lpj@lpj-linux:~$ which ls
/bin/ls
2. 用命令搜索该软件所在包，代码如下：
lpj@lpj-linux:~$ dpkg -S /bin/ls
coreutils: /bin/ls
3. 从上一步中可以知道ls命令的实现在包coreutils中，用apt安装(说安装有些歧义，主要是区分apt-get -d)该包的源代码然后解压，代码如下：
sudo apt-get source coreutils
cd /usr/src/coreutils-XXX #XXX表示版本号
sudo tar zxvf coreutils-XXX.tar.gz
或者只下载源码，然后手动打补丁再解压，代码如下：
sudo apt-get -d source coreutils
cd /usr/src
tar zxvf coreutils-XXX.tar.gz
gzip -d coreutils-XXX.diff.gz #这一步会生成coreutils-XXX.diff文件
patch -p0 <coreutils-XXX.diff
cd coreutils-XXX
tar zxvf coreutils-XXX.tar.gz
OK，这几步执行完后，就可以进入/usr/src/coreutils-XXX/coreutils-XXX/src中查看各命令对应的源代码了
mutex相关的函数并不是linux kernel实现的，而是glibc实现的，源码位于nptl目录下。
http://ftp.gnu.org/pub/gnu/glibc/glibc-2.3.5.tar.gz
首先说数据结构：
typedef union
{
struct
{
int __lock
unsigned int __count
int __owner
unsigned int __nusers
/* KIND must stay at this position in the structure to maintain
binary compatibility. */
int __kind
int __spins
} __data
char __size[__SIZEOF_PTHREAD_MUTEX_T]
long int __align
} pthread_mutex_t
int __lock 资源竞争引用计数
int __kind锁类型，init 函数中mutexattr 参数传递，该参数可以为NULL，一般为 PTHREAD_MUTEX_NORMAL
结构体其他元素暂时不了解，以后更新。
/*nptl/pthread_mutex_init.c*/
int
__pthread_mutex_init (mutex, mutexattr)
pthread_mutex_t *mutex
const pthread_mutexattr_t *mutexattr
{
const struct pthread_mutexattr *imutexattr
assert (sizeof (pthread_mutex_t) <= __SIZEOF_PTHREAD_MUTEX_T)
imutexattr = (const struct pthread_mutexattr *) mutexattr ?: &default_attr
/* Clear the whole variable. */
memset (mutex, '\0', __SIZEOF_PTHREAD_MUTEX_T)
/* Copy the values from the attribute. */
mutex->__data.__kind = imutexattr->mutexkind &~0x80000000
/* Default values: mutex not used yet. */
// mutex->__count = 0 already done by memset
// mutex->__owner = 0 already done by memset
// mutex->__nusers = 0 already done by memset
// mutex->__spins = 0 already done by memset
return 0
}
init函数就比较简单了，将mutex结构体清零，设置结构体中__kind属性。
/*nptl/pthread_mutex_lock.c*/
int
__pthread_mutex_lock (mutex)
pthread_mutex_t *mutex
{
assert (sizeof (mutex->__size) >= sizeof (mutex->__data))
pid_t id = THREAD_GETMEM (THREAD_SELF, tid)
switch (__builtin_expect (mutex->__data.__kind, PTHREAD_MUTEX_TIMED_NP))
{
…
default:
/* Correct code cannot set any other type. */
case PTHREAD_MUTEX_TIMED_NP:
simple:
/* Normal mutex. */
LLL_MUTEX_LOCK (mutex->__data.__lock)
break
…
}
/* Record the ownership. */
assert (mutex->__data.__owner == 0)
mutex->__data.__owner = id
#ifndef NO_INCR
++mutex->__data.__nusers
#endif
return 0
}
该函数主要是调用LLL_MUTEX_LOCK， 省略部分为根据mutex结构体__kind属性不同值做些处理。
宏定义函数LLL_MUTEX_LOCK最终调用，将结构体mutex的__lock属性作为参数传递进来
#define __lll_mutex_lock(futex) \
((void) ({ \
int *__futex = (futex) \
if (atomic_compare_and_exchange_bool_acq (__futex, 1, 0) != 0) \
__lll_lock_wait (__futex) \
}))
atomic_compare_and_exchange_bool_acq (__futex, 1, 0)宏定义为：
#define atomic_compare_and_exchange_bool_acq(mem, newval, oldval) \
({ __typeof (mem) __gmemp = (mem) \
__typeof (*mem) __gnewval = (newval) \
\
*__gmemp == (oldval) ? (*__gmemp = __gnewval, 0) : 1})
这个宏实现的功能是：
如果mem的值等于oldval，则把newval赋值给mem，放回0，否则不做任何处理，返回1.
由此可以看出，当mutex锁限制的资源没有竞争时，__lock 属性被置为1，并返回0，不会调用__lll_lock_wait (__futex)当存在竞争时，再次调用lock函数，该宏不做任何处理，返回1，调用__lll_lock_wait (__futex)
void
__lll_lock_wait (int *futex)
{
do
{
int oldval = atomic_compare_and_exchange_val_acq (futex, 2, 1)
if (oldval != 0)
lll_futex_wait (futex, 2)
}
while (atomic_compare_and_exchange_bool_acq (futex, 2, 0) != 0)
}
atomic_compare_and_exchange_val_acq (futex, 2, 1)宏定义：
/* The only basic operation needed is compare and exchange. */
#define atomic_compare_and_exchange_val_acq(mem, newval, oldval) \
({ __typeof (mem) __gmemp = (mem) \
__typeof (*mem) __gret = *__gmemp \
__typeof (*mem) __gnewval = (newval) \
\
if (__gret == (oldval)) \
*__gmemp = __gnewval \
__gret})
这个宏实现的功能是，当mem等于oldval时，将mem置为newval，始终返回mem原始值。
此时，futex等于1，futex将被置为2，并且返回1. 进而调用
lll_futex_wait (futex, 2)
#define lll_futex_timed_wait(ftx, val, timespec) \
({ \
DO_INLINE_SYSCALL(futex, 4, (long) (ftx), FUTEX_WAIT, (int) (val), \
(long) (timespec)) \
_r10 == -1 ? -_retval : _retval \
})
该宏对于不同的平台架构会用不同的实现，采用汇编语言实现系统调用。不过确定的是调用了Linux kernel的futex系统调用。
futex在linux kernel的实现位于：kernel/futex.c
SYSCALL_DEFINE6(futex, u32 __user *, uaddr, int, op, u32, val,
struct timespec __user *, utime, u32 __user *, uaddr2,
u32, val3)
{
struct timespec ts
ktime_t t, *tp = NULL
u32 val2 = 0
int cmd = op &FUTEX_CMD_MASK
if (utime &&(cmd == FUTEX_WAIT || cmd == FUTEX_LOCK_PI ||
cmd == FUTEX_WAIT_BITSET ||
cmd == FUTEX_WAIT_REQUEUE_PI)) {
if (copy_from_user(&ts, utime, sizeof(ts)) != 0)
return -EFAULT
if (!timespec_valid(&ts))
return -EINVAL
t = timespec_to_ktime(ts)
if (cmd == FUTEX_WAIT)
t = ktime_add_safe(ktime_get(), t)
tp = &t
}
/*
* requeue parameter in 'utime' if cmd == FUTEX_*_REQUEUE_*.
* number of waiters to wake in 'utime' if cmd == FUTEX_WAKE_OP.
*/
if (cmd == FUTEX_REQUEUE || cmd == FUTEX_CMP_REQUEUE ||
cmd == FUTEX_CMP_REQUEUE_PI || cmd == FUTEX_WAKE_OP)
val2 = (u32) (unsigned long) utime
return do_futex(uaddr, op, val, tp, uaddr2, val2, val3)
}
futex具有六个形参，pthread_mutex_lock最终只关注了前四个。futex函数对参数进行判断和转化之后，直接调用do_futex。
long do_futex(u32 __user *uaddr, int op, u32 val, ktime_t *timeout,
u32 __user *uaddr2, u32 val2, u32 val3)
{
int clockrt, ret = -ENOSYS
int cmd = op &FUTEX_CMD_MASK
int fshared = 0
if (!(op &FUTEX_PRIVATE_FLAG))
fshared = 1
clockrt = op &FUTEX_CLOCK_REALTIME
if (clockrt &&cmd != FUTEX_WAIT_BITSET &&cmd != FUTEX_WAIT_REQUEUE_PI)
return -ENOSYS
switch (cmd) {
case FUTEX_WAIT:
val3 = FUTEX_BITSET_MATCH_ANY
case FUTEX_WAIT_BITSET:
ret = futex_wait(uaddr, fshared, val, timeout, val3, clockrt)
break
…
default:
ret = -ENOSYS
}
return ret
}
省略部分为对其他cmd的处理，pthread_mutex_lock函数最终传入的cmd参数为FUTEX_WAIT，所以在此只关注此分之，分析futex_wait函数的实现。
static int futex_wait(u32 __user *uaddr, int fshared,
u32 val, ktime_t *abs_time, u32 bitset, int clockrt)
{
struct hrtimer_sleeper timeout, *to = NULL
struct restart_block *restart
struct futex_hash_bucket *hb
struct futex_q q
int ret
… … //delete parameters check and convertion
retry:
/* Prepare to wait on uaddr. */
ret = futex_wait_setup(uaddr, val, fshared, &q, &hb)
if (ret)
goto out
/* queue_me and wait for wakeup, timeout, or a signal. */
futex_wait_queue_me(hb, &q, to)
… … //other handlers
return ret
}
futex_wait_setup 将线程放进休眠队列中，
futex_wait_queue_me(hb, &q, to)将本线程休眠，等待唤醒。
唤醒后，__lll_lock_wait函数中的while (atomic_compare_and_exchange_bool_acq (futex, 2, 0) != 0)语句将被执行，由于此时futex在pthread_mutex_unlock中置为0，所以atomic_compare_and_exchange_bool_acq (futex, 2, 0)语句将futex置为2，返回0. 退出循环，访问用户控件的临界资源。
/*nptl/pthread_mutex_unlock.c*/
int
internal_function attribute_hidden
__pthread_mutex_unlock_usercnt (mutex, decr)
pthread_mutex_t *mutex
int decr
{
switch (__builtin_expect (mutex->__data.__kind, PTHREAD_MUTEX_TIMED_NP))
{
… …
default:
/* Correct code cannot set any other type. */
case PTHREAD_MUTEX_TIMED_NP:
case PTHREAD_MUTEX_ADAPTIVE_NP:
/* Normal mutex. Nothing special to do. */
break
}
/* Always reset the owner field. */
mutex->__data.__owner = 0
if (decr)
/* One less user. */
--mutex->__data.__nusers
/* Unlock. */
lll_mutex_unlock (mutex->__data.__lock)
return 0
}
省略部分是针对不同的__kind属性值做的一些处理，最终调用 lll_mutex_unlock。
该宏函数最终的定义为：
#define __lll_mutex_unlock(futex) \
((void) ({ \
int *__futex = (futex) \
int __val = atomic_exchange_rel (__futex, 0) \
\
if (__builtin_expect (__val >1, 0)) \
lll_futex_wake (__futex, 1) \
}))
atomic_exchange_rel (__futex, 0)宏为：
#define atomic_exchange_rel(mem, value) \
(__sync_synchronize (), __sync_lock_test_and_set (mem, value))
实现功能为：将mem设置为value，返回原始mem值。
__builtin_expect (__val >1, 0) 是编译器优化语句，告诉编译器期望值，也就是大多数情况下__val >1 ？是0，其逻辑判断依然为if(__val >1)为真的话执行 lll_futex_wake。
现在分析，在资源没有被竞争的情况下，__futex 为1，那么返回值__val则为1，那么 lll_futex_wake (__futex, 1) 不会被执行，不产生系统调用。 当资源产生竞争的情况时，根据对pthread_mutex_lock 函数的分析，__futex为2， __val则为2，执行 lll_futex_wake (__futex, 1)从而唤醒等在临界资源的线程。
lll_futex_wake (__futex, 1)最终会调动同一个系统调用，即futex, 只是传递的cmd参数为FUTEX_WAKE。
在linux kernel的futex实现中，调用
static int futex_wake(u32 __user *uaddr, int fshared, int nr_wake, u32 bitset)
{
struct futex_hash_bucket *hb
struct futex_q *this, *next
struct plist_head *head
union futex_key key = FUTEX_KEY_INIT
int ret
if (!bitset)
return -EINVAL
ret = get_futex_key(uaddr, fshared, &key)
if (unlikely(ret != 0))
goto out
hb = hash_futex(&key)
spin_lock(&hb->lock)
head = &hb->chain
plist_for_each_entry_safe(this, next, head, list) {
if (match_futex (&this->key, &key)) {
if (this->pi_state || this->rt_waiter) {
ret = -EINVAL
break
}
/* Check if one of the bits is set in both bitsets */
if (!(this->bitset &bitset))
continue
wake_futex(this)
if (++ret >= nr_wake)
break
}
}
spin_unlock(&hb->lock)
put_futex_key(fshared, &key)
out:
return ret
}
该函数遍历在该mutex上休眠的所有线程，调用wake_futex进行唤醒，
static void wake_futex(struct futex_q *q)
{
struct task_struct *p = q->task
/*
* We set q->lock_ptr = NULL _before_ we wake up the task. If
* a non futex wake up happens on another CPU then the task
* might exit and p would dereference a non existing task
* struct. Prevent this by holding a reference on p across the
* wake up.
*/
get_task_struct(p)
plist_del(&q->list, &q->list.plist)
/*
* The waiting task can free the futex_q as soon as
* q->lock_ptr = NULL is written, without taking any locks. A
* memory barrier is required here to prevent the following
* store to lock_ptr from getting ahead of the plist_del.
*/
smp_wmb()
q->lock_ptr = NULL
wake_up_state(p, TASK_NORMAL)
put_task_struct(p)
}
wake_up_state(p, TASK_NORMAL) 的实现位于kernel/sched.c中，属于linux进程调度的技术。
欢迎分享，转载请注明来源：内存溢出
微信扫一扫
支付宝扫一扫
评论列表（0条）