MP regression ? (sequel to: CPU1 failed to become ready)

[Vincent <10.50%free.fr@localhost>]

Well, I am still stuck with my Core 2 Duo that will not start CPU 1.

However, up to .62, I could workaround by setting MPDEBUG to yes. After 
the message "cpu1 failed to become ready", the kernel fell back in DDB, 
and typing cont woke up CPU1, and everything was nominal onwards.

But with a kernel built this morning, this hack ceased to work. After a 
few seconds, I get a "CPU1 fails to start", drop again in DDB and cannot 
do anything else because the keyboard is dead.

Hints?

Thanks,
Vincent