Message-ID: <d2c31606e6bb9b83a02ed4835d65191b38d4ba12.camel@j-davis.com>
Subject: Is this a problem in GenericXLogFinish()?
From: Jeff Davis <pgsql@j-davis.com>
To: pgsql-hackers@postgresql.org
Date: Fri, 22 Sep 2023 13:52:48 -0700
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.44.4-0ubuntu2 
MIME-Version: 1.0
Archived-At: 
 <https://www.postgresql.org/message-id/d2c31606e6bb9b83a02ed4835d65191b38d4ba12.camel%40j-davis.com>
Precedence: bulk


src/backend/transam/README says:

  ...
  4. Mark the shared buffer(s) as dirty with MarkBufferDirty().  (This=C2=
=A0
  must happen before the WAL record is inserted; see notes in=C2=A0
  SyncOneBuffer().)
  ...

But GenericXLogFinish() does this:

  ...
  /* Insert xlog record */
  lsn =3D XLogInsert(RM_GENERIC_ID, 0);

  /* Set LSN and mark buffers dirty */
  for (i =3D 0; i < MAX_GENERIC_XLOG_PAGES; i++)
  {
      PageData   *pageData =3D &state->pages[i];

      if (BufferIsInvalid(pageData->buffer))
          continue;
      PageSetLSN(BufferGetPage(pageData->buffer), lsn);
      MarkBufferDirty(pageData->buffer);
  }
  END_CRIT_SECTION();

Am I missing something or is that a problem?

Regards,
	Jeff Davis