现在的锁方案是flock?

[friskfly]

To avoid data crupption, we use a read-write lock to ensure that data modification is exclusive. But when a program is writting data when something bad, for example, a SIGKILL, happens that crashes the program before the write operation is complete and lock is released, other processes may not be able to enter the exclusive region again. I do not use an auto recovery lock such as flock, which will automatically release when process exits, just in case that wrong data is returned when performing a reading operation, or even, causing a segment fault.

我看到描述中，进程退出是不能释放锁的，但看到代码实现中使用的是flock，是这样吗？

[kyriosli]

是的，我本来不想用flock因为进程出错后它自动释放会导致写入不完全，但后来加入了一个dirty位解决了这个问题。当进程获得写锁开始写操作前，先将dirty置为1，写成功释放锁之前，将dirty置为0. 这样如果一个进程获得了写锁，发现dirty为1，说明有一个进程写操作没完成就退出了，当前进程可以选择检查并恢复数据完整性，或者初始化整个内存区域（目前只实现了后者）。

[friskfly]

好的 thx