On Thu, Mar 25, 2010 at 11:26:56AM +0100, Arnd Bergmann wrote:
Yeah, the comment gives this:
/*
* This lock_kernel fixes a subtle race with suid exec
*/
But there is no lock_kernel() in the exec path, may be I missed it...
so this may be an old lock_kernel() that doesn't exist anymore, and
the bkl in the ptrace path is not going to help in any way.
What remain to check are the possible unintended racy places that
this bkl might protect.
I'm going to check a first pass and if it looks fine, I'll just submit
to Oleg and Roland.
--