Hello Robin, On 15 May 2018 at 22:10, Robin Kuzmin <kuzmin.robin@xxxxxxxxx> wrote: > Michael, > Is the assumption correct that `select.2` ignores the `O_NONBLOCK` > flag in the file descriptors it operates on? > > Details: > I assume that `select.2` > (http://man7.org/linux/man-pages/man2/select.2.html) completely > ignores the fact that some (or all) of the file descriptors (the > `select.2` operates on) can be marked as non-blocking (O_NONBLOCK). > > I mean if > + the `select()` is called for a single socket, > + and that socket is non-blocking, > + and the `select()` time-out is non-zero, > then > + `select()` will not return immediately (returning (-1) and `errno` > equal to `EWOULDBLOCK` or `EAGAIN`) > + but will be suspended for the duration of the `select()` time-out. > > If the assumption is correct then it would save me some time if I saw > an explicit note about that in > http://man7.org/linux/man-pages/man2/select.2.html. I added this text to the description of O_NONBLOCK in open(2): Note that ithe setting of this flag has no effect on the operation of poll(2), select(2), epoll(7), and similar, since those interfaces merely inform the caller about whether a file descriptor is "ready", meaning that an I/O operation performed on the file descriptor with the O_NON‐ BLOCK flag clear would not block. This seems better than a note in select(2), since there are several related APIs where people might be confused about this point. Cheers, Michael -- Michael Kerrisk Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/ Linux/UNIX System Programming Training: http://man7.org/training/ -- To unsubscribe from this list: send the line "unsubscribe linux-man" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
