On 03/29, Gregory Price wrote:

Last note on this before I push up another patch set.

The change from __get_user to get_user also introduces a call to
might_fault() which adds a larger callstack for every syscall /
dispatch.  This turns into a might_sleep and might_reschedule, which
represent a very different pattern of execution from before.

might_fault() is nop unless CONFIG_PROVE_LOCKING || DEBUG_ATOMIC_SLEEP.

Again, I won't really argue with task_access_ok(). Just I am not sure
2/4 gives enough justification for this new helper with unclear semantics
(until we ensure that access_ok() doesn't depend on current).

Oleg.