From c7afa4fec330539186fb8ceb6f539b6ccc987403 Mon Sep 17 00:00:00 2001 From: Hasso Tepper Date: Wed, 30 Jan 2008 12:35:18 +0000 Subject: [PATCH] Import libevent-1.3e. --- contrib/libevent/buffer.c | 460 +++++ contrib/libevent/evbuffer.c | 413 ++++ contrib/libevent/evdns.3 | 322 +++ contrib/libevent/evdns.c | 3116 +++++++++++++++++++++++++++++ contrib/libevent/evdns.h | 367 ++++ contrib/libevent/event-internal.h | 61 + contrib/libevent/event.3 | 592 ++++++ contrib/libevent/event.c | 928 +++++++++ contrib/libevent/event.h | 340 ++++ contrib/libevent/event_tagging.c | 372 ++++ contrib/libevent/evsignal.h | 44 + contrib/libevent/kqueue.c | 419 ++++ contrib/libevent/log.c | 219 ++ contrib/libevent/log.h | 43 + contrib/libevent/poll.c | 391 ++++ contrib/libevent/select.c | 373 ++++ contrib/libevent/signal.c | 195 ++ 17 files changed, 8655 insertions(+) create mode 100644 contrib/libevent/buffer.c create mode 100644 contrib/libevent/evbuffer.c create mode 100644 contrib/libevent/evdns.3 create mode 100644 contrib/libevent/evdns.c create mode 100644 contrib/libevent/evdns.h create mode 100644 contrib/libevent/event-internal.h create mode 100644 contrib/libevent/event.3 create mode 100644 contrib/libevent/event.c create mode 100644 contrib/libevent/event.h create mode 100644 contrib/libevent/event_tagging.c create mode 100644 contrib/libevent/evsignal.h create mode 100644 contrib/libevent/kqueue.c create mode 100644 contrib/libevent/log.c create mode 100644 contrib/libevent/log.h create mode 100644 contrib/libevent/poll.c create mode 100644 contrib/libevent/select.c create mode 100644 contrib/libevent/signal.c diff --git a/contrib/libevent/buffer.c b/contrib/libevent/buffer.c new file mode 100644 index 0000000000..0327eb5493 --- /dev/null +++ b/contrib/libevent/buffer.c @@ -0,0 +1,460 @@ +/* + * Copyright (c) 2002, 2003 Niels Provos + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#ifdef HAVE_VASPRINTF +/* If we have vasprintf, we need to define this before we include stdio.h. */ +#define _GNU_SOURCE +#endif + +#include + +#ifdef HAVE_SYS_TIME_H +#include +#endif + +#ifdef HAVE_SYS_IOCTL_H +#include +#endif + +#include +#include +#include +#include +#include +#ifdef HAVE_STDARG_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "event.h" + +struct evbuffer * +evbuffer_new(void) +{ + struct evbuffer *buffer; + + buffer = calloc(1, sizeof(struct evbuffer)); + + return (buffer); +} + +void +evbuffer_free(struct evbuffer *buffer) +{ + if (buffer->orig_buffer != NULL) + free(buffer->orig_buffer); + free(buffer); +} + +/* + * This is a destructive add. The data from one buffer moves into + * the other buffer. + */ + +#define SWAP(x,y) do { \ + (x)->buffer = (y)->buffer; \ + (x)->orig_buffer = (y)->orig_buffer; \ + (x)->misalign = (y)->misalign; \ + (x)->totallen = (y)->totallen; \ + (x)->off = (y)->off; \ +} while (0) + +int +evbuffer_add_buffer(struct evbuffer *outbuf, struct evbuffer *inbuf) +{ + int res; + + /* Short cut for better performance */ + if (outbuf->off == 0) { + struct evbuffer tmp; + size_t oldoff = inbuf->off; + + /* Swap them directly */ + SWAP(&tmp, outbuf); + SWAP(outbuf, inbuf); + SWAP(inbuf, &tmp); + + /* + * Optimization comes with a price; we need to notify the + * buffer if necessary of the changes. oldoff is the amount + * of data that we transfered from inbuf to outbuf + */ + if (inbuf->off != oldoff && inbuf->cb != NULL) + (*inbuf->cb)(inbuf, oldoff, inbuf->off, inbuf->cbarg); + if (oldoff && outbuf->cb != NULL) + (*outbuf->cb)(outbuf, 0, oldoff, outbuf->cbarg); + + return (0); + } + + res = evbuffer_add(outbuf, inbuf->buffer, inbuf->off); + if (res == 0) { + /* We drain the input buffer on success */ + evbuffer_drain(inbuf, inbuf->off); + } + + return (res); +} + +int +evbuffer_add_vprintf(struct evbuffer *buf, const char *fmt, va_list ap) +{ + char *buffer; + size_t space; + size_t oldoff = buf->off; + int sz; + va_list aq; + + /* make sure that at least some space is available */ + evbuffer_expand(buf, 64); + for (;;) { + size_t used = buf->misalign + buf->off; + buffer = (char *)buf->buffer + buf->off; + assert(buf->totallen >= used); + space = buf->totallen - used; + +#ifndef va_copy +#define va_copy(dst, src) memcpy(&(dst), &(src), sizeof(va_list)) +#endif + va_copy(aq, ap); + +#ifdef WIN32 + sz = vsnprintf(buffer, space - 1, fmt, aq); + buffer[space - 1] = '\0'; +#else + sz = vsnprintf(buffer, space, fmt, aq); +#endif + + va_end(aq); + + if (sz < 0) + return (-1); + if (sz < space) { + buf->off += sz; + if (buf->cb != NULL) + (*buf->cb)(buf, oldoff, buf->off, buf->cbarg); + return (sz); + } + if (evbuffer_expand(buf, sz + 1) == -1) + return (-1); + + } + /* NOTREACHED */ +} + +int +evbuffer_add_printf(struct evbuffer *buf, const char *fmt, ...) +{ + int res = -1; + va_list ap; + + va_start(ap, fmt); + res = evbuffer_add_vprintf(buf, fmt, ap); + va_end(ap); + + return (res); +} + +/* Reads data from an event buffer and drains the bytes read */ + +int +evbuffer_remove(struct evbuffer *buf, void *data, size_t datlen) +{ + size_t nread = datlen; + if (nread >= buf->off) + nread = buf->off; + + memcpy(data, buf->buffer, nread); + evbuffer_drain(buf, nread); + + return (nread); +} + +/* + * Reads a line terminated by either '\r\n', '\n\r' or '\r' or '\n'. + * The returned buffer needs to be freed by the called. + */ + +char * +evbuffer_readline(struct evbuffer *buffer) +{ + u_char *data = EVBUFFER_DATA(buffer); + size_t len = EVBUFFER_LENGTH(buffer); + char *line; + unsigned int i; + + for (i = 0; i < len; i++) { + if (data[i] == '\r' || data[i] == '\n') + break; + } + + if (i == len) + return (NULL); + + if ((line = malloc(i + 1)) == NULL) { + fprintf(stderr, "%s: out of memory\n", __func__); + evbuffer_drain(buffer, i); + return (NULL); + } + + memcpy(line, data, i); + line[i] = '\0'; + + /* + * Some protocols terminate a line with '\r\n', so check for + * that, too. + */ + if ( i < len - 1 ) { + char fch = data[i], sch = data[i+1]; + + /* Drain one more character if needed */ + if ( (sch == '\r' || sch == '\n') && sch != fch ) + i += 1; + } + + evbuffer_drain(buffer, i + 1); + + return (line); +} + +/* Adds data to an event buffer */ + +static void +evbuffer_align(struct evbuffer *buf) +{ + memmove(buf->orig_buffer, buf->buffer, buf->off); + buf->buffer = buf->orig_buffer; + buf->misalign = 0; +} + +/* Expands the available space in the event buffer to at least datlen */ + +int +evbuffer_expand(struct evbuffer *buf, size_t datlen) +{ + size_t need = buf->misalign + buf->off + datlen; + + /* If we can fit all the data, then we don't have to do anything */ + if (buf->totallen >= need) + return (0); + + /* + * If the misalignment fulfills our data needs, we just force an + * alignment to happen. Afterwards, we have enough space. + */ + if (buf->misalign >= datlen) { + evbuffer_align(buf); + } else { + void *newbuf; + size_t length = buf->totallen; + + if (length < 256) + length = 256; + while (length < need) + length <<= 1; + + if (buf->orig_buffer != buf->buffer) + evbuffer_align(buf); + if ((newbuf = realloc(buf->buffer, length)) == NULL) + return (-1); + + buf->orig_buffer = buf->buffer = newbuf; + buf->totallen = length; + } + + return (0); +} + +int +evbuffer_add(struct evbuffer *buf, const void *data, size_t datlen) +{ + size_t need = buf->misalign + buf->off + datlen; + size_t oldoff = buf->off; + + if (buf->totallen < need) { + if (evbuffer_expand(buf, datlen) == -1) + return (-1); + } + + memcpy(buf->buffer + buf->off, data, datlen); + buf->off += datlen; + + if (datlen && buf->cb != NULL) + (*buf->cb)(buf, oldoff, buf->off, buf->cbarg); + + return (0); +} + +void +evbuffer_drain(struct evbuffer *buf, size_t len) +{ + size_t oldoff = buf->off; + + if (len >= buf->off) { + buf->off = 0; + buf->buffer = buf->orig_buffer; + buf->misalign = 0; + goto done; + } + + buf->buffer += len; + buf->misalign += len; + + buf->off -= len; + + done: + /* Tell someone about changes in this buffer */ + if (buf->off != oldoff && buf->cb != NULL) + (*buf->cb)(buf, oldoff, buf->off, buf->cbarg); + +} + +/* + * Reads data from a file descriptor into a buffer. + */ + +#define EVBUFFER_MAX_READ 4096 + +int +evbuffer_read(struct evbuffer *buf, int fd, int howmuch) +{ + u_char *p; + size_t oldoff = buf->off; + int n = EVBUFFER_MAX_READ; +#ifdef WIN32 + DWORD dwBytesRead; +#endif + +#ifdef FIONREAD + if (ioctl(fd, FIONREAD, &n) == -1 || n == 0) { + n = EVBUFFER_MAX_READ; + } else if (n > EVBUFFER_MAX_READ && n > howmuch) { + /* + * It's possible that a lot of data is available for + * reading. We do not want to exhaust resources + * before the reader has a chance to do something + * about it. If the reader does not tell us how much + * data we should read, we artifically limit it. + */ + if (n > buf->totallen << 2) + n = buf->totallen << 2; + if (n < EVBUFFER_MAX_READ) + n = EVBUFFER_MAX_READ; + } +#endif + if (howmuch < 0 || howmuch > n) + howmuch = n; + + /* If we don't have FIONREAD, we might waste some space here */ + if (evbuffer_expand(buf, howmuch) == -1) + return (-1); + + /* We can append new data at this point */ + p = buf->buffer + buf->off; + +#ifndef WIN32 + n = read(fd, p, howmuch); + if (n == -1) + return (-1); + if (n == 0) + return (0); +#else + n = ReadFile((HANDLE)fd, p, howmuch, &dwBytesRead, NULL); + if (n == 0) + return (-1); + if (dwBytesRead == 0) + return (0); + n = dwBytesRead; +#endif + + buf->off += n; + + /* Tell someone about changes in this buffer */ + if (buf->off != oldoff && buf->cb != NULL) + (*buf->cb)(buf, oldoff, buf->off, buf->cbarg); + + return (n); +} + +int +evbuffer_write(struct evbuffer *buffer, int fd) +{ + int n; +#ifdef WIN32 + DWORD dwBytesWritten; +#endif + +#ifndef WIN32 + n = write(fd, buffer->buffer, buffer->off); + if (n == -1) + return (-1); + if (n == 0) + return (0); +#else + n = WriteFile((HANDLE)fd, buffer->buffer, buffer->off, &dwBytesWritten, NULL); + if (n == 0) + return (-1); + if (dwBytesWritten == 0) + return (0); + n = dwBytesWritten; +#endif + evbuffer_drain(buffer, n); + + return (n); +} + +u_char * +evbuffer_find(struct evbuffer *buffer, const u_char *what, size_t len) +{ + u_char *search = buffer->buffer, *end = search + buffer->off; + u_char *p; + + while (search < end && + (p = memchr(search, *what, end - search)) != NULL) { + if (p + len > end) + break; + if (memcmp(p, what, len) == 0) + return (p); + search = p + 1; + } + + return (NULL); +} + +void evbuffer_setcb(struct evbuffer *buffer, + void (*cb)(struct evbuffer *, size_t, size_t, void *), + void *cbarg) +{ + buffer->cb = cb; + buffer->cbarg = cbarg; +} diff --git a/contrib/libevent/evbuffer.c b/contrib/libevent/evbuffer.c new file mode 100644 index 0000000000..52712bce58 --- /dev/null +++ b/contrib/libevent/evbuffer.c @@ -0,0 +1,413 @@ +/* + * Copyright (c) 2002-2004 Niels Provos + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include + +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#ifdef HAVE_SYS_TIME_H +#include +#endif + +#include +#include +#include +#include +#ifdef HAVE_STDARG_H +#include +#endif + +#include "event.h" + +/* prototypes */ + +void bufferevent_setwatermark(struct bufferevent *, short, size_t, size_t); +void bufferevent_read_pressure_cb(struct evbuffer *, size_t, size_t, void *); + +static int +bufferevent_add(struct event *ev, int timeout) +{ + struct timeval tv, *ptv = NULL; + + if (timeout) { + timerclear(&tv); + tv.tv_sec = timeout; + ptv = &tv; + } + + return (event_add(ev, ptv)); +} + +/* + * This callback is executed when the size of the input buffer changes. + * We use it to apply back pressure on the reading side. + */ + +void +bufferevent_read_pressure_cb(struct evbuffer *buf, size_t old, size_t now, + void *arg) { + struct bufferevent *bufev = arg; + /* + * If we are below the watermark then reschedule reading if it's + * still enabled. + */ + if (bufev->wm_read.high == 0 || now < bufev->wm_read.high) { + evbuffer_setcb(buf, NULL, NULL); + + if (bufev->enabled & EV_READ) + bufferevent_add(&bufev->ev_read, bufev->timeout_read); + } +} + +static void +bufferevent_readcb(int fd, short event, void *arg) +{ + struct bufferevent *bufev = arg; + int res = 0; + short what = EVBUFFER_READ; + size_t len; + int howmuch = -1; + + if (event == EV_TIMEOUT) { + what |= EVBUFFER_TIMEOUT; + goto error; + } + + /* + * If we have a high watermark configured then we don't want to + * read more data than would make us reach the watermark. + */ + if (bufev->wm_read.high != 0) + howmuch = bufev->wm_read.high; + + res = evbuffer_read(bufev->input, fd, howmuch); + if (res == -1) { + if (errno == EAGAIN || errno == EINTR) + goto reschedule; + /* error case */ + what |= EVBUFFER_ERROR; + } else if (res == 0) { + /* eof case */ + what |= EVBUFFER_EOF; + } + + if (res <= 0) + goto error; + + bufferevent_add(&bufev->ev_read, bufev->timeout_read); + + /* See if this callbacks meets the water marks */ + len = EVBUFFER_LENGTH(bufev->input); + if (bufev->wm_read.low != 0 && len < bufev->wm_read.low) + return; + if (bufev->wm_read.high != 0 && len > bufev->wm_read.high) { + struct evbuffer *buf = bufev->input; + event_del(&bufev->ev_read); + + /* Now schedule a callback for us */ + evbuffer_setcb(buf, bufferevent_read_pressure_cb, bufev); + return; + } + + /* Invoke the user callback - must always be called last */ + if (bufev->readcb != NULL) + (*bufev->readcb)(bufev, bufev->cbarg); + return; + + reschedule: + bufferevent_add(&bufev->ev_read, bufev->timeout_read); + return; + + error: + (*bufev->errorcb)(bufev, what, bufev->cbarg); +} + +static void +bufferevent_writecb(int fd, short event, void *arg) +{ + struct bufferevent *bufev = arg; + int res = 0; + short what = EVBUFFER_WRITE; + + if (event == EV_TIMEOUT) { + what |= EVBUFFER_TIMEOUT; + goto error; + } + + if (EVBUFFER_LENGTH(bufev->output)) { + res = evbuffer_write(bufev->output, fd); + if (res == -1) { +#ifndef WIN32 +/*todo. evbuffer uses WriteFile when WIN32 is set. WIN32 system calls do not + *set errno. thus this error checking is not portable*/ + if (errno == EAGAIN || + errno == EINTR || + errno == EINPROGRESS) + goto reschedule; + /* error case */ + what |= EVBUFFER_ERROR; + +#else + goto reschedule; +#endif + + } else if (res == 0) { + /* eof case */ + what |= EVBUFFER_EOF; + } + if (res <= 0) + goto error; + } + + if (EVBUFFER_LENGTH(bufev->output) != 0) + bufferevent_add(&bufev->ev_write, bufev->timeout_write); + + /* + * Invoke the user callback if our buffer is drained or below the + * low watermark. + */ + if (bufev->writecb != NULL && + EVBUFFER_LENGTH(bufev->output) <= bufev->wm_write.low) + (*bufev->writecb)(bufev, bufev->cbarg); + + return; + + reschedule: + if (EVBUFFER_LENGTH(bufev->output) != 0) + bufferevent_add(&bufev->ev_write, bufev->timeout_write); + return; + + error: + (*bufev->errorcb)(bufev, what, bufev->cbarg); +} + +/* + * Create a new buffered event object. + * + * The read callback is invoked whenever we read new data. + * The write callback is invoked whenever the output buffer is drained. + * The error callback is invoked on a write/read error or on EOF. + * + * Both read and write callbacks maybe NULL. The error callback is not + * allowed to be NULL and have to be provided always. + */ + +struct bufferevent * +bufferevent_new(int fd, evbuffercb readcb, evbuffercb writecb, + everrorcb errorcb, void *cbarg) +{ + struct bufferevent *bufev; + + if ((bufev = calloc(1, sizeof(struct bufferevent))) == NULL) + return (NULL); + + if ((bufev->input = evbuffer_new()) == NULL) { + free(bufev); + return (NULL); + } + + if ((bufev->output = evbuffer_new()) == NULL) { + evbuffer_free(bufev->input); + free(bufev); + return (NULL); + } + + event_set(&bufev->ev_read, fd, EV_READ, bufferevent_readcb, bufev); + event_set(&bufev->ev_write, fd, EV_WRITE, bufferevent_writecb, bufev); + + bufev->readcb = readcb; + bufev->writecb = writecb; + bufev->errorcb = errorcb; + + bufev->cbarg = cbarg; + + /* + * Set to EV_WRITE so that using bufferevent_write is going to + * trigger a callback. Reading needs to be explicitly enabled + * because otherwise no data will be available. + */ + bufev->enabled = EV_WRITE; + + return (bufev); +} + +int +bufferevent_priority_set(struct bufferevent *bufev, int priority) +{ + if (event_priority_set(&bufev->ev_read, priority) == -1) + return (-1); + if (event_priority_set(&bufev->ev_write, priority) == -1) + return (-1); + + return (0); +} + +/* Closing the file descriptor is the responsibility of the caller */ + +void +bufferevent_free(struct bufferevent *bufev) +{ + event_del(&bufev->ev_read); + event_del(&bufev->ev_write); + + evbuffer_free(bufev->input); + evbuffer_free(bufev->output); + + free(bufev); +} + +/* + * Returns 0 on success; + * -1 on failure. + */ + +int +bufferevent_write(struct bufferevent *bufev, void *data, size_t size) +{ + int res; + + res = evbuffer_add(bufev->output, data, size); + + if (res == -1) + return (res); + + /* If everything is okay, we need to schedule a write */ + if (size > 0 && (bufev->enabled & EV_WRITE)) + bufferevent_add(&bufev->ev_write, bufev->timeout_write); + + return (res); +} + +int +bufferevent_write_buffer(struct bufferevent *bufev, struct evbuffer *buf) +{ + int res; + + res = bufferevent_write(bufev, buf->buffer, buf->off); + if (res != -1) + evbuffer_drain(buf, buf->off); + + return (res); +} + +size_t +bufferevent_read(struct bufferevent *bufev, void *data, size_t size) +{ + struct evbuffer *buf = bufev->input; + + if (buf->off < size) + size = buf->off; + + /* Copy the available data to the user buffer */ + memcpy(data, buf->buffer, size); + + if (size) + evbuffer_drain(buf, size); + + return (size); +} + +int +bufferevent_enable(struct bufferevent *bufev, short event) +{ + if (event & EV_READ) { + if (bufferevent_add(&bufev->ev_read, bufev->timeout_read) == -1) + return (-1); + } + if (event & EV_WRITE) { + if (bufferevent_add(&bufev->ev_write, bufev->timeout_write) == -1) + return (-1); + } + + bufev->enabled |= event; + return (0); +} + +int +bufferevent_disable(struct bufferevent *bufev, short event) +{ + if (event & EV_READ) { + if (event_del(&bufev->ev_read) == -1) + return (-1); + } + if (event & EV_WRITE) { + if (event_del(&bufev->ev_write) == -1) + return (-1); + } + + bufev->enabled &= ~event; + return (0); +} + +/* + * Sets the read and write timeout for a buffered event. + */ + +void +bufferevent_settimeout(struct bufferevent *bufev, + int timeout_read, int timeout_write) { + bufev->timeout_read = timeout_read; + bufev->timeout_write = timeout_write; +} + +/* + * Sets the water marks + */ + +void +bufferevent_setwatermark(struct bufferevent *bufev, short events, + size_t lowmark, size_t highmark) +{ + if (events & EV_READ) { + bufev->wm_read.low = lowmark; + bufev->wm_read.high = highmark; + } + + if (events & EV_WRITE) { + bufev->wm_write.low = lowmark; + bufev->wm_write.high = highmark; + } + + /* If the watermarks changed then see if we should call read again */ + bufferevent_read_pressure_cb(bufev->input, + 0, EVBUFFER_LENGTH(bufev->input), bufev); +} + +int +bufferevent_base_set(struct event_base *base, struct bufferevent *bufev) +{ + int res; + + res = event_base_set(base, &bufev->ev_read); + if (res == -1) + return (res); + + res = event_base_set(base, &bufev->ev_write); + return (res); +} diff --git a/contrib/libevent/evdns.3 b/contrib/libevent/evdns.3 new file mode 100644 index 0000000000..10414fa2ef --- /dev/null +++ b/contrib/libevent/evdns.3 @@ -0,0 +1,322 @@ +.\" +.\" Copyright (c) 2006 Niels Provos +.\" All rights reserved. +.\" +.\" Redistribution and use in source and binary forms, with or without +.\" modification, are permitted provided that the following conditions +.\" are met: +.\" +.\" 1. Redistributions of source code must retain the above copyright +.\" notice, this list of conditions and the following disclaimer. +.\" 2. Redistributions in binary form must reproduce the above copyright +.\" notice, this list of conditions and the following disclaimer in the +.\" documentation and/or other materials provided with the distribution. +.\" 3. The name of the author may not be used to endorse or promote products +.\" derived from this software without specific prior written permission. +.\" +.\" THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, +.\" INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY +.\" AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL +.\" THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, +.\" EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, +.\" PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; +.\" OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, +.\" WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR +.\" OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF +.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +.\" +.Dd October 7, 2006 +.Dt EVDNS 3 +.Os +.Sh NAME +.Nm evdns_init +.Nm evdns_shutdown +.Nm evdns_err_to_string +.Nm evdns_nameserver_add +.Nm evdns_count_nameservers +.Nm evdns_clear_nameservers_and_suspend +.Nm evdns_resume +.Nm evdns_nameserver_ip_add +.Nm evdns_resolve_ipv4 +.Nm evdns_resolve_reverse +.Nm evdns_resolv_conf_parse +.Nm evdns_config_windows_nameservers +.Nm evdns_search_clear +.Nm evdns_search_add +.Nm evdns_search_ndots_set +.Nm evdns_set_log_fn +.Nd asynchronous functions for DNS resolution. +.Sh SYNOPSIS +.Fd #include +.Fd #include +.Fd #include +.Ft int +.Fn evdns_init +.Ft void +.Fn evdns_shutdown "int fail_requests" +.Ft "const char *" +.Fn evdns_err_to_string "int err" +.Ft int +.Fn evdns_nameserver_add "unsigned long int address" +.Ft int +.Fn evdns_count_nameservers +.Ft int +.Fn evdns_clear_nameservers_and_suspend +.Ft int +.Fn evdns_resume +.Ft int +.Fn evdns_nameserver_ip_add(const char *ip_as_string); +.Ft int +.Fn evdns_resolve_ipv4 "const char *name" "int flags" "evdns_callback_type callback" "void *ptr" +.Ft int +.Fn evdns_resolve_reverse "struct in_addr *in" "int flags" "evdns_callback_type callback" "void *ptr" +.Ft int +.Fn evdns_resolv_conf_parse "int flags" "const char *" +.Ft void +.Fn evdns_search_clear +.Ft void +.Fn evdns_search_add "const char *domain" +.Ft void +.Fn evdns_search_ndots_set "const int ndots" +.Ft void +.Fn evdns_set_log_fn "evdns_debug_log_fn_type fn" +.Ft int +.Fn evdns_config_windows_nameservers +.Sh DESCRIPTION +Welcome, gentle reader +.Pp +Async DNS lookups are really a whole lot harder than they should be, +mostly stemming from the fact that the libc resolver has never been +very good at them. Before you use this library you should see if libc +can do the job for you with the modern async call getaddrinfo_a +(see http://www.imperialviolet.org/page25.html#e498). Otherwise, +please continue. +.Pp +This code is based on libevent and you must call event_init before +any of the APIs in this file. You must also seed the OpenSSL random +source if you are using OpenSSL for ids (see below). +.Pp +This library is designed to be included and shipped with your source +code. You statically link with it. You should also test for the +existence of strtok_r and define HAVE_STRTOK_R if you have it. +.Pp +The DNS protocol requires a good source of id numbers and these +numbers should be unpredictable for spoofing reasons. There are +three methods for generating them here and you must define exactly +one of them. In increasing order of preference: +.Pp +.Bl -tag -width "DNS_USE_GETTIMEOFDAY_FOR_ID" -compact -offset indent +.It DNS_USE_GETTIMEOFDAY_FOR_ID +Using the bottom 16 bits of the usec result from gettimeofday. This +is a pretty poor solution but should work anywhere. +.It DNS_USE_CPU_CLOCK_FOR_ID +Using the bottom 16 bits of the nsec result from the CPU's time +counter. This is better, but may not work everywhere. Requires +POSIX realtime support and you'll need to link against -lrt on +glibc systems at least. +.It DNS_USE_OPENSSL_FOR_ID +Uses the OpenSSL RAND_bytes call to generate the data. You must +have seeded the pool before making any calls to this library. +.El +.Pp +The library keeps track of the state of nameservers and will avoid +them when they go down. Otherwise it will round robin between them. +.Pp +Quick start guide: + #include "evdns.h" + void callback(int result, char type, int count, int ttl, + void *addresses, void *arg); + evdns_resolv_conf_parse(DNS_OPTIONS_ALL, "/etc/resolv.conf"); + evdns_resolve("www.hostname.com", 0, callback, NULL); +.Pp +When the lookup is complete the callback function is called. The +first argument will be one of the DNS_ERR_* defines in evdns.h. +Hopefully it will be DNS_ERR_NONE, in which case type will be +DNS_IPv4_A, count will be the number of IP addresses, ttl is the time +which the data can be cached for (in seconds), addresses will point +to an array of uint32_t's and arg will be whatever you passed to +evdns_resolve. +.Pp +Searching: +.Pp +In order for this library to be a good replacement for glibc's resolver it +supports searching. This involves setting a list of default domains, in +which names will be queried for. The number of dots in the query name +determines the order in which this list is used. +.Pp +Searching appears to be a single lookup from the point of view of the API, +although many DNS queries may be generated from a single call to +evdns_resolve. Searching can also drastically slow down the resolution +of names. +.Pp +To disable searching: +.Bl -enum -compact -offset indent +.It +Never set it up. If you never call +.Fn evdns_resolv_conf_parse, +.Fn evdns_init, +or +.Fn evdns_search_add +then no searching will occur. +.It +If you do call +.Fn evdns_resolv_conf_parse +then don't pass +.Va DNS_OPTION_SEARCH +(or +.Va DNS_OPTIONS_ALL, +which implies it). +.It +When calling +.Fn evdns_resolve, +pass the +.Va DNS_QUERY_NO_SEARCH +flag. +.El +.Pp +The order of searches depends on the number of dots in the name. If the +number is greater than the ndots setting then the names is first tried +globally. Otherwise each search domain is appended in turn. +.Pp +The ndots setting can either be set from a resolv.conf, or by calling +evdns_search_ndots_set. +.Pp +For example, with ndots set to 1 (the default) and a search domain list of +["myhome.net"]: + Query: www + Order: www.myhome.net, www. +.Pp + Query: www.abc + Order: www.abc., www.abc.myhome.net +.Pp +.Sh API reference +.Pp +.Bl -tag -width 0123456 +.It Ft int Fn evdns_init +Initializes support for non-blocking name resolution by calling +.Fn evdns_resolv_conf_parse +on UNIX and +.Fn evdns_config_windows_nameservers +on Windows. +.It Ft int Fn evdns_nameserver_add "unsigned long int address" +Add a nameserver. The address should be an IP address in +network byte order. The type of address is chosen so that +it matches in_addr.s_addr. +Returns non-zero on error. +.It Ft int Fn evdns_nameserver_ip_add "const char *ip_as_string" +This wraps the above function by parsing a string as an IP +address and adds it as a nameserver. +Returns non-zero on error +.It Ft int Fn evdns_resolve "const char *name" "int flags" "evdns_callback_type callback" "void *ptr" +Resolve a name. The name parameter should be a DNS name. +The flags parameter should be 0, or DNS_QUERY_NO_SEARCH +which disables searching for this query. (see defn of +searching above). +.Pp +The callback argument is a function which is called when +this query completes and ptr is an argument which is passed +to that callback function. +.Pp +Returns non-zero on error +.It Ft void Fn evdns_search_clear +Clears the list of search domains +.It Ft void Fn evdns_search_add "const char *domain" +Add a domain to the list of search domains +.It Ft void Fn evdns_search_ndots_set "int ndots" +Set the number of dots which, when found in a name, causes +the first query to be without any search domain. +.It Ft int Fn evdns_count_nameservers "void" +Return the number of configured nameservers (not necessarily the +number of running nameservers). This is useful for double-checking +whether our calls to the various nameserver configuration functions +have been successful. +.It Ft int Fn evdns_clear_nameservers_and_suspend "void" +Remove all currently configured nameservers, and suspend all pending +resolves. Resolves will not necessarily be re-attempted until +evdns_resume() is called. +.It Ft int Fn evdns_resume "void" +Re-attempt resolves left in limbo after an earlier call to +evdns_clear_nameservers_and_suspend(). +.It Ft int Fn evdns_config_windows_nameservers "void" +Attempt to configure a set of nameservers based on platform settings on +a win32 host. Preferentially tries to use GetNetworkParams; if that fails, +looks in the registry. Returns 0 on success, nonzero on failure. +.It Ft int Fn evdns_resolv_conf_parse "int flags" "const char *filename" +Parse a resolv.conf like file from the given filename. +.Pp +See the man page for resolv.conf for the format of this file. +The flags argument determines what information is parsed from +this file: +.Bl -tag -width "DNS_OPTION_NAMESERVERS" -offset indent -compact -nested +.It DNS_OPTION_SEARCH +domain, search and ndots options +.It DNS_OPTION_NAMESERVERS +nameserver lines +.It DNS_OPTION_MISC +timeout and attempts options +.It DNS_OPTIONS_ALL +all of the above +.El +.Pp +The following directives are not parsed from the file: + sortlist, rotate, no-check-names, inet6, debug +.Pp +Returns non-zero on error: +.Bl -tag -width "0" -offset indent -compact -nested +.It 0 +no errors +.It 1 +failed to open file +.It 2 +failed to stat file +.It 3 +file too large +.It 4 +out of memory +.It 5 +short read from file +.El +.El +.Sh Internals: +Requests are kept in two queues. The first is the inflight queue. In +this queue requests have an allocated transaction id and nameserver. +They will soon be transmitted if they haven't already been. +.Pp +The second is the waiting queue. The size of the inflight ring is +limited and all other requests wait in waiting queue for space. This +bounds the number of concurrent requests so that we don't flood the +nameserver. Several algorithms require a full walk of the inflight +queue and so bounding its size keeps thing going nicely under huge +(many thousands of requests) loads. +.Pp +If a nameserver loses too many requests it is considered down and we +try not to use it. After a while we send a probe to that nameserver +(a lookup for google.com) and, if it replies, we consider it working +again. If the nameserver fails a probe we wait longer to try again +with the next probe. +.Sh SEE ALSO +.Xr event 3 , +.Xr gethostbyname 3 , +.Xr resolv.conf 5 +.Sh HISTORY +The +.Nm evdns +API was developed by Adam Langley on top of the +.Nm libevent +API. +The code was integrate into +.Nm Tor +by Nick Mathewson and finally put into +.Nm libevent +itself by Niels Provos. +.Sh AUTHORS +The +.Nm evdns +API and code was written by Adam Langley with significant +contributions by Nick Mathewson. +.Sh BUGS +This documentation is neither complete nor authoritative. +If you are in doubt about the usage of this API then +check the source code to find out how it works, write +up the missing piece of documentation and send it to +me for inclusion in this man page. diff --git a/contrib/libevent/evdns.c b/contrib/libevent/evdns.c new file mode 100644 index 0000000000..1fc2075a4c --- /dev/null +++ b/contrib/libevent/evdns.c @@ -0,0 +1,3116 @@ +/* $Id: evdns.c 6979 2006-08-04 18:31:13Z nickm $ */ + +/* The original version of this module was written by Adam Langley; for + * a history of modifications, check out the subversion logs. + * + * When editing this module, try to keep it re-mergeable by Adam. Don't + * reformat the whitespace, add Tor dependencies, or so on. + * + * TODO: + * - Support IPv6 and PTR records. + * - Replace all externally visible magic numbers with #defined constants. + * - Write doccumentation for APIs of all external functions. + */ + +/* Async DNS Library + * Adam Langley + * http://www.imperialviolet.org/eventdns.html + * Public Domain code + * + * This software is Public Domain. To view a copy of the public domain dedication, + * visit http://creativecommons.org/licenses/publicdomain/ or send a letter to + * Creative Commons, 559 Nathan Abbott Way, Stanford, California 94305, USA. + * + * I ask and expect, but do not require, that all derivative works contain an + * attribution similar to: + * Parts developed by Adam Langley + * + * You may wish to replace the word "Parts" with something else depending on + * the amount of original code. + * + * (Derivative works does not include programs which link against, run or include + * the source verbatim in their source distributions) + * + * Version: 0.1b + */ + +#include +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#ifdef WIN32 +#include "misc.h" +#endif + +/* #define NDEBUG */ + +#ifndef DNS_USE_CPU_CLOCK_FOR_ID +#ifndef DNS_USE_GETTIMEOFDAY_FOR_ID +#ifndef DNS_USE_OPENSSL_FOR_ID +#error Must configure at least one id generation method. +#error Please see the documentation. +#endif +#endif +#endif + +/* #define _POSIX_C_SOURCE 200507 */ +#define _GNU_SOURCE + +#ifdef DNS_USE_CPU_CLOCK_FOR_ID +#ifdef DNS_USE_OPENSSL_FOR_ID +#error Multiple id options selected +#endif +#ifdef DNS_USE_GETTIMEOFDAY_FOR_ID +#error Multiple id options selected +#endif +#include +#endif + +#ifdef DNS_USE_OPENSSL_FOR_ID +#ifdef DNS_USE_GETTIMEOFDAY_FOR_ID +#error Multiple id options selected +#endif +#include +#endif + +#define _FORTIFY_SOURCE 3 + +#include +#include +#include +#ifdef HAVE_STDINT_H +#include +#endif +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "evdns.h" +#include "log.h" +#ifdef WIN32 +#include +#include +#include +#else +#include +#include +#include +#endif + +#ifdef HAVE_NETINET_IN6_H +#include +#endif + +#ifdef WIN32 +typedef int socklen_t; +#endif + +#define EVDNS_LOG_DEBUG 0 +#define EVDNS_LOG_WARN 1 + +#ifndef HOST_NAME_MAX +#define HOST_NAME_MAX 255 +#endif + +#ifndef NDEBUG +#include +#endif + +#undef MIN +#define MIN(a,b) ((a)<(b)?(a):(b)) + +#ifdef __USE_ISOC99B +/* libevent doesn't work without this */ +typedef uint8_t u_char; +typedef unsigned int uint; +#endif +#include + +#define u64 uint64_t +#define u32 uint32_t +#define u16 uint16_t +#define u8 uint8_t + +#define MAX_ADDRS 4 /* maximum number of addresses from a single packet */ +/* which we bother recording */ + +#define TYPE_A EVDNS_TYPE_A +#define TYPE_CNAME 5 +#define TYPE_PTR EVDNS_TYPE_PTR +#define TYPE_AAAA EVDNS_TYPE_AAAA + +#define CLASS_INET EVDNS_CLASS_INET + +struct request { + u8 *request; /* the dns packet data */ + unsigned int request_len; + int reissue_count; + int tx_count; /* the number of times that this packet has been sent */ + unsigned int request_type; /* TYPE_PTR or TYPE_A */ + void *user_pointer; /* the pointer given to us for this request */ + evdns_callback_type user_callback; + struct nameserver *ns; /* the server which we last sent it */ + + /* elements used by the searching code */ + int search_index; + struct search_state *search_state; + char *search_origname; /* needs to be free()ed */ + int search_flags; + + /* these objects are kept in a circular list */ + struct request *next, *prev; + + struct event timeout_event; + + u16 trans_id; /* the transaction id */ + char request_appended; /* true if the request pointer is data which follows this struct */ + char transmit_me; /* needs to be transmitted */ +}; + +#ifndef HAVE_STRUCT_IN6_ADDR +struct in6_addr { + u8 s6_addr[16]; +}; +#endif + +struct reply { + unsigned int type; + unsigned int have_answer; + union { + struct { + u32 addrcount; + u32 addresses[MAX_ADDRS]; + } a; + struct { + u32 addrcount; + struct in6_addr addresses[MAX_ADDRS]; + } aaaa; + struct { + char name[HOST_NAME_MAX]; + } ptr; + } data; +}; + +struct nameserver { + int socket; /* a connected UDP socket */ + u32 address; + int failed_times; /* number of times which we have given this server a chance */ + int timedout; /* number of times in a row a request has timed out */ + struct event event; + /* these objects are kept in a circular list */ + struct nameserver *next, *prev; + struct event timeout_event; /* used to keep the timeout for */ + /* when we next probe this server. */ + /* Valid if state == 0 */ + char state; /* zero if we think that this server is down */ + char choked; /* true if we have an EAGAIN from this server's socket */ + char write_waiting; /* true if we are waiting for EV_WRITE events */ +}; + +static struct request *req_head = NULL, *req_waiting_head = NULL; +static struct nameserver *server_head = NULL; + +/* Represents a local port where we're listening for DNS requests. Right now, */ +/* only UDP is supported. */ +struct evdns_server_port { + int socket; /* socket we use to read queries and write replies. */ + int refcnt; /* reference count. */ + char choked; /* Are we currently blocked from writing? */ + char closing; /* Are we trying to close this port, pending writes? */ + evdns_request_callback_fn_type user_callback; /* Fn to handle requests */ + void *user_data; /* Opaque pointer passed to user_callback */ + struct event event; /* Read/write event */ + /* circular list of replies that we want to write. */ + struct server_request *pending_replies; +}; + +/* Represents part of a reply being built. (That is, a single RR.) */ +struct server_reply_item { + struct server_reply_item *next; /* next item in sequence. */ + char *name; /* name part of the RR */ + u16 type : 16; /* The RR type */ + u16 class : 16; /* The RR class (usually CLASS_INET) */ + u32 ttl; /* The RR TTL */ + char is_name; /* True iff data is a label */ + u16 datalen; /* Length of data; -1 if data is a label */ + void *data; /* The contents of the RR */ +}; + +/* Represents a request that we've received as a DNS server, and holds */ +/* the components of the reply as we're constructing it. */ +struct server_request { + /* Pointers to the next and previous entries on the list of replies */ + /* that we're waiting to write. Only set if we have tried to respond */ + /* and gotten EAGAIN. */ + struct server_request *next_pending; + struct server_request *prev_pending; + + u16 trans_id; /* Transaction id. */ + struct evdns_server_port *port; /* Which port received this request on? */ + struct sockaddr_storage addr; /* Where to send the response */ + socklen_t addrlen; /* length of addr */ + + int n_answer; /* how many answer RRs have been set? */ + int n_authority; /* how many authority RRs have been set? */ + int n_additional; /* how many additional RRs have been set? */ + + struct server_reply_item *answer; /* linked list of answer RRs */ + struct server_reply_item *authority; /* linked list of authority RRs */ + struct server_reply_item *additional; /* linked list of additional RRs */ + + /* Constructed response. Only set once we're ready to send a reply. */ + /* Once this is set, the RR fields are cleared, and no more should be set. */ + char *response; + size_t response_len; + + /* Caller-visible fields: flags, questions. */ + struct evdns_server_request base; +}; + +/* helper macro */ +#define OFFSET_OF(st, member) ((off_t) (((char*)&((st*)0)->member)-(char*)0)) + +/* Given a pointer to an evdns_server_request, get the corresponding */ +/* server_request. */ +#define TO_SERVER_REQUEST(base_ptr) \ + ((struct server_request*) \ + (((char*)(base_ptr) - OFFSET_OF(struct server_request, base)))) + +/* The number of good nameservers that we have */ +static int global_good_nameservers = 0; + +/* inflight requests are contained in the req_head list */ +/* and are actually going out across the network */ +static int global_requests_inflight = 0; +/* requests which aren't inflight are in the waiting list */ +/* and are counted here */ +static int global_requests_waiting = 0; + +static int global_max_requests_inflight = 64; + +static struct timeval global_timeout = {5, 0}; /* 5 seconds */ +static int global_max_reissues = 1; /* a reissue occurs when we get some errors from the server */ +static int global_max_retransmits = 3; /* number of times we'll retransmit a request which timed out */ +/* number of timeouts in a row before we consider this server to be down */ +static int global_max_nameserver_timeout = 3; + +/* These are the timeout values for nameservers. If we find a nameserver is down */ +/* we try to probe it at intervals as given below. Values are in seconds. */ +static const struct timeval global_nameserver_timeouts[] = {{10, 0}, {60, 0}, {300, 0}, {900, 0}, {3600, 0}}; +static const int global_nameserver_timeouts_length = sizeof(global_nameserver_timeouts)/sizeof(struct timeval); + +static struct nameserver *nameserver_pick(void); +static void evdns_request_insert(struct request *req, struct request **head); +static void nameserver_ready_callback(int fd, short events, void *arg); +static int evdns_transmit(void); +static int evdns_request_transmit(struct request *req); +static void nameserver_send_probe(struct nameserver *const ns); +static void search_request_finished(struct request *const); +static int search_try_next(struct request *const req); +static int search_request_new(int type, const char *const name, int flags, evdns_callback_type user_callback, void *user_arg); +static void evdns_requests_pump_waiting_queue(void); +static u16 transaction_id_pick(void); +static struct request *request_new(int type, const char *name, int flags, evdns_callback_type callback, void *ptr); +static void request_submit(struct request *req); + +static int server_request_free(struct server_request *req); +static void server_request_free_answers(struct server_request *req); +static void server_port_free(struct evdns_server_port *port); +static void server_port_ready_callback(int fd, short events, void *arg); + +static int strtoint(const char *const str); + +#ifdef WIN32 +static int +last_error(int sock) +{ + int optval, optvallen=sizeof(optval); + int err = WSAGetLastError(); + if (err == WSAEWOULDBLOCK && sock >= 0) { + if (getsockopt(sock, SOL_SOCKET, SO_ERROR, (void*)&optval, + &optvallen)) + return err; + if (optval) + return optval; + } + return err; + +} +static int +error_is_eagain(int err) +{ + return err == EAGAIN || err == WSAEWOULDBLOCK; +} +static int +inet_aton(const char *c, struct in_addr *addr) +{ + uint32_t r; + if (strcmp(c, "255.255.255.255") == 0) { + addr->s_addr = 0xffffffffu; + } else { + r = inet_addr(c); + if (r == INADDR_NONE) + return 0; + addr->s_addr = r; + } + return 1; +} +#define CLOSE_SOCKET(x) closesocket(x) +#else +#define last_error(sock) (errno) +#define error_is_eagain(err) ((err) == EAGAIN) +#define CLOSE_SOCKET(x) close(x) +#endif + +#define ISSPACE(c) isspace((int)(unsigned char)(c)) +#define ISDIGIT(c) isdigit((int)(unsigned char)(c)) + +#ifndef NDEBUG +static const char * +debug_ntoa(u32 address) +{ + static char buf[32]; + u32 a = ntohl(address); + snprintf(buf, sizeof(buf), "%d.%d.%d.%d", + (int)(u8)((a>>24)&0xff), + (int)(u8)((a>>16)&0xff), + (int)(u8)((a>>8 )&0xff), + (int)(u8)((a )&0xff)); + return buf; +} +#endif + +static evdns_debug_log_fn_type evdns_log_fn = NULL; + +void +evdns_set_log_fn(evdns_debug_log_fn_type fn) +{ + evdns_log_fn = fn; +} + +#ifdef __GNUC__ +#define EVDNS_LOG_CHECK __attribute__ ((format(printf, 2, 3))) +#else +#define EVDNS_LOG_CHECK +#endif + +static void _evdns_log(int warn, const char *fmt, ...) EVDNS_LOG_CHECK; +static void +_evdns_log(int warn, const char *fmt, ...) +{ + va_list args; + static char buf[512]; + if (!evdns_log_fn) + return; + va_start(args,fmt); +#ifdef WIN32 + _vsnprintf(buf, sizeof(buf), fmt, args); +#else + vsnprintf(buf, sizeof(buf), fmt, args); +#endif + buf[sizeof(buf)-1] = '\0'; + evdns_log_fn(warn, buf); + va_end(args); +} + +#define log _evdns_log + +/* This walks the list of inflight requests to find the */ +/* one with a matching transaction id. Returns NULL on */ +/* failure */ +static struct request * +request_find_from_trans_id(u16 trans_id) { + struct request *req = req_head, *const started_at = req_head; + + if (req) { + do { + if (req->trans_id == trans_id) return req; + req = req->next; + } while (req != started_at); + } + + return NULL; +} + +/* a libevent callback function which is called when a nameserver */ +/* has gone down and we want to test if it has came back to life yet */ +static void +nameserver_prod_callback(int fd, short events, void *arg) { + struct nameserver *const ns = (struct nameserver *) arg; + (void)fd; + (void)events; + + nameserver_send_probe(ns); +} + +/* a libevent callback which is called when a nameserver probe (to see if */ +/* it has come back to life) times out. We increment the count of failed_times */ +/* and wait longer to send the next probe packet. */ +static void +nameserver_probe_failed(struct nameserver *const ns) { + const struct timeval * timeout; + (void) evtimer_del(&ns->timeout_event); + if (ns->state == 1) { + /* This can happen if the nameserver acts in a way which makes us mark */ + /* it as bad and then starts sending good replies. */ + return; + } + + timeout = + &global_nameserver_timeouts[MIN(ns->failed_times, + global_nameserver_timeouts_length - 1)]; + ns->failed_times++; + + evtimer_set(&ns->timeout_event, nameserver_prod_callback, ns); + if (evtimer_add(&ns->timeout_event, (struct timeval *) timeout) < 0) { + log(EVDNS_LOG_WARN, + "Error from libevent when adding timer event for %s", + debug_ntoa(ns->address)); + /* ???? Do more? */ + } +} + +/* called when a nameserver has been deemed to have failed. For example, too */ +/* many packets have timed out etc */ +static void +nameserver_failed(struct nameserver *const ns, const char *msg) { + struct request *req, *started_at; + /* if this nameserver has already been marked as failed */ + /* then don't do anything */ + if (!ns->state) return; + + log(EVDNS_LOG_WARN, "Nameserver %s has failed: %s", + debug_ntoa(ns->address), msg); + global_good_nameservers--; + assert(global_good_nameservers >= 0); + if (global_good_nameservers == 0) { + log(EVDNS_LOG_WARN, "All nameservers have failed"); + } + + ns->state = 0; + ns->failed_times = 1; + + evtimer_set(&ns->timeout_event, nameserver_prod_callback, ns); + if (evtimer_add(&ns->timeout_event, (struct timeval *) &global_nameserver_timeouts[0]) < 0) { + log(EVDNS_LOG_WARN, + "Error from libevent when adding timer event for %s", + debug_ntoa(ns->address)); + /* ???? Do more? */ + } + + /* walk the list of inflight requests to see if any can be reassigned to */ + /* a different server. Requests in the waiting queue don't have a */ + /* nameserver assigned yet */ + + /* if we don't have *any* good nameservers then there's no point */ + /* trying to reassign requests to one */ + if (!global_good_nameservers) return; + + req = req_head; + started_at = req_head; + if (req) { + do { + if (req->tx_count == 0 && req->ns == ns) { + /* still waiting to go out, can be moved */ + /* to another server */ + req->ns = nameserver_pick(); + } + req = req->next; + } while (req != started_at); + } +} + +static void +nameserver_up(struct nameserver *const ns) { + if (ns->state) return; + log(EVDNS_LOG_WARN, "Nameserver %s is back up", + debug_ntoa(ns->address)); + evtimer_del(&ns->timeout_event); + ns->state = 1; + ns->failed_times = 0; + ns->timedout = 0; + global_good_nameservers++; +} + +static void +request_trans_id_set(struct request *const req, const u16 trans_id) { + req->trans_id = trans_id; + *((u16 *) req->request) = htons(trans_id); +} + +/* Called to remove a request from a list and dealloc it. */ +/* head is a pointer to the head of the list it should be */ +/* removed from or NULL if the request isn't in a list. */ +static void +request_finished(struct request *const req, struct request **head) { + if (head) { + if (req->next == req) { + /* only item in the list */ + *head = NULL; + } else { + req->next->prev = req->prev; + req->prev->next = req->next; + if (*head == req) *head = req->next; + } + } + + log(EVDNS_LOG_DEBUG, "Removing timeout for request %lx", + (unsigned long) req); + evtimer_del(&req->timeout_event); + + search_request_finished(req); + global_requests_inflight--; + + if (!req->request_appended) { + /* need to free the request data on it's own */ + free(req->request); + } else { + /* the request data is appended onto the header */ + /* so everything gets free()ed when we: */ + } + + free(req); + + evdns_requests_pump_waiting_queue(); +} + +/* This is called when a server returns a funny error code. */ +/* We try the request again with another server. */ +/* */ +/* return: */ +/* 0 ok */ +/* 1 failed/reissue is pointless */ +static int +request_reissue(struct request *req) { + const struct nameserver *const last_ns = req->ns; + /* the last nameserver should have been marked as failing */ + /* by the caller of this function, therefore pick will try */ + /* not to return it */ + req->ns = nameserver_pick(); + if (req->ns == last_ns) { + /* ... but pick did return it */ + /* not a lot of point in trying again with the */ + /* same server */ + return 1; + } + + req->reissue_count++; + req->tx_count = 0; + req->transmit_me = 1; + + return 0; +} + +/* this function looks for space on the inflight queue and promotes */ +/* requests from the waiting queue if it can. */ +static void +evdns_requests_pump_waiting_queue(void) { + while (global_requests_inflight < global_max_requests_inflight && + global_requests_waiting) { + struct request *req; + /* move a request from the waiting queue to the inflight queue */ + assert(req_waiting_head); + if (req_waiting_head->next == req_waiting_head) { + /* only one item in the queue */ + req = req_waiting_head; + req_waiting_head = NULL; + } else { + req = req_waiting_head; + req->next->prev = req->prev; + req->prev->next = req->next; + req_waiting_head = req->next; + } + + global_requests_waiting--; + global_requests_inflight++; + + req->ns = nameserver_pick(); + request_trans_id_set(req, transaction_id_pick()); + + evdns_request_insert(req, &req_head); + evdns_request_transmit(req); + evdns_transmit(); + } +} + +static void +reply_callback(struct request *const req, u32 ttl, u32 err, struct reply *reply) { + switch (req->request_type) { + case TYPE_A: + if (reply) + req->user_callback(DNS_ERR_NONE, DNS_IPv4_A, + reply->data.a.addrcount, ttl, + reply->data.a.addresses, + req->user_pointer); + else + req->user_callback(err, 0, 0, 0, NULL, req->user_pointer); + return; + case TYPE_PTR: + if (reply) { + char *name = reply->data.ptr.name; + req->user_callback(DNS_ERR_NONE, DNS_PTR, 1, ttl, + &name, req->user_pointer); + } else { + req->user_callback(err, 0, 0, 0, NULL, + req->user_pointer); + } + return; + case TYPE_AAAA: + if (reply) + req->user_callback(DNS_ERR_NONE, DNS_IPv6_AAAA, + reply->data.aaaa.addrcount, ttl, + reply->data.aaaa.addresses, + req->user_pointer); + else + req->user_callback(err, 0, 0, 0, NULL, req->user_pointer); + return; + } + assert(0); +} + +/* this processes a parsed reply packet */ +static void +reply_handle(struct request *const req, u16 flags, u32 ttl, struct reply *reply) { + int error; + static const int error_codes[] = {DNS_ERR_FORMAT, DNS_ERR_SERVERFAILED, DNS_ERR_NOTEXIST, DNS_ERR_NOTIMPL, DNS_ERR_REFUSED}; + + if (flags & 0x020f || !reply || !reply->have_answer) { + /* there was an error */ + if (flags & 0x0200) { + error = DNS_ERR_TRUNCATED; + } else { + u16 error_code = (flags & 0x000f) - 1; + if (error_code > 4) { + error = DNS_ERR_UNKNOWN; + } else { + error = error_codes[error_code]; + } + } + + switch(error) { + case DNS_ERR_NOTIMPL: + case DNS_ERR_REFUSED: + /* we regard these errors as marking a bad nameserver */ + if (req->reissue_count < global_max_reissues) { + char msg[64]; + snprintf(msg, sizeof(msg), "Bad response %d (%s)", + error, evdns_err_to_string(error)); + nameserver_failed(req->ns, msg); + if (!request_reissue(req)) return; + } + break; + case DNS_ERR_SERVERFAILED: + /* rcode 2 (servfailed) sometimes means "we are broken" and + * sometimes (with some binds) means "that request was very + * confusing." Treat this as a timeout, not a failure. + */ + log(EVDNS_LOG_DEBUG, "Got a SERVERFAILED from nameserver %s; " + "will allow the request to time out.", + debug_ntoa(req->ns->address)); + break; + default: + /* we got a good reply from the nameserver */ + nameserver_up(req->ns); + } + + if (req->search_state && req->request_type != TYPE_PTR) { + /* if we have a list of domains to search in, try the next one */ + if (!search_try_next(req)) { + /* a new request was issued so this request is finished and */ + /* the user callback will be made when that request (or a */ + /* child of it) finishes. */ + request_finished(req, &req_head); + return; + } + } + + /* all else failed. Pass the failure up */ + reply_callback(req, 0, error, NULL); + request_finished(req, &req_head); + } else { + /* all ok, tell the user */ + reply_callback(req, ttl, 0, reply); + nameserver_up(req->ns); + request_finished(req, &req_head); + } +} + +static int +name_parse(u8 *packet, int length, int *idx, char *name_out, int name_out_len) { + int name_end = -1; + int j = *idx; + int ptr_count = 0; +#define GET32(x) do { if (j + 4 > length) goto err; memcpy(&_t32, packet + j, 4); j += 4; x = ntohl(_t32); } while(0) +#define GET16(x) do { if (j + 2 > length) goto err; memcpy(&_t, packet + j, 2); j += 2; x = ntohs(_t); } while(0) +#define GET8(x) do { if (j >= length) goto err; x = packet[j++]; } while(0) + + char *cp = name_out; + const char *const end = name_out + name_out_len; + + /* Normally, names are a series of length prefixed strings terminated */ + /* with a length of 0 (the lengths are u8's < 63). */ + /* However, the length can start with a pair of 1 bits and that */ + /* means that the next 14 bits are a pointer within the current */ + /* packet. */ + + for(;;) { + u8 label_len; + if (j >= length) return -1; + GET8(label_len); + if (!label_len) break; + if (label_len & 0xc0) { + u8 ptr_low; + GET8(ptr_low); + if (name_end < 0) name_end = j; + j = (((int)label_len & 0x3f) << 8) + ptr_low; + /* Make sure that the target offset is in-bounds. */ + if (j < 0 || j >= length) return -1; + /* If we've jumped more times than there are characters in the + * message, we must have a loop. */ + if (++ptr_count > length) return -1; + continue; + } + if (label_len > 63) return -1; + if (cp != name_out) { + if (cp + 1 >= end) return -1; + *cp++ = '.'; + } + if (cp + label_len >= end) return -1; + memcpy(cp, packet + j, label_len); + cp += label_len; + j += label_len; + } + if (cp >= end) return -1; + *cp = '\0'; + if (name_end < 0) + *idx = j; + else + *idx = name_end; + return 0; + err: + return -1; +} + +/* parses a raw request from a nameserver */ +static int +reply_parse(u8 *packet, int length) { + int j = 0; /* index into packet */ + u16 _t; /* used by the macros */ + u32 _t32; /* used by the macros */ + char tmp_name[256]; /* used by the macros */ + + u16 trans_id, questions, answers, authority, additional, datalength; + u16 flags = 0; + u32 ttl, ttl_r = 0xffffffff; + struct reply reply; + struct request *req = NULL; + unsigned int i; + + GET16(trans_id); + GET16(flags); + GET16(questions); + GET16(answers); + GET16(authority); + GET16(additional); + (void) authority; /* suppress "unused variable" warnings. */ + (void) additional; /* suppress "unused variable" warnings. */ + + req = request_find_from_trans_id(trans_id); + if (!req) return -1; + + memset(&reply, 0, sizeof(reply)); + + /* If it's not an answer, it doesn't correspond to any request. */ + if (!(flags & 0x8000)) return -1; /* must be an answer */ + if (flags & 0x020f) { + /* there was an error */ + goto err; + } + /* if (!answers) return; */ /* must have an answer of some form */ + + /* This macro skips a name in the DNS reply. */ +#define SKIP_NAME \ + do { tmp_name[0] = '\0'; \ + if (name_parse(packet, length, &j, tmp_name, sizeof(tmp_name))<0) \ + goto err; \ + } while(0); + + reply.type = req->request_type; + + /* skip over each question in the reply */ + for (i = 0; i < questions; ++i) { + /* the question looks like + * + */ + SKIP_NAME; + j += 4; + if (j >= length) goto err; + } + + /* now we have the answer section which looks like + * + */ + + for (i = 0; i < answers; ++i) { + u16 type, class; + + SKIP_NAME; + GET16(type); + GET16(class); + GET32(ttl); + GET16(datalength); + + if (type == TYPE_A && class == CLASS_INET) { + int addrcount, addrtocopy; + if (req->request_type != TYPE_A) { + j += datalength; continue; + } + if ((datalength & 3) != 0) /* not an even number of As. */ + goto err; + addrcount = datalength >> 2; + addrtocopy = MIN(MAX_ADDRS - reply.data.a.addrcount, (unsigned)addrcount); + + ttl_r = MIN(ttl_r, ttl); + /* we only bother with the first four addresses. */ + if (j + 4*addrtocopy > length) goto err; + memcpy(&reply.data.a.addresses[reply.data.a.addrcount], + packet + j, 4*addrtocopy); + j += 4*addrtocopy; + reply.data.a.addrcount += addrtocopy; + reply.have_answer = 1; + if (reply.data.a.addrcount == MAX_ADDRS) break; + } else if (type == TYPE_PTR && class == CLASS_INET) { + if (req->request_type != TYPE_PTR) { + j += datalength; continue; + } + if (name_parse(packet, length, &j, reply.data.ptr.name, + sizeof(reply.data.ptr.name))<0) + goto err; + ttl_r = MIN(ttl_r, ttl); + reply.have_answer = 1; + break; + } else if (type == TYPE_AAAA && class == CLASS_INET) { + int addrcount, addrtocopy; + if (req->request_type != TYPE_AAAA) { + j += datalength; continue; + } + if ((datalength & 15) != 0) /* not an even number of AAAAs. */ + goto err; + addrcount = datalength >> 4; /* each address is 16 bytes long */ + addrtocopy = MIN(MAX_ADDRS - reply.data.aaaa.addrcount, (unsigned)addrcount); + ttl_r = MIN(ttl_r, ttl); + + /* we only bother with the first four addresses. */ + if (j + 16*addrtocopy > length) goto err; + memcpy(&reply.data.aaaa.addresses[reply.data.aaaa.addrcount], + packet + j, 16*addrtocopy); + reply.data.aaaa.addrcount += addrtocopy; + j += 16*addrtocopy; + reply.have_answer = 1; + if (reply.data.aaaa.addrcount == MAX_ADDRS) break; + } else { + /* skip over any other type of resource */ + j += datalength; + } + } + + reply_handle(req, flags, ttl_r, &reply); + return 0; + err: + if (req) + reply_handle(req, flags, 0, NULL); + return -1; +} + +/* Parse a raw request (packet,length) sent to a nameserver port (port) from */ +/* a DNS client (addr,addrlen), and if it's well-formed, call the corresponding */ +/* callback. */ +static int +request_parse(u8 *packet, int length, struct evdns_server_port *port, struct sockaddr *addr, socklen_t addrlen) +{ + int j = 0; /* index into packet */ + u16 _t; /* used by the macros */ + char tmp_name[256]; /* used by the macros */ + + int i; + u16 trans_id, flags, questions, answers, authority, additional; + struct server_request *server_req = NULL; + + /* Get the header fields */ + GET16(trans_id); + GET16(flags); + GET16(questions); + GET16(answers); + GET16(authority); + GET16(additional); + + if (flags & 0x8000) return -1; /* Must not be an answer. */ + if (flags & 0x7800) return -1; /* only standard queries are supported */ + flags &= 0x0300; /* Only TC and RD get preserved. */ + + server_req = malloc(sizeof(struct server_request)); + if (server_req == NULL) return -1; + memset(server_req, 0, sizeof(struct server_request)); + + server_req->trans_id = trans_id; + memcpy(&server_req->addr, addr, addrlen); + server_req->addrlen = addrlen; + + server_req->base.flags = flags; + server_req->base.nquestions = 0; + server_req->base.questions = malloc(sizeof(struct evdns_server_question *) * questions); + if (server_req->base.questions == NULL) + goto err; + + for (i = 0; i < questions; ++i) { + u16 type, class; + struct evdns_server_question *q; + int namelen; + if (name_parse(packet, length, &j, tmp_name, sizeof(tmp_name))<0) + goto err; + GET16(type); + GET16(class); + namelen = strlen(tmp_name); + q = malloc(sizeof(struct evdns_server_question) + namelen); + if (!q) + goto err; + q->type = type; + q->class = class; + memcpy(q->name, tmp_name, namelen+1); + server_req->base.questions[server_req->base.nquestions++] = q; + } + + /* Ignore answers, authority, and additional. */ + + server_req->port = port; + port->refcnt++; + port->user_callback(&(server_req->base), port->user_data); + + return 0; +err: + if (server_req) { + if (server_req->base.questions) { + for (i = 0; i < server_req->base.nquestions; ++i) + free(server_req->base.questions[i]); + free(server_req->base.questions); + } + free(server_req); + } + return -1; + +#undef SKIP_NAME +#undef GET32 +#undef GET16 +#undef GET8 +} + +/* Try to choose a strong transaction id which isn't already in flight */ +static u16 +transaction_id_pick(void) { + for (;;) { + const struct request *req = req_head, *started_at; +#ifdef DNS_USE_CPU_CLOCK_FOR_ID + struct timespec ts; + u16 trans_id; +#ifdef CLOCK_MONOTONIC + if (clock_gettime(CLOCK_MONOTONIC, &ts) == -1) +#else + if (clock_gettime(CLOCK_REALTIME, &ts) == -1) +#endif + event_err(1, "clock_gettime"); + trans_id = ts.tv_nsec & 0xffff; +#endif + +#ifdef DNS_USE_GETTIMEOFDAY_FOR_ID + struct timeval tv; + u16 trans_id; + gettimeofday(&tv, NULL); + trans_id = tv.tv_usec & 0xffff; +#endif + +#ifdef DNS_USE_OPENSSL_FOR_ID + u16 trans_id; + if (RAND_pseudo_bytes((u8 *) &trans_id, 2) == -1) { + /* in the case that the RAND call fails we back */ + /* down to using gettimeofday. */ + struct timeval tv; + gettimeofday(&tv, NULL); + trans_id = tv.tv_usec & 0xffff; */ + abort(); + } +#endif + + if (trans_id == 0xffff) continue; + /* now check to see if that id is already inflight */ + req = started_at = req_head; + if (req) { + do { + if (req->trans_id == trans_id) break; + req = req->next; + } while (req != started_at); + } + /* we didn't find it, so this is a good id */ + if (req == started_at) return trans_id; + } +} + +/* choose a namesever to use. This function will try to ignore */ +/* nameservers which we think are down and load balance across the rest */ +/* by updating the server_head global each time. */ +static struct nameserver * +nameserver_pick(void) { + struct nameserver *started_at = server_head, *picked; + if (!server_head) return NULL; + + /* if we don't have any good nameservers then there's no */ + /* point in trying to find one. */ + if (!global_good_nameservers) { + server_head = server_head->next; + return server_head; + } + + /* remember that nameservers are in a circular list */ + for (;;) { + if (server_head->state) { + /* we think this server is currently good */ + picked = server_head; + server_head = server_head->next; + return picked; + } + + server_head = server_head->next; + if (server_head == started_at) { + /* all the nameservers seem to be down */ + /* so we just return this one and hope for the */ + /* best */ + assert(global_good_nameservers == 0); + picked = server_head; + server_head = server_head->next; + return picked; + } + } +} + +/* this is called when a namesever socket is ready for reading */ +static void +nameserver_read(struct nameserver *ns) { + u8 packet[1500]; + + for (;;) { + const int r = recv(ns->socket, packet, sizeof(packet), 0); + if (r < 0) { + int err = last_error(ns->socket); + if (error_is_eagain(err)) return; + nameserver_failed(ns, strerror(err)); + return; + } + ns->timedout = 0; + reply_parse(packet, r); + } +} + +/* Read a packet from a DNS client on a server port s, parse it, and */ +/* act accordingly. */ +static void +server_port_read(struct evdns_server_port *s) { + u8 packet[1500]; + struct sockaddr_storage addr; + socklen_t addrlen; + int r; + + for (;;) { + addrlen = sizeof(struct sockaddr_storage); + r = recvfrom(s->socket, packet, sizeof(packet), 0, + (struct sockaddr*) &addr, &addrlen); + if (r < 0) { + int err = last_error(s->socket); + if (error_is_eagain(err)) return; + log(EVDNS_LOG_WARN, "Error %s (%d) while reading request.", + strerror(err), err); + return; + } + request_parse(packet, r, s, (struct sockaddr*) &addr, addrlen); + } +} + +/* Try to write all pending replies on a given DNS server port. */ +static void +server_port_flush(struct evdns_server_port *port) +{ + while (port->pending_replies) { + struct server_request *req = port->pending_replies; + int r = sendto(port->socket, req->response, req->response_len, 0, + (struct sockaddr*) &req->addr, req->addrlen); + if (r < 0) { + int err = last_error(port->socket); + if (error_is_eagain(err)) + return; + log(EVDNS_LOG_WARN, "Error %s (%d) while writing response to port; dropping", strerror(err), err); + } + if (server_request_free(req)) { + /* we released the last reference to req->port. */ + return; + } + } + + /* We have no more pending requests; stop listening for 'writeable' events. */ + (void) event_del(&port->event); + event_set(&port->event, port->socket, EV_READ | EV_PERSIST, + server_port_ready_callback, port); + if (event_add(&port->event, NULL) < 0) { + log(EVDNS_LOG_WARN, "Error from libevent when adding event for DNS server."); + /* ???? Do more? */ + } +} + +/* set if we are waiting for the ability to write to this server. */ +/* if waiting is true then we ask libevent for EV_WRITE events, otherwise */ +/* we stop these events. */ +static void +nameserver_write_waiting(struct nameserver *ns, char waiting) { + if (ns->write_waiting == waiting) return; + + ns->write_waiting = waiting; + (void) event_del(&ns->event); + event_set(&ns->event, ns->socket, EV_READ | (waiting ? EV_WRITE : 0) | EV_PERSIST, + nameserver_ready_callback, ns); + if (event_add(&ns->event, NULL) < 0) { + log(EVDNS_LOG_WARN, "Error from libevent when adding event for %s", + debug_ntoa(ns->address)); + /* ???? Do more? */ + } +} + +/* a callback function. Called by libevent when the kernel says that */ +/* a nameserver socket is ready for writing or reading */ +static void +nameserver_ready_callback(int fd, short events, void *arg) { + struct nameserver *ns = (struct nameserver *) arg; + (void)fd; + + if (events & EV_WRITE) { + ns->choked = 0; + if (!evdns_transmit()) { + nameserver_write_waiting(ns, 0); + } + } + if (events & EV_READ) { + nameserver_read(ns); + } +} + +/* a callback function. Called by libevent when the kernel says that */ +/* a server socket is ready for writing or reading. */ +static void +server_port_ready_callback(int fd, short events, void *arg) { + struct evdns_server_port *port = (struct evdns_server_port *) arg; + (void) fd; + + if (events & EV_WRITE) { + port->choked = 0; + server_port_flush(port); + } + if (events & EV_READ) { + server_port_read(port); + } +} + +/* This is an inefficient representation; only use it via the dnslabel_table_* + * functions, so that is can be safely replaced with something smarter later. */ +#define MAX_LABELS 128 +/* Structures used to implement name compression */ +struct dnslabel_entry { char *v; off_t pos; }; +struct dnslabel_table { + int n_labels; /* number of current entries */ + /* map from name to position in message */ + struct dnslabel_entry labels[MAX_LABELS]; +}; + +/* Initialize dnslabel_table. */ +static void +dnslabel_table_init(struct dnslabel_table *table) +{ + table->n_labels = 0; +} + +/* Free all storage held by table, but not the table itself. */ +static void +dnslabel_clear(struct dnslabel_table *table) +{ + int i; + for (i = 0; i < table->n_labels; ++i) + free(table->labels[i].v); + table->n_labels = 0; +} + +/* return the position of the label in the current message, or -1 if the label */ +/* hasn't been used yet. */ +static int +dnslabel_table_get_pos(const struct dnslabel_table *table, const char *label) +{ + int i; + for (i = 0; i < table->n_labels; ++i) { + if (!strcmp(label, table->labels[i].v)) + return table->labels[i].pos; + } + return -1; +} + +/* remember that we've used the label at position pos */ +static int +dnslabel_table_add(struct dnslabel_table *table, const char *label, off_t pos) +{ + char *v; + int p; + if (table->n_labels == MAX_LABELS) + return (-1); + v = strdup(label); + if (v == NULL) + return (-1); + p = table->n_labels++; + table->labels[p].v = v; + table->labels[p].pos = pos; + + return (0); +} + +/* Converts a string to a length-prefixed set of DNS labels, starting */ +/* at buf[j]. name and buf must not overlap. name_len should be the length */ +/* of name. table is optional, and is used for compression. */ +/* */ +/* Input: abc.def */ +/* Output: <3>abc<3>def<0> */ +/* */ +/* Returns the first index after the encoded name, or negative on error. */ +/* -1 label was > 63 bytes */ +/* -2 name too long to fit in buffer. */ +/* */ +static off_t +dnsname_to_labels(u8 *const buf, size_t buf_len, off_t j, + const char *name, const int name_len, + struct dnslabel_table *table) { + const char *end = name + name_len; + int ref = 0; + u16 _t; + +#define APPEND16(x) do { \ + if (j + 2 > (off_t)buf_len) \ + goto overflow; \ + _t = htons(x); \ + memcpy(buf + j, &_t, 2); \ + j += 2; \ + } while (0) +#define APPEND32(x) do { \ + if (j + 4 > (off_t)buf_len) \ + goto overflow; \ + _t32 = htonl(x); \ + memcpy(buf + j, &_t32, 4); \ + j += 4; \ + } while (0) + + if (name_len > 255) return -2; + + for (;;) { + const char *const start = name; + if (table && (ref = dnslabel_table_get_pos(table, name)) >= 0) { + APPEND16(ref | 0xc000); + return j; + } + name = strchr(name, '.'); + if (!name) { + const unsigned int label_len = end - start; + if (label_len > 63) return -1; + if ((size_t)(j+label_len+1) > buf_len) return -2; + if (table) dnslabel_table_add(table, start, j); + buf[j++] = label_len; + + memcpy(buf + j, start, end - start); + j += end - start; + break; + } else { + /* append length of the label. */ + const unsigned int label_len = name - start; + if (label_len > 63) return -1; + if ((size_t)(j+label_len+1) > buf_len) return -2; + if (table) dnslabel_table_add(table, start, j); + buf[j++] = label_len; + + memcpy(buf + j, start, name - start); + j += name - start; + /* hop over the '.' */ + name++; + } + } + + /* the labels must be terminated by a 0. */ + /* It's possible that the name ended in a . */ + /* in which case the zero is already there */ + if (!j || buf[j-1]) buf[j++] = 0; + return j; + overflow: + return (-2); +} + +/* Finds the length of a dns request for a DNS name of the given */ +/* length. The actual request may be smaller than the value returned */ +/* here */ +static int +evdns_request_len(const int name_len) { + return 96 + /* length of the DNS standard header */ + name_len + 2 + + 4; /* space for the resource type */ +} + +/* build a dns request packet into buf. buf should be at least as long */ +/* as evdns_request_len told you it should be. */ +/* */ +/* Returns the amount of space used. Negative on error. */ +static int +evdns_request_data_build(const char *const name, const int name_len, + const u16 trans_id, const u16 type, const u16 class, + u8 *const buf, size_t buf_len) { + off_t j = 0; /* current offset into buf */ + u16 _t; /* used by the macros */ + + APPEND16(trans_id); + APPEND16(0x0100); /* standard query, recusion needed */ + APPEND16(1); /* one question */ + APPEND16(0); /* no answers */ + APPEND16(0); /* no authority */ + APPEND16(0); /* no additional */ + + j = dnsname_to_labels(buf, buf_len, j, name, name_len, NULL); + if (j < 0) { + return (int)j; + } + + APPEND16(type); + APPEND16(class); + + return (int)j; + overflow: + return (-1); +} + +/* exported function */ +struct evdns_server_port * +evdns_add_server_port(int socket, int is_tcp, evdns_request_callback_fn_type cb, void *user_data) +{ + struct evdns_server_port *port; + if (!(port = malloc(sizeof(struct evdns_server_port)))) + return NULL; + memset(port, 0, sizeof(struct evdns_server_port)); + + assert(!is_tcp); /* TCP sockets not yet implemented */ + port->socket = socket; + port->refcnt = 1; + port->choked = 0; + port->closing = 0; + port->user_callback = cb; + port->user_data = user_data; + port->pending_replies = NULL; + + event_set(&port->event, port->socket, EV_READ | EV_PERSIST, + server_port_ready_callback, port); + event_add(&port->event, NULL); /* check return. */ + return port; +} + +/* exported function */ +void +evdns_close_server_port(struct evdns_server_port *port) +{ + if (--port->refcnt == 0) + server_port_free(port); + port->closing = 1; +} + +/* exported function */ +int +evdns_server_request_add_reply(struct evdns_server_request *_req, int section, const char *name, int type, int class, int ttl, int datalen, int is_name, const char *data) +{ + struct server_request *req = TO_SERVER_REQUEST(_req); + struct server_reply_item **itemp, *item; + int *countp; + + if (req->response) /* have we already answered? */ + return (-1); + + switch (section) { + case EVDNS_ANSWER_SECTION: + itemp = &req->answer; + countp = &req->n_answer; + break; + case EVDNS_AUTHORITY_SECTION: + itemp = &req->authority; + countp = &req->n_authority; + break; + case EVDNS_ADDITIONAL_SECTION: + itemp = &req->additional; + countp = &req->n_additional; + break; + default: + return (-1); + } + while (*itemp) { + itemp = &((*itemp)->next); + } + item = malloc(sizeof(struct server_reply_item)); + if (!item) + return -1; + item->next = NULL; + if (!(item->name = strdup(name))) { + free(item); + return -1; + } + item->type = type; + item->class = class; + item->ttl = ttl; + item->is_name = is_name != 0; + item->datalen = 0; + item->data = NULL; + if (data) { + if (item->is_name) { + if (!(item->data = strdup(data))) { + free(item->name); + free(item); + return -1; + } + item->datalen = (u16)-1; + } else { + if (!(item->data = malloc(datalen))) { + free(item->name); + free(item); + return -1; + } + item->datalen = datalen; + memcpy(item->data, data, datalen); + } + } + + *itemp = item; + ++(*countp); + return 0; +} + +/* exported function */ +int +evdns_server_request_add_a_reply(struct evdns_server_request *req, const char *name, int n, void *addrs, int ttl) +{ + return evdns_server_request_add_reply( + req, EVDNS_ANSWER_SECTION, name, TYPE_A, CLASS_INET, + ttl, n*4, 0, addrs); +} + +/* exported function */ +int +evdns_server_request_add_aaaa_reply(struct evdns_server_request *req, const char *name, int n, void *addrs, int ttl) +{ + return evdns_server_request_add_reply( + req, EVDNS_ANSWER_SECTION, name, TYPE_AAAA, CLASS_INET, + ttl, n*16, 0, addrs); +} + +/* exported function */ +int +evdns_server_request_add_ptr_reply(struct evdns_server_request *req, struct in_addr *in, const char *inaddr_name, const char *hostname, int ttl) +{ + u32 a; + char buf[32]; + assert(in || inaddr_name); + assert(!(in && inaddr_name)); + if (in) { + a = ntohl(in->s_addr); + snprintf(buf, sizeof(buf), "%d.%d.%d.%d.in-addr.arpa", + (int)(u8)((a )&0xff), + (int)(u8)((a>>8 )&0xff), + (int)(u8)((a>>16)&0xff), + (int)(u8)((a>>24)&0xff)); + inaddr_name = buf; + } + return evdns_server_request_add_reply( + req, EVDNS_ANSWER_SECTION, inaddr_name, TYPE_PTR, CLASS_INET, + ttl, -1, 1, hostname); +} + +/* exported function */ +int +evdns_server_request_add_cname_reply(struct evdns_server_request *req, const char *name, const char *cname, int ttl) +{ + return evdns_server_request_add_reply( + req, EVDNS_ANSWER_SECTION, name, TYPE_A, CLASS_INET, + ttl, -1, 1, cname); +} + + +static int +evdns_server_request_format_response(struct server_request *req, int err) +{ + unsigned char buf[1500]; + size_t buf_len = sizeof(buf); + off_t j = 0, r; + u16 _t; + u32 _t32; + int i; + u16 flags; + struct dnslabel_table table; + + if (err < 0 || err > 15) return -1; + + /* Set response bit and error code; copy OPCODE and RD fields from + * question; copy RA and AA if set by caller. */ + flags = req->base.flags; + flags |= (0x8000 | err); + + dnslabel_table_init(&table); + APPEND16(req->trans_id); + APPEND16(flags); + APPEND16(req->base.nquestions); + APPEND16(req->n_answer); + APPEND16(req->n_authority); + APPEND16(req->n_additional); + + /* Add questions. */ + for (i=0; i < req->base.nquestions; ++i) { + const char *s = req->base.questions[i]->name; + j = dnsname_to_labels(buf, buf_len, j, s, strlen(s), &table); + if (j < 0) { + dnslabel_clear(&table); + return (int) j; + } + APPEND16(req->base.questions[i]->type); + APPEND16(req->base.questions[i]->class); + } + + /* Add answer, authority, and additional sections. */ + for (i=0; i<3; ++i) { + struct server_reply_item *item; + if (i==0) + item = req->answer; + else if (i==1) + item = req->authority; + else + item = req->additional; + while (item) { + r = dnsname_to_labels(buf, buf_len, j, item->name, strlen(item->name), &table); + if (r < 0) + goto overflow; + j = r; + + APPEND16(item->type); + APPEND16(item->class); + APPEND32(item->ttl); + if (item->is_name) { + off_t len_idx = j, name_start; + j += 2; + name_start = j; + r = dnsname_to_labels(buf, buf_len, j, item->data, strlen(item->data), &table); + if (r < 0) + goto overflow; + j = r; + _t = htons( (j-name_start) ); + memcpy(buf+len_idx, &_t, 2); + } else { + APPEND16(item->datalen); + if (j+item->datalen > (off_t)buf_len) + goto overflow; + memcpy(buf+j, item->data, item->datalen); + j += item->datalen; + } + item = item->next; + } + } + + if (j > 512) { +overflow: + j = 512; + buf[3] |= 0x02; /* set the truncated bit. */ + } + + req->response_len = j; + + if (!(req->response = malloc(req->response_len))) { + server_request_free_answers(req); + dnslabel_clear(&table); + return (-1); + } + memcpy(req->response, buf, req->response_len); + server_request_free_answers(req); + dnslabel_clear(&table); + return (0); +} + +/* exported function */ +int +evdns_server_request_respond(struct evdns_server_request *_req, int err) +{ + struct server_request *req = TO_SERVER_REQUEST(_req); + struct evdns_server_port *port = req->port; + int r; + if (!req->response) { + if ((r = evdns_server_request_format_response(req, err))<0) + return r; + } + + r = sendto(port->socket, req->response, req->response_len, 0, + (struct sockaddr*) &req->addr, req->addrlen); + if (r<0) { + int err = last_error(port->socket); + if (! error_is_eagain(err)) + return -1; + + if (port->pending_replies) { + req->prev_pending = port->pending_replies->prev_pending; + req->next_pending = port->pending_replies; + req->prev_pending->next_pending = + req->next_pending->prev_pending = req; + } else { + req->prev_pending = req->next_pending = req; + port->pending_replies = req; + port->choked = 1; + + (void) event_del(&port->event); + event_set(&port->event, port->socket, (port->closing?0:EV_READ) | EV_WRITE | EV_PERSIST, server_port_ready_callback, port); + + if (event_add(&port->event, NULL) < 0) { + log(EVDNS_LOG_WARN, "Error from libevent when adding event for DNS server"); + } + + } + + return 1; + } + if (server_request_free(req)) + return 0; + + if (port->pending_replies) + server_port_flush(port); + + return 0; +} + +/* Free all storage held by RRs in req. */ +static void +server_request_free_answers(struct server_request *req) +{ + struct server_reply_item *victim, *next, **list; + int i; + for (i = 0; i < 3; ++i) { + if (i==0) + list = &req->answer; + else if (i==1) + list = &req->authority; + else + list = &req->additional; + + victim = *list; + while (victim) { + next = victim->next; + free(victim->name); + if (victim->data) + free(victim->data); + free(victim); + victim = next; + } + *list = NULL; + } +} + +/* Free all storage held by req, and remove links to it. */ +/* return true iff we just wound up freeing the server_port. */ +static int +server_request_free(struct server_request *req) +{ + int i, rc=1; + if (req->base.questions) { + for (i = 0; i < req->base.nquestions; ++i) + free(req->base.questions[i]); + free(req->base.questions); + } + + if (req->port) { + if (req->port->pending_replies == req) { + if (req->next_pending) + req->port->pending_replies = req->next_pending; + else + req->port->pending_replies = NULL; + } + rc = --req->port->refcnt; + } + + if (req->response) { + free(req->response); + } + + server_request_free_answers(req); + + if (req->next_pending && req->next_pending != req) { + req->next_pending->prev_pending = req->prev_pending; + req->prev_pending->next_pending = req->next_pending; + } + + if (rc == 0) { + server_port_free(req->port); + free(req); + return (1); + } + free(req); + return (0); +} + +/* Free all storage held by an evdns_server_port. Only called when */ +static void +server_port_free(struct evdns_server_port *port) +{ + assert(port); + assert(!port->refcnt); + assert(!port->pending_replies); + if (port->socket > 0) { + CLOSE_SOCKET(port->socket); + port->socket = -1; + } + (void) event_del(&port->event); + /* XXXX actually free the port? -NM */ +} + +/* exported function */ +int +evdns_server_request_drop(struct evdns_server_request *_req) +{ + struct server_request *req = TO_SERVER_REQUEST(_req); + server_request_free(req); + return 0; +} + +/* exported function */ +int +evdns_server_request_get_requesting_addr(struct evdns_server_request *_req, struct sockaddr *sa, int addr_len) +{ + struct server_request *req = TO_SERVER_REQUEST(_req); + if (addr_len < (int)req->addrlen) + return -1; + memcpy(sa, &(req->addr), req->addrlen); + return req->addrlen; +} + +#undef APPEND16 +#undef APPEND32 + +/* this is a libevent callback function which is called when a request */ +/* has timed out. */ +static void +evdns_request_timeout_callback(int fd, short events, void *arg) { + struct request *const req = (struct request *) arg; + (void) fd; + (void) events; + + log(EVDNS_LOG_DEBUG, "Request %lx timed out", (unsigned long) arg); + + req->ns->timedout++; + if (req->ns->timedout > global_max_nameserver_timeout) { + req->ns->timedout = 0; + nameserver_failed(req->ns, "request timed out."); + } + + (void) evtimer_del(&req->timeout_event); + if (req->tx_count >= global_max_retransmits) { + /* this request has failed */ + reply_callback(req, 0, DNS_ERR_TIMEOUT, NULL); + request_finished(req, &req_head); + } else { + /* retransmit it */ + evdns_request_transmit(req); + } +} + +/* try to send a request to a given server. */ +/* */ +/* return: */ +/* 0 ok */ +/* 1 temporary failure */ +/* 2 other failure */ +static int +evdns_request_transmit_to(struct request *req, struct nameserver *server) { + const int r = send(server->socket, req->request, req->request_len, 0); + if (r < 0) { + int err = last_error(server->socket); + if (error_is_eagain(err)) return 1; + nameserver_failed(req->ns, strerror(err)); + return 2; + } else if (r != (int)req->request_len) { + return 1; /* short write */ + } else { + return 0; + } +} + +/* try to send a request, updating the fields of the request */ +/* as needed */ +/* */ +/* return: */ +/* 0 ok */ +/* 1 failed */ +static int +evdns_request_transmit(struct request *req) { + int retcode = 0, r; + + /* if we fail to send this packet then this flag marks it */ + /* for evdns_transmit */ + req->transmit_me = 1; + if (req->trans_id == 0xffff) abort(); + + if (req->ns->choked) { + /* don't bother trying to write to a socket */ + /* which we have had EAGAIN from */ + return 1; + } + + r = evdns_request_transmit_to(req, req->ns); + switch (r) { + case 1: + /* temp failure */ + req->ns->choked = 1; + nameserver_write_waiting(req->ns, 1); + return 1; + case 2: + /* failed in some other way */ + retcode = 1; + /* fall through */ + default: + /* all ok */ + log(EVDNS_LOG_DEBUG, + "Setting timeout for request %lx", (unsigned long) req); + evtimer_set(&req->timeout_event, evdns_request_timeout_callback, req); + if (evtimer_add(&req->timeout_event, &global_timeout) < 0) { + log(EVDNS_LOG_WARN, + "Error from libevent when adding timer for request %lx", + (unsigned long) req); + /* ???? Do more? */ + } + req->tx_count++; + req->transmit_me = 0; + return retcode; + } +} + +static void +nameserver_probe_callback(int result, char type, int count, int ttl, void *addresses, void *arg) { + struct nameserver *const ns = (struct nameserver *) arg; + (void) type; + (void) count; + (void) ttl; + (void) addresses; + + if (result == DNS_ERR_NONE || result == DNS_ERR_NOTEXIST) { + /* this is a good reply */ + nameserver_up(ns); + } else nameserver_probe_failed(ns); +} + +static void +nameserver_send_probe(struct nameserver *const ns) { + struct request *req; + /* here we need to send a probe to a given nameserver */ + /* in the hope that it is up now. */ + + log(EVDNS_LOG_DEBUG, "Sending probe to %s", debug_ntoa(ns->address)); + + req = request_new(TYPE_A, "www.google.com", DNS_QUERY_NO_SEARCH, nameserver_probe_callback, ns); + if (!req) return; + /* we force this into the inflight queue no matter what */ + request_trans_id_set(req, transaction_id_pick()); + req->ns = ns; + request_submit(req); +} + +/* returns: */ +/* 0 didn't try to transmit anything */ +/* 1 tried to transmit something */ +static int +evdns_transmit(void) { + char did_try_to_transmit = 0; + + if (req_head) { + struct request *const started_at = req_head, *req = req_head; + /* first transmit all the requests which are currently waiting */ + do { + if (req->transmit_me) { + did_try_to_transmit = 1; + evdns_request_transmit(req); + } + + req = req->next; + } while (req != started_at); + } + + return did_try_to_transmit; +} + +/* exported function */ +int +evdns_count_nameservers(void) +{ + const struct nameserver *server = server_head; + int n = 0; + if (!server) + return 0; + do { + ++n; + server = server->next; + } while (server != server_head); + return n; +} + +/* exported function */ +int +evdns_clear_nameservers_and_suspend(void) +{ + struct nameserver *server = server_head, *started_at = server_head; + struct request *req = req_head, *req_started_at = req_head; + + if (!server) + return 0; + while (1) { + struct nameserver *next = server->next; + (void) event_del(&server->event); + (void) evtimer_del(&server->timeout_event); + if (server->socket >= 0) + CLOSE_SOCKET(server->socket); + free(server); + if (next == started_at) + break; + server = next; + } + server_head = NULL; + global_good_nameservers = 0; + + while (req) { + struct request *next = req->next; + req->tx_count = req->reissue_count = 0; + req->ns = NULL; + /* ???? What to do about searches? */ + (void) evtimer_del(&req->timeout_event); + req->trans_id = 0; + req->transmit_me = 0; + + global_requests_waiting++; + evdns_request_insert(req, &req_waiting_head); + /* We want to insert these suspended elements at the front of + * the waiting queue, since they were pending before any of + * the waiting entries were added. This is a circular list, + * so we can just shift the start back by one.*/ + req_waiting_head = req_waiting_head->prev; + + if (next == req_started_at) + break; + req = next; + } + req_head = NULL; + global_requests_inflight = 0; + + return 0; +} + + +/* exported function */ +int +evdns_resume(void) +{ + evdns_requests_pump_waiting_queue(); + return 0; +} + +static int +_evdns_nameserver_add_impl(unsigned long int address, int port) { + /* first check to see if we already have this nameserver */ + + const struct nameserver *server = server_head, *const started_at = server_head; + struct nameserver *ns; + struct sockaddr_in sin; + int err = 0; + if (server) { + do { + if (server->address == address) return 3; + server = server->next; + } while (server != started_at); + } + + ns = (struct nameserver *) malloc(sizeof(struct nameserver)); + if (!ns) return -1; + + memset(ns, 0, sizeof(struct nameserver)); + + ns->socket = socket(PF_INET, SOCK_DGRAM, 0); + if (ns->socket < 0) { err = 1; goto out1; } +#ifdef WIN32 + { + u_long nonblocking = 1; + ioctlsocket(ns->socket, FIONBIO, &nonblocking); + } +#else + fcntl(ns->socket, F_SETFL, O_NONBLOCK); +#endif + sin.sin_addr.s_addr = address; + sin.sin_port = htons(port); + sin.sin_family = AF_INET; + if (connect(ns->socket, (struct sockaddr *) &sin, sizeof(sin)) != 0) { + err = 2; + goto out2; + } + + ns->address = address; + ns->state = 1; + event_set(&ns->event, ns->socket, EV_READ | EV_PERSIST, nameserver_ready_callback, ns); + if (event_add(&ns->event, NULL) < 0) { + err = 2; + goto out2; + } + + log(EVDNS_LOG_DEBUG, "Added nameserver %s", debug_ntoa(address)); + + /* insert this nameserver into the list of them */ + if (!server_head) { + ns->next = ns->prev = ns; + server_head = ns; + } else { + ns->next = server_head->next; + ns->prev = server_head; + server_head->next = ns; + if (server_head->prev == server_head) { + server_head->prev = ns; + } + } + + global_good_nameservers++; + + return 0; + +out2: + CLOSE_SOCKET(ns->socket); +out1: + free(ns); + log(EVDNS_LOG_WARN, "Unable to add nameserver %s: error %d", debug_ntoa(address), err); + return err; +} + +/* exported function */ +int +evdns_nameserver_add(unsigned long int address) { + return _evdns_nameserver_add_impl(address, 53); +} + +/* exported function */ +int +evdns_nameserver_ip_add(const char *ip_as_string) { + struct in_addr ina; + int port; + char buf[20]; + const char *cp; + cp = strchr(ip_as_string, ':'); + if (! cp) { + cp = ip_as_string; + port = 53; + } else { + port = strtoint(cp+1); + if (port < 0 || port > 65535) { + return 4; + } + if ((cp-ip_as_string) >= (int)sizeof(buf)) { + return 4; + } + memcpy(buf, ip_as_string, cp-ip_as_string); + buf[cp-ip_as_string] = '\0'; + cp = buf; + } + if (!inet_aton(cp, &ina)) { + return 4; + } + return _evdns_nameserver_add_impl(ina.s_addr, port); +} + +/* insert into the tail of the queue */ +static void +evdns_request_insert(struct request *req, struct request **head) { + if (!*head) { + *head = req; + req->next = req->prev = req; + return; + } + + req->prev = (*head)->prev; + req->prev->next = req; + req->next = *head; + (*head)->prev = req; +} + +static int +string_num_dots(const char *s) { + int count = 0; + while ((s = strchr(s, '.'))) { + s++; + count++; + } + return count; +} + +static struct request * +request_new(int type, const char *name, int flags, + evdns_callback_type callback, void *user_ptr) { + const char issuing_now = + (global_requests_inflight < global_max_requests_inflight) ? 1 : 0; + + const int name_len = strlen(name); + const int request_max_len = evdns_request_len(name_len); + const u16 trans_id = issuing_now ? transaction_id_pick() : 0xffff; + /* the request data is alloced in a single block with the header */ + struct request *const req = + (struct request *) malloc(sizeof(struct request) + request_max_len); + int rlen; + (void) flags; + + if (!req) return NULL; + memset(req, 0, sizeof(struct request)); + + /* request data lives just after the header */ + req->request = ((u8 *) req) + sizeof(struct request); + /* denotes that the request data shouldn't be free()ed */ + req->request_appended = 1; + rlen = evdns_request_data_build(name, name_len, trans_id, + type, CLASS_INET, req->request, request_max_len); + if (rlen < 0) + goto err1; + req->request_len = rlen; + req->trans_id = trans_id; + req->tx_count = 0; + req->request_type = type; + req->user_pointer = user_ptr; + req->user_callback = callback; + req->ns = issuing_now ? nameserver_pick() : NULL; + req->next = req->prev = NULL; + + return req; +err1: + free(req); + return NULL; +} + +static void +request_submit(struct request *const req) { + if (req->ns) { + /* if it has a nameserver assigned then this is going */ + /* straight into the inflight queue */ + evdns_request_insert(req, &req_head); + global_requests_inflight++; + evdns_request_transmit(req); + } else { + evdns_request_insert(req, &req_waiting_head); + global_requests_waiting++; + } +} + +/* exported function */ +int evdns_resolve_ipv4(const char *name, int flags, + evdns_callback_type callback, void *ptr) { + log(EVDNS_LOG_DEBUG, "Resolve requested for %s", name); + if (flags & DNS_QUERY_NO_SEARCH) { + struct request *const req = + request_new(TYPE_A, name, flags, callback, ptr); + if (req == NULL) + return (1); + request_submit(req); + return (0); + } else { + return (search_request_new(TYPE_A, name, flags, callback, ptr)); + } +} + +/* exported function */ +int evdns_resolve_ipv6(const char *name, int flags, + evdns_callback_type callback, void *ptr) { + log(EVDNS_LOG_DEBUG, "Resolve requested for %s", name); + if (flags & DNS_QUERY_NO_SEARCH) { + struct request *const req = + request_new(TYPE_AAAA, name, flags, callback, ptr); + if (req == NULL) + return (1); + request_submit(req); + return (0); + } else { + return (search_request_new(TYPE_AAAA, name, flags, callback, ptr)); + } +} + +int evdns_resolve_reverse(struct in_addr *in, int flags, evdns_callback_type callback, void *ptr) { + char buf[32]; + struct request *req; + u32 a; + assert(in); + a = ntohl(in->s_addr); + snprintf(buf, sizeof(buf), "%d.%d.%d.%d.in-addr.arpa", + (int)(u8)((a )&0xff), + (int)(u8)((a>>8 )&0xff), + (int)(u8)((a>>16)&0xff), + (int)(u8)((a>>24)&0xff)); + log(EVDNS_LOG_DEBUG, "Resolve requested for %s (reverse)", buf); + req = request_new(TYPE_PTR, buf, flags, callback, ptr); + if (!req) return 1; + request_submit(req); + return 0; +} + +int evdns_resolve_reverse_ipv6(struct in6_addr *in, int flags, evdns_callback_type callback, void *ptr) { + char buf[64]; + char *cp; + struct request *req; + int i; + assert(in); + cp = buf; + for (i=15; i >= 0; --i) { + u8 byte = in->s6_addr[i]; + *cp++ = "0123456789abcdef"[byte & 0x0f]; + *cp++ = '.'; + *cp++ = "0123456789abcdef"[byte >> 4]; + *cp++ = '.'; + } + assert(cp + strlen(".ip6.arpa") < buf+sizeof(buf)); + memcpy(cp, ".ip6.arpa", strlen(".ip6.arpa")+1); + log(EVDNS_LOG_DEBUG, "Resolve requested for %s (reverse)", buf); + req = request_new(TYPE_PTR, buf, flags, callback, ptr); + if (!req) return 1; + request_submit(req); + return 0; +} + +/*/////////////////////////////////////////////////////////////////// */ +/* Search support */ +/* */ +/* the libc resolver has support for searching a number of domains */ +/* to find a name. If nothing else then it takes the single domain */ +/* from the gethostname() call. */ +/* */ +/* It can also be configured via the domain and search options in a */ +/* resolv.conf. */ +/* */ +/* The ndots option controls how many dots it takes for the resolver */ +/* to decide that a name is non-local and so try a raw lookup first. */ + +struct search_domain { + int len; + struct search_domain *next; + /* the text string is appended to this structure */ +}; + +struct search_state { + int refcount; + int ndots; + int num_domains; + struct search_domain *head; +}; + +static struct search_state *global_search_state = NULL; + +static void +search_state_decref(struct search_state *const state) { + if (!state) return; + state->refcount--; + if (!state->refcount) { + struct search_domain *next, *dom; + for (dom = state->head; dom; dom = next) { + next = dom->next; + free(dom); + } + free(state); + } +} + +static struct search_state * +search_state_new(void) { + struct search_state *state = (struct search_state *) malloc(sizeof(struct search_state)); + if (!state) return NULL; + memset(state, 0, sizeof(struct search_state)); + state->refcount = 1; + state->ndots = 1; + + return state; +} + +static void +search_postfix_clear(void) { + search_state_decref(global_search_state); + + global_search_state = search_state_new(); +} + +/* exported function */ +void +evdns_search_clear(void) { + search_postfix_clear(); +} + +static void +search_postfix_add(const char *domain) { + int domain_len; + struct search_domain *sdomain; + while (domain[0] == '.') domain++; + domain_len = strlen(domain); + + if (!global_search_state) global_search_state = search_state_new(); + if (!global_search_state) return; + global_search_state->num_domains++; + + sdomain = (struct search_domain *) malloc(sizeof(struct search_domain) + domain_len); + if (!sdomain) return; + memcpy( ((u8 *) sdomain) + sizeof(struct search_domain), domain, domain_len); + sdomain->next = global_search_state->head; + sdomain->len = domain_len; + + global_search_state->head = sdomain; +} + +/* reverse the order of members in the postfix list. This is needed because, */ +/* when parsing resolv.conf we push elements in the wrong order */ +static void +search_reverse(void) { + struct search_domain *cur, *prev = NULL, *next; + cur = global_search_state->head; + while (cur) { + next = cur->next; + cur->next = prev; + prev = cur; + cur = next; + } + + global_search_state->head = prev; +} + +/* exported function */ +void +evdns_search_add(const char *domain) { + search_postfix_add(domain); +} + +/* exported function */ +void +evdns_search_ndots_set(const int ndots) { + if (!global_search_state) global_search_state = search_state_new(); + if (!global_search_state) return; + global_search_state->ndots = ndots; +} + +static void +search_set_from_hostname(void) { + char hostname[HOST_NAME_MAX + 1], *domainname; + + search_postfix_clear(); + if (gethostname(hostname, sizeof(hostname))) return; + domainname = strchr(hostname, '.'); + if (!domainname) return; + search_postfix_add(domainname); +} + +/* warning: returns malloced string */ +static char * +search_make_new(const struct search_state *const state, int n, const char *const base_name) { + const int base_len = strlen(base_name); + const char need_to_append_dot = base_name[base_len - 1] == '.' ? 0 : 1; + struct search_domain *dom; + + for (dom = state->head; dom; dom = dom->next) { + if (!n--) { + /* this is the postfix we want */ + /* the actual postfix string is kept at the end of the structure */ + const u8 *const postfix = ((u8 *) dom) + sizeof(struct search_domain); + const int postfix_len = dom->len; + char *const newname = (char *) malloc(base_len + need_to_append_dot + postfix_len + 1); + if (!newname) return NULL; + memcpy(newname, base_name, base_len); + if (need_to_append_dot) newname[base_len] = '.'; + memcpy(newname + base_len + need_to_append_dot, postfix, postfix_len); + newname[base_len + need_to_append_dot + postfix_len] = 0; + return newname; + } + } + + /* we ran off the end of the list and still didn't find the requested string */ + abort(); + return NULL; /* unreachable; stops warnings in some compilers. */ +} + +static int +search_request_new(int type, const char *const name, int flags, evdns_callback_type user_callback, void *user_arg) { + assert(type == TYPE_A || type == TYPE_AAAA); + if ( ((flags & DNS_QUERY_NO_SEARCH) == 0) && + global_search_state && + global_search_state->num_domains) { + /* we have some domains to search */ + struct request *req; + if (string_num_dots(name) >= global_search_state->ndots) { + req = request_new(type, name, flags, user_callback, user_arg); + if (!req) return 1; + req->search_index = -1; + } else { + char *const new_name = search_make_new(global_search_state, 0, name); + if (!new_name) return 1; + req = request_new(type, new_name, flags, user_callback, user_arg); + free(new_name); + if (!req) return 1; + req->search_index = 0; + } + req->search_origname = strdup(name); + req->search_state = global_search_state; + req->search_flags = flags; + global_search_state->refcount++; + request_submit(req); + return 0; + } else { + struct request *const req = request_new(type, name, flags, user_callback, user_arg); + if (!req) return 1; + request_submit(req); + return 0; + } +} + +/* this is called when a request has failed to find a name. We need to check */ +/* if it is part of a search and, if so, try the next name in the list */ +/* returns: */ +/* 0 another request has been submitted */ +/* 1 no more requests needed */ +static int +search_try_next(struct request *const req) { + if (req->search_state) { + /* it is part of a search */ + char *new_name; + struct request *newreq; + req->search_index++; + if (req->search_index >= req->search_state->num_domains) { + /* no more postfixes to try, however we may need to try */ + /* this name without a postfix */ + if (string_num_dots(req->search_origname) < req->search_state->ndots) { + /* yep, we need to try it raw */ + struct request *const newreq = request_new(req->request_type, req->search_origname, req->search_flags, req->user_callback, req->user_pointer); + log(EVDNS_LOG_DEBUG, "Search: trying raw query %s", req->search_origname); + if (newreq) { + request_submit(newreq); + return 0; + } + } + return 1; + } + + new_name = search_make_new(req->search_state, req->search_index, req->search_origname); + if (!new_name) return 1; + log(EVDNS_LOG_DEBUG, "Search: now trying %s (%d)", new_name, req->search_index); + newreq = request_new(req->request_type, new_name, req->search_flags, req->user_callback, req->user_pointer); + free(new_name); + if (!newreq) return 1; + newreq->search_origname = req->search_origname; + req->search_origname = NULL; + newreq->search_state = req->search_state; + newreq->search_flags = req->search_flags; + newreq->search_index = req->search_index; + newreq->search_state->refcount++; + request_submit(newreq); + return 0; + } + return 1; +} + +static void +search_request_finished(struct request *const req) { + if (req->search_state) { + search_state_decref(req->search_state); + req->search_state = NULL; + } + if (req->search_origname) { + free(req->search_origname); + req->search_origname = NULL; + } +} + +/*/////////////////////////////////////////////////////////////////// */ +/* Parsing resolv.conf files */ + +static void +evdns_resolv_set_defaults(int flags) { + /* if the file isn't found then we assume a local resolver */ + if (flags & DNS_OPTION_SEARCH) search_set_from_hostname(); + if (flags & DNS_OPTION_NAMESERVERS) evdns_nameserver_ip_add("127.0.0.1"); +} + +#ifndef HAVE_STRTOK_R +static char * +strtok_r(char *s, const char *delim, char **state) { + return strtok(s, delim); +} +#endif + +/* helper version of atoi which returns -1 on error */ +static int +strtoint(const char *const str) { + char *endptr; + const int r = strtol(str, &endptr, 10); + if (*endptr) return -1; + return r; +} + +/* helper version of atoi that returns -1 on error and clips to bounds. */ +static int +strtoint_clipped(const char *const str, int min, int max) +{ + int r = strtoint(str); + if (r == -1) + return r; + else if (rmax) + return max; + else + return r; +} + +/* exported function */ +int +evdns_set_option(const char *option, const char *val, int flags) +{ + if (!strncmp(option, "ndots:", 6)) { + const int ndots = strtoint(val); + if (ndots == -1) return -1; + if (!(flags & DNS_OPTION_SEARCH)) return 0; + log(EVDNS_LOG_DEBUG, "Setting ndots to %d", ndots); + if (!global_search_state) global_search_state = search_state_new(); + if (!global_search_state) return -1; + global_search_state->ndots = ndots; + } else if (!strncmp(option, "timeout:", 8)) { + const int timeout = strtoint(val); + if (timeout == -1) return -1; + if (!(flags & DNS_OPTION_MISC)) return 0; + log(EVDNS_LOG_DEBUG, "Setting timeout to %d", timeout); + global_timeout.tv_sec = timeout; + } else if (!strncmp(option, "max-timeouts:", 12)) { + const int maxtimeout = strtoint_clipped(val, 1, 255); + if (maxtimeout == -1) return -1; + if (!(flags & DNS_OPTION_MISC)) return 0; + log(EVDNS_LOG_DEBUG, "Setting maximum allowed timeouts to %d", + maxtimeout); + global_max_nameserver_timeout = maxtimeout; + } else if (!strncmp(option, "max-inflight:", 13)) { + const int maxinflight = strtoint_clipped(val, 1, 65000); + if (maxinflight == -1) return -1; + if (!(flags & DNS_OPTION_MISC)) return 0; + log(EVDNS_LOG_DEBUG, "Setting maximum inflight requests to %d", + maxinflight); + global_max_requests_inflight = maxinflight; + } else if (!strncmp(option, "attempts:", 9)) { + int retries = strtoint(val); + if (retries == -1) return -1; + if (retries > 255) retries = 255; + if (!(flags & DNS_OPTION_MISC)) return 0; + log(EVDNS_LOG_DEBUG, "Setting retries to %d", retries); + global_max_retransmits = retries; + } + return 0; +} + +static void +resolv_conf_parse_line(char *const start, int flags) { + char *strtok_state; + static const char *const delims = " \t"; +#define NEXT_TOKEN strtok_r(NULL, delims, &strtok_state) + + char *const first_token = strtok_r(start, delims, &strtok_state); + if (!first_token) return; + + if (!strcmp(first_token, "nameserver") && (flags & DNS_OPTION_NAMESERVERS)) { + const char *const nameserver = NEXT_TOKEN; + struct in_addr ina; + + if (inet_aton(nameserver, &ina)) { + /* address is valid */ + evdns_nameserver_add(ina.s_addr); + } + } else if (!strcmp(first_token, "domain") && (flags & DNS_OPTION_SEARCH)) { + const char *const domain = NEXT_TOKEN; + if (domain) { + search_postfix_clear(); + search_postfix_add(domain); + } + } else if (!strcmp(first_token, "search") && (flags & DNS_OPTION_SEARCH)) { + const char *domain; + search_postfix_clear(); + + while ((domain = NEXT_TOKEN)) { + search_postfix_add(domain); + } + search_reverse(); + } else if (!strcmp(first_token, "options")) { + const char *option; + while ((option = NEXT_TOKEN)) { + const char *val = strchr(option, ':'); + evdns_set_option(option, val ? val+1 : "", flags); + } + } +#undef NEXT_TOKEN +} + +/* exported function */ +/* returns: */ +/* 0 no errors */ +/* 1 failed to open file */ +/* 2 failed to stat file */ +/* 3 file too large */ +/* 4 out of memory */ +/* 5 short read from file */ +int +evdns_resolv_conf_parse(int flags, const char *const filename) { + struct stat st; + int fd, n, r; + u8 *resolv; + char *start; + int err = 0; + + log(EVDNS_LOG_DEBUG, "Parsing resolv.conf file %s", filename); + + fd = open(filename, O_RDONLY); + if (fd < 0) { + evdns_resolv_set_defaults(flags); + return 1; + } + + if (fstat(fd, &st)) { err = 2; goto out1; } + if (!st.st_size) { + evdns_resolv_set_defaults(flags); + err = (flags & DNS_OPTION_NAMESERVERS) ? 6 : 0; + goto out1; + } + if (st.st_size > 65535) { err = 3; goto out1; } /* no resolv.conf should be any bigger */ + + resolv = (u8 *) malloc((size_t)st.st_size + 1); + if (!resolv) { err = 4; goto out1; } + + n = 0; + while ((r = read(fd, resolv+n, (size_t)st.st_size-n)) > 0) { + n += r; + if (n == st.st_size) + break; + assert(n < st.st_size); + } + if (r < 0) { err = 5; goto out2; } + resolv[n] = 0; /* we malloced an extra byte; this should be fine. */ + + start = (char *) resolv; + for (;;) { + char *const newline = strchr(start, '\n'); + if (!newline) { + resolv_conf_parse_line(start, flags); + break; + } else { + *newline = 0; + resolv_conf_parse_line(start, flags); + start = newline + 1; + } + } + + if (!server_head && (flags & DNS_OPTION_NAMESERVERS)) { + /* no nameservers were configured. */ + evdns_nameserver_ip_add("127.0.0.1"); + err = 6; + } + if (flags & DNS_OPTION_SEARCH && (!global_search_state || global_search_state->num_domains == 0)) { + search_set_from_hostname(); + } + +out2: + free(resolv); +out1: + close(fd); + return err; +} + +#ifdef WIN32 +/* Add multiple nameservers from a space-or-comma-separated list. */ +static int +evdns_nameserver_ip_add_line(const char *ips) { + const char *addr; + char *buf; + int r; + while (*ips) { + while (ISSPACE(*ips) || *ips == ',' || *ips == '\t') + ++ips; + addr = ips; + while (ISDIGIT(*ips) || *ips == '.' || *ips == ':') + ++ips; + buf = malloc(ips-addr+1); + if (!buf) return 4; + memcpy(buf, addr, ips-addr); + buf[ips-addr] = '\0'; + r = evdns_nameserver_ip_add(buf); + free(buf); + if (r) return r; + } + return 0; +} + +typedef DWORD(WINAPI *GetNetworkParams_fn_t)(FIXED_INFO *, DWORD*); + +/* Use the windows GetNetworkParams interface in iphlpapi.dll to */ +/* figure out what our nameservers are. */ +static int +load_nameservers_with_getnetworkparams(void) +{ + /* Based on MSDN examples and inspection of c-ares code. */ + FIXED_INFO *fixed; + HMODULE handle = 0; + ULONG size = sizeof(FIXED_INFO); + void *buf = NULL; + int status = 0, r, added_any; + IP_ADDR_STRING *ns; + GetNetworkParams_fn_t fn; + + if (!(handle = LoadLibrary("iphlpapi.dll"))) { + log(EVDNS_LOG_WARN, "Could not open iphlpapi.dll"); + status = -1; + goto done; + } + if (!(fn = (GetNetworkParams_fn_t) GetProcAddress(handle, "GetNetworkParams"))) { + log(EVDNS_LOG_WARN, "Could not get address of function."); + status = -1; + goto done; + } + + buf = malloc(size); + if (!buf) { status = 4; goto done; } + fixed = buf; + r = fn(fixed, &size); + if (r != ERROR_SUCCESS && r != ERROR_BUFFER_OVERFLOW) { + status = -1; + goto done; + } + if (r != ERROR_SUCCESS) { + free(buf); + buf = malloc(size); + if (!buf) { status = 4; goto done; } + fixed = buf; + r = fn(fixed, &size); + if (r != ERROR_SUCCESS) { + log(EVDNS_LOG_DEBUG, "fn() failed."); + status = -1; + goto done; + } + } + + assert(fixed); + added_any = 0; + ns = &(fixed->DnsServerList); + while (ns) { + r = evdns_nameserver_ip_add_line(ns->IpAddress.String); + if (r) { + log(EVDNS_LOG_DEBUG,"Could not add nameserver %s to list,error: %d", + (ns->IpAddress.String),(int)GetLastError()); + status = r; + goto done; + } else { + log(EVDNS_LOG_DEBUG,"Succesfully added %s as nameserver",ns->IpAddress.String); + } + + added_any++; + ns = ns->Next; + } + + if (!added_any) { + log(EVDNS_LOG_DEBUG, "No nameservers added."); + status = -1; + } + + done: + if (buf) + free(buf); + if (handle) + FreeLibrary(handle); + return status; +} + +static int +config_nameserver_from_reg_key(HKEY key, const char *subkey) +{ + char *buf; + DWORD bufsz = 0, type = 0; + int status = 0; + + if (RegQueryValueEx(key, subkey, 0, &type, NULL, &bufsz) + != ERROR_MORE_DATA) + return -1; + if (!(buf = malloc(bufsz))) + return -1; + + if (RegQueryValueEx(key, subkey, 0, &type, (LPBYTE)buf, &bufsz) + == ERROR_SUCCESS && bufsz > 1) { + status = evdns_nameserver_ip_add_line(buf); + } + + free(buf); + return status; +} + +#define SERVICES_KEY "System\\CurrentControlSet\\Services\\" +#define WIN_NS_9X_KEY SERVICES_KEY "VxD\\MSTCP" +#define WIN_NS_NT_KEY SERVICES_KEY "Tcpip\\Parameters" + +static int +load_nameservers_from_registry(void) +{ + int found = 0; + int r; +#define TRY(k, name) \ + if (!found && config_nameserver_from_reg_key(k,name) == 0) { \ + log(EVDNS_LOG_DEBUG,"Found nameservers in %s/%s",#k,name); \ + found = 1; \ + } else if (!found) { \ + log(EVDNS_LOG_DEBUG,"Didn't find nameservers in %s/%s", \ + #k,#name); \ + } + + if (((int)GetVersion()) > 0) { /* NT */ + HKEY nt_key = 0, interfaces_key = 0; + + if (RegOpenKeyEx(HKEY_LOCAL_MACHINE, WIN_NS_NT_KEY, 0, + KEY_READ, &nt_key) != ERROR_SUCCESS) { + log(EVDNS_LOG_DEBUG,"Couldn't open nt key, %d",(int)GetLastError()); + return -1; + } + r = RegOpenKeyEx(nt_key, "Interfaces", 0, + KEY_QUERY_VALUE|KEY_ENUMERATE_SUB_KEYS, + &interfaces_key); + if (r != ERROR_SUCCESS) { + log(EVDNS_LOG_DEBUG,"Couldn't open interfaces key, %d",(int)GetLastError()); + return -1; + } + TRY(nt_key, "NameServer"); + TRY(nt_key, "DhcpNameServer"); + TRY(interfaces_key, "NameServer"); + TRY(interfaces_key, "DhcpNameServer"); + RegCloseKey(interfaces_key); + RegCloseKey(nt_key); + } else { + HKEY win_key = 0; + if (RegOpenKeyEx(HKEY_LOCAL_MACHINE, WIN_NS_9X_KEY, 0, + KEY_READ, &win_key) != ERROR_SUCCESS) { + log(EVDNS_LOG_DEBUG, "Couldn't open registry key, %d", (int)GetLastError()); + return -1; + } + TRY(win_key, "NameServer"); + RegCloseKey(win_key); + } + + if (found == 0) { + log(EVDNS_LOG_WARN,"Didn't find any nameservers."); + } + + return found ? 0 : -1; +#undef TRY +} + +int +evdns_config_windows_nameservers(void) +{ + if (load_nameservers_with_getnetworkparams() == 0) + return 0; + return load_nameservers_from_registry(); +} +#endif + +int +evdns_init(void) +{ + int res = 0; +#ifdef WIN32 + evdns_config_windows_nameservers(); +#else + res = evdns_resolv_conf_parse(DNS_OPTIONS_ALL, "/etc/resolv.conf"); +#endif + + return (res); +} + +const char * +evdns_err_to_string(int err) +{ + switch (err) { + case DNS_ERR_NONE: return "no error"; + case DNS_ERR_FORMAT: return "misformatted query"; + case DNS_ERR_SERVERFAILED: return "server failed"; + case DNS_ERR_NOTEXIST: return "name does not exist"; + case DNS_ERR_NOTIMPL: return "query not implemented"; + case DNS_ERR_REFUSED: return "refused"; + + case DNS_ERR_TRUNCATED: return "reply truncated or ill-formed"; + case DNS_ERR_UNKNOWN: return "unknown"; + case DNS_ERR_TIMEOUT: return "request timed out"; + case DNS_ERR_SHUTDOWN: return "dns subsystem shut down"; + default: return "[Unknown error code]"; + } +} + +void +evdns_shutdown(int fail_requests) +{ + struct nameserver *server, *server_next; + struct search_domain *dom, *dom_next; + + while (req_head) { + if (fail_requests) + reply_callback(req_head, 0, DNS_ERR_SHUTDOWN, NULL); + request_finished(req_head, &req_head); + } + while (req_waiting_head) { + if (fail_requests) + reply_callback(req_waiting_head, 0, DNS_ERR_SHUTDOWN, NULL); + request_finished(req_waiting_head, &req_waiting_head); + } + global_requests_inflight = global_requests_waiting = 0; + + for (server = server_head; server; server = server_next) { + server_next = server->next; + if (server->socket >= 0) + CLOSE_SOCKET(server->socket); + (void) event_del(&server->event); + if (server->state == 0) + (void) event_del(&server->timeout_event); + free(server); + if (server_next == server_head) + break; + } + server_head = NULL; + global_good_nameservers = 0; + + if (global_search_state) { + for (dom = global_search_state->head; dom; dom = dom_next) { + dom_next = dom->next; + free(dom); + } + free(global_search_state); + global_search_state = NULL; + } + evdns_log_fn = NULL; +} + +#ifdef EVDNS_MAIN +void +main_callback(int result, char type, int count, int ttl, + void *addrs, void *orig) { + char *n = (char*)orig; + int i; + for (i = 0; i < count; ++i) { + if (type == DNS_IPv4_A) { + printf("%s: %s\n", n, debug_ntoa(((u32*)addrs)[i])); + } else if (type == DNS_PTR) { + printf("%s: %s\n", n, ((char**)addrs)[i]); + } + } + if (!count) { + printf("%s: No answer (%d)\n", n, result); + } + fflush(stdout); +} +void +evdns_server_callback(struct evdns_server_request *req, void *data) +{ + int i, r; + (void)data; + /* dummy; give 192.168.11.11 as an answer for all A questions, + * give foo.bar.example.com as an answer for all PTR questions. */ + for (i = 0; i < req->nquestions; ++i) { + u32 ans = htonl(0xc0a80b0bUL); + if (req->questions[i]->type == EVDNS_TYPE_A && + req->questions[i]->class == EVDNS_CLASS_INET) { + printf(" -- replying for %s (A)\n", req->questions[i]->name); + r = evdns_server_request_add_a_reply(req, req->questions[i]->name, + 1, &ans, 10); + if (r<0) + printf("eeep, didn't work.\n"); + } else if (req->questions[i]->type == EVDNS_TYPE_PTR && + req->questions[i]->class == EVDNS_CLASS_INET) { + printf(" -- replying for %s (PTR)\n", req->questions[i]->name); + r = evdns_server_request_add_ptr_reply(req, NULL, req->questions[i]->name, + "foo.bar.example.com", 10); + } else { + printf(" -- skipping %s [%d %d]\n", req->questions[i]->name, + req->questions[i]->type, req->questions[i]->class); + } + } + + r = evdns_request_respond(req, 0); + if (r<0) + printf("eeek, couldn't send reply.\n"); +} + +void +logfn(int is_warn, const char *msg) { + (void) is_warn; + fprintf(stderr, "%s\n", msg); +} +int +main(int c, char **v) { + int idx; + int reverse = 0, verbose = 1, servertest = 0; + if (c<2) { + fprintf(stderr, "syntax: %s [-x] [-v] hostname\n", v[0]); + fprintf(stderr, "syntax: %s [-servertest]\n", v[0]); + return 1; + } + idx = 1; + while (idx < c && v[idx][0] == '-') { + if (!strcmp(v[idx], "-x")) + reverse = 1; + else if (!strcmp(v[idx], "-v")) + verbose = 1; + else if (!strcmp(v[idx], "-servertest")) + servertest = 1; + else + fprintf(stderr, "Unknown option %s\n", v[idx]); + ++idx; + } + event_init(); + if (verbose) + evdns_set_log_fn(logfn); + evdns_resolv_conf_parse(DNS_OPTION_NAMESERVERS, "/etc/resolv.conf"); + if (servertest) { + int sock; + struct sockaddr_in my_addr; + sock = socket(PF_INET, SOCK_DGRAM, 0); + fcntl(sock, F_SETFL, O_NONBLOCK); + my_addr.sin_family = AF_INET; + my_addr.sin_port = htons(10053); + my_addr.sin_addr.s_addr = INADDR_ANY; + if (bind(sock, (struct sockaddr*)&my_addr, sizeof(my_addr))<0) { + perror("bind"); + exit(1); + } + evdns_add_server_port(sock, 0, evdns_server_callback, NULL); + } + for (; idx < c; ++idx) { + if (reverse) { + struct in_addr addr; + if (!inet_aton(v[idx], &addr)) { + fprintf(stderr, "Skipping non-IP %s\n", v[idx]); + continue; + } + fprintf(stderr, "resolving %s...\n",v[idx]); + evdns_resolve_reverse(&addr, 0, main_callback, v[idx]); + } else { + fprintf(stderr, "resolving (fwd) %s...\n",v[idx]); + evdns_resolve_ipv4(v[idx], 0, main_callback, v[idx]); + } + } + fflush(stdout); + event_dispatch(); + return 0; +} +#endif diff --git a/contrib/libevent/evdns.h b/contrib/libevent/evdns.h new file mode 100644 index 0000000000..af7962e193 --- /dev/null +++ b/contrib/libevent/evdns.h @@ -0,0 +1,367 @@ +/* + * Copyright (c) 2006 Niels Provos + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +/* + * The original DNS code is due to Adam Langley with heavy + * modifications by Nick Mathewson. Adam put his DNS software in the + * public domain. You can find his original copyright below. Please, + * aware that the code as part of libevent is governed by the 3-clause + * BSD license above. + * + * This software is Public Domain. To view a copy of the public domain dedication, + * visit http://creativecommons.org/licenses/publicdomain/ or send a letter to + * Creative Commons, 559 Nathan Abbott Way, Stanford, California 94305, USA. + * + * I ask and expect, but do not require, that all derivative works contain an + * attribution similar to: + * Parts developed by Adam Langley + * + * You may wish to replace the word "Parts" with something else depending on + * the amount of original code. + * + * (Derivative works does not include programs which link against, run or include + * the source verbatim in their source distributions) + */ + +/* + * Welcome, gentle reader + * + * Async DNS lookups are really a whole lot harder than they should be, + * mostly stemming from the fact that the libc resolver has never been + * very good at them. Before you use this library you should see if libc + * can do the job for you with the modern async call getaddrinfo_a + * (see http://www.imperialviolet.org/page25.html#e498). Otherwise, + * please continue. + * + * This code is based on libevent and you must call event_init before + * any of the APIs in this file. You must also seed the OpenSSL random + * source if you are using OpenSSL for ids (see below). + * + * This library is designed to be included and shipped with your source + * code. You statically link with it. You should also test for the + * existence of strtok_r and define HAVE_STRTOK_R if you have it. + * + * The DNS protocol requires a good source of id numbers and these + * numbers should be unpredictable for spoofing reasons. There are + * three methods for generating them here and you must define exactly + * one of them. In increasing order of preference: + * + * DNS_USE_GETTIMEOFDAY_FOR_ID: + * Using the bottom 16 bits of the usec result from gettimeofday. This + * is a pretty poor solution but should work anywhere. + * DNS_USE_CPU_CLOCK_FOR_ID: + * Using the bottom 16 bits of the nsec result from the CPU's time + * counter. This is better, but may not work everywhere. Requires + * POSIX realtime support and you'll need to link against -lrt on + * glibc systems at least. + * DNS_USE_OPENSSL_FOR_ID: + * Uses the OpenSSL RAND_bytes call to generate the data. You must + * have seeded the pool before making any calls to this library. + * + * The library keeps track of the state of nameservers and will avoid + * them when they go down. Otherwise it will round robin between them. + * + * Quick start guide: + * #include "evdns.h" + * void callback(int result, char type, int count, int ttl, + * void *addresses, void *arg); + * evdns_resolv_conf_parse(DNS_OPTIONS_ALL, "/etc/resolv.conf"); + * evdns_resolve("www.hostname.com", 0, callback, NULL); + * + * When the lookup is complete the callback function is called. The + * first argument will be one of the DNS_ERR_* defines in evdns.h. + * Hopefully it will be DNS_ERR_NONE, in which case type will be + * DNS_IPv4_A, count will be the number of IP addresses, ttl is the time + * which the data can be cached for (in seconds), addresses will point + * to an array of uint32_t's and arg will be whatever you passed to + * evdns_resolve. + * + * Searching: + * + * In order for this library to be a good replacement for glibc's resolver it + * supports searching. This involves setting a list of default domains, in + * which names will be queried for. The number of dots in the query name + * determines the order in which this list is used. + * + * Searching appears to be a single lookup from the point of view of the API, + * although many DNS queries may be generated from a single call to + * evdns_resolve. Searching can also drastically slow down the resolution + * of names. + * + * To disable searching: + * 1. Never set it up. If you never call evdns_resolv_conf_parse or + * evdns_search_add then no searching will occur. + * + * 2. If you do call evdns_resolv_conf_parse then don't pass + * DNS_OPTION_SEARCH (or DNS_OPTIONS_ALL, which implies it). + * + * 3. When calling evdns_resolve, pass the DNS_QUERY_NO_SEARCH flag. + * + * The order of searches depends on the number of dots in the name. If the + * number is greater than the ndots setting then the names is first tried + * globally. Otherwise each search domain is appended in turn. + * + * The ndots setting can either be set from a resolv.conf, or by calling + * evdns_search_ndots_set. + * + * For example, with ndots set to 1 (the default) and a search domain list of + * ["myhome.net"]: + * Query: www + * Order: www.myhome.net, www. + * + * Query: www.abc + * Order: www.abc., www.abc.myhome.net + * + * API reference: + * + * int evdns_nameserver_add(unsigned long int address) + * Add a nameserver. The address should be an IP address in + * network byte order. The type of address is chosen so that + * it matches in_addr.s_addr. + * Returns non-zero on error. + * + * int evdns_nameserver_ip_add(const char *ip_as_string) + * This wraps the above function by parsing a string as an IP + * address and adds it as a nameserver. + * Returns non-zero on error + * + * int evdns_resolve(const char *name, int flags, + * evdns_callback_type callback, + * void *ptr) + * Resolve a name. The name parameter should be a DNS name. + * The flags parameter should be 0, or DNS_QUERY_NO_SEARCH + * which disables searching for this query. (see defn of + * searching above). + * + * The callback argument is a function which is called when + * this query completes and ptr is an argument which is passed + * to that callback function. + * + * Returns non-zero on error + * + * void evdns_search_clear() + * Clears the list of search domains + * + * void evdns_search_add(const char *domain) + * Add a domain to the list of search domains + * + * void evdns_search_ndots_set(int ndots) + * Set the number of dots which, when found in a name, causes + * the first query to be without any search domain. + * + * int evdns_count_nameservers(void) + * Return the number of configured nameservers (not necessarily the + * number of running nameservers). This is useful for double-checking + * whether our calls to the various nameserver configuration functions + * have been successful. + * + * int evdns_clear_nameservers_and_suspend(void) + * Remove all currently configured nameservers, and suspend all pending + * resolves. Resolves will not necessarily be re-attempted until + * evdns_resume() is called. + * + * int evdns_resume(void) + * Re-attempt resolves left in limbo after an earlier call to + * evdns_clear_nameservers_and_suspend(). + * + * int evdns_config_windows_nameservers(void) + * Attempt to configure a set of nameservers based on platform settings on + * a win32 host. Preferentially tries to use GetNetworkParams; if that fails, + * looks in the registry. Returns 0 on success, nonzero on failure. + * + * int evdns_resolv_conf_parse(int flags, const char *filename) + * Parse a resolv.conf like file from the given filename. + * + * See the man page for resolv.conf for the format of this file. + * The flags argument determines what information is parsed from + * this file: + * DNS_OPTION_SEARCH - domain, search and ndots options + * DNS_OPTION_NAMESERVERS - nameserver lines + * DNS_OPTION_MISC - timeout and attempts options + * DNS_OPTIONS_ALL - all of the above + * The following directives are not parsed from the file: + * sortlist, rotate, no-check-names, inet6, debug + * + * Returns non-zero on error: + * 0 no errors + * 1 failed to open file + * 2 failed to stat file + * 3 file too large + * 4 out of memory + * 5 short read from file + * 6 no nameservers in file + * + * Internals: + * + * Requests are kept in two queues. The first is the inflight queue. In + * this queue requests have an allocated transaction id and nameserver. + * They will soon be transmitted if they haven't already been. + * + * The second is the waiting queue. The size of the inflight ring is + * limited and all other requests wait in waiting queue for space. This + * bounds the number of concurrent requests so that we don't flood the + * nameserver. Several algorithms require a full walk of the inflight + * queue and so bounding its size keeps thing going nicely under huge + * (many thousands of requests) loads. + * + * If a nameserver loses too many requests it is considered down and we + * try not to use it. After a while we send a probe to that nameserver + * (a lookup for google.com) and, if it replies, we consider it working + * again. If the nameserver fails a probe we wait longer to try again + * with the next probe. + */ + +#ifndef EVENTDNS_H +#define EVENTDNS_H + +#ifdef __cplusplus +extern "C" { +#endif + +/* Error codes 0-5 are as described in RFC 1035. */ +#define DNS_ERR_NONE 0 +/* The name server was unable to interpret the query */ +#define DNS_ERR_FORMAT 1 +/* The name server was unable to process this query due to a problem with the + * name server */ +#define DNS_ERR_SERVERFAILED 2 +/* The domain name does not exist */ +#define DNS_ERR_NOTEXIST 3 +/* The name server does not support the requested kind of query */ +#define DNS_ERR_NOTIMPL 4 +/* The name server refuses to reform the specified operation for policy + * reasons */ +#define DNS_ERR_REFUSED 5 +/* The reply was truncated or ill-formated */ +#define DNS_ERR_TRUNCATED 65 +/* An unknown error occurred */ +#define DNS_ERR_UNKNOWN 66 +/* Communication with the server timed out */ +#define DNS_ERR_TIMEOUT 67 +/* The request was canceled because the DNS subsystem was shut down. */ +#define DNS_ERR_SHUTDOWN 68 + +#define DNS_IPv4_A 1 +#define DNS_PTR 2 +#define DNS_IPv6_AAAA 3 + +#define DNS_QUERY_NO_SEARCH 1 + +#define DNS_OPTION_SEARCH 1 +#define DNS_OPTION_NAMESERVERS 2 +#define DNS_OPTION_MISC 4 +#define DNS_OPTIONS_ALL 7 + +/* + * The callback that contains the results from a lookup. + * - type is either DNS_IPv4_A or DNS_PTR or DNS_IPv6_AAAA + * - count contains the number of addresses of form type + * - ttl is the number of seconds the resolution may be cached for. + * - addresses needs to be cast according to type + */ +typedef void (*evdns_callback_type) (int result, char type, int count, int ttl, void *addresses, void *arg); + +int evdns_init(void); +void evdns_shutdown(int fail_requests); +const char *evdns_err_to_string(int err); +int evdns_nameserver_add(unsigned long int address); +int evdns_count_nameservers(void); +int evdns_clear_nameservers_and_suspend(void); +int evdns_resume(void); +int evdns_nameserver_ip_add(const char *ip_as_string); +int evdns_resolve_ipv4(const char *name, int flags, evdns_callback_type callback, void *ptr); +int evdns_resolve_ipv6(const char *name, int flags, evdns_callback_type callback, void *ptr); +struct in_addr; +struct in6_addr; +int evdns_resolve_reverse(struct in_addr *in, int flags, evdns_callback_type callback, void *ptr); +int evdns_resolve_reverse_ipv6(struct in6_addr *in, int flags, evdns_callback_type callback, void *ptr); +int evdns_set_option(const char *option, const char *val, int flags); +int evdns_resolv_conf_parse(int flags, const char *); +#ifdef MS_WINDOWS +int evdns_config_windows_nameservers(void); +#endif +void evdns_search_clear(void); +void evdns_search_add(const char *domain); +void evdns_search_ndots_set(const int ndots); + +typedef void (*evdns_debug_log_fn_type)(int is_warning, const char *msg); +void evdns_set_log_fn(evdns_debug_log_fn_type fn); + +#define DNS_NO_SEARCH 1 + +#ifdef __cplusplus +} +#endif + +/* + * Structures and functions used to implement a DNS server. + */ + +struct evdns_server_request { + int flags; + int nquestions; + struct evdns_server_question **questions; +}; +struct evdns_server_question { + int type; + int class; + char name[1]; +}; +typedef void (*evdns_request_callback_fn_type)(struct evdns_server_request *, void *); +#define EVDNS_ANSWER_SECTION 0 +#define EVDNS_AUTHORITY_SECTION 1 +#define EVDNS_ADDITIONAL_SECTION 2 + +#define EVDNS_TYPE_A 1 +#define EVDNS_TYPE_NS 2 +#define EVDNS_TYPE_CNAME 5 +#define EVDNS_TYPE_SOA 6 +#define EVDNS_TYPE_PTR 12 +#define EVDNS_TYPE_MX 15 +#define EVDNS_TYPE_TXT 16 +#define EVDNS_TYPE_AAAA 28 + +#define EVDNS_QTYPE_AXFR 252 +#define EVDNS_QTYPE_ALL 255 + +#define EVDNS_CLASS_INET 1 + +struct evdns_server_port *evdns_add_server_port(int socket, int is_tcp, evdns_request_callback_fn_type callback, void *user_data); +void evdns_close_server_port(struct evdns_server_port *port); + +int evdns_server_request_add_reply(struct evdns_server_request *req, int section, const char *name, int type, int class, int ttl, int datalen, int is_name, const char *data); +int evdns_server_request_add_a_reply(struct evdns_server_request *req, const char *name, int n, void *addrs, int ttl); +int evdns_server_request_add_aaaa_reply(struct evdns_server_request *req, const char *name, int n, void *addrs, int ttl); +int evdns_server_request_add_ptr_reply(struct evdns_server_request *req, struct in_addr *in, const char *inaddr_name, const char *hostname, int ttl); +int evdns_server_request_add_cname_reply(struct evdns_server_request *req, const char *name, const char *cname, int ttl); + +int evdns_server_request_respond(struct evdns_server_request *req, int err); +int evdns_server_request_drop(struct evdns_server_request *req); +struct sockaddr; +int evdns_server_request_get_requesting_addr(struct evdns_server_request *_req, struct sockaddr *sa, int addr_len); + +#endif /* !EVENTDNS_H */ diff --git a/contrib/libevent/event-internal.h b/contrib/libevent/event-internal.h new file mode 100644 index 0000000000..7fd4b6c690 --- /dev/null +++ b/contrib/libevent/event-internal.h @@ -0,0 +1,61 @@ +/* + * Copyright (c) 2000-2004 Niels Provos + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#ifndef _EVENT_INTERNAL_H_ +#define _EVENT_INTERNAL_H_ + +#ifdef __cplusplus +extern "C" { +#endif + +#include "evsignal.h" + +struct event_base { + const struct eventop *evsel; + void *evbase; + int event_count; /* counts number of total events */ + int event_count_active; /* counts number of active events */ + + int event_gotterm; /* Set to terminate loop */ + + /* active event management */ + struct event_list **activequeues; + int nactivequeues; + + /* signal handling info */ + struct evsignal_info sig; + + struct event_list eventqueue; + struct timeval event_tv; + + RB_HEAD(event_tree, event) timetree; +}; + +#ifdef __cplusplus +} +#endif + +#endif /* _EVENT_INTERNAL_H_ */ diff --git a/contrib/libevent/event.3 b/contrib/libevent/event.3 new file mode 100644 index 0000000000..a6648e9a9d --- /dev/null +++ b/contrib/libevent/event.3 @@ -0,0 +1,592 @@ +.\" $OpenBSD: event.3,v 1.4 2002/07/12 18:50:48 provos Exp $ +.\" +.\" Copyright (c) 2000 Artur Grabowski +.\" All rights reserved. +.\" +.\" Redistribution and use in source and binary forms, with or without +.\" modification, are permitted provided that the following conditions +.\" are met: +.\" +.\" 1. Redistributions of source code must retain the above copyright +.\" notice, this list of conditions and the following disclaimer. +.\" 2. Redistributions in binary form must reproduce the above copyright +.\" notice, this list of conditions and the following disclaimer in the +.\" documentation and/or other materials provided with the distribution. +.\" 3. The name of the author may not be used to endorse or promote products +.\" derived from this software without specific prior written permission. +.\" +.\" THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, +.\" INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY +.\" AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL +.\" THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, +.\" EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, +.\" PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; +.\" OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, +.\" WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR +.\" OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF +.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +.\" +.Dd August 8, 2000 +.Dt EVENT 3 +.Os +.Sh NAME +.Nm event_init , +.Nm event_dispatch , +.Nm event_loop , +.Nm event_loopexit , +.Nm event_set , +.Nm event_base_dispatch , +.Nm event_base_loop , +.Nm event_base_loopexit , +.Nm event_base_set , +.Nm event_base_free , +.Nm event_add , +.Nm event_del , +.Nm event_once , +.Nm event_base_once , +.Nm event_pending , +.Nm event_initialized , +.Nm event_priority_init , +.Nm event_priority_set , +.Nm evtimer_set , +.Nm evtimer_add , +.Nm evtimer_del , +.Nm evtimer_pending , +.Nm evtimer_initialized , +.Nm signal_set , +.Nm signal_add , +.Nm signal_del , +.Nm signal_pending , +.Nm signal_initialized , +.Nm bufferevent_new , +.Nm bufferevent_free , +.Nm bufferevent_write , +.Nm bufferevent_write_buffer , +.Nm bufferevent_read , +.Nm bufferevent_enable , +.Nm bufferevent_disable , +.Nm bufferevent_settimeout , +.Nm bufferevent_base_set , +.Nm evbuffer_new , +.Nm evbuffer_free , +.Nm evbuffer_add , +.Nm evbuffer_add_buffer , +.Nm evbuffer_add_printf , +.Nm evbuffer_add_vprintf , +.Nm evbuffer_drain , +.Nm evbuffer_write , +.Nm evbuffer_read , +.Nm evbuffer_find , +.Nm evbuffer_readline , +.Nm evhttp_start , +.Nm evhttp_free +.Nd execute a function when a specific event occurs +.Sh SYNOPSIS +.Fd #include +.Fd #include +.Ft "struct event_base *" +.Fn "event_init" "void" +.Ft int +.Fn "event_dispatch" "void" +.Ft int +.Fn "event_loop" "int flags" +.Ft int +.Fn "event_loopexit" "struct timeval *tv" +.Ft void +.Fn "event_set" "struct event *ev" "int fd" "short event" "void (*fn)(int, short, void *)" "void *arg" +.Ft int +.Fn "event_base_dispatch" "struct event_base *base" +.Ft int +.Fn "event_base_loop" "struct event_base *base" "int flags" +.Ft int +.Fn "event_base_loopexit" "struct event_base *base" "struct timeval *tv" +.Ft int +.Fn "event_base_set" "struct event_base *base" "struct event *" +.Ft void +.Fn "event_base_free" "struct event_base *base" +.Ft int +.Fn "event_add" "struct event *ev" "struct timeval *tv" +.Ft int +.Fn "event_del" "struct event *ev" +.Ft int +.Fn "event_once" "int fd" "short event" "void (*fn)(int, short, void *)" "void *arg" "struct timeval *tv" +.Ft int +.Fn "event_base_once" "struct event_base *base" "int fd" "short event" "void (*fn)(int, short, void *)" "void *arg" "struct timeval *tv" +.Ft int +.Fn "event_pending" "struct event *ev" "short event" "struct timeval *tv" +.Ft int +.Fn "event_initialized" "struct event *ev" +.Ft int +.Fn "event_priority_init" "int npriorities" +.Ft int +.Fn "event_priority_set" "struct event *ev" "int priority" +.Ft void +.Fn "evtimer_set" "struct event *ev" "void (*fn)(int, short, void *)" "void *arg" +.Ft void +.Fn "evtimer_add" "struct event *ev" "struct timeval *" +.Ft void +.Fn "evtimer_del" "struct event *ev" +.Ft int +.Fn "evtimer_pending" "struct event *ev" "struct timeval *tv" +.Ft int +.Fn "evtimer_initialized" "struct event *ev" +.Ft void +.Fn "signal_set" "struct event *ev" "int signal" "void (*fn)(int, short, void *)" "void *arg" +.Ft void +.Fn "signal_add" "struct event *ev" "struct timeval *" +.Ft void +.Fn "signal_del" "struct event *ev" +.Ft int +.Fn "signal_pending" "struct event *ev" "struct timeval *tv" +.Ft int +.Fn "signal_initialized" "struct event *ev" +.Ft "struct bufferevent *" +.Fn "bufferevent_new" "int fd" "evbuffercb readcb" "evbuffercb writecb" "everrorcb" "void *cbarg" +.Ft void +.Fn "bufferevent_free" "struct bufferevent *bufev" +.Ft int +.Fn "bufferevent_write" "struct bufferevent *bufev" "void *data" "size_t size" +.Ft int +.Fn "bufferevent_write_buffer" "struct bufferevent *bufev" "struct evbuffer *buf" +.Ft size_t +.Fn "bufferevent_read" "struct bufferevent *bufev" "void *data" "size_t size" +.Ft int +.Fn "bufferevent_enable" "struct bufferevent *bufev" "short event" +.Ft int +.Fn "bufferevent_disable" "struct bufferevent *bufev" "short event" +.Ft void +.Fn "bufferevent_settimeout" "struct bufferevent *bufev" "int timeout_read" "int timeout_write" +.Ft int +.Fn "bufferevent_base_set" "struct event_base *base" "struct bufferevent *bufev" +.Ft "struct evbuffer *" +.Fn "evbuffer_new" "void" +.Ft void +.Fn "evbuffer_free" "struct evbuffer *buf" +.Ft int +.Fn "evbuffer_add" "struct evbuffer *buf" "const void *data" "size_t size" +.Ft int +.Fn "evbuffer_add_buffer" "struct evbuffer *dst" "struct evbuffer *src" +.Ft int +.Fn "evbuffer_add_printf" "struct evbuffer *buf" "const char *fmt" "..." +.Ft int +.Fn "evbuffer_add_vprintf" "struct evbuffer *buf" "const char *fmt" "va_list ap" +.Ft void +.Fn "evbuffer_drain" "struct evbuffer *buf" "size_t size" +.Ft int +.Fn "evbuffer_write" "struct evbuffer *buf" "int fd" +.Ft int +.Fn "evbuffer_read" "struct evbuffer *buf" "int fd" "int size" +.Ft "u_char *" +.Fn "evbuffer_find" "struct evbuffer *buf" "const u_char *data" "size_t size" +.Ft "char *" +.Fn "evbuffer_readline" "struct evbuffer *buf" +.Ft "struct evhttp *" +.Fn "evhttp_start" "const char *address" "u_short port" +.Ft "void" +.Fn "evhttp_free" "struct evhttp* http" +.Ft int +.Fa (*event_sigcb)(void) ; +.Ft volatile sig_atomic_t +.Fa event_gotsig ; +.Sh DESCRIPTION +The +.Nm event +API provides a mechanism to execute a function when a specific event +on a file descriptor occurs or after a given time has passed. +.Pp +The +.Nm event +API needs to be initialized with +.Fn event_init +before it can be used. +.Pp +In order to process events, an application needs to call +.Fn event_dispatch . +This function only returns on error, and should replace the event core +of the application program. +.Pp +In order to avoid races in signal handlers, the +.Nm event +API provides two variables: +.Va event_sigcb +and +.Va event_gotsig . +A signal handler +sets +.Va event_gotsig +to indicate that a signal has been received. +The application sets +.Va event_sigcb +to a callback function. +After the signal handler sets +.Va event_gotsig , +.Nm event_dispatch +will execute the callback function to process received signals. +The callback returns 1 when no events are registered any more. +It can return \-1 to indicate an error to the +.Nm event +library, causing +.Fn event_dispatch +to terminate with +.Va errno +set to +.Er EINTR . +.Pp +The +.Nm event_loop +function provides an interface for single pass execution of pending +events. +The flags +.Va EVLOOP_ONCE +and +.Va EVLOOP_NONBLOCK +are recognized. +The +.Nm event_loopexit +function allows the loop to be terminated after some amount of time +has passed. +The parameter indicates the time after which the loop should terminate. +.Pp +It is the responsibility of the caller to provide these functions with +pre-allocated event structures. +.Pp +The function +.Fn event_set +prepares the event structure +.Fa ev +to be used in future calls to +.Fn event_add +and +.Fn event_del . +The event will be prepared to call the function specified by the +.Fa fn +argument with an +.Fa int +argument indicating the file descriptor, a +.Fa short +argument indicating the type of event, and a +.Fa void * +argument given in the +.Fa arg +argument. +The +.Fa fd +indicates the file descriptor that should be monitored for events. +The events can be either +.Va EV_READ , +.Va EV_WRITE , +or both, +indicating that an application can read or write from the file descriptor +respectively without blocking. +.Pp +The function +.Fa fn +will be called with the file descriptor that triggered the event and +the type of event which will be either +.Va EV_TIMEOUT , +.Va EV_SIGNAL , +.Va EV_READ , +or +.Va EV_WRITE . +The additional flag +.Va EV_PERSIST +makes an +.Fn event_add +persistent until +.Fn event_del +has been called. +.Pp +Once initialized, the +.Fa ev +structure can be used repeatedly with +.Fn event_add +and +.Fn event_del +and does not need to be reinitialized unless the function called and/or +the argument to it are to be changed. +However, when an +.Fa ev +structure has been added to libevent using +.Fn event_add +the structure must persist until the event occurs (assuming +.Fa EV_PERSIST +is not set) or is removed +using +.Fn event_del . +You may not reuse the same +.Fa ev +structure for multiple monitored descriptors; each descriptor +needs its own +.Fa ev . +.Pp +The function +.Fn event_add +schedules the execution of the +.Fa ev +event when the event specified in +.Fn event_set +occurs or in at least the time specified in the +.Fa tv . +If +.Fa tv +is +.Dv NULL , +no timeout occurs and the function will only be called +if a matching event occurs on the file descriptor. +The event in the +.Fa ev +argument must be already initialized by +.Fn event_set +and may not be used in calls to +.Fn event_set +until it has timed out or been removed with +.Fn event_del . +If the event in the +.Fa ev +argument already has a scheduled timeout, the old timeout will be +replaced by the new one. +.Pp +The function +.Fn event_del +will cancel the event in the argument +.Fa ev . +If the event has already executed or has never been added +the call will have no effect. +.Pp +The function +.Fn event_once +is similar to +.Fn event_set . +However, it schedules a callback to be called exactly once and does not +require the caller to prepare an +.Fa event +structure. +This function supports +.Fa EV_TIMEOUT , +.Fa EV_READ , +and +.Fa EV_WRITE . +.Pp +The +.Fn event_pending +function can be used to check if the event specified by +.Fa event +is pending to run. +If +.Va EV_TIMEOUT +was specified and +.Fa tv +is not +.Dv NULL , +the expiration time of the event will be returned in +.Fa tv . +.Pp +The +.Fn event_initialized +macro can be used to check if an event has been initialized. +.Pp +The functions +.Fn evtimer_set , +.Fn evtimer_add , +.Fn evtimer_del , +.Fn evtimer_initialized , +and +.Fn evtimer_pending +are abbreviations for common situations where only a timeout is required. +The file descriptor passed will be \-1, and the event type will be +.Va EV_TIMEOUT . +.Pp +The functions +.Fn signal_set , +.Fn signal_add , +.Fn signal_del , +.Fn signal_initialized , +and +.Fn signal_pending +are abbreviations. +The event type will be a persistent +.Va EV_SIGNAL . +That means +.Fn signal_set +adds +.Va EV_PERSIST . +.Pp +It is possible to disable support for +.Va epoll , kqueue , devpoll , poll +or +.Va select +by setting the environment variable +.Va EVENT_NOEPOLL , EVENT_NOKQUEUE , EVENT_NODEVPOLL , EVENT_NOPOLL +or +.Va EVENT_NOSELECT , +respectively. +By setting the environment variable +.Va EVENT_SHOW_METHOD , +.Nm libevent +displays the kernel notification method that it uses. +.Sh EVENT PRIORITIES +By default +.Nm libevent +schedules all active events with the same priority. +However, sometimes it is desirable to process some events with a higher +priority than others. +For that reason, +.Nm libevent +supports strict priority queues. +Active events with a lower priority are always processed before events +with a higher priority. +.Pp +The number of different priorities can be set initially with the +.Fn event_priority_init +function. +This function should be called before the first call to +.Fn event_dispatch . +The +.Fn event_priority_set +function can be used to assign a priority to an event. +By default, +.Nm libevent +assigns the middle priority to all events unless their priority +is explicitly set. +.Sh THREAD SAFE EVENTS +.Nm Libevent +has experimental support for thread-safe events. +When initializing the library via +.Fn event_init , +an event base is returned. +This event base can be used in conjunction with calls to +.Fn event_base_set , +.Fn event_base_dispatch , +.Fn event_base_loop , +.Fn event_base_loopexit , +.Fn bufferevent_base_set +and +.Fn event_base_free . +.Fn event_base_set +should be called after preparing an event with +.Fn event_set , +as +.Fn event_set +assigns the provided event to the most recently created event base. +.Fn bufferevent_base_set +should be called after preparing a bufferevent with +.Fn bufferevent_new . +.Fn event_base_free +should be used to free memory associated with the event base +when it is no longer needed. +.Sh BUFFERED EVENTS +.Nm libevent +provides an abstraction on top of the regular event callbacks. +This abstraction is called a +.Va "buffered event" . +A buffered event provides input and output buffers that get filled +and drained automatically. +The user of a buffered event no longer deals directly with the IO, +but instead is reading from input and writing to output buffers. +.Pp +A new bufferevent is created by +.Fn bufferevent_new . +The parameter +.Fa fd +specifies the file descriptor from which data is read and written to. +This file descriptor is not allowed to be a +.Xr pipe 2 . +The next three parameters are callbacks. +The read and write callback have the following form: +.Ft void +.Fn "(*cb)" "struct bufferevent *bufev" "void *arg" . +The error callback has the following form: +.Ft void +.Fn "(*cb)" "struct bufferevent *bufev" "short what" "void *arg" . +The argument is specified by the fourth parameter +.Fa "cbarg" . +A +.Fa bufferevent struct +pointer is returned on success, NULL on error. +Both the read and the write callback may be NULL. +The error callback has to be always provided. +.Pp +Once initialized, the bufferevent structure can be used repeatedly with +bufferevent_enable() and bufferevent_disable(). +The flags parameter can be a combination of +.Va EV_READ +and +.Va EV_WRITE . +When read enabled the bufferevent will try to read from the file +descriptor and call the read callback. +The write callback is executed +whenever the output buffer is drained below the write low watermark, +which is +.Va 0 +by default. +.Pp +The +.Fn bufferevent_write +function can be used to write data to the file descriptor. +The data is appended to the output buffer and written to the descriptor +automatically as it becomes available for writing. +.Fn bufferevent_write +returns 0 on success or \-1 on failure. +The +.Fn bufferevent_read +function is used to read data from the input buffer, +returning the amount of data read. +.Pp +If multiple bases are in use, bufferevent_base_set() must be called before +enabling the bufferevent for the first time. +.Sh NON-BLOCKING HTTP SUPPORT +.Nm libevent +provides a very thin HTTP layer that can be used both to host an HTTP +server and also to make HTTP requests. +An HTTP server can be created by calling +.Fn evhttp_start . +When the HTTP server is no longer used, it can be freed via +.Fn evhttp_free . +.Pp +To be notified of HTTP requests, a user needs to register callbacks with the +HTTP server. +This can be done by calling +.Fn evhttp_set_cb . +The second argument is the URI for which a callback is being registered. +The corresponding callback will receive an +.Va struct evhttp_request +object that contains all information about the request. +.Pp +This section does not document all the possible function calls; please +check +.Va event.h +for the public interfaces. +.Sh RETURN VALUES +Upon successful completion +.Fn event_add +and +.Fn event_del +return 0. +Otherwise, \-1 is returned and the global variable errno is +set to indicate the error. +.Sh SEE ALSO +.Xr kqueue 2 , +.Xr poll 2 , +.Xr select 2 , +.Xr evdns 3 , +.Xr timeout 9 +.Sh HISTORY +The +.Nm event +API manpage is based on the +.Xr timeout 9 +manpage by Artur Grabowski. +The port of +.Nm libevent +to Windows is due to Michael A. Davis. +Support for real-time signals is due to Taral. +.Sh AUTHORS +The +.Nm event +library was written by Niels Provos. +.Sh BUGS +This documentation is neither complete nor authoritative. +If you are in doubt about the usage of this API then +check the source code to find out how it works, write +up the missing piece of documentation and send it to +me for inclusion in this man page. diff --git a/contrib/libevent/event.c b/contrib/libevent/event.c new file mode 100644 index 0000000000..15bf14cf3b --- /dev/null +++ b/contrib/libevent/event.c @@ -0,0 +1,928 @@ +/* + * Copyright (c) 2000-2004 Niels Provos + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#ifdef WIN32 +#define WIN32_LEAN_AND_MEAN +#include +#undef WIN32_LEAN_AND_MEAN +#include "misc.h" +#endif +#include +#include +#ifdef HAVE_SYS_TIME_H +#include +#else +#include +#endif +#include +#include +#include +#ifndef WIN32 +#include +#endif +#include +#include +#include +#include +#include + +#include "event.h" +#include "event-internal.h" +#include "log.h" + +#ifdef HAVE_EVENT_PORTS +extern const struct eventop evportops; +#endif +#ifdef HAVE_SELECT +extern const struct eventop selectops; +#endif +#ifdef HAVE_POLL +extern const struct eventop pollops; +#endif +#ifdef HAVE_RTSIG +extern const struct eventop rtsigops; +#endif +#ifdef HAVE_EPOLL +extern const struct eventop epollops; +#endif +#ifdef HAVE_WORKING_KQUEUE +extern const struct eventop kqops; +#endif +#ifdef HAVE_DEVPOLL +extern const struct eventop devpollops; +#endif +#ifdef WIN32 +extern const struct eventop win32ops; +#endif + +/* In order of preference */ +const struct eventop *eventops[] = { +#ifdef HAVE_EVENT_PORTS + &evportops, +#endif +#ifdef HAVE_WORKING_KQUEUE + &kqops, +#endif +#ifdef HAVE_EPOLL + &epollops, +#endif +#ifdef HAVE_DEVPOLL + &devpollops, +#endif +#ifdef HAVE_RTSIG + &rtsigops, +#endif +#ifdef HAVE_POLL + &pollops, +#endif +#ifdef HAVE_SELECT + &selectops, +#endif +#ifdef WIN32 + &win32ops, +#endif + NULL +}; + +/* Global state */ +struct event_base *current_base = NULL; +extern struct event_base *evsignal_base; +static int use_monotonic; + +/* Handle signals - This is a deprecated interface */ +int (*event_sigcb)(void); /* Signal callback when gotsig is set */ +volatile sig_atomic_t event_gotsig; /* Set in signal handler */ + +/* Prototypes */ +static void event_queue_insert(struct event_base *, struct event *, int); +static void event_queue_remove(struct event_base *, struct event *, int); +static int event_haveevents(struct event_base *); + +static void event_process_active(struct event_base *); + +static int timeout_next(struct event_base *, struct timeval **); +static void timeout_process(struct event_base *); +static void timeout_correct(struct event_base *, struct timeval *); + +static int +compare(struct event *a, struct event *b) +{ + if (timercmp(&a->ev_timeout, &b->ev_timeout, <)) + return (-1); + else if (timercmp(&a->ev_timeout, &b->ev_timeout, >)) + return (1); + if (a < b) + return (-1); + else if (a > b) + return (1); + return (0); +} + +static void +detect_monotonic(void) +{ +#if defined(HAVE_CLOCK_GETTIME) && defined(CLOCK_MONOTONIC) + struct timespec ts; + + if (clock_gettime(CLOCK_MONOTONIC, &ts) == 0) + use_monotonic = 1; +#endif +} + +static int +gettime(struct timeval *tp) +{ +#if defined(HAVE_CLOCK_GETTIME) && defined(CLOCK_MONOTONIC) + struct timespec ts; + + if (use_monotonic) { + if (clock_gettime(CLOCK_MONOTONIC, &ts) == -1) + return (-1); + + tp->tv_sec = ts.tv_sec; + tp->tv_usec = ts.tv_nsec / 1000; + return (0); + } +#endif + + return (gettimeofday(tp, NULL)); +} + +RB_PROTOTYPE(event_tree, event, ev_timeout_node, compare); + +RB_GENERATE(event_tree, event, ev_timeout_node, compare); + + +void * +event_init(void) +{ + int i; + struct event_base *base; + + if ((base = calloc(1, sizeof(struct event_base))) == NULL) + event_err(1, "%s: calloc"); + + event_sigcb = NULL; + event_gotsig = 0; + + detect_monotonic(); + gettime(&base->event_tv); + + RB_INIT(&base->timetree); + TAILQ_INIT(&base->eventqueue); + TAILQ_INIT(&base->sig.signalqueue); + base->sig.ev_signal_pair[0] = -1; + base->sig.ev_signal_pair[1] = -1; + + base->evbase = NULL; + for (i = 0; eventops[i] && !base->evbase; i++) { + base->evsel = eventops[i]; + + base->evbase = base->evsel->init(base); + } + + if (base->evbase == NULL) + event_errx(1, "%s: no event mechanism available", __func__); + + if (getenv("EVENT_SHOW_METHOD")) + event_msgx("libevent using: %s\n", + base->evsel->name); + + /* allocate a single active event queue */ + event_base_priority_init(base, 1); + + current_base = base; + return (base); +} + +void +event_base_free(struct event_base *base) +{ + int i; + + if (base == NULL && current_base) + base = current_base; + if (base == current_base) + current_base = NULL; + + assert(base); + if (base->evsel->dealloc != NULL) + base->evsel->dealloc(base, base->evbase); + for (i=0; i < base->nactivequeues; ++i) + assert(TAILQ_EMPTY(base->activequeues[i])); + + assert(RB_EMPTY(&base->timetree)); + + for (i = 0; i < base->nactivequeues; ++i) + free(base->activequeues[i]); + free(base->activequeues); + + assert(TAILQ_EMPTY(&base->eventqueue)); + + free(base); +} + +int +event_priority_init(int npriorities) +{ + return event_base_priority_init(current_base, npriorities); +} + +int +event_base_priority_init(struct event_base *base, int npriorities) +{ + int i; + + if (base->event_count_active) + return (-1); + + if (base->nactivequeues && npriorities != base->nactivequeues) { + for (i = 0; i < base->nactivequeues; ++i) { + free(base->activequeues[i]); + } + free(base->activequeues); + } + + /* Allocate our priority queues */ + base->nactivequeues = npriorities; + base->activequeues = (struct event_list **)calloc(base->nactivequeues, + npriorities * sizeof(struct event_list *)); + if (base->activequeues == NULL) + event_err(1, "%s: calloc", __func__); + + for (i = 0; i < base->nactivequeues; ++i) { + base->activequeues[i] = malloc(sizeof(struct event_list)); + if (base->activequeues[i] == NULL) + event_err(1, "%s: malloc", __func__); + TAILQ_INIT(base->activequeues[i]); + } + + return (0); +} + +int +event_haveevents(struct event_base *base) +{ + return (base->event_count > 0); +} + +/* + * Active events are stored in priority queues. Lower priorities are always + * process before higher priorities. Low priority events can starve high + * priority ones. + */ + +static void +event_process_active(struct event_base *base) +{ + struct event *ev; + struct event_list *activeq = NULL; + int i; + short ncalls; + + if (!base->event_count_active) + return; + + for (i = 0; i < base->nactivequeues; ++i) { + if (TAILQ_FIRST(base->activequeues[i]) != NULL) { + activeq = base->activequeues[i]; + break; + } + } + + assert(activeq != NULL); + + for (ev = TAILQ_FIRST(activeq); ev; ev = TAILQ_FIRST(activeq)) { + event_queue_remove(base, ev, EVLIST_ACTIVE); + + /* Allows deletes to work */ + ncalls = ev->ev_ncalls; + ev->ev_pncalls = &ncalls; + while (ncalls) { + ncalls--; + ev->ev_ncalls = ncalls; + (*ev->ev_callback)((int)ev->ev_fd, ev->ev_res, ev->ev_arg); + if (event_gotsig) + return; + } + } +} + +/* + * Wait continously for events. We exit only if no events are left. + */ + +int +event_dispatch(void) +{ + return (event_loop(0)); +} + +int +event_base_dispatch(struct event_base *event_base) +{ + return (event_base_loop(event_base, 0)); +} + +static void +event_loopexit_cb(int fd, short what, void *arg) +{ + struct event_base *base = arg; + base->event_gotterm = 1; +} + +/* not thread safe */ +int +event_loopexit(struct timeval *tv) +{ + return (event_once(-1, EV_TIMEOUT, event_loopexit_cb, + current_base, tv)); +} + +int +event_base_loopexit(struct event_base *event_base, struct timeval *tv) +{ + return (event_base_once(event_base, -1, EV_TIMEOUT, event_loopexit_cb, + event_base, tv)); +} + +/* not thread safe */ + +int +event_loop(int flags) +{ + return event_base_loop(current_base, flags); +} + +int +event_base_loop(struct event_base *base, int flags) +{ + const struct eventop *evsel = base->evsel; + void *evbase = base->evbase; + struct timeval tv; + struct timeval *tv_p; + int res, done; + +#ifndef WIN32 + if(!TAILQ_EMPTY(&base->sig.signalqueue)) + evsignal_base = base; +#endif + done = 0; + while (!done) { + /* Calculate the initial events that we are waiting for */ + if (evsel->recalc(base, evbase, 0) == -1) + return (-1); + + /* Terminate the loop if we have been asked to */ + if (base->event_gotterm) { + base->event_gotterm = 0; + break; + } + + /* You cannot use this interface for multi-threaded apps */ + while (event_gotsig) { + event_gotsig = 0; + if (event_sigcb) { + res = (*event_sigcb)(); + if (res == -1) { + errno = EINTR; + return (-1); + } + } + } + + timeout_correct(base, &tv); + + tv_p = &tv; + if (!base->event_count_active && !(flags & EVLOOP_NONBLOCK)) { + timeout_next(base, &tv_p); + } else { + /* + * if we have active events, we just poll new events + * without waiting. + */ + timerclear(&tv); + } + + /* If we have no events, we just exit */ + if (!event_haveevents(base)) { + event_debug(("%s: no events registered.", __func__)); + return (1); + } + + res = evsel->dispatch(base, evbase, tv_p); + + + if (res == -1) + return (-1); + + timeout_process(base); + + if (base->event_count_active) { + event_process_active(base); + if (!base->event_count_active && (flags & EVLOOP_ONCE)) + done = 1; + } else if (flags & EVLOOP_NONBLOCK) + done = 1; + } + + event_debug(("%s: asked to terminate loop.", __func__)); + return (0); +} + +/* Sets up an event for processing once */ + +struct event_once { + struct event ev; + + void (*cb)(int, short, void *); + void *arg; +}; + +/* One-time callback, it deletes itself */ + +static void +event_once_cb(int fd, short events, void *arg) +{ + struct event_once *eonce = arg; + + (*eonce->cb)(fd, events, eonce->arg); + free(eonce); +} + +/* not threadsafe, event scheduled once. */ +int +event_once(int fd, short events, + void (*callback)(int, short, void *), void *arg, struct timeval *tv) +{ + return event_base_once(current_base, fd, events, callback, arg, tv); +} + +/* Schedules an event once */ +int +event_base_once(struct event_base *base, int fd, short events, + void (*callback)(int, short, void *), void *arg, struct timeval *tv) +{ + struct event_once *eonce; + struct timeval etv; + int res; + + /* We cannot support signals that just fire once */ + if (events & EV_SIGNAL) + return (-1); + + if ((eonce = calloc(1, sizeof(struct event_once))) == NULL) + return (-1); + + eonce->cb = callback; + eonce->arg = arg; + + if (events == EV_TIMEOUT) { + if (tv == NULL) { + timerclear(&etv); + tv = &etv; + } + + evtimer_set(&eonce->ev, event_once_cb, eonce); + } else if (events & (EV_READ|EV_WRITE)) { + events &= EV_READ|EV_WRITE; + + event_set(&eonce->ev, fd, events, event_once_cb, eonce); + } else { + /* Bad event combination */ + free(eonce); + return (-1); + } + + res = event_base_set(base, &eonce->ev); + if (res == 0) + res = event_add(&eonce->ev, tv); + if (res != 0) { + free(eonce); + return (res); + } + + return (0); +} + +void +event_set(struct event *ev, int fd, short events, + void (*callback)(int, short, void *), void *arg) +{ + /* Take the current base - caller needs to set the real base later */ + ev->ev_base = current_base; + + ev->ev_callback = callback; + ev->ev_arg = arg; + ev->ev_fd = fd; + ev->ev_events = events; + ev->ev_res = 0; + ev->ev_flags = EVLIST_INIT; + ev->ev_ncalls = 0; + ev->ev_pncalls = NULL; + + /* by default, we put new events into the middle priority */ + if(current_base) + ev->ev_pri = current_base->nactivequeues/2; +} + +int +event_base_set(struct event_base *base, struct event *ev) +{ + /* Only innocent events may be assigned to a different base */ + if (ev->ev_flags != EVLIST_INIT) + return (-1); + + ev->ev_base = base; + ev->ev_pri = base->nactivequeues/2; + + return (0); +} + +/* + * Set's the priority of an event - if an event is already scheduled + * changing the priority is going to fail. + */ + +int +event_priority_set(struct event *ev, int pri) +{ + if (ev->ev_flags & EVLIST_ACTIVE) + return (-1); + if (pri < 0 || pri >= ev->ev_base->nactivequeues) + return (-1); + + ev->ev_pri = pri; + + return (0); +} + +/* + * Checks if a specific event is pending or scheduled. + */ + +int +event_pending(struct event *ev, short event, struct timeval *tv) +{ + struct timeval now, res; + int flags = 0; + + if (ev->ev_flags & EVLIST_INSERTED) + flags |= (ev->ev_events & (EV_READ|EV_WRITE)); + if (ev->ev_flags & EVLIST_ACTIVE) + flags |= ev->ev_res; + if (ev->ev_flags & EVLIST_TIMEOUT) + flags |= EV_TIMEOUT; + if (ev->ev_flags & EVLIST_SIGNAL) + flags |= EV_SIGNAL; + + event &= (EV_TIMEOUT|EV_READ|EV_WRITE|EV_SIGNAL); + + /* See if there is a timeout that we should report */ + if (tv != NULL && (flags & event & EV_TIMEOUT)) { + gettime(&now); + timersub(&ev->ev_timeout, &now, &res); + /* correctly remap to real time */ + gettimeofday(&now, NULL); + timeradd(&now, &res, tv); + } + + return (flags & event); +} + +int +event_add(struct event *ev, struct timeval *tv) +{ + struct event_base *base = ev->ev_base; + const struct eventop *evsel = base->evsel; + void *evbase = base->evbase; + + event_debug(( + "event_add: event: %p, %s%s%scall %p", + ev, + ev->ev_events & EV_READ ? "EV_READ " : " ", + ev->ev_events & EV_WRITE ? "EV_WRITE " : " ", + tv ? "EV_TIMEOUT " : " ", + ev->ev_callback)); + + assert(!(ev->ev_flags & ~EVLIST_ALL)); + + if (tv != NULL) { + struct timeval now; + + if (ev->ev_flags & EVLIST_TIMEOUT) + event_queue_remove(base, ev, EVLIST_TIMEOUT); + + /* Check if it is active due to a timeout. Rescheduling + * this timeout before the callback can be executed + * removes it from the active list. */ + if ((ev->ev_flags & EVLIST_ACTIVE) && + (ev->ev_res & EV_TIMEOUT)) { + /* See if we are just active executing this + * event in a loop + */ + if (ev->ev_ncalls && ev->ev_pncalls) { + /* Abort loop */ + *ev->ev_pncalls = 0; + } + + event_queue_remove(base, ev, EVLIST_ACTIVE); + } + + gettime(&now); + timeradd(&now, tv, &ev->ev_timeout); + + event_debug(( + "event_add: timeout in %d seconds, call %p", + tv->tv_sec, ev->ev_callback)); + + event_queue_insert(base, ev, EVLIST_TIMEOUT); + } + + if ((ev->ev_events & (EV_READ|EV_WRITE)) && + !(ev->ev_flags & (EVLIST_INSERTED|EVLIST_ACTIVE))) { + event_queue_insert(base, ev, EVLIST_INSERTED); + + return (evsel->add(evbase, ev)); + } else if ((ev->ev_events & EV_SIGNAL) && + !(ev->ev_flags & EVLIST_SIGNAL)) { + event_queue_insert(base, ev, EVLIST_SIGNAL); + + return (evsel->add(evbase, ev)); + } + + return (0); +} + +int +event_del(struct event *ev) +{ + struct event_base *base; + const struct eventop *evsel; + void *evbase; + + event_debug(("event_del: %p, callback %p", + ev, ev->ev_callback)); + + /* An event without a base has not been added */ + if (ev->ev_base == NULL) + return (-1); + + base = ev->ev_base; + evsel = base->evsel; + evbase = base->evbase; + + assert(!(ev->ev_flags & ~EVLIST_ALL)); + + /* See if we are just active executing this event in a loop */ + if (ev->ev_ncalls && ev->ev_pncalls) { + /* Abort loop */ + *ev->ev_pncalls = 0; + } + + if (ev->ev_flags & EVLIST_TIMEOUT) + event_queue_remove(base, ev, EVLIST_TIMEOUT); + + if (ev->ev_flags & EVLIST_ACTIVE) + event_queue_remove(base, ev, EVLIST_ACTIVE); + + if (ev->ev_flags & EVLIST_INSERTED) { + event_queue_remove(base, ev, EVLIST_INSERTED); + return (evsel->del(evbase, ev)); + } else if (ev->ev_flags & EVLIST_SIGNAL) { + event_queue_remove(base, ev, EVLIST_SIGNAL); + return (evsel->del(evbase, ev)); + } + + return (0); +} + +void +event_active(struct event *ev, int res, short ncalls) +{ + /* We get different kinds of events, add them together */ + if (ev->ev_flags & EVLIST_ACTIVE) { + ev->ev_res |= res; + return; + } + + ev->ev_res = res; + ev->ev_ncalls = ncalls; + ev->ev_pncalls = NULL; + event_queue_insert(ev->ev_base, ev, EVLIST_ACTIVE); +} + +static int +timeout_next(struct event_base *base, struct timeval **tv_p) +{ + struct timeval now; + struct event *ev; + struct timeval *tv = *tv_p; + + if ((ev = RB_MIN(event_tree, &base->timetree)) == NULL) { + /* if no time-based events are active wait for I/O */ + *tv_p = NULL; + return (0); + } + + if (gettime(&now) == -1) + return (-1); + + if (timercmp(&ev->ev_timeout, &now, <=)) { + timerclear(tv); + return (0); + } + + timersub(&ev->ev_timeout, &now, tv); + + assert(tv->tv_sec >= 0); + assert(tv->tv_usec >= 0); + + event_debug(("timeout_next: in %d seconds", tv->tv_sec)); + return (0); +} + +/* + * Determines if the time is running backwards by comparing the current + * time against the last time we checked. Not needed when using clock + * monotonic. + */ + +static void +timeout_correct(struct event_base *base, struct timeval *tv) +{ + struct event *ev; + struct timeval off; + + if (use_monotonic) + return; + + /* Check if time is running backwards */ + gettime(tv); + if (timercmp(tv, &base->event_tv, >=)) { + base->event_tv = *tv; + return; + } + + event_debug(("%s: time is running backwards, corrected", + __func__)); + timersub(&base->event_tv, tv, &off); + + /* + * We can modify the key element of the node without destroying + * the key, beause we apply it to all in the right order. + */ + RB_FOREACH(ev, event_tree, &base->timetree) + timersub(&ev->ev_timeout, &off, &ev->ev_timeout); +} + +void +timeout_process(struct event_base *base) +{ + struct timeval now; + struct event *ev, *next; + + gettime(&now); + + for (ev = RB_MIN(event_tree, &base->timetree); ev; ev = next) { + if (timercmp(&ev->ev_timeout, &now, >)) + break; + next = RB_NEXT(event_tree, &base->timetree, ev); + + event_queue_remove(base, ev, EVLIST_TIMEOUT); + + /* delete this event from the I/O queues */ + event_del(ev); + + event_debug(("timeout_process: call %p", + ev->ev_callback)); + event_active(ev, EV_TIMEOUT, 1); + } +} + +void +event_queue_remove(struct event_base *base, struct event *ev, int queue) +{ + int docount = 1; + + if (!(ev->ev_flags & queue)) + event_errx(1, "%s: %p(fd %d) not on queue %x", __func__, + ev, ev->ev_fd, queue); + + if (ev->ev_flags & EVLIST_INTERNAL) + docount = 0; + + if (docount) + base->event_count--; + + ev->ev_flags &= ~queue; + switch (queue) { + case EVLIST_ACTIVE: + if (docount) + base->event_count_active--; + TAILQ_REMOVE(base->activequeues[ev->ev_pri], + ev, ev_active_next); + break; + case EVLIST_SIGNAL: + TAILQ_REMOVE(&base->sig.signalqueue, ev, ev_signal_next); + break; + case EVLIST_TIMEOUT: + RB_REMOVE(event_tree, &base->timetree, ev); + break; + case EVLIST_INSERTED: + TAILQ_REMOVE(&base->eventqueue, ev, ev_next); + break; + default: + event_errx(1, "%s: unknown queue %x", __func__, queue); + } +} + +void +event_queue_insert(struct event_base *base, struct event *ev, int queue) +{ + int docount = 1; + + if (ev->ev_flags & queue) { + /* Double insertion is possible for active events */ + if (queue & EVLIST_ACTIVE) + return; + + event_errx(1, "%s: %p(fd %d) already on queue %x", __func__, + ev, ev->ev_fd, queue); + } + + if (ev->ev_flags & EVLIST_INTERNAL) + docount = 0; + + if (docount) + base->event_count++; + + ev->ev_flags |= queue; + switch (queue) { + case EVLIST_ACTIVE: + if (docount) + base->event_count_active++; + TAILQ_INSERT_TAIL(base->activequeues[ev->ev_pri], + ev,ev_active_next); + break; + case EVLIST_SIGNAL: + TAILQ_INSERT_TAIL(&base->sig.signalqueue, ev, ev_signal_next); + break; + case EVLIST_TIMEOUT: { + struct event *tmp = RB_INSERT(event_tree, &base->timetree, ev); + assert(tmp == NULL); + break; + } + case EVLIST_INSERTED: + TAILQ_INSERT_TAIL(&base->eventqueue, ev, ev_next); + break; + default: + event_errx(1, "%s: unknown queue %x", __func__, queue); + } +} + +/* Functions for debugging */ + +const char * +event_get_version(void) +{ + return (VERSION); +} + +/* + * No thread-safe interface needed - the information should be the same + * for all threads. + */ + +const char * +event_get_method(void) +{ + return (current_base->evsel->name); +} diff --git a/contrib/libevent/event.h b/contrib/libevent/event.h new file mode 100644 index 0000000000..4c39939cc5 --- /dev/null +++ b/contrib/libevent/event.h @@ -0,0 +1,340 @@ +/* + * Copyright (c) 2000-2004 Niels Provos + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#ifndef _EVENT_H_ +#define _EVENT_H_ + +#ifdef __cplusplus +extern "C" { +#endif + +#include +#include +#include + +#ifdef WIN32 +#define WIN32_LEAN_AND_MEAN +#include +#undef WIN32_LEAN_AND_MEAN +typedef unsigned char u_char; +typedef unsigned short u_short; +#endif + +#define EVLIST_TIMEOUT 0x01 +#define EVLIST_INSERTED 0x02 +#define EVLIST_SIGNAL 0x04 +#define EVLIST_ACTIVE 0x08 +#define EVLIST_INTERNAL 0x10 +#define EVLIST_INIT 0x80 + +/* EVLIST_X_ Private space: 0x1000-0xf000 */ +#define EVLIST_ALL (0xf000 | 0x9f) + +#define EV_TIMEOUT 0x01 +#define EV_READ 0x02 +#define EV_WRITE 0x04 +#define EV_SIGNAL 0x08 +#define EV_PERSIST 0x10 /* Persistant event */ + +/* Fix so that ppl dont have to run with */ +#ifndef TAILQ_ENTRY +#define _EVENT_DEFINED_TQENTRY +#define TAILQ_ENTRY(type) \ +struct { \ + struct type *tqe_next; /* next element */ \ + struct type **tqe_prev; /* address of previous next element */ \ +} +#endif /* !TAILQ_ENTRY */ +#ifndef RB_ENTRY +#define _EVENT_DEFINED_RBENTRY +#define RB_ENTRY(type) \ +struct { \ + struct type *rbe_left; /* left element */ \ + struct type *rbe_right; /* right element */ \ + struct type *rbe_parent; /* parent element */ \ + int rbe_color; /* node color */ \ +} +#endif /* !RB_ENTRY */ + +struct event_base; +struct event { + TAILQ_ENTRY (event) ev_next; + TAILQ_ENTRY (event) ev_active_next; + TAILQ_ENTRY (event) ev_signal_next; + RB_ENTRY (event) ev_timeout_node; + + struct event_base *ev_base; + int ev_fd; + short ev_events; + short ev_ncalls; + short *ev_pncalls; /* Allows deletes in callback */ + + struct timeval ev_timeout; + + int ev_pri; /* smaller numbers are higher priority */ + + void (*ev_callback)(int, short, void *arg); + void *ev_arg; + + int ev_res; /* result passed to event callback */ + int ev_flags; +}; + +#define EVENT_SIGNAL(ev) (int)(ev)->ev_fd +#define EVENT_FD(ev) (int)(ev)->ev_fd + +/* + * Key-Value pairs. Can be used for HTTP headers but also for + * query argument parsing. + */ +struct evkeyval { + TAILQ_ENTRY(evkeyval) next; + + char *key; + char *value; +}; + +#ifdef _EVENT_DEFINED_TQENTRY +#undef TAILQ_ENTRY +struct event_list; +struct evkeyvalq; +#undef _EVENT_DEFINED_TQENTRY +#else +TAILQ_HEAD (event_list, event); +TAILQ_HEAD (evkeyvalq, evkeyval); +#endif /* _EVENT_DEFINED_TQENTRY */ +#ifdef _EVENT_DEFINED_RBENTRY +#undef RB_ENTRY +#undef _EVENT_DEFINED_RBENTRY +#endif /* _EVENT_DEFINED_RBENTRY */ + +struct eventop { + char *name; + void *(*init)(struct event_base *); + int (*add)(void *, struct event *); + int (*del)(void *, struct event *); + int (*recalc)(struct event_base *, void *, int); + int (*dispatch)(struct event_base *, void *, struct timeval *); + void (*dealloc)(struct event_base *, void *); +}; + +void *event_init(void); +int event_dispatch(void); +int event_base_dispatch(struct event_base *); +void event_base_free(struct event_base *); + +#define _EVENT_LOG_DEBUG 0 +#define _EVENT_LOG_MSG 1 +#define _EVENT_LOG_WARN 2 +#define _EVENT_LOG_ERR 3 +typedef void (*event_log_cb)(int severity, const char *msg); +void event_set_log_callback(event_log_cb cb); + +/* Associate a different event base with an event */ +int event_base_set(struct event_base *, struct event *); + +#define EVLOOP_ONCE 0x01 +#define EVLOOP_NONBLOCK 0x02 +int event_loop(int); +int event_base_loop(struct event_base *, int); +int event_loopexit(struct timeval *); /* Causes the loop to exit */ +int event_base_loopexit(struct event_base *, struct timeval *); + +#define evtimer_add(ev, tv) event_add(ev, tv) +#define evtimer_set(ev, cb, arg) event_set(ev, -1, 0, cb, arg) +#define evtimer_del(ev) event_del(ev) +#define evtimer_pending(ev, tv) event_pending(ev, EV_TIMEOUT, tv) +#define evtimer_initialized(ev) ((ev)->ev_flags & EVLIST_INIT) + +#define timeout_add(ev, tv) event_add(ev, tv) +#define timeout_set(ev, cb, arg) event_set(ev, -1, 0, cb, arg) +#define timeout_del(ev) event_del(ev) +#define timeout_pending(ev, tv) event_pending(ev, EV_TIMEOUT, tv) +#define timeout_initialized(ev) ((ev)->ev_flags & EVLIST_INIT) + +#define signal_add(ev, tv) event_add(ev, tv) +#define signal_set(ev, x, cb, arg) \ + event_set(ev, x, EV_SIGNAL|EV_PERSIST, cb, arg) +#define signal_del(ev) event_del(ev) +#define signal_pending(ev, tv) event_pending(ev, EV_SIGNAL, tv) +#define signal_initialized(ev) ((ev)->ev_flags & EVLIST_INIT) + +void event_set(struct event *, int, short, void (*)(int, short, void *), void *); +int event_once(int, short, void (*)(int, short, void *), void *, struct timeval *); +int event_base_once(struct event_base *, int, short, void (*)(int, short, void *), void *, struct timeval *); + +int event_add(struct event *, struct timeval *); +int event_del(struct event *); +void event_active(struct event *, int, short); + +int event_pending(struct event *, short, struct timeval *); + +#ifdef WIN32 +#define event_initialized(ev) ((ev)->ev_flags & EVLIST_INIT && (ev)->ev_fd != (int)INVALID_HANDLE_VALUE) +#else +#define event_initialized(ev) ((ev)->ev_flags & EVLIST_INIT) +#endif + +/* Some simple debugging functions */ +const char *event_get_version(void); +const char *event_get_method(void); + +/* These functions deal with event priorities */ + +int event_priority_init(int); +int event_base_priority_init(struct event_base *, int); +int event_priority_set(struct event *, int); + +/* These functions deal with buffering input and output */ + +struct evbuffer { + u_char *buffer; + u_char *orig_buffer; + + size_t misalign; + size_t totallen; + size_t off; + + void (*cb)(struct evbuffer *, size_t, size_t, void *); + void *cbarg; +}; + +/* Just for error reporting - use other constants otherwise */ +#define EVBUFFER_READ 0x01 +#define EVBUFFER_WRITE 0x02 +#define EVBUFFER_EOF 0x10 +#define EVBUFFER_ERROR 0x20 +#define EVBUFFER_TIMEOUT 0x40 + +struct bufferevent; +typedef void (*evbuffercb)(struct bufferevent *, void *); +typedef void (*everrorcb)(struct bufferevent *, short what, void *); + +struct event_watermark { + size_t low; + size_t high; +}; + +struct bufferevent { + struct event ev_read; + struct event ev_write; + + struct evbuffer *input; + struct evbuffer *output; + + struct event_watermark wm_read; + struct event_watermark wm_write; + + evbuffercb readcb; + evbuffercb writecb; + everrorcb errorcb; + void *cbarg; + + int timeout_read; /* in seconds */ + int timeout_write; /* in seconds */ + + short enabled; /* events that are currently enabled */ +}; + +struct bufferevent *bufferevent_new(int fd, + evbuffercb readcb, evbuffercb writecb, everrorcb errorcb, void *cbarg); +int bufferevent_base_set(struct event_base *base, struct bufferevent *bufev); +int bufferevent_priority_set(struct bufferevent *bufev, int pri); +void bufferevent_free(struct bufferevent *bufev); +int bufferevent_write(struct bufferevent *bufev, void *data, size_t size); +int bufferevent_write_buffer(struct bufferevent *bufev, struct evbuffer *buf); +size_t bufferevent_read(struct bufferevent *bufev, void *data, size_t size); +int bufferevent_enable(struct bufferevent *bufev, short event); +int bufferevent_disable(struct bufferevent *bufev, short event); +void bufferevent_settimeout(struct bufferevent *bufev, + int timeout_read, int timeout_write); + +#define EVBUFFER_LENGTH(x) (x)->off +#define EVBUFFER_DATA(x) (x)->buffer +#define EVBUFFER_INPUT(x) (x)->input +#define EVBUFFER_OUTPUT(x) (x)->output + +struct evbuffer *evbuffer_new(void); +void evbuffer_free(struct evbuffer *); +int evbuffer_expand(struct evbuffer *, size_t); +int evbuffer_add(struct evbuffer *, const void *, size_t); +int evbuffer_remove(struct evbuffer *, void *, size_t); +char *evbuffer_readline(struct evbuffer *); +int evbuffer_add_buffer(struct evbuffer *, struct evbuffer *); +int evbuffer_add_printf(struct evbuffer *, const char *fmt, ...); +int evbuffer_add_vprintf(struct evbuffer *, const char *fmt, va_list ap); +void evbuffer_drain(struct evbuffer *, size_t); +int evbuffer_write(struct evbuffer *, int); +int evbuffer_read(struct evbuffer *, int, int); +u_char *evbuffer_find(struct evbuffer *, const u_char *, size_t); +void evbuffer_setcb(struct evbuffer *, void (*)(struct evbuffer *, size_t, size_t, void *), void *); + +/* + * Marshaling tagged data - We assume that all tags are inserted in their + * numeric order - so that unknown tags will always be higher than the + * known ones - and we can just ignore the end of an event buffer. + */ + +void evtag_init(void); + +void evtag_marshal(struct evbuffer *evbuf, uint8_t tag, const void *data, + uint32_t len); + +void encode_int(struct evbuffer *evbuf, uint32_t number); + +void evtag_marshal_int(struct evbuffer *evbuf, uint8_t tag, uint32_t integer); + +void evtag_marshal_string(struct evbuffer *buf, uint8_t tag, + const char *string); + +void evtag_marshal_timeval(struct evbuffer *evbuf, uint8_t tag, + struct timeval *tv); + +void evtag_test(void); + +int evtag_unmarshal(struct evbuffer *src, uint8_t *ptag, struct evbuffer *dst); +int evtag_peek(struct evbuffer *evbuf, uint8_t *ptag); +int evtag_peek_length(struct evbuffer *evbuf, uint32_t *plength); +int evtag_payload_length(struct evbuffer *evbuf, uint32_t *plength); +int evtag_consume(struct evbuffer *evbuf); + +int evtag_unmarshal_int(struct evbuffer *evbuf, uint8_t need_tag, + uint32_t *pinteger); + +int evtag_unmarshal_fixed(struct evbuffer *src, uint8_t need_tag, void *data, + size_t len); + +int evtag_unmarshal_string(struct evbuffer *evbuf, uint8_t need_tag, + char **pstring); + +int evtag_unmarshal_timeval(struct evbuffer *evbuf, uint8_t need_tag, + struct timeval *ptv); + +#ifdef __cplusplus +} +#endif + +#endif /* _EVENT_H_ */ diff --git a/contrib/libevent/event_tagging.c b/contrib/libevent/event_tagging.c new file mode 100644 index 0000000000..3266d5435a --- /dev/null +++ b/contrib/libevent/event_tagging.c @@ -0,0 +1,372 @@ +/* + * Copyright (c) 2003, 2004 Niels Provos + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include +#include + +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#ifdef WIN32 +#define WIN32_LEAN_AND_MEAN +#include +#undef WIN32_LEAN_AND_MEAN +#else +#include +#endif + +#include +#include +#ifdef HAVE_SYS_TIME_H +#include +#endif + +#include +#include +#include +#include +#ifndef WIN32 +#include +#endif +#include + +#include "event.h" +#include "log.h" + +int decode_int(uint32_t *pnumber, struct evbuffer *evbuf); + +static struct evbuffer *_buf; /* not thread safe */ + +void +evtag_init() +{ + if (_buf != NULL) + return; + + if ((_buf = evbuffer_new()) == NULL) + event_err(1, "%s: malloc", __func__); +} + +/* + * We encode integer's by nibbles; the first nibble contains the number + * of significant nibbles - 1; this allows us to encode up to 64-bit + * integers. This function is byte-order independent. + */ + +void +encode_int(struct evbuffer *evbuf, uint32_t number) +{ + int off = 1, nibbles = 0; + uint8_t data[5]; + + memset(data, 0, sizeof(data)); + while (number) { + if (off & 0x1) + data[off/2] = (data[off/2] & 0xf0) | (number & 0x0f); + else + data[off/2] = (data[off/2] & 0x0f) | + ((number & 0x0f) << 4); + number >>= 4; + off++; + } + + if (off > 2) + nibbles = off - 2; + + /* Off - 1 is the number of encoded nibbles */ + data[0] = (data[0] & 0x0f) | ((nibbles & 0x0f) << 4); + + evbuffer_add(evbuf, data, (off + 1) / 2); +} + +/* + * Marshal a data type, the general format is as follows: + * + * tag number: one byte; length: var bytes; payload: var bytes + */ + +void +evtag_marshal(struct evbuffer *evbuf, uint8_t tag, + const void *data, uint32_t len) +{ + evbuffer_add(evbuf, &tag, sizeof(tag)); + encode_int(evbuf, len); + evbuffer_add(evbuf, (void *)data, len); +} + +/* Marshaling for integers */ +void +evtag_marshal_int(struct evbuffer *evbuf, uint8_t tag, uint32_t integer) +{ + evbuffer_drain(_buf, EVBUFFER_LENGTH(_buf)); + encode_int(_buf, integer); + + evbuffer_add(evbuf, &tag, sizeof(tag)); + encode_int(evbuf, EVBUFFER_LENGTH(_buf)); + evbuffer_add_buffer(evbuf, _buf); +} + +void +evtag_marshal_string(struct evbuffer *buf, uint8_t tag, const char *string) +{ + evtag_marshal(buf, tag, string, strlen(string)); +} + +void +evtag_marshal_timeval(struct evbuffer *evbuf, uint8_t tag, struct timeval *tv) +{ + evbuffer_drain(_buf, EVBUFFER_LENGTH(_buf)); + + encode_int(_buf, tv->tv_sec); + encode_int(_buf, tv->tv_usec); + + evtag_marshal(evbuf, tag, EVBUFFER_DATA(_buf), + EVBUFFER_LENGTH(_buf)); +} + +static int +decode_int_internal(uint32_t *pnumber, struct evbuffer *evbuf, int dodrain) +{ + uint32_t number = 0; + uint8_t *data = EVBUFFER_DATA(evbuf); + int len = EVBUFFER_LENGTH(evbuf); + int nibbles = 0, off; + + if (!len) + return (-1); + + nibbles = ((data[0] & 0xf0) >> 4) + 1; + if (nibbles > 8 || (nibbles >> 1) > len - 1) + return (-1); + + off = nibbles; + while (off > 0) { + number <<= 4; + if (off & 0x1) + number |= data[off >> 1] & 0x0f; + else + number |= (data[off >> 1] & 0xf0) >> 4; + off--; + } + + len = (nibbles >> 1) + 1; + if (dodrain) + evbuffer_drain(evbuf, len); + + *pnumber = number; + + return (len); +} + +int +decode_int(uint32_t *pnumber, struct evbuffer *evbuf) +{ + return (decode_int_internal(pnumber, evbuf, 1) == -1 ? -1 : 0); +} + +int +evtag_peek(struct evbuffer *evbuf, uint8_t *ptag) +{ + if (EVBUFFER_LENGTH(evbuf) < 2) + return (-1); + *ptag = EVBUFFER_DATA(evbuf)[0]; + + return (0); +} + +int +evtag_peek_length(struct evbuffer *evbuf, uint32_t *plength) +{ + struct evbuffer tmp; + int res; + + if (EVBUFFER_LENGTH(evbuf) < 2) + return (-1); + + tmp = *evbuf; + tmp.buffer += 1; + tmp.off -= 1; + + res = decode_int_internal(plength, &tmp, 0); + if (res == -1) + return (-1); + + *plength += res + 1; + + return (0); +} + +int +evtag_payload_length(struct evbuffer *evbuf, uint32_t *plength) +{ + struct evbuffer tmp; + int res; + + if (EVBUFFER_LENGTH(evbuf) < 2) + return (-1); + + tmp = *evbuf; + tmp.buffer += 1; + tmp.off -= 1; + + res = decode_int_internal(plength, &tmp, 0); + if (res == -1) + return (-1); + + return (0); +} + +int +evtag_consume(struct evbuffer *evbuf) +{ + uint32_t len; + evbuffer_drain(evbuf, 1); + if (decode_int(&len, evbuf) == -1) + return (-1); + evbuffer_drain(evbuf, len); + + return (0); +} + +/* Reads the data type from an event buffer */ + +int +evtag_unmarshal(struct evbuffer *src, uint8_t *ptag, struct evbuffer *dst) +{ + uint8_t tag; + uint32_t len; + uint32_t integer; + + if (evbuffer_remove(src, &tag, sizeof(tag)) != sizeof(tag)) + return (-1); + if (decode_int(&integer, src) == -1) + return (-1); + len = integer; + + if (EVBUFFER_LENGTH(src) < len) + return (-1); + + if (evbuffer_add(dst, EVBUFFER_DATA(src), len) == -1) + return (-1); + + evbuffer_drain(src, len); + + *ptag = tag; + return (len); +} + +/* Marshaling for integers */ + +int +evtag_unmarshal_int(struct evbuffer *evbuf, uint8_t need_tag, + uint32_t *pinteger) +{ + uint8_t tag; + uint32_t len; + uint32_t integer; + + if (evbuffer_remove(evbuf, &tag, sizeof(tag)) != sizeof(tag) || + tag != need_tag) + return (-1); + if (decode_int(&integer, evbuf) == -1) + return (-1); + len = integer; + + if (EVBUFFER_LENGTH(evbuf) < len) + return (-1); + + evbuffer_drain(_buf, EVBUFFER_LENGTH(_buf)); + if (evbuffer_add(_buf, EVBUFFER_DATA(evbuf), len) == -1) + return (-1); + + evbuffer_drain(evbuf, len); + + return (decode_int(pinteger, _buf)); +} + +/* Unmarshal a fixed length tag */ + +int +evtag_unmarshal_fixed(struct evbuffer *src, uint8_t need_tag, void *data, + size_t len) +{ + uint8_t tag; + + /* Initialize this event buffer so that we can read into it */ + evbuffer_drain(_buf, EVBUFFER_LENGTH(_buf)); + + /* Now unmarshal a tag and check that it matches the tag we want */ + if (evtag_unmarshal(src, &tag, _buf) == -1 || tag != need_tag) + return (-1); + + if (EVBUFFER_LENGTH(_buf) != len) + return (-1); + + memcpy(data, EVBUFFER_DATA(_buf), len); + return (0); +} + +int +evtag_unmarshal_string(struct evbuffer *evbuf, uint8_t need_tag, + char **pstring) +{ + uint8_t tag; + + evbuffer_drain(_buf, EVBUFFER_LENGTH(_buf)); + + if (evtag_unmarshal(evbuf, &tag, _buf) == -1 || tag != need_tag) + return (-1); + + *pstring = calloc(EVBUFFER_LENGTH(_buf) + 1, 1); + if (*pstring == NULL) + event_err(1, "%s: calloc", __func__); + evbuffer_remove(_buf, *pstring, EVBUFFER_LENGTH(_buf)); + + return (0); +} + +int +evtag_unmarshal_timeval(struct evbuffer *evbuf, uint8_t need_tag, + struct timeval *ptv) +{ + uint8_t tag; + uint32_t integer; + + evbuffer_drain(_buf, EVBUFFER_LENGTH(_buf)); + if (evtag_unmarshal(evbuf, &tag, _buf) == -1 || tag != need_tag) + return (-1); + + if (decode_int(&integer, _buf) == -1) + return (-1); + ptv->tv_sec = integer; + if (decode_int(&integer, _buf) == -1) + return (-1); + ptv->tv_usec = integer; + + return (0); +} diff --git a/contrib/libevent/evsignal.h b/contrib/libevent/evsignal.h new file mode 100644 index 0000000000..7efbcabec8 --- /dev/null +++ b/contrib/libevent/evsignal.h @@ -0,0 +1,44 @@ +/* + * Copyright 2000-2002 Niels Provos + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#ifndef _EVSIGNAL_H_ +#define _EVSIGNAL_H_ + +struct evsignal_info { + struct event_list signalqueue; + struct event ev_signal; + int ev_signal_pair[2]; + int ev_signal_added; + volatile sig_atomic_t evsignal_caught; + sig_atomic_t evsigcaught[NSIG]; +}; +void evsignal_init(struct event_base *); +void evsignal_process(struct event_base *); +int evsignal_add(struct event *); +int evsignal_del(struct event *); +void evsignal_dealloc(struct event_base *); + +#endif /* _EVSIGNAL_H_ */ diff --git a/contrib/libevent/kqueue.c b/contrib/libevent/kqueue.c new file mode 100644 index 0000000000..059d94a0b4 --- /dev/null +++ b/contrib/libevent/kqueue.c @@ -0,0 +1,419 @@ +/* $OpenBSD: kqueue.c,v 1.5 2002/07/10 14:41:31 art Exp $ */ + +/* + * Copyright 2000-2002 Niels Provos + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#include +#ifdef HAVE_SYS_TIME_H +#include +#else +#include +#endif +#include +#include +#include +#include +#include +#include +#include +#include +#ifdef HAVE_INTTYPES_H +#include +#endif + +/* Some platforms apparently define the udata field of struct kevent as + * ntptr_t, whereas others define it as void*. There doesn't seem to be an + * easy way to tell them apart via autoconf, so we need to use OS macros. */ +#if defined(HAVE_INTTYPES_H) && !defined(__OpenBSD__) && !defined(__FreeBSD__) && !defined(__darwin__) && !defined(__APPLE__) +#define PTR_TO_UDATA(x) ((intptr_t)(x)) +#else +#define PTR_TO_UDATA(x) (x) +#endif + +#include "event.h" +#include "log.h" + +#define EVLIST_X_KQINKERNEL 0x1000 + +#define NEVENT 64 + +struct kqop { + struct kevent *changes; + int nchanges; + struct kevent *events; + int nevents; + int kq; +}; + +void *kq_init (struct event_base *); +int kq_add (void *, struct event *); +int kq_del (void *, struct event *); +int kq_recalc (struct event_base *, void *, int); +int kq_dispatch (struct event_base *, void *, struct timeval *); +int kq_insert (struct kqop *, struct kevent *); +void kq_dealloc (struct event_base *, void *); + +const struct eventop kqops = { + "kqueue", + kq_init, + kq_add, + kq_del, + kq_recalc, + kq_dispatch, + kq_dealloc +}; + +void * +kq_init(struct event_base *base) +{ + int kq; + struct kqop *kqueueop; + + /* Disable kqueue when this environment variable is set */ + if (getenv("EVENT_NOKQUEUE")) + return (NULL); + + if (!(kqueueop = calloc(1, sizeof(struct kqop)))) + return (NULL); + + /* Initalize the kernel queue */ + + if ((kq = kqueue()) == -1) { + event_warn("kqueue"); + free (kqueueop); + return (NULL); + } + + kqueueop->kq = kq; + + /* Initalize fields */ + kqueueop->changes = malloc(NEVENT * sizeof(struct kevent)); + if (kqueueop->changes == NULL) { + free (kqueueop); + return (NULL); + } + kqueueop->events = malloc(NEVENT * sizeof(struct kevent)); + if (kqueueop->events == NULL) { + free (kqueueop->changes); + free (kqueueop); + return (NULL); + } + kqueueop->nevents = NEVENT; + + /* Check for Mac OS X kqueue bug. */ + kqueueop->changes[0].ident = -1; + kqueueop->changes[0].filter = EVFILT_READ; + kqueueop->changes[0].flags = EV_ADD; + /* + * If kqueue works, then kevent will succeed, and it will + * stick an error in events[0]. If kqueue is broken, then + * kevent will fail. + */ + if (kevent(kq, + kqueueop->changes, 1, kqueueop->events, NEVENT, NULL) != 1 || + kqueueop->events[0].ident != -1 || + kqueueop->events[0].flags != EV_ERROR) { + event_warn("%s: detected broken kqueue; not using.", __func__); + free(kqueueop->changes); + free(kqueueop->events); + free(kqueueop); + close(kq); + return (NULL); + } + + return (kqueueop); +} + +int +kq_recalc(struct event_base *base, void *arg, int max) +{ + return (0); +} + +int +kq_insert(struct kqop *kqop, struct kevent *kev) +{ + int nevents = kqop->nevents; + + if (kqop->nchanges == nevents) { + struct kevent *newchange; + struct kevent *newresult; + + nevents *= 2; + + newchange = realloc(kqop->changes, + nevents * sizeof(struct kevent)); + if (newchange == NULL) { + event_warn("%s: malloc", __func__); + return (-1); + } + kqop->changes = newchange; + + newresult = realloc(kqop->events, + nevents * sizeof(struct kevent)); + + /* + * If we fail, we don't have to worry about freeing, + * the next realloc will pick it up. + */ + if (newresult == NULL) { + event_warn("%s: malloc", __func__); + return (-1); + } + kqop->events = newresult; + + kqop->nevents = nevents; + } + + memcpy(&kqop->changes[kqop->nchanges++], kev, sizeof(struct kevent)); + + event_debug(("%s: fd %d %s%s", + __func__, kev->ident, + kev->filter == EVFILT_READ ? "EVFILT_READ" : "EVFILT_WRITE", + kev->flags == EV_DELETE ? " (del)" : "")); + + return (0); +} + +static void +kq_sighandler(int sig) +{ + /* Do nothing here */ +} + +int +kq_dispatch(struct event_base *base, void *arg, struct timeval *tv) +{ + struct kqop *kqop = arg; + struct kevent *changes = kqop->changes; + struct kevent *events = kqop->events; + struct event *ev; + struct timespec ts, *ts_p = NULL; + int i, res; + + if (tv != NULL) { + TIMEVAL_TO_TIMESPEC(tv, &ts); + ts_p = &ts; + } + + res = kevent(kqop->kq, changes, kqop->nchanges, + events, kqop->nevents, ts_p); + kqop->nchanges = 0; + if (res == -1) { + if (errno != EINTR) { + event_warn("kevent"); + return (-1); + } + + return (0); + } + + event_debug(("%s: kevent reports %d", __func__, res)); + + for (i = 0; i < res; i++) { + int which = 0; + + if (events[i].flags & EV_ERROR) { + /* + * Error messages that can happen, when a delete fails. + * EBADF happens when the file discriptor has been + * closed, + * ENOENT when the file discriptor was closed and + * then reopened. + * EINVAL for some reasons not understood; EINVAL + * should not be returned ever; but FreeBSD does :-\ + * An error is also indicated when a callback deletes + * an event we are still processing. In that case + * the data field is set to ENOENT. + */ + if (events[i].data == EBADF || + events[i].data == EINVAL || + events[i].data == ENOENT) + continue; + errno = events[i].data; + return (-1); + } + + ev = (struct event *)events[i].udata; + + if (events[i].filter == EVFILT_READ) { + which |= EV_READ; + } else if (events[i].filter == EVFILT_WRITE) { + which |= EV_WRITE; + } else if (events[i].filter == EVFILT_SIGNAL) { + which |= EV_SIGNAL; + } + + if (!which) + continue; + + if (!(ev->ev_events & EV_PERSIST)) + event_del(ev); + + event_active(ev, which, + ev->ev_events & EV_SIGNAL ? events[i].data : 1); + } + + return (0); +} + + +int +kq_add(void *arg, struct event *ev) +{ + struct kqop *kqop = arg; + struct kevent kev; + + if (ev->ev_events & EV_SIGNAL) { + int nsignal = EVENT_SIGNAL(ev); + + memset(&kev, 0, sizeof(kev)); + kev.ident = nsignal; + kev.filter = EVFILT_SIGNAL; + kev.flags = EV_ADD; + if (!(ev->ev_events & EV_PERSIST)) + kev.flags |= EV_ONESHOT; + kev.udata = PTR_TO_UDATA(ev); + + if (kq_insert(kqop, &kev) == -1) + return (-1); + + if (signal(nsignal, kq_sighandler) == SIG_ERR) + return (-1); + + ev->ev_flags |= EVLIST_X_KQINKERNEL; + return (0); + } + + if (ev->ev_events & EV_READ) { + memset(&kev, 0, sizeof(kev)); + kev.ident = ev->ev_fd; + kev.filter = EVFILT_READ; +#ifdef NOTE_EOF + /* Make it behave like select() and poll() */ + kev.fflags = NOTE_EOF; +#endif + kev.flags = EV_ADD; + if (!(ev->ev_events & EV_PERSIST)) + kev.flags |= EV_ONESHOT; + kev.udata = PTR_TO_UDATA(ev); + + if (kq_insert(kqop, &kev) == -1) + return (-1); + + ev->ev_flags |= EVLIST_X_KQINKERNEL; + } + + if (ev->ev_events & EV_WRITE) { + memset(&kev, 0, sizeof(kev)); + kev.ident = ev->ev_fd; + kev.filter = EVFILT_WRITE; + kev.flags = EV_ADD; + if (!(ev->ev_events & EV_PERSIST)) + kev.flags |= EV_ONESHOT; + kev.udata = PTR_TO_UDATA(ev); + + if (kq_insert(kqop, &kev) == -1) + return (-1); + + ev->ev_flags |= EVLIST_X_KQINKERNEL; + } + + return (0); +} + +int +kq_del(void *arg, struct event *ev) +{ + struct kqop *kqop = arg; + struct kevent kev; + + if (!(ev->ev_flags & EVLIST_X_KQINKERNEL)) + return (0); + + if (ev->ev_events & EV_SIGNAL) { + int nsignal = EVENT_SIGNAL(ev); + + memset(&kev, 0, sizeof(kev)); + kev.ident = nsignal; + kev.filter = EVFILT_SIGNAL; + kev.flags = EV_DELETE; + + if (kq_insert(kqop, &kev) == -1) + return (-1); + + if (signal(nsignal, SIG_DFL) == SIG_ERR) + return (-1); + + ev->ev_flags &= ~EVLIST_X_KQINKERNEL; + return (0); + } + + if (ev->ev_events & EV_READ) { + memset(&kev, 0, sizeof(kev)); + kev.ident = ev->ev_fd; + kev.filter = EVFILT_READ; + kev.flags = EV_DELETE; + + if (kq_insert(kqop, &kev) == -1) + return (-1); + + ev->ev_flags &= ~EVLIST_X_KQINKERNEL; + } + + if (ev->ev_events & EV_WRITE) { + memset(&kev, 0, sizeof(kev)); + kev.ident = ev->ev_fd; + kev.filter = EVFILT_WRITE; + kev.flags = EV_DELETE; + + if (kq_insert(kqop, &kev) == -1) + return (-1); + + ev->ev_flags &= ~EVLIST_X_KQINKERNEL; + } + + return (0); +} + +void +kq_dealloc(struct event_base *base, void *arg) +{ + struct kqop *kqop = arg; + + if (kqop->changes) + free(kqop->changes); + if (kqop->events) + free(kqop->events); + if (kqop->kq) + close(kqop->kq); + memset(kqop, 0, sizeof(struct kqop)); + free(kqop); +} diff --git a/contrib/libevent/log.c b/contrib/libevent/log.c new file mode 100644 index 0000000000..c9275e363f --- /dev/null +++ b/contrib/libevent/log.c @@ -0,0 +1,219 @@ +/* $OpenBSD: err.c,v 1.2 2002/06/25 15:50:15 mickey Exp $ */ + +/* + * log.c + * + * Based on err.c, which was adapted from OpenBSD libc *err* *warn* code. + * + * Copyright (c) 2005 Nick Mathewson + * + * Copyright (c) 2000 Dug Song + * + * Copyright (c) 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#ifdef WIN32 +#define WIN32_LEAN_AND_MEAN +#include +#undef WIN32_LEAN_AND_MEAN +#include "misc.h" +#endif +#include +#include +#ifdef HAVE_SYS_TIME_H +#include +#else +#include +#endif +#include +#include +#include +#include +#include +#include "event.h" + +#include "log.h" + +static void _warn_helper(int severity, int log_errno, const char *fmt, + va_list ap); +static void event_log(int severity, const char *msg); + +static int +event_vsnprintf(char *str, size_t size, const char *format, va_list args) +{ + int r; + if (size == 0) + return -1; +#ifdef WIN32 + r = _vsnprintf(str, size, format, args); +#else + r = vsnprintf(str, size, format, args); +#endif + str[size-1] = '\0'; + if (r < 0 || ((size_t)r) >= size) { + /* different platforms behave differently on overflow; + * handle both kinds. */ + return -1; + } + return r; +} + +static int +event_snprintf(char *str, size_t size, const char *format, ...) +{ + va_list ap; + int r; + va_start(ap, format); + r = event_vsnprintf(str, size, format, ap); + va_end(ap); + return r; +} + +void +event_err(int eval, const char *fmt, ...) +{ + va_list ap; + + va_start(ap, fmt); + _warn_helper(_EVENT_LOG_ERR, errno, fmt, ap); + va_end(ap); + exit(eval); +} + +void +event_warn(const char *fmt, ...) +{ + va_list ap; + + va_start(ap, fmt); + _warn_helper(_EVENT_LOG_WARN, errno, fmt, ap); + va_end(ap); +} + +void +event_errx(int eval, const char *fmt, ...) +{ + va_list ap; + + va_start(ap, fmt); + _warn_helper(_EVENT_LOG_ERR, -1, fmt, ap); + va_end(ap); + exit(eval); +} + +void +event_warnx(const char *fmt, ...) +{ + va_list ap; + + va_start(ap, fmt); + _warn_helper(_EVENT_LOG_WARN, -1, fmt, ap); + va_end(ap); +} + +void +event_msgx(const char *fmt, ...) +{ + va_list ap; + + va_start(ap, fmt); + _warn_helper(_EVENT_LOG_MSG, -1, fmt, ap); + va_end(ap); +} + +void +_event_debugx(const char *fmt, ...) +{ + va_list ap; + + va_start(ap, fmt); + _warn_helper(_EVENT_LOG_DEBUG, -1, fmt, ap); + va_end(ap); +} + +static void +_warn_helper(int severity, int log_errno, const char *fmt, va_list ap) +{ + char buf[1024]; + size_t len; + + if (fmt != NULL) + event_vsnprintf(buf, sizeof(buf), fmt, ap); + else + buf[0] = '\0'; + + if (log_errno >= 0) { + len = strlen(buf); + if (len < sizeof(buf) - 3) { + event_snprintf(buf + len, sizeof(buf) - len, ": %s", + strerror(log_errno)); + } + } + + event_log(severity, buf); +} + +static event_log_cb log_fn = NULL; + +void +event_set_log_callback(event_log_cb cb) +{ + log_fn = cb; +} + +static void +event_log(int severity, const char *msg) +{ + if (log_fn) + log_fn(severity, msg); + else { + const char *severity_str; + switch (severity) { + case _EVENT_LOG_DEBUG: + severity_str = "debug"; + break; + case _EVENT_LOG_MSG: + severity_str = "msg"; + break; + case _EVENT_LOG_WARN: + severity_str = "warn"; + break; + case _EVENT_LOG_ERR: + severity_str = "err"; + break; + default: + severity_str = "???"; + break; + } + (void)fprintf(stderr, "[%s] %s\n", severity_str, msg); + } +} diff --git a/contrib/libevent/log.h b/contrib/libevent/log.h new file mode 100644 index 0000000000..1f843cf984 --- /dev/null +++ b/contrib/libevent/log.h @@ -0,0 +1,43 @@ +/* + * Copyright (c) 2000-2004 Niels Provos + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#ifndef _LOG_H_ +#define _LOG_H_ + +void event_err(int eval, const char *fmt, ...); +void event_warn(const char *fmt, ...); +void event_errx(int eval, const char *fmt, ...); +void event_warnx(const char *fmt, ...); +void event_msgx(const char *fmt, ...); +void _event_debugx(const char *fmt, ...); + +#ifdef USE_DEBUG +#define event_debug(x) _event_debugx x +#else +#define event_debug(x) do {;} while (0) +#endif + +#endif diff --git a/contrib/libevent/poll.c b/contrib/libevent/poll.c new file mode 100644 index 0000000000..123d36a560 --- /dev/null +++ b/contrib/libevent/poll.c @@ -0,0 +1,391 @@ +/* $OpenBSD: poll.c,v 1.2 2002/06/25 15:50:15 mickey Exp $ */ + +/* + * Copyright 2000-2003 Niels Provos + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#include +#ifdef HAVE_SYS_TIME_H +#include +#else +#include +#endif +#include +#include +#include +#include +#include +#include +#include +#include +#include +#ifdef CHECK_INVARIANTS +#include +#endif + +#include "event.h" +#include "event-internal.h" +#include "evsignal.h" +#include "log.h" + +struct pollop { + int event_count; /* Highest number alloc */ + int nfds; /* Size of event_* */ + int fd_count; /* Size of idxplus1_by_fd */ + struct pollfd *event_set; + struct event **event_r_back; + struct event **event_w_back; + int *idxplus1_by_fd; /* Index into event_set by fd; we add 1 so + * that 0 (which is easy to memset) can mean + * "no entry." */ +}; + +void *poll_init (struct event_base *); +int poll_add (void *, struct event *); +int poll_del (void *, struct event *); +int poll_recalc (struct event_base *, void *, int); +int poll_dispatch (struct event_base *, void *, struct timeval *); +void poll_dealloc (struct event_base *, void *); + +const struct eventop pollops = { + "poll", + poll_init, + poll_add, + poll_del, + poll_recalc, + poll_dispatch, + poll_dealloc +}; + +void * +poll_init(struct event_base *base) +{ + struct pollop *pollop; + + /* Disable poll when this environment variable is set */ + if (getenv("EVENT_NOPOLL")) + return (NULL); + + if (!(pollop = calloc(1, sizeof(struct pollop)))) + return (NULL); + + evsignal_init(base); + + return (pollop); +} + +/* + * Called with the highest fd that we know about. If it is 0, completely + * recalculate everything. + */ + +int +poll_recalc(struct event_base *base, void *arg, int max) +{ + return (0); +} + +#ifdef CHECK_INVARIANTS +static void +poll_check_ok(struct pollop *pop) +{ + int i, idx; + struct event *ev; + + for (i = 0; i < pop->fd_count; ++i) { + idx = pop->idxplus1_by_fd[i]-1; + if (idx < 0) + continue; + assert(pop->event_set[idx].fd == i); + if (pop->event_set[idx].events & POLLIN) { + ev = pop->event_r_back[idx]; + assert(ev); + assert(ev->ev_events & EV_READ); + assert(ev->ev_fd == i); + } + if (pop->event_set[idx].events & POLLOUT) { + ev = pop->event_w_back[idx]; + assert(ev); + assert(ev->ev_events & EV_WRITE); + assert(ev->ev_fd == i); + } + } + for (i = 0; i < pop->nfds; ++i) { + struct pollfd *pfd = &pop->event_set[i]; + assert(pop->idxplus1_by_fd[pfd->fd] == i+1); + } +} +#else +#define poll_check_ok(pop) +#endif + +int +poll_dispatch(struct event_base *base, void *arg, struct timeval *tv) +{ + int res, i, msec = -1, nfds; + struct pollop *pop = arg; + + poll_check_ok(pop); + + if (tv != NULL) + msec = tv->tv_sec * 1000 + (tv->tv_usec + 999) / 1000; + + nfds = pop->nfds; + res = poll(pop->event_set, nfds, msec); + + if (res == -1) { + if (errno != EINTR) { + event_warn("poll"); + return (-1); + } + + evsignal_process(base); + return (0); + } else if (base->sig.evsignal_caught) { + evsignal_process(base); + } + + event_debug(("%s: poll reports %d", __func__, res)); + + if (res == 0) + return (0); + + for (i = 0; i < nfds; i++) { + int what = pop->event_set[i].revents; + struct event *r_ev = NULL, *w_ev = NULL; + if (!what) + continue; + + res = 0; + + /* If the file gets closed notify */ + if (what & (POLLHUP|POLLERR)) + what |= POLLIN|POLLOUT; + if (what & POLLIN) { + res |= EV_READ; + r_ev = pop->event_r_back[i]; + } + if (what & POLLOUT) { + res |= EV_WRITE; + w_ev = pop->event_w_back[i]; + } + if (res == 0) + continue; + + if (r_ev && (res & r_ev->ev_events)) { + if (!(r_ev->ev_events & EV_PERSIST)) + event_del(r_ev); + event_active(r_ev, res & r_ev->ev_events, 1); + } + if (w_ev && w_ev != r_ev && (res & w_ev->ev_events)) { + if (!(w_ev->ev_events & EV_PERSIST)) + event_del(w_ev); + event_active(w_ev, res & w_ev->ev_events, 1); + } + } + + return (0); +} + +int +poll_add(void *arg, struct event *ev) +{ + struct pollop *pop = arg; + struct pollfd *pfd = NULL; + int i; + + if (ev->ev_events & EV_SIGNAL) + return (evsignal_add(ev)); + if (!(ev->ev_events & (EV_READ|EV_WRITE))) + return (0); + + poll_check_ok(pop); + if (pop->nfds + 1 >= pop->event_count) { + struct pollfd *tmp_event_set; + struct event **tmp_event_r_back; + struct event **tmp_event_w_back; + int tmp_event_count; + + if (pop->event_count < 32) + tmp_event_count = 32; + else + tmp_event_count = pop->event_count * 2; + + /* We need more file descriptors */ + tmp_event_set = realloc(pop->event_set, + tmp_event_count * sizeof(struct pollfd)); + if (tmp_event_set == NULL) { + event_warn("realloc"); + return (-1); + } + pop->event_set = tmp_event_set; + + tmp_event_r_back = realloc(pop->event_r_back, + tmp_event_count * sizeof(struct event *)); + if (tmp_event_r_back == NULL) { + /* event_set overallocated; that's okay. */ + event_warn("realloc"); + return (-1); + } + pop->event_r_back = tmp_event_r_back; + + tmp_event_w_back = realloc(pop->event_w_back, + tmp_event_count * sizeof(struct event *)); + if (tmp_event_w_back == NULL) { + /* event_set and event_r_back overallocated; that's + * okay. */ + event_warn("realloc"); + return (-1); + } + pop->event_w_back = tmp_event_w_back; + + pop->event_count = tmp_event_count; + } + if (ev->ev_fd >= pop->fd_count) { + int *tmp_idxplus1_by_fd; + int new_count; + if (pop->fd_count < 32) + new_count = 32; + else + new_count = pop->fd_count * 2; + while (new_count <= ev->ev_fd) + new_count *= 2; + tmp_idxplus1_by_fd = + realloc(pop->idxplus1_by_fd, new_count * sizeof(int)); + if (tmp_idxplus1_by_fd == NULL) { + event_warn("realloc"); + return (-1); + } + pop->idxplus1_by_fd = tmp_idxplus1_by_fd; + memset(pop->idxplus1_by_fd + pop->fd_count, + 0, sizeof(int)*(new_count - pop->fd_count)); + pop->fd_count = new_count; + } + + i = pop->idxplus1_by_fd[ev->ev_fd] - 1; + if (i >= 0) { + pfd = &pop->event_set[i]; + } else { + i = pop->nfds++; + pfd = &pop->event_set[i]; + pfd->events = 0; + pfd->fd = ev->ev_fd; + pop->event_w_back[i] = pop->event_r_back[i] = NULL; + pop->idxplus1_by_fd[ev->ev_fd] = i + 1; + } + + pfd->revents = 0; + if (ev->ev_events & EV_WRITE) { + pfd->events |= POLLOUT; + pop->event_w_back[i] = ev; + } + if (ev->ev_events & EV_READ) { + pfd->events |= POLLIN; + pop->event_r_back[i] = ev; + } + poll_check_ok(pop); + + return (0); +} + +/* + * Nothing to be done here. + */ + +int +poll_del(void *arg, struct event *ev) +{ + struct pollop *pop = arg; + struct pollfd *pfd = NULL; + int i; + + if (ev->ev_events & EV_SIGNAL) + return (evsignal_del(ev)); + + if (!(ev->ev_events & (EV_READ|EV_WRITE))) + return (0); + + poll_check_ok(pop); + i = pop->idxplus1_by_fd[ev->ev_fd] - 1; + if (i < 0) + return (-1); + + /* Do we still want to read or write? */ + pfd = &pop->event_set[i]; + if (ev->ev_events & EV_READ) { + pfd->events &= ~POLLIN; + pop->event_r_back[i] = NULL; + } + if (ev->ev_events & EV_WRITE) { + pfd->events &= ~POLLOUT; + pop->event_w_back[i] = NULL; + } + poll_check_ok(pop); + if (pfd->events) + /* Another event cares about that fd. */ + return (0); + + /* Okay, so we aren't interested in that fd anymore. */ + pop->idxplus1_by_fd[ev->ev_fd] = 0; + + --pop->nfds; + if (i != pop->nfds) { + /* + * Shift the last pollfd down into the now-unoccupied + * position. + */ + memcpy(&pop->event_set[i], &pop->event_set[pop->nfds], + sizeof(struct pollfd)); + pop->event_r_back[i] = pop->event_r_back[pop->nfds]; + pop->event_w_back[i] = pop->event_w_back[pop->nfds]; + pop->idxplus1_by_fd[pop->event_set[i].fd] = i + 1; + } + + poll_check_ok(pop); + return (0); +} + +void +poll_dealloc(struct event_base *base, void *arg) +{ + struct pollop *pop = arg; + + evsignal_dealloc(base); + if (pop->event_set) + free(pop->event_set); + if (pop->event_r_back) + free(pop->event_r_back); + if (pop->event_w_back) + free(pop->event_w_back); + if (pop->idxplus1_by_fd) + free(pop->idxplus1_by_fd); + + memset(pop, 0, sizeof(struct pollop)); + free(pop); +} diff --git a/contrib/libevent/select.c b/contrib/libevent/select.c new file mode 100644 index 0000000000..d645f1a370 --- /dev/null +++ b/contrib/libevent/select.c @@ -0,0 +1,373 @@ +/* $OpenBSD: select.c,v 1.2 2002/06/25 15:50:15 mickey Exp $ */ + +/* + * Copyright 2000-2002 Niels Provos + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#include +#ifdef HAVE_SYS_TIME_H +#include +#else +#include +#endif +#ifdef HAVE_SYS_SELECT_H +#include +#endif +#include +#include +#include +#include +#include +#include +#include +#include +#ifdef CHECK_INVARIANTS +#include +#endif + +#include "event.h" +#include "event-internal.h" +#include "evsignal.h" +#include "log.h" + +#ifndef howmany +#define howmany(x, y) (((x)+((y)-1))/(y)) +#endif + +struct selectop { + int event_fds; /* Highest fd in fd set */ + int event_fdsz; + fd_set *event_readset_in; + fd_set *event_writeset_in; + fd_set *event_readset_out; + fd_set *event_writeset_out; + struct event **event_r_by_fd; + struct event **event_w_by_fd; +}; + +void *select_init (struct event_base *); +int select_add (void *, struct event *); +int select_del (void *, struct event *); +int select_recalc (struct event_base *, void *, int); +int select_dispatch (struct event_base *, void *, struct timeval *); +void select_dealloc (struct event_base *, void *); + +const struct eventop selectops = { + "select", + select_init, + select_add, + select_del, + select_recalc, + select_dispatch, + select_dealloc +}; + +static int select_resize(struct selectop *sop, int fdsz); + +void * +select_init(struct event_base *base) +{ + struct selectop *sop; + + /* Disable select when this environment variable is set */ + if (getenv("EVENT_NOSELECT")) + return (NULL); + + if (!(sop = calloc(1, sizeof(struct selectop)))) + return (NULL); + + select_resize(sop, howmany(32 + 1, NFDBITS)*sizeof(fd_mask)); + + evsignal_init(base); + + return (sop); +} + +#ifdef CHECK_INVARIANTS +static void +check_selectop(struct selectop *sop) +{ + int i; + for (i = 0; i <= sop->event_fds; ++i) { + if (FD_ISSET(i, sop->event_readset_in)) { + assert(sop->event_r_by_fd[i]); + assert(sop->event_r_by_fd[i]->ev_events & EV_READ); + assert(sop->event_r_by_fd[i]->ev_fd == i); + } else { + assert(! sop->event_r_by_fd[i]); + } + if (FD_ISSET(i, sop->event_writeset_in)) { + assert(sop->event_w_by_fd[i]); + assert(sop->event_w_by_fd[i]->ev_events & EV_WRITE); + assert(sop->event_w_by_fd[i]->ev_fd == i); + } else { + assert(! sop->event_w_by_fd[i]); + } + } + +} +#else +#define check_selectop(sop) do { (void) sop; } while (0) +#endif + +/* + * Called with the highest fd that we know about. If it is 0, completely + * recalculate everything. + */ + +int +select_recalc(struct event_base *base, void *arg, int max) +{ + struct selectop *sop = arg; + + check_selectop(sop); + + return (0); +} + +int +select_dispatch(struct event_base *base, void *arg, struct timeval *tv) +{ + int res, i; + struct selectop *sop = arg; + + check_selectop(sop); + + memcpy(sop->event_readset_out, sop->event_readset_in, + sop->event_fdsz); + memcpy(sop->event_writeset_out, sop->event_writeset_in, + sop->event_fdsz); + + res = select(sop->event_fds + 1, sop->event_readset_out, + sop->event_writeset_out, NULL, tv); + + check_selectop(sop); + + if (res == -1) { + if (errno != EINTR) { + event_warn("select"); + return (-1); + } + + evsignal_process(base); + return (0); + } else if (base->sig.evsignal_caught) { + evsignal_process(base); + } + + event_debug(("%s: select reports %d", __func__, res)); + + check_selectop(sop); + for (i = 0; i <= sop->event_fds; ++i) { + struct event *r_ev = NULL, *w_ev = NULL; + res = 0; + if (FD_ISSET(i, sop->event_readset_out)) { + r_ev = sop->event_r_by_fd[i]; + res |= EV_READ; + } + if (FD_ISSET(i, sop->event_writeset_out)) { + w_ev = sop->event_w_by_fd[i]; + res |= EV_WRITE; + } + if (r_ev && (res & r_ev->ev_events)) { + if (!(r_ev->ev_events & EV_PERSIST)) + event_del(r_ev); + event_active(r_ev, res & r_ev->ev_events, 1); + } + if (w_ev && w_ev != r_ev && (res & w_ev->ev_events)) { + if (!(w_ev->ev_events & EV_PERSIST)) + event_del(w_ev); + event_active(w_ev, res & w_ev->ev_events, 1); + } + } + check_selectop(sop); + + return (0); +} + + +static int +select_resize(struct selectop *sop, int fdsz) +{ + int n_events, n_events_old; + + fd_set *readset_in = NULL; + fd_set *writeset_in = NULL; + fd_set *readset_out = NULL; + fd_set *writeset_out = NULL; + struct event **r_by_fd = NULL; + struct event **w_by_fd = NULL; + + n_events = (fdsz/sizeof(fd_mask)) * NFDBITS; + n_events_old = (sop->event_fdsz/sizeof(fd_mask)) * NFDBITS; + + if (sop->event_readset_in) + check_selectop(sop); + + if ((readset_in = realloc(sop->event_readset_in, fdsz)) == NULL) + goto error; + sop->event_readset_in = readset_in; + if ((readset_out = realloc(sop->event_readset_out, fdsz)) == NULL) + goto error; + sop->event_readset_out = readset_out; + if ((writeset_in = realloc(sop->event_writeset_in, fdsz)) == NULL) + goto error; + sop->event_writeset_in = writeset_in; + if ((writeset_out = realloc(sop->event_writeset_out, fdsz)) == NULL) + goto error; + sop->event_writeset_out = writeset_out; + if ((r_by_fd = realloc(sop->event_r_by_fd, + n_events*sizeof(struct event*))) == NULL) + goto error; + sop->event_r_by_fd = r_by_fd; + if ((w_by_fd = realloc(sop->event_w_by_fd, + n_events * sizeof(struct event*))) == NULL) + goto error; + sop->event_w_by_fd = w_by_fd; + + memset((char *)sop->event_readset_in + sop->event_fdsz, 0, + fdsz - sop->event_fdsz); + memset((char *)sop->event_writeset_in + sop->event_fdsz, 0, + fdsz - sop->event_fdsz); + memset(sop->event_r_by_fd + n_events_old, 0, + (n_events-n_events_old) * sizeof(struct event*)); + memset(sop->event_w_by_fd + n_events_old, 0, + (n_events-n_events_old) * sizeof(struct event*)); + + sop->event_fdsz = fdsz; + check_selectop(sop); + + return (0); + + error: + event_warn("malloc"); + return (-1); +} + + +int +select_add(void *arg, struct event *ev) +{ + struct selectop *sop = arg; + + if (ev->ev_events & EV_SIGNAL) + return (evsignal_add(ev)); + + check_selectop(sop); + /* + * Keep track of the highest fd, so that we can calculate the size + * of the fd_sets for select(2) + */ + if (sop->event_fds < ev->ev_fd) { + int fdsz = sop->event_fdsz; + + if (fdsz < sizeof(fd_mask)) + fdsz = sizeof(fd_mask); + + while (fdsz < + (howmany(ev->ev_fd + 1, NFDBITS) * sizeof(fd_mask))) + fdsz *= 2; + + if (fdsz != sop->event_fdsz) { + if (select_resize(sop, fdsz)) { + check_selectop(sop); + return (-1); + } + } + + sop->event_fds = ev->ev_fd; + } + + if (ev->ev_events & EV_READ) { + FD_SET(ev->ev_fd, sop->event_readset_in); + sop->event_r_by_fd[ev->ev_fd] = ev; + } + if (ev->ev_events & EV_WRITE) { + FD_SET(ev->ev_fd, sop->event_writeset_in); + sop->event_w_by_fd[ev->ev_fd] = ev; + } + check_selectop(sop); + + return (0); +} + +/* + * Nothing to be done here. + */ + +int +select_del(void *arg, struct event *ev) +{ + struct selectop *sop = arg; + + check_selectop(sop); + if (ev->ev_events & EV_SIGNAL) + return (evsignal_del(ev)); + + if (sop->event_fds < ev->ev_fd) { + check_selectop(sop); + return (0); + } + + if (ev->ev_events & EV_READ) { + FD_CLR(ev->ev_fd, sop->event_readset_in); + sop->event_r_by_fd[ev->ev_fd] = NULL; + } + + if (ev->ev_events & EV_WRITE) { + FD_CLR(ev->ev_fd, sop->event_writeset_in); + sop->event_w_by_fd[ev->ev_fd] = NULL; + } + + check_selectop(sop); + return (0); +} + +void +select_dealloc(struct event_base *base, void *arg) +{ + struct selectop *sop = arg; + + evsignal_dealloc(base); + if (sop->event_readset_in) + free(sop->event_readset_in); + if (sop->event_writeset_in) + free(sop->event_writeset_in); + if (sop->event_readset_out) + free(sop->event_readset_out); + if (sop->event_writeset_out) + free(sop->event_writeset_out); + if (sop->event_r_by_fd) + free(sop->event_r_by_fd); + if (sop->event_w_by_fd) + free(sop->event_w_by_fd); + + memset(sop, 0, sizeof(struct selectop)); + free(sop); +} diff --git a/contrib/libevent/signal.c b/contrib/libevent/signal.c new file mode 100644 index 0000000000..6c0953d9e1 --- /dev/null +++ b/contrib/libevent/signal.c @@ -0,0 +1,195 @@ +/* $OpenBSD: select.c,v 1.2 2002/06/25 15:50:15 mickey Exp $ */ + +/* + * Copyright 2000-2002 Niels Provos + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the author may not be used to endorse or promote products + * derived from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#include +#include +#ifdef HAVE_SYS_TIME_H +#include +#else +#include +#endif +#include +#include +#include +#include +#include +#include +#include +#include +#ifdef HAVE_FCNTL_H +#include +#endif +#include + +#include "event.h" +#include "event-internal.h" +#include "evsignal.h" +#include "log.h" + +struct event_base *evsignal_base = NULL; + +static void evsignal_handler(int sig); + +/* Callback for when the signal handler write a byte to our signaling socket */ +static void +evsignal_cb(int fd, short what, void *arg) +{ + static char signals[100]; + struct event *ev = arg; + ssize_t n; + + n = read(fd, signals, sizeof(signals)); + if (n == -1) + event_err(1, "%s: read", __func__); + event_add(ev, NULL); +} + +#ifdef HAVE_SETFD +#define FD_CLOSEONEXEC(x) do { \ + if (fcntl(x, F_SETFD, 1) == -1) \ + event_warn("fcntl(%d, F_SETFD)", x); \ +} while (0) +#else +#define FD_CLOSEONEXEC(x) +#endif + +void +evsignal_init(struct event_base *base) +{ + /* + * Our signal handler is going to write to one end of the socket + * pair to wake up our event loop. The event loop then scans for + * signals that got delivered. + */ + if (socketpair(AF_UNIX, SOCK_STREAM, 0, base->sig.ev_signal_pair) == -1) + event_err(1, "%s: socketpair", __func__); + + FD_CLOSEONEXEC(base->sig.ev_signal_pair[0]); + FD_CLOSEONEXEC(base->sig.ev_signal_pair[1]); + base->sig.evsignal_caught = 0; + memset(&base->sig.evsigcaught, 0, sizeof(sig_atomic_t)*NSIG); + + fcntl(base->sig.ev_signal_pair[0], F_SETFL, O_NONBLOCK); + + event_set(&base->sig.ev_signal, base->sig.ev_signal_pair[1], EV_READ, + evsignal_cb, &base->sig.ev_signal); + base->sig.ev_signal.ev_base = base; + base->sig.ev_signal.ev_flags |= EVLIST_INTERNAL; +} + +int +evsignal_add(struct event *ev) +{ + int evsignal; + struct sigaction sa; + struct event_base *base = ev->ev_base; + + if (ev->ev_events & (EV_READ|EV_WRITE)) + event_errx(1, "%s: EV_SIGNAL incompatible use", __func__); + evsignal = EVENT_SIGNAL(ev); + + memset(&sa, 0, sizeof(sa)); + sa.sa_handler = evsignal_handler; + sigfillset(&sa.sa_mask); + sa.sa_flags |= SA_RESTART; + /* catch signals if they happen quickly */ + evsignal_base = base; + + if (sigaction(evsignal, &sa, NULL) == -1) + return (-1); + + if (!base->sig.ev_signal_added) { + base->sig.ev_signal_added = 1; + event_add(&base->sig.ev_signal, NULL); + } + + return (0); +} + +int +evsignal_del(struct event *ev) +{ + return (sigaction(EVENT_SIGNAL(ev),(struct sigaction *)SIG_DFL, NULL)); +} + +static void +evsignal_handler(int sig) +{ + int save_errno = errno; + + if(evsignal_base == NULL) { + event_warn( + "%s: received signal %s, but have no base configured", + __func__, sig); + return; + } + + evsignal_base->sig.evsigcaught[sig]++; + evsignal_base->sig.evsignal_caught = 1; + + /* Wake up our notification mechanism */ + write(evsignal_base->sig.ev_signal_pair[0], "a", 1); + errno = save_errno; +} + +void +evsignal_process(struct event_base *base) +{ + struct event *ev; + sig_atomic_t ncalls; + + base->sig.evsignal_caught = 0; + TAILQ_FOREACH(ev, &base->sig.signalqueue, ev_signal_next) { + ncalls = base->sig.evsigcaught[EVENT_SIGNAL(ev)]; + if (ncalls) { + if (!(ev->ev_events & EV_PERSIST)) + event_del(ev); + event_active(ev, EV_SIGNAL, ncalls); + base->sig.evsigcaught[EVENT_SIGNAL(ev)] = 0; + } + } +} + +void +evsignal_dealloc(struct event_base *base) +{ + if(base->sig.ev_signal_added) { + event_del(&base->sig.ev_signal); + base->sig.ev_signal_added = 0; + } + assert(TAILQ_EMPTY(&base->sig.signalqueue)); + + close(base->sig.ev_signal_pair[0]); + base->sig.ev_signal_pair[0] = -1; + close(base->sig.ev_signal_pair[1]); + base->sig.ev_signal_pair[1] = -1; +} -- 2.41.0