Index: server/mpm_common.c
===================================================================
RCS file: /home/cvs/httpd-2.0/server/mpm_common.c,v
retrieving revision 1.121
diff -u -r1.121 mpm_common.c
--- server/mpm_common.c	14 Aug 2004 10:49:43 -0000	1.121
+++ server/mpm_common.c	15 Sep 2004 17:40:30 -0000
@@ -61,22 +61,55 @@
 #ifdef AP_MPM_WANT_RECLAIM_CHILD_PROCESSES
 void ap_reclaim_child_processes(int terminate)
 {
-    int i;
-    long int waittime = 1024 * 16;      /* in usecs */
+    apr_time_t waittime = 1024 * 16;
     apr_status_t waitret;
-    int tries;
+    int i;
     int not_dead_yet;
     int max_daemons;
+    apr_time_t starttime = apr_time_now();
+    /* this table of actions and elapsed times tells what action is taken
+     * at which elapsed time from starting the reclaim
+     */
+    struct {
+        enum {DO_NOTHING, SEND_SIGTERM, SEND_SIGKILL, GIVEUP} action;
+        apr_time_t action_time;
+    } action_table[] = {
+        {DO_NOTHING, 0}, /* dummy entry for iterations where we reap
+                          * children but take no action against
+                          * stragglers
+                          */
+        {SEND_SIGTERM, apr_time_from_sec(3)},
+        {SEND_SIGTERM, apr_time_from_sec(4)},
+        {SEND_SIGTERM, apr_time_from_sec(5)},
+        {SEND_SIGKILL, apr_time_from_sec(7)},
+        {GIVEUP,       apr_time_from_sec(9)}
+    };
+    int cur_action;      /* index of action we decided to take this
+                          * iteration
+                          */
+    int next_action = 1; /* index of first real action */
 
     ap_mpm_query(AP_MPMQ_MAX_DAEMON_USED, &max_daemons);
 
-    for (tries = terminate ? 4 : 1; tries <= 9; ++tries) {
-        /* don't want to hold up progress any more than
-         * necessary, but we need to allow children a few moments to exit.
-         * Set delay with an exponential backoff.
-         */
+    while (1) {
         apr_sleep(waittime);
+        /* don't let waittime get longer than 1 second; otherwise, we don't
+         * react quickly to the last child exiting, and taking action can
+         * be delayed
+         */
         waittime = waittime * 4;
+        if (waittime > apr_time_from_sec(1)) {
+            waittime = apr_time_from_sec(1);
+        }
+
+        /* see what action to take, if any */
+        if (action_table[next_action].action_time <= apr_time_now() - starttime) {
+            cur_action = next_action;
+            ++next_action;
+        }
+        else {
+            cur_action = 0; /* nothing to do */
+        }
 
         /* now see who is done */
         not_dead_yet = 0;
@@ -95,16 +128,11 @@
             }
 
             ++not_dead_yet;
-            switch (tries) {
-            case 1:     /*  16ms */
-            case 2:     /*  82ms */
-            case 3:     /* 344ms */
-            case 4:     /*  16ms */
+            switch(action_table[cur_action].action) {
+            case DO_NOTHING:
                 break;
-
-            case 5:     /*  82ms */
-            case 6:     /* 344ms */
-            case 7:     /* 1.4sec */
+                
+            case SEND_SIGTERM:
                 /* ok, now it's being annoying */
                 ap_log_error(APLOG_MARK, APLOG_WARNING,
                              0, ap_server_conf,
@@ -113,9 +141,8 @@
                              (long)pid);
                 kill(pid, SIGTERM);
                 break;
-
-            case 8:     /*  6 sec */
-                /* die child scum */
+                
+            case SEND_SIGKILL:
                 ap_log_error(APLOG_MARK, APLOG_ERR,
                              0, ap_server_conf,
                              "child process %ld still did not exit, "
@@ -133,8 +160,8 @@
                 kill_thread(pid);
 #endif
                 break;
-
-            case 9:     /* 14 sec */
+                
+            case GIVEUP:
                 /* gave it our best shot, but alas...  If this really
                  * is a child we are trying to kill and it really hasn't
                  * exited, we will likely fail to bind to the port
@@ -153,8 +180,9 @@
         apr_proc_other_child_refresh_all(APR_OC_REASON_RESTART);
 #endif
 
-        if (!not_dead_yet) {
-            /* nothing left to wait for */
+        if (!not_dead_yet ||
+            action_table[cur_action].action == GIVEUP) {
+            /* nothing left to wait for, or we gave up */
             break;
         }
     }