A few months ago, Davide Libenzi wrote this message on lkml: http://marc.info/?l=linux-kernel&m=119368443105320&w=2 but apparently hasn't found time yet to review the man page. So I figured it might work better if I propose a patch and get Davide to ack it or suggest further changes. My proposed patch to epoll(7) is below. Feedback would be appreciated. In particular, I removed reference to threads in A1 because I was unable to find a test case where I didn't get EEXIST. Maybe I'm missing something. Chris --- man-pages-2.78/man7/epoll.7.orig 2008-02-11 06:53:44.000000000 -0500 +++ man-pages-2.78/man7/epoll.7 2008-02-21 23:43:53.000000000 -0500 @@ -234,11 +234,11 @@ What happens if you add the same file descriptor to an epoll set twice? .TP .B A1 -You will probably get +You will get .BR EEXIST . -However, it is possible that two -threads may add the same file descriptor twice. -This is a harmless condition. +However, it is possible to add a +.BR dup (2) +of that file descriptor to the epoll set. .TP .B Q2 Can two @@ -285,7 +285,15 @@ sets automatically? .TP .B A6 -Yes. +Only if the underlying open file description (see +.BR open (2)) +is closed. +For example, if the file descriptor was duplicated with a +.BR dup (2) +or +.BR fork (2) +call, it will only be removed from epoll sets when all duplicate +descriptors are closed. .TP .B Q7 If more than one event occurs between - 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
