sched/completions: Fix complete_all() semantics
authorPeter Zijlstra <peterz@infradead.org>
Wed, 21 Dec 2016 08:05:02 +0000 (09:05 +0100)
committerIngo Molnar <mingo@kernel.org>
Sat, 14 Jan 2017 10:30:01 +0000 (11:30 +0100)
commitda9647e076d440190fff4b4ee0f4b829dd6e8c4f
tree7bedc99c5a41500084b0ef26673cb64ab751de0e
parent59f8c2989283bbd3df9fcfb22494d84f4852e536
sched/completions: Fix complete_all() semantics

Documentation/scheduler/completion.txt says this about complete_all():

  "calls complete_all() to signal all current and future waiters."

Which doesn't strictly match the current semantics. Currently
complete_all() is equivalent to UINT_MAX/2 complete() invocations,
which is distinctly less than 'all current and future waiters'
(enumerable vs innumerable), although it has worked in practice.

However, Dmitry had a weird case where it might matter, so change
completions to use saturation semantics for complete()/complete_all().
Once done hits UINT_MAX (and complete_all() sets it there) it will
never again be decremented.

Requested-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: der.herr@hofr.at
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/sched/completion.c