Thanks, AS.
Actually, My -kjp1 kernel is the only one of four attempts to compile that even makes it to a desktop. The other three, with various configurations based on my custom 2.6.36 to the standard pclos config, all hang just after starting xinetd and cannot be switched to a terminal or start X. Even an R-E-I-S-U-B reboot has to be tried many times before it finally works.
So I think the problem is really a hang during start-up. Both my main computers have Intel graphics and both fail to finish booting even though they work well in graphical mode, but if I try to switch to a virtual terminal I find they won't start and are occupied with boot messages.
The 2.6.38 kernel appears to have a regression not in the 2.6.36 to an old KMS fb bug with Intel chips, which may be why Texstar disabled the KMS. You can see the error in my fuzzy photo, which although taken during shutdown is actually still showing boot messages. Unfortunately, my laptop needs KMS.
There is an old patch and I will take a look at the kernel code to see whether it really is a regression.