netdev - [Regression] 6.0.16-6.0.18 kernel no longer return EADDRINUSE from bind

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAFsF8vL4CGFzWMb38_XviiEgxoKX0GYup=JiUFXUOmagdk9CRg@mail.gmail.com>
Date:   Wed, 11 Jan 2023 16:52:21 +0100
From:   Paul Holzinger <pholzing@...hat.com>
To:     stable@...r.kernel.org
Cc:     regressions@...ts.linux.dev, netdev@...r.kernel.org
Subject: [Regression] 6.0.16-6.0.18 kernel no longer return EADDRINUSE from bind

Hi all,

Since updating to 6.0.16 the bind() system call no longer fails with
EADDRINUSE when the address is already in use.
Instead bind() returns 1 in such a case, which is not a valid return
value for this system call.

It works with the 6.0.15 kernel and earlier, 6.1.4 and 6.2-rc3 also
seem to work.

Fedora bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=2159066

To reproduce you can just run `ncat -l 5000` two times, the second one
should fail. However it just uses a random port instead.

As far as I can tell this problem is caused by
https://lore.kernel.org/stable/20221228144337.512799851@linuxfoundation.org/
which did not backport commit 7a7160edf1bf properly.
The line `int ret = -EADDRINUSE, port = snum, l3mdev;` is missing in
net/ipv4/inet_connection_sock.c.
This is the working 6.1 patch:
https://lore.kernel.org/all/20221228144339.969733443@linuxfoundation.org/

Best regards,
Paul