Merge pull request #374 from fermionic/diaspora-prevent-fcontact-race-conditions

Prevent fcontact race conditions

boot.php

@@ -12,7 +12,7 @@ require_once('include/cache.php');
 define ( 'FRIENDICA_PLATFORM',     'Friendica');
 define ( 'FRIENDICA_VERSION',      '3.0.1394' );
 define ( 'DFRN_PROTOCOL_VERSION',  '2.23'    );
-define ( 'DB_UPDATE_VERSION',      1151      );
+define ( 'DB_UPDATE_VERSION',      1152      );
 
 define ( 'EOL',                    "<br />\r\n"     );
 define ( 'ATOM_TIME',              'Y-m-d\TH:i:s\Z' );

database.sql

@@ -598,6 +598,19 @@ CREATE TABLE IF NOT EXISTS `item_id` (
 
 -- --------------------------------------------------------
 
+--
+-- Table structure for table `locks`
+--
+
+CREATE TABLE `locks` (
+  `id` int(11) NOT NULL AUTO_INCREMENT,
+  `name` char(128) NOT NULL,
+  `locked` tinyint(1) NOT NULL DEFAULT '0',
+  PRIMARY KEY (`id`)
+) ENGINE=MyISAM DEFAULT CHARSET=utf8;
+
+-- --------------------------------------------------------
+
 --
 -- Table structure for table `mail`
 --

include/diaspora.php

@@ -5,6 +5,7 @@ require_once('include/items.php');
 require_once('include/bb2diaspora.php');
 require_once('include/contact_selectors.php');
 require_once('include/queue_fn.php');
+require_once('include/lock.php');
 
 
 function diaspora_dispatch_public($msg) {
@@ -113,27 +114,83 @@ function diaspora_get_contact_by_handle($uid,$handle) {
 }
 
 function find_diaspora_person_by_handle($handle) {
+
+	$person = false;
 	$update = false;
+	$got_lock = false;
+
+	do {
 		$r = q("select * from fcontact where network = '%s' and addr = '%s' limit 1",
 			dbesc(NETWORK_DIASPORA),
 			dbesc($handle)
 		);
 		if(count($r)) {
+			$person = $r[0];
 			logger('find_diaspora_person_by handle: in cache ' . print_r($r,true), LOGGER_DEBUG);
+
 			// update record occasionally so it doesn't get stale
-		$d = strtotime($r[0]['updated'] . ' +00:00');
+			$d = strtotime($person['updated'] . ' +00:00');
-		if($d > strtotime('now - 14 days'))
+			if($d < strtotime('now - 14 days'))
-			return $r[0];
 				$update = true;
 		}
-	logger('find_diaspora_person_by_handle: refresh',LOGGER_DEBUG);
+
+
+		// FETCHING PERSON INFORMATION FROM REMOTE SERVER
+		//
+		// If the person isn't in our 'fcontact' table, or if he/she is but
+		// his/her information hasn't been updated for more than 14 days, then
+		// we want to fetch the person's information from the remote server.
+		//
+		// Note that $person isn't changed by this block of code unless the
+		// person's information has been successfully fetched from the remote
+		// server. So if $person was 'false' to begin with (because he/she wasn't
+		// in the local cache), it'll stay false, and if $person held the local
+		// cache information to begin with, it'll keep that information. That way
+		// if there's a problem with the remote fetch, we can at least use our
+		// cached information--it's better than nothing.
+
+		if((! $person) || ($update))  {
+			// Lock the function to prevent race conditions if multiple items
+			// come in at the same time from a person who doesn't exist in
+			// fcontact
+			$got_lock = lock_function('find_diaspora_person_by_handle', false);
+
+			if($got_lock) {
+				logger('find_diaspora_person_by_handle: create or refresh', LOGGER_DEBUG);
 				require_once('include/Scrape.php');
 				$r = probe_url($handle, PROBE_DIASPORA);
+
+				// Note that Friendica contacts can return a "Diaspora person"
+				// if Diaspora connectivity is enabled on their server
 				if((count($r)) && ($r['network'] === NETWORK_DIASPORA)) {
 					add_fcontact($r,$update);
-		return ($r);
+					$person = ($r);
 				}
-	return false;
+
+				unlock_function('find_diaspora_person_by_handle');
+			}
+			else {
+				logger('find_diaspora_person_by_handle: couldn\'t lock function', LOGGER_DEBUG);
+				if(! $person)
+					block_on_function_lock('find_diaspora_person_by_handle');
+			}
+		}
+	} while((! $person) && (! $got_lock));
+	// We need to try again if the person wasn't in 'fcontact' but the function was locked.
+	// The fact that the function was locked may mean that another process was creating the
+	// person's record. It could also mean another process was creating or updating an unrelated
+	// person.
+	//
+	// At any rate, we need to keep trying until we've either got the person or had a chance to
+	// try to fetch his/her remote information. But we don't want to block on locking the
+	// function, because if the other process is creating the record, then when we acquire the lock
+	// we'll dive right into creating another, duplicate record. We DO want to at least wait
+	// until the lock is released, so we don't flood the database with requests.
+	//
+	// If the person was in the 'fcontact' table, don't try again. It's not worth the time, since
+	// we do have some information for the person
+
+	return $person;
 }
 
 

include/lock.php

@@ -0,0 +1,76 @@
+<?php
+
+// Provide some ability to lock a PHP function so that multiple processes
+// can't run the function concurrently
+if(! function_exists('lock_function')) {
+function lock_function($fn_name, $block = true, $wait_sec = 2) {
+	if( $wait_sec == 0 )
+		$wait_sec = 2;	// don't let the user pick a value that's likely to crash the system
+
+	$got_lock = false;
+
+	do {
+		q("LOCK TABLE locks WRITE");
+		$r = q("SELECT locked FROM locks WHERE name = '%s' LIMIT 1",
+			dbesc($fn_name)
+		);
+
+		if((count($r)) && (! $r[0]['locked'])) {
+			q("UPDATE locks SET locked = 1 WHERE name = '%s' LIMIT 1",
+				dbesc($fn_name)
+			);
+			$got_lock = true;
+		}
+		elseif(! $r) { // the Boolean value for count($r) should be equivalent to the Boolean value of $r
+			q("INSERT INTO locks ( name, locked ) VALUES ( '%s', 1 )",
+				dbesc($fn_name)
+			);
+			$got_lock = true;
+		}
+
+		q("UNLOCK TABLES");
+
+		if(($block) && (! $got_lock))
+			sleep($wait_sec);
+
+	} while(($block) && (! $got_lock));
+
+	logger('lock_function: function ' . $fn_name . ' with blocking = ' . $block . ' got_lock = ' . $got_lock, LOGGER_DEBUG);
+	
+	return $got_lock;
+}}
+
+
+if(! function_exists('block_on_function_lock')) {
+function block_on_function_lock($fn_name, $wait_sec = 2) {
+	if( $wait_sec == 0 )
+		$wait_sec = 2;	// don't let the user pick a value that's likely to crash the system
+
+	do {
+		$r = q("SELECT locked FROM locks WHERE name = '%s' LIMIT 1",
+				dbesc($fn_name)
+		     );
+
+		if(count($r) && $r[0]['locked'])
+			sleep($wait_sec);
+
+	} while(count($r) && $r[0]['locked']);
+
+	return;
+}}
+
+
+if(! function_exists('unlock_function')) {
+function unlock_function($fn_name) {
+	//$r = q("LOCK TABLE lock WRITE");
+	$r = q("UPDATE locks SET locked = 0 WHERE name = '%s' LIMIT 1",
+			dbesc($fn_name)
+	     );
+	//$r = q("UNLOCK TABLES");
+
+	logger('unlock_function: released lock for function ' . $fn_name, LOGGER_DEBUG);
+
+	return;
+}}
+
+?>

update.php

@@ -1,6 +1,6 @@
 <?php
 
-define( 'UPDATE_VERSION' , 1151 );
+define( 'UPDATE_VERSION' , 1152 );
 
 /**
  *
@@ -1307,3 +1307,15 @@ function update_1150() {
 	return UPDATE_SUCCESS;
 }
 
+
+function update_1151() {
+	$r = q("CREATE TABLE IF NOT EXISTS locks (
+			id INT NOT NULL AUTO_INCREMENT PRIMARY KEY ,
+			name CHAR( 128 ) NOT NULL ,
+			locked TINYINT( 1 ) NOT NULL DEFAULT '0'
+		  ) ENGINE = MYISAM DEFAULT CHARSET=utf8 ");
+	if (!$r)
+		return UPDATE_FAILED;
+	return UPDATE_SUCCESS;
+}
+