On 02/06/2013 10:41 PM, Martin Sustrik wrote:
> When implementing network protocols in user space, one has to implement
> fake user-space file descriptors to represent the sockets for the protocol.
>
> While all the BSD socket API functionality for such descriptors may be faked as
> well (myproto_send(), myproto_recv() etc.) this approach doesn't work for
> polling (select, poll, epoll). For polling, real system-level file descriptor
> is needed.
>
> In theory, eventfd may be used for this purpose, except that it is well suited
> only for signaling POLLIN. With some hacking it can be also used to signal
> POLLOUT and POLLERR, however:
>
> I. There's no way to signal POLLPRI, POLLHUP etc.
> II. There's no way to signal arbitraty combination of POLL* flags. Most notably,
> !POLLIN & !POLLOUT, which is a perfectly valid combination for a network
> protocol (rx buffer is empty and tx buffer is full), cannot be signaled
> using current implementation of eventfd.
>
> This patch implements new EFD_MASK flag which attempts to solve this problem.
>
> Additionally, when implementing network protocols in user space, there's a
> need to associate user-space state with the each "socket". If eventfd object is
> used as a reference to the socket, it should be possible to associate an opaque
> pointer to user-space data with it.
>
> The semantics of EFD_MASK are as follows:
>
> eventfd(2):
>
> If eventfd is created with EFD_MASK flag set, it is initialised in such a way
> as to signal no events on the file descriptor when it is polled on. 'initval'
> argument is ignored.
>
> write(2):
>
> User is allowed to write only buffers containing the following structure:
>
> struct efd_mask {
> short events;
> void *ptr;
> };
IMO that should be u64 ptr to avoid compat problems.
>
> The value of 'events' should be any combination of event flags as defined by
> poll(2) function (POLLIN, POLLOUT, POLLERR, POLLHUP etc.) Specified events will
> be signaled when polling (select, poll, epoll) on the eventfd is done later on.
> 'ptr' is an opaque pointer that is not interpreted by eventfd object.
How does this interact with EPOLLET?
--Andy
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
