Catalyst 1.1.8 doesn't set the livecd USE flag when emerging genkernel during a livecd-stage2 build. So you can't use the --lvm2 argument to genkernel, because the genkernel ebuild will not have copied the device-mapper and LVM2 source tarballs into it's private source stash. It seems reasonable that the livecd USE flag be set when building a livecd? This is with genkernel 3.1.5 being used in the chroot.
For the release, we had genkernel in livecd-stage1, so it was merged with USE=livecd as well. If genkernel was detected already in the system, then it was not merged again in livecd-stage2. Anyway, the solution is to put genkernel in livecd-stage1. The USE=livecd stuff was done *after* catalyst 1.1.8 was even released. Anyway, I've changed this behavior in CVS and it'll show up in catalyst 1.1.9, which should be out in a couple weeks.
Thanks for making the change! I was avoiding putting it in livecd-stage1 because I didn't actually need it there or want it on the livecd -- it was just required to build the kernel to boot with. This will get me by for the moment, though. Thanks, again.
Fixed with catalyst 1.1.9