internals/HASH.md \
internals/LLIST.md \
internals/MQTT.md \
+ internals/MULTI-EV.md \
internals/NEW-PROTOCOL.md \
internals/README.md \
internals/SPLAY.md \
--- /dev/null
+<!--
+Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
+
+SPDX-License-Identifier: curl
+-->
+
+# Multi Event Based
+
+A libcurl multi is operating "event based" when the application uses
+and event library like `libuv` to monitor the sockets and file descriptors
+libcurl uses to trigger transfer operations. How that works from the
+applications point of view is described in libcurl-multi(3).
+
+This documents is about the internal handling.
+
+## Source Locations
+
+All code related to event based handling is found in `lib/multi_ev.c`
+and `lib/multi_ev.h`. The header defines a set of internal functions
+and `struct curl_multi_ev` that is embedded in each multi handle.
+
+There is `Curl_multi_ev_init()` and `Curl_multi_ev_cleanup()` to manage
+the overall life cycle, call on creation and destruction of the multi
+handle.
+
+## Tracking Events
+
+First, the various functions in `lib/multi_ev.h` only ever really do
+something when the libcurl application has registered its callback
+in `multi->socket_cb`.
+
+This is important as this callback gets informed about *changes* to sockets.
+When a new socket is added, an existing is removed, or the `POLLIN/OUT`
+flags change, `multi->socket_cb` needs to be invoked. `multi_ev` has to
+track what it already reported to detect changes.
+
+Most applications are expected to go "event based" right from the start,
+but the libcurl API does not prohibit an application to start another
+way and then go for events later on, even in the middle of a transfer.
+
+### Transfer Events
+
+Most event that happen are in connection with a transfer. A transfer
+opens a connection, which opens a socket, and waits for this socket
+to become writable (`POLLOUT`) when using TCP, for example.
+
+The multi then calls `Curl_multi_ev_assess_xfer(multi, data)` to
+let the multi event code detect what sockets the transfer is interested in.
+If indeed a `multi->socket_cb` is set, the *current* transfer pollset is
+retrieved via `Curl_multi_getsock()`. This current pollset is then
+compared to the *previous* pollset. If relevant changes are detected,
+`multi->socket_cb` gets informed about those. These can be:
+
+ * a socket is in the current set, but not the previous one
+ * a socket was also in the previous one, but IN/OUT flags changed
+ * a socket in the previous one is no longer part of the current
+
+`multi_ev.c` keeps a `struct mev_sh_entry` for each sockets in a hash
+with the socket as key. It tracks in each entry which transfers are
+interested in this particular socket. How many transfer want to read
+and/or write and what the summarized `POLLIN/POLLOUT` action, that
+had been reported to `multi->socket_cb` was.
+
+This is necessary as a socket may be in use by several transfers
+at the same time (think HTTP/2 on the same connection). When a transfer
+is done and gets removed from the socket entry, it decrements
+the reader and/or writer count (depending on what it was last
+interested in). This *may* result in the entry's summarized action
+to change, or not.
+
+### Connection Events
+
+There are also events not connected to any transfer that need to be tracked.
+The multi connection cache, concerned with clean shutdowns of connections,
+is interested in socket events during the shutdown.
+
+To allow use of the libcurl infrastructure, the connection cache operates
+using an *internal* easy handle that is not a transfer as such. The
+internal handle is used for all connection shutdown operations, being tied
+to a particular connection only for a short time. This means tracking
+the last pollset for an internal handle is useless.
+
+Instead, the connection cache uses `Curl_multi_ev_assess_conn()` to have
+multi event handling check the connection and track a "last pollset"
+for the connection alone.
+
+## Event Processing
+
+When the libcurl application is informed by the event library that
+a particular socket has an event, it calls `curl_multi_socket_action()`
+to make libcurl react to it. This internally invokes
+`Curl_multi_ev_expire_xfers()` which expires all transfers that
+are interested in the given socket, so the multi handle runs them.
+
+In addition `Curl_multi_ev_expire_xfers()` returns a `bool` to let
+the multi know that connections are also interested in the socket, so
+the connection pool should be informed as well.
+
+## All Things Pass
+
+When a transfer is done, e.g. removed from its multi handle, the
+multi calls `Curl_multi_ev_xfer_done()`. This cleans up the pollset
+tracking for the transfer.
+
+When a connection is done, and before it is destroyed,
+`Curl_multi_ev_conn_done()` is called. This cleans up the pollset
+tracking for this connection.
+
+When a socket is about to be closed, `Curl_multi_ev_socket_done()`
+is called to cleanup the socket entry and all information kept there.
+
+These calls do not have to happen in any particular order. A transfer's
+socket may be around while the transfer is ongoing. Or it might disappear
+in the middle of things. Also, a transfer might be interested in several
+sockets at the same time (resolving, eye balling, ftp are all examples of
+those).
+
+### And Come Again
+
+While transfer and connection identifier are practically unique in a
+libcurl application, sockets are not. Operating systems are keen on reusing
+their resources, and the next socket may get the same identifier as
+one just having been closed with high likelihood.
+
+This means that multi event handling needs to be informed *before* a close,
+clean up all its tracking and be ready to see that same socket identifier
+again right after.
Tracing of DNS operations to resolve hostnames and HTTPS records.
+## `lib-ids`
+
+Adds transfer and connection identifiers as prefix to every call to
+CURLOPT_DEBUGFUNCTION(3). The format is `[n-m]` where `n` is the identifier
+of the transfer and `m` is the identifier of the connection. A literal `x`
+is used for internal transfers or when no connection is assigned.
+
+For example, `[5-x]` is the prefix for transfer 5 that has no
+connection. The command line tool `curl`uses the same format for its
+`--trace-ids` option.
+
+`lib-ids` is intended for libcurl applications that handle multiple
+transfers but have no own way to identify in trace output which transfer
+a trace event is connected to.
+
## `doh`
Former name for DNS-over-HTTP operations. Now an alias for `dns`.
+## `multi`
+
+Traces multi operations managing transfers' state changes and sockets poll
+states.
+
## `read`
-Traces reading of upload data from the application in order to send it to the server.
+Traces reading of upload data from the application in order to send it to the
+server.
## `ssls`
mprintf.c \
mqtt.c \
multi.c \
+ multi_ev.c \
netrc.c \
nonblock.c \
noproxy.c \
mime.h \
mqtt.h \
multihandle.h \
+ multi_ev.h \
multiif.h \
netrc.h \
nonblock.h \
struct Curl_easy *easy = data;
if(!readable && !writable) {
DEBUGASSERT(easy);
- Curl_multi_closed(easy, socket_fd);
+ Curl_multi_will_close(easy, socket_fd);
}
}
* ensure CURLMOPT_SOCKETFUNCTION fires CURL_POLL_REMOVE
* before the FD is invalidated to avoid EBADF on EPOLL_CTL_DEL
*/
- Curl_multi_closed(data, sock_rd);
+ Curl_multi_will_close(data, sock_rd);
wakeup_close(sock_rd);
#endif
/* Curl_resolver_getsock()
*
- * This function is called from the multi_getsock() function. 'sock' is a
+ * This function is called from the Curl_multi_getsock() function. 'sock' is a
* pointer to an array to hold the file descriptors, with 'numsock' being the
* size of that array (in number of entries). This function is supposed to
* return bitmask indicating what file descriptors (referring to array indexes
if(use_callback && conn && conn->fclosesocket) {
int rc;
- Curl_multi_closed(data, sock);
+ Curl_multi_will_close(data, sock);
Curl_set_in_callback(data, TRUE);
rc = conn->fclosesocket(conn->closesocket_client, sock);
Curl_set_in_callback(data, FALSE);
if(conn)
/* tell the multi-socket code about this */
- Curl_multi_closed(data, sock);
+ Curl_multi_will_close(data, sock);
sclose(sock);
struct cf_socket_ctx *ctx = cf->ctx;
if(ctx && CURL_SOCKET_BAD != ctx->sock) {
- CURL_TRC_CF(data, cf, "cf_socket_close(%" FMT_SOCKET_T ")", ctx->sock);
+ CURL_TRC_CF(data, cf, "cf_socket_close, fd=%" FMT_SOCKET_T, ctx->sock);
if(ctx->sock == cf->conn->sock[cf->sockindex])
cf->conn->sock[cf->sockindex] = CURL_SOCKET_BAD;
socket_close(data, cf->conn, !ctx->accepted, ctx->sock);
if(cf->connected) {
struct cf_socket_ctx *ctx = cf->ctx;
- CURL_TRC_CF(data, cf, "cf_socket_shutdown(%" FMT_SOCKET_T ")", ctx->sock);
+ CURL_TRC_CF(data, cf, "cf_socket_shutdown, fd=%" FMT_SOCKET_T, ctx->sock);
/* On TCP, and when the socket looks well and non-blocking mode
* can be enabled, receive dangling bytes before close to avoid
* entering RST states unnecessarily. */
*done = FALSE;
now = Curl_now();
if(!Curl_shutdown_started(data, sockindex)) {
- DEBUGF(infof(data, "shutdown start on%s connection",
- sockindex ? " secondary" : ""));
+ CURL_TRC_M(data, "shutdown start on%s connection",
+ sockindex ? " secondary" : "");
Curl_shutdown_start(data, sockindex, &now);
}
else {
/* In general, we want to send after connect, wait on that. */
if(sockfd != CURL_SOCKET_BAD)
Curl_pollset_set_out_only(data, &ps, sockfd);
- Curl_conn_adjust_pollset(data, &ps);
+ Curl_conn_adjust_pollset(data, data->conn, &ps);
result = Curl_pollfds_add_ps(&cpfds, &ps);
if(result)
goto out;
}
void Curl_conn_adjust_pollset(struct Curl_easy *data,
- struct easy_pollset *ps)
+ struct connectdata *conn,
+ struct easy_pollset *ps)
{
int i;
DEBUGASSERT(data);
- DEBUGASSERT(data->conn);
+ DEBUGASSERT(conn);
for(i = 0; i < 2; ++i) {
- Curl_conn_cf_adjust_pollset(data->conn->cfilter[i], data, ps);
+ Curl_conn_cf_adjust_pollset(conn->cfilter[i], data, ps);
}
}
* Adjust pollset from filters installed at transfer's connection.
*/
void Curl_conn_adjust_pollset(struct Curl_easy *data,
- struct easy_pollset *ps);
+ struct connectdata *conn,
+ struct easy_pollset *ps);
/**
* Curl_poll() the filter chain at `cf` with timeout `timeout_ms`.
#include "cfilters.h"
#include "progress.h"
#include "multiif.h"
+#include "multi_ev.h"
#include "sendf.h"
#include "conncache.h"
#include "http_negotiate.h"
bool *done);
static void cpool_run_conn_shutdown_handler(struct Curl_easy *data,
struct connectdata *conn);
-static CURLMcode cpool_update_shutdown_ev(struct Curl_multi *multi,
- struct Curl_easy *data,
+static CURLMcode cpool_update_shutdown_ev(struct cpool *cpool,
+ struct Curl_multi *multi,
struct connectdata *conn);
static void cpool_shutdown_all(struct cpool *cpool,
struct Curl_easy *data, int timeout_ms);
}
int Curl_cpool_init(struct cpool *cpool,
- Curl_cpool_disconnect_cb *disconnect_cb,
- struct Curl_multi *multi,
- struct Curl_share *share,
- size_t size)
+ Curl_cpool_disconnect_cb *disconnect_cb,
+ struct Curl_multi *multi,
+ struct Curl_share *share,
+ size_t size)
{
DEBUGASSERT(!!multi != !!share); /* either one */
Curl_hash_init(&cpool->dest2bundle, size, Curl_hash_str,
if(!oldest_idle)
break;
/* disconnect the old conn and continue */
- DEBUGF(infof(data, "Discarding connection #%"
+ CURL_TRC_M(data, "Discarding connection #%"
FMT_OFF_T " from %zu to reach destination "
"limit of %zu", oldest_idle->connection_id,
- Curl_llist_count(&bundle->conns), dest_limit));
+ Curl_llist_count(&bundle->conns), dest_limit);
Curl_cpool_disconnect(data, oldest_idle, FALSE);
/* in case the bundle was destroyed in disconnect, look it up again */
if(!oldest_idle)
break;
/* disconnect the old conn and continue */
- DEBUGF(infof(data, "Discarding connection #%"
- FMT_OFF_T " from %zu to reach total "
- "limit of %zu",
- oldest_idle->connection_id, cpool->num_conn, total_limit));
+ CURL_TRC_M(data, "Discarding connection #%"
+ FMT_OFF_T " from %zu to reach total "
+ "limit of %zu",
+ oldest_idle->connection_id, cpool->num_conn, total_limit);
Curl_cpool_disconnect(data, oldest_idle, FALSE);
shutdowns = Curl_llist_count(&cpool->shutdowns);
}
while(e) {
conn = Curl_node_elem(e);
Curl_node_remove(e);
- DEBUGF(infof(cpool->idata, "discard connection #%" FMT_OFF_T,
- conn->connection_id));
cpool_close_and_destroy(cpool, conn, NULL, FALSE);
e = Curl_llist_head(&cpool->shutdowns);
}
* are other users of it
*/
if(CONN_INUSE(conn) && !aborted) {
- DEBUGF(infof(data, "[CCACHE] not discarding #%" FMT_OFF_T
- " still in use by %zu transfers", conn->connection_id,
- CONN_INUSE(conn)));
+ CURL_TRC_M(data, "[CPOOL] not discarding #%" FMT_OFF_T
+ " still in use by %zu transfers", conn->connection_id,
+ CONN_INUSE(conn));
return;
}
/* Attempt to shutdown the connection right away. */
Curl_attach_connection(data, conn);
cpool_run_conn_shutdown(data, conn, &done);
- DEBUGF(infof(data, "[CCACHE] shutdown #%" FMT_OFF_T ", done=%d",
- conn->connection_id, done));
+ CURL_TRC_M(data, "[CPOOL] shutdown, done=%d", done);
Curl_detach_connection(data);
}
if(data->multi && data->multi->max_total_connections > 0 &&
(data->multi->max_total_connections <=
(long)(cpool->num_conn + Curl_llist_count(&cpool->shutdowns)))) {
- DEBUGF(infof(data, "[CCACHE] discarding oldest shutdown connection "
- "due to connection limit of %ld",
- data->multi->max_total_connections));
+ CURL_TRC_M(data, "[CPOOL] discarding oldest shutdown connection "
+ "due to connection limit of %ld",
+ data->multi->max_total_connections);
cpool_shutdown_destroy_oldest(cpool);
}
if(data->multi && data->multi->socket_cb) {
DEBUGASSERT(cpool == &data->multi->cpool);
- /* Start with an empty shutdown pollset, so out internal closure handle
- * is added to the sockets. */
- memset(&conn->shutdown_poll, 0, sizeof(conn->shutdown_poll));
- if(cpool_update_shutdown_ev(data->multi, cpool->idata, conn)) {
- DEBUGF(infof(data, "[CCACHE] update events for shutdown failed, "
- "discarding #%" FMT_OFF_T,
- conn->connection_id));
+ if(cpool_update_shutdown_ev(cpool, data->multi, conn)) {
+ CURL_TRC_M(data, "[CPOOL] update events failed, discarding #%"
+ FMT_OFF_T, conn->connection_id);
cpool_close_and_destroy(cpool, conn, data, FALSE);
return;
}
}
Curl_llist_append(&cpool->shutdowns, conn, &conn->cpool_node);
- DEBUGF(infof(data, "[CCACHE] added #%" FMT_OFF_T
- " to shutdowns, now %zu conns in shutdown",
- conn->connection_id, Curl_llist_count(&cpool->shutdowns)));
+ CURL_TRC_M(data, "[CPOOL] added #%" FMT_OFF_T
+ " to shutdowns, now %zu conns in shutdown",
+ conn->connection_id, Curl_llist_count(&cpool->shutdowns));
}
void Curl_cpool_disconnect(struct Curl_easy *data,
conn = Curl_node_elem(e);
memset(&ps, 0, sizeof(ps));
Curl_attach_connection(cpool->idata, conn);
- Curl_conn_adjust_pollset(cpool->idata, &ps);
+ Curl_conn_adjust_pollset(cpool->idata, conn, &ps);
Curl_detach_connection(cpool->idata);
result = Curl_pollfds_add_ps(cpfds, &ps);
conn = Curl_node_elem(e);
memset(&ps, 0, sizeof(ps));
Curl_attach_connection(cpool->idata, conn);
- Curl_conn_adjust_pollset(cpool->idata, &ps);
+ Curl_conn_adjust_pollset(cpool->idata, conn, &ps);
Curl_detach_connection(cpool->idata);
need += Curl_waitfds_add_ps(cwfds, &ps);
struct connectdata *conn = Curl_node_elem(e);
memset(&ps, 0, sizeof(ps));
Curl_attach_connection(cpool->idata, conn);
- Curl_conn_adjust_pollset(cpool->idata, &ps);
+ Curl_conn_adjust_pollset(cpool->idata, conn, &ps);
Curl_detach_connection(cpool->idata);
for(i = 0; i < ps.num; i++) {
return;
DEBUGASSERT(data);
- DEBUGF(infof(data, "[CCACHE] perform, %zu connections being shutdown",
- Curl_llist_count(&cpool->shutdowns)));
+ CURL_TRC_M(data, "[CPOOL] perform, %zu connections being shutdown",
+ Curl_llist_count(&cpool->shutdowns));
while(e) {
enext = Curl_node_next(e);
conn = Curl_node_elem(e);
Curl_attach_connection(data, conn);
cpool_run_conn_shutdown(data, conn, &done);
- DEBUGF(infof(data, "[CCACHE] shutdown #%" FMT_OFF_T ", done=%d",
- conn->connection_id, done));
+ CURL_TRC_M(data, "[CPOOL] shutdown, done=%d", done);
Curl_detach_connection(data);
if(done) {
Curl_node_remove(e);
Curl_expire(data, next_from_now_ms, EXPIRE_RUN_NOW);
}
-void Curl_cpool_multi_perform(struct Curl_multi *multi)
-{
- CPOOL_LOCK(&multi->cpool);
- cpool_perform(&multi->cpool);
- CPOOL_UNLOCK(&multi->cpool);
-}
-
-
/*
* Close and destroy the connection. Run the shutdown sequence once,
* of so requested.
}
if(cpool)
- DEBUGF(infof(data, "[CCACHE] closing #%" FMT_OFF_T,
- conn->connection_id));
+ CURL_TRC_M(data, "[CPOOL] closing connection");
else
DEBUGF(infof(data, "closing connection #%" FMT_OFF_T,
conn->connection_id));
Curl_conn_close(data, FIRSTSOCKET);
Curl_detach_connection(data);
+ if(cpool && cpool->multi)
+ Curl_multi_ev_conn_done(cpool->multi, data, conn);
+ else if(data->multi)
+ Curl_multi_ev_conn_done(data->multi, data, conn);
+
Curl_conn_free(data, conn);
if(cpool && cpool->multi) {
- DEBUGF(infof(data, "[CCACHE] trigger multi connchanged"));
+ CURL_TRC_M(data, "[CPOOL] trigger multi connchanged");
Curl_multi_connchanged(cpool->multi);
}
}
-static CURLMcode cpool_update_shutdown_ev(struct Curl_multi *multi,
- struct Curl_easy *data,
+static CURLMcode cpool_update_shutdown_ev(struct cpool *cpool,
+ struct Curl_multi *multi,
struct connectdata *conn)
{
- struct easy_pollset ps;
CURLMcode mresult;
- DEBUGASSERT(data);
+ DEBUGASSERT(cpool);
DEBUGASSERT(multi);
DEBUGASSERT(multi->socket_cb);
- memset(&ps, 0, sizeof(ps));
- Curl_attach_connection(data, conn);
- Curl_conn_adjust_pollset(data, &ps);
- Curl_detach_connection(data);
-
- mresult = Curl_multi_pollset_ev(multi, data, &ps, &conn->shutdown_poll);
-
- if(!mresult) /* Remember for next time */
- memcpy(&conn->shutdown_poll, &ps, sizeof(ps));
+ Curl_attach_connection(cpool->idata, conn);
+ mresult = Curl_multi_ev_assess_conn(multi, cpool->idata, conn);
+ Curl_detach_connection(cpool->idata);
return mresult;
}
-void Curl_cpool_multi_socket(struct Curl_multi *multi,
- curl_socket_t s, int ev_bitmask)
+static void cpool_multi_socket(struct Curl_multi *multi, curl_socket_t s)
{
struct cpool *cpool = &multi->cpool;
- struct Curl_easy *data = cpool->idata;
struct Curl_llist_node *e;
struct connectdata *conn;
bool done;
- (void)ev_bitmask;
DEBUGASSERT(multi->socket_cb);
CPOOL_LOCK(cpool);
e = Curl_llist_head(&cpool->shutdowns);
while(e) {
conn = Curl_node_elem(e);
if(s == conn->sock[FIRSTSOCKET] || s == conn->sock[SECONDARYSOCKET]) {
- Curl_attach_connection(data, conn);
- cpool_run_conn_shutdown(data, conn, &done);
- DEBUGF(infof(data, "[CCACHE] shutdown #%" FMT_OFF_T ", done=%d",
- conn->connection_id, done));
- Curl_detach_connection(data);
- if(done || cpool_update_shutdown_ev(multi, data, conn)) {
+ Curl_attach_connection(cpool->idata, conn);
+ cpool_run_conn_shutdown(cpool->idata, conn, &done);
+ CURL_TRC_M(cpool->idata, "[CPOOL] shutdown, done=%d", done);
+ Curl_detach_connection(cpool->idata);
+ if(done || cpool_update_shutdown_ev(cpool, multi, conn)) {
Curl_node_remove(e);
cpool_close_and_destroy(cpool, conn, NULL, FALSE);
}
CPOOL_UNLOCK(cpool);
}
+void Curl_cpool_multi_perform(struct Curl_multi *multi, curl_socket_t s)
+{
+ CPOOL_LOCK(&multi->cpool);
+ if((s == CURL_SOCKET_TIMEOUT) || (!multi->socket_cb))
+ cpool_perform(&multi->cpool);
+ else
+ cpool_multi_socket(multi, s);
+ CPOOL_UNLOCK(&multi->cpool);
+}
+
+
#define NUM_POLLS_ON_STACK 10
static CURLcode cpool_shutdown_wait(struct cpool *cpool, int timeout_ms)
return;
(void)data;
- DEBUGF(infof(data, "cpool shutdown all"));
+ CURL_TRC_M(data, "[CPOOL] shutdown all");
/* Move all connections into the shutdown queue */
for(conn = cpool_get_live_conn(cpool); conn;
conn = cpool_get_live_conn(cpool)) {
/* Move conn from live set to shutdown or destroy right away */
- DEBUGF(infof(data, "moving connection #%" FMT_OFF_T
- " to shutdown queue", conn->connection_id));
+ CURL_TRC_M(data, "[CPOOL] moving connection to shutdown queue");
cpool_remove_conn(cpool, conn);
cpool_discard_conn(cpool, data, conn, FALSE);
}
cpool_perform(cpool);
if(!Curl_llist_head(&cpool->shutdowns)) {
- DEBUGF(infof(data, "cpool shutdown ok"));
+ CURL_TRC_M(data, "[CPOOL] shutdown finished cleanly");
break;
}
/* wait for activity, timeout or "nothing" */
timespent = Curl_timediff(Curl_now(), started);
if(timespent >= (timediff_t)timeout_ms) {
- DEBUGF(infof(data, "cpool shutdown %s",
- (timeout_ms > 0) ? "timeout" : "best effort done"));
+ CURL_TRC_M(data, "[CPOOL] shutdown finished, %s",
+ (timeout_ms > 0) ? "timeout" : "best effort done");
break;
}
remain_ms = timeout_ms - (int)timespent;
if(cpool_shutdown_wait(cpool, remain_ms)) {
- DEBUGF(infof(data, "cpool shutdown all, abort"));
+ CURL_TRC_M(data, "[CPOOL] shutdown finished, aborted");
break;
}
}
int *maxfd);
/**
- * Perform maintenance on connections in the pool. Specifically,
- * progress the shutdown of connections in the queue.
+ * Run connections on socket. If socket is CURL_SOCKET_TIMEOUT, run
+ * maintenance on all connections.
*/
-void Curl_cpool_multi_perform(struct Curl_multi *multi);
-
-void Curl_cpool_multi_socket(struct Curl_multi *multi,
- curl_socket_t s, int ev_bitmask);
-
+void Curl_cpool_multi_perform(struct Curl_multi *multi, curl_socket_t s);
#endif /* HEADER_CURL_CONNCACHE_H */
#include "curl_memory.h"
#include "memdebug.h"
-void Curl_debug(struct Curl_easy *data, curl_infotype type,
- char *ptr, size_t size)
+static void trc_write(struct Curl_easy *data, curl_infotype type,
+ char *ptr, size_t size)
{
if(data->set.verbose) {
- static const char s_infotype[CURLINFO_END][3] = {
- "* ", "< ", "> ", "{ ", "} ", "{ ", "} " };
if(data->set.fdebug) {
bool inCallback = Curl_is_in_callback(data);
Curl_set_in_callback(data, TRUE);
Curl_set_in_callback(data, inCallback);
}
else {
+ static const char s_infotype[CURLINFO_END][3] = {
+ "* ", "< ", "> ", "{ ", "} ", "{ ", "} " };
switch(type) {
case CURLINFO_TEXT:
case CURLINFO_HEADER_OUT:
}
}
+/* max length we trace before ending in '...' */
+#define TRC_LINE_MAX 2048
+
+#define CURL_TRC_FMT_IDSC "[x-%" CURL_FORMAT_CURL_OFF_T "] "
+#define CURL_TRC_FMT_IDSD "[%" CURL_FORMAT_CURL_OFF_T "-x] "
+#define CURL_TRC_FMT_IDSDC "[%" CURL_FORMAT_CURL_OFF_T "-%" \
+ CURL_FORMAT_CURL_OFF_T "] "
+
+static struct curl_trc_feat Curl_trc_feat_ids = {
+ "LIB-IDS",
+ CURL_LOG_LVL_NONE,
+};
+#define CURL_TRC_IDS(data) \
+ (Curl_trc_is_verbose(data) && \
+ Curl_trc_feat_ids.log_level >= CURL_LOG_LVL_INFO)
+
+static size_t trc_print_ids(struct Curl_easy *data, char *buf, size_t maxlen)
+{
+ curl_off_t cid = data->conn ?
+ data->conn->connection_id : data->state.recent_conn_id;
+ if(data->id >= 0) {
+ if(cid >= 0)
+ return msnprintf(buf, maxlen, CURL_TRC_FMT_IDSDC, data->id, cid);
+ else
+ return msnprintf(buf, maxlen, CURL_TRC_FMT_IDSD, data->id);
+ }
+ else if(cid >= 0)
+ return msnprintf(buf, maxlen, CURL_TRC_FMT_IDSC, cid);
+ else {
+ return msnprintf(buf, maxlen, "[x-x] ");
+ }
+}
+
+static size_t trc_end_buf(char *buf, size_t len, size_t maxlen, bool addnl)
+{
+ /* make sure we end the trace line in `buf` properly. It needs
+ * to end with a terminating '\0' or '\n\0' */
+ if(len >= (maxlen - (addnl ? 2 : 1))) {
+ len = maxlen - 5;
+ buf[len++] = '.';
+ buf[len++] = '.';
+ buf[len++] = '.';
+ buf[len++] = '\n';
+ }
+ else if(addnl)
+ buf[len++] = '\n';
+ buf[len] = '\0';
+ return len;
+}
+
+void Curl_debug(struct Curl_easy *data, curl_infotype type,
+ char *ptr, size_t size)
+{
+ if(data->set.verbose) {
+ static const char s_infotype[CURLINFO_END][3] = {
+ "* ", "< ", "> ", "{ ", "} ", "{ ", "} " };
+ char buf[TRC_LINE_MAX];
+ size_t len;
+ if(data->set.fdebug) {
+ bool inCallback = Curl_is_in_callback(data);
+
+ if(CURL_TRC_IDS(data) && (size < TRC_LINE_MAX)) {
+ len = trc_print_ids(data, buf, TRC_LINE_MAX);
+ len += msnprintf(buf + len, TRC_LINE_MAX - len, "%.*s",
+ (int)size, ptr);
+ len = trc_end_buf(buf, len, TRC_LINE_MAX, FALSE);
+ Curl_set_in_callback(data, TRUE);
+ (void)(*data->set.fdebug)(data, type, buf, len, data->set.debugdata);
+ Curl_set_in_callback(data, inCallback);
+ }
+ else {
+ Curl_set_in_callback(data, TRUE);
+ (void)(*data->set.fdebug)(data, type, ptr, size, data->set.debugdata);
+ Curl_set_in_callback(data, inCallback);
+ }
+ }
+ else {
+ switch(type) {
+ case CURLINFO_TEXT:
+ case CURLINFO_HEADER_OUT:
+ case CURLINFO_HEADER_IN:
+ if(CURL_TRC_IDS(data)) {
+ len = trc_print_ids(data, buf, TRC_LINE_MAX);
+ fwrite(buf, len, 1, data->set.err);
+ }
+ fwrite(s_infotype[type], 2, 1, data->set.err);
+ fwrite(ptr, size, 1, data->set.err);
+ break;
+ default: /* nada */
+ break;
+ }
+ }
+ }
+}
/* Curl_failf() is for messages stating why we failed.
* The message SHALL NOT include any LF or CR.
DEBUGASSERT(!strchr(fmt, '\n'));
if(data->set.verbose || data->set.errorbuffer) {
va_list ap;
- int len;
+ size_t len;
char error[CURL_ERROR_SIZE + 2];
va_start(ap, fmt);
len = mvsnprintf(error, CURL_ERROR_SIZE, fmt, ap);
}
error[len++] = '\n';
error[len] = '\0';
- Curl_debug(data, CURLINFO_TEXT, error, len);
+ trc_write(data, CURLINFO_TEXT, error, len);
va_end(ap);
}
}
#if !defined(CURL_DISABLE_VERBOSE_STRINGS)
-/* Curl_infof() is for info message along the way */
-#define MAXINFO 2048
-static void trc_infof(struct Curl_easy *data, struct curl_trc_feat *feat,
- const char * const fmt, va_list ap) CURL_PRINTF(3, 0);
+static void trc_infof(struct Curl_easy *data,
+ struct curl_trc_feat *feat,
+ const char *opt_id, int opt_id_idx,
+ const char * const fmt, va_list ap) CURL_PRINTF(5, 0);
-static void trc_infof(struct Curl_easy *data, struct curl_trc_feat *feat,
+static void trc_infof(struct Curl_easy *data,
+ struct curl_trc_feat *feat,
+ const char *opt_id, int opt_id_idx,
const char * const fmt, va_list ap)
{
- int len = 0;
- char buffer[MAXINFO + 5];
+ size_t len = 0;
+ char buf[TRC_LINE_MAX];
+
+ if(CURL_TRC_IDS(data))
+ len += trc_print_ids(data, buf + len, TRC_LINE_MAX - len);
if(feat)
- len = msnprintf(buffer, (MAXINFO + 1), "[%s] ", feat->name);
- len += mvsnprintf(buffer + len, (MAXINFO + 1) - len, fmt, ap);
- if(len >= MAXINFO) { /* too long, shorten with '...' */
- --len;
- buffer[len++] = '.';
- buffer[len++] = '.';
- buffer[len++] = '.';
+ len += msnprintf(buf + len, TRC_LINE_MAX - len, "[%s] ", feat->name);
+ if(opt_id) {
+ if(opt_id_idx > 0)
+ len += msnprintf(buf + len, TRC_LINE_MAX - len, "[%s-%d] ",
+ opt_id, opt_id_idx);
+ else
+ len += msnprintf(buf + len, TRC_LINE_MAX - len, "[%s] ", opt_id);
}
- buffer[len++] = '\n';
- buffer[len] = '\0';
- Curl_debug(data, CURLINFO_TEXT, buffer, len);
+ len += mvsnprintf(buf + len, TRC_LINE_MAX - len, fmt, ap);
+ len = trc_end_buf(buf, len, TRC_LINE_MAX, TRUE);
+ trc_write(data, CURLINFO_TEXT, buf, len);
}
void Curl_infof(struct Curl_easy *data, const char *fmt, ...)
if(Curl_trc_is_verbose(data)) {
va_list ap;
va_start(ap, fmt);
- trc_infof(data, data->state.feat, fmt, ap);
+ trc_infof(data, data->state.feat, NULL, 0, fmt, ap);
va_end(ap);
}
}
DEBUGASSERT(cf);
if(Curl_trc_cf_is_verbose(cf, data)) {
va_list ap;
- int len = 0;
- char buffer[MAXINFO + 2];
- if(data->state.feat)
- len += msnprintf(buffer + len, MAXINFO - len, "[%s] ",
- data->state.feat->name);
- if(cf->sockindex)
- len += msnprintf(buffer + len, MAXINFO - len, "[%s-%d] ",
- cf->cft->name, cf->sockindex);
- else
- len += msnprintf(buffer + len, MAXINFO - len, "[%s] ", cf->cft->name);
va_start(ap, fmt);
- len += mvsnprintf(buffer + len, MAXINFO - len, fmt, ap);
+ trc_infof(data, data->state.feat, cf->cft->name, cf->sockindex, fmt, ap);
va_end(ap);
- buffer[len++] = '\n';
- buffer[len] = '\0';
- Curl_debug(data, CURLINFO_TEXT, buffer, len);
}
}
+struct curl_trc_feat Curl_trc_feat_multi = {
+ "MULTI",
+ CURL_LOG_LVL_NONE,
+};
struct curl_trc_feat Curl_trc_feat_read = {
"READ",
CURL_LOG_LVL_NONE,
};
+static const char * const Curl_trc_mstate_names[]={
+ "INIT",
+ "PENDING",
+ "SETUP",
+ "CONNECT",
+ "RESOLVING",
+ "CONNECTING",
+ "TUNNELING",
+ "PROTOCONNECT",
+ "PROTOCONNECTING",
+ "DO",
+ "DOING",
+ "DOING_MORE",
+ "DID",
+ "PERFORMING",
+ "RATELIMITING",
+ "DONE",
+ "COMPLETED",
+ "MSGSENT",
+};
+
+const char *Curl_trc_mstate_name(int state)
+{
+ if((state >= 0) && ((size_t)state < CURL_ARRAYSIZE(Curl_trc_mstate_names)))
+ return Curl_trc_mstate_names[(size_t)state];
+ return "?";
+}
+
+void Curl_trc_multi(struct Curl_easy *data, const char *fmt, ...)
+{
+ DEBUGASSERT(!strchr(fmt, '\n'));
+ if(Curl_trc_ft_is_verbose(data, &Curl_trc_feat_multi)) {
+ const char *sname = (data->id >= 0) ?
+ Curl_trc_mstate_name(data->mstate) : NULL;
+ va_list ap;
+ va_start(ap, fmt);
+ trc_infof(data, &Curl_trc_feat_multi, sname, 0, fmt, ap);
+ va_end(ap);
+ }
+}
+
void Curl_trc_read(struct Curl_easy *data, const char *fmt, ...)
{
DEBUGASSERT(!strchr(fmt, '\n'));
if(Curl_trc_ft_is_verbose(data, &Curl_trc_feat_read)) {
va_list ap;
va_start(ap, fmt);
- trc_infof(data, &Curl_trc_feat_read, fmt, ap);
+ trc_infof(data, &Curl_trc_feat_read, NULL, 0, fmt, ap);
va_end(ap);
}
}
if(Curl_trc_ft_is_verbose(data, &Curl_trc_feat_write)) {
va_list ap;
va_start(ap, fmt);
- trc_infof(data, &Curl_trc_feat_write, fmt, ap);
+ trc_infof(data, &Curl_trc_feat_write, NULL, 0, fmt, ap);
va_end(ap);
}
}
if(Curl_trc_ft_is_verbose(data, &Curl_trc_feat_dns)) {
va_list ap;
va_start(ap, fmt);
- trc_infof(data, &Curl_trc_feat_dns, fmt, ap);
+ trc_infof(data, &Curl_trc_feat_dns, NULL, 0, fmt, ap);
va_end(ap);
}
}
if(Curl_trc_ft_is_verbose(data, &Curl_trc_feat_ftp)) {
va_list ap;
va_start(ap, fmt);
- trc_infof(data, &Curl_trc_feat_ftp, fmt, ap);
+ trc_infof(data, &Curl_trc_feat_ftp, NULL, 0, fmt, ap);
va_end(ap);
}
}
if(Curl_trc_ft_is_verbose(data, &Curl_trc_feat_smtp)) {
va_list ap;
va_start(ap, fmt);
- trc_infof(data, &Curl_trc_feat_smtp, fmt, ap);
+ trc_infof(data, &Curl_trc_feat_smtp, NULL, 0, fmt, ap);
va_end(ap);
}
}
if(Curl_trc_ft_is_verbose(data, &Curl_trc_feat_ssls)) {
va_list ap;
va_start(ap, fmt);
- trc_infof(data, &Curl_trc_feat_ssls, fmt, ap);
+ trc_infof(data, &Curl_trc_feat_ssls, NULL, 0, fmt, ap);
va_end(ap);
}
}
if(Curl_trc_ft_is_verbose(data, &Curl_trc_feat_ws)) {
va_list ap;
va_start(ap, fmt);
- trc_infof(data, &Curl_trc_feat_ws, fmt, ap);
+ trc_infof(data, &Curl_trc_feat_ws, NULL, 0, fmt, ap);
va_end(ap);
}
}
#define TRC_CT_PROTOCOL (1<<(0))
#define TRC_CT_NETWORK (1<<(1))
#define TRC_CT_PROXY (1<<(2))
+#define TRC_CT_INTERNALS (1<<(3))
struct trc_feat_def {
struct curl_trc_feat *feat;
};
static struct trc_feat_def trc_feats[] = {
+ { &Curl_trc_feat_ids, TRC_CT_INTERNALS },
+ { &Curl_trc_feat_multi, TRC_CT_NETWORK },
{ &Curl_trc_feat_read, TRC_CT_NONE },
{ &Curl_trc_feat_write, TRC_CT_NONE },
{ &Curl_trc_feat_dns, TRC_CT_NETWORK },
struct curl_trc_feat;
+void Curl_trc_multi(struct Curl_easy *data, const char *fmt, ...)
+{
+ (void)data; (void)fmt;
+}
+
void Curl_trc_write(struct Curl_easy *data, const char *fmt, ...)
{
(void)data; (void)fmt;
*/
void Curl_trc_cf_infof(struct Curl_easy *data, struct Curl_cfilter *cf,
const char *fmt, ...) CURL_PRINTF(3, 4);
+void Curl_trc_multi(struct Curl_easy *data,
+ const char *fmt, ...) CURL_PRINTF(2, 3);
+const char *Curl_trc_mstate_name(int state);
+#define CURL_MSTATE_NAME(s) Curl_trc_mstate_name((int)(s))
void Curl_trc_write(struct Curl_easy *data,
const char *fmt, ...) CURL_PRINTF(2, 3);
void Curl_trc_read(struct Curl_easy *data,
#define infof(data, ...) \
do { if(Curl_trc_is_verbose(data)) \
Curl_infof(data, __VA_ARGS__); } while(0)
+#define CURL_TRC_M(data, ...) \
+ do { if(Curl_trc_ft_is_verbose(data, &Curl_trc_feat_multi)) \
+ Curl_trc_multi(data, __VA_ARGS__); } while(0)
#define CURL_TRC_CF(data, cf, ...) \
do { if(Curl_trc_cf_is_verbose(cf, data)) \
Curl_trc_cf_infof(data, cf, __VA_ARGS__); } while(0)
#else /* CURL_HAVE_C99 */
#define infof Curl_infof
+#define CURL_TRC_M Curl_trc_multi
#define CURL_TRC_CF Curl_trc_cf_infof
#define CURL_TRC_WRITE Curl_trc_write
#define CURL_TRC_READ Curl_trc_read
const char *name;
int log_level;
};
+extern struct curl_trc_feat Curl_trc_feat_multi;
extern struct curl_trc_feat Curl_trc_feat_read;
extern struct curl_trc_feat Curl_trc_feat_write;
extern struct curl_trc_feat Curl_trc_feat_dns;
#define Curl_trc_is_verbose(d) (FALSE)
#define Curl_trc_cf_is_verbose(x,y) (FALSE)
#define Curl_trc_ft_is_verbose(x,y) (FALSE)
+#define CURL_MSTATE_NAME(x) ((void)(x), "-")
#endif /* !defined(CURL_DISABLE_VERBOSE_STRINGS) */
}
out:
- if(!result && !data->state.done && keep_changed)
- /* This transfer may have been moved in or out of the bundle, update the
- corresponding socket callback, if used */
- result = Curl_updatesocket(data);
+ if(!result && !data->state.done && keep_changed && data->multi)
+ /* pause/unpausing may result in multi event changes */
+ if(Curl_multi_ev_assess_xfer(data->multi, data))
+ result = CURLE_ABORTED_BY_CALLBACK;
if(recursive)
/* this might have called a callback recursively which might have set this
#include "share.h"
#include "psl.h"
#include "multiif.h"
+#include "multi_ev.h"
#include "sendf.h"
#include "timeval.h"
#include "http.h"
static void move_pending_to_connect(struct Curl_multi *multi,
struct Curl_easy *data);
-static CURLMcode singlesocket(struct Curl_multi *multi,
- struct Curl_easy *data);
static CURLMcode add_next_timeout(struct curltime now,
struct Curl_multi *multi,
struct Curl_easy *d);
long *timeout_ms);
static void process_pending_handles(struct Curl_multi *multi);
static void multi_xfer_bufs_free(struct Curl_multi *multi);
-static void expire_ex(struct Curl_easy *data, const struct curltime *nowp,
- timediff_t milli, expire_id id);
-
-#if defined( DEBUGBUILD) && !defined(CURL_DISABLE_VERBOSE_STRINGS)
-static const char * const multi_statename[]={
- "INIT",
- "PENDING",
- "SETUP",
- "CONNECT",
- "RESOLVING",
- "CONNECTING",
- "TUNNELING",
- "PROTOCONNECT",
- "PROTOCONNECTING",
- "DO",
- "DOING",
- "DOING_MORE",
- "DID",
- "PERFORMING",
- "RATELIMITING",
- "DONE",
- "COMPLETED",
- "MSGSENT",
-};
-#endif
/* function pointer called once when switching TO a state */
typedef void (*init_multistate_func)(struct Curl_easy *data);
NULL /* MSGSENT */
};
-#if defined(DEBUGBUILD) && defined(CURL_DISABLE_VERBOSE_STRINGS)
- (void) lineno;
-#endif
-
if(oldstate == state)
/* do not bother when the new state is the same as the old state */
return;
- data->mstate = state;
-
-#if defined(DEBUGBUILD) && !defined(CURL_DISABLE_VERBOSE_STRINGS)
- if(data->mstate >= MSTATE_PENDING &&
- data->mstate < MSTATE_COMPLETED) {
- infof(data,
- "STATE: %s => %s handle %p; line %d",
- multi_statename[oldstate], multi_statename[data->mstate],
- (void *)data, lineno);
- }
+#ifdef DEBUGBUILD
+ CURL_TRC_M(data, "-> [%s] (line %d)", CURL_MSTATE_NAME(state), lineno);
+#else
+ CURL_TRC_M(data, "-> [%s]", CURL_MSTATE_NAME(state));
#endif
+ data->mstate = state;
+
if(state == MSTATE_COMPLETED) {
/* changing to COMPLETED means there is one less easy handle 'alive' */
DEBUGASSERT(data->multi->num_alive > 0);
#define multistate(x,y) mstate(x,y, __LINE__)
#endif
-/*
- * We add one of these structs to the sockhash for each socket
- */
-
-struct Curl_sh_entry {
- struct Curl_hash transfers; /* hash of transfers using this socket */
- unsigned int action; /* what combined action READ/WRITE this socket waits
- for */
- unsigned int users; /* number of transfers using this */
- void *socketp; /* settable by users with curl_multi_assign() */
- unsigned int readers; /* this many transfers want to read */
- unsigned int writers; /* this many transfers want to write */
-};
-
-/* look up a given socket in the socket hash, skip invalid sockets */
-static struct Curl_sh_entry *sh_getentry(struct Curl_hash *sh,
- curl_socket_t s)
-{
- if(s != CURL_SOCKET_BAD) {
- /* only look for proper sockets */
- return Curl_hash_pick(sh, (char *)&s, sizeof(curl_socket_t));
- }
- return NULL;
-}
-
-#define TRHASH_SIZE 13
-
-/* the given key here is a struct Curl_easy pointer */
-static size_t trhash(void *key, size_t key_length, size_t slots_num)
-{
- unsigned char bytes = ((unsigned char *)key)[key_length - 1] ^
- ((unsigned char *)key)[0];
- return (bytes % slots_num);
-}
-
-static size_t trhash_compare(void *k1, size_t k1_len, void *k2, size_t k2_len)
-{
- (void)k2_len;
- return !memcmp(k1, k2, k1_len);
-}
-
-static void trhash_dtor(void *nada)
-{
- (void)nada;
-}
-
-/*
- * The sockhash has its own separate subhash in each entry that need to be
- * safely destroyed first.
- */
-static void sockhash_destroy(struct Curl_hash *h)
-{
- struct Curl_hash_iterator iter;
- struct Curl_hash_element *he;
-
- DEBUGASSERT(h);
- Curl_hash_start_iterate(h, &iter);
- he = Curl_hash_next_element(&iter);
- while(he) {
- struct Curl_sh_entry *sh = (struct Curl_sh_entry *)he->ptr;
- Curl_hash_destroy(&sh->transfers);
- he = Curl_hash_next_element(&iter);
- }
- Curl_hash_destroy(h);
-}
-
-
-/* make sure this socket is present in the hash for this handle */
-static struct Curl_sh_entry *sh_addentry(struct Curl_hash *sh,
- curl_socket_t s)
-{
- struct Curl_sh_entry *there = sh_getentry(sh, s);
- struct Curl_sh_entry *check;
-
- if(there) {
- /* it is present, return fine */
- return there;
- }
-
- /* not present, add it */
- check = calloc(1, sizeof(struct Curl_sh_entry));
- if(!check)
- return NULL; /* major failure */
-
- Curl_hash_init(&check->transfers, TRHASH_SIZE, trhash, trhash_compare,
- trhash_dtor);
-
- /* make/add new hash entry */
- if(!Curl_hash_add(sh, (char *)&s, sizeof(curl_socket_t), check)) {
- Curl_hash_destroy(&check->transfers);
- free(check);
- return NULL; /* major failure */
- }
-
- return check; /* things are good in sockhash land */
-}
-
-
-/* delete the given socket + handle from the hash */
-static void sh_delentry(struct Curl_sh_entry *entry,
- struct Curl_hash *sh, curl_socket_t s)
-{
- Curl_hash_destroy(&entry->transfers);
-
- /* We remove the hash entry. This will end up in a call to
- sh_freeentry(). */
- Curl_hash_delete(sh, (char *)&s, sizeof(curl_socket_t));
-}
-
-/*
- * free a sockhash entry
- */
-static void sh_freeentry(void *freethis)
-{
- struct Curl_sh_entry *p = (struct Curl_sh_entry *) freethis;
-
- free(p);
-}
-
-static size_t fd_key_compare(void *k1, size_t k1_len, void *k2, size_t k2_len)
-{
- (void) k1_len; (void) k2_len;
-
- return (*((curl_socket_t *) k1)) == (*((curl_socket_t *) k2));
-}
-
-static size_t hash_fd(void *key, size_t key_length, size_t slots_num)
-{
- curl_socket_t fd = *((curl_socket_t *) key);
- (void) key_length;
-
- return (fd % (curl_socket_t)slots_num);
-}
-
-/*
- * sh_init() creates a new socket hash and returns the handle for it.
- *
- * Quote from README.multi_socket:
- *
- * "Some tests at 7000 and 9000 connections showed that the socket hash lookup
- * is somewhat of a bottle neck. Its current implementation may be a bit too
- * limiting. It simply has a fixed-size array, and on each entry in the array
- * it has a linked list with entries. The hash only checks which list to scan
- * through. The code I had used so for used a list with merely 7 slots (as
- * that is what the DNS hash uses) but with 7000 connections that would make
- * an average of 1000 nodes in each list to run through. I upped that to 97
- * slots (I believe a prime is suitable) and noticed a significant speed
- * increase. I need to reconsider the hash implementation or use a rather
- * large default value like this. At 9000 connections I was still below 10us
- * per call."
- *
- */
-static void sh_init(struct Curl_hash *hash, size_t hashsize)
-{
- Curl_hash_init(hash, hashsize, hash_fd, fd_key_compare,
- sh_freeentry);
-}
/* multi->proto_hash destructor. Should never be called as elements
* MUST be added with their own destructor */
Curl_llist_append(&multi->msglist, msg, &msg->list);
}
-struct Curl_multi *Curl_multi_handle(size_t hashsize, /* socket hash */
+struct Curl_multi *Curl_multi_handle(size_t ev_hashsize, /* event hash */
size_t chashsize, /* connection hash */
size_t dnssize, /* dns hash */
size_t sesssize) /* TLS session cache */
multi->magic = CURL_MULTI_HANDLE;
Curl_init_dnscache(&multi->hostcache, dnssize);
-
- sh_init(&multi->sockhash, hashsize);
+ Curl_multi_ev_init(multi, ev_hashsize);
Curl_hash_init(&multi->proto_hash, 23,
Curl_hash_str, Curl_str_key_compare, ph_freeentry);
error:
- sockhash_destroy(&multi->sockhash);
+ Curl_multi_ev_cleanup(multi);
Curl_hash_destroy(&multi->proto_hash);
Curl_hash_destroy(&multi->hostcache);
Curl_cpool_destroy(&multi->cpool);
Curl_cpool_xfer_init(data);
multi_warn_debug(multi, data);
+ CURL_TRC_M(data, "added, transfers=%u", multi->num_easy);
return CURLM_OK;
}
if(CONN_INUSE(conn)) {
/* Stop if still used. */
- DEBUGF(infof(data, "Connection still in use %zu, "
- "no more multi_done now!",
- Curl_llist_count(&conn->easyq)));
+ CURL_TRC_M(data, "Connection still in use %zu, no more multi_done now!",
+ Curl_llist_count(&conn->easyq));
return;
}
#endif
) || conn->bits.close
|| (mdctx->premature && !Curl_conn_is_multiplex(conn, FIRSTSOCKET))) {
- DEBUGF(infof(data, "multi_done, not reusing connection=%"
- FMT_OFF_T ", forbid=%d"
- ", close=%d, premature=%d, conn_multiplex=%d",
- conn->connection_id, data->set.reuse_forbid,
- conn->bits.close, mdctx->premature,
- Curl_conn_is_multiplex(conn, FIRSTSOCKET)));
+ CURL_TRC_M(data, "multi_done, not reusing connection=%"
+ FMT_OFF_T ", forbid=%d"
+ ", close=%d, premature=%d, conn_multiplex=%d",
+ conn->connection_id, data->set.reuse_forbid,
+ conn->bits.close, mdctx->premature,
+ Curl_conn_is_multiplex(conn, FIRSTSOCKET));
connclose(conn, "disconnecting");
Curl_cpool_disconnect(data, conn, mdctx->premature);
}
memset(&mdctx, 0, sizeof(mdctx));
-#if defined(DEBUGBUILD) && !defined(CURL_DISABLE_VERBOSE_STRINGS)
- DEBUGF(infof(data, "multi_done[%s]: status: %d prem: %d done: %d",
- multi_statename[data->mstate],
- (int)status, (int)premature, data->state.done));
-#else
- DEBUGF(infof(data, "multi_done: status: %d prem: %d done: %d",
- (int)status, (int)premature, data->state.done));
-#endif
+ CURL_TRC_M(data, "multi_done: status: %d prem: %d done: %d",
+ (int)status, (int)premature, data->state.done);
if(data->state.done)
/* Stop if multi_done() has already been called */
Curl_wildcard_dtor(&data->wildcard);
- /* change state without using multistate(), only to make singlesocket() do
- what we want */
data->mstate = MSTATE_COMPLETED;
- /* This ignores the return code even in case of problems because there is
- nothing more to do about that, here */
- (void)singlesocket(multi, data); /* to let the application know what sockets
- that vanish with this handle */
-
/* Remove the association between the connection and the handle */
Curl_detach_connection(data);
+ /* Tell event handling that this transfer is definitely going away */
+ Curl_multi_ev_xfer_done(multi, data);
+
if(data->set.connect_only && !data->multi_easy) {
/* This removes a handle that was part the multi interface that used
CONNECT_ONLY, that connection is now left alive but since this handle
if(rc)
return rc;
}
+
+ CURL_TRC_M(data, "removed, transfers=%u", multi->num_easy);
return CURLM_OK;
}
/* Initializes `poll_set` with the current socket poll actions needed
* for transfer `data`. */
-static void multi_getsock(struct Curl_easy *data,
- struct easy_pollset *ps)
+void Curl_multi_getsock(struct Curl_easy *data,
+ struct easy_pollset *ps,
+ const char *caller)
{
bool expect_sockets = TRUE;
- /* The no connection case can happen when this is called from
- curl_multi_remove_handle() => singlesocket() => multi_getsock().
- */
+
+ /* If the transfer has no connection, this is fine. Happens when
+ called via curl_multi_remove_handle() => Curl_multi_ev_assess() =>
+ Curl_multi_getsock(). */
Curl_pollset_reset(data, ps);
if(!data->conn)
return;
case MSTATE_CONNECTING:
case MSTATE_TUNNELING:
Curl_pollset_add_socks(data, ps, connecting_getsock);
- Curl_conn_adjust_pollset(data, ps);
+ Curl_conn_adjust_pollset(data, data->conn, ps);
break;
case MSTATE_PROTOCONNECT:
case MSTATE_PROTOCONNECTING:
Curl_pollset_add_socks(data, ps, protocol_getsock);
- Curl_conn_adjust_pollset(data, ps);
+ Curl_conn_adjust_pollset(data, data->conn, ps);
break;
case MSTATE_DO:
case MSTATE_DOING:
Curl_pollset_add_socks(data, ps, doing_getsock);
- Curl_conn_adjust_pollset(data, ps);
+ Curl_conn_adjust_pollset(data, data->conn, ps);
break;
case MSTATE_DOING_MORE:
Curl_pollset_add_socks(data, ps, domore_getsock);
- Curl_conn_adjust_pollset(data, ps);
+ Curl_conn_adjust_pollset(data, data->conn, ps);
break;
case MSTATE_DID: /* same as PERFORMING in regard to polling */
case MSTATE_PERFORMING:
Curl_pollset_add_socks(data, ps, perform_getsock);
- Curl_conn_adjust_pollset(data, ps);
+ Curl_conn_adjust_pollset(data, data->conn, ps);
break;
case MSTATE_RATELIMITING:
break;
}
+ switch(ps->num) {
+ case 0:
+ CURL_TRC_M(data, "%s pollset[], timeouts=%zu, paused %d/%d (r/w)",
+ caller, Curl_llist_count(&data->state.timeoutlist),
+ Curl_creader_is_paused(data), Curl_cwriter_is_paused(data));
+ break;
+ case 1:
+ CURL_TRC_M(data, "%s pollset[fd=%" FMT_SOCKET_T " %s%s], timeouts=%zu",
+ caller, ps->sockets[0],
+ (ps->actions[0] & CURL_POLL_IN) ? "IN" : "",
+ (ps->actions[0] & CURL_POLL_OUT) ? "OUT" : "",
+ Curl_llist_count(&data->state.timeoutlist));
+ break;
+ case 2:
+ CURL_TRC_M(data, "%s pollset[fd=%" FMT_SOCKET_T " %s%s, "
+ "fd=%" FMT_SOCKET_T " %s%s], timeouts=%zu",
+ caller, ps->sockets[0],
+ (ps->actions[0] & CURL_POLL_IN) ? "IN" : "",
+ (ps->actions[0] & CURL_POLL_OUT) ? "OUT" : "",
+ ps->sockets[1],
+ (ps->actions[1] & CURL_POLL_IN) ? "IN" : "",
+ (ps->actions[1] & CURL_POLL_OUT) ? "OUT" : "",
+ Curl_llist_count(&data->state.timeoutlist));
+ break;
+ default:
+ CURL_TRC_M(data, "%s pollset[fds=%u], timeouts=%zu",
+ caller, ps->num, Curl_llist_count(&data->state.timeoutlist));
+ break;
+ }
if(expect_sockets && !ps->num &&
!Curl_llist_count(&data->state.timeoutlist) &&
!Curl_cwriter_is_paused(data) && !Curl_creader_is_paused(data) &&
for(e = Curl_llist_head(&multi->process); e; e = Curl_node_next(e)) {
struct Curl_easy *data = Curl_node_elem(e);
+ struct easy_pollset ps;
- multi_getsock(data, &data->last_poll);
+ Curl_multi_getsock(data, &ps, "curl_multi_fdset");
- for(i = 0; i < data->last_poll.num; i++) {
- if(!FDSET_SOCK(data->last_poll.sockets[i]))
+ for(i = 0; i < ps.num; i++) {
+ if(!FDSET_SOCK(ps.sockets[i]))
/* pretend it does not exist */
continue;
#if defined(__DJGPP__)
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Warith-conversion"
#endif
- if(data->last_poll.actions[i] & CURL_POLL_IN)
- FD_SET(data->last_poll.sockets[i], read_fd_set);
- if(data->last_poll.actions[i] & CURL_POLL_OUT)
- FD_SET(data->last_poll.sockets[i], write_fd_set);
+ if(ps.actions[i] & CURL_POLL_IN)
+ FD_SET(ps.sockets[i], read_fd_set);
+ if(ps.actions[i] & CURL_POLL_OUT)
+ FD_SET(ps.sockets[i], write_fd_set);
#if defined(__DJGPP__)
#pragma GCC diagnostic pop
#endif
- if((int)data->last_poll.sockets[i] > this_max_fd)
- this_max_fd = (int)data->last_poll.sockets[i];
+ if((int)ps.sockets[i] > this_max_fd)
+ this_max_fd = (int)ps.sockets[i];
}
}
Curl_waitfds_init(&cwfds, ufds, size);
for(e = Curl_llist_head(&multi->process); e; e = Curl_node_next(e)) {
struct Curl_easy *data = Curl_node_elem(e);
- multi_getsock(data, &data->last_poll);
- need += Curl_waitfds_add_ps(&cwfds, &data->last_poll);
+ struct easy_pollset ps;
+
+ Curl_multi_getsock(data, &ps, "curl_multi_waitfds");
+ need += Curl_waitfds_add_ps(&cwfds, &ps);
}
need += Curl_cpool_add_waitfds(&multi->cpool, &cwfds);
/* Add the curl handles to our pollfds first */
for(e = Curl_llist_head(&multi->process); e; e = Curl_node_next(e)) {
struct Curl_easy *data = Curl_node_elem(e);
+ struct easy_pollset ps;
- multi_getsock(data, &data->last_poll);
- if(Curl_pollfds_add_ps(&cpfds, &data->last_poll)) {
+ Curl_multi_getsock(data, &ps, "multi_wait");
+ if(Curl_pollfds_add_ps(&cpfds, &ps)) {
result = CURLM_OUT_OF_MEMORY;
goto out;
}
#ifdef USE_WINSOCK
/* Count up all our own sockets that had activity,
and remove them from the event. */
- if(curl_nfds) {
- for(e = Curl_llist_head(&multi->process); e && !result;
- e = Curl_node_next(e)) {
- struct Curl_easy *data = Curl_node_elem(e);
-
- for(i = 0; i < data->last_poll.num; i++) {
- wsa_events.lNetworkEvents = 0;
- if(WSAEnumNetworkEvents(data->last_poll.sockets[i], NULL,
- &wsa_events) == 0) {
- if(ret && !pollrc && wsa_events.lNetworkEvents)
- retcode++;
- }
- WSAEventSelect(data->last_poll.sockets[i], multi->wsa_event, 0);
- }
+ for(i = 0; i < curl_nfds; ++i) {
+ wsa_events.lNetworkEvents = 0;
+ if(WSAEnumNetworkEvents(cpfds.pfds[i].fd, NULL, &wsa_events) == 0) {
+ if(ret && !pollrc && wsa_events.lNetworkEvents)
+ retcode++;
}
+ WSAEventSelect(cpfds.pfds[i].fd, multi->wsa_event, 0);
}
-
WSAResetEvent(multi->wsa_event);
#else
#ifdef ENABLE_WAKEUP
socket(s) will again be used further down. If the name has not yet been
resolved, it is likely that new sockets have been opened in an attempt to
contact another resolver. */
- rc = singlesocket(multi, data);
+ rc = Curl_multi_ev_assess_xfer(multi, data);
if(rc)
return rc;
rc = CURLM_OK;
if(multi_ischanged(multi, TRUE)) {
- DEBUGF(infof(data, "multi changed, check CONNECT_PEND queue"));
+ CURL_TRC_M(data, "multi changed, check CONNECT_PEND queue");
process_pending_handles(multi); /* multiplexed */
}
struct Curl_llist_node *e;
struct Curl_llist_node *n = NULL;
struct Curl_multi *multi = m;
+ bool first = TRUE;
SIGPIPE_VARIABLE(pipe_st);
if(!GOOD_MULTI_HANDLE(multi))
CURLMcode result;
/* Do the loop and only alter the signal ignore state if the next handle
has a different NO_SIGNAL state than the previous */
+ if(first) {
+ CURL_TRC_M(data, "multi_perform(running=%u)", multi->num_alive);
+ first = FALSE;
+ }
/* the current node might be unlinked in multi_runsingle(), get the next
pointer now */
n = Curl_node_next(e);
- if(data != multi->cpool.idata) {
+ if(data && data != multi->cpool.idata) {
/* connection pool handle is processed below */
sigpipe_apply(data, &pipe_st);
result = multi_runsingle(multi, &now, data);
}
sigpipe_apply(multi->cpool.idata, &pipe_st);
- Curl_cpool_multi_perform(multi);
+ Curl_cpool_multi_perform(multi, CURL_SOCKET_TIMEOUT);
sigpipe_restore(&pipe_st);
if(multi_ischanged(m, TRUE))
multi->magic = 0; /* not good anymore */
- sockhash_destroy(&multi->sockhash);
+ Curl_multi_ev_cleanup(multi);
Curl_hash_destroy(&multi->proto_hash);
Curl_hash_destroy(&multi->hostcache);
Curl_psl_destroy(&multi->psl);
return NULL;
}
-/*
- * singlesocket() checks what sockets we deal with and their "action state"
- * and if we have a different state in any of those sockets from last time we
- * call the callback accordingly.
- */
-static CURLMcode singlesocket(struct Curl_multi *multi,
- struct Curl_easy *data)
-{
- struct easy_pollset cur_poll;
- CURLMcode mresult;
-
- /* Fill in the 'current' struct with the state as it is now: what sockets to
- supervise and for what actions */
- multi_getsock(data, &cur_poll);
- mresult = Curl_multi_pollset_ev(multi, data, &cur_poll, &data->last_poll);
- if(!mresult) /* Remember for next time */
- memcpy(&data->last_poll, &cur_poll, sizeof(cur_poll));
- return mresult;
-}
-
-CURLMcode Curl_multi_pollset_ev(struct Curl_multi *multi,
- struct Curl_easy *data,
- struct easy_pollset *ps,
- struct easy_pollset *last_ps)
-{
- unsigned int i;
- struct Curl_sh_entry *entry;
- curl_socket_t s;
- int rc;
-
- /* We have 0 .. N sockets already and we get to know about the 0 .. M
- sockets we should have from now on. Detect the differences, remove no
- longer supervised ones and add new ones */
-
- /* walk over the sockets we got right now */
- for(i = 0; i < ps->num; i++) {
- unsigned char cur_action = ps->actions[i];
- unsigned char last_action = 0;
- int comboaction;
-
- s = ps->sockets[i];
-
- /* get it from the hash */
- entry = sh_getentry(&multi->sockhash, s);
- if(entry) {
- /* check if new for this transfer */
- unsigned int j;
- for(j = 0; j < last_ps->num; j++) {
- if(s == last_ps->sockets[j]) {
- last_action = last_ps->actions[j];
- break;
- }
- }
- }
- else {
- /* this is a socket we did not have before, add it to the hash! */
- entry = sh_addentry(&multi->sockhash, s);
- if(!entry)
- /* fatal */
- return CURLM_OUT_OF_MEMORY;
- }
- if(last_action && (last_action != cur_action)) {
- /* Socket was used already, but different action now */
- if(last_action & CURL_POLL_IN) {
- DEBUGASSERT(entry->readers);
- entry->readers--;
- }
- if(last_action & CURL_POLL_OUT) {
- DEBUGASSERT(entry->writers);
- entry->writers--;
- }
- if(cur_action & CURL_POLL_IN) {
- entry->readers++;
- }
- if(cur_action & CURL_POLL_OUT)
- entry->writers++;
- }
- else if(!last_action &&
- !Curl_hash_pick(&entry->transfers, (char *)&data, /* hash key */
- sizeof(struct Curl_easy *))) {
- DEBUGASSERT(entry->users < 100000); /* detect weird values */
- /* a new transfer using this socket */
- entry->users++;
- if(cur_action & CURL_POLL_IN)
- entry->readers++;
- if(cur_action & CURL_POLL_OUT)
- entry->writers++;
- /* add 'data' to the transfer hash on this socket! */
- if(!Curl_hash_add(&entry->transfers, (char *)&data, /* hash key */
- sizeof(struct Curl_easy *), data)) {
- Curl_hash_destroy(&entry->transfers);
- return CURLM_OUT_OF_MEMORY;
- }
- }
-
- comboaction = (entry->writers ? CURL_POLL_OUT : 0) |
- (entry->readers ? CURL_POLL_IN : 0);
-
- /* socket existed before and has the same action set as before */
- if(last_action && ((int)entry->action == comboaction))
- /* same, continue */
- continue;
-
- if(multi->socket_cb) {
- set_in_callback(multi, TRUE);
- rc = multi->socket_cb(data, s, comboaction, multi->socket_userp,
- entry->socketp);
-
- set_in_callback(multi, FALSE);
- if(rc == -1) {
- multi->dead = TRUE;
- return CURLM_ABORTED_BY_CALLBACK;
- }
- }
-
- /* store the current action state */
- entry->action = (unsigned int)comboaction;
- }
-
- /* Check for last_poll.sockets that no longer appear in ps->sockets.
- * Need to remove the easy handle from the multi->sockhash->transfers and
- * remove multi->sockhash entry when this was the last transfer */
- for(i = 0; i < last_ps->num; i++) {
- unsigned int j;
- bool stillused = FALSE;
- s = last_ps->sockets[i];
- for(j = 0; j < ps->num; j++) {
- if(s == ps->sockets[j]) {
- /* this is still supervised */
- stillused = TRUE;
- break;
- }
- }
- if(stillused)
- continue;
-
- entry = sh_getentry(&multi->sockhash, s);
- /* if this is NULL here, the socket has been closed and notified so
- already by Curl_multi_closed() */
- if(entry) {
- unsigned char oldactions = last_ps->actions[i];
- /* this socket has been removed. Decrease user count */
- DEBUGASSERT(entry->users);
- entry->users--;
- if(oldactions & CURL_POLL_OUT)
- entry->writers--;
- if(oldactions & CURL_POLL_IN)
- entry->readers--;
- if(!entry->users) {
- bool dead = FALSE;
- if(multi->socket_cb) {
- set_in_callback(multi, TRUE);
- rc = multi->socket_cb(data, s, CURL_POLL_REMOVE,
- multi->socket_userp, entry->socketp);
- set_in_callback(multi, FALSE);
- if(rc == -1)
- dead = TRUE;
- }
- sh_delentry(entry, &multi->sockhash, s);
- if(dead) {
- multi->dead = TRUE;
- return CURLM_ABORTED_BY_CALLBACK;
- }
- }
- else {
- /* still users, but remove this handle as a user of this socket */
- if(Curl_hash_delete(&entry->transfers, (char *)&data,
- sizeof(struct Curl_easy *))) {
- DEBUGASSERT(NULL);
- }
- }
- }
- } /* for loop over num */
-
- return CURLM_OK;
-}
-
-CURLcode Curl_updatesocket(struct Curl_easy *data)
-{
- if(singlesocket(data->multi, data))
- return CURLE_ABORTED_BY_CALLBACK;
- return CURLE_OK;
-}
-
-
-/*
- * Curl_multi_closed()
- *
- * Used by the connect code to tell the multi_socket code that one of the
- * sockets we were using is about to be closed. This function will then
- * remove it from the sockethash for this handle to make the multi_socket API
- * behave properly, especially for the case when libcurl will create another
- * socket again and it gets the same file descriptor number.
- */
-
-void Curl_multi_closed(struct Curl_easy *data, curl_socket_t s)
+void Curl_multi_will_close(struct Curl_easy *data, curl_socket_t s)
{
if(data) {
- /* if there is still an easy handle associated with this connection */
struct Curl_multi *multi = data->multi;
- DEBUGF(infof(data, "Curl_multi_closed, fd=%" FMT_SOCKET_T
- " multi is %p", s, (void *)multi));
if(multi) {
- /* this is set if this connection is part of a handle that is added to
- a multi handle, and only then this is necessary */
- struct Curl_sh_entry *entry = sh_getentry(&multi->sockhash, s);
-
- DEBUGF(infof(data, "Curl_multi_closed, fd=%" FMT_SOCKET_T
- " entry is %p", s, (void *)entry));
- if(entry) {
- int rc = 0;
- if(multi->socket_cb) {
- set_in_callback(multi, TRUE);
- rc = multi->socket_cb(data, s, CURL_POLL_REMOVE,
- multi->socket_userp, entry->socketp);
- set_in_callback(multi, FALSE);
- }
-
- /* now remove it from the socket hash */
- sh_delentry(entry, &multi->sockhash, s);
- if(rc == -1)
- /* This just marks the multi handle as "dead" without returning an
- error code primarily because this function is used from many
- places where propagating an error back is tricky. */
- multi->dead = TRUE;
- }
+ CURL_TRC_M(data, "Curl_multi_will_close fd=%" FMT_SOCKET_T, s);
+ Curl_multi_ev_socket_done(multi, data, s);
}
}
}
result = multi_runsingle(multi, &mrc->now, data);
if(CURLM_OK >= result) {
- /* get the socket(s) and check if the state has been changed since
- last */
- result = singlesocket(multi, data);
+ /* reassess event handling of data */
+ result = Curl_multi_ev_assess_xfer(multi, data);
if(result)
goto out;
}
int *running_handles)
{
CURLMcode result = CURLM_OK;
- struct Curl_easy *data = NULL;
struct multi_run_ctx mrc;
(void)ev_bitmask;
sigpipe_init(&mrc.pipe_st);
if(checkall) {
- struct Curl_llist_node *e;
/* *perform() deals with running_handles on its own */
result = curl_multi_perform(multi, running_handles);
- /* walk through each easy handle and do the socket state change magic
- and callbacks */
if(result != CURLM_BAD_HANDLE) {
- for(e = Curl_llist_head(&multi->process); e && !result;
- e = Curl_node_next(e)) {
- result = singlesocket(multi, Curl_node_elem(e));
- }
+ /* Reassess event status of all active transfers */
+ result = Curl_multi_ev_assess_xfer_list(multi, &multi->process);
}
mrc.run_cpool = TRUE;
goto out;
}
if(s != CURL_SOCKET_TIMEOUT) {
- struct Curl_sh_entry *entry = sh_getentry(&multi->sockhash, s);
-
- if(!entry) {
- /* Unmatched socket, we cannot act on it but we ignore this fact. In
- real-world tests it has been proved that libevent can in fact give
- the application actions even though the socket was just previously
- asked to get removed, so thus we better survive stray socket actions
- and just move on. */
- /* The socket might come from a connection that is being shut down
- * by the multi's connection pool. */
- Curl_cpool_multi_socket(multi, s, ev_bitmask);
- }
- else {
- struct Curl_hash_iterator iter;
- struct Curl_hash_element *he;
-
- /* the socket can be shared by many transfers, iterate */
- Curl_hash_start_iterate(&entry->transfers, &iter);
- for(he = Curl_hash_next_element(&iter); he;
- he = Curl_hash_next_element(&iter)) {
- data = (struct Curl_easy *)he->ptr;
- DEBUGASSERT(data);
- DEBUGASSERT(data->magic == CURLEASY_MAGIC_NUMBER);
-
- if(data == multi->cpool.idata)
- mrc.run_cpool = TRUE;
- else {
- /* Expire with out current now, so we will get it below when
- * asking the splaytree for expired transfers. */
- expire_ex(data, &mrc.now, 0, EXPIRE_RUN_NOW);
- }
- }
- }
+ Curl_multi_ev_expire_xfers(multi, s, &mrc.now, &mrc.run_cpool);
}
else {
/* Asked to run due to time-out. Clear the 'last_expire_ts' variable to
handles the case when the application asks libcurl to run the timeout
prematurely. */
memset(&multi->last_expire_ts, 0, sizeof(multi->last_expire_ts));
+ mrc.run_cpool = TRUE;
}
result = multi_run_expired(&mrc);
out:
if(mrc.run_cpool) {
sigpipe_apply(multi->cpool.idata, &mrc.pipe_st);
- Curl_cpool_multi_perform(multi);
+ Curl_cpool_multi_perform(multi, s);
}
sigpipe_restore(&mrc.pipe_st);
return CURLM_OK;
}
-static void expire_ex(struct Curl_easy *data,
- const struct curltime *nowp,
- timediff_t milli, expire_id id)
+void Curl_expire_ex(struct Curl_easy *data,
+ const struct curltime *nowp,
+ timediff_t milli, expire_id id)
{
struct Curl_multi *multi = data->multi;
struct curltime *curr_expire = &data->state.expiretime;
void Curl_expire(struct Curl_easy *data, timediff_t milli, expire_id id)
{
struct curltime now = Curl_now();
- expire_ex(data, &now, milli, id);
+ Curl_expire_ex(data, &now, milli, id);
}
/*
/* clear the timeout list too */
Curl_llist_destroy(list, NULL);
-#ifdef DEBUGBUILD
- infof(data, "Expire cleared");
-#endif
+ CURL_TRC_M(data, "Expire cleared");
nowp->tv_sec = 0;
nowp->tv_usec = 0;
return TRUE;
CURLMcode curl_multi_assign(CURLM *m, curl_socket_t s,
void *hashp)
{
- struct Curl_sh_entry *there = NULL;
struct Curl_multi *multi = m;
if(!GOOD_MULTI_HANDLE(multi))
return CURLM_BAD_HANDLE;
- there = sh_getentry(&multi->sockhash, s);
-
- if(!there)
- return CURLM_BAD_SOCKET;
-
- there->socketp = hashp;
-
- return CURLM_OK;
+ return Curl_multi_ev_assign(multi, s, hashp);
}
static void move_pending_to_connect(struct Curl_multi *multi,
--- /dev/null
+/***************************************************************************
+ * _ _ ____ _
+ * Project ___| | | | _ \| |
+ * / __| | | | |_) | |
+ * | (__| |_| | _ <| |___
+ * \___|\___/|_| \_\_____|
+ *
+ * Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
+ *
+ * This software is licensed as described in the file COPYING, which
+ * you should have received as part of this distribution. The terms
+ * are also available at https://curl.se/docs/copyright.html.
+ *
+ * You may opt to use, copy, modify, merge, publish, distribute and/or sell
+ * copies of the Software, and permit persons to whom the Software is
+ * furnished to do so, under the terms of the COPYING file.
+ *
+ * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
+ * KIND, either express or implied.
+ *
+ * SPDX-License-Identifier: curl
+ *
+ ***************************************************************************/
+
+#include "curl_setup.h"
+
+#include <curl/curl.h>
+
+#include "urldata.h"
+#include "cfilters.h"
+#include "curl_trc.h"
+#include "multiif.h"
+#include "timeval.h"
+#include "multi_ev.h"
+#include "select.h"
+#include "warnless.h"
+#include "multihandle.h"
+#include "socks.h"
+/* The last 3 #include files should be in this order */
+#include "curl_printf.h"
+#include "curl_memory.h"
+#include "memdebug.h"
+
+
+static void mev_in_callback(struct Curl_multi *multi, bool value)
+{
+ multi->in_callback = value;
+}
+
+#define CURL_MEV_XFER_HASH_SIZE 13
+#define CURL_MEV_CONN_HASH_SIZE 3
+
+/* Information about a socket for which we inform the libcurl application
+ * what to supervise (CURL_POLL_IN/CURL_POLL_OUT/CURL_POLL_REMOVE)
+ */
+struct mev_sh_entry {
+ struct Curl_hash xfers; /* hash of transfers using this socket */
+ struct Curl_hash conns; /* hash of connections using this socket */
+ void *user_data; /* libcurl app data via curl_multi_assign() */
+ unsigned int action; /* CURL_POLL_IN/CURL_POLL_OUT we last told the
+ * libcurl application to watch out for */
+ unsigned int readers; /* this many transfers want to read */
+ unsigned int writers; /* this many transfers want to write */
+};
+
+static size_t mev_sh_entry_hash(void *key, size_t key_length, size_t slots_num)
+{
+ curl_socket_t fd = *((curl_socket_t *) key);
+ (void) key_length;
+ return (fd % (curl_socket_t)slots_num);
+}
+
+static size_t mev_sh_entry_compare(void *k1, size_t k1_len,
+ void *k2, size_t k2_len)
+{
+ (void) k1_len; (void) k2_len;
+ return (*((curl_socket_t *) k1)) == (*((curl_socket_t *) k2));
+}
+
+/* sockhash entry destructor callback */
+static void mev_sh_entry_dtor(void *freethis)
+{
+ struct mev_sh_entry *entry = (struct mev_sh_entry *)freethis;
+ Curl_hash_destroy(&entry->xfers);
+ Curl_hash_destroy(&entry->conns);
+ free(entry);
+}
+
+/* look up a given socket in the socket hash, skip invalid sockets */
+static struct mev_sh_entry *
+mev_sh_entry_get(struct Curl_hash *sh, curl_socket_t s)
+{
+ if(s != CURL_SOCKET_BAD) {
+ /* only look for proper sockets */
+ return Curl_hash_pick(sh, (char *)&s, sizeof(curl_socket_t));
+ }
+ return NULL;
+}
+
+static void mev_nop_dtor(void *e)
+{
+ (void)e; /* does nothing */
+}
+
+/* make sure this socket is present in the hash for this handle */
+static struct mev_sh_entry *
+mev_sh_entry_add(struct Curl_hash *sh, curl_socket_t s)
+{
+ struct mev_sh_entry *there = mev_sh_entry_get(sh, s);
+ struct mev_sh_entry *check;
+
+ if(there) {
+ /* it is present, return fine */
+ return there;
+ }
+
+ /* not present, add it */
+ check = calloc(1, sizeof(struct mev_sh_entry));
+ if(!check)
+ return NULL; /* major failure */
+
+ Curl_hash_offt_init(&check->xfers, CURL_MEV_XFER_HASH_SIZE, mev_nop_dtor);
+ Curl_hash_offt_init(&check->conns, CURL_MEV_CONN_HASH_SIZE, mev_nop_dtor);
+
+ /* make/add new hash entry */
+ if(!Curl_hash_add(sh, (char *)&s, sizeof(curl_socket_t), check)) {
+ mev_sh_entry_dtor(check);
+ return NULL; /* major failure */
+ }
+
+ return check; /* things are good in sockhash land */
+}
+
+/* delete the given socket entry from the hash */
+static void mev_sh_entry_kill(struct Curl_multi *multi, curl_socket_t s)
+{
+ Curl_hash_delete(&multi->ev.sh_entries, (char *)&s, sizeof(curl_socket_t));
+}
+
+static size_t mev_sh_entry_user_count(struct mev_sh_entry *e)
+{
+ return Curl_hash_count(&e->xfers) + Curl_hash_count(&e->conns);
+}
+
+static bool mev_sh_entry_xfer_known(struct mev_sh_entry *e,
+ struct Curl_easy *data)
+{
+ return !!Curl_hash_offt_get(&e->xfers, data->id);
+}
+
+static bool mev_sh_entry_conn_known(struct mev_sh_entry *e,
+ struct connectdata *conn)
+{
+ return !!Curl_hash_offt_get(&e->conns, conn->connection_id);
+}
+
+static bool mev_sh_entry_xfer_add(struct mev_sh_entry *e,
+ struct Curl_easy *data)
+{
+ /* detect weird values */
+ DEBUGASSERT(mev_sh_entry_user_count(e) < 100000);
+ return !!Curl_hash_offt_set(&e->xfers, data->id, data);
+}
+
+static bool mev_sh_entry_conn_add(struct mev_sh_entry *e,
+ struct connectdata *conn)
+{
+ /* detect weird values */
+ DEBUGASSERT(mev_sh_entry_user_count(e) < 100000);
+ return !!Curl_hash_offt_set(&e->conns, conn->connection_id, conn);
+}
+
+
+static bool mev_sh_entry_xfer_remove(struct mev_sh_entry *e,
+ struct Curl_easy *data)
+{
+ return !Curl_hash_offt_remove(&e->xfers, data->id);
+}
+
+/* Purge any information about socket `s`.
+ * Let the socket callback know as well when necessary */
+static CURLMcode mev_forget_socket(struct Curl_multi *multi,
+ struct Curl_easy *data,
+ curl_socket_t s,
+ const char *cause)
+{
+ struct mev_sh_entry *entry = mev_sh_entry_get(&multi->ev.sh_entries, s);
+ int rc = 0;
+
+ if(!entry) /* we never knew or already forgot about this socket */
+ return CURLM_OK;
+
+ /* We managed this socket before, tell the socket callback to forget it. */
+ if(multi->socket_cb) {
+ CURL_TRC_M(data, "ev %s, call(fd=%" FMT_SOCKET_T ", ev=REMOVE)",
+ cause, s);
+ mev_in_callback(multi, TRUE);
+ rc = multi->socket_cb(data, s, CURL_POLL_REMOVE,
+ multi->socket_userp, entry->user_data);
+ mev_in_callback(multi, FALSE);
+ }
+
+ mev_sh_entry_kill(multi, s);
+ if(rc == -1) {
+ multi->dead = TRUE;
+ return CURLM_ABORTED_BY_CALLBACK;
+ }
+ return CURLM_OK;
+}
+
+static CURLMcode mev_sh_entry_update(struct Curl_multi *multi,
+ struct Curl_easy *data,
+ struct mev_sh_entry *entry,
+ curl_socket_t s,
+ unsigned char last_action,
+ unsigned char cur_action)
+{
+ int rc, comboaction;
+
+ /* we should only be called when the callback exists */
+ DEBUGASSERT(multi->socket_cb);
+ if(!multi->socket_cb)
+ return CURLM_OK;
+
+ /* Transfer `data` goes from `last_action` to `cur_action` on socket `s`
+ * with `multi->ev.sh_entries` entry `entry`. Update `entry` and trigger
+ * `multi->socket_cb` on change, if the callback is set. */
+ if(last_action == cur_action) /* nothing from `data` changed */
+ return CURLM_OK;
+
+ if(last_action & CURL_POLL_IN) {
+ DEBUGASSERT(entry->readers);
+ if(!(cur_action & CURL_POLL_IN))
+ entry->readers--;
+ }
+ else if(cur_action & CURL_POLL_IN)
+ entry->readers++;
+
+ if(last_action & CURL_POLL_OUT) {
+ DEBUGASSERT(entry->writers);
+ if(!(cur_action & CURL_POLL_OUT))
+ entry->writers--;
+ }
+ else if(cur_action & CURL_POLL_OUT)
+ entry->writers++;
+
+ DEBUGASSERT(entry->readers <= mev_sh_entry_user_count(entry));
+ DEBUGASSERT(entry->writers <= mev_sh_entry_user_count(entry));
+ DEBUGASSERT(entry->writers + entry->readers);
+
+ CURL_TRC_M(data, "ev update fd=%" FMT_SOCKET_T ", action '%s%s' -> '%s%s'"
+ " (%d/%d r/w)", s,
+ (last_action & CURL_POLL_IN) ? "IN" : "",
+ (last_action & CURL_POLL_OUT) ? "OUT" : "",
+ (cur_action & CURL_POLL_IN) ? "IN" : "",
+ (cur_action & CURL_POLL_OUT) ? "OUT" : "",
+ entry->readers, entry->writers);
+
+ comboaction = (entry->writers ? CURL_POLL_OUT : 0) |
+ (entry->readers ? CURL_POLL_IN : 0);
+ if(((int)entry->action == comboaction)) /* nothing for socket changed */
+ return CURLM_OK;
+
+ CURL_TRC_M(data, "ev update call(fd=%" FMT_SOCKET_T ", ev=%s%s)",
+ s, (comboaction & CURL_POLL_IN) ? "IN" : "",
+ (comboaction & CURL_POLL_OUT) ? "OUT" : "");
+ mev_in_callback(multi, TRUE);
+ rc = multi->socket_cb(data, s, comboaction, multi->socket_userp,
+ entry->user_data);
+
+ mev_in_callback(multi, FALSE);
+ if(rc == -1) {
+ multi->dead = TRUE;
+ return CURLM_ABORTED_BY_CALLBACK;
+ }
+ entry->action = (unsigned int)comboaction;
+ return CURLM_OK;
+}
+
+static CURLMcode mev_pollset_diff(struct Curl_multi *multi,
+ struct Curl_easy *data,
+ struct connectdata *conn,
+ struct easy_pollset *ps,
+ struct easy_pollset *prev_ps)
+{
+ struct mev_sh_entry *entry;
+ curl_socket_t s;
+ unsigned int i, j;
+ CURLMcode mresult;
+
+ /* The transfer `data` reports in `ps` the sockets it is interested
+ * in and which combinatino of CURL_POLL_IN/CURL_POLL_OUT it wants
+ * to have monitored for events.
+ * There can be more than 1 transfer interested in the same socket
+ * and 1 transfer might be interested in more than 1 socket.
+ * `prev_ps` is the pollset copy from the previous call here. On
+ * the 1st call it will be empty.
+ */
+ DEBUGASSERT(ps);
+ DEBUGASSERT(prev_ps);
+
+ /* Handle changes to sockets the transfer is interested in. */
+ for(i = 0; i < ps->num; i++) {
+ unsigned char last_action;
+ bool first_time = FALSE; /* data/conn appears first time on socket */
+
+ s = ps->sockets[i];
+ /* Have we handled this socket before? */
+ entry = mev_sh_entry_get(&multi->ev.sh_entries, s);
+ if(!entry) {
+ /* new socket, add new entry */
+ first_time = TRUE;
+ entry = mev_sh_entry_add(&multi->ev.sh_entries, s);
+ if(!entry) /* fatal */
+ return CURLM_OUT_OF_MEMORY;
+ CURL_TRC_M(data, "ev new entry fd=%" FMT_SOCKET_T, s);
+ }
+ else if(conn) {
+ first_time = !mev_sh_entry_conn_known(entry, data->conn);
+ }
+ else {
+ first_time = !mev_sh_entry_xfer_known(entry, data);
+ }
+
+ /* What was the previous action the transfer had regarding this socket?
+ * If the transfer is new to the socket, disregard the information
+ * in `last_poll`, because the socket might have been destroyed and
+ * reopened. We'd have cleared the sh_entry for that, but the socket
+ * might still be mentioned in the hashed pollsets. */
+ last_action = 0;
+ if(first_time) {
+ if(conn) {
+ if(!mev_sh_entry_conn_add(entry, data->conn))
+ return CURLM_OUT_OF_MEMORY;
+ }
+ else {
+ if(!mev_sh_entry_xfer_add(entry, data))
+ return CURLM_OUT_OF_MEMORY;
+ }
+ CURL_TRC_M(data, "ev entry fd=%" FMT_SOCKET_T ", added %s #%" FMT_OFF_T
+ ", total=%zu/%zu (xfer/conn)", s,
+ conn ? "connection" : "transfer",
+ conn ? conn->connection_id : data->id,
+ Curl_hash_count(&entry->xfers),
+ Curl_hash_count(&entry->conns));
+ }
+ else {
+ for(j = 0; j < prev_ps->num; j++) {
+ if(s == prev_ps->sockets[j]) {
+ last_action = prev_ps->actions[j];
+ break;
+ }
+ }
+ }
+ /* track readers/writers changes and report to socket callback */
+ mresult = mev_sh_entry_update(multi, data, entry, s,
+ last_action, ps->actions[i]);
+ if(mresult)
+ return mresult;
+ }
+
+ /* Handle changes to sockets the transfer is NO LONGER interested in. */
+ for(i = 0; i < prev_ps->num; i++) {
+ bool stillused = FALSE;
+
+ s = prev_ps->sockets[i];
+ for(j = 0; j < ps->num; j++) {
+ if(s == ps->sockets[j]) {
+ /* socket is still supervised */
+ stillused = TRUE;
+ break;
+ }
+ }
+ if(stillused)
+ continue;
+
+ entry = mev_sh_entry_get(&multi->ev.sh_entries, s);
+ /* if entry does not exist, we were either never told about it or
+ * have already cleaned up this socket via Curl_multi_ev_socket_done().
+ * In other words: this is perfectly normal */
+ if(!entry)
+ continue;
+
+ if(!mev_sh_entry_xfer_remove(entry, data)) {
+ /* `data` says in `prev_ps` that it had been using a socket,
+ * but `data` has not been registered for it.
+ * This should not happen if our book-keeping is correct? */
+ CURL_TRC_M(data, "ev entry fd=%" FMT_SOCKET_T ", transfer lost "
+ "interest but is not registered", s);
+ DEBUGASSERT(NULL);
+ continue;
+ }
+
+ if(mev_sh_entry_user_count(entry)) {
+ /* track readers/writers changes and report to socket callback */
+ mresult = mev_sh_entry_update(multi, data, entry, s,
+ prev_ps->actions[i], 0);
+ if(mresult)
+ return mresult;
+ CURL_TRC_M(data, "ev entry fd=%" FMT_SOCKET_T ", removed transfer, "
+ "total=%zu/%zu (xfer/conn)", s,
+ Curl_hash_count(&entry->xfers),
+ Curl_hash_count(&entry->conns));
+ }
+ else {
+ mresult = mev_forget_socket(multi, data, s, "last user gone");
+ if(mresult)
+ return mresult;
+ }
+ } /* for loop over num */
+
+ /* Remember for next time */
+ memcpy(prev_ps, ps, sizeof(*prev_ps));
+ return CURLM_OK;
+}
+
+static struct easy_pollset*
+mev_add_new_pollset(struct Curl_hash *h, curl_off_t id)
+{
+ struct easy_pollset *ps;
+
+ ps = calloc(1, sizeof(*ps));
+ if(!ps)
+ return NULL;
+ if(!Curl_hash_offt_set(h, id, ps)) {
+ free(ps);
+ return NULL;
+ }
+ return ps;
+}
+
+static struct easy_pollset *
+mev_get_last_pollset(struct Curl_multi *multi,
+ struct Curl_easy *data,
+ struct connectdata *conn)
+{
+ if(data) {
+ if(conn)
+ return Curl_hash_offt_get(&multi->ev.conn_pollsets,
+ conn->connection_id);
+ else if(data)
+ return Curl_hash_offt_get(&multi->ev.xfer_pollsets, data->id);
+ }
+ return NULL;
+}
+
+static void mev_init_cur_pollset(struct easy_pollset *ps,
+ struct Curl_easy *data,
+ struct connectdata *conn)
+{
+ memset(ps, 0, sizeof(*ps));
+ if(conn)
+ Curl_conn_adjust_pollset(data, conn, ps);
+ else if(data)
+ Curl_multi_getsock(data, ps, "ev assess");
+}
+
+static CURLMcode mev_assess(struct Curl_multi *multi,
+ struct Curl_easy *data,
+ struct connectdata *conn)
+{
+ if(multi && multi->socket_cb) {
+ struct easy_pollset ps, *last_ps;
+
+ mev_init_cur_pollset(&ps, data, conn);
+ last_ps = mev_get_last_pollset(multi, data, conn);
+
+ if(!last_ps && ps.num) {
+ if(conn)
+ last_ps = mev_add_new_pollset(&multi->ev.conn_pollsets,
+ data->conn->connection_id);
+ else
+ last_ps = mev_add_new_pollset(&multi->ev.xfer_pollsets, data->id);
+ if(!last_ps)
+ return CURLM_OUT_OF_MEMORY;
+ }
+
+ if(last_ps)
+ return mev_pollset_diff(multi, data, conn, &ps, last_ps);
+ else
+ DEBUGASSERT(!ps.num);
+ }
+ return CURLM_OK;
+}
+
+CURLMcode Curl_multi_ev_assess_xfer(struct Curl_multi *multi,
+ struct Curl_easy *data)
+{
+ return mev_assess(multi, data, NULL);
+}
+
+CURLMcode Curl_multi_ev_assess_conn(struct Curl_multi *multi,
+ struct Curl_easy *data,
+ struct connectdata *conn)
+{
+ return mev_assess(multi, data, conn);
+}
+
+CURLMcode Curl_multi_ev_assess_xfer_list(struct Curl_multi *multi,
+ struct Curl_llist *list)
+{
+ struct Curl_llist_node *e;
+ CURLMcode result = CURLM_OK;
+
+ if(multi && multi->socket_cb) {
+ for(e = Curl_llist_head(list); e && !result; e = Curl_node_next(e)) {
+ result = Curl_multi_ev_assess_xfer(multi, Curl_node_elem(e));
+ }
+ }
+ return result;
+}
+
+
+CURLMcode Curl_multi_ev_assign(struct Curl_multi *multi,
+ curl_socket_t s,
+ void *user_data)
+{
+ struct mev_sh_entry *e = mev_sh_entry_get(&multi->ev.sh_entries, s);
+ if(!e)
+ return CURLM_BAD_SOCKET;
+ e->user_data = user_data;
+ return CURLM_OK;
+}
+
+
+void Curl_multi_ev_expire_xfers(struct Curl_multi *multi,
+ curl_socket_t s,
+ const struct curltime *nowp,
+ bool *run_cpool)
+{
+ struct mev_sh_entry *entry;
+
+ DEBUGASSERT(s != CURL_SOCKET_TIMEOUT);
+ entry = mev_sh_entry_get(&multi->ev.sh_entries, s);
+
+ /* Unmatched socket, we cannot act on it but we ignore this fact. In
+ real-world tests it has been proved that libevent can in fact give
+ the application actions even though the socket was just previously
+ asked to get removed, so thus we better survive stray socket actions
+ and just move on. */
+ if(entry) {
+ struct Curl_hash_iterator iter;
+ struct Curl_hash_element *he;
+
+ /* the socket can be shared by many transfers, iterate */
+ Curl_hash_start_iterate(&entry->xfers, &iter);
+ for(he = Curl_hash_next_element(&iter); he;
+ he = Curl_hash_next_element(&iter)) {
+ struct Curl_easy *data = (struct Curl_easy *)he->ptr;
+ DEBUGASSERT(data);
+ DEBUGASSERT(data->magic == CURLEASY_MAGIC_NUMBER);
+ DEBUGASSERT(data->id >= 0); /* we should not track internal handles */
+
+ /* Expire with out current now, so we will get it below when
+ * asking the splaytree for expired transfers. */
+ Curl_expire_ex(data, nowp, 0, EXPIRE_RUN_NOW);
+ }
+
+ if(Curl_hash_count(&entry->conns))
+ *run_cpool = TRUE;
+ }
+}
+
+void Curl_multi_ev_socket_done(struct Curl_multi *multi,
+ struct Curl_easy *data, curl_socket_t s)
+{
+ mev_forget_socket(multi, data, s, "socket done");
+}
+
+void Curl_multi_ev_xfer_done(struct Curl_multi *multi,
+ struct Curl_easy *data)
+{
+ DEBUGASSERT(!data->conn); /* transfer should have been detached */
+ if(data->id >= 0) {
+ (void)mev_assess(multi, data, NULL);
+ Curl_hash_offt_remove(&multi->ev.xfer_pollsets, data->id);
+ }
+}
+
+void Curl_multi_ev_conn_done(struct Curl_multi *multi,
+ struct Curl_easy *data,
+ struct connectdata *conn)
+{
+ (void)mev_assess(multi, data, conn);
+ Curl_hash_offt_remove(&multi->ev.conn_pollsets, conn->connection_id);
+}
+
+#define CURL_MEV_PS_HASH_SLOTS (991) /* nice prime */
+
+static void mev_hash_pollset_free(void *entry)
+{
+ free(entry);
+}
+
+void Curl_multi_ev_init(struct Curl_multi *multi, size_t hashsize)
+{
+ Curl_hash_init(&multi->ev.sh_entries, hashsize, mev_sh_entry_hash,
+ mev_sh_entry_compare, mev_sh_entry_dtor);
+ Curl_hash_offt_init(&multi->ev.xfer_pollsets,
+ CURL_MEV_PS_HASH_SLOTS, mev_hash_pollset_free);
+ Curl_hash_offt_init(&multi->ev.conn_pollsets,
+ CURL_MEV_PS_HASH_SLOTS, mev_hash_pollset_free);
+}
+
+void Curl_multi_ev_cleanup(struct Curl_multi *multi)
+{
+ Curl_hash_destroy(&multi->ev.sh_entries);
+ Curl_hash_destroy(&multi->ev.xfer_pollsets);
+ Curl_hash_destroy(&multi->ev.conn_pollsets);
+}
--- /dev/null
+#ifndef HEADER_CURL_MULTI_EV_H
+#define HEADER_CURL_MULTI_EV_H
+/***************************************************************************
+ * _ _ ____ _
+ * Project ___| | | | _ \| |
+ * / __| | | | |_) | |
+ * | (__| |_| | _ <| |___
+ * \___|\___/|_| \_\_____|
+ *
+ * Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
+ *
+ * This software is licensed as described in the file COPYING, which
+ * you should have received as part of this distribution. The terms
+ * are also available at https://curl.se/docs/copyright.html.
+ *
+ * You may opt to use, copy, modify, merge, publish, distribute and/or sell
+ * copies of the Software, and permit persons to whom the Software is
+ * furnished to do so, under the terms of the COPYING file.
+ *
+ * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
+ * KIND, either express or implied.
+ *
+ * SPDX-License-Identifier: curl
+ *
+ ***************************************************************************/
+
+struct Curl_easy;
+struct Curl_multi;
+struct easy_pollset;
+
+struct curl_multi_ev {
+ struct Curl_hash sh_entries;
+ struct Curl_hash xfer_pollsets;
+ struct Curl_hash conn_pollsets;
+};
+
+/* Setup/teardown of multi event book-keeping. */
+void Curl_multi_ev_init(struct Curl_multi *multi, size_t hashsize);
+void Curl_multi_ev_cleanup(struct Curl_multi *multi);
+
+/* Assign a 'user_data' to be passed to the socket callback when
+ * invoked with the given socket. This will fail if this socket
+ * is not active, e.g. the application has not been told to monitor it. */
+CURLMcode Curl_multi_ev_assign(struct Curl_multi *multi, curl_socket_t s,
+ void *user_data);
+
+/* Assess the transfer by getting its current pollset, compute
+ * any changes to the last one and inform the application's socket
+ * callback if things have changed. */
+CURLMcode Curl_multi_ev_assess_xfer(struct Curl_multi *multi,
+ struct Curl_easy *data);
+/* Assess all easy handles on the list */
+CURLMcode Curl_multi_ev_assess_xfer_list(struct Curl_multi *multi,
+ struct Curl_llist *list);
+/* Assess the connection by getting its current pollset */
+CURLMcode Curl_multi_ev_assess_conn(struct Curl_multi *multi,
+ struct Curl_easy *data,
+ struct connectdata *conn);
+
+/* Expire all transfers tied to the given socket */
+void Curl_multi_ev_expire_xfers(struct Curl_multi *multi,
+ curl_socket_t s,
+ const struct curltime *nowp,
+ bool *run_cpool);
+
+/* Socket will be closed, forget anything we know about it. */
+void Curl_multi_ev_socket_done(struct Curl_multi *multi,
+ struct Curl_easy *data, curl_socket_t s);
+
+/* Transfer is removed from the multi */
+void Curl_multi_ev_xfer_done(struct Curl_multi *multi,
+ struct Curl_easy *data);
+
+/* Connection is being destroyed */
+void Curl_multi_ev_conn_done(struct Curl_multi *multi,
+ struct Curl_easy *data,
+ struct connectdata *conn);
+
+#endif /* HEADER_CURL_MULTI_EV_H */
#include "llist.h"
#include "hash.h"
#include "conncache.h"
+#include "multi_ev.h"
#include "psl.h"
#include "socketpair.h"
struct CURLMsg extmsg;
};
-/* NOTE: if you add a state here, add the name to the statename[] array as
- well!
-*/
+/* NOTE: if you add a state here, add the name to the statenames[] array
+ * in curl_trc.c as well!
+ */
typedef enum {
MSTATE_INIT, /* 0 - start in this state */
MSTATE_PENDING, /* 1 - no connections, waiting for one */
char *xfer_sockbuf; /* the actual buffer */
size_t xfer_sockbuf_len; /* the allocated length */
- /* 'sockhash' is the lookup hash for socket descriptor => easy handles (note
- the pluralis form, there can be more than one easy handle waiting on the
- same actual socket) */
- struct Curl_hash sockhash;
+ /* multi event related things */
+ struct curl_multi_ev ev;
+
/* `proto_hash` is a general key-value store for protocol implementations
* with the lifetime of the multi handle. The number of elements kept here
* should be in the order of supported protocols (and sub-protocols like
* Prototypes for library-wide functions provided by multi.c
*/
-CURLcode Curl_updatesocket(struct Curl_easy *data);
void Curl_expire(struct Curl_easy *data, timediff_t milli, expire_id);
+void Curl_expire_ex(struct Curl_easy *data,
+ const struct curltime *nowp,
+ timediff_t milli, expire_id id);
bool Curl_expire_clear(struct Curl_easy *data);
void Curl_expire_done(struct Curl_easy *data, expire_id id);
CURLMcode Curl_update_timer(struct Curl_multi *multi) WARN_UNUSED_RESULT;
/* mask for checking if read and/or write is set for index x */
#define GETSOCK_MASK_RW(x) (GETSOCK_READSOCK(x)|GETSOCK_WRITESOCK(x))
-/*
- * Curl_multi_closed()
- *
- * Used by the connect code to tell the multi_socket code that one of the
- * sockets we were using is about to be closed. This function will then
- * remove it from the sockethash for this handle to make the multi_socket API
- * behave properly, especially for the case when libcurl will create another
- * socket again and it gets the same file descriptor number.
- */
-
-void Curl_multi_closed(struct Curl_easy *data, curl_socket_t s);
-
-/* Compare the two pollsets to notify the multi_socket API of changes
- * in socket polling, e.g calling multi->socket_cb() with the changes if
- * differences are seen.
+/**
+ * Let the multi handle know that the socket is about to be closed.
+ * The multi will then remove anything it knows about the socket, so
+ * when the OS is using this socket (number) again subsequently,
+ * the internal book keeping will not get confused.
*/
-CURLMcode Curl_multi_pollset_ev(struct Curl_multi *multi,
- struct Curl_easy *data,
- struct easy_pollset *ps,
- struct easy_pollset *last_ps);
+void Curl_multi_will_close(struct Curl_easy *data, curl_socket_t s);
/*
* Add a handle and move it into PERFORM state at once. For pushed streams.
/* Return the value of the CURLMOPT_MAX_CONCURRENT_STREAMS option */
unsigned int Curl_multi_max_concurrent_streams(struct Curl_multi *multi);
+void Curl_multi_getsock(struct Curl_easy *data,
+ struct easy_pollset *ps,
+ const char *caller);
+
/**
* Borrow the transfer buffer from the multi, suitable
* for the given transfer `data`. The buffer may only be used in one
struct curltime start[2]; /* when filter shutdown started */
unsigned int timeout_ms; /* 0 means no timeout */
} shutdown;
- /* Last pollset used in connection shutdown. Used to detect changes
- * for multi_socket API. */
- struct easy_pollset shutdown_poll;
struct ssl_primary_config ssl_config;
#ifndef CURL_DISABLE_PROXY
struct Curl_message msg; /* A single posted message. */
- /* Array with the plain socket numbers this handle takes care of, in no
- particular order. Note that all sockets are added to the sockhash, where
- the state etc are also kept. This array is mostly used to detect when a
- socket is to be removed from the hash. See singlesocket(). */
- struct easy_pollset last_poll;
-
struct Names dns;
struct Curl_multi *multi; /* if non-NULL, points to the multi handle
struct to which this "belongs" when used by
m = re.match(r'^([0-9:.]+) \[0-[0x]] .+', line)
if m is None:
assert False, f'no match: {line}'
- m = re.match(r'^([0-9:.]+) \[0-[0x]] . \[TCP].+', line)
+ m = re.match(r'^([0-9:.]+) \[0-[0x]] .+ \[TCP].+', line)
if m is not None:
found_tcp = True
if not found_tcp:
###########################################################################
#
import logging
+import os
import re
import pytest
def test_19_01_check_tcp_rst(self, env: Env, httpd, proto):
if env.ci_run:
pytest.skip("seems not to work in CI")
- curl = CurlClient(env=env)
+ run_env = os.environ.copy()
+ if 'CURL_DEBUG' in run_env:
+ del run_env['CURL_DEBUG']
+ curl = CurlClient(env=env, run_env=run_env)
url = f'https://{env.authority_for(env.domain1, proto)}/data.json?[0-1]'
r = curl.http_download(urls=[url], alpn_proto=proto, with_tcpdump=True, extra_args=[
'--parallel'
def test_19_02_check_shutdown(self, env: Env, httpd, proto):
if not env.curl_is_debug():
pytest.skip('only works for curl debug builds')
- curl = CurlClient(env=env, run_env={
+ run_env = os.environ.copy()
+ run_env.update({
'CURL_GRACEFUL_SHUTDOWN': '2000',
- 'CURL_DEBUG': 'ssl,tcp'
+ 'CURL_DEBUG': 'ssl,tcp,lib-ids,multi'
})
+ curl = CurlClient(env=env, run_env=run_env)
url = f'https://{env.authority_for(env.domain1, proto)}/data.json?[0-1]'
r = curl.http_download(urls=[url], alpn_proto=proto, with_tcpdump=True, extra_args=[
'--parallel'
count = 10
curl = CurlClient(env=env, run_env={
'CURL_GRACEFUL_SHUTDOWN': '2000',
- 'CURL_DEBUG': 'ssl'
+ 'CURL_DEBUG': 'ssl,multi'
})
url = f'https://{env.authority_for(env.domain1, proto)}/curltest/tweak/?'\
f'id=[0-{count-1}]&with_cl&close'
r = curl.http_download(urls=[url], alpn_proto=proto)
r.check_response(http_status=200, count=count)
shutdowns = [line for line in r.trace_lines
- if re.match(r'.*CCACHE\] shutdown #\d+, done=1', line)]
+ if re.match(r'.*\[CPOOL\] shutdown, done=1', line)]
assert len(shutdowns) == count, f'{shutdowns}'
# run downloads with CURLOPT_FORBID_REUSE set, meaning *we* close
url = f'https://localhost:{env.https_port}/{docname}'
client = LocalClient(name='hx-download', env=env, run_env={
'CURL_GRACEFUL_SHUTDOWN': '2000',
- 'CURL_DEBUG': 'ssl'
+ 'CURL_DEBUG': 'ssl,multi'
})
if not client.exists():
pytest.skip(f'example client not built: {client.name}')
])
r.check_exit_code(0)
shutdowns = [line for line in r.trace_lines
- if re.match(r'.*CCACHE\] shutdown #\d+, done=1', line)]
+ if re.match(r'.*CPOOL\] shutdown, done=1', line)]
assert len(shutdowns) == count, f'{shutdowns}'
# run event-based downloads with CURLOPT_FORBID_REUSE set, meaning *we* close
if not env.curl_is_debug():
pytest.skip('only works for curl debug builds')
count = 10
- curl = CurlClient(env=env, run_env={
- # forbid connection reuse to trigger shutdowns after transfer
- 'CURL_FORBID_REUSE': '1',
- # make socket receives block 50% of the time to delay shutdown
- 'CURL_DBG_SOCK_RBLOCK': '50',
- 'CURL_DEBUG': 'ssl'
- })
+ run_env = os.environ.copy()
+ # forbid connection reuse to trigger shutdowns after transfer
+ run_env['CURL_FORBID_REUSE'] = '1'
+ # make socket receives block 50% of the time to delay shutdown
+ run_env['CURL_DBG_SOCK_RBLOCK'] = '50'
+ run_env['CURL_DEBUG'] = 'ssl,multi,lib-ids'
+ curl = CurlClient(env=env, run_env=run_env)
url = f'https://{env.authority_for(env.domain1, proto)}/curltest/tweak/?'\
f'id=[0-{count-1}]&with_cl&'
r = curl.http_download(urls=[url], alpn_proto=proto, extra_args=[
r.check_response(http_status=200, count=count)
# check that we closed all connections
closings = [line for line in r.trace_lines
- if re.match(r'.*CCACHE\] closing #\d+', line)]
+ if re.match(r'.*CPOOL\] closing', line)]
assert len(closings) == count, f'{closings}'
# check that all connection sockets were removed from event
removes = [line for line in r.trace_lines
r.check_response(http_status=200, count=2)
# check connection cache closings
shutdowns = [line for line in r.trace_lines
- if re.match(r'.*CCACHE\] shutdown #\d+, done=1', line)]
+ if re.match(r'.*CPOOL\] shutdown, done=1', line)]
assert len(shutdowns) == 1, f'{shutdowns}'
Curl_infof(testdata, "%s", (char *)NULL);
fail_unless(verify(output, "(nil)") == 0, "Passing NULL as string");
+/* Note: libcurl's tracebuffer hold 2048 bytes, so the max strlen() we
+ * get out of it is 2047, since we need a \0 at the end.
+ * Curl_infof() in addition adds a \n at the end, making the effective
+ * output 2046 characters.
+ * Any input that long or longer will truncated, ending in '...\n'.
+ */
+
/* A string just long enough to not be truncated */
memset(input, '\0', sizeof(input));
-memset(input, 'A', 2047);
+memset(input, 'A', 2045);
Curl_infof(testdata, "%s", input);
-fail_unless(strlen(output) == 2048, "No truncation of infof input");
+fprintf(stderr, "output len %d: %s", (int)strlen(output), output);
+/* output is input + \n */
+fail_unless(strlen(output) == 2046, "No truncation of infof input");
fail_unless(verify(output, input) == 0, "No truncation of infof input");
fail_unless(output[sizeof(output) - 1] == '\0',
"No truncation of infof input");
/* Just over the limit without newline for truncation via '...' */
-memset(input + 2047, 'A', 4);
+memset(input + 2045, 'A', 4);
Curl_infof(testdata, "%s", input);
-fail_unless(strlen(output) == 2051, "Truncation of infof input 1");
+fprintf(stderr, "output len %d: %s", (int)strlen(output), output);
+fail_unless(strlen(output) == 2047, "Truncation of infof input 1");
fail_unless(output[sizeof(output) - 1] == '\0', "Truncation of infof input 1");
/* Just over the limit with newline for truncation via '...' */
-memset(input + 2047, 'A', 4);
-memset(input + 2047 + 4, '\n', 1);
+memset(input + 2045, 'A', 4);
+memset(input + 2045 + 4, '\n', 1);
Curl_infof(testdata, "%s", input);
-fail_unless(strlen(output) == 2051, "Truncation of infof input 2");
+fprintf(stderr, "output len %d: %s", (int)strlen(output), output);
+fail_unless(strlen(output) == 2047, "Truncation of infof input 2");
fail_unless(output[sizeof(output) - 1] == '\0', "Truncation of infof input 2");
/* Way over the limit for truncation via '...' */
memset(input, '\0', sizeof(input));
memset(input, 'A', sizeof(input) - 1);
Curl_infof(testdata, "%s", input);
-fail_unless(strlen(output) == 2051, "Truncation of infof input 3");
+fprintf(stderr, "output len %d: %s", (int)strlen(output), output);
+fail_unless(strlen(output) == 2047, "Truncation of infof input 3");
fail_unless(output[sizeof(output) - 1] == '\0', "Truncation of infof input 3");
#if defined(CURL_GNUC_DIAG) && !defined(__clang__)