Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Reword some unclear comments
authorAlvaro Herrera <alvherre@alvh.no-ip.org>
Tue, 8 Aug 2017 22:46:16 +0000 (18:46 -0400)
committerAlvaro Herrera <alvherre@alvh.no-ip.org>
Tue, 8 Aug 2017 22:48:24 +0000 (18:48 -0400)
src/backend/access/heap/heapam.c

index 26d96fd7fc0ce0b786d9e0367aa03b563306d2f6..af101689282c4905ef08fd20148c6e3362efafd3 100644 (file)
@@ -3594,7 +3594,7 @@ l2:
 
        /*
         * To prevent concurrent sessions from updating the tuple, we have to
-        * temporarily mark it locked, while we release the lock.
+        * temporarily mark it locked, while we release the page-level lock.
         *
         * To satisfy the rule that any xid potentially appearing in a buffer
         * written out to disk, we unfortunately have to WAL log this
@@ -3606,8 +3606,9 @@ l2:
 
        /*
         * Compute xmax / infomask appropriate for locking the tuple. This has
-        * to be done separately from the lock, because the potentially
-        * created multixact would otherwise be wrong.
+        * to be done separately from the combo that's going to be used for
+        * updating, because the potentially created multixact would otherwise
+        * be wrong.
         */
        compute_new_xmax_infomask(HeapTupleHeaderGetRawXmax(oldtup.t_data),
                                  oldtup.t_data->t_infomask,