[PATCH 10/10] drm/i915: terrible hack for ci

[Daniel Vetter]

CI results for my series shows regressions in

igt at i915_selftest@live at execlists on some (but not all) icl and tgl
machines.

This is the best I came up with as an explanation. It will be
terrifying if it holds out to be true.

Signed-off-by: Daniel Vetter <daniel.vetter at intel.com>
---
 drivers/gpu/drm/i915/gt/intel_context.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/gt/intel_context.c b/drivers/gpu/drm/i915/gt/intel_context.c
index 9e33efb594dd..076a83d954ac 100644
--- a/drivers/gpu/drm/i915/gt/intel_context.c
+++ b/drivers/gpu/drm/i915/gt/intel_context.c
@@ -25,6 +25,8 @@ static void rcu_context_free(struct rcu_head *rcu)
 {
 	struct intel_context *ce = container_of(rcu, typeof(*ce), rcu);
 
+	i915_vm_put(ce->vm);
+
 	trace_intel_context_free(ce);
 	kmem_cache_free(slab_ce, ce);
 }
@@ -434,7 +436,6 @@ void intel_context_fini(struct intel_context *ce)
 {
 	if (ce->timeline)
 		intel_timeline_put(ce->timeline);
-	i915_vm_put(ce->vm);
 
 	mutex_destroy(&ce->pin_mutex);
 	i915_active_fini(&ce->active);
-- 
2.32.0