What does poll() do with a timeout of 0?

Question

I'm looking at the poll() man page, and it tells me the behavior of poll() when positive and negative values are passed in for the timeout parameter. It doesn't doesn't tell me what happens if timeout is 0. Any ideas?

Looking at the epoll_wait() man page, it tells me that with a timeout value of 0, it will return right away, even if there are no events available. Is it safe to assume that poll() would behave the same way?

Solution 1

It will return immediately:

If timeout is greater than zero, it specifies a maximum interval (in milliseconds) to wait for any file descriptor to become ready. If timeout is zero, then poll() will return without blocking. If the value of timeout is -1, the poll blocks indefinitely.

, as of Mac OS X 10.5;

Maximum interval to wait for the poll to complete, in milliseconds. If this value is 0, poll() will return immediately. If this value is INFTIM (-1), poll() will block indefinitely until a condition is found.

, as of OpenBSD 3.8

Solution 2

As I see it, waiting for a timeout means "having" a timeout. This way I would expect that poll() actually checks the file descriptors, and then waits if no one is ready to a timeout of 0 milliseconds (no wait at all). But the case is that it will just signal if a fd is available.

I also checked linux source code and to my knowledge, this is the way it works: first calculates the "future" waiting point, then checks the file descriptors, then if none available, waits for the timeout specified time.

Regards,