linux内核初始化卡死,armlinux内核启动－－内存初始化管理

linux版本:2.6.36

相关数据结构

arch/arm/include/asm/setup.h

#ifdef CONFIG_ARCH_LH7A40X

# define NR_BANKS 16

#else

# define NR_BANKS 8

#endif

struct membank {

unsigned long start;

unsigned long size;

unsigned int highmem;

};

struct meminfo {

int nr_banks;

struct membank bank[NR_BANKS];

};

linux内核的内存管理分三个阶段。

1. 启动---->bootmem初始化完成为第一阶段。此阶段只能使用memblock_reserve函数分配内存。

此阶段结束标志为：init_bootmem_done = 1.

2. bootmem初始化完--->buddy完成前。结束标志为mem_init_done = 1.

3. 全部内存初始化完毕，可以用cache和buddy分配内存。

内存初始化步骤：

1. start_kernel---->setup_arch->pageing_init

|-->setup_per_cpu_areas

|-->build_all_zonelists

|-->mem_init

-->setup_per_cpu_pageset

setup_arch-->arm_memblock_init-->memblock_init

内存原始数据由u-boot传入,对照本开发板uboot部分内存初始化函数，我们知道uboot传递过来的tag->u.mem.start, tag->u.mem.size分别为0x30000000,0x4000000，即内存起始地址是0x30000000，大小为64M，start_arch获取u-boot传递的参数地址后，调用了parse_tag_mem32函数对传递的内存参数处理：

556 static int __init parse_tag_mem32(const struct tag *tag)

557 {

558         return arm_add_memory(tag->u.mem.start, tag->u.mem.size);

559 }

560

561 __tagtable(ATAG_MEM, parse_tag_mem32);

如上可见，parse_tag_mem32函数调用arm_add_memory函数把RAM的start和size等参数保存到了meminfo结构的 meminfo结构体中。现在再来分析arm_add_memory

arm_add_memory定义如下(arch/arm/kernel/setup.c)

static int __init arm_add_memory(unsigned long start, unsigned long size)

{

struct membank *bank = &meminfo.bank[meminfo.nr_banks];

if (meminfo.nr_banks >= NR_BANKS) {

printk(KERN_CRIT "NR_BANKS too low, "

"ignoring memory at %#lx\n", start);

return -EINVAL;

}

/*

* Ensure that start/size are aligned to a page boundary.

* Size is appropriately rounded down, start is rounded up.

*/

size -= start & ~PAGE_MASK;

bank->start = PAGE_ALIGN(start);

bank->size  = size & PAGE_MASK;

/*

* Check whether this memory region has non-zero size or

* invalid node number.

*/

if (bank->size == 0)

return -EINVAL;

meminfo.nr_banks++;

return 0;

}

经过这样的处理，setup.c文件中的meminfo可就不再是

struct meminfo meminfo  = { 0, };

而是

struct meminfo meminfo  = { 1,{0x30000000,0x4000000,0},{}, };

表示当前有一个内存区域，物理地址是从0x30000000开始，大小是64M,节点是0

处理完这些数据后，start_arch会调用第822行调用arm_memblock_init(&meminfo, mdesc);

先分析第1阶段。

1. 初始化：setup_arch-->arm_memblock_init-->memblock_init

arch/arm/mm/init.c

270 void __init arm_memblock_init(struct meminfo *mi, struct machine_desc *mdesc)

271 {

272         int i;

273

274         memblock_init();

275         for (i = 0; i < mi->nr_banks; i++)

276                 memblock_add(mi->bank[i].start, mi->bank[i].size);

277

278         /* Register the kernel text, kernel data and initrd with memblock. */

279 #ifdef CONFIG_XIP_KERNEL //配置内核XIP方式运行，避免把内核从Nor 拷贝到主存 空间，让内核运行在低功耗的NOR 上，节省系统启动时间，降低系统对SDRAM 的需要，减少电能消耗，使产品能够持续使用更长时间

280         memblock_reserve(__pa(_data), _end - _data);

281 #else

282         memblock_reserve(__pa(_stext), _end - _stext);

/*第282行，在０节点中保留内核镜像部分的内存，大概是0x30108000到以后的5M左右(解压后的内核)。

参看arch/arm/kernel/vmlinux.lds文件，

375 SECTIONS

376 {

377  . = 0xC0000000 + 0x00108000;

378  .init : { /* Init code and data                */

379   _stext = .;

所以stext等于c0108000,对应物理地址30108000,

对_end再参看arch/arm/kernel/vmlinux.lds.S文件,它也在SECTIONS区，在233行定义

232         BSS_SECTION(0, 0, 0)

233         _end = .;

对于我移植的2.3.36内核，_end等于c0555b60,对应物理地址30555b60，在此语句后添加打印信息得到

_end - _stext=44db60，4.512608M大小,所以这一句功能就是把0x30108000　～　0x30555b60这段(4.512608M)空间保留下来。

*/

283 #endif

284 #ifdef CONFIG_BLK_DEV_INITRD //本开发板没配置

285         if (phys_initrd_size) {

286                 memblock_reserve(phys_initrd_start, phys_initrd_size);

287

288                 /* Now convert initrd to virtual addresses */

289                 initrd_start = __phys_to_virt(phys_initrd_start);

290                 initrd_end = initrd_start + phys_initrd_size;

291         }

292 #endif

293

294         arm_mm_memblock_reserve();//就是把0x30104000　～　0x30108000这段(16K)用于页目录的内容保留下来。

295

296         /* reserve any platform specific memblock areas */

297         if (mdesc->reserve)//对st2410中没有定义

298                 mdesc->reserve();

299

300         memblock_analyze();//计算内存块大小

301         memblock_dump_all();//显示

302 }

内存原始数据由u-boot传入，在初始化完memblock_init后，arm_memblock_init函数中memblock_add调用 memblock_add_region加入原始内存数据，我的板子上配了64M内存，即：0x0000 0000->0x40000000,加完后的配置如下：

MEMBLOCK configuration:

rmo_size    = 0x0

memory.size = 0x4000000

memory.cnt  = 0x1

memory[0x0]    0x0000000030000000 - 0x0000000033ffffff, 0x4000000 bytes

reserved.cnt  = 0x1

reserved[0x0]  0x0000000030104000 - 0x0000000030555b5f, 0x451b60 bytes

memblock_init()在mm/memblock.c里面被定义。

void __init memblock_init(void)

{

/* Create a dummy zero size MEMBLOCK which will get coalesced away later.

* This simplifies the memblock_add() code below...

*/

memblock.memory.region[0].base = 0;

memblock.memory.region[0].size = 0;

memblock.memory.cnt = 1;

/* Ditto. */

memblock.reserved.region[0].base = 0;

memblock.reserved.region[0].size = 0;

memblock.reserved.cnt = 1;

}

其作用就是初始化memblock这个结构。将他们清空。memblock包含两个重要的成员，分别是memblock.memory和memblock.reserved.其分别代表系统中可用的内存和已经被保留的内存。

memblock.memory和memblock.reserved被定义为以下结构：include/linux/memblock.h

#define MAX_MEMBLOCK_REGIONS 128

struct memblock_property {

u64 base;

u64 size;

};

struct memblock_region {

unsigned long cnt;

u64 size;

struct memblock_property region[MAX_MEMBLOCK_REGIONS+1];

};

struct memblock {

unsigned long debug;

u64 rmo_size;

struct memblock_region memory;

struct memblock_region reserved;

};

mm/memblock.c

long memblock_add(u64 base, u64 size)

{

struct memblock_region *_rgn = &memblock.memory;

/* On pSeries LPAR systems, the first MEMBLOCK is our RMO region. */

if (base == 0)

memblock.rmo_size = size;

return memblock_add_region(_rgn, base, size);

}

static long memblock_add_region(struct memblock_region *rgn, u64 base, u64 size)

{

unsigned long coalesced = 0;

long adjacent, i;

if ((rgn->cnt == 1) && (rgn->region[0].size == 0)) {

rgn->region[0].base = base;

rgn->region[0].size = size;

return 0;

}

/* First try and coalesce this MEMBLOCK with another. */

for (i = 0; i < rgn->cnt; i++) {

u64 rgnbase = rgn->region[i].base;

u64 rgnsize = rgn->region[i].size;

if ((rgnbase == base) && (rgnsize == size))

/* Already have this region, so we're done */

return 0;

adjacent = memblock_addrs_adjacent(base, size, rgnbase, rgnsize);

if (adjacent > 0) {

rgn->region[i].base -= size;

rgn->region[i].size += size;

coalesced++;

break;

} else if (adjacent < 0) {

rgn->region[i].size += size;

coalesced++;

break;

}

}

if ((i < rgn->cnt - 1) && memblock_regions_adjacent(rgn, i, i+1)) {

memblock_coalesce_regions(rgn, i, i+1);

coalesced++;

}

if (coalesced)

return coalesced;

if (rgn->cnt >= MAX_MEMBLOCK_REGIONS)

return -1;

/* Couldn't coalesce the MEMBLOCK, so add it to the sorted table. */

for (i = rgn->cnt - 1; i >= 0; i--) {

if (base < rgn->region[i].base) {

rgn->region[i+1].base = rgn->region[i].base;

rgn->region[i+1].size = rgn->region[i].size;

} else {

rgn->region[i+1].base = base;

rgn->region[i+1].size = size;

break;

}

}

if (base < rgn->region[0].base) {

rgn->region[0].base = base;

rgn->region[0].size = size;

}

rgn->cnt++;

return 0;

}

memblock_add_region函数作用是将给定的物理地址所指定的memory region加入到指定的memblock(memblock.reserved或者是memblock.memory)中。新加入的memory region需要经过检查，如果与原先的memory region有重叠，则需要合并在原先的memory region中，否则的话就新建一个memory region.

3. memblock_reserve用来分配内存页。

以分配内核本身占用的内存为例：

/* Register the kernel text, kernel data and initrd with memblock. */

memblock_reserve(__pa(_stext), _end - _stext);

_stext,_end参见arch/arm/kernel/vmlinux.lds.S 链接脚本

long __init memblock_reserve(u64 base, u64 size)

{

struct memblock_region *_rgn = &memblock.reserved;

BUG_ON(0 == size);

return memblock_add_region(_rgn, base, size);

}

分配完之后的内存配置为：

MEMBLOCK configuration:

rmo_size    = 0x40000000

memory.size = 0x0

memory.cnt  = 0x1

memory[0x0]    0x0000000000000000 - 0x000000003fffffff, 0x40000000 bytes

reserved.cnt  = 0x1

reserved[0x0]  0x0000000000000000 - 0x0000000000xxxxx,  _end - _stext bytes

如果分配有连续则进行合并。

几次分配后的配置如下：

MEMBLOCK configuration:

rmo_size    = 0x80000000

memory.size = 0x80000000

memory.cnt  = 0x1

memory[0x0]    0x0000000000000000 - 0x000000007fffffff, 0x80000000 bytes

reserved.cnt  = 0x6

reserved[0x0]  0x0000000000000000 - 0x00000000006b0fff, 0x6b1000 bytes

reserved[0x1]  0x0000000000ffa000 - 0x0000000000ffcfff, 0x3000 bytes

reserved[0x2]  0x000000002fbc4000 - 0x000000002fbdefff, 0x1b000 bytes

reserved[0x3]  0x000000002fbdfa88 - 0x000000002ffff4cc, 0x41fa45 bytes

reserved[0x4]  0x000000002fbe4000 - 0x000000002ffff4cd, 0x41b4ce bytes

reserved[0x5]  0x000000007ffff000 - 0x000000007fffffff, 0x1000 bytes

init

arch/arm/mm/mmu.c

/*

* Reserve the special regions of memory

*/

void __init arm_mm_memblock_reserve(void)

{

/*

* Reserve the page tables.  These are already in use,

* and can only be in node 0.

*/

memblock_reserve(__pa(swapper_pg_dir), PTRS_PER_PGD * sizeof(pgd_t));

/*

相关的信息

swapper_pg_dir是初始化页表虚拟地址，它在

./arch/arm/kernel/head.S定义

44:    .globl  swapper_pg_dir

45:    .equ    swapper_pg_dir, KERNEL_RAM_VADDR - 0x4000

./arch/arm/kernel/head.S:55:#define KERNEL_START        KERNEL_RAM_VADDR

所以swapper_pg_dir等于0xc0104000，对应物理地址0x30104000

PTRS_PER_PGD在arch/arm/include/asm/pgtable.h第103行定义

#define PTRS_PER_PGD        2048

而pgd_t定义为

typedef unsigned long pgd_t[2];

PTRS_PER_PGD * sizeof(pgd_t)＝2048*8=16384的大小为0x00004000 (16K)

就是把0x30104000　～　0x30108000这段(16K)用于页目录的内容保留下来。

*/

#ifdef CONFIG_SA1111 //2410没定义

/*

* Because of the SA1111 DMA bug, we want to preserve our

* precious DMA-able memory...

*/

memblock_reserve(PHYS_OFFSET, __pa(swapper_pg_dir) - PHYS_OFFSET);

#endif

}

arch/arm/mm/init.c

#define MLK(b, t) b, t, ((t) - (b)) >> 10 //右移10位为K

#define MLM(b, t) b, t, ((t) - (b)) >> 20  //右移20位为M

#define MLK_ROUNDUP(b, t) b, t, DIV_ROUND_UP(((t) - (b)), SZ_1K)

printk(KERN_NOTICE "Virtual kernel memory layout:\n"

"    vector  : 0x%08lx - 0x%08lx   (%4ld kB)\n"

"    fixmap  : 0x%08lx - 0x%08lx   (%4ld kB)\n"

#ifdef CONFIG_MMU

"    DMA     : 0x%08lx - 0x%08lx   (%4ld MB)\n"

#endif

"    vmalloc : 0x%08lx - 0x%08lx   (%4ld MB)\n"

"    lowmem  : 0x%08lx - 0x%08lx   (%4ld MB)\n"

#ifdef CONFIG_HIGHMEM

"    pkmap   : 0x%08lx - 0x%08lx   (%4ld MB)\n"

#endif

"    modules : 0x%08lx - 0x%08lx   (%4ld MB)\n"

"      .init : 0x%p" " - 0x%p" "   (%4d kB)\n"

"      .text : 0x%p" " - 0x%p" "   (%4d kB)\n"

"      .data : 0x%p" " - 0x%p" "   (%4d kB)\n",

MLK(UL(CONFIG_VECTORS_BASE), UL(CONFIG_VECTORS_BASE) +

(PAGE_SIZE)),

MLK(FIXADDR_START, FIXADDR_TOP),

#ifdef CONFIG_MMU

MLM(CONSISTENT_BASE, CONSISTENT_END),

#endif

MLM(VMALLOC_START, VMALLOC_END),

MLM(PAGE_OFFSET, (unsigned long)high_memory),

#ifdef CONFIG_HIGHMEM

MLM(PKMAP_BASE, (PKMAP_BASE) + (LAST_PKMAP) *

(PAGE_SIZE)),

#endif

MLM(MODULES_VADDR, MODULES_END),

MLK_ROUNDUP(__init_begin, __init_end),

MLK_ROUNDUP(_text, _etext),

MLK_ROUNDUP(_data, _edata));

系统启动的时候打印出：

stext:c0108000,__pa_stext:30108000, _end - _stext=44db60

_end:c0555b60,__pa(_end)=30555b60

Memory: 64MB = 64MB total

Memory: 60536k/60536k available, 5000k reserved, 0K highmem

Virtual kernel memory layout:

vector  : 0xffff0000 - 0xffff1000   (   4 kB)

fixmap  : 0xfff00000 - 0xfffe0000   ( 896 kB)

DMA     : 0xffc00000 - 0xffe00000   (   2 MB)

vmalloc : 0xc4800000 - 0xe0000000   ( 440 MB)

lowmem  : 0xc0000000 - 0xc4000000   (  64 MB)

modules : 0xbf000000 - 0xc0000000   (  16 MB)

.init : 0xc0108000 - 0xc012e000   ( 152 kB)---->(系统启动过程中 以__init宏标识的函数占用的空间 被vmlinux.lds标识为 __init_begin __init_end 启动init进程前被释放掉 :Freeing init memory: 152K)

.text : 0xc012e000 - 0xc04dd000   (3772 kB)

.data : 0xc04fe000 - 0xc0527500   ( 166 kB)

Hierarchical RCU implementation.

启动后查看内存命令显示如下：

[root@localhost /]# free

total         used         free       shared      buffers

Mem:        60688         6644        54044            0            0

Swap:            0            0            0

Total:        60688         6644        54044

[root@localhost /]#

保留内存5000K中 linux内核占用了152+3772+166=4090K

加上boot的时候参数等平台空间占用32K = 3149K

其它还差3464-3149=315K 没有着落

其它还包括影射的向量表4K

还有可能是系统管理内存的页表占用空间等其它的占用

系统启动成功后 Freeing init memory: 100K

所以用free命令看会多了100K

而free命令中看到的used基本是缓存 buffer占用的,为了提高i/o速度的缓存,很多都并非真正应用

在我应用中去试图malloc更多的内存的时候 used中很多都能被malloc出来的.

mm/memblock.c

void __init memblock_analyze(void)

{

int i;

memblock.memory.size = 0;

for (i = 0; i < memblock.memory.cnt; i++)

memblock.memory.size += memblock.memory.region[i].size;

}

mm/memblock.c

static void memblock_dump(struct memblock_region *region, char *name)

{

unsigned long long base, size;

int i;

pr_info(" %s.cnt  = 0x%lx\n", name, region->cnt);

for (i = 0; i < region->cnt; i++) {

base = region->region[i].base;

size = region->region[i].size;

pr_info(" %s[0x%x]\t0x%016llx - 0x%016llx, 0x%llx bytes\n",

name, i, base, base + size - 1, size);

}

}

mm/memblock.c

void memblock_dump_all(void)

{

if (!memblock_debug)

return;

pr_info("MEMBLOCK configuration:\n");

pr_info(" rmo_size    = 0x%llx\n", (unsigned long long)memblock.rmo_size);

pr_info(" memory.size = 0x%llx\n", (unsigned long long)memblock.memory.size);

memblock_dump(&memblock.memory, "memory");

memblock_dump(&memblock.reserved, "reserved");

}

22 static int memblock_debug;

23

24 static int __init early_memblock(char *p)

25 {

26         if (p && strstr(p, "debug"))

27                 memblock_debug = 1;

28         return 0;

29 }

30 early_param("memblock", early_memblock);

阅读(7987) | 评论(0) | 转发(7) |