- curl_multi_timeout() could return a timeout value of 0 even though …

…nothing

  was actually ready to get done, as the internal time resolution is higher
  than the returned millisecond timer. Therefore it could cause applications
  running on fast processors to do short bursts of busy-loops.
  curl_multi_timeout() will now only return 0 if the timeout is actually
  alreay triggered.

CHANGES

@@ -7,6 +7,13 @@
                                   Changelog
 
 Daniel Stenberg (19 Dec 2008)
+- curl_multi_timeout() could return a timeout value of 0 even though nothing
+  was actually ready to get done, as the internal time resolution is higher
+  than the returned millisecond timer. Therefore it could cause applications
+  running on fast processors to do short bursts of busy-loops.
+  curl_multi_timeout() will now only return 0 if the timeout is actually
+  alreay triggered.
+
 - Using the libssh2 0.19 function libssh2_session_block_directions(), libcurl
   now has an improved ability to do right when the multi interface (both
   "regular" and multi_socket) is used for SCP and SFTP transfers. This should

RELEASE-NOTES

@@ -34,6 +34,8 @@ This release includes the following bugfixes:
  o improved connection re-use for subsequent SCP and SFTP transfers
  o multi interface does less busy-loops for SCP and SFTP transfers with libssh2
    0.19 or later
+ o curl_multi_timeout() no longer returns timeout 0 when there's nothing to do
+   yet
 
 This release includes the following known bugs:
 

lib/multi.c

@@ -1541,6 +1541,7 @@ CURLMcode curl_multi_perform(CURLM *multi_handle, int *running_handles)
 
   if( CURLM_OK >= returncode )
     update_timer(multi);
+
   return returncode;
 }
 
@@ -1951,9 +1952,19 @@ static CURLMcode multi_timeout(struct Curl_multi *multi,
     /* splay the lowest to the bottom */
     multi->timetree = Curl_splay(tv_zero, multi->timetree);
 
-    if(Curl_splaycomparekeys(multi->timetree->key, now) > 0)
+    if(Curl_splaycomparekeys(multi->timetree->key, now) > 0) {
       /* some time left before expiration */
       *timeout_ms = curlx_tvdiff(multi->timetree->key, now);
+      if(!*timeout_ms)
+        /*
+         * Since we only provide millisecond resolution on the returned value
+         * and the diff might be less than one millisecond here, we don't
+         * return zero as that may cause short bursts of busyloops on fast
+         * processors while the diff is still present but less than one
+         * millisecond! instead we return 1 until the time is ripe.
+         */
+        *timeout_ms=1;
+    }
     else
       /* 0 means immediately */
       *timeout_ms = 0;