In response to CVE-2017-7500 and CVE-2017-7501, it was decided that the policy of RPM is "Only follow directory symlinks owned by target directory owner or root." . This check was implemented in a way that is subject to race conditions. If an attacker manages to change things between the call to lstat() that finds a safe symlink and the open() that creates a new file, the policy is not enforced.
Exploits are tricky because of the narrow timing window between the calls, but mazes  could probably be used to delay the stat() long enough for a reliable exploit. Fixing this would require opening the directory with O_PATH|O_NOFOLLOW, followed by fstat() to check ownership and openat() to create the final file.
Created rpm tracking bugs for this issue:
Affects: fedora-all [bug 1977846]
This flaw, along with CVE-2021-35938 and CVE-2021-35939, belong to a set of complex issues that may allow an unprivileged user to trick RPM into modifying root-owned files during installation, due to race conditions and/or symlink attacks. These issues do not have a solution upstream. Fixing would require rather involved refactoring of RPM internals.
Note that in this context, unprivileged users are actually system accounts (like the pcpqa user mentioned in one of the SUSE bugs) that are usually more tightly controlled than ordinary users. In general, access to files and directories installed by RPMs requires high privileges. Regular users should not be allowed to manipulate RPM artifacts during installation. A local attacker would first need to compromise a system account in order to exploit these flaws, thus reducing the overall impact considerably.
This is considered fixed in RPM 4.18 (https://rpm.org/wiki/Releases/4.18.0) which is currently in alpha stage of the release process, final version is expected in Q3.
(In reply to Panu Matilainen from comment #9)
> This is considered fixed in RPM 4.18 (https://rpm.org/wiki/Releases/4.18.0)
> which is currently in alpha stage of the release process, final version is
> expected in Q3.
Is there a patch for this issue anywhere?
(In reply to John Helmert III from comment #10)
> Is there a patch for this issue anywhere?
No. As indicated in the comments, fixing the set of related issues is a complicated refactoring job, some of which can be found here: https://github.com/rpm-software-management/rpm/pull/1919.