[PATCH 1/2] sched: Don't defer CPU pick to migration_cpu_stop()

[Valentin Schneider]

Will reported that the 'XXX __migrate_task() can fail' in migration_cpu_stop()
can happen, and it *is* sort of a big deal. Looking at it some more, one
will note there is a glaring hole in the deferred CPU selection:

  (w/ CONFIG_CPUSET=n, so that the affinity mask passed via taskset doesn't
  get AND'd with cpu_online_mask)

  $ taskset -pc 0-2 $PID
  # offline CPUs 3-4
  $ taskset -pc 3-5 $PID
    `\
      $PID may stay on 0-2 due to the cpumask_any_distribute() picking an
      offline CPU and __migrate_task() refusing to do anything due to
      cpu_is_allowed().

set_cpus_allowed_ptr() goes to some length to pick a dest_cpu that matches
the right constraints vs affinity and the online/active state of the
CPUs. Reuse that instead of discarding it in the affine_move_task() case.

Fixes: 6d337eab041d ("sched: Fix migrate_disable() vs set_cpus_allowed_ptr()")
Reported-by: Will Deacon <will@xxxxxxxxxx>
Signed-off-by: Valentin Schneider <valentin.schneider@xxxxxxx>
---
 kernel/sched/core.c | 20 ++++++++++++--------
 1 file changed, 12 insertions(+), 8 deletions(-)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 3d2527239c3e..d40511c55e80 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -2273,7 +2273,6 @@ static int migration_cpu_stop(void *data)
 	struct migration_arg *arg = data;
 	struct set_affinity_pending *pending = arg->pending;
 	struct task_struct *p = arg->task;
-	int dest_cpu = arg->dest_cpu;
 	struct rq *rq = this_rq();
 	bool complete = false;
 	struct rq_flags rf;
@@ -2311,19 +2310,15 @@ static int migration_cpu_stop(void *data)
 		if (pending) {
 			p->migration_pending = NULL;
 			complete = true;
-		}
 
-		if (dest_cpu < 0) {
 			if (cpumask_test_cpu(task_cpu(p), &p->cpus_mask))
 				goto out;
-
-			dest_cpu = cpumask_any_distribute(&p->cpus_mask);
 		}
 
 		if (task_on_rq_queued(p))
-			rq = __migrate_task(rq, &rf, p, dest_cpu);
+			rq = __migrate_task(rq, &rf, p, arg->dest_cpu);
 		else
-			p->wake_cpu = dest_cpu;
+			p->wake_cpu = arg->dest_cpu;
 
 		/*
 		 * XXX __migrate_task() can fail, at which point we might end
@@ -2606,7 +2601,7 @@ static int affine_move_task(struct rq *rq, struct task_struct *p, struct rq_flag
 			init_completion(&my_pending.done);
 			my_pending.arg = (struct migration_arg) {
 				.task = p,
-				.dest_cpu = -1,		/* any */
+				.dest_cpu = dest_cpu,
 				.pending = &my_pending,
 			};
 
@@ -2614,6 +2609,15 @@ static int affine_move_task(struct rq *rq, struct task_struct *p, struct rq_flag
 		} else {
 			pending = p->migration_pending;
 			refcount_inc(&pending->refs);
+			/*
+			 * Affinity has changed, but we've already installed a
+			 * pending. migration_cpu_stop() *must* see this, else
+			 * we risk a completion of the pending despite having a
+			 * task on a disallowed CPU.
+			 *
+			 * Serialized by p->pi_lock, so this is safe.
+			 */
+			pending->arg.dest_cpu = dest_cpu;
 		}
 	}
 	pending = p->migration_pending;
-- 
2.25.1