POLLHUP vs. POLLRDHUP?

Question

According to the poll man page, the poll function can return POLLHUP and POLLRDHUP events. From what I understand, only POLLHUP is POSIX compliant, and POLLRDHUP is a Linux non-standard extension. Howerver, both seem to signal that the write end of a connection is closed, so I don't understand the added value of POLLRDHUP over POLLHUP. Would someone please explain the difference between the two?

Henton

It means the remote end did a partial shutdown of the socket and won't be sending any more data but can still receive. – Organology

@Organology When you say "it", do you mean POLLHUP but not POLLRDHUP? Or POLLRDHUP but not POLLHUP? Or do you mean they both mean this? Or what? – Betide

"it" is the added value of POLLRDHUP. Although for TCP, I don't see how the two events can be distinguished. TCP doesn't send any indication when the remote socket has closed for reading, only when it closes for writing. – Apteryx

@Shawn, Would POLLRDHUP ever be set for a file descriptor representing a FIFO? – Henton

@Apteryx The indication that the remove socket has closed for reading is that it has acknoledged our FIN packet. – Adventurism

@Henton no, POLLRDHUP will not be set for a FIFO. – Adventurism

@mosvy That's not true. The ACK just indicates that the FIN was received and the socket is still open. But you can't tell the difference between fully open and half-closed. – Apteryx

@Apteryx What's "not true"? That's how it works in Linux. When the remote end has acked the FIN, the kernel with turn the SEND_SHUTDOWN flag on the local socket. Please refer to the net/ipv4/tcp_input.c file in the kernel source. When the POLLRDHUP flag is set in revents but not POLLHUP, the socket is half-closed: you can still write but not read from it. – Adventurism

@mosvy I don't dispute that the receiver of the ACK does this, but I don't see how the ACK means that the remote end has half-closed the socket. – Apteryx

Answer 1

No, when poll()ing a socket, POLLHUP will signal that the connection was closed in both directions.

POLLRDHUP will be set when the other end has called shutdown(SHUT_WR) or when this end has called shutdown(SHUT_RD), but the connection may still be alive in the other direction.

You can have a look at net/ipv4/tcp.c the kernel source:

        if (sk->sk_shutdown == SHUTDOWN_MASK || state == TCP_CLOSE)
                mask |= EPOLLHUP;
        if (sk->sk_shutdown & RCV_SHUTDOWN)
                mask |= EPOLLIN | EPOLLRDNORM | EPOLLRDHUP;

SHUTDOWN_MASK is RCV_SHUTDOWN|SEND_SHUTDOWN. RCV_SHUTDOWN is set when a FIN packet is received, and SEND_SHUTDOWN is set when a FIN packet is acknowledged by the other end, and the socket moves to the FIN-WAIT2 state.

[except for the TCP_CLOSE part, that snippet is replicated by all protocols; and the whole thing works similarly for unix sockets, etc]

There are other important differences -- POLLRDHUP (unlike POLLHUP) has to be set explicitly in .events in order to be returned in .revents.

And POLLRDHUP only works on sockets, not on fifos/pipes or ttys.

Adventurism