Fix comment that Vadim found confusing.
authorTom Lane
Thu, 5 Apr 2001 16:55:21 +0000 (16:55 +0000)
committerTom Lane
Thu, 5 Apr 2001 16:55:21 +0000 (16:55 +0000)
src/backend/access/transam/xlog.c

index 282841c57202cf783ee73e116721cb027e431680..f6394b115d70649e53e7e40e82ecc11acd788c60 100644 (file)
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Header: /cvsroot/pgsql/src/backend/access/transam/xlog.c,v 1.64 2001/04/05 09:34:32 vadim Exp $
+ * $Header: /cvsroot/pgsql/src/backend/access/transam/xlog.c,v 1.65 2001/04/05 16:55:21 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -2467,6 +2467,7 @@ StartupXLOG(void)
            Insert->currpage->xlp_sui = ThisStartUpID;
        else
            Insert->currpage->xlp_sui = ThisStartUpID + 1;
+       /* rest of buffer was zeroed in XLOGShmemInit */
    }
    else
    {
@@ -2475,10 +2476,8 @@ StartupXLOG(void)
            ((EndOfLog.xrecoff - 1) / BLCKSZ + 1) * BLCKSZ;
        /*
         * Tricky point here: readBuf contains the *last* block that the
-        * LastRec record spans, not the one it starts in, which is what we
-        * want.
-        *
-        * XXX - why would we want block LastRec starts in?
+        * LastRec record spans, not the one it starts in.  The last block
+        * is indeed the one we want to use.
         */
        Assert(readOff == (XLogCtl->xlblocks[0].xrecoff - BLCKSZ) % XLogSegSize);
        memcpy((char *) Insert->currpage, readBuf, BLCKSZ);