From bdeee61530b25a5196b2ac16933b18067d84d502 Mon Sep 17 00:00:00 2001 From: Heikki Linnakangas Date: Mon, 22 Nov 2021 12:13:46 +0200 Subject: [PATCH v7 2/5] Use correct WAL position in error message on invalid XLOG page header. We're using EndRecPtr as an argument to emode_for_corrupt_record(), which is all about suppressing duplicate complaints about the same LSN. But if the xlogreader has been repositioned using XLogBeginRead() since the last call to ReadRecord(), or if there are no preceding calls to ReadRecord(), then the value of EndRecPtr here is left over from the previous read position and is not particularly related to the record we're reading now. xlogreader->EndRecPtr, OTOH, is. This doesn't seem worth backpatching, but I'm doing this now as a separate commit because it is needed by the next big refactoring commit. Author: Robert Haas Discussion: https://www.postgresql.org/message-id/CA+Tgmoao96EuNeSPd+hspRKcsCddu=b1h-QNRuKfY8VmfNQdfg@mail.gmail.com --- src/backend/access/transam/xlog.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index c42accfa53f..cfccf950dd5 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -12582,7 +12582,7 @@ retry: * errmsg_internal() because the message was already translated. */ if (xlogreader->errormsg_buf[0]) - ereport(emode_for_corrupt_record(emode, EndRecPtr), + ereport(emode_for_corrupt_record(emode, xlogreader->EndRecPtr), (errmsg_internal("%s", xlogreader->errormsg_buf))); /* reset any error XLogReaderValidatePageHeader() might have set */ -- 2.30.2