-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Dear Qubes Community,
We have just published Qubes Security Bulletin (QSB) #056: Insufficient anti-spoofing firewall rules. The text of this QSB is reproduced below. This QSB and its accompanying signatures will always be available in the Qubes Security Pack (qubes-secpack). View QSB #056 in the qubes-secpack: https://github.com/QubesOS/qubes-secpack/blob/master/QSBs/qsb-056-2019.txt Learn about the qubes-secpack, including how to obtain, verify, and read it: https://www.qubes-os.org/security/pack/ View all past QSBs: https://www.qubes-os.org/security/bulletins/ View the Xen Security Advisory (XSA) Tracker: https://www.qubes-os.org/security/xsa/ ``` ---===[ Qubes Security Bulletin #56]===--- 2019-12-25 Insufficient anti-spoofing firewall rules Summary ======= The firewall configuration in Qubes OS prevents IP address spoofing in downstream interfaces (e.g., network-providing qubes, network-consuming qubes, and `vif*` interfaces). However, it does not prevent IP spoofing in upstream interfaces (normally `eth0`, but in the case of VPNs or other configuration, there may also be others). Impact ====== Configurations with inter-VM networking allowed [1] or additional interfaces created (e.g., VPNs) are vulnerable to IP spoofing. Combined with other vulnerabilities, such as the procedure described in the CVE-2019-14899 report [2], this could allow an upstream qube (e.g., sys-net) to inject data into an established connection. Discussion ========== The anti-spoofing firewall rules in a network-providing qube look like this: *raw ... -A PREROUTING ! -s 10.137.0.5/32 -i vif12.0 -j DROP -A PREROUTING ! -s 10.137.0.6/32 -i vif17.0 -j DROP -A PREROUTING ! -s 10.137.0.7/32 -i vif18.0 -j DROP -A PREROUTING ! -s 10.137.0.8/32 -i vif21.0 -j DROP COMMIT Each `vif*` interface drops packets if its source IP does not match the one assigned to the qube behind that interface. However, it does not ensure that the source IP does not appear on any other (non-`vif`) interface. The other property could, in theory, be achieved by this FORWARD chain: *filter ... -A FORWARD -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT -A FORWARD -j QBS-FORWARD -A FORWARD -i vif+ -o vif+ -j DROP -A FORWARD -i vif+ -j ACCEPT -A FORWARD -j DROP COMMIT These rules should reject packets not belonging to established connections on non-vif interfaces. Moreover, without seeing other packets in the connection, it should be prohibitively difficult to forge packets that would be considered to be part of an established connection. However, methods like the one described in the CVE-2019-14899 report [2] allow one to guess the required parameters. Note that if a connection normally goes through a given qube (without any further protection like TLS), that qube can always manipulate the traffic without guessing anything. The default Qubes configuration is secure, since network traffic either goes directly to the upstream qube (which, by definition, has access to that traffic), or it is an inter-VM connection attempt, which is prevented by the third rule (meaning that there are no connections in the conntrack table that the upstream qube could try to hijack). However, once the user departs from the default configuration, e.g., by introducing inter-VM communications [1] (allowing traffic between some `vif*` interfaces), or VPN-like interfaces, the default rules are no longer sufficient, since an upstream qube can inject packets (by spoofing the source IP) into connections that normally do not pass through it in the clear. Our solution to this problem is twofold: 1. For Qubes OS 4.0, whenever a running qube is connected to a network-providing qube, an additional firewall rule is added that blocks the running qube's IP as a source on other network interfaces. 2. For Qubes OS 4.1 and later, we will modify the firewall mechanism so that it maintains aa list of connected qubes and their addresses, even when they are not running. All such addresses will be rejected on upstream network interfaces. The main difference between these two solutions is that fix for Qubes OS 4.0 does not protect against spoofing the addresses of qubes that are not running. However, since 4.0 is a stable release, we must consider the impact of such a solution on the stability of this release. This fix is a much simpler change that carries a considerably lower risk of introducing a regression. Patching ======== The specific packages that resolve the problems discussed in this bulletin are as follows: For Qubes OS 4.0: - qubes-core-agent version 4.0.51 The packages for domUs are to be installed in TemplateVMs and StandaloneVMs via the Qube Manager or via their respective package managers: For updates to Fedora from the stable repository (not immediately available): $ sudo dnf update For updates to Fedora from the security-testing repository: $ sudo dnf update --enablerepo=qubes-vm-*-security-testing For updates to Debian from the stable repository (not immediately available): $ sudo apt update && sudo apt dist-upgrade For updates to Debian from the security-testing repository: First, uncomment the line below "Qubes security updates testing repository" in: /etc/apt/sources.list.d/qubes-r*.list Then: $ sudo apt update && sudo apt dist-upgrade A restart is required for these changes to take effect. This entails shutting down the TemplateVM before restarting all the TemplateBasedVMs based on that TemplateVM. These packages will migrate from the security-testing repositories to their respective current (stable) repositories over the next two weeks after being tested by the community. Credits ======== The issue was reported by Demi Marie Obenour. References ========== [1] https://www.qubes-os.org/doc/firewall/#enabling-networking-between-two-qubes [2] https://nvd.nist.gov/vuln/detail/CVE-2019-14899 - -- The Qubes Security Team https://www.qubes-os.org/security/ ``` - -- Best Regards, Marek Marczykowski-Górecki Invisible Things Lab A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? -----BEGIN PGP SIGNATURE----- iQEzBAEBCAAdFiEEhrpukzGPukRmQqkK24/THMrX1ywFAl4DmQkACgkQ24/THMrX 1yyTrgf8DB9+TOy89Gn9kwDYm15nXqCuxOm0k3Zsv3FWJCz1NobTpDJ14+LI0qcf YR1jXT+XqUvfeIJ2NlJ+DQ4454cd3m27nEP7B0G7A2PU3jbonIyBE9Qe7PpQ8kmU FtI+GoknrMlUEP+QNwceRg1Q9OPaB6Zzzq0VE6C58rxnL6oGNUXVgrsXV/Jtl1pZ quf5c8x7cZqRqUbFBkaE2P5deYRfCIj/Vt3N3uhsvEKAay2qwgMnnZmQv2Qhp+cq gephG2LgrczBjvjlZ/0zt2+7N4LPyDCeP5dVJlFSz/85uNBo0vmTecyFhaUJEhn1 2JzJh9rUVQFNTwetTTsh2M2q6rubnQ== =29/a -----END PGP SIGNATURE----- -- You received this message because you are subscribed to the Google Groups "qubes-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to qubes-users+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/qubes-users/20191225171448.GA11736%40mail-itl.