高端内存映射之kmap_atomic固定映射--Linux内存管理(二十一)

1 固定映射

1.1 数据结构

linux高端内存中的临时内存区为固定内存区的一部分, 对于固定内存在linux内核中有下面描述

x86	arm	arm64
arch/x86/include/asm/fixmap.h?v=4.7, line 67	arch/arm/include/asm/fixmap.h?v=4.7, line 11	arch/arm64/include/asm/fixmap.h?v=4.7, line 36

/*
 * Here we define all the compile-time 'special' virtual
 * addresses. The point is to have a constant address at
 * compile time, but to set the physical address only
 * in the boot process.
 *
 * These 'compile-time allocated' memory buffers are
 * page-sized. Use set_fixmap(idx,phys) to associate
 * physical memory with fixmap indices.
 *
 */
enum fixed_addresses {
    FIX_HOLE,

    /*
     * Reserve a virtual window for the FDT that is 2 MB larger than the
     * maximum supported size, and put it at the top of the fixmap region.
     * The additional space ensures that any FDT that does not exceed
     * MAX_FDT_SIZE can be mapped regardless of whether it crosses any
     * 2 MB alignment boundaries.
     *
     * Keep this at the top so it remains 2 MB aligned.
     */
#define FIX_FDT_SIZE        (MAX_FDT_SIZE + SZ_2M)
    FIX_FDT_END,
    FIX_FDT = FIX_FDT_END + FIX_FDT_SIZE / PAGE_SIZE - 1,

    FIX_EARLYCON_MEM_BASE,
    FIX_TEXT_POKE0,
    __end_of_permanent_fixed_addresses,

    /*
     * Temporary boot-time mappings, used by early_ioremap(),
     * before ioremap() is functional.
     */
#define NR_FIX_BTMAPS       (SZ_256K / PAGE_SIZE)
#define FIX_BTMAPS_SLOTS    7
#define TOTAL_FIX_BTMAPS    (NR_FIX_BTMAPS * FIX_BTMAPS_SLOTS)

    FIX_BTMAP_END = __end_of_permanent_fixed_addresses,
    FIX_BTMAP_BEGIN = FIX_BTMAP_END + TOTAL_FIX_BTMAPS - 1,

    /*
     * Used for kernel page table creation, so unmapped memory may be used
     * for tables.
     */
    FIX_PTE,
    FIX_PMD,
    FIX_PUD,
    FIX_PGD,

    __end_of_fixed_addresses
};

1.2 固定映射

ioremap的作用是将IO和BIOS以及物理地址空间映射到在896M至1G的128M的地址空间内, 使得kernel能够访问该空间并进行相应的读写操作。

start_kernel()->setup_arch()->early_ioremap_init()

然后arm和arm64上early_ioremap_init又是early_ioremap_setup的前端

函数	x86	arm	arm64
early_ioremap_init	arch/x86/mm/ioremap.c?v=4.7, line 445	arch/arm/mm/ioremap.c?v=4.7, line 489	arch/arm64/mm/ioremap.c?v=4.7, line 110
early_ioremap_setup	mm/early_ioremap.c?v=4.7, line 67	体系结构无关	体系结构无关

/*
 * Must be called after early_fixmap_init
 */
void __init early_ioremap_init(void)
{
    early_ioremap_setup();
}

但是arm和arm64下的setup_arch函数则会先调用early_fixmap_init函数来填充fixmap. 参见arch/arm/kernel/setup.c?v=4.7, line 1058和arch/arm64/kernel/setup.c?v=4.7, line 229.

void __init setup_arch(char **cmdline_p)
{
    early_fixmap_init();
    early_ioremap_init();
}

early_fixmap_init函数的定义在

arm	arm64
arch/arm/mm/mmu.c?v=4.7, line 385	arch/arm64/mm/mmu.c?v=4.7, line 676

其中arm架构的定义如下所示, 在arch/arm/mm/mmu.c?v=4.7, line 385

void __init early_fixmap_init(void)
{
    pmd_t *pmd;

    /*
     * The early fixmap range spans multiple pmds, for which
     * we are not prepared:
     */
    BUILD_BUG_ON((__fix_to_virt(__end_of_early_ioremap_region) >> PMD_SHIFT)
             != FIXADDR_TOP >> PMD_SHIFT);

    /*得到固定映射区的pmd
    ，此pmd为虚拟地址转换为物理地址的pmd*/
    pmd = fixmap_pmd(FIXADDR_TOP);
     /*将bm_pte页表设置为固定映射区开始地址的pmd的第一个页表；*/
    pmd_populate_kernel(&init_mm, pmd, bm_pte);

    pte_offset_fixmap = pte_offset_early_fixmap;
}

1.3 ioremap函数

对于ioremap的使用需要通过early_memremap和early_iounmap进行.

由于对应于ioremap的内存空间是有限的, 所以对于ioremap空间的使用遵照使用结束马上释放的原则. 这就是说early_memremap和early_iounmap必须配对使用并且访问结束必须马上执行unmap

2 临时内核映射

刚才描述的kmap函数不能用于中断处理程序, 因为它可能进入睡眠状态. 如果pkmap数组中没有空闲位置, 该函数会进入睡眠状态, 直至情形有所改善.

因此内核提供了一个备选的映射函数, 其执行是原子的, 逻辑上称为kmap_atomic. 该函数的一个主要优点是它比普通的kmap快速. 但它不能用于可能进入睡眠的代码. 因此, 它对于很快就需要一个临时页的简短代码，是非常理想的.

kmap_atomic的定义在IA-32, PPC, Sparc32上是特定于体系结构的, 但这3种实现只有非常细微的差别. 其原型是相同的.

2.1 kmap_atomic函数

//  http://lxr.free-electrons.com/source/arch/arm/mm/highmem.c?v=4.7#L55
void *kmap_atomic(struct page *page)

page是一个指向高端内存页的管理结构的指针, 而早期的内核中, 增加了一个类型为enum km_type的type参数, 用于指定所需的映射类型

//  http://lxr.free-electrons.com/source/arch/arm/mm/highmem.c?v=2.6.32#L39
void *kmap_atomic(struct page *page, enum km_type type)

而在新的内核中, 删除了这个标识, 但是保留了km_type的最大值KM_TYPE_NR

void *kmap_atomic(struct page *page)
{
    unsigned int idx;
    unsigned long vaddr;
    void *kmap;
    int type;

    preempt_disable();
    pagefault_disable();
    if (!PageHighMem(page))
        return page_address(page);

#ifdef CONFIG_DEBUG_HIGHMEM
    /*
     * There is no cache coherency issue when non VIVT, so force the
     * dedicated kmap usage for better debugging purposes in that case.
     */
    if (!cache_is_vivt())
        kmap = NULL;
    else
#endif
        kmap = kmap_high_get(page);
    if (kmap)
        return kmap;

    type = kmap_atomic_idx_push();

    idx = FIX_KMAP_BEGIN + type + KM_TYPE_NR * smp_processor_id();
    vaddr = __fix_to_virt(idx);
#ifdef CONFIG_DEBUG_HIGHMEM
    /*
     * With debugging enabled, kunmap_atomic forces that entry to 0.
     * Make sure it was indeed properly unmapped.
     */
    BUG_ON(!pte_none(get_fixmap_pte(vaddr)));
#endif
    /*
     * When debugging is off, kunmap_atomic leaves the previous mapping
     * in place, so the contained TLB flush ensures the TLB is updated
     * with the new mapping.
     */
    set_fixmap_pte(idx, mk_pte(page, kmap_prot));

    return (void *)vaddr;
}
EXPORT_SYMBOL(kmap_atomic);

这个函数不会被阻塞, 因此可以用在中断上下文和起亚不能重新调度的地方. 它也禁止内核抢占, 这是有必要的, 因此映射对每个处理器都是唯一的(调度可能对哪个处理器执行哪个进程做变动).

2.2 kunmap_atomic函数

可以通过函数kunmap_atomic取消映射

/*
 * Prevent people trying to call kunmap_atomic() as if it were kunmap()
 * kunmap_atomic() should get the return value of kmap_atomic, not the page.
 */
#define kunmap_atomic(addr)                     
do {                                
    BUILD_BUG_ON(__same_type((addr), struct page *));       
    __kunmap_atomic(addr);                  
} while (0)

这个函数也不会阻塞. 在很多体系结构中, 除非激活了内核抢占, 否则kunmap_atomic根本无事可做, 因为只有在下一个临时映射到来前上一个临时映射才有效. 因此, 内核完全可以”忘掉”kmap_atomic映射, kunmap_atomic也无需做什么实际的事情. 下一个原子映射将自动覆盖前一个映射.