On my Acer Aspire One netbook (where an 8GB SSD replaces a conventional HDD) I dual-boot Ubuntu and TC using grub to choose. All TC materials are on the SSD. In this setup the boot option waitusb is essential for 2.6rc3 to notice the extensions in /dev/sda1/tce in time to load them. (When they don't get noticed, the boot is extremely quick!) This issue is new with 2.6rc3; everything up to 2.6rc2 needed no waitusb. I know that waitusb=3 works and that no waitusb parameter doesn't; I haven't experimented on how low I can go. [This extra wrinkle may confuse newcomers to TC, but IMHO it's a good sign that the init scripts are getting ever faster and more efficient. So please receive these remarks as a report and not at all a complaint.]
A confusing behaviour: even with waitusb=3 or 5, 2.6rc3 seems not to find the file mydata.tgz on the root of device /dev/sda1. The result is that /opt does not get populated with my preferred scripts, and /opt/bootlocal.sh does not get recalled from the backup, so a bunch of user-specified initialization is not happening. I know that the file mydata.tgz is still intact because booting TC2.5 still loads it OK.
It may be useful to mention that the kernel in 2.6rc3 is the same one used in all versions 2.0-2.5. Users (like me) who kept the old one around can just go back to using it ... the experimental kernel in earlier RC's for 2.6 was functionally identical, only compressed for size. [I think this means that users like me for whom the upx-compressed kernel actually worked could keep using it instead of reverting, but that would slightly violate the spirit of helping test the full RC as proposed.]
With thanks, as always.