Added logic to the worker MPM to skip the poll before accept

if there's only one listener socket (based on the implementation
in the prefork MPM).  This also enables us to do single-listener
unserialized accept

git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/trunk@93096 13f79535-47bb-0310-9956-ffa450edef68

diff --git a/CHANGES b/CHANGES
index 25a6de18bf24eb09dce38bb4208066abe2a439c0..daba118f8f4271ccc6a0142e4c9de7cea93c2630 100644 (file)
--- a/CHANGES
+++ b/CHANGES
@@ -1,5 +1,8 @@
 Changes with Apache 2.0.31-dev
 
+  *) Added single-listener unserialized accept support to the
+     worker MPM [Brian Pane]
+
   *) New Directive for mod_proxy: 'ProxyPreserveHost'. This passes
      the incoming host header through to the proxied server
      [Geoff <g.russell@ieee.org>]

diff --git a/STATUS b/STATUS
index dad73d49e8fd443b640ff3ba383a886084061b03..9d71ce2a8ddaeecbbfa41af6c0a10236d63bb8b3 100644 (file)
--- a/STATUS
+++ b/STATUS
@@ -1,5 +1,5 @@
 APACHE 2.0 STATUS:                                             -*-text-*-
-Last modified at [$Date: 2002/01/28 00:41:31 $]
+Last modified at [$Date: 2002/01/30 06:33:18 $]
 
 Release:
 
@@ -287,10 +287,6 @@ RELEASE NON-SHOWSTOPPERS BUT WOULD BE REAL NICE TO WRAP THESE UP:
                 also thinks that the https filter logic may be sufficiently
                 decoupled now, but isn't really sure.
 
-    * Performance: Get the SINGLE_LISTEN_UNSERIALIZED_ACCEPT
-      optimization working in worker.  prefork's new design for how
-      to notice data on the pod should be sufficient.
-
     * Performance & Debug: Eliminate most (and perhaps all) of the 
       malloc/free calls in the bucket brigade code.  Need some 
       light weight memory management functions that allow freeing 

diff --git a/server/mpm/worker/worker.c b/server/mpm/worker/worker.c
index 52259b56d8b25bfad6ccd35c094d942b7067465c..00660cd7f87e77b3015ba78ca8891a15b2f5c9b5 100644 (file)
--- a/server/mpm/worker/worker.c
+++ b/server/mpm/worker/worker.c
@@ -196,10 +196,7 @@ typedef struct {
  */
 int ap_max_daemons_limit = -1;
 
-static apr_file_t *pipe_of_death_in = NULL;
-static apr_file_t *pipe_of_death_out = NULL;
-/* insures that a child process only consumes one character */
-static apr_thread_mutex_t *pipe_of_death_mutex;
+static ap_pod_t *pod;
 
 /* *Non*-shared http_main globals... */
 
@@ -237,8 +234,8 @@ static apr_thread_mutex_t *worker_thread_count_mutex;
 /* Locks for accept serialization */
 static apr_proc_mutex_t *accept_mutex;
 
-#ifdef NO_SERIALIZED_ACCEPT
-#define SAFE_ACCEPT(stmt) APR_SUCCESS
+#ifdef SINGLE_LISTEN_UNSERIALIZED_ACCEPT
+#define SAFE_ACCEPT(stmt) (ap_listeners->next ? (stmt) : APR_SUCCESS)
 #else
 #define SAFE_ACCEPT(stmt) (stmt)
 #endif
@@ -588,35 +585,6 @@ static void check_infinite_requests(void)
     }
 }
 
-/* Sets workers_may_exit if we received a character on the pipe_of_death */
-static apr_status_t check_pipe_of_death(void **csd, ap_listen_rec *lr, 
-                                        apr_pool_t *ptrans)
-{
-    *csd = NULL;
-    apr_thread_mutex_lock(pipe_of_death_mutex);
-    if (!workers_may_exit) {
-        apr_status_t ret;
-        char pipe_read_char;
-        apr_size_t n = 1;
-
-        ret = apr_recv(lr->sd, &pipe_read_char, &n);
-        if (APR_STATUS_IS_EAGAIN(ret)) {
-            /* It lost the lottery. It must continue to suffer
-             * through a life of servitude. */
-        }
-        else {
-            /* It won the lottery (or something else is very
-             * wrong). Embrace death with open arms. */
-            signal_workers();
-        }
-    }
-    apr_thread_mutex_unlock(pipe_of_death_mutex);
-    /* This is a hack to get us back to the top of the accept loop.
-     * we should probably have a better way to do this though.
-     */
-    return APR_EINTR;
-}
-
 static void *listener_thread(apr_thread_t *thd, void * dummy)
 {
     proc_info * ti = dummy;
@@ -657,31 +625,30 @@ static void *listener_thread(apr_thread_t *thd, void * dummy)
             signal_workers();
         }
 
-        while (!workers_may_exit) {
-            apr_status_t ret;
-            apr_int16_t event;
+        if (!ap_listeners->next) {
+            /* Only one listener, so skip the poll */
+            lr = ap_listeners;
+        }
+        else {
+            while (!workers_may_exit) {
+                apr_status_t ret;
+                apr_int16_t event;
+
+                ret = apr_poll(pollset, &n, -1);
+                if (ret != APR_SUCCESS) {
+                    if (APR_STATUS_IS_EINTR(ret)) {
+                        continue;
+                    }
 
-            ret = apr_poll(pollset, &n, -1);
-            if (ret != APR_SUCCESS) {
-                if (APR_STATUS_IS_EINTR(ret)) {
-                    continue;
+                    /* apr_poll() will only return errors in catastrophic
+                     * circumstances. Let's try exiting gracefully, for now. */
+                    ap_log_error(APLOG_MARK, APLOG_ERR, ret, (const server_rec *)
+                                 ap_server_conf, "apr_poll: (listen)");
+                    signal_workers();
                 }
 
-                /* apr_poll() will only return errors in catastrophic
-                 * circumstances. Let's try exiting gracefully, for now. */
-                ap_log_error(APLOG_MARK, APLOG_ERR, ret, (const server_rec *)
-                             ap_server_conf, "apr_poll: (listen)");
-                signal_workers();
-            }
+                if (workers_may_exit) break;
 
-            if (workers_may_exit) break;
-
-            if (ap_listeners->next == NULL) {
-                /* only one listener */
-                lr = ap_listeners;
-                goto got_fd;
-            }
-            else {
                 /* find a listener */
                 lr = last_lr;
                 do {
@@ -728,6 +695,10 @@ static void *listener_thread(apr_thread_t *thd, void * dummy)
                                  rv);
                 }
             }
+            if (ap_mpm_pod_check(pod) == APR_SUCCESS) {
+                signal_workers();
+                break;
+            }
         }
         else {
             if ((rv = SAFE_ACCEPT(apr_proc_mutex_unlock(accept_mutex)))
@@ -951,8 +922,6 @@ static void child_main(int child_num_arg)
     worker_thread_count = 0;
     apr_thread_mutex_create(&worker_thread_count_mutex,
                             APR_THREAD_MUTEX_DEFAULT, pchild);
-    apr_thread_mutex_create(&pipe_of_death_mutex,
-                            APR_THREAD_MUTEX_DEFAULT, pchild);
 
     ts = (thread_starter *)apr_palloc(pchild, sizeof(*ts));
 
@@ -1071,15 +1040,7 @@ static void wake_up_and_die(void)
     apr_size_t one = 1;
     apr_status_t rv;
     
-    for (i = 0; i < ap_daemons_limit;) {
-        if ((rv = apr_file_write(pipe_of_death_out, &char_of_death, &one)) 
-                                 != APR_SUCCESS) {
-            if (APR_STATUS_IS_EINTR(rv)) continue;
-            ap_log_error(APLOG_MARK, APLOG_WARNING, rv, ap_server_conf, 
-                         "write pipe_of_death");
-        }
-        i++;
-    }
+    ap_mpm_pod_killpg(pod, ap_daemons_limit);
 }
 
 /* start up a bunch of children */
@@ -1202,13 +1163,8 @@ static void perform_idle_server_maintenance(void)
     ap_max_daemons_limit = last_non_dead + 1;
 
     if (idle_thread_count > max_spare_threads) {
-        char char_of_death = '!';
         /* Kill off one child */
-        if ((rv = apr_file_write(pipe_of_death_out, &char_of_death, 
-                                 &one)) != APR_SUCCESS) {
-            ap_log_error(APLOG_MARK, APLOG_WARNING, rv, ap_server_conf, 
-                         "write pipe_of_death");
-        }
+        ap_mpm_pod_signal(pod);
         idle_spawn_rate = 1;
     }
     else if (idle_thread_count < min_spare_threads) {
@@ -1334,41 +1290,6 @@ static void server_main_loop(int remaining_children_to_start)
     }
 }
 
-static void make_pipe_of_death(int *num_listeners, apr_pool_t *p)
-{
-    ap_listen_rec *lr = apr_palloc(p, sizeof(*lr));
-    int filedes;
-    apr_socket_t *sd = NULL;
-    apr_status_t rv;
-
-    rv = apr_file_pipe_create(&pipe_of_death_in, &pipe_of_death_out, p);
-    if (rv != APR_SUCCESS) {
-        ap_log_error(APLOG_MARK, APLOG_ERR, rv,
-                     (const server_rec*) ap_server_conf,
-                     "apr_file_pipe_create (pipe_of_death)");
-        exit(1);
-    }
-
-    if ((rv = apr_file_pipe_timeout_set(pipe_of_death_in, 0)) != APR_SUCCESS) {
-        ap_log_error(APLOG_MARK, APLOG_ERR, rv,
-                     (const server_rec*) ap_server_conf,
-                     "apr_file_pipe_timeout_set (pipe_of_death)");
-        exit(1);
-    }
-
-    apr_os_file_get(&filedes, pipe_of_death_in);
-    apr_os_sock_put(&sd, &filedes, p);
-
-    lr->sd = sd;
-    lr->active = 1;
-    lr->accept_func = check_pipe_of_death;
-    /* We are not bound to a real address.  So, indicate that. */
-    lr->bind_addr = 0;
-    lr->next = ap_listeners;
-    ap_listeners = lr;
-    (*num_listeners)++;
-}
-
 int ap_mpm_run(apr_pool_t *_pconf, apr_pool_t *plog, server_rec *s)
 {
     int remaining_children_to_start;
@@ -1391,7 +1312,13 @@ int ap_mpm_run(apr_pool_t *_pconf, apr_pool_t *plog, server_rec *s)
             "no listening sockets available, shutting down");
         return 1;
     }
-    make_pipe_of_death(&num_listensocks, pconf);
+
+    if ((rv = ap_mpm_pod_open(pconf, &pod))) {
+        ap_log_error(APLOG_MARK, APLOG_CRIT, rv, s,
+                     "Could not open pipe-of-death.");
+        return 1;
+    }
+
     ap_log_pid(pconf, ap_pid_fname);
 
     /* Initialize cross-process accept lock */
@@ -1533,10 +1460,6 @@ int ap_mpm_run(apr_pool_t *_pconf, apr_pool_t *plog, server_rec *s)
                     "SIGHUP received.  Attempting to restart");
     }
 
-    /* get the pipe of death out of the listen_rec list */
-    ap_assert(ap_listeners->bind_addr == 0);
-    ap_listeners = ap_listeners->next;
-
     return 0;
 }