With lockdep-crossrelease we get deadlock reports that span cpu-up and
cpu-down chains. Such deadlocks cannot possibly happen because cpu-up
and cpu-down are globally serialized.

  CPU0                  CPU1                    CPU2
  cpuhp_up_callbacks:   takedown_cpu:           cpuhp_thread_fun:

  cpuhp_state
                        irq_lock_sparse()
    irq_lock_sparse()
                        wait_for_completion()
                                                cpuhp_state
                                                complete()

Now that we have consistent AP state, we can trivially separate the
AP-work class between up and down using st->bringup.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
---
 kernel/cpu.c |   41 ++++++++++++++++++++++++++++++++---------
 1 file changed, 32 insertions(+), 9 deletions(-)

--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -68,9 +68,26 @@ struct cpuhp_cpu_state {
 static DEFINE_PER_CPU(struct cpuhp_cpu_state, cpuhp_state);
 
 #if defined(CONFIG_LOCKDEP) && defined(CONFIG_SMP)
-static struct lock_class_key cpuhp_state_key;
-static struct lockdep_map cpuhp_state_lock_map =
-	STATIC_LOCKDEP_MAP_INIT("cpuhp_state", &cpuhp_state_key);
+static struct lockdep_map cpuhp_state_up_map =
+	STATIC_LOCKDEP_MAP_INIT("cpuhp_state-up", &cpuhp_state_up_map);
+static struct lockdep_map cpuhp_state_down_map =
+	STATIC_LOCKDEP_MAP_INIT("cpuhp_state-down", &cpuhp_state_down_map);
+
+
+static void inline cpuhp_lock_acquire(bool bringup)
+{
+	lock_map_acquire(bringup ? &cpuhp_state_up_map : &cpuhp_state_down_map);
+}
+
+static void inline cpuhp_lock_release(bool bringup)
+{
+	lock_map_release(bringup ? &cpuhp_state_up_map : &cpuhp_state_down_map);
+}
+#else
+
+static void inline cpuhp_lock_acquire(bool bringup) { }
+static void inline cpuhp_lock_release(bool bringup) { }
+
 #endif
 
 /**
@@ -512,7 +529,7 @@ static void cpuhp_thread_fun(unsigned in
 	if (WARN_ON_ONCE(!st->should_run))
 		return;
 
-	lock_map_acquire(&cpuhp_state_lock_map);
+	cpuhp_lock_acquire(bringup);
 
 	if (st->single) {
 		state = st->cb_state;
@@ -564,7 +581,7 @@ static void cpuhp_thread_fun(unsigned in
 	}
 
 next:
-	lock_map_release(&cpuhp_state_lock_map);
+	cpuhp_lock_release(bringup);
 
 	if (!st->should_run)
 		complete(&st->done);
@@ -581,8 +598,11 @@ cpuhp_invoke_ap_callback(int cpu, enum c
 	if (!cpu_online(cpu))
 		return 0;
 
-	lock_map_acquire(&cpuhp_state_lock_map);
-	lock_map_release(&cpuhp_state_lock_map);
+	cpuhp_lock_acquire(false);
+	cpuhp_lock_release(false);
+
+	cpuhp_lock_acquire(true);
+	cpuhp_lock_release(true);
 
 	/*
 	 * If we are up and running, use the hotplug thread. For early calls
@@ -620,8 +640,11 @@ static int cpuhp_kick_ap_work(unsigned i
 	enum cpuhp_state prev_state = st->state;
 	int ret;
 
-	lock_map_acquire(&cpuhp_state_lock_map);
-	lock_map_release(&cpuhp_state_lock_map);
+	cpuhp_lock_acquire(false);
+	cpuhp_lock_release(false);
+
+	cpuhp_lock_acquire(true);
+	cpuhp_lock_release(true);
 
 	trace_cpuhp_enter(cpu, st->target, prev_state, cpuhp_kick_ap_work);
 	ret = cpuhp_kick_ap(st, st->target);