请教为什么在radio中对SPI的互斥访问保护不用mutex而用sem呢？

发布于 2011-12-23 16:24:52 浏览：3904 订阅该版

今天在移植touch.c的时候发现对spi的并发访问保护是用的sem，请教各位一下为什么不用mutex呢？ 不是说mutex专门为这种情况设计的吗，可以避免优先级反转。难道是因为代码粒度的原因？
下面是我搜索spi1_lock的结果，表明codec touch spi-flash都是通过sem来保护SPI访问。
```
Searching for 'spi1_lock'...
.at45dbxx.c(151):     rt_sem_take(&spi1_lock, RT_WAITING_FOREVER);
.at45dbxx.c(178):     rt_sem_release(&spi1_lock);
.at45dbxx.c(182):     rt_sem_take(&spi1_lock, RT_WAITING_FOREVER);
.at45dbxx.c(206):     rt_sem_release(&spi1_lock);
.at45dbxx.c(214):     rt_sem_take(&spi1_lock, RT_WAITING_FOREVER);
.at45dbxx.c(234):     rt_sem_release(&spi1_lock);
.oard.c(21): struct rt_semaphore spi1_lock;
.oard.c(319):         if (rt_sem_init(&spi1_lock, "spi1lock", 1, RT_IPC_FLAG_FIFO) != RT_EOK)
.codec.c(214):     rt_sem_take(&spi1_lock, RT_WAITING_FOREVER);
.codec.c(223):     rt_sem_release(&spi1_lock);
.sst25vfxx.c(45): #define spi_lock()    rt_sem_take(&spi1_lock, RT_WAITING_FOREVER);
.sst25vfxx.c(46): #define spi_unlock()  rt_sem_release(&spi1_lock);
.	ouch.c(63):         rt_sem_take(&spi1_lock, RT_WAITING_FOREVER);
.	ouch.c(122):         rt_sem_release(&spi1_lock);
.oard.h(93): extern struct rt_semaphore spi1_lock;
15 occurrence(s) have been found.
```