Remove malloc from entry appending

Added documentation to mention that user is responsible for managing
memory.

Author: willemt

include/raft.h

@@ -206,7 +206,7 @@ typedef int (
     );
 
 /** Callback for saving who we voted for to disk.
- * This callback MUST flush the change to disk.
+ * For safety reasons this callback MUST flush the change to disk.
  * @param[in] raft The Raft server making this callback
  * @param[in] user_data User data that is passed from Raft server
  * @param[in] voted_for The node we voted for
@@ -220,16 +220,21 @@ typedef int (
     );
 
 /** Callback for saving log entry changes.
+ *
  * This callback is used for:
  * <ul>
  *      <li>Adding entries to the log (ie. offer)
  *      <li>Removing the first entry from the log (ie. polling)
  *      <li>Removing the last entry from the log (ie. popping)
  * </ul>
- * This callback MUST flush the change to disk.
+ *
+ * For safety reasons this callback MUST flush the change to disk.
+ *
  * @param[in] raft The Raft server making this callback
  * @param[in] user_data User data that is passed from Raft server
- * @param[in] entry The entry that the event is happening to
+ * @param[in] entry The entry that the event is happening to.
+ *    The user is allowed to change the memory pointed to in the
+ *    raft_entry_data_t struct. This MUST be done if the memory is temporary.
  * @param[in] entry_idx The entries index in the log
  * @return 0 on success */
 typedef int (
@@ -253,23 +258,27 @@ typedef struct
     func_applylog_f applylog;
 
     /** Callback for persisting vote data
-     * This callback MUST flush the change to disk. */
+     * For safety reasons this callback MUST flush the change to disk. */
     func_persist_int_f persist_vote;
 
     /** Callback for persisting term data
-     * This callback MUST flush the change to disk. */
+     * For safety reasons this callback MUST flush the change to disk. */
     func_persist_int_f persist_term;
 
     /** Callback for adding an entry to the log
-     * This callback MUST flush the change to disk. */
+     * For safety reasons this callback MUST flush the change to disk. */
     func_logentry_event_f log_offer;
 
     /** Callback for removing the oldest entry from the log
-     * This callback MUST flush the change to disk. */
+     * For safety reasons this callback MUST flush the change to disk.
+     * @note If memory was malloc'd in log_offer then this should be the right
+     *  time to free the memory. */
     func_logentry_event_f log_poll;
 
     /** Callback for removing the youngest entry from the log
-     * This callback MUST flush the change to disk. */
+     * For safety reasons this callback MUST flush the change to disk.
+     * @note If memory was malloc'd in log_offer then this should be the right
+     *  time to free the memory. */
     func_logentry_event_f log_pop;
 
     /** Callback for catching debugging log messages
@@ -318,11 +327,11 @@ __attribute__ ((deprecated));
 /** Add node.
  *
  * @note This library does not yet support membership changes.
- * Once raft_periodic has been run this will fail.
+ *  Once raft_periodic has been run this will fail.
  *
  * @note The order this call is made is important.
- * This call MUST be made in the same order as the other raft nodes.
- * This is because the node ID is assigned depending on when this call is made
+ *  This call MUST be made in the same order as the other raft nodes.
+ *  This is because the node ID is assigned depending on when this call is made
  *
  * @param[in] user_data The user data for the node.
  *  This is obtained using raft_node_get_udata.
@@ -351,7 +360,18 @@ void raft_set_request_timeout(raft_server_t* me, int msec);
 int raft_periodic(raft_server_t* me, int msec_elapsed);
 
 /** Receive an appendentries message.
- * This function will block if it needs to append the message.
+ *
+ * Will block (ie. by syncing to disk) if we need to append a message.
+ *
+ * Might call malloc once to increase the log entry array size.
+ *
+ * The log_offer callback will be called.
+ *
+ * @note The memory pointer (ie. raft_entry_data_t) for each msg_entry_t is
+ *   copied directly. If the memory is temporary you MUST either make the
+ *   memory permanent (ie. via malloc) OR re-assign the memory within the
+ *   log_offer callback.
+ *
  * @param[in] node Index of the node who sent us this message
  * @param[in] ae The appendentries message
  * @param[out] r The resulting response
@@ -387,11 +407,20 @@ int raft_recv_requestvote_response(raft_server_t* me,
                                    int node,
                                    msg_requestvote_response_t* r);
 
-/** Receive an entry message from client.
+/** Receive an entry message from the client.
  *
  * Append the entry to the log and send appendentries to followers.
  *
- * This function will block if it needs to append the message.
+ * Will block (ie. by syncing to disk) if we need to append a message.
+ *
+ * Might call malloc once to increase the log entry array size.
+ *
+ * The log_offer callback will be called.
+ *
+ * @note The memory pointer (ie. raft_entry_data_t) in msg_entry_t is
+ *  copied directly. If the memory is temporary you MUST either make the
+ *  memory permanent (ie. via malloc) OR re-assign the memory within the
+ *  log_offer callback.
  *
  * Will fail:
  * <ul>

src/raft_server.c

@@ -314,14 +314,11 @@ int raft_recv_appendentries(
     {
         msg_entry_t* cmd = &ae->entries[i];
 
-        /* TODO: replace malloc with mempoll/arena */
-        raft_entry_t* c = (raft_entry_t*)malloc(sizeof(raft_entry_t));
-        c->term = cmd->term;
-        memcpy(&c->data, &cmd->data, sizeof(raft_entry_data_t));
-        c->data.buf = (unsigned char*)malloc(cmd->data.len);
-        memcpy(c->data.buf, cmd->data.buf, cmd->data.len);
-        c->id = cmd->id;
-        int e = raft_append_entry(me_, c);
+        raft_entry_t ety;
+        ety.term = cmd->term;
+        ety.id = cmd->id;
+        memcpy(&ety.data, &cmd->data, sizeof(raft_entry_data_t));
+        int e = raft_append_entry(me_, &ety);
         if (-1 == e)
         {
             __log(me_, "AE failure; couldn't append entry");
@@ -423,19 +420,17 @@ int raft_recv_entry(raft_server_t* me_, int node, msg_entry_t* e,
                     msg_entry_response_t *r)
 {
     raft_server_private_t* me = (raft_server_private_t*)me_;
-    raft_entry_t ety;
     int i;
 
     if (!raft_is_leader(me_))
         return -1;
 
     __log(me_, "received entry from: %d", node);
 
+    raft_entry_t ety;
     ety.term = me->current_term;
     ety.id = e->id;
-    ety.data.len = e->data.len;
-    ety.data.buf = malloc(e->data.len);
-    memcpy(ety.data.buf, e->data.buf, e->data.len);
+    memcpy(&ety.data, &e->data, sizeof(raft_entry_data_t));
     raft_append_entry(me_, &ety);
     for (i = 0; i < me->num_nodes; i++)
         if (me->nodeid != i)
@@ -463,10 +458,10 @@ int raft_send_requestvote(raft_server_t* me_, int node)
     return 0;
 }
 
-int raft_append_entry(raft_server_t* me_, raft_entry_t* c)
+int raft_append_entry(raft_server_t* me_, raft_entry_t* ety)
 {
     raft_server_private_t* me = (raft_server_private_t*)me_;
-    return log_append_entry(me->log, c);
+    return log_append_entry(me->log, ety);
 }
 
 int raft_apply_entry(raft_server_t* me_)