drm/i915: Individualize fences before adding to dma_resv obj

_i915_vma_move_to_active() can receive > 1 fences for
multiple batch buffers submission. Because dma_resv_add_fence()
can only accept one fence at a time, change _i915_vma_move_to_active()
to be aware of multiple fences so that it can add individual
fences to the dma resv object.

v6: fix multi-line comment.
v5: remove double fence reservation for batch VMAs.
v4: Reserve fences for composite_fence on multi-batch contexts and
    also reserve fence slots to composite_fence for each VMAs.
v3: dma_resv_reserve_fences is not cumulative so pass num_fences.
v2: make sure to reserve enough fence slots before adding.

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/5614
Fixes: 544460c3 ("drm/i915: Multi-BB execbuf")
Cc: <stable@vger.kernel.org> # v5.16+
Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220525095955.15371-1-nirmoy.das@intel.com

drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c

@@ -999,7 +999,8 @@ static int eb_validate_vmas(struct i915_execbuffer *eb)
 			}
 		}
 
-		err = dma_resv_reserve_fences(vma->obj->base.resv, 1);
+		/* Reserve enough slots to accommodate composite fences */
+		err = dma_resv_reserve_fences(vma->obj->base.resv, eb->num_batches);
 		if (err)
 			return err;
 

drivers/gpu/drm/i915/i915_vma.c

@@ -23,6 +23,7 @@
  */
 
 #include <linux/sched/mm.h>
+#include <linux/dma-fence-array.h>
 #include <drm/drm_gem.h>
 
 #include "display/intel_frontbuffer.h"
@@ -1823,6 +1824,21 @@ int _i915_vma_move_to_active(struct i915_vma *vma,
 	if (unlikely(err))
 		return err;
 
+	/*
+	 * Reserve fences slot early to prevent an allocation after preparing
+	 * the workload and associating fences with dma_resv.
+	 */
+	if (fence && !(flags & __EXEC_OBJECT_NO_RESERVE)) {
+		struct dma_fence *curr;
+		int idx;
+
+		dma_fence_array_for_each(curr, idx, fence)
+			;
+		err = dma_resv_reserve_fences(vma->obj->base.resv, idx);
+		if (unlikely(err))
+			return err;
+	}
+
 	if (flags & EXEC_OBJECT_WRITE) {
 		struct intel_frontbuffer *front;
 
@@ -1832,31 +1848,23 @@ int _i915_vma_move_to_active(struct i915_vma *vma,
 				i915_active_add_request(&front->write, rq);
 			intel_frontbuffer_put(front);
 		}
-
-		if (!(flags & __EXEC_OBJECT_NO_RESERVE)) {
-			err = dma_resv_reserve_fences(vma->obj->base.resv, 1);
-			if (unlikely(err))
-				return err;
 	}
 
 	if (fence) {
-			dma_resv_add_fence(vma->obj->base.resv, fence,
-					   DMA_RESV_USAGE_WRITE);
-			obj->write_domain = I915_GEM_DOMAIN_RENDER;
+		struct dma_fence *curr;
+		enum dma_resv_usage usage;
+		int idx;
+
 		obj->read_domains = 0;
-		}
+		if (flags & EXEC_OBJECT_WRITE) {
+			usage = DMA_RESV_USAGE_WRITE;
+			obj->write_domain = I915_GEM_DOMAIN_RENDER;
 		} else {
-		if (!(flags & __EXEC_OBJECT_NO_RESERVE)) {
-			err = dma_resv_reserve_fences(vma->obj->base.resv, 1);
-			if (unlikely(err))
-				return err;
+			usage = DMA_RESV_USAGE_READ;
 		}
 
-		if (fence) {
-			dma_resv_add_fence(vma->obj->base.resv, fence,
-					   DMA_RESV_USAGE_READ);
-			obj->write_domain = 0;
-		}
+		dma_fence_array_for_each(curr, idx, fence)
+			dma_resv_add_fence(vma->obj->base.resv, curr, usage);
 	}
 
 	if (flags & EXEC_OBJECT_NEEDS_FENCE && vma->fence)