如何建设网站子页网络营销推广的特点

Linux内核中 spinlock相关数据结构和代码实现涉及的文件还是挺多的，这篇博客尝试从文件的角度来梳理一下 spinlock的相关数据结构和代码实现，适合想大概了解 Linux内核中 spinlock从上层 API到底层实现间的调用路径和传参变化，尤其适合了解 spinlock和 qspinlock之间的关系。

这里只以 spinlock的 加锁函数 为切入点，梳理该函数从上层 API到底层实现涉及的文件。

1. 关注 spin_lock()

spinlock的加锁函数是 spin_lock()
spin_lock()的位置：include/linux/spinlock.h
spin_lock()的内容：

static __always_inline void spin_lock(spinlock_t *lock)
{raw_spin_lock(&lock->rlock);
}

看到这里，需要关注该函数的入参类型：spinlock_t
spinlock_t的位置： include/linux/spinlock_types.h
spinlock_t的内容：

typedef struct spinlock {union {struct raw_spinlock rlock;#ifdef CONFIG_DEBUG_LOCK_ALLOC
# define LOCK_PADSIZE (offsetof(struct raw_spinlock, dep_map))struct {u8 __padding[LOCK_PADSIZE];struct lockdep_map dep_map;};
#endif};
} spinlock_t;

2. 关注 __raw_spin_lock()

从上面可以看到，spin_lock()调用了 raw_spin_lock(), 而 raw_spin_lock()是一个宏，将它展开后得到：__raw_spin_lock()，同时假定 CONFIG_INLINE_SPIN_LOCK配置项打开，则
__raw_spin_lock()的位置：include/linux/spinlock_api_smp.h
__raw_spin_lock()的内容：

static inline void __raw_spin_lock(raw_spinlock_t *lock)                                       
{preempt_disable();spin_acquire(&lock->dep_map, 0, 0, _RET_IP_);LOCK_CONTENDED(lock, do_raw_spin_trylock, do_raw_spin_lock);
}

__raw_spin_lock()的参数类型是：raw_spinlock_t，它和 struct raw_spinlock是同一个类型，对应 spinlock_t中 rlock成员变量的类型
raw_spinlock_t的位置：include/linux/spinlock_types.h
raw_spinloct_t的内容：

typedef struct raw_spinlock {arch_spinlock_t raw_lock;
#ifdef CONFIG_DEBUG_SPINLOCK          unsigned int magic, owner_cpu;void *owner;                
#endif                                
#ifdef CONFIG_DEBUG_LOCK_ALLOC        struct lockdep_map dep_map;
#endif                                
} raw_spinlock_t;

3. 关注 do_raw_spin_lock()

从上面可以看到， __raw_spin_lock()调用了 do_raw_spin_lock()
do_raw_spin_lock()的位置：include/linux/spinlock.h
do_raw_spin_lock()的内容：

static inline void do_raw_spin_lock(raw_spinlock_t *lock) __acquires(lock)
{__acquire(lock);arch_spin_lock(&lock->raw_lock);
}

4. 关注 arch_spin_lock() 和 arch_spinlock_t

从上面可以看到，do_raw_spin_lock()调用了 arch_spin_lock()
arch_spin_lock()和它的入参 arch_spinlock_t是与体系结构相关的，这里面基本能分成两派，
一派是自己实现 arch_spin_lock()和 arch_spinlock_t，
一派是使用通用的 qspinlock实现，也即 #define arch_spin_lock(l) queued_spin_lock(l)和 typedef struct qspinlock {...} arch_spinlock_t

体系结构自己实现的架构包括：alpha, arc, arm, hexagon, ia64, parisc, powerpc, riscv, s390, sh, sparc-32, xtensa
具体列举其中的 5 个：
alpha架构
arch_spin_lock()的位置：arch/alpha/include/asm/spinlock.h
arch_spinlock_t的位置：arch/alpha/include/asm/spinlock_types.h

arc架构
arch_spin_lock()的位置：arch/arc/include/asm/spinlock.h
arch_spinlock_t的位置：arch/arc/include/asm/spinlock_types.h

arm架构
arch_spin_lock()的位置：arch/arm/include/asm/spinlock.h
arch_spinlock_t的位置：arch/arm/include/asm/spinlock_types.h

hexagon架构
arch_spin_lock()的位置：arch/hexagon/include/asm/spinlock.h
arch_spinlock_t的位置：arch/hexagon/include/asm/spinlock_types.h

ia64架构
arch_spin_lock()的位置：arch/ia64/include/asm/spinlock.h
arch_spinlock_t的位置：arch/ia64/include/asm/spinlock_types.h

使用 qspinlock实现的架构包括：mips, sparc-64, x86, arm64, openrisc
queued_spin_lock()的位置：include/asm-generic/qspinlock.h
queued_spin_lock()的内容是：

static __always_inline void queued_spin_lock(struct qspinlock *lock)
{u32 val;val = atomic_cmpxchg_acquire(&lock->val, 0, _Q_LOCKED_VAL);if (likely(val == 0)) return;queued_spin_lock_slowpath(lock, val);
}

struct qspinlock的位置：include/asm-generic/qspinlock_types.h
struct qspinlock的内容是：

typedef struct qspinlock {union {atomic_t val;/** By using the whole 2nd least significant byte for the* pending bit, we can allow better optimization of the lock* acquisition for the pending bit holder.*/
#ifdef __LITTLE_ENDIANstruct {u8  locked;u8  pending;};struct {u16 locked_pending;u16 tail;};
#elsestruct {u16 tail;u16 locked_pending;};struct {u8  reserved[2];u8  pending;u8  locked;};
#endif};
} arch_spinlock_t;