From: Nick Mathewson Date: Fri, 27 Nov 2009 18:16:54 +0000 (-0500) Subject: Rate-limiting for bufferevents; group and individual limits are supported. X-Git-Tag: release-2.0.4-alpha~93^2~4 X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=737c9cd87b453d7371ea0b9774aa2bc51124c4eb;p=libevent Rate-limiting for bufferevents; group and individual limits are supported. The fairness algorithms are not the best, not every bufferevent type is supported, and some of the locking tricks here are simply absurd. Still, this code should be a good first step. --- diff --git a/Makefile.am b/Makefile.am index 08fb0fdd..52b15023 100644 --- a/Makefile.am +++ b/Makefile.am @@ -105,7 +105,7 @@ event-config.h: config.h CORE_SRC = event.c evthread.c buffer.c \ bufferevent.c bufferevent_sock.c bufferevent_filter.c \ - bufferevent_pair.c listener.c \ + bufferevent_pair.c listener.c bufferevent_ratelim.c \ evmap.c log.c evutil.c strlcpy.c $(SYS_SRC) EXTRA_SRC = event_tagging.c http.c evdns.c evrpc.c @@ -136,7 +136,8 @@ noinst_HEADERS = util-internal.h mm-internal.h ipv6-internal.h \ evrpc-internal.h strlcpy-internal.h evbuffer-internal.h \ bufferevent-internal.h http-internal.h event-internal.h \ evthread-internal.h ht-internal.h defer-internal.h \ - minheap-internal.h log-internal.h evsignal-internal.h evmap-internal.h + minheap-internal.h log-internal.h evsignal-internal.h evmap-internal.h \ + ratelim-internal.h include_HEADERS = event.h evhttp.h evdns.h evrpc.h evutil.h diff --git a/bufferevent-internal.h b/bufferevent-internal.h index c3c33eca..220cdeb7 100644 --- a/bufferevent-internal.h +++ b/bufferevent-internal.h @@ -35,6 +35,7 @@ extern "C" { #include "defer-internal.h" #include "evthread-internal.h" #include "event2/thread.h" +#include "ratelim-internal.h" /* These flags are reasons that we might be declining to actually enable reading or writing on a bufferevent. @@ -43,22 +44,77 @@ extern "C" { /* On a all bufferevents, for reading: used when we have read up to the watermark value. - On a filtering bufferxevent, for writing: used when the underlying + On a filtering bufferevent, for writing: used when the underlying bufferevent's write buffer has been filled up to its watermark value. */ #define BEV_SUSPEND_WM 0x01 -/* On a base bufferevent: when we have used up our bandwidth buckets. */ +/* On a base bufferevent: when we have emptied a bandwidth buckets */ #define BEV_SUSPEND_BW 0x02 +/* On a base bufferevent: when we have emptied the group's bandwidth bucket. */ +#define BEV_SUSPEND_BW_GROUP 0x04 /* On a socket bufferevent: we aren't going to try reading until the * connect operation is done. */ -#define BEV_SUSPEND_CONNECTING 0x04 +#define BEV_SUSPEND_CONNECTING 0x08 + +struct bufferevent_rate_limit_group { + /** List of all members in the group */ + TAILQ_HEAD(rlim_group_member_list, bufferevent_private) members; + /** Current limits for the group. */ + struct ev_token_bucket rate_limit; + struct ev_token_bucket_cfg rate_limit_cfg; + + /** True iff we don't want to read from any member of the group.until + * the token bucket refills. */ + unsigned read_suspended : 1; + /** True iff we don't want to write from any member of the group.until + * the token bucket refills. */ + unsigned write_suspended : 1; + /** True iff we were unable to suspend one of the bufferevents in the + * group for reading the last time we tried, and we should try + * again. */ + unsigned pending_unsuspend_read : 1; + /** True iff we were unable to suspend one of the bufferevents in the + * group for writing the last time we tried, and we should try + * again. */ + unsigned pending_unsuspend_write : 1; + + /** The number of bufferevents in the group. */ + int n_members; + + /** The smallest number of bytes that any member of the group should + * be limited to read or write at a time. */ + ev_uint32_t min_share; + /** Timeout event that goes off once a tick, when the bucket is ready + * to refill. */ + struct event master_refill_event; + /** Lock to protect the members of this group. This lock should nest + * within every bufferevent lock: if you are holding this lock, do + * not assume you can lock another bufferevent. */ + void *lock; +}; -struct token_bucket { - ev_uint32_t limit; - ev_uint32_t rate; - ev_uint32_t burst; - unsigned last_updated; +/** Fields for rate-limiting a single bufferevent. */ +struct bufferevent_rate_limit { + /* Linked-list elements for storing this bufferevent_private in a + * group. + * + * Note that this field is supposed to be protected by the group + * lock */ + TAILQ_ENTRY(bufferevent_private) next_in_group; + /** The rate-limiting group for this bufferevent, or NULL if it is + * only rate-limited on its own. */ + struct bufferevent_rate_limit_group *group; + + /* This bufferevent's current limits. */ + struct ev_token_bucket limit; + /* Pointer to the rate-limit configuration for this bufferevent. + * Can be shared. XXX reference-count this? */ + struct ev_token_bucket_cfg *cfg; + + /* Timeout event used when one this bufferevent's buckets are + * empty. */ + struct event refill_bucket_event; }; /** Parts of the bufferevent structure that are shared among all bufferevent @@ -111,6 +167,9 @@ struct bufferevent_private { /** Lock for this bufferevent. Shared by the inbuf and the outbuf. * If NULL, locking is disabled. */ void *lock; + + /** Rate-limiting information for this bufferevent */ + struct bufferevent_rate_limit *rate_limiting; }; /** Possible operations for a control callback. */ @@ -170,6 +229,7 @@ struct bufferevent_ops { /** Called to access miscellaneous fields. */ int (*ctrl)(struct bufferevent *, enum bufferevent_ctrl_op, union bufferevent_ctrl_data *); + }; extern const struct bufferevent_ops bufferevent_ops_socket; @@ -287,6 +347,15 @@ void _bufferevent_generic_adj_timeouts(struct bufferevent *bev); EVLOCK_UNLOCK(locking->lock, 0); \ } while(0) +/* ==== For rate-limiting. */ + +int _bufferevent_decrement_write_buckets(struct bufferevent_private *bev, + int bytes); +int _bufferevent_decrement_read_buckets(struct bufferevent_private *bev, + int bytes); +int _bufferevent_get_read_max(struct bufferevent_private *bev); +int _bufferevent_get_write_max(struct bufferevent_private *bev); + #ifdef __cplusplus } #endif diff --git a/bufferevent.c b/bufferevent.c index 96b8ec7d..94a00714 100644 --- a/bufferevent.c +++ b/bufferevent.c @@ -525,6 +525,15 @@ _bufferevent_decref_and_unlock(struct bufferevent *bufev) evbuffer_free(bufev->input); evbuffer_free(bufev->output); + if (bufev_private->rate_limiting) { + if (bufev_private->rate_limiting->group) + bufferevent_remove_from_rate_limit_group(bufev); + if (event_initialized(&bufev_private->rate_limiting->refill_bucket_event)) + event_del(&bufev_private->rate_limiting->refill_bucket_event); + mm_free(bufev_private->rate_limiting); + bufev_private->rate_limiting = NULL; + } + BEV_UNLOCK(bufev); if (bufev_private->own_lock) EVTHREAD_FREE_LOCK(bufev_private->lock, diff --git a/bufferevent_async.c b/bufferevent_async.c index a8e92b70..2075301b 100644 --- a/bufferevent_async.c +++ b/bufferevent_async.c @@ -127,6 +127,8 @@ upcast_write(struct event_overlapped *eo) static void bev_async_consider_writing(struct bufferevent_async *b) { + size_t at_most; + int limit; /* Don't write if there's a write in progress, or we do not * want to write. */ if (!b->ok || b->write_in_progress || !(b->bev.bev.enabled&EV_WRITE)) @@ -135,8 +137,18 @@ bev_async_consider_writing(struct bufferevent_async *b) if (!evbuffer_get_length(b->bev.bev.output)) return; + at_most = evbuffer_get_length(b->bev.bev.output); + + /* XXXX This over-commits. */ + limit = _bufferevent_get_write_max(&b->bev); + if (at_most >= limit) + at_most = limit; + + if (b->bev.write_suspended) + return; + /* XXXX doesn't respect low-water mark very well. */ - if (evbuffer_launch_write(b->bev.bev.output, -1, + if (evbuffer_launch_write(b->bev.bev.output, at_most, &b->write_overlapped)) { EVUTIL_ASSERT(0);/* XXX act sensibly. */ } else { @@ -150,6 +162,7 @@ bev_async_consider_reading(struct bufferevent_async *b) size_t cur_size; size_t read_high; size_t at_most; + int limit; /* Don't read if there is a read in progress, or we do not * want to read. */ if (!b->ok || b->read_in_progress || !(b->bev.bev.enabled&EV_READ)) @@ -166,6 +179,14 @@ bev_async_consider_reading(struct bufferevent_async *b) at_most = 16384; /* FIXME totally magic. */ } + /* XXXX This over-commits. */ + limit = _bufferevent_get_read_max(&b->bev); + if (at_most >= limit) + at_most = limit; + + if (b->bev.read_suspended) + return; + if (evbuffer_launch_read(b->bev.bev.input, at_most, &b->read_overlapped)) { EVUTIL_ASSERT(0); @@ -304,6 +325,7 @@ read_complete(struct event_overlapped *eo, uintptr_t key, if (ok && nbytes) { BEV_RESET_GENERIC_READ_TIMEOUT(bev); + _bufferevent_derement_read_buckets(&bev_a->bev, nbytes); if (bev->readcb != NULL && evbuffer_get_length(bev->input) >= bev->wm_read.low) _bufferevent_run_readcb(bev); @@ -337,7 +359,8 @@ write_complete(struct event_overlapped *eo, uintptr_t key, if (ok && nbytes) { BEV_RESET_GENERIC_WRITE_TIMEOUT(bev); - if (bev->writecb != NULL && + _bufferevent_derement_write_buckets(&bev_a->bev, nbytes); + if (bev->writecb != NULL && evbuffer_get_length(bev->output) <= bev->wm_write.low) _bufferevent_run_writecb(bev); bev_async_consider_writing(bev_a); diff --git a/bufferevent_openssl.c b/bufferevent_openssl.c index f121c5be..e2cb28b1 100644 --- a/bufferevent_openssl.c +++ b/bufferevent_openssl.c @@ -524,20 +524,29 @@ do_read(struct bufferevent_openssl *bev_ssl, int n_to_read) /* Requires lock */ struct bufferevent *bev = &bev_ssl->bev.bev; struct evbuffer *input = bev->input; - int r, n, i, n_used = 0, blocked = 0; + int r, n, i, n_used = 0, blocked = 0, atmost; struct evbuffer_iovec space[2]; + atmost = _bufferevent_get_read_max(&bev_ssl->bev); + if (n_to_read > atmost) + n_to_read = atmost; + n = evbuffer_reserve_space(input, n_to_read, space, 2); if (n < 0) return -1; for (i=0; ibev.read_suspended) + break; r = SSL_read(bev_ssl->ssl, space[i].iov_base, space[i].iov_len); if (r>0) { if (bev_ssl->read_blocked_on_write) clear_rbow(bev_ssl); ++n_used; space[i].iov_len = r; + /* Not exactly right; we probably want to do + * our rate-limiting on the underlying bytes. */ + _bufferevent_decrement_read_buckets(&bev_ssl->bev, r); } else { int err = SSL_get_error(bev_ssl->ssl, r); print_err(err); @@ -585,6 +594,8 @@ do_write(struct bufferevent_openssl *bev_ssl, int atmost) if (bev_ssl->last_write > 0) atmost = bev_ssl->last_write; + else + atmost = _bufferevent_get_write_max(&bev_ssl->bev); n = evbuffer_peek(output, atmost, NULL, space, 8); if (n < 0) @@ -593,6 +604,9 @@ do_write(struct bufferevent_openssl *bev_ssl, int atmost) if (n > 8) n = 8; for (i=0; i < n; ++i) { + if (bev_ssl->bev.write_suspended) + break; + r = SSL_write(bev_ssl->ssl, space[i].iov_base, space[i].iov_len); if (r > 0) { @@ -600,6 +614,9 @@ do_write(struct bufferevent_openssl *bev_ssl, int atmost) clear_wbor(bev_ssl); n_written += r; bev_ssl->last_write = -1; + /* Not exactly right; we probably want to do + * our rate-limiting on the underlying bytes. */ + _bufferevent_decrement_write_buckets(&bev_ssl->bev, r); } else { int err = SSL_get_error(bev_ssl->ssl, r); print_err(err); diff --git a/bufferevent_ratelim.c b/bufferevent_ratelim.c new file mode 100644 index 00000000..564367cb --- /dev/null +++ b/bufferevent_ratelim.c @@ -0,0 +1,654 @@ +/* + * Copyright (c) 2007-2009 Niels Provos and Nick Mathewson + * Copyright (c) 2002-2006 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. + */ + +#include +#include +#include +#include + +#include "event2/event.h" +#include "event2/event_struct.h" +#include "event2/util.h" +#include "event2/bufferevent.h" +#include "event2/bufferevent_struct.h" +#include "event2/buffer.h" + +#include "ratelim-internal.h" + +#include "bufferevent-internal.h" +#include "mm-internal.h" +#include "util-internal.h" + +int +ev_token_bucket_init(struct ev_token_bucket *bucket, + const struct ev_token_bucket_cfg *cfg, + ev_uint32_t current_tick, + int reinitialize) +{ + if (reinitialize) { + /* on reinitialization, we only clip downwards, since we've + already used who-knows-how-much bandwidth this tick. We + leave "last_updated" as it is; the next update will add the + appropriate amount of bandwidth to the bucket. + */ + if (bucket->read_limit > cfg->read_maximum) + bucket->read_limit = cfg->read_maximum; + if (bucket->write_limit > cfg->write_maximum) + bucket->write_limit = cfg->write_maximum; + } else { + bucket->read_limit = cfg->read_rate; + bucket->write_limit = cfg->write_rate; + bucket->last_updated = current_tick; + } + return 0; +} + +int +ev_token_bucket_update(struct ev_token_bucket *bucket, + const struct ev_token_bucket_cfg *cfg, + ev_uint32_t current_tick) +{ + /* It's okay if the tick number overflows, since we'll just + * wrap around when we do the unsigned substraction. */ + unsigned n_ticks = current_tick - bucket->last_updated; + + /* Make sure some ticks actually happened, and that time didn't + * roll back. */ + if (n_ticks == 0 || n_ticks > INT_MAX) + return 0; + + /* Naively, we would say + bucket->limit += n_ticks * cfg->rate; + + if (bucket->limit > cfg->maximum) + bucket->limit = cfg->maximum; + + But we're worried about overflow, so we do it like this: + */ + + if ((cfg->read_maximum - bucket->read_limit) / n_ticks < cfg->read_rate) + bucket->read_limit = cfg->read_maximum; + else + bucket->read_limit += n_ticks * cfg->read_rate; + + + if ((cfg->write_maximum - bucket->write_limit) / n_ticks < cfg->write_rate) + bucket->write_limit = cfg->write_maximum; + else + bucket->write_limit += n_ticks * cfg->write_rate; + + + bucket->last_updated = current_tick; + + return 1; +} + +ev_uint32_t +ev_token_bucket_get_tick(const struct timeval *tv, + const struct ev_token_bucket_cfg *cfg) +{ + /* This computation uses two multiplies and a divide. We could do + * fewer if we knew that the tick length was an integer number of + * seconds, or if we knew it divided evenly into a second. We should + * investigate that more. + */ + + /* We cast to an ev_uint64_t first, since we don't want to overflow + * before we do the final divide. */ + ev_uint64_t msec = (ev_uint64_t)tv->tv_sec * 1000 + tv->tv_usec / 1000; + return (unsigned)(msec / cfg->msec_per_tick); +} + +struct ev_token_bucket_cfg * +ev_token_bucket_cfg_new(ev_uint32_t read_rate, ev_uint32_t read_burst, + ev_uint32_t write_rate, ev_uint32_t write_burst, + const struct timeval *tick_len) +{ + struct ev_token_bucket_cfg *r; + struct timeval g; + if (! tick_len) { + g.tv_sec = 1; + g.tv_usec = 0; + tick_len = &g; + } + if (read_rate > read_burst || write_rate > write_burst || + read_rate < 1 || write_rate < 1) + return NULL; + r = mm_calloc(1, sizeof(struct ev_token_bucket_cfg)); + if (!r) + return NULL; + r->read_rate = read_rate; + r->write_rate = write_rate; + r->read_maximum = read_burst; + r->write_maximum = write_burst; + memcpy(&r->tick_timeout, tick_len, sizeof(struct timeval)); + r->msec_per_tick = (tick_len->tv_sec * 1000) + tick_len->tv_usec/1000; + return r; +} + +void +ev_token_bucket_cfg_free(struct ev_token_bucket_cfg *cfg) +{ + mm_free(cfg); +} + +/* No matter how big our bucket gets, don't try to read more than this + * much in a single read operation. */ +#define MAX_TO_READ_EVER 16384 +/* No matter how big our bucket gets, don't try to write more than this + * much in a single write operation. */ +#define MAX_TO_WRITE_EVER 16384 + +#define LOCK_GROUP(g) EVLOCK_LOCK((g)->lock, 0) +#define UNLOCK_GROUP(g) EVLOCK_UNLOCK((g)->lock, 0) + +static int _bev_group_suspend_reading(struct bufferevent_rate_limit_group *g); +static int _bev_group_suspend_writing(struct bufferevent_rate_limit_group *g); + +/** Helper: figure out the maximum amount we should write if is_write, or + the maximum amount we should read if is_read. Return that maximum, or + 0 if our bucket is wholly exhausted. + */ +static inline int +_bufferevent_get_rlim_max(struct bufferevent_private *bev, int is_write) +{ + /* needs lock on bev. */ + int max_so_far = is_write?MAX_TO_WRITE_EVER:MAX_TO_READ_EVER; + struct timeval now; + +#define LIM(x) \ + (is_write ? (x).write_limit : (x).read_limit) + +#define GROUP_SUSPENDED(g) \ + (is_write ? (g)->write_suspended : (g)->read_suspended) + + /* Sets max_so_far to MIN(x, max_so_far) */ +#define CLAMPTO(x) \ + do { \ + if (max_so_far > (x)) \ + max_so_far = (x); \ + } while (0); + + if (!bev->rate_limiting) + return max_so_far; + + /* If rate-limiting is enabled at all, update the appropriate + bucket, and take the smaller of our rate limit and the group + rate limit. + */ + + if (bev->rate_limiting->cfg) { + unsigned tick; + + event_base_gettimeofday_cached(bev->bev.ev_base, &now); + tick = ev_token_bucket_get_tick(&now, bev->rate_limiting->cfg); + ev_token_bucket_update(&bev->rate_limiting->limit, + bev->rate_limiting->cfg, tick); + max_so_far = LIM(bev->rate_limiting->limit); + } + if (bev->rate_limiting->group) { + struct bufferevent_rate_limit_group *g = + bev->rate_limiting->group; + ev_uint32_t share; + LOCK_GROUP(g); + if (GROUP_SUSPENDED(g)) { + /* We can get here if we failed to lock this + * particular bufferevent while suspending the whole + * group. */ + if (is_write) + bufferevent_suspend_write(&bev->bev, + BEV_SUSPEND_BW_GROUP); + else + bufferevent_suspend_read(&bev->bev, + BEV_SUSPEND_BW_GROUP); + share = 0; + } else { + /* XXXX probably we should divide among the active + * members, not the total members. */ + share = LIM(g->rate_limit) / g->n_members; + if (share < g->min_share) + share = g->min_share; + } + UNLOCK_GROUP(g); + CLAMPTO(share); + } + + return max_so_far; +} + +int +_bufferevent_get_read_max(struct bufferevent_private *bev) +{ + return _bufferevent_get_rlim_max(bev, 0); +} + +int +_bufferevent_get_write_max(struct bufferevent_private *bev) +{ + return _bufferevent_get_rlim_max(bev, 1); +} + +int +_bufferevent_decrement_read_buckets(struct bufferevent_private *bev, int bytes) +{ + /* need to hold lock on bev */ + if (!bev->rate_limiting) + return 0; + + if (bev->rate_limiting->cfg) { + bev->rate_limiting->limit.read_limit -= bytes; + if (bev->rate_limiting->limit.read_limit <= 0) { + bufferevent_suspend_read(&bev->bev, BEV_SUSPEND_BW); + event_add(&bev->rate_limiting->refill_bucket_event, + &bev->rate_limiting->cfg->tick_timeout); + } + } + + if (bev->rate_limiting->group) { + LOCK_GROUP(bev->rate_limiting->group); + bev->rate_limiting->group->rate_limit.read_limit -= bytes; + if (bev->rate_limiting->group->rate_limit.read_limit <= 0) { + _bev_group_suspend_reading(bev->rate_limiting->group); + } + UNLOCK_GROUP(bev->rate_limiting->group); + } + + return 0; +} + +int +_bufferevent_decrement_write_buckets(struct bufferevent_private *bev, int bytes) +{ + /* need to hold lock */ + if (!bev->rate_limiting) + return 0; + + if (bev->rate_limiting->cfg) { + bev->rate_limiting->limit.write_limit -= bytes; + if (bev->rate_limiting->limit.write_limit <= 0) { + bufferevent_suspend_write(&bev->bev, BEV_SUSPEND_BW); + event_add(&bev->rate_limiting->refill_bucket_event, + &bev->rate_limiting->cfg->tick_timeout); + } + } + + if (bev->rate_limiting->group) { + LOCK_GROUP(bev->rate_limiting->group); + bev->rate_limiting->group->rate_limit.write_limit -= bytes; + if (bev->rate_limiting->group->rate_limit.write_limit <= 0) { + _bev_group_suspend_writing(bev->rate_limiting->group); + } + UNLOCK_GROUP(bev->rate_limiting->group); + } + + return 0; +} + +/** Stop reading on every bufferevent in g */ +static int +_bev_group_suspend_reading(struct bufferevent_rate_limit_group *g) +{ + /* Needs group lock */ + struct bufferevent_private *bev; + g->read_suspended = 1; + g->pending_unsuspend_read = 0; + + /* Note that in this loop we call EVLOCK_TRY_LOCK instead of BEV_LOCK, + to prevent a deadlock. (Ordinarily, the group lock nests inside + the bufferevent locks. If we are unable to lock any individual + bufferevent, it will find out later when it looks at its limit + and sees that its group is suspended. + */ + TAILQ_FOREACH(bev, &g->members, rate_limiting->next_in_group) { + if (EVLOCK_TRY_LOCK(bev->lock)) { + bufferevent_suspend_read(&bev->bev, + BEV_SUSPEND_BW_GROUP); + EVLOCK_UNLOCK(bev->lock, 0); + } + } + return 0; +} + +/** Stop writing on every bufferevent in g */ +static int +_bev_group_suspend_writing(struct bufferevent_rate_limit_group *g) +{ + /* Needs group lock */ + struct bufferevent_private *bev; + g->write_suspended = 1; + g->pending_unsuspend_write = 0; + TAILQ_FOREACH(bev, &g->members, rate_limiting->next_in_group) { + if (EVLOCK_TRY_LOCK(bev->lock)) { + bufferevent_suspend_write(&bev->bev, + BEV_SUSPEND_BW_GROUP); + EVLOCK_UNLOCK(bev->lock, 0); + } + } + return 0; +} + +/** Timer callback invoked on a single bufferevent with one or more exhausted + buckets when they are ready to refill. */ +static void +_bev_refill_callback(evutil_socket_t fd, short what, void *arg) +{ + unsigned tick; + struct timeval now; + struct bufferevent_private *bev = arg; + int again = 0; + BEV_LOCK(&bev->bev); + if (!bev->rate_limiting || !bev->rate_limiting->cfg) { + BEV_UNLOCK(&bev->bev); + return; + } + + /* First, update the bucket */ + event_base_gettimeofday_cached(bev->bev.ev_base, &now); + tick = ev_token_bucket_get_tick(&now, + bev->rate_limiting->cfg); + ev_token_bucket_update(&bev->rate_limiting->limit, + bev->rate_limiting->cfg, + tick); + + /* Now unsuspend any read/write operations as appropriate. */ + if ((bev->read_suspended & BEV_SUSPEND_BW)) { + if (bev->rate_limiting->limit.read_limit > 0) + bufferevent_unsuspend_read(&bev->bev, BEV_SUSPEND_BW); + else + again = 1; + } + if ((bev->write_suspended & BEV_SUSPEND_BW)) { + if (bev->rate_limiting->limit.write_limit > 0) + bufferevent_unsuspend_write(&bev->bev, BEV_SUSPEND_BW); + else + again = 1; + } + if (again) { + /* One or more of the buckets may need another refill if they + started negative. + + XXXX if we need to be quiet for more ticks, we should + maybe figure out what timeout we really want. + */ + event_add(&bev->rate_limiting->refill_bucket_event, + &bev->rate_limiting->cfg->tick_timeout); + } + BEV_UNLOCK(&bev->bev); +} + +/** Helper: grab a random element from a bufferevent group. */ +static struct bufferevent_private * +_bev_group_random_element(struct bufferevent_rate_limit_group *group) +{ + int which; + struct bufferevent_private *bev; + + /* requires group lock */ + + if (!group->n_members) + return NULL; + + EVUTIL_ASSERT(! TAILQ_EMPTY(&group->members)); + + which = random() % group->n_members; + + bev = TAILQ_FIRST(&group->members); + while (which--) + bev = TAILQ_NEXT(bev, rate_limiting->next_in_group); + + return bev; +} + +/** Iterate over the elements of a rate-limiting group 'g' with a random + starting point, assigning each to the variable 'bev', and executing the + block 'block'. + + We do this in a half-baked effort to get fairness among group members. + XXX Round-robin or some kind of priority queue would be even more fair. + */ +#define FOREACH_RANDOM_ORDER(block) \ + do { \ + first = _bev_group_random_element(g); \ + for (bev = first; bev != TAILQ_END(&g->members); \ + bev = TAILQ_NEXT(bev, rate_limiting->next_in_group)) { \ + block ; \ + } \ + for (bev = TAILQ_FIRST(&g->members); bev && bev != first; \ + bev = TAILQ_NEXT(bev, rate_limiting->next_in_group)) { \ + block ; \ + } \ + } while (0) + +/** Callback invoked every tick to add more elements to the group bucket + and unsuspend group members as needed. + */ +static void +_bev_group_refill_callback(evutil_socket_t fd, short what, void *arg) +{ + struct bufferevent_rate_limit_group *g = arg; + unsigned tick; + struct timeval now; + int again = 0; + struct bufferevent_private *bev, *first; + + event_base_gettimeofday_cached(event_get_base(&g->master_refill_event), &now); + + LOCK_GROUP(g); + tick = ev_token_bucket_get_tick(&now, &g->rate_limit_cfg); + ev_token_bucket_update(&g->rate_limit, &g->rate_limit_cfg, tick); + + if (g->pending_unsuspend_read || + (g->read_suspended && (g->rate_limit.read_limit >= g->min_share))) { + g->read_suspended = 0; + again = 0; + FOREACH_RANDOM_ORDER({ + if (EVLOCK_TRY_LOCK(bev->lock)) { + bufferevent_unsuspend_read(&bev->bev, + BEV_SUSPEND_BW_GROUP); + EVLOCK_UNLOCK(bev->lock, 0); + } else { + again = 1; + } + }); + g->pending_unsuspend_read = again; + } + if (g->pending_unsuspend_write || + (g->write_suspended && (g->rate_limit.write_limit >= g->min_share))){ + g->write_suspended = 0; + again = 0; + FOREACH_RANDOM_ORDER({ + if (EVLOCK_TRY_LOCK(bev->lock)) { + bufferevent_unsuspend_write(&bev->bev, + BEV_SUSPEND_BW_GROUP); + EVLOCK_UNLOCK(bev->lock, 0); + } else { + again = 1; + } + }); + g->pending_unsuspend_write = again; + } + + /* XXXX Rather than waiting to the next tick to unsuspend stuff + * with pending_unsuspend_write/read, we should do it on the + * next iteration of the mainloop. + */ + + UNLOCK_GROUP(g); +} + +int +bufferevent_set_rate_limit(struct bufferevent *bev, + struct ev_token_bucket_cfg *cfg) +{ + struct bufferevent_private *bevp = + EVUTIL_UPCAST(bev, struct bufferevent_private, bev); + int r = -1; + struct bufferevent_rate_limit *rlim; + struct timeval now; + ev_uint32_t tick; + /* XXX reference-count cfg */ + + BEV_LOCK(bev); + + if (cfg == NULL) { + if (bevp->rate_limiting) { + bevp->rate_limiting->cfg = NULL; + bufferevent_unsuspend_read(bev, BEV_SUSPEND_BW); + bufferevent_unsuspend_write(bev, BEV_SUSPEND_BW); + } + r = 0; + goto done; + } + + event_base_gettimeofday_cached(bev->ev_base, &now); + tick = ev_token_bucket_get_tick(&now, cfg); + + if (bevp->rate_limiting && bevp->rate_limiting->cfg == cfg) { + ; + } else if (bevp->rate_limiting) { + bevp->rate_limiting->cfg = cfg; + ev_token_bucket_init(&bevp->rate_limiting->limit, cfg, tick, 1); + if (bevp->rate_limiting->limit.read_limit > 0) + bufferevent_unsuspend_read(bev, BEV_SUSPEND_BW); + else + bufferevent_suspend_read(bev, BEV_SUSPEND_BW); + if (bevp->rate_limiting->limit.write_limit > 0) + bufferevent_unsuspend_write(bev, BEV_SUSPEND_BW); + else + bufferevent_suspend_write(bev, BEV_SUSPEND_BW); + } else { + rlim = mm_calloc(1, sizeof(struct bufferevent_rate_limit)); + if (!rlim) + goto done; + rlim->cfg = cfg; + ev_token_bucket_init(&rlim->limit, cfg, tick, 0); + evtimer_assign(&rlim->refill_bucket_event, bev->ev_base, + _bev_refill_callback, bevp); + bevp->rate_limiting = rlim; + } + r = 0; +done: + BEV_UNLOCK(bev); + return r; +} + +struct bufferevent_rate_limit_group * +bufferevent_rate_limit_group_new(struct event_base *base, + const struct ev_token_bucket_cfg *cfg) +{ + struct bufferevent_rate_limit_group *g; + struct timeval now; + ev_uint32_t tick; + + event_base_gettimeofday_cached(base, &now); + tick = ev_token_bucket_get_tick(&now, cfg); + + g = mm_calloc(1, sizeof(struct bufferevent_rate_limit_group)); + if (!g) + return NULL; + memcpy(&g->rate_limit_cfg, cfg, sizeof(g->rate_limit_cfg)); + TAILQ_INIT(&g->members); + + ev_token_bucket_init(&g->rate_limit, cfg, tick, 0); + + g->min_share = 64; + event_assign(&g->master_refill_event, base, -1, EV_PERSIST, + _bev_group_refill_callback, g); + event_add(&g->master_refill_event, &cfg->tick_timeout); + + EVTHREAD_ALLOC_LOCK(g->lock, EVTHREAD_LOCKTYPE_RECURSIVE); + return g; +} + +int +bufferevent_add_to_rate_limit_group(struct bufferevent *bev, + struct bufferevent_rate_limit_group *g) +{ + int wsuspend, rsuspend; + struct bufferevent_private *bevp = + EVUTIL_UPCAST(bev, struct bufferevent_private, bev); + BEV_LOCK(bev); + + if (!bevp->rate_limiting) { + struct bufferevent_rate_limit *rlim; + rlim = mm_calloc(1, sizeof(struct bufferevent_rate_limit)); + if (!rlim) { + BEV_UNLOCK(bev); + return -1; + } + evtimer_assign(&rlim->refill_bucket_event, bev->ev_base, + _bev_refill_callback, bevp); + bevp->rate_limiting = rlim; + } + + if (bevp->rate_limiting->group == g) { + BEV_UNLOCK(bev); + return 0; + } + if (bevp->rate_limiting->group) + bufferevent_remove_from_rate_limit_group(bev); + + LOCK_GROUP(g); + bevp->rate_limiting->group = g; + ++g->n_members; + TAILQ_INSERT_TAIL(&g->members, bevp, rate_limiting->next_in_group); + + rsuspend = g->read_suspended; + wsuspend = g->write_suspended; + + UNLOCK_GROUP(g); + + if (rsuspend) + bufferevent_suspend_read(bev, BEV_SUSPEND_BW_GROUP); + if (wsuspend) + bufferevent_suspend_write(bev, BEV_SUSPEND_BW_GROUP); + + BEV_UNLOCK(bev); + return 0; +} + +int +bufferevent_remove_from_rate_limit_group(struct bufferevent *bev) +{ + struct bufferevent_private *bevp = + EVUTIL_UPCAST(bev, struct bufferevent_private, bev); + BEV_LOCK(bev); + if (bevp->rate_limiting && bevp->rate_limiting->group) { + struct bufferevent_rate_limit_group *g = + bevp->rate_limiting->group; + LOCK_GROUP(g); + bevp->rate_limiting->group = NULL; + --g->n_members; + TAILQ_REMOVE(&g->members, bevp, rate_limiting->next_in_group); + UNLOCK_GROUP(g); + } + bufferevent_unsuspend_read(bev, BEV_SUSPEND_BW_GROUP); + bufferevent_unsuspend_write(bev, BEV_SUSPEND_BW_GROUP); + BEV_UNLOCK(bev); + return 0; +} diff --git a/bufferevent_sock.c b/bufferevent_sock.c index 61a369f6..f53689ee 100644 --- a/bufferevent_sock.c +++ b/bufferevent_sock.c @@ -118,10 +118,12 @@ static void bufferevent_readcb(evutil_socket_t fd, short event, void *arg) { struct bufferevent *bufev = arg; + struct bufferevent_private *bufev_p = + EVUTIL_UPCAST(bufev, struct bufferevent_private, bev); struct evbuffer *input; int res = 0; short what = BEV_EVENT_READING; - int howmuch = -1; + int howmuch = -1, readmax=-1; _bufferevent_incref_and_lock(bufev); @@ -144,6 +146,12 @@ bufferevent_readcb(evutil_socket_t fd, short event, void *arg) goto done; } } + readmax = _bufferevent_get_read_max(bufev_p); + if (howmuch < 0 || howmuch > readmax) /* The use of -1 for "unlimited" + * uglifies this code. */ + howmuch = readmax; + if (bufev_p->read_suspended) + goto done; evbuffer_unfreeze(input, 0); res = evbuffer_read(input, fd, howmuch); @@ -163,6 +171,7 @@ bufferevent_readcb(evutil_socket_t fd, short event, void *arg) if (res <= 0) goto error; + _bufferevent_decrement_read_buckets(bufev_p, res); /* Invoke the user callback - must always be called last */ if (evbuffer_get_length(input) >= bufev->wm_read.low && @@ -191,6 +200,7 @@ bufferevent_writecb(evutil_socket_t fd, short event, void *arg) int res = 0; short what = BEV_EVENT_WRITING; int connected = 0; + int atmost = -1; _bufferevent_incref_and_lock(bufev); @@ -232,9 +242,14 @@ bufferevent_writecb(evutil_socket_t fd, short event, void *arg) } } + atmost = _bufferevent_get_write_max(bufev_p); + + if (bufev_p->write_suspended) + goto done; + if (evbuffer_get_length(bufev->output)) { evbuffer_unfreeze(bufev->output, 1); - res = evbuffer_write(bufev->output, fd); + res = evbuffer_write_atmost(bufev->output, fd, atmost); evbuffer_freeze(bufev->output, 1); if (res == -1) { int err = evutil_socket_geterror(fd); @@ -250,6 +265,8 @@ bufferevent_writecb(evutil_socket_t fd, short event, void *arg) } if (res <= 0) goto error; + + _bufferevent_decrement_write_buckets(bufev_p, res); } if (evbuffer_get_length(bufev->output) == 0) diff --git a/include/event2/bufferevent.h b/include/event2/bufferevent.h index 523a7fce..48d56bd6 100644 --- a/include/event2/bufferevent.h +++ b/include/event2/bufferevent.h @@ -499,6 +499,97 @@ int bufferevent_pair_new(struct event_base *base, int options, struct bufferevent *pair[2]); + +/** + Abstract type used to configure rate-limiting on a bufferevent or a group + of bufferevents. + */ +struct ev_token_bucket_cfg; +/** + A group of bufferevents which are configured to respect the same rate + limit. +*/ +struct bufferevent_rate_limit_group; + +/** + Initialize and return a new object to configure the rate-limiting behavior + of bufferevents. + + @param read_rate The maximum number of bytes to read per tick on + average. + @param read_burst The maximum number of bytes to read in any single tick. + @param write_rate The maximum number of bytes to write per tick on + average. + @param write_burst The maximum number of bytes to write in any single tick. + @param tick_len The length of a single tick. Defaults to one second. + Any fractions of a millisecond are ignored. + + Note that all rate-limits hare are currently best-effort: future versions + of Libevent may implement them more tightly. + */ +struct ev_token_bucket_cfg *ev_token_bucket_cfg_new( + ev_uint32_t read_rate, ev_uint32_t read_burst, + ev_uint32_t write_rate, ev_uint32_t write_burst, + const struct timeval *tick_len); + +/** Free all storage held in 'cfg'. + + Note: 'cfg' is not currently reference-counted; it is not safe to free it + until no bufferevent is using it. + */ +void ev_token_bucket_cfg_free(struct ev_token_bucket_cfg *cfg); + +/** + Set the rate-limit of a the bufferevent 'bev' to the one specified in + 'cfg'. If 'cfg' is NULL, disable any per-bufferevent rate-limiting on + 'bev'. + + Note that only some bufferevent types currently respect rate-limiting. + They are: socket-based bufferevents (normal and IOCP-based), and SSL-based + bufferevents. + + Return 0 on sucess, -1 on failure. + */ +int bufferevent_set_rate_limit(struct bufferevent *bev, + struct ev_token_bucket_cfg *cfg); +/** + Create a new rate-limit group for bufferevents. A rate-limit group + constrains the maximum number of bytes sent and received, in toto, + by all of its bufferevents. + + @param base An event_base to run any necessary timeouts for the group. + Note that all bufferevents in the group do not necessarily need to share + this event_base. + @param cfg The rate-limit for this group. + + Note that all rate-limits hare are currently best-effort: future versions + of Libevent may implement them more tightly. + + Note also that only some bufferevent types currently respect rate-limiting. + They are: socket-based bufferevents (normal and IOCP-based), and SSL-based + bufferevents. + */ +struct bufferevent_rate_limit_group *bufferevent_rate_limit_group_new( + struct event_base *base, + const struct ev_token_bucket_cfg *cfg); +/*XXX we need a bufferevent_rate_limit_group_set_cfg */ + +/** + Add 'bev' to the list of bufferevents whose aggregate reading and writing + is restricted by 'g'. If 'g' is NULL, remove 'bev' from its current group. + + A bufferevent may belong to no more than one rate-limit group at a time. + If 'bev' is already a member of a group, it will be removed from its old + group before being added to 'g'. + + Return 0 on success and -1 on failure. + */ +int bufferevent_add_to_rate_limit_group(struct bufferevent *bev, + struct bufferevent_rate_limit_group *g); + +/** Remove 'bev' from its current rate-limit group (if any). */ +int bufferevent_remove_from_rate_limit_group(struct bufferevent *bev); + #ifdef __cplusplus } #endif diff --git a/include/event2/util.h b/include/event2/util.h index 98c26b61..e50e434b 100644 --- a/include/event2/util.h +++ b/include/event2/util.h @@ -93,12 +93,16 @@ extern "C" { #ifdef _EVENT_HAVE_UINT32_T #define ev_uint32_t uint32_t +#define ev_int32_t int32_t #elif defined(WIN32) #define ev_uint32_t unsigned int +#define ev_int32_t signed int #elif _EVENT_SIZEOF_LONG == 4 #define ev_uint32_t unsigned long +#define ev_int32_t signed long #elif _EVENT_SIZEOF_INT == 4 #define ev_uint32_t unsigned int +#define ev_int32_t signed int #else #error "No way to define ev_uint32_t" #endif diff --git a/ratelim-internal.h b/ratelim-internal.h new file mode 100644 index 00000000..105798ae --- /dev/null +++ b/ratelim-internal.h @@ -0,0 +1,102 @@ +/* + * Copyright (c) 2009 Niels Provos and Nick Mathewson + * + * 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 _RATELIM_INTERNAL_H_ +#define _RATELIM_INTERNAL_H_ + +#ifdef __cplusplus +extern "C" { +#endif + +#include + +/** A token bucket is an internal structure that tracks how many bytes we are + * currently willing to read or write on a given bufferevent or group of + * bufferevents */ +struct ev_token_bucket { + /** How many bytes are we willing to read or write right now? These + * values are signed so that we can do "defecit spending" */ + ev_int32_t read_limit, write_limit; + /** When was this bucket last updated? Measured in abstract 'ticks' + * relative to the token bucket configuration. */ + ev_uint32_t last_updated; +}; + +/** Configuration info for a token bucket or set of token buckets. */ +struct ev_token_bucket_cfg { + /** How many bytes are we willing to read on average per tick? */ + ev_uint32_t read_rate; + /** How many bytes are we willing to read at most in any one tick? */ + ev_uint32_t read_maximum; + /** How many bytes are we willing to write on average per tick? */ + ev_uint32_t write_rate; + /** How many bytes are we willing to write at most in any one tick? */ + ev_uint32_t write_maximum; + + /* How long is a tick? Note that fractions of a millisecond are + * ignored. */ + struct timeval tick_timeout; + + /* How long is a tick, in milliseconds? Derived from tick_timeout. */ + unsigned msec_per_tick; +}; + +/** The current tick is 'current_tick': add bytes to 'bucket' as specified in + * 'cfg'. */ +int ev_token_bucket_update(struct ev_token_bucket *bucket, + const struct ev_token_bucket_cfg *cfg, + ev_uint32_t current_tick); + +/** In which tick does 'tv' fall according to 'cfg'? Note that ticks can + * overflow easily; your code needs to handle this. */ +ev_uint32_t ev_token_bucket_get_tick(const struct timeval *tv, + const struct ev_token_bucket_cfg *cfg); + +/** Adjust 'bucket' to respect 'cfg', and note that it was last updated in + * 'current_tick'. If 'reinitialize' is true, we are changing the + * configuration of 'bucket'; otherwise, we are setting it up for the first + * time. + */ +int ev_token_bucket_init(struct ev_token_bucket *bucket, + const struct ev_token_bucket_cfg *cfg, + ev_uint32_t current_tick, + int reinitialize); + +/** Decrease the read limit of 'b' by 'n' bytes */ +#define ev_token_bucket_decrement_read(b,n) \ + do { \ + (b)->read_limit -= (n); \ + } while (0) +/** Decrease the write limit of 'b' by 'n' bytes */ +#define ev_token_bucket_decrement_write(b,n) \ + do { \ + (b)->write_limit -= (n); \ + } while (0) + +#ifdef __cplusplus +} +#endif + +#endif