diff --git a/drivers/gpu/drm/i915/i915_request.c b/drivers/gpu/drm/i915/i915_request.c
index e9c2094ab8ea..82094b9f5ba7 100644
--- a/drivers/gpu/drm/i915/i915_request.c
+++ b/drivers/gpu/drm/i915/i915_request.c
@@ -783,6 +783,12 @@ emit_semaphore_wait(struct i915_request *to,
 	GEM_BUG_ON(!from->timeline->has_initial_breadcrumb);
 	GEM_BUG_ON(INTEL_GEN(to->i915) < 8);
 
+	/* Just emit the first semaphore we see as request space is limited. */
+	if (to->sched.semaphores & from->engine->mask)
+		return i915_sw_fence_await_dma_fence(&to->submit,
+						     &from->fence, 0,
+						     I915_FENCE_GFP);
+
 	/* We need to pin the signaler's HWSP until we are finished reading. */
 	err = i915_timeline_read_hwsp(from, to, &hwsp_offset);
 	if (err)
@@ -814,7 +820,8 @@ emit_semaphore_wait(struct i915_request *to,
 	*cs++ = 0;
 
 	intel_ring_advance(to, cs);
-	to->sched.flags |= I915_SCHED_HAS_SEMAPHORE;
+	to->sched.semaphores |= from->engine->mask;
+	to->sched.flags |= I915_SCHED_HAS_SEMAPHORE_CHAIN;
 	return 0;
 }
 
@@ -1126,7 +1133,7 @@ void i915_request_add(struct i915_request *request)
 		 * far in the distance past over useful work, we keep a history
 		 * of any semaphore use along our dependency chain.
 		 */
-		if (!(request->sched.flags & I915_SCHED_HAS_SEMAPHORE))
+		if (!(request->sched.flags & I915_SCHED_HAS_SEMAPHORE_CHAIN))
 			attr.priority |= I915_PRIORITY_NOSEMAPHORE;
 
 		/*
diff --git a/drivers/gpu/drm/i915/i915_scheduler.c b/drivers/gpu/drm/i915/i915_scheduler.c
index e0f609d01564..77dbf7d74e12 100644
--- a/drivers/gpu/drm/i915/i915_scheduler.c
+++ b/drivers/gpu/drm/i915/i915_scheduler.c
@@ -41,6 +41,7 @@ void i915_sched_node_init(struct i915_sched_node *node)
 	INIT_LIST_HEAD(&node->waiters_list);
 	INIT_LIST_HEAD(&node->link);
 	node->attr.priority = I915_PRIORITY_INVALID;
+	node->semaphores = 0;
 	node->flags = 0;
 }
 
@@ -73,9 +74,9 @@ bool __i915_sched_node_add_dependency(struct i915_sched_node *node,
 		dep->flags = flags;
 
 		/* Keep track of whether anyone on this chain has a semaphore */
-		if (signal->flags & I915_SCHED_HAS_SEMAPHORE &&
+		if (signal->flags & I915_SCHED_HAS_SEMAPHORE_CHAIN &&
 		    !node_started(signal))
-			node->flags |= I915_SCHED_HAS_SEMAPHORE;
+			node->flags |= I915_SCHED_HAS_SEMAPHORE_CHAIN;
 
 		ret = true;
 	}
diff --git a/drivers/gpu/drm/i915/i915_scheduler_types.h b/drivers/gpu/drm/i915/i915_scheduler_types.h
index 76a004a17b2e..f1af3916a808 100644
--- a/drivers/gpu/drm/i915/i915_scheduler_types.h
+++ b/drivers/gpu/drm/i915/i915_scheduler_types.h
@@ -10,6 +10,7 @@
 #include <linux/list.h>
 
 #include "i915_priolist_types.h"
+#include "intel_engine_types.h"
 
 struct drm_i915_private;
 struct i915_request;
@@ -55,7 +56,8 @@ struct i915_sched_node {
 	struct list_head link;
 	struct i915_sched_attr attr;
 	unsigned int flags;
-#define I915_SCHED_HAS_SEMAPHORE	BIT(0)
+#define I915_SCHED_HAS_SEMAPHORE_CHAIN	BIT(0)
+	intel_engine_mask_t semaphores;
 };
 
 struct i915_dependency {