Can you try to sprinkle printk's around to see where it actually ends up
stopping?
Since all special commands should now go through the taskfile interface
(Bartlomiej - correct?), I guess just a printk() at the top of
do_rw_taskfile() should at least get the ball rolling. Even something that
just prints out task->data_phase and tf->command would start to unravel at
least what the last command is before it hangs..
Linus
--