haproxy/include/types/fd.h
Willy Tarreau 037d2c1f8f MAJOR: sepoll: make the poller totally event-driven
At the moment sepoll is not 100% event-driven, because a call to fd_set()
on an event which is already being polled will not change its state.

This causes issues with OpenSSL because if some I/O processing is interrupted
after clearing the I/O event (eg: read all data from a socket, can't put it
all into the buffer), then there is no way to call the SSL_read() again once
the buffer releases some space.

The only real solution is to go 100% event-driven. The principle is to use
the spec list as an event cache and that each time an I/O event is reported
by epoll_wait(), this event is automatically scheduled for addition to the
spec list for future calls until the consumer explicitly asks for polling
or stopping.

Doing this is a bit tricky because sepoll used to provide a substantial
number of optimizations such as event merging. These optimizations have
been maintained : a dedicated update list is affected when events change,
but not the event list, so that updates may cancel themselves without any
side effect such as displacing events. A specific case was considered for
handling newly created FDs as soon as they are detected from within the
poll loop. This ensures that their read or write operation will always be
attempted as soon as possible, thus reducing the number of poll loops and
process_session wakeups. This is especially true for newly accepted fds
which immediately perform their first recv() call.

Two new flags were added to the fdtab[] struct to tag the fact that a file
descriptor already exists in the update list. One flag indicates that a
file descriptor is new and has just been created (fdtab[].new) and the other
one indicates that a file descriptor is already referenced by the update list
(fdtab[].updated). Even if the FD state changes during operations or if the
fd is closed and replaced, it's not an issue because the update flag remains
and is easily spotted during list walks. The flag must absolutely reflect the
presence of the fd in the update list in order to avoid overflowing the update
list with more events than there are distinct fds.

Note that this change also recovers the small performance loss introduced
by its connection counter-part and goes even beyond.
2012-11-10 00:17:27 +01:00

114 lines
4.8 KiB
C

/*
* include/types/fd.h
* File descriptors states.
*
* Copyright (C) 2000-2009 Willy Tarreau - w@1wt.eu
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation, version 2.1
* exclusively.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#ifndef _TYPES_FD_H
#define _TYPES_FD_H
#include <common/config.h>
#include <types/port_range.h>
enum {
DIR_RD=0,
DIR_WR=1,
DIR_SIZE
};
/*
* FD_POLL_IN remains set as long as some data is pending for read.
* FD_POLL_OUT remains set as long as the fd accepts to write data.
* FD_POLL_ERR and FD_POLL_ERR remain set forever (until processed).
*/
#define FD_POLL_IN 0x01
#define FD_POLL_PRI 0x02
#define FD_POLL_OUT 0x04
#define FD_POLL_ERR 0x08
#define FD_POLL_HUP 0x10
#define FD_POLL_DATA (FD_POLL_IN | FD_POLL_OUT)
#define FD_POLL_STICKY (FD_POLL_ERR | FD_POLL_HUP)
/* info about one given fd */
struct fdtab {
int (*iocb)(int fd); /* I/O handler, returns FD_WAIT_* */
void *owner; /* the connection or listener associated with this fd, NULL if closed */
unsigned int spec_p; /* speculative polling: position in spec list+1. 0=not in list. */
unsigned char spec_e; /* speculative polling: read and write events status. 4 bits */
unsigned char ev; /* event seen in return of poll() : FD_POLL_* */
unsigned char new:1; /* 1 if this fd has just been created */
unsigned char updated:1; /* 1 if this fd is already in the update list */
};
/* less often used information */
struct fdinfo {
struct port_range *port_range; /* optional port range to bind to */
int local_port; /* optional local port */
};
/*
* Poller descriptors.
* - <name> is initialized by the poller's register() function, and should not
* be allocated, just linked to.
* - <pref> is initialized by the poller's register() function. It is set to 0
* by default, meaning the poller is disabled. init() should set it to 0 in
* case of failure. term() must set it to 0. A generic unoptimized select()
* poller should set it to 100.
* - <private> is initialized by the poller's init() function, and cleaned by
* the term() function.
* - clo() should be used to do indicate the poller that fd will be closed. It
* may be the same as rem() on some pollers.
* - poll() calls the poller, expiring at <exp>
*/
struct poller {
void *private; /* any private data for the poller */
int REGPRM2 (*is_set)(const int fd, int dir); /* check if <fd> is being polled for dir <dir> */
void REGPRM2 (*set)(const int fd, int dir); /* set polling on <fd> for <dir> */
void REGPRM2 (*clr)(const int fd, int dir); /* clear polling on <fd> for <dir> */
void REGPRM2 (*wai)(const int fd, int dir); /* wait for polling on <fd> for <dir> */
void REGPRM1 (*rem)(const int fd); /* remove any polling on <fd> */
void REGPRM1 (*clo)(const int fd); /* mark <fd> as closed */
void REGPRM2 (*poll)(struct poller *p, int exp); /* the poller itself */
int REGPRM1 (*init)(struct poller *p); /* poller initialization */
void REGPRM1 (*term)(struct poller *p); /* termination of this poller */
int REGPRM1 (*test)(struct poller *p); /* pre-init check of the poller */
int REGPRM1 (*fork)(struct poller *p); /* post-fork re-opening */
const char *name; /* poller name */
int pref; /* try pollers with higher preference first */
};
extern struct poller cur_poller; /* the current poller */
extern int nbpollers;
#define MAX_POLLERS 10
extern struct poller pollers[MAX_POLLERS]; /* all registered pollers */
extern struct fdtab *fdtab; /* array of all the file descriptors */
extern struct fdinfo *fdinfo; /* less-often used infos for file descriptors */
extern int maxfd; /* # of the highest fd + 1 */
extern int totalconn; /* total # of terminated sessions */
extern int actconn; /* # of active sessions */
#endif /* _TYPES_FD_H */
/*
* Local variables:
* c-indent-level: 8
* c-basic-offset: 8
* End:
*/