arm linux多核同时获取读写锁, 会死循环吗, 还是都得到写锁

2017-01-18

文来自本人的旧博客: http://blog.163.com/awaken_ing/blog/static/120613197201631133058159

本篇只提问, 不解答, 祝开心.

如果读写锁是unlocked, 然后cpu 0和cpu 1都想获取写锁, 如表格所示, CPU 0在t2时刻执行的是wfene, 这时, CPU 1执行的是ldrex, 然后是否会有cpu能够获取到锁? 源码来自 linux-3.10.86/arch/arm/include/asm/spinlock.h archwritelock()

时间	CPU0	时间	CPU1
t0/t6	1: ldrex %0, [%1]
t1/t7	teq %0, #0
t2	WFE("ne")	t2	1: ldrex %0, [%1]
t3	strexeq %0, %2, [%1]	t3	teq %0, #0
t4	teq %0, #0	t4	WFE("ne")
t5	bne 1b	t5	strexeq %0, %2, [%1]
t6		t6	teq %0, #0
t7		t7	bne 1b

CPU 0和CPU 1都拿到写锁? 两个CPU都死循环? 好像不应该啊? 本篇只提问, 不解答, 祝开心. ^_^

附 linux-3.10.86/arch/arm/include/asm/spinlock.h

static inline void arch_write_lock(arch_rwlock_t *rw)
{
    unsigned long tmp;

    __asm__ __volatile__(
"1: ldrex   %0, [%1]\n"
"   teq %0, #0\n"
    WFE("ne")
"   strexeq %0, %2, [%1]\n"
"   teq %0, #0\n"
"   bne 1b"
    : "=&r" (tmp)
    : "r" (&rw->lock), "r" (0x80000000)
    : "cc");

    smp_mb();
}

本文地址: https://awakening-fong.github.io/posts/arm/arm_smp_rwlock

转载请注明出处: https://awakening-fong.github.io

ARM 12

ARM 12

若无法评论, 请打开JavaScript, 并通过proxy.