From: Tom Lane <tgl@sss.pgh.pa.us>
To: pgsql-hackers@lists.postgresql.org
Subject: Dubious code in pg_rewind's process_target_file()
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1221795.1599329320.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Sat, 05 Sep 2020 14:08:40 -0400
Message-ID: <1221796.1599329320@sss.pgh.pa.us>
Precedence: bulk

scan-build complains that "exists =3D false" is a dead store,
which it is:

process_target_file(const char *path, file_type_t type, size_t oldsize,
                    const char *link_target)
{
    bool        exists;
    ...

    if (lstat(localpath, &statbuf) < 0)
    {
        if (errno !=3D ENOENT)
            pg_fatal("could not stat file \"%s\": %s\n",
                     localpath, strerror(errno));

        exists =3D false;
    }

    ...
    exists =3D (bsearch(&key_ptr, map->array, map->narray, sizeof(file_ent=
ry_t *),
                      path_cmp) !=3D NULL);

It looks to me like we could replace "exists =3D false" with "return",
rather than uselessly constructing a FILE_ACTION_REMOVE entry for
a file we've already proven is not there.  This seems to have been
copy-and-pasted from process_source_file, without much thought for
the fact that the situations are quite different.

			regards, tom lane