Linus Torvalds writes: (Summary) wrote:
Actually, let's make it even *more* obvious, and even easier for lockdep (and for humans) to see what's going on.
lockdep (and for humans) to see what's going on.
So I think the best model would be something like this: So I think the best model would be something like this: - T1:
mutex_lock(&lock)
...
mutex_transfer(&lock)
mutex_transfer(&lock)
- T2:
mutex_receive(&lock);
...
mutex_unlock(&lock);
mutex_unlock(&lock);
where the "mutex_transfer() -> And then "mutex_receive()" can verify that (a) the ownership was that "in tranfer" thing and that it's still held, before it then sets the ownership to thread 2.
it then sets the ownership to thread 2.
Wouldn't that be easier for lockdep?
[...]
operations are the fancy ones.Actually, let's make it even *more* obvious, and even easier for lockdep (and for humans) to see what's going on.
lockdep (and for humans) to see what's going on.
So I think the best model would be something like this: So I think the best model would be something like this: - T1:
mutex_lock(&lock)
...
mutex_transfer(&lock)
mutex_transfer(&lock)
- T2:
mutex_receive(&lock);
...
mutex_unlock(&lock);
mutex_unlock(&lock);
where the "mutex_transfer() -> And then "mutex_receive()" can verify that (a) the ownership was that "in tranfer" thing and that it's still held, before it then sets the ownership to thread 2.
it then sets the ownership to thread 2.
Wouldn't that be easier for lockdep?