It's not a Bug, It's a Weapon | TechSNAP 179

Google leverages Chrome’s marketshare to push web security forward. Are we about to see zero day exploits reclassified as weapons & ZFS gets the green light on Linux for production.

Then it’s a great batch of your questions, our answers & much, much more!

Thanks to:

Direct Download:

RSS Feeds:

Become a supporter on Patreon:

— Show Notes: —

Killing off SHA-1 in SSL certificates

“The SHA-1 cryptographic hash algorithm has been known to be considerably weaker than it was designed to be since at least 2005 — 9 years ago”
“That’s why Chrome will start the process of sunsetting SHA-1 (as used in certificate signatures for HTTPS) with Chrome 39 in November. HTTPS sites whose certificate chains use SHA-1 and are valid past 1 January 2017 will no longer appear to be fully trustworthy in Chrome’s user interface.”
The CA/Browser forum, the group made up of Google, Mozilla, Microsoft, Apple, Opera, and most of the Certificate Authorities, and sets the policies for the group
The forum is how the browsers decide which CAs to include in their trust store
Part of the problem was that older browsers and devices only supported SHA-1, and none of the SHA-2 (SHA256, SHA512) algorithms
The CA/Browser Forum officially deprecated SHA-1 in 2011, no new certificates can be issued that use SHA-1
Google is proposing to add increasingly severe warning messages for visitors to site using SHA-1 certificates that have an expiration date after the end of 2016
Upgrades may still be complicated. Windows Server 2003 and Windows XP SP2 does not support SHA-256, only SHA-1. Servers would need to be upgraded, and Windows XP clients would need to install SP3. Android before 2.3 only supports SHA-1, 2.2 is still quite popular
Support for running 2 certificates, an upgraded one for clients that support it, and a legacy certificates for ones that do not, is being worked on. Apache supports it now, and work is underway to add support to NGINX and Apache Traffic Server.
GlobalSign’s SHA-256 compatibility matrix
It is nice to see the steps being taken with plenty of time for everyone to update gracefully. In the past, the move away from MD5 was much less smooth, only finally spurred on by the real danger of rogue certificates via MD5 collisions
The CA/Browser forum similarly disallowed new 1024 bit certificates in 2010, with no certificate to have an expiration date later than Dec 31st 2013. Mozilla recently pulled the plug on 1024 bit certificates, leaving 107,000 “valid” certificates no longer trusted
SSL Labs breaks down what you need to know
Additional Coverage: Why Google is Hurrying to kill SHA-1

Will selling 0-day exploits soon be considered “Arms Dealing” and be illegal?

VUPEN and others are now following the Wassenaar Arrangement that classifies their 0-days and exploits as regulated and export-controlled “dual-use” technologies. Going forward they will only sell to approved government agencies in approved countries.
The latest version of the agreement included 0-days, exploits, and backdoors as regulated and export-controlled “dual-use” technologies. Previously, the US wasn’t recognizing these most recent additions but that is all changing come later this month according to a recent Federal Register notice (pdf). The notice states that the US will be adopting changes made to the list of dual-use items made in December 2013 as of August 4th.
The big question is where the government will draw the line in terms of defining “dual-use.” Will day-to-day security tools (e.g., Nessus and Nmap) fit into this category? What about a quick bash script you write up to bruteforce web application session ids?

The state of ZFS on Linux

ZFS on Linux is now “officially” production ready
Key ZFS data integrity features work on Linux like they do on other platforms
ZFS runtime stability on Linux is comparable to other filesystems, with certain exceptions
ZoL is at near feature parity with ZFS on other platforms.
ZoL does not lose data
changes to the disk format are forward compatible
Updates are always flawless
Up until now, it was mostly the “on Linux” part that was at question, OpenZFS (the open source fork used in IllumOS, FreeBSD, SmartOS, and elsewhere) has been stable for many years
“Data loss can be defined as the occurrence of either of two events. The first is failing to store some information. The second is attempting to retrieve information that was successfully stored and getting either something else or nothing at all”
“The ZFS on Linux kernel driver performs the same block device operations as its counterparts on other platforms. As a consequence, its ability to ensure data integrity is equivalent to its counterparts on other platforms and this ability far exceeds that of any other Linux filesystem for direct attached storage”
ZoL is missing 9 of the newest features in OpenZFS, including LZ4 compression, Spacemap histographs (speed improvements under heavy fragmentation), Feature Flag enabled TXG (support for rolling back and upgrade), Hole Birth (improved replication performance) and ZFS Bookmarks (resumable zfs send/recv)
Also, there are 9 other features missing from ZoL, including integration for iSCSI (also missing on FreeBSD, as until recently FreeBSD did not have a kernel iSCSI target daemon), Integration with Containers (Linux doesn’t really have a feature similar to Solaris Zones or FreeBSD Jails), Boot Loader integration, etc.
“The current release is 0.6.3 and the next release will be 0.6.4 later this year. The plan is to continue performing 0.6.x releases with distribution maintainers doing backports until the /dev/zfs ioctl interface is stabilized. At that point, the project will release 1.0. New releases will be 1.x while 1.x.y maintenance releases will be done to back port fixes like is done by the Linux kernel stable maintainers”

It’s not a Bug, It’s a Weapon | TechSNAP 179