Remove obsolete comment.
authorThomas Munro <tmunro@postgresql.org>
Fri, 14 Feb 2025 00:16:05 +0000 (13:16 +1300)
committerThomas Munro <tmunro@postgresql.org>
Fri, 14 Feb 2025 00:16:05 +0000 (13:16 +1300)
Commit 755a4c10d19d prevented StartReadBuffers() from crossing md.c
segment boundaries in one operation, but a comment about that
possibility remained.

src/backend/storage/buffer/bufmgr.c

index b5938f1b473d935e043aec002f5987e8322733a0..80b0d0c5ded61a76fd59ac7ea5e2975ef4df5102 100644 (file)
@@ -1330,10 +1330,7 @@ StartReadBuffersImpl(ReadBuffersOperation *operation,
         * StartReadBuffers() were made for the same blocks before
         * WaitReadBuffers(), only the first would issue the advice. That'd be
         * a better simulation of true asynchronous I/O, which would only
-        * start the I/O once, but isn't done here for simplicity.  Note also
-        * that the following call might actually issue two advice calls if we
-        * cross a segment boundary; in a true asynchronous version we might
-        * choose to process only one real I/O at a time in that case.
+        * start the I/O once, but isn't done here for simplicity.
         */
        smgrprefetch(operation->smgr,
                     operation->forknum,