BugLink: https://bugs.launchpad.net/bugs/1821259

Matt reported the following deadlock:

CPU0					CPU1

schedule(.prev=migrate/0)		<fault>
  pick_next_task()			  ...
    idle_balance()			    migrate_swap()
      active_balance()			      stop_two_cpus()
						spin_lock(stopper0->lock)
						spin_lock(stopper1->lock)
						ttwu(migrate/0)
						  smp_cond_load_acquire() -- waits for schedule()
        stop_one_cpu(1)
	  spin_lock(stopper1->lock) -- waits for stopper lock

Fix this deadlock by taking the wakeups out from under stopper->lock.
This allows the active_balance() to queue the stop work and finish the
context switch, which in turn allows the wakeup from migrate_swap() to
observe the context and complete the wakeup.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reported-by: Matt Fleming <matt@codeblueprint.co.uk>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Galbraith <umgwanakikbuti@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20180420095005.GH4064@hirez.programming.kicks-ass.netSigned-off-by: Ingo Molnar <mingo@kernel.org>
(backported from commit 0b26351b)
[mfo: backport:
 - hunk 1:
   - refresh context lines
   - include 'linux/sched.h' instead of 'linux/sched/wake_q.h' which is
     code moved by upstream commit eb61baf6 ("sched/headers: Move the
     wake-queue types and interfaces from sched.h into <linux/sched/wake_q.h>")
 - hunk 2:
   - refresh context lines
   - s/bool/void/ return type of cpu_stop_queue_work() and other 2 changes
     which are not relevant to this fix, due to lack of upstream commits:
     - 'enabled' variable:
       commit 1b034bd9 ("stop_machine: Make cpu_stop_queue_work() and
       stop_one_cpu_nowait() return bool")
     - else 'if (work->done)' condition
       commit dd2e3121 ("stop_machine: Shift the 'done != NULL' check
       from cpu_stop_signal_done() to callers")
   - s/DEFINE_WAKE_Q/WAKE_Q/ due to lack of upstream commit 194a6b5b
     ("sched/wake_q: Rename WAKE_Q to DEFINE_WAKE_Q")
 - hunk 3:
   - refresh context lines
   - s/DEFINE_WAKE_Q/WAKE_Q/ due to lack of upstream commit 194a6b5b
     ("sched/wake_q: Rename WAKE_Q to DEFINE_WAKE_Q")
 - hunk 4:
   - refresh context lines
 - hunk 5:
   - merge with hunk 4
   - refresh context lines]
Signed-off-by: Mauricio Faria de Oliveira <mfo@canonical.com>
Acked-by: Khalid Elmously <khalid.elmously@canonical.com>
Acked-by: Marcelo Henrique Cerri <marcelo.cerri@canonical.com>
Signed-off-by: Khalid Elmously <khalid.elmously@canonical.com>