path: root/nsock/src/engine_epoll.c

/***************************************************************************
 * engine_epoll.c -- epoll(7) based IO engine.                             *
 *                                                                         *
 ***********************IMPORTANT NSOCK LICENSE TERMS***********************
 *
 * The nsock parallel socket event library is (C) 1999-2023 Nmap Software LLC
 * This library is free software; you may redistribute and/or modify it under
 * the terms of the GNU General Public License as published by the Free Software
 * Foundation; Version 2. This guarantees your right to use, modify, and
 * redistribute this software under certain conditions. If this license is
 * unacceptable to you, Nmap Software LLC may be willing to sell alternative
 * licenses (contact sales@nmap.com ).
 *
 * As a special exception to the GPL terms, Nmap Software LLC grants permission
 * to link the code of this program with any version of the OpenSSL library
 * which is distributed under a license identical to that listed in the included
 * docs/licenses/OpenSSL.txt file, and distribute linked combinations including
 * the two. You must obey the GNU GPL in all respects for all of the code used
 * other than OpenSSL. If you modify this file, you may extend this exception to
 * your version of the file, but you are not obligated to do so.
 *
 * If you received these files with a written license agreement stating terms
 * other than the (GPL) terms above, then that alternative license agreement
 * takes precedence over this comment.
 *
 * Source is provided to this software because we believe users have a right to
 * know exactly what a program is going to do before they run it. This also
 * allows you to audit the software for security holes.
 *
 * Source code also allows you to port Nmap to new platforms, fix bugs, and add
 * new features. You are highly encouraged to send your changes to the
 * dev@nmap.org mailing list for possible incorporation into the main
 * distribution. By sending these changes to Fyodor or one of the Insecure.Org
 * development mailing lists, or checking them into the Nmap source code
 * repository, it is understood (unless you specify otherwise) that you are
 * offering the Nmap Project (Nmap Software LLC) the unlimited, non-exclusive
 * right to reuse, modify, and relicense the code. Nmap will always be available
 * Open Source, but this is important because the inability to relicense code
 * has caused devastating problems for other Free Software projects (such as KDE
 * and NASM). We also occasionally relicense the code to third parties as
 * discussed above. If you wish to specify special license conditions of your
 * contributions, just say so when you send them.
 *
 * This program 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 General Public License v2.0 for more
 * details (http://www.gnu.org/licenses/gpl-2.0.html).
 *
 ***************************************************************************/

/* $Id$ */

#ifdef HAVE_CONFIG_H
#include "nsock_config.h"
#endif

#if HAVE_EPOLL

#include <sys/epoll.h>
#include <errno.h>

#include "nsock_internal.h"
#include "nsock_log.h"

#if HAVE_PCAP
#include "nsock_pcap.h"
#endif

#define INITIAL_EV_COUNT  128

#define EPOLL_R_FLAGS (EPOLLIN | EPOLLPRI)
#define EPOLL_W_FLAGS EPOLLOUT

/* EPOLLRDHUP was introduced later and might be unavailable on older systems. */
#ifndef EPOLLRDHUP
  #define EPOLLRDHUP 0
#endif
#define EPOLL_X_FLAGS (EPOLLERR | EPOLLRDHUP| EPOLLHUP)


/* --- ENGINE INTERFACE PROTOTYPES --- */
static int epoll_init(struct npool *nsp);
static void epoll_destroy(struct npool *nsp);
static int epoll_iod_register(struct npool *nsp, struct niod *iod, struct nevent *nse, int ev);
static int epoll_iod_unregister(struct npool *nsp, struct niod *iod);
static int epoll_iod_modify(struct npool *nsp, struct niod *iod, struct nevent *nse, int ev_set, int ev_clr);
static int epoll_loop(struct npool *nsp, int msec_timeout);

extern struct io_operations posix_io_operations;

/* ---- ENGINE DEFINITION ---- */
struct io_engine engine_epoll = {
  "epoll",
  epoll_init,
  epoll_destroy,
  epoll_iod_register,
  epoll_iod_unregister,
  epoll_iod_modify,
  epoll_loop,
  &posix_io_operations
};


/* --- INTERNAL PROTOTYPES --- */
static void iterate_through_event_lists(struct npool *nsp, int evcount);

/* defined in nsock_core.c */
void process_iod_events(struct npool *nsp, struct niod *nsi, int ev);
void process_event(struct npool *nsp, gh_list_t *evlist, struct nevent *nse, int ev);
void process_expired_events(struct npool *nsp);
#if HAVE_PCAP
#ifndef PCAP_CAN_DO_SELECT
int pcap_read_on_nonselect(struct npool *nsp);
#endif
#endif

/* defined in nsock_event.c */
void update_first_events(struct nevent *nse);


extern struct timeval nsock_tod;


/*
 * Engine specific data structure
 */
struct epoll_engine_info {
  /* file descriptor corresponding to our epoll instance */
  int epfd;
  /* number of epoll_events we can deal with */
  int evlen;
  /* list of epoll events, resized if necessary (when polling over large numbers of IODs) */
  struct epoll_event *events;
};


int epoll_init(struct npool *nsp) {
  struct epoll_engine_info *einfo;

  einfo = (struct epoll_engine_info *)safe_malloc(sizeof(struct epoll_engine_info));

  einfo->epfd = epoll_create(10); /* argument is ignored */
  einfo->evlen = INITIAL_EV_COUNT;
  einfo->events = (struct epoll_event *)safe_malloc(einfo->evlen * sizeof(struct epoll_event));

  nsp->engine_data = (void *)einfo;

  return 1;
}

void epoll_destroy(struct npool *nsp) {
  struct epoll_engine_info *einfo = (struct epoll_engine_info *)nsp->engine_data;

  assert(einfo != NULL);
  close(einfo->epfd);
  free(einfo->events);
  free(einfo);
}

int epoll_iod_register(struct npool *nsp, struct niod *iod, struct nevent *nse, int ev) {
  int sd;
  struct epoll_event epev;
  struct epoll_engine_info *einfo = (struct epoll_engine_info *)nsp->engine_data;

  assert(!IOD_PROPGET(iod, IOD_REGISTERED));

  iod->watched_events = ev;

  memset(&epev, 0x00, sizeof(struct epoll_event));
  epev.events = EPOLLET;
  epev.data.ptr = (void *)iod;

  if (ev & EV_READ)
    epev.events |= EPOLL_R_FLAGS;
  if (ev & EV_WRITE)
    epev.events |= EPOLL_W_FLAGS;

  sd = nsock_iod_get_sd(iod);
  if (epoll_ctl(einfo->epfd, EPOLL_CTL_ADD, sd, &epev) < 0)
    fatal("Unable to register IOD #%lu: %s", iod->id, strerror(errno));

  IOD_PROPSET(iod, IOD_REGISTERED);
  return 1;
}

int epoll_iod_unregister(struct npool *nsp, struct niod *iod) {
  iod->watched_events = EV_NONE;

  /* some IODs can be unregistered here if they're associated to an event that was
   * immediately completed */
  if (IOD_PROPGET(iod, IOD_REGISTERED)) {
    struct epoll_engine_info *einfo = (struct epoll_engine_info *)nsp->engine_data;
    int sd;

    sd = nsock_iod_get_sd(iod);
    epoll_ctl(einfo->epfd, EPOLL_CTL_DEL, sd, NULL);

    IOD_PROPCLR(iod, IOD_REGISTERED);
  }
  return 1;
}

int epoll_iod_modify(struct npool *nsp, struct niod *iod, struct nevent *nse, int ev_set, int ev_clr) {
  int sd;
  struct epoll_event epev;
  int new_events;
  struct epoll_engine_info *einfo = (struct epoll_engine_info *)nsp->engine_data;

  assert((ev_set & ev_clr) == 0);
  assert(IOD_PROPGET(iod, IOD_REGISTERED));

  memset(&epev, 0x00, sizeof(struct epoll_event));
  epev.events = EPOLLET;
  epev.data.ptr = (void *)iod;

  new_events = iod->watched_events;
  new_events |= ev_set;
  new_events &= ~ev_clr;

  if (new_events == iod->watched_events)
    return 1; /* nothing to do */

  iod->watched_events = new_events;

  /* regenerate the current set of events for this IOD */
  if (iod->watched_events & EV_READ)
    epev.events |= EPOLL_R_FLAGS;
  if (iod->watched_events & EV_WRITE)
    epev.events |= EPOLL_W_FLAGS;

  sd = nsock_iod_get_sd(iod);

  if (epoll_ctl(einfo->epfd, EPOLL_CTL_MOD, sd, &epev) < 0)
    fatal("Unable to update events for IOD #%lu: %s", iod->id, strerror(errno));

  return 1;
}

int epoll_loop(struct npool *nsp, int msec_timeout) {
  int results_left = 0;
  int event_msecs; /* msecs before an event goes off */
  int combined_msecs;
  int sock_err = 0;
  unsigned int iod_count;
  struct epoll_engine_info *einfo = (struct epoll_engine_info *)nsp->engine_data;

  assert(msec_timeout >= -1);

  if (nsp->events_pending == 0)
    return 0; /* No need to wait on 0 events ... */


  iod_count = gh_list_count(&nsp->active_iods);
  if (iod_count > einfo->evlen) {
    einfo->evlen = iod_count * 2;
    einfo->events = (struct epoll_event *)safe_realloc(einfo->events, einfo->evlen * sizeof(struct epoll_event));
  }

  do {
    struct nevent *nse;

    nsock_log_debug_all("wait for events");

    nse = next_expirable_event(nsp);
    if (!nse)
      event_msecs = -1; /* None of the events specified a timeout */
    else
      event_msecs = MAX(0, TIMEVAL_MSEC_SUBTRACT(nse->timeout, nsock_tod));

#if HAVE_PCAP
#ifndef PCAP_CAN_DO_SELECT
    /* Force a low timeout when capturing packets on systems where
     * the pcap descriptor is not select()able. */
    if (gh_list_count(&nsp->pcap_read_events) > 0)
      if (event_msecs > PCAP_POLL_INTERVAL)
        event_msecs = PCAP_POLL_INTERVAL;
#endif
#endif

    /* We cast to unsigned because we want -1 to be very high (since it means no
     * timeout) */
    combined_msecs = MIN((unsigned)event_msecs, (unsigned)msec_timeout);

#if HAVE_PCAP
#ifndef PCAP_CAN_DO_SELECT
    /* do non-blocking read on pcap devices that doesn't support select()
     * If there is anything read, just leave this loop. */
    if (pcap_read_on_nonselect(nsp)) {
      /* okay, something was read. */
    } else
#endif
#endif
    {
      results_left = epoll_wait(einfo->epfd, einfo->events, einfo->evlen, combined_msecs);
      if (results_left == -1)
        sock_err = socket_errno();
    }

    gettimeofday(&nsock_tod, NULL); /* Due to epoll delay */
  } while (results_left == -1 && sock_err == EINTR); /* repeat only if signal occurred */

  if (results_left == -1 && sock_err != EINTR) {
    nsock_log_error("nsock_loop error %d: %s", sock_err, socket_strerror(sock_err));
    nsp->errnum = sock_err;
    return -1;
  }

  iterate_through_event_lists(nsp, results_left);

  return 1;
}


/* ---- INTERNAL FUNCTIONS ---- */
static inline int get_evmask(struct epoll_engine_info *einfo, int n) {
  int evmask = EV_NONE;

  if (einfo->events[n].events & EPOLL_R_FLAGS)
    evmask |= EV_READ;
  if (einfo->events[n].events & EPOLL_W_FLAGS)
    evmask |= EV_WRITE;
  if (einfo->events[n].events & EPOLL_X_FLAGS)
    evmask |= EV_EXCEPT;

  return evmask;
}

/* Iterate through all the event lists (such as connect_events, read_events,
 * timer_events, etc) and take action for those that have completed (due to
 * timeout, i/o, etc) */
void iterate_through_event_lists(struct npool *nsp, int evcount) {
  struct epoll_engine_info *einfo = (struct epoll_engine_info *)nsp->engine_data;
  int n;

  for (n = 0; n < evcount; n++) {
    struct niod *nsi = (struct niod *)einfo->events[n].data.ptr;

    assert(nsi);

    /* process all the pending events for this IOD */
    process_iod_events(nsp, nsi, get_evmask(einfo, n));

    if (nsi->state == NSIOD_STATE_DELETED) {
      gh_list_remove(&nsp->active_iods, &nsi->nodeq);
      gh_list_prepend(&nsp->free_iods, &nsi->nodeq);
    }
  }

  /* iterate through timers and expired events */
  process_expired_events(nsp);
}

#endif /* HAVE_EPOLL */