After a cold boot I reproducibly got into an infinite loop of the gray start screen. The boot-up sound would be played every 3 minutes continuously.
I pressed Apple-V to see what is going on and I got several messages indicating that the journalling on my root disk was corrupted:
jnl: replay_journal: bad block list header @ ...
jnl: journal_open: Error replaying the journal!
hfs: early jnl init: failed to open/create the journal (retval 0)
Booting in single user mode (Apple+S), doing an fsck -fy and an exit solved the problem temporarily. It would boot up without problems. But after a day and another cold reboot the problem was there again. My MacBook Pro would not boot and was stuck in the endless gray boot screen loop.
I think I now solved the problem by making it boot once (see above), then disabling the journal by doing
sudo diskutil disableJournal /dev/disk0s2
(This cannot be done in single user mode, you need to boot Mac OS X for this.)
Then I rebooted. No problem. Then I activated journalling using the disk utility and it works again. Rebooting cold now works fine. Let's see what the future brings.
But: Apple: Please. How hard can it be. Please make your stupid fsck program deal with such a situation instead of continuously rebooting. This is a showstopper bug for a lot of users. Filesystem checks must never assume the filesystem including any journal is OK!!! This is sounds rather obvious to me.