文章代码分析基于linux-5.19.13,架构基于aarch64(ARM64)。
涉及页表代码分析部分:
(1)假设页表映射层级是4,即配置CONFIG_ARM64_PGTABLE_LEVELS=4;
(2)虚拟地址宽度是48,即配置CONFIG_ARM64_VA_BITS=48;
(3)物理地址宽度是48,即配置CONFIG_ARM64_PA_BITS=48;
1. 入口分析
1.1 链接脚本arch/arm64/kernel/vmlinux.lds.S
这里只列举与内存初始化相关的定义,其它的采用“......”省略。
......
OUTPUT_ARCH(aarch64) '指定一个特定的输出机器架构为aarch64'
ENTRY(_text) '设置入口地址,实现在arch/arm64/kernel/head.S'
......
SECTIONS
{
......
'在5.8内核版本发现TEXT_OFFSET没有任何作用,因此,被重新定义为0x0'
. = KIMAGE_VADDR; '内核映像虚拟的起始地址(在5.8内核之前这里为KIMAGE_VADDR + TEXT_OFFSET)'
.head.text : { '早期汇编代码的text段'
_text = .; '入口地址'
HEAD_TEXT 定义在include/asm-generic/vmlinux.lds.h'#define HEAD_TEXT KEEP(*(.head.text))'
}
.text : ALIGN(SEGMENT_ALIGN) { /* Real text segment */
_stext = .; /* Text and read-only data */ 'text段起始'
......
}
......
. = ALIGN(SEGMENT_ALIGN);
_etext = .; /* End of text section */ 'text段结束'
/* everything from this point to __init_begin will be marked RO NX */
RO_DATA(PAGE_SIZE) '只读数据段'
......
idmap_pg_dir = .; '恒等映射一级页表地址'
. += IDMAP_DIR_SIZE;
idmap_pg_end = .;
#ifdef CONFIG_UNMAP_KERNEL_AT_EL0
tramp_pg_dir = .; '熔断(安全漏洞引入)'
. += PAGE_SIZE;
#endif
reserved_pg_dir = .;
. += PAGE_SIZE;
swapper_pg_dir = .;
. += PAGE_SIZE;
. = ALIGN(SEGMENT_ALIGN);
__init_begin = .; 'init段起始'
__inittext_begin = .;
......
. = ALIGN(SEGMENT_ALIGN);
__initdata_end = .;
__init_end = .; 'init段结束'
_data = .;
_sdata = .; '数据段起始'
RW_DATA(L1_CACHE_BYTES, PAGE_SIZE, THREAD_ALIGN)
_edata = .; '数据段结束'
BSS_SECTION(SBSS_ALIGN, 0, 0) --- 'BSS段'
. = ALIGN(PAGE_SIZE);
init_pg_dir = .;
. += INIT_DIR_SIZE;
init_pg_end = .;
......
}
1.2 入口
#arch/arm64/kernel/head.S
/*
* Kernel startup entry point.
* ---------------------------
*
* The requirements are:
* MMU = off, D-cache = off, I-cache = on or off,
* x0 = physical address to the FDT blob.
*
* This code is mostly position independent so you call this at
* __pa(PAGE_OFFSET).
*
* Note that the callee-saved registers are used for storing variables
* that are useful before the MMU is enabled. The allocations are described
* in the entry routines.
*/
__HEAD --- 定义在include/linux/init.h中'#define __HEAD .section ".head.text","ax"',紧接着_text
/*
* DO NOT MODIFY. Image header expected by Linux boot-loaders.
*/
efi_signature_nop // special NOP to identity as PE/COFF executable
b primary_entry // branch to kernel start, magic '要重点关注分析的启动汇编代码'
......
1.3 启动 AArch64 Linux的调用约定
内核从上电开始到执行到内核入口"_text",中间要经过bootloader或者bios的引导。引导程序会做一些初始化内存,设置device tree,解压内核,跳转到内核等等。在跳转到内核之前,有一些标准的约定,参见Documentation/translations/zh_CN/arm64/booting.txt。这里仅列出在跳转入内核前,必须符合以下章节的状态:
在跳转入内核前,必须符合以下状态:
- 停止所有 DMA 设备,这样内存数据就不会因为虚假网络包或磁盘数据而 被破坏。这可能可以节省你许多的调试时间。
- 主 CPU 通用寄存器设置 x0 = 系统 RAM 中设备树数据块(dtb)的物理地址。
x1 = 0 (保留,将来可能使用)
x2 = 0 (保留,将来可能使用)
x3 = 0 (保留,将来可能使用)
- CPU 模式 所有形式的中断必须在 PSTATE.DAIF 中被屏蔽(Debug、SError、IRQ 和 FIQ)。
CPU 必须处于 EL2(推荐,可访问虚拟化扩展)或非安全 EL1 模式下。'bootloader来切'
- 高速缓存、MMU MMU 必须关闭。 'mmu关闭,指令高速缓存一般可以打开,数据高速缓存必须关闭'
指令缓存开启或关闭皆可。 已载入的内核映像的相应内存区必须被清理,以达到缓存一致性点(PoC)。 当存在系统缓存或其他使能缓存的一致性主控器时,通常需使用虚拟地址 维护其缓存,而非 set/way 操作。 遵从通过虚拟地址操作维护构架缓存的系统缓存必须被配置,并可以被使能。 而不通过虚拟地址操作维护构架缓存的系统缓存(不推荐),必须被配置且 禁用。
*译者注:对于 PoC 以及缓存相关内容,请参考 ARMv8 构架参考手册 ARM DDI 0487A
- 架构计时器 CNTFRQ 必须设定为计时器的频率,且 CNTVOFF 必须设定为对所有 CPU 都一致的值。如果在 EL1 模式下进入内核,则 CNTHCTL_EL2 中的 EL1PCTEN (bit 0) 必须置位。
- 一致性 通过内核启动的所有 CPU 在内核入口地址上必须处于相同的一致性域中。 这可能要根据具体实现来定义初始化过程,以使能每个CPU上对维护操作的 接收。
- 系统寄存器 在进入内核映像的异常级中,所有构架中可写的系统寄存器必须通过软件 在一个更高的异常级别下初始化,以防止在 未知 状态下运行。
对于拥有 GICv3 中断控制器并以 v3 模式运行的系统:
- 如果 EL3 存在: ICC_SRE_EL3.Enable (位 3) 必须初始化为 0b1。
ICC_SRE_EL3.SRE (位 0) 必须初始化为 0b1。
- 若内核运行在 EL1: ICC_SRE_EL2.Enable (位 3) 必须初始化为 0b1。
ICC_SRE_EL2.SRE (位 0) 必须初始化为 0b1。
- 设备树(DT)或 ACPI 表必须描述一个 GICv3 中断控制器。
对于拥有 GICv3 中断控制器并以兼容(v2)模式运行的系统:
- 如果 EL3 存在: ICC_SRE_EL3.SRE (位 0) 必须初始化为 0b0。
- 若内核运行在 EL1: ICC_SRE_EL2.SRE (位 0) 必须初始化为 0b0。
- 设备树(DT)或 ACPI 表必须描述一个 GICv2 中断控制器。
这里有个很关键的问题:为什么跳转到内核时指令高速缓存可以打开,数据高速缓存必须关闭?
(1)CPU启动取数据的时候首先去访问数据高速缓存,这个数据高速缓存有可能缓存了bootloader的数据,这个数据对于内核可能是错误的。因此数据高速缓存必须关闭。
(2)bootloader和内核的指令无冲突。因为bootloader指令运行完成后不会再次运行,直接运行内核的指令。因此指令高速缓存可以不关闭。
2. 启动汇编接口primary_entry分析
/*
* The following callee saved general purpose registers are used on the
* primary lowlevel boot path:
*
* Register Scope Purpose
* x21 primary_entry() .. start_kernel() FDT pointer passed at boot in x0
* x23 primary_entry() .. start_kernel() physical misalignment/KASLR offset
* x28 __create_page_tables() callee preserved temp register
* x19/x20 __primary_switch() callee preserved temp registers
* x24 __primary_switch() .. relocate_kernel() current RELR displacement
*/
SYM_CODE_START(primary_entry)
bl preserve_boot_args
bl init_kernel_el // w0=cpu_boot_mode
adrp x23, __PHYS_OFFSET --- '__PHYS_OFFSET加载到x23寄存器'
and x23, x23, MIN_KIMG_ALIGN - 1 // KASLR offset, defaults to 0
bl set_cpu_boot_mode_flag
bl __create_page_tables
/*
* The following calls CPU setup code, see arch/arm64/mm/proc.S for
* details.
* On return, the CPU will be ready for the MMU to be turned on and
* the TCR will have been set.
*/
bl __cpu_setup // initialise processor
b __primary_switch
SYM_CODE_END(primary_entry)
2.1 preserve_boot_args
功能:把bootloader传进来的x0 .. x3保存到boot_args数组中。
/*
* Preserve the arguments passed by the bootloader in x0 .. x3
*/
SYM_CODE_START_LOCAL(preserve_boot_args)
mov x21, x0 // x21=FDT(x0寄存器保存devicetree的地址),devicetree保存到x21寄存器
adr_l x0, boot_args // record the contents of. boot_args数组地址保存到x0
stp x21, x1, [x0] // x0 .. x3 at kernel entry
stp x2, x3, [x0, #16] '参数x0 .. x3保存到boot_args数组中'
dmb sy // needed before dc ivac with.内存屏障指令(+sy表示全系统高速缓存范围内做一次内存屏障)
// MMU off
add x1, x0, #0x20 // 4 x 8 bytes
b dcache_inval_poc // tail call 清除boot_args数组对应的高速缓存
SYM_CODE_END(preserve_boot_args)
2.2 init_kernel_el
判断启动的模式是EL2还是非安全模式的EL1,并进行相关级别的系统配置(ARMv8中EL2是hypervisor模式,EL1是标准的内核模式),然后使用w0返回启动模式(BOOT_CPU_MODE_EL1或BOOT_CPU_MODE_EL2)。通常来讲系统启动时运行在EL3,uboot会把处理器置于EL2,内核运行到init_kernel_el会设为EL1。
/*
* Starting from EL2 or EL1, configure the CPU to execute at the highest
* reachable EL supported by the kernel in a chosen default state. If dropping
* from EL2 to EL1, configure EL2 before configuring EL1.
*
* Since we cannot always rely on ERET synchronizing writes to sysregs (e.g. if
* SCTLR_ELx.EOS is clear), we place an ISB prior to ERET.
*
* Returns either BOOT_CPU_MODE_EL1 or BOOT_CPU_MODE_EL2 in w0 if
* booted in EL1 or EL2 respectively.
*/
SYM_FUNC_START(init_kernel_el)
mrs x0, CurrentEL '获取当前PSTATE异常等级'
cmp x0, #CurrentEL_EL2
b.eq init_el2 '如果PSTATE异常等级为EL2,则跳转到init_el2'
SYM_INNER_LABEL(init_el1, SYM_L_LOCAL)
mov_q x0, INIT_SCTLR_EL1_MMU_OFF
msr sctlr_el1, x0
isb '因为前面修改了系统控制器'
mov_q x0, INIT_PSTATE_EL1
msr spsr_el1, x0
msr elr_el1, lr
mov w0, #BOOT_CPU_MODE_EL1
eret 'Return from exception'
SYM_INNER_LABEL(init_el2, SYM_L_LOCAL) --- 'EL2切向EL1'
......
msr elr_el1, x0
eret
1:
......
mov w0, #BOOT_CPU_MODE_EL2
eret
__cpu_stick_to_vhe:
mov x0, #HVC_VHE_RESTART
hvc #0
mov x0, #BOOT_CPU_MODE_EL2
ret
SYM_FUNC_END(init_kernel_el)
2.3 set_cpu_boot_mode_flag
根据w0中传递的cpu启动模式设置__boot_cpu_mode标志。
/*
* Sets the __boot_cpu_mode flag depending on the CPU boot mode passed
* in w0. See arch/arm64/include/asm/virt.h for more info.
*/
SYM_FUNC_START_LOCAL(set_cpu_boot_mode_flag)
adr_l x1, __boot_cpu_mode //x1记录__boot_cpu_mode[]的地址
cmp w0, #BOOT_CPU_MODE_EL2 //w0记录启动时的异常等级
b.ne 1f //如果不是从EL2启动,则跳转到1处
add x1, x1, #4 // 如果是从EL2启动,地址指向__boot_cpu_mode[1]
1: str w0, [x1] // Save CPU boot mode 保存启动模式到x1指向的地址,如果是从EL1启动,地址指向__boot_cpu_mode[0]
dmb sy // 保证str指令执行完成
dc ivac, x1 // Invalidate potentially stale cache line 使高速缓存失效
ret
SYM_FUNC_END(set_cpu_boot_mode_flag)
2.4 __create_page_tables
/*
* Setup the initial page tables. We only setup the barest amount which is
* required to get the kernel running. The following sections are required:
* - identity mapping to enable the MMU (low address, TTBR0) (1)恒等映射
* - first few MB of the kernel linear mapping to jump to once the MMU has
* been enabled (2)内核image映射
*/
SYM_FUNC_START_LOCAL(__create_page_tables)
...
SYM_FUNC_END(__create_page_tables)
2.4.1 保存LR值
mov x28, lr //#把LR的值存放到X28
2.4.2 使初始化页表无效、并清空初始化页表
/*
* Invalidate the init page tables to avoid potential dirty cache lines
* being evicted. Other page tables are allocated in rodata as part of
* the kernel image, and thus are clean to the PoC per the boot
* protocol.
*/
adrp x0, init_pg_dir //把init_pg_dir的物理地址赋值给x0
adrp x1, init_pg_end //把init_pg_end的物理地址赋值给x1
bl dcache_inval_poc //把init_pg_dir页表对应的高速缓存清掉(入参是x0和x1)
/*
* Clear the init page tables.//把这个页表内容清零
*/
adrp x0, init_pg_dir
adrp x1, init_pg_end
sub x1, x1, x0
1: stp xzr, xzr, [x0], #16 //xzr是零寄存器
stp xzr, xzr, [x0], #16
stp xzr, xzr, [x0], #16
stp xzr, xzr, [x0], #16
subs x1, x1, #64
b.ne 1b
(1)init_pg_dir和init_pg_end定义在arch/arm64/kernel/vmlinux.lds.S链接文件中:
#arch/arm64/kernel/vmlinux.lds.S
. = ALIGN(PAGE_SIZE);
init_pg_dir = .;
. += INIT_DIR_SIZE;
init_pg_end = .;
(2)adrp指令
作用:以页为单位的大范围的地址读取指令,这里的P就是page的意思。
原理:符号扩展一个21位的offset(immhi+immlo), 向左移动12位,PC的值的低12位清零,然后把这两者相加,结果写入到Xd寄存器,用来得到一块含有lable的4KB对齐内存区域的base地址(也就是说lable所在的地址,一定落在这个4KB的内存区域里,指令助记符里Page也就是这个意思), 可用来寻址 +/- 4GB的范围(2^33次幂)。
通俗来讲,ADRP指令就是先进行PC+imm(偏移值)。然后找到lable所在的一个4KB的页,然后取得label的基址,再进行偏移去寻址。
ADRP {cond} Rd label
其中:Rd加载的目标寄存器。lable为地址表达式。
(3)使用adrp指令获取init_pg_dir和init_pg_end的地址,页大小为4KB,由于内核启动的时候MMU还未打开(PC为物理地址),因此此时获取的地址也为物理地址。
(4)adrp通过当前PC地址的偏移地址计算目标地址,和实际的物理地址无关,因此属于位置无关码。
2.4.3 保存SWAPPER_MM_MMUFLAGS到x7寄存器
mov_q x7, SWAPPER_MM_MMUFLAGS
SWAPPER_MM_MMUFLAGS宏描述了段映射的属性,它实现在arch/arm64/include/asm/kernel-pgtable.h头文件中:
/*
* Initial memory map attributes.
*/
#define SWAPPER_PTE_FLAGS (PTE_TYPE_PAGE | PTE_AF | PTE_SHARED | PTE_UXN)
#define SWAPPER_PMD_FLAGS (PMD_TYPE_SECT | PMD_SECT_AF | PMD_SECT_S | PMD_SECT_UXN)
#if ARM64_KERNEL_USES_PMD_MAPS
#define SWAPPER_MM_MMUFLAGS (PMD_ATTRINDX(MT_NORMAL) | SWAPPER_PMD_FLAGS) //段映射,这里要使用的
#else
#define SWAPPER_MM_MMUFLAGS (PTE_ATTRINDX(MT_NORMAL) | SWAPPER_PTE_FLAGS) //页映射
#endif
2.4.4 创建恒等映射
/*
* Create the identity mapping.
*/
adrp x0, idmap_pg_dir ---(1)
adrp x3, __idmap_text_start // __pa(__idmap_text_start) ---(2)
#ifdef CONFIG_ARM64_VA_BITS_52 ---(3)
mrs_s x6, SYS_ID_AA64MMFR2_EL1
and x6, x6, #(0xf << ID_AA64MMFR2_LVA_SHIFT)
mov x5, #52
cbnz x6, 1f
#endif
mov x5, #VA_BITS_MIN ---(4)
1:
adr_l x6, vabits_actual ---(5)
str x5, [x6]
dmb sy //内存屏障
dc ivac, x6 // Invalidate potentially stale cache line 把vabits_actual变量对应的缓存给clean掉
/*
* VA_BITS may be too small to allow for an ID mapping to be created
* that covers system RAM if that is located sufficiently high in the
* physical address space. So for the ID map, use an extended virtual
* range in that case, and configure an additional translation level
* if needed.
*
* Calculate the maximum allowed value for TCR_EL1.T0SZ so that the
* entire ID map region can be mapped. As T0SZ == (64 - #bits used),
* this number conveniently equals the number of leading zeroes in
* the physical address of __idmap_text_end.
*/
adrp x5, __idmap_text_end ---(6)
clz x5, x5 //前导0计数:第一个1前0的个数 ---(6)
cmp x5, TCR_T0SZ(VA_BITS_MIN) // default T0SZ small enough? ---(6)
b.ge 1f // .. then skip VA range extension
adr_l x6, idmap_t0sz
str x5, [x6]
dmb sy
dc ivac, x6 // Invalidate potentially stale cache line
#if (VA_BITS < 48)
#define EXTRA_SHIFT (PGDIR_SHIFT + PAGE_SHIFT - 3)
#define EXTRA_PTRS (1 << (PHYS_MASK_SHIFT - EXTRA_SHIFT))
/*
* If VA_BITS < 48, we have to configure an additional table level.
* First, we have to verify our assumption that the current value of
* VA_BITS was chosen such that all translation levels are fully
* utilised, and that lowering T0SZ will always result in an additional
* translation level to be configured.
*/
#if VA_BITS != EXTRA_SHIFT
#error "Mismatch between VA_BITS and page size/number of translation levels"
#endif
mov x4, EXTRA_PTRS
create_table_entry x0, x3, EXTRA_SHIFT, x4, x5, x6
#else
/*
* If VA_BITS == 48, we don't have to configure an additional
* translation level, but the top-level table has more entries.
*/
mov x4, #1 << (PHYS_MASK_SHIFT - PGDIR_SHIFT)
str_l x4, idmap_ptrs_per_pgd, x5
#endif
1:
ldr_l x4, idmap_ptrs_per_pgd ---(7)
adr_l x6, __idmap_text_end // __pa(__idmap_text_end) ---(8)
map_memory x0, x1, x3, x6, x7, x3, x4, x10, x11, x12, x13, x14 ---(9)
- 将加载idmap_pg_dir的物理地址到x0寄存器,idmap_pg_dir是恒等映射的一级页表起始地址,其定义在vmlinux.lds.S链接文件中
idmap_pg_dir = .;
. += IDMAP_DIR_SIZE;
idmap_pg_end = .;
这里分配给idmap_pg_dir的页面大小为IDMAP_DIR_SIZE,而IDMAP_DIR_SIZE实现在arch/arm64/include/asm/kernel-pgtable.h头文件中,通常是3个连续的大小为4K页面。计算参考如下:
<1>#arch/arm64/include/asm/pgtable-hwdef.h
/*
* Number of page-table levels required to address 'va_bits' wide
* address, without section mapping. We resolve the top (va_bits - PAGE_SHIFT)
* bits with (PAGE_SHIFT - 3) bits at each page table level. Hence:
*
* levels = DIV_ROUND_UP((va_bits - PAGE_SHIFT), (PAGE_SHIFT - 3))
*
* where DIV_ROUND_UP(n, d) => (((n) + (d) - 1) / (d))
*
* We cannot include linux/kernel.h which defines DIV_ROUND_UP here
* due to build issues. So we open code DIV_ROUND_UP here:
*
* ((((va_bits) - PAGE_SHIFT) + (PAGE_SHIFT - 3) - 1) / (PAGE_SHIFT - 3))
*
* which gets simplified as :
*/
#define ARM64_HW_PGTABLE_LEVELS(va_bits) (((va_bits) - 4) / (PAGE_SHIFT - 3))
...
/*
* Highest possible physical address supported.
*/
#define PHYS_MASK_SHIFT (CONFIG_ARM64_PA_BITS) //48
<2>#arch/arm64/include/asm/kernel-pgtable.h
#if ARM64_KERNEL_USES_PMD_MAPS //段映射一般走这个
#define SWAPPER_PGTABLE_LEVELS (CONFIG_PGTABLE_LEVELS - 1)
#define IDMAP_PGTABLE_LEVELS (ARM64_HW_PGTABLE_LEVELS(PHYS_MASK_SHIFT) - 1) // {((48-12)+(12-3)-1) / (12-3) = (36+9-1)/9 = 44/9 = 4}-1 =3
#else
#define SWAPPER_PGTABLE_LEVELS (CONFIG_PGTABLE_LEVELS)
#define IDMAP_PGTABLE_LEVELS (ARM64_HW_PGTABLE_LEVELS(PHYS_MASK_SHIFT)) //3
#endif
...
#define IDMAP_DIR_SIZE (IDMAP_PGTABLE_LEVELS * PAGE_SIZE)
这里的CONFIG_ARM64_PA_BITS配置的是48. 这里的含义是,计算采用section mapping的话,需要几个页来存放table。ARM64_HW_PGTABLE_LEVELS,很关键,根据配置的物理地址线的宽度计算需要的页面数,注意注释处的计算方法:
((((va_bits) - PAGE_SHIFT) + (PAGE_SHIFT - 3) - 1) / (PAGE_SHIFT - 3))
结合vmlinux.lds,上面的公式就是: ((48-12)+(12-3)-1) / (12-3) = (36+9-1)/9 = 44/9 = 4,最终IDMAP_DIR_SIZE为3个页面,即一次性在连续的地址上分配三个页表---PGD/PUD/PMD页表,每一级页表占据一个页面。
这里需要注意一下我们在这里只建立了一个2MB大小的段映射,也就是说对于恒等映射,2M的段映射已经够用。
- 将__idmap_text_start的物理地址放到x3寄存器中, __idmap_text_start标号定义在arch/arm64/kernel/vmlinux.lds.S中,是我们要进行恒等映射的起始地址(物理 == 虚拟地址):
#define IDMAP_TEXT \
. = ALIGN(SZ_4K); \
__idmap_text_start = .; \
*(.idmap.text) \
__idmap_text_end = .;
.text : ALIGN(SEGMENT_ALIGN) { /* Real text segment */
_stext = .; /* Text and read-only data */
IRQENTRY_TEXT
SOFTIRQENTRY_TEXT
ENTRY_TEXT
TEXT_TEXT
SCHED_TEXT
CPUIDLE_TEXT
LOCK_TEXT
KPROBES_TEXT
HYPERVISOR_TEXT
IDMAP_TEXT '.idmap.text段'
*(.gnu.warning)
. = ALIGN(16);
*(.got) /* Global offset table */
}
除了在开机启动时打开MMU外,内核里还有很对场景需要恒等映射,我们通过.section把这些函数都放在.idmap.text段中:
# arch/arm64/kernel/head.S
/*
* end early head section, begin head code that is also used for
* hotplug and needs to have the same protections as the text region
*/
.section ".idmap.text","awx"
这些处于.idmap.text段中函数也可以通过System.map看到:
ffffffc00952f000 T __idmap_text_start //
ffffffc00952f000 T init_kernel_el
ffffffc00952f010 t init_el1
ffffffc00952f038 t init_el2
ffffffc00952f270 t __cpu_stick_to_vhe
ffffffc00952f280 t set_cpu_boot_mode_flag
ffffffc00952f2a8 T secondary_holding_pen
ffffffc00952f2d0 t pen
ffffffc00952f2e4 T secondary_entry
ffffffc00952f2f4 t secondary_startup
ffffffc00952f314 t __secondary_switched
ffffffc00952f3b8 t __secondary_too_slow
ffffffc00952f3c8 T __enable_mmu //重点关注
ffffffc00952f42c T __cpu_secondary_check52bitva
ffffffc00952f434 t __no_granule_support
ffffffc00952f45c t __relocate_kernel
ffffffc00952f4a8 t __primary_switch //重点关注
ffffffc00952f530 t enter_vhe
ffffffc00952f568 T cpu_resume
ffffffc00952f590 T cpu_soft_restart
ffffffc00952f5c4 T cpu_do_resume
ffffffc00952f66c T idmap_cpu_replace_ttbr1
ffffffc00952f6a4 t __idmap_kpti_flag
ffffffc00952f6a8 T idmap_kpti_install_ng_mappings
ffffffc00952f6e8 t do_pgd
ffffffc00952f700 t next_pgd
ffffffc00952f710 t skip_pgd
ffffffc00952f750 t walk_puds
ffffffc00952f758 t next_pud
ffffffc00952f75c t walk_pmds
ffffffc00952f764 t do_pmd
ffffffc00952f77c t next_pmd
ffffffc00952f78c t skip_pmd
ffffffc00952f79c t walk_ptes
ffffffc00952f7a4 t do_pte
ffffffc00952f7c8 t skip_pte
ffffffc00952f7d8 t __idmap_kpti_secondary
ffffffc00952f820 T __cpu_setup
ffffffc00952f974 T __idmap_text_end //
- 假设虚拟地址位宽为48(我们定义的是CONFIG_ARM64_VA_BITS_48);
- 虚拟地址位宽(48)保存到X5寄存器;
- 把立即数VA_BITS_MIN(48)保存到全局变量vabits_actual中;
- 将__idmap_text_end的物理地址放到x5寄存器中,计算__idmap_text_end地址第一个1前0的个数。并判断__idmap_text_end地址是否超过VA_BITS_MIN所能表达的地址范围。其中TCR_T0SZ(VA_BITS_MIN) 表示TTBR0页表所能映射的大小,因为稍后我们创建的页表会填充到TTBR0寄存器里面;
- 把PGD页表包含的页表项保存到x4寄存器中(2^9);
- 把__idmap_text_end的物理地址放到x6寄存器中;
- 调用map_memory宏来创建这段恒等映射的页表;
map_memory x0, x1, x3, x6, x7, x3, x4, x10, x11, x12, x13, x14
(1) x0 --- idmap_pg_dir
(2) x1 --- 无效值,会在map_memory中根据tbl的值重新计算
(3) x3 --- __idmap_text_start
(4) x6 --- __idmap_text_end
(5) x7 --- SWAPPER_MM_MMUFLAGS
(6) x3 --- __idmap_text_start
(7) x4 --- idmap_ptrs_per_pgd
2.4.5 map_memory宏的解析
map_memory宏一共12个参数,参数的解释在下面代码的批注中解释的非常清楚。重要参数说明如下:
- tbl : 页表起始地址(pgd)
- rtbl : 下级页表起始地址(typically tbl + PAGE_SIZE)
- vstart: 要映射虚拟地址的起始地址
- vend : 要映射虚拟地址的结束地址
- flags : 最后一级页表的属性
- phys : 要映射物理地址的起始地址
- flags : pgd entries个数
/*
* Map memory for specified virtual address range. Each level of page table needed supports
* multiple entries. If a level requires n entries the next page table level is assumed to be
* formed from n pages.
*
* tbl: location of page table 页表起始地址(pgd)
* rtbl: address to be used for first level page table entry (typically tbl + PAGE_SIZE)下级页表起始地址
* vstart: virtual address of start of range 要映射虚拟地址的起始地址
* vend: virtual address of end of range - we map [vstart, vend - 1]要映射虚拟地址的结束地址
* flags: flags to use to map last level entries 最后一级页表的属性
* phys: physical address corresponding to vstart - physical memory is contiguous 要映射物理地址的起始地址
* pgds: the number of pgd entries :pgd entries个数
*
* Temporaries: istart, iend, tmp, count, sv - these need to be different registers
* Preserves: vstart, flags
* Corrupts: tbl, rtbl, vend, istart, iend, tmp, count, sv
*/
.macro map_memory, tbl, rtbl, vstart, vend, flags, phys, pgds, istart, iend, tmp, count, sv
sub \vend, \vend, #1
add \rtbl, \tbl, #PAGE_SIZE ---(1)
mov \sv, \rtbl
mov \count, #0
compute_indices \vstart, \vend, #PGDIR_SHIFT, \pgds, \istart, \iend, \count ---(2)
populate_entries \tbl, \rtbl, \istart, \iend, #PMD_TYPE_TABLE, #PAGE_SIZE, \tmp ---(3)
mov \tbl, \sv
mov \sv, \rtbl
#if SWAPPER_PGTABLE_LEVELS > 3 //我们这里不成立
compute_indices \vstart, \vend, #PUD_SHIFT, #PTRS_PER_PUD, \istart, \iend, \count
populate_entries \tbl, \rtbl, \istart, \iend, #PMD_TYPE_TABLE, #PAGE_SIZE, \tmp
mov \tbl, \sv
mov \sv, \rtbl
#endif
#if SWAPPER_PGTABLE_LEVELS > 2 ---(4)
compute_indices \vstart, \vend, #SWAPPER_TABLE_SHIFT, #PTRS_PER_PMD, \istart, \iend, \count
populate_entries \tbl, \rtbl, \istart, \iend, #PMD_TYPE_TABLE, #PAGE_SIZE, \tmp
mov \tbl, \sv
#endif
compute_indices \vstart, \vend, #SWAPPER_BLOCK_SHIFT, #PTRS_PER_PTE, \istart, \iend, \count ---(5)
bic \count, \phys, #SWAPPER_BLOCK_SIZE - 1
populate_entries \tbl, \count, \istart, \iend, \flags, #SWAPPER_BLOCK_SIZE, \tmp
.endm
-
计算PUD基地址,rtbl是下级页表地址:PUD = PGD+PAGE_SIZE
-
compute_indices宏的功能:根据虚拟地址计算各级页表的索引值index
/*
* Compute indices of table entries from virtual address range. If multiple entries
* were needed in the previous page table level then the next page table level is assumed
* to be composed of multiple pages. (This effectively scales the end index).
*
* vstart: virtual address of start of range
* vend: virtual address of end of range - we map [vstart, vend]
* shift: shift used to transform virtual address into index
* ptrs: number of entries in page table
* istart: index in table corresponding to vstart
* iend: index in table corresponding to vend
* count: On entry: how many extra entries were required in previous level, scales
* our end index.
* On exit: returns how many extra entries required for next page table level
*
* Preserves: vstart, vend, shift, ptrs
* Returns: istart, iend, count
*/
.macro compute_indices, vstart, vend, shift, ptrs, istart, iend, count
lsr \iend, \vend, \shift
mov \istart, \ptrs
sub \istart, \istart, #1
and \iend, \iend, \istart // iend = (vend >> shift) & (ptrs - 1)
mov \istart, \ptrs
mul \istart, \istart, \count
add \iend, \iend, \istart // iend += count * ptrs
// our entries span multiple tables
lsr \istart, \vstart, \shift
mov \count, \ptrs
sub \count, \count, #1
and \istart, \istart, \count
sub \count, \iend, \istart
.endm
- populate_entries宏的功能:填充索引值index对应的页表项
/*
* Macro to populate page table entries, these entries can be pointers to the next level
* or last level entries pointing to physical memory.
*
* tbl: page table address
* rtbl: pointer to page table or physical memory
* index: start index to write
* eindex: end index to write - [index, eindex] written to
* flags: flags for pagetable entry to or in
* inc: increment to rtbl between each entry
* tmp1: temporary variable
*
* Preserves: tbl, eindex, flags, inc
* Corrupts: index, tmp1
* Returns: rtbl
*/
.macro populate_entries, tbl, rtbl, index, eindex, flags, inc, tmp1
.Lpe\@: phys_to_pte \tmp1, \rtbl
orr \tmp1, \tmp1, \flags // tmp1 = table entry
str \tmp1, [\tbl, \index, lsl #3]
add \rtbl, \rtbl, \inc // rtbl = pa next level
add \index, \index, #1
cmp \index, \eindex
b.ls .Lpe\@
.endm
- 设置PUD页表项 ;
- 设置PMD页表项(因为我们用的是段映射,因此这里是最后一级,没有PTE);
2.4.6 创建内核image的映射
/*
* Map the kernel image (starting with PHYS_OFFSET).
*/
adrp x0, init_pg_dir ---(1)
mov_q x5, KIMAGE_VADDR // compile time __va(_text) ---(2)
add x5, x5, x23 // add KASLR displacement //x23 = __PHYS_OFFSET
mov x4, PTRS_PER_PGD
adrp x6, _end // runtime __pa(_end) ---(3)内核映像结束物理地址
adrp x3, _text // runtime __pa(_text) ---(4)内核映像起始物理地址
sub x6, x6, x3 // _end - _text //内核映像的大小
add x6, x6, x5 // runtime __va(_end) ---(5)内核映像结束地址
map_memory x0, x1, x5, x6, x7, x3, x4, x10, x11, x12, x13, x14 ---(6)
- 这里是加载init_pg_dir的物理地址到x0寄存器,init_pg_dir是kernel image的映射使用的页表起始地址(与恒等映射不同),其定义在vmlinux.lds.S链接文件中。
BSS_SECTION(SBSS_ALIGN, 0, 0)
. = ALIGN(PAGE_SIZE);
init_pg_dir = .;
. += INIT_DIR_SIZE;
init_pg_end = .;
- 加载内核映像虚拟的起始地址KIMAGE_VADDR到x5寄存器,注意这里使用的是mov_q指令。KIMAGE_VADDR定义在vmlinux.lds.S链接文件中。
SECTIONS
{
......
'在5.8内核版本发现TEXT_OFFSET没有任何作用,因此,被重新定义为0x0'
. = KIMAGE_VADDR; '内核映像虚拟的起始地址(在5.8内核之前这里为KIMAGE_VADDR + TEXT_OFFSET)'
.head.text : { '早期汇编代码的text段'
_text = .; '入口地址'
HEAD_TEXT 定义在include/asm-generic/vmlinux.lds.h'#define HEAD_TEXT KEEP(*(.head.text))'
}
- 这里是加载内核映像结束物理地址到x3寄存器;
- 这里是加载内核映像起始物理地址到x6寄存器;
- 换算得到内核映像起始虚拟地址,并加载到x6寄存器;
- 调用map_memory宏来创建这段内核映像映射的页表;