将检测结果做掩码
- ARM( addeq pc, r12, r3 ) @ 如果匹配就调用缓存函数
- THUMB( addeq r12, r3 )
- THUMB( moveq pc, r12 ) @ call cache function
- add r12, r12, #PROC_ENTRY_SIZE @ 如果不匹配就跳过这个入口,进入下个测试
- b 1b
- /*
- * 缓存操作表. 这些是最基本的:
- * - CPU ID 匹配
- * - CPU ID 掩码
- * - ‘cache on‘ 方法代码
- * - ‘cache off‘ 方法代码
- * - ‘cache flush‘ 方法代码
- *
- * 我们通过这个公式匹配入口: ((real_id ^ match) & mask) == 0
- *
- * 写通式缓存一般只需要 ‘on‘ 和 ‘off‘ 方法
- * 回写式缓存必须有 flush 方法定义
- *
- */
- .align 2
- .type proc_types,#object
- proc_types:
- .word 0x41560600 @ ARM6/610
- .word 0xffffffe0
- W(b) __arm6_mmu_cache_off @ 可以使用但是较慢
- W(b) __arm6_mmu_cache_off
- mov pc, lr
- THUMB( nop )
- @ b __arm6_mmu_cache_on @ 未测试
- @ b __arm6_mmu_cache_off
- @ b __armv3_mmu_cache_flush
- .word 0x00000000 @ old ARM ID
- .word 0x0000f000
- mov pc, lr
- THUMB( nop )
- mov pc, lr
- THUMB( nop )
- mov pc, lr
- THUMB( nop )
- .word 0x41007000 @ ARM7/710
- .word 0xfff8fe00
- W(b) __arm7_mmu_cache_off
- W(b) __arm7_mmu_cache_off
- mov pc, lr
- THUMB( nop )
- .word 0x41807200 @ ARM720T (写通式)
- .word 0xffffff00
- W(b) __armv4_mmu_cache_on
- W(b) __armv4_mmu_cache_off
- mov pc, lr
- THUMB( nop )
- .word 0x41007400 @ ARM74x
- .word 0xff00ff00
- W(b) __armv3_mpu_cache_on
- W(b) __armv3_mpu_cache_off
- W(b) __armv3_mpu_cache_flush
- .word 0x41009400 @ ARM94x
- .word 0xff00ff00
- W(b) __armv4_mpu_cache_on
- W(b) __armv4_mpu_cache_off
- W(b) __armv4_mpu_cache_flush
- .word 0x41069260 @ ARM926EJ-S (v5TEJ)
- .word 0xff0ffff0
- W(b) __arm926ejs_mmu_cache_on
- W(b) __armv4_mmu_cache_off
- W(b) __armv5tej_mmu_cache_flush
- .word 0x00007000 @ ARM7 IDs
- .word 0x0000f000
- mov pc, lr
- THUMB( nop )
- mov pc, lr
- THUMB( nop )
- mov pc, lr
- THUMB( nop )
- @ 以下使用新的 ID 系统.
- .word 0x4401a100 @ sa110 / sa1100
- .word 0xffffffe0
- W(b) __armv4_mmu_cache_on
- W(b) __armv4_mmu_cache_off
- W(b) __armv4_mmu_cache_flush
- .word 0x6901b110 @ sa1110
- .word 0xfffffff0
- W(b) __armv4_mmu_cache_on
- W(b) __armv4_mmu_cache_off
- W(b) __armv4_mmu_cache_flush
- .word 0x56056900
- .word 0xffffff00 @ PXA9xx
- W(b) __armv4_mmu_cache_on
- W(b) __armv4_mmu_cache_off
- W(b) __armv4_mmu_cache_flush
- .word 0x56158000 @ PXA168
- .word 0xfffff000
- W(b) __armv4_mmu_cache_on
- W(b) __armv4_mmu_cache_off
- W(b) __armv5tej_mmu_cache_flush
- .word 0x56050000 @ Feroceon
- .word 0xff0f0000
- W(b) __armv4_mmu_cache_on
- W(b) __armv4_mmu_cache_off
- W(b) __armv5tej_mmu_cache_flush
- #ifdef CONFIG_CPU_FEROCEON_OLD_ID
- /* this conflicts with the standard ARMv5TE entry */
- .long 0x41009260 @ Old Feroceon
- .long 0xff00fff0
- b __armv4_mmu_cache_on
- b __armv4_mmu_cache_off
- b __armv5tej_mmu_cache_flush
- #endif
- .word 0x66015261 @ FA526
- .word 0xff01fff1
- W(b) __fa526_cache_on
- W(b) __armv4_mmu_cache_off
- W(b) __fa526_cache_flush
- @ 这些匹配构架ID
- .word 0x00020000 @ ARMv4T
- .word 0x000f0000
- W(b) __armv4_mmu_cache_on
- W(b) __armv4_mmu_cache_off
- W(b) __armv4_mmu_cache_flush
- .word 0x00050000 @ ARMv5TE
- .word 0x000f0000
- W(b) __armv4_mmu_cache_on
- W(b) __armv4_mmu_cache_off
- W(b) __armv4_mmu_cache_flush
- .word 0x00060000 @ ARMv5TEJ
- .word 0x000f0000
- W(b) __armv4_mmu_cache_on
- W(b) __armv4_mmu_cache_off
- W(b) __armv5tej_mmu_cache_flush
- .word 0x0007b000 @ ARMv6
- .word 0x000ff000
- W(b) __armv4_mmu_cache_on
- W(b) __armv4_mmu_cache_off
- W(b) __armv6_mmu_cache_flush
- .word 0x000f0000 @ new CPU Id
- .word 0x000f0000
- W(b) __armv7_mmu_cache_on
- W(b) __armv7_mmu_cache_off
- W(b) __armv7_mmu_cache_flush
- .word 0 @ 未识别类型
- .word 0
- mov pc, lr
- THUMB( nop )
- mov pc, lr
- THUMB( nop )
- mov pc, lr
- THUMB( nop )
- .size proc_types, . - proc_types
- /*
- * 如果你获得了一个 "非常量的表达式".如果汇编器从这行返回" 申明"错误
- * 请检查下你是否偶尔在应该使用“W(b)”的地方写了"b"指令
- * 这是一个缓存方法跳转表的对齐检查机制
- * 在写汇编的时候可以借鉴
- */
- .if (. - proc_types) % PROC_ENTRY_SIZE != 0
- .error "The size of one or more proc_types entries is wrong."
- .endif
- /*
- * 关闭缓存和MMU. ARMv3不支持控制寄存器的读取,
- * 但ARMv4支持.
- *
- * 在退出时,
- * r0, r1, r2, r3, r9, r12 被篡改
- * 这个例程必须保护:
- * r4, r7, r8
- */
- .align 5
- cache_off: mov r3, #12 @ 缓存关闭函数
- b call_cache_fn
- __armv4_mpu_cache_off:
- mrc p15, 0, r0, c1, c0
- bic r0, r0, #0x000d
- mcr p15, 0, r0, c1, c0 @ turn MPU and cache off
- mov r0, #0
- mcr p15, 0, r0, c7, c10, 4 @ drain write buffer
- mcr p15, 0, r0, c7, c6, 0 @ flush D-Cache
- mcr p15, 0, r0, c7, c5, 0 @ flush I-Cache
- mov pc, lr
- __armv3_mpu_cache_off:
- mrc p15, 0, r0, c1, c0
- bic r0, r0, #0x000d
- mcr p15, 0, r0, c1, c0, 0 @ turn MPU and cache off
- mov r0, #0
- mcr p15, 0, r0, c7, c0, 0 @ invalidate whole cache v3
- mov pc, lr
- __armv4_mmu_cache_off:
- #ifdef CONFIG_MMU
- mrc p15, 0, r0, c1, c0
- bic r0, r0, #0x000d
- mcr p15, 0, r0, c1, c0 @ turn MMU and cache off
- mov r0, #0
- mcr p15, 0, r0, c7, c7 @ invalidate whole cache v4
- mcr p15, 0, r0, c8, c7 @ invalidate whole TLB v4
- #endif
- mov pc, lr
- __armv7_mmu_cache_off:
- mrc p15, 0, r0, c1, c0 @ 读取系统控制寄存器SCTLR
- #ifdef CONFIG_MMU
- bic r0, r0, #0x000d @ 清零MMU和cache使能位
- #else
- bic r0, r0, #0x000c @ 清零cache使能位
- #endif
- mcr p15, 0, r0, c1, c0 @ 关闭MMU和cache
- mov r12, lr @ 保存lr到r12
- bl __armv7_mmu_cache_flush
- mov r0, #0
- #ifdef CONFIG_MMU
- mcr p15, 0, r0, c8, c7, 0 @ 废止整个TLB
- #endif
- mcr p15, 0, r0, c7, c5, 6 @ 废止BTC
- mcr p15, 0, r0, c7, c10, 4 @ 数据同步屏障
- mcr p15, 0, r0, c7, c5, 4 @ 指令同步屏障(确保上面指令完成才返回)
- mov pc, r12
- __arm6_mmu_cache_off:
- mov r0, #0x00000030 @ ARM6 control reg.
- b __armv3_mmu_cache_off
- __arm7_mmu_cache_off:
- mov r0, #0x00000070 @ ARM7 control reg.
- b __armv3_mmu_cache_off
- __armv3_mmu_cache_off:
- mcr p15, 0, r0, c1, c0, 0 @ turn MMU and cache off
- mov r0, #0
- mcr p15, 0, r0, c7, c0, 0 @ invalidate whole cache v3
- mcr p15, 0, r0, c5, c0, 0 @ invalidate whole TLB v3
- mov pc, lr
- /*
- * 清空和flush缓存以保持一致性
- *
- * 退出时,
- * r1, r2, r3, r9, r10, r11, r12 被篡改
- * 这个例程必须保护:
- * r4, r6, r7, r8
- */
- .align 5
- cache_clean_flush:
- mov r3, #16
- b call_cache_fn
- __armv4_mpu_cache_flush:
- mov r2, #1
- mov r3, #0
- mcr p15, 0, ip, c7, c6, 0 @ invalidate D cache
- mov r1, #7 << 5 @ 8 segments
- 1: orr r3, r1, #63 << 26 @ 64 entries
- 2: mcr p15, 0, r3, c7, c14, 2 @ clean & invalidate D index
- subs r3, r3, #1 << 26
- bcs 2b @ entries 63 to 0
- subs r1, r1, #1 << 5
- bcs 1b @ segments 7 to 0
- teq r2, #0
- mcrne p15, 0, ip, c7, c5, 0 @ invalidate I cache
- mcr p15, 0, ip, c7, c10, 4 @ drain WB
- mov pc, lr
- __fa526_cache_flush:
- mov r1, #0
- mcr p15, 0, r1, c7, c14, 0 @ clean and invalidate D cache
- mcr p15, 0, r1, c7, c5, 0 @ flush I cache
- mcr p15, 0, r1, c7, c10, 4 @ drain WB
- mov pc, lr
- __armv6_mmu_cache_flush:
- mov r1, #0
- mcr p15, 0, r1, c7, c14, 0 @ clean+invalidate D
- mcr p15, 0, r1, c7, c5, 0 @ invalidate I+BTB
- mcr p15, 0, r1, c7, c15, 0 @ clean+invalidate unified
- mcr p15, 0, r1, c7, c10, 4 @ drain WB
- mov pc, lr
- __armv7_mmu_cache_flush:
- mrc p15, 0, r10, c0, c1, 5 @ read ID_MMFR1
- tst r10, #0xf << 16 @ hierarchical cache (ARMv7)
- mov r10, #0
- beq hierarchical
- mcr p15, 0, r10, c7, c14, 0 @ clean+invalidate D
- b iflush
- hierarchical:
- mcr p15, 0, r10, c7, c10, 5 @ DMB
- stmfd sp!, {r0-r7, r9-r11}
- mrc p15, 1, r0, c0, c0, 1 @ read clidr
- ands r3, r0, #0x7000000 @ extract loc from clidr
- mov r3, r3, lsr #23 @ left align loc bit field
- beq finished @ if loc is 0, then no need to clean
- mov r10, #0 @ start clean at cache level 0
- loop1:
- add r2, r10, r10, lsr #1 @ work out 3x current cache level
- mov r1, r0, lsr r2 @ extract cache type bits from clidr
- and r1, r1, #7 @ mask of the bits for current cache only
- cmp r1, #2 @ see what cache we have at this level
- blt skip @ skip if no cache, or just i-cache
- mcr p15, 2, r10, c0, c0, 0 @ select current cache level in cssr
- mcr p15, 0, r10, c7, c5, 4 @ isb to sych the new cssr&csidr
- mrc p15, 1, r1, c0, c0, 0 @ read the new csidr
- and r2, r1, #7 @ extract the length of the cache lines
- add r2, r2, #4 @ add 4 (line length offset)
- ldr r4, =0x3ff
- ands r4, r4, r1, lsr #3 @ find maximum number on the way size
- clz r5, r4 @ find bit position of way size increment
- ldr r7, =0x7fff
- ands r7, r7, r1, lsr #13 @ extract max number of the index size
- loop2:
- mov r9, r4 @ create working copy of max way size
- loop3:
- ARM( orr r11, r10, r9, lsl r5 ) @ factor way and cache number into r11
- ARM( orr r11, r11, r7, lsl r2 ) @ factor index number into r11
- THUMB( lsl r6, r9, r5 )
- THUMB( orr r11, r10, r6 ) @ factor way and cache number into r11
- THUMB( lsl r6, r7, r2 )
- THUMB( orr r11, r11, r6 ) @ factor index number into r11
- mcr p15, 0, r11, c7, c14, 2 @ clean & invalidate by set/way
- subs r9, r9, #1 @ decrement the way
- bge loop3
- subs r7, r7, #1 @ decrement the index
- bge loop2
- skip:
- add r10, r10, #2 @ increment cache number
- cmp r3, r10
- bgt loop1
- finished:
- ldmfd sp!, {r0-r7, r9-r11}
- mov r10, #0 @ swith back to cache level 0
- mcr p15, 2, r10, c0, c0, 0 @ select current cache level in cssr
- iflush:
- mcr p15, 0, r10, c7, c10, 4 @ DSB
- mcr p15, 0, r10, c7, c5, 0 @ invalidate I+BTB
- mcr p15, 0, r10, c7, c10, 4 @ DSB
- mcr p15, 0, r10, c7, c5, 4 @ ISB
- mov pc, lr
- __armv5tej_mmu_cache_flush:
- 1: mrc p15, 0, r15, c7, c14, 3 @ test,clean,invalidate D cache
- bne 1b
- mcr p15, 0, r0, c7, c5, 0 @ flush I cache
- mcr p15, 0, r0, c7, c10, 4 @ drain WB
- mov pc, lr
- __armv4_mmu_cache_flush:
- mov r2, #64*1024 @ default: 32K dcache size (*2)
- mov r11, #32 @ default: 32 byte line size
- mrc p15, 0, r3, c0, c0, 1 @ read cache type
- teq r3, r9 @ cache ID register present?
- beq no_cache_id
- mov r1, r3, lsr #18
- and r1, r1, #7
- mov r2, #1024
- mov r2, r2, lsl r1 @ base dcache size *2
- tst r3, #1 << 14 @ test M bit
- addne r2, r2, r2, lsr #1 @ +1/2 size if M == 1
- mov r3, r3, lsr #12
- and r3, r3, #3
- mov r11, #8
- mov r11, r11, lsl r3 @ cache line size in bytes
- no_cache_id:
- mov r1, pc
- bic r1, r1, #63 @ align to longest cache line
- add r2, r1, r2
- 1:
- ARM( ldr r3, [r1], r11 ) @ s/w flush D cache
- THUMB( ldr r3, [r1] ) @ s/w flush D cache
- THUMB( add r1, r1, r11 )
- teq r1, r2
- bne 1b
- mcr p15, 0, r1, c7, c5, 0 @ flush I cache
- mcr p15, 0, r1, c7, c6, 0 @ flush D cache
- mcr p15, 0, r1, c7, c10, 4 @ drain WB
- mov pc, lr
- __armv3_mmu_cache_flush:
- __armv3_mpu_cache_flush:
- mov r1, #0
- mcr p15, 0, r1, c7, c0, 0 @ invalidate whole cache v3
- mov pc, lr
- /*
- * Various debugging routines for printing hex characters and
- * memory, which again must be relocatable.
- */
- #ifdef DEBUG
- .align 2
- .type phexbuf,#object
- phexbuf: .space 12
- .size phexbuf, . - phexbuf
- @ phex corrupts {r0, r1, r2, r3}
- phex: adr r3, phexbuf
- mov r2, #0
- strb r2, [r3, r1]
- 1: subs r1, r1, #1
- movmi r0, r3
- bmi puts
- and r2, r0, #15
- mov r0, r0, lsr #4
- cmp r2, #10
- addge r2, r2, #7
- add r2, r2, #‘0‘
- strb r2, [r3, r1]
- b 1b
- @ puts corrupts {r0, r1, r2, r3}
- puts: loadsp r3, r1
- 1: ldrb r2, [r0], #1
- teq r2, #0
- moveq pc, lr
- 2: writeb r2, r3
- mov r1, #0x00020000
- 3: subs r1, r1, #1
- bne 3b
- teq r2, #‘\n‘
- moveq r2, #‘\r‘
- beq 2b
- teq r0, #0
- bne 1b
- mov pc, lr
- @ putc corrupts {r0, r1, r2, r3}
- putc:
- mov r2, r0
- mov r0, #0
- loadsp r3, r1
- b 2b
- @ memdump corrupts {r0, r1, r2, r3, r10, r11, r12, lr}
- memdump: mov r12, r0
- mov r10, lr
- mov r11, #0
- 2: mov r0, r11, lsl #2
- add r0, r0, r12
- mov r1, #8
- bl phex
- mov r0, #‘:‘
- bl putc
- 1: mov r0, #‘ ‘
- bl putc
- ldr r0, [r12, r11, lsl #2]
- mov r1, #8
- bl phex
- and r0, r11, #7
- teq r0, #3
- moveq r0, #‘ ‘
- bleq putc
- and r0, r11, #7
- add r11, r11, #1
- teq r0, #7
- bne 1b
- mov r0, #‘\n‘
- bl putc
- cmp r11, #64
- blt 2b
- mov pc, r10
- #endif
- .ltorg
- reloc_code_end:
- .align
- .section ".stack", "aw", %nobits
- .L_user_stack: .space 4096
- .L_user_stack_end:
看了上面的源码,可能就算是分析过了也是比较模糊的,通过下面的一个代码流程图,大家就可以清楚的了解内核自解压的全过程了: