ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libev/ev.pod
(Generate patch)

Comparing libev/ev.pod (file contents):
Revision 1.448 by root, Sun Jun 23 02:02:24 2019 UTC vs.
Revision 1.451 by root, Mon Jun 24 00:19:26 2019 UTC

… …
584be well worth enabling it - if it isn't available in your kernel this will 584be well worth enabling it - if it isn't available in your kernel this will
585be detected and this backend will be skipped. 585be detected and this backend will be skipped.
586 586
587This backend can batch oneshot requests and supports a user-space ring 587This backend can batch oneshot requests and supports a user-space ring
588buffer to receive events. It also doesn't suffer from most of the design 588buffer to receive events. It also doesn't suffer from most of the design
589problems of epoll (such as not being able to remove event sources from 589problems of epoll (such as not being able to remove event sources from the
590the epoll set), and generally sounds too good to be true. Because, this 590epoll set), and generally sounds too good to be true. Because, this being
591being the linux kernel, of course it suffers from a whole new set of 591the linux kernel, of course it suffers from a whole new set of limitations.
592limitations.
593 592
594For one, it is not easily embeddable (but probably could be done using 593For one, it is not easily embeddable (but probably could be done using
595an event fd at some extra overhead). It also is subject to a system wide 594an event fd at some extra overhead). It also is subject to a system wide
596limit that can be configured in F</proc/sys/fs/aio-max-nr> - each loop 595limit that can be configured in F</proc/sys/fs/aio-max-nr> - each loop
597currently requires C<61> of this number. If no aio requests are left, this 596currently requires C<61> of this number. If no aio requests are left, this
598backend will be skipped during initialisation. 597backend will be skipped during initialisation.
599 598
600Most problematic in practise, however, is that not all file descriptors 599Most problematic in practise, however, is that not all file descriptors
601work with it. For example, in linux 5.1, tcp sockets, pipes, event fds, 600work with it. For example, in linux 5.1, tcp sockets, pipes, event fds,
602files, F</dev/null> and a few others are supported, but ttys do not work 601files, F</dev/null> and a few others are supported, but ttys do not work
603(probably because of a bug), so this is not (yet?) a generic event polling 602properly (a known bug that the kernel developers don't care about, see
604interface. 603L<https://lore.kernel.org/patchwork/patch/1047453/>), so this is not
604(yet?) a generic event polling interface.
605 605
606Overall, it seems the linux developers just don't want it to have a
607generic event handling mechanism other than C<select> or C<poll>.
608
606To work around this latter problem, the current version of libev uses 609To work around the fd type problem, the current version of libev uses
607epoll as a fallback for file deescriptor types that do not work. Epoll 610epoll as a fallback for file deescriptor types that do not work. Epoll
608is used in, kind of, slow mode that hopefully avoids most of its design 611is used in, kind of, slow mode that hopefully avoids most of its design
609problems. 612problems and requires 1-3 extra syscalls per active fd every iteration.
610 613
611This backend maps C<EV_READ> and C<EV_WRITE> in the same way as 614This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
612C<EVBACKEND_POLL>. 615C<EVBACKEND_POLL>.
613 616
614=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones) 617=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines