VIA DRM - While there's long been available independently developed code (mostly developed single-handedly by James Simmons) to provide a VIA DRM / KMS graphics driver for Linux, it's still not part of the mainline tree. Back during the summer the VIA driver was proposed for mainlining but all issues haven't been worked out yet and there doesn't appear to be much motivation to actually get it in the tree. It's also getting harder and harder these days to simply find VIA x86 hardware still in production use.
Reiser4 - The ReiserFS successor is still being maintained out-of-tree and there's patches for it up through Linux 3.11 at the moment. However, there isn't much motivation to get it mainlined. Originally mainlining Reiser4 was held up by wanting more technical review and evaluation of the file-system done, but these days it's attributed to having no vendor support. Edward Shishkin doesn't see a mainlined Reiser4 file-system going far without having any vendor backing and his time (as one of the few remaining Reiser4 developers) is spent mostly fixing bugs in the code rather than working on a mainline proposal or adding any new features. You can see my Reiser4 file-system benchmarks from earlier in the year if you're interested in this solution.
ZFS - While there continues to be a lot of community interest in ZFS and now the Open ZFS project, the Sun Microsystems / Oracle file-system doesn't stand a chance yet of being mainlined as the CDDL-licensed file-system code is incompatible with the GPL. Until Oracle is to re-license the code or there's a clean-room implementation done that is GPL compatible, ZFS won't happen in mainline but at least the ZFS On Linux project is pushing along and their out-of-tree DKMS modules tend to work fairly well. I do have ZFS Linux benchmarks from a few months ago.
Tux3 - Tux3 is another interesting file-system that's not in the mainline kernel. Tux3 is still working on new features and the file-system claims to be really fast after the project was restarted months ago. While Tux3 is open-source, it seems there's still a lot of work to be done before the project will commit to pushing it in the mainline kernel and maintaining its code.
Microsoft exFAT - Last but not least for file-systems, the other FS drawing attention in recent months has been Microsoft's exFAT. Samsung open-sourced their exFAT driver after it was accidentally leaked earlier in the year. There's been some work by developers/users to mainline it, but Samsung doesn't appear interested.
BFS - Don't look for the Brain Fuck Scheduler to be appearing in mainline anytime soon, but it's still maintained out-of-tree by Con Kolivas.
Nouveau Re-Clocking - While the Linux 3.13 kernel has Nouveau re-clocking improvements, including a first stab at Fermi/Kepler support, it doesn't yet handle video memory re-clocking well for these new GPUs and isn't yet ready for end-users. This is a big deal as when re-clocking can be figured out correctly, the open-source Nouveau driver should be able to compete much better with the NVIDIA binary driver. Having re-clocking for better performance while also conserving power and heat output will allow a lot more NVIDIA Linux users to try out this driver.
There's some other smaller features too we'd like to see out of the kernel, but what else would you have liked to see in Linux 3.13? Let us know in the forums! Overall though, the Linux 3.13 kernel should be a very great release and kudos to all the developers and stakeholders involved.