drm/i915: Add support for FBs requiring a POT stride alignment

An upcoming platform has a restriction that the FB stride must be
power-of-two aligned. To support framebuffer layouts that are not in
this layout add a logic that pads the tile rows to the POT aligned size.

The HW won't read the padding PTEs, so these don't have to point to an
allocated address, or even have their valid flag set. So use a NULL PTE
instead for instance the scratch page, which is simple and keeps the SG
table compact.

v2:
- Simplify plane_view_dst_stride(). (Ville)
- Pass pitch_tiles as unsigned int.
v3:
- Drop unintentional s/plane_state->rotation/plane_config->rotation/
  change.
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210325214808.2071517-24-imre.deak@intel.com

drivers/gpu/drm/i915/display/intel_display.c

@@ -933,7 +933,7 @@ unsigned int intel_rotation_info_size(const struct intel_rotation_info *rot_info
 	int i;
 
 	for (i = 0 ; i < ARRAY_SIZE(rot_info->plane); i++)
-		size += rot_info->plane[i].width * rot_info->plane[i].height;
+		size += rot_info->plane[i].dst_stride * rot_info->plane[i].width;
 
 	return size;
 }
@@ -944,7 +944,7 @@ unsigned int intel_remapped_info_size(const struct intel_remapped_info *rem_info
 	int i;
 
 	for (i = 0 ; i < ARRAY_SIZE(rem_info->plane); i++)
-		size += rem_info->plane[i].width * rem_info->plane[i].height;
+		size += rem_info->plane[i].dst_stride * rem_info->plane[i].height;
 
 	return size;
 }

drivers/gpu/drm/i915/display/intel_display_types.h

@@ -128,6 +128,7 @@ struct intel_framebuffer {
 	/* Params to remap the FB pages and program the plane registers in each view. */
 	struct intel_fb_view normal_view;
 	struct intel_fb_view rotated_view;
+	struct intel_fb_view remapped_view;
 };
 
 struct intel_fbdev {

drivers/gpu/drm/i915/display/intel_fb.c

@@ -486,10 +486,17 @@ static bool intel_plane_can_remap(const struct intel_plane_state *plane_state)
 	return true;
 }
 
+static bool intel_fb_needs_pot_stride_remap(const struct intel_framebuffer *fb)
+{
+	return false;
+}
+
 static int intel_fb_pitch(const struct intel_framebuffer *fb, int color_plane, unsigned int rotation)
 {
 	if (drm_rotation_90_or_270(rotation))
 		return fb->rotated_view.color_plane[color_plane].stride;
+	else if (intel_fb_needs_pot_stride_remap(fb))
+		return fb->remapped_view.color_plane[color_plane].stride;
 	else
 		return fb->normal_view.color_plane[color_plane].stride;
 }
@@ -597,6 +604,16 @@ plane_view_src_stride_tiles(const struct intel_framebuffer *fb, int color_plane,
 			    dims->tile_width * fb->base.format->cpp[color_plane]);
 }
 
+static unsigned int
+plane_view_dst_stride_tiles(const struct intel_framebuffer *fb, int color_plane,
+			    unsigned int pitch_tiles)
+{
+	if (intel_fb_needs_pot_stride_remap(fb))
+		return roundup_pow_of_two(pitch_tiles);
+	else
+		return pitch_tiles;
+}
+
 static unsigned int
 plane_view_width_tiles(const struct intel_framebuffer *fb, int color_plane,
 		       const struct fb_plane_view_dims *dims,
@@ -629,8 +646,8 @@ static u32 calc_plane_remap_info(const struct intel_framebuffer *fb, int color_p
 	unsigned int tile_width = dims->tile_width;
 	unsigned int tile_height = dims->tile_height;
 	unsigned int tile_size = intel_tile_size(i915);
-	unsigned int pitch_tiles;
 	struct drm_rect r;
+	u32 size;
 
 	assign_chk_ovf(i915, remap_info->offset, obj_offset);
 	assign_chk_ovf(i915, remap_info->src_stride, plane_view_src_stride_tiles(fb, color_plane, dims));
@@ -640,6 +657,9 @@ static u32 calc_plane_remap_info(const struct intel_framebuffer *fb, int color_p
 	if (view->gtt.type == I915_GGTT_VIEW_ROTATED) {
 		check_array_bounds(i915, view->gtt.rotated.plane, color_plane);
 
+		assign_chk_ovf(i915, remap_info->dst_stride,
+			       plane_view_dst_stride_tiles(fb, color_plane, remap_info->height));
+
 		/* rotate the x/y offsets to match the GTT view */
 		drm_rect_init(&r, x, y, dims->width, dims->height);
 		drm_rect_rotate(&r,
@@ -650,8 +670,9 @@ static u32 calc_plane_remap_info(const struct intel_framebuffer *fb, int color_p
 		color_plane_info->x = r.x1;
 		color_plane_info->y = r.y1;
 
-		pitch_tiles = remap_info->height;
-		color_plane_info->stride = pitch_tiles * tile_height;
+		color_plane_info->stride = remap_info->dst_stride * tile_height;
+
+		size = remap_info->dst_stride * remap_info->width;
 
 		/* rotate the tile dimensions to match the GTT view */
 		swap(tile_width, tile_height);
@@ -660,12 +681,16 @@ static u32 calc_plane_remap_info(const struct intel_framebuffer *fb, int color_p
 
 		check_array_bounds(i915, view->gtt.remapped.plane, color_plane);
 
+		assign_chk_ovf(i915, remap_info->dst_stride,
+			       plane_view_dst_stride_tiles(fb, color_plane, remap_info->width));
+
 		color_plane_info->x = x;
 		color_plane_info->y = y;
 
-		pitch_tiles = remap_info->width;
-		color_plane_info->stride = pitch_tiles * tile_width *
-					  fb->base.format->cpp[color_plane];
+		color_plane_info->stride = remap_info->dst_stride * tile_width *
+					   fb->base.format->cpp[color_plane];
+
+		size = remap_info->dst_stride * remap_info->height;
 	}
 
 	/*
@@ -675,10 +700,10 @@ static u32 calc_plane_remap_info(const struct intel_framebuffer *fb, int color_p
 	 */
 	intel_adjust_tile_offset(&color_plane_info->x, &color_plane_info->y,
 				 tile_width, tile_height,
-				 tile_size, pitch_tiles,
+				 tile_size, remap_info->dst_stride,
 				 gtt_offset * tile_size, 0);
 
-	return remap_info->width * remap_info->height;
+	return size;
 }
 
 #undef assign_chk_ovf
@@ -723,12 +748,14 @@ int intel_fill_fb_info(struct drm_i915_private *i915, struct drm_framebuffer *fb
 	struct intel_framebuffer *intel_fb = to_intel_framebuffer(fb);
 	struct drm_i915_gem_object *obj = intel_fb_obj(fb);
 	u32 gtt_offset_rotated = 0;
+	u32 gtt_offset_remapped = 0;
 	unsigned int max_size = 0;
 	int i, num_planes = fb->format->num_planes;
 	unsigned int tile_size = intel_tile_size(i915);
 
 	intel_fb_view_init(&intel_fb->normal_view, I915_GGTT_VIEW_NORMAL);
 	intel_fb_view_init(&intel_fb->rotated_view, I915_GGTT_VIEW_ROTATED);
+	intel_fb_view_init(&intel_fb->remapped_view, I915_GGTT_VIEW_REMAPPED);
 
 	for (i = 0; i < num_planes; i++) {
 		struct fb_plane_view_dims view_dims;
@@ -776,6 +803,11 @@ int intel_fill_fb_info(struct drm_i915_private *i915, struct drm_framebuffer *fb
 								    offset, gtt_offset_rotated, x, y,
 								    &intel_fb->rotated_view);
 
+		if (intel_fb_needs_pot_stride_remap(intel_fb))
+			gtt_offset_remapped += calc_plane_remap_info(intel_fb, i, &view_dims,
+								     offset, gtt_offset_remapped, x, y,
+								     &intel_fb->remapped_view);
+
 		size = calc_plane_normal_size(intel_fb, i, &view_dims, x, y);
 		/* how many tiles in total needed in the bo */
 		max_size = max(max_size, offset + size);
@@ -859,6 +891,8 @@ void intel_fb_fill_view(const struct intel_framebuffer *fb, unsigned int rotatio
 {
 	if (drm_rotation_90_or_270(rotation))
 		*view = fb->rotated_view;
+	else if (intel_fb_needs_pot_stride_remap(fb))
+		*view = fb->remapped_view;
 	else
 		*view = fb->normal_view;
 }

drivers/gpu/drm/i915/gt/intel_ggtt.c

@@ -1247,14 +1247,16 @@ void i915_ggtt_resume(struct i915_ggtt *ggtt)
 static struct scatterlist *
 rotate_pages(struct drm_i915_gem_object *obj, unsigned int offset,
 	     unsigned int width, unsigned int height,
-	     unsigned int stride,
+	     unsigned int src_stride, unsigned int dst_stride,
 	     struct sg_table *st, struct scatterlist *sg)
 {
 	unsigned int column, row;
 	unsigned int src_idx;
 
 	for (column = 0; column < width; column++) {
-		src_idx = stride * (height - 1) + column + offset;
+		unsigned int left;
+
+		src_idx = src_stride * (height - 1) + column + offset;
 		for (row = 0; row < height; row++) {
 			st->nents++;
 			/*
@@ -1267,8 +1269,25 @@ rotate_pages(struct drm_i915_gem_object *obj, unsigned int offset,
 				i915_gem_object_get_dma_address(obj, src_idx);
 			sg_dma_len(sg) = I915_GTT_PAGE_SIZE;
 			sg = sg_next(sg);
-			src_idx -= stride;
+			src_idx -= src_stride;
 		}
+
+		left = (dst_stride - height) * I915_GTT_PAGE_SIZE;
+
+		if (!left)
+			continue;
+
+		st->nents++;
+
+		/*
+		 * The DE ignores the PTEs for the padding tiles, the sg entry
+		 * here is just a conenience to indicate how many padding PTEs
+		 * to insert at this spot.
+		 */
+		sg_set_page(sg, NULL, left, 0);
+		sg_dma_address(sg) = 0;
+		sg_dma_len(sg) = left;
+		sg = sg_next(sg);
 	}
 
 	return sg;
@@ -1297,11 +1316,12 @@ intel_rotate_pages(struct intel_rotation_info *rot_info,
 	st->nents = 0;
 	sg = st->sgl;
 
-	for (i = 0 ; i < ARRAY_SIZE(rot_info->plane); i++) {
+	for (i = 0 ; i < ARRAY_SIZE(rot_info->plane); i++)
 		sg = rotate_pages(obj, rot_info->plane[i].offset,
 				  rot_info->plane[i].width, rot_info->plane[i].height,
-				  rot_info->plane[i].src_stride, st, sg);
-	}
+				  rot_info->plane[i].src_stride,
+				  rot_info->plane[i].dst_stride,
+				  st, sg);
 
 	return st;
 
@@ -1319,7 +1339,7 @@ intel_rotate_pages(struct intel_rotation_info *rot_info,
 static struct scatterlist *
 remap_pages(struct drm_i915_gem_object *obj, unsigned int offset,
 	    unsigned int width, unsigned int height,
-	    unsigned int stride,
+	    unsigned int src_stride, unsigned int dst_stride,
 	    struct sg_table *st, struct scatterlist *sg)
 {
 	unsigned int row;
@@ -1352,7 +1372,24 @@ remap_pages(struct drm_i915_gem_object *obj, unsigned int offset,
 			left -= length;
 		}
 
-		offset += stride - width;
+		offset += src_stride - width;
+
+		left = (dst_stride - width) * I915_GTT_PAGE_SIZE;
+
+		if (!left)
+			continue;
+
+		st->nents++;
+
+		/*
+		 * The DE ignores the PTEs for the padding tiles, the sg entry
+		 * here is just a conenience to indicate how many padding PTEs
+		 * to insert at this spot.
+		 */
+		sg_set_page(sg, NULL, left, 0);
+		sg_dma_address(sg) = 0;
+		sg_dma_len(sg) = left;
+		sg = sg_next(sg);
 	}
 
 	return sg;
@@ -1384,7 +1421,8 @@ intel_remap_pages(struct intel_remapped_info *rem_info,
 	for (i = 0 ; i < ARRAY_SIZE(rem_info->plane); i++) {
 		sg = remap_pages(obj, rem_info->plane[i].offset,
 				 rem_info->plane[i].width, rem_info->plane[i].height,
-				 rem_info->plane[i].src_stride, st, sg);
+				 rem_info->plane[i].src_stride, rem_info->plane[i].dst_stride,
+				 st, sg);
 	}
 
 	i915_sg_trim(st);

drivers/gpu/drm/i915/i915_debugfs.c

@@ -173,26 +173,30 @@ i915_debugfs_describe_obj(struct seq_file *m, struct drm_i915_gem_object *obj)
 				break;
 
 			case I915_GGTT_VIEW_ROTATED:
-				seq_printf(m, ", rotated [(%ux%u, stride=%u, offset=%u), (%ux%u, stride=%u, offset=%u)]",
+				seq_printf(m, ", rotated [(%ux%u, src_stride=%u, dst_stride=%u, offset=%u), (%ux%u, src_stride=%u, dst_stride=%u, offset=%u)]",
 					   vma->ggtt_view.rotated.plane[0].width,
 					   vma->ggtt_view.rotated.plane[0].height,
 					   vma->ggtt_view.rotated.plane[0].src_stride,
+					   vma->ggtt_view.rotated.plane[0].dst_stride,
 					   vma->ggtt_view.rotated.plane[0].offset,
 					   vma->ggtt_view.rotated.plane[1].width,
 					   vma->ggtt_view.rotated.plane[1].height,
 					   vma->ggtt_view.rotated.plane[1].src_stride,
+					   vma->ggtt_view.rotated.plane[1].dst_stride,
 					   vma->ggtt_view.rotated.plane[1].offset);
 				break;
 
 			case I915_GGTT_VIEW_REMAPPED:
-				seq_printf(m, ", remapped [(%ux%u, stride=%u, offset=%u), (%ux%u, stride=%u, offset=%u)]",
+				seq_printf(m, ", remapped [(%ux%u, src_stride=%u, dst_stride=%u, offset=%u), (%ux%u, src_stride=%u, dst_stride=%u, offset=%u)]",
 					   vma->ggtt_view.remapped.plane[0].width,
 					   vma->ggtt_view.remapped.plane[0].height,
 					   vma->ggtt_view.remapped.plane[0].src_stride,
+					   vma->ggtt_view.remapped.plane[0].dst_stride,
 					   vma->ggtt_view.remapped.plane[0].offset,
 					   vma->ggtt_view.remapped.plane[1].width,
 					   vma->ggtt_view.remapped.plane[1].height,
 					   vma->ggtt_view.remapped.plane[1].src_stride,
+					   vma->ggtt_view.remapped.plane[1].dst_stride,
 					   vma->ggtt_view.remapped.plane[1].offset);
 				break;
 

drivers/gpu/drm/i915/i915_vma_types.h

@@ -101,7 +101,7 @@ struct intel_remapped_plane_info {
 	u16 width;
 	u16 height;
 	u16 src_stride;
-	u16 unused_mbz;
+	u16 dst_stride;
 } __packed;
 
 struct intel_remapped_info {

drivers/gpu/drm/i915/selftests/i915_vma.c

@@ -528,6 +528,15 @@ static int igt_vma_rotate_remap(void *arg)
 			GEM_BUG_ON(max_offset > max_pages);
 			max_offset = max_pages - max_offset;
 
+			if (!plane_info[0].dst_stride)
+				plane_info[0].dst_stride = view.type == I915_GGTT_VIEW_ROTATED ?
+									plane_info[0].height :
+									plane_info[0].width;
+			if (!plane_info[1].dst_stride)
+				plane_info[1].dst_stride = view.type == I915_GGTT_VIEW_ROTATED ?
+									plane_info[1].height :
+									plane_info[1].width;
+
 			for_each_prime_number_from(plane_info[0].offset, 0, max_offset) {
 				for_each_prime_number_from(plane_info[1].offset, 0, max_offset) {
 					struct scatterlist *sg;
@@ -902,6 +911,10 @@ static int igt_vma_remapped_gtt(void *arg)
 			if (err)
 				goto out;
 
+			if (!plane_info[0].dst_stride)
+				plane_info[0].dst_stride = *t == I915_GGTT_VIEW_ROTATED ?
+								 p->height : p->width;
+
 			vma = i915_gem_object_ggtt_pin(obj, &view, 0, 0, PIN_MAPPABLE);
 			if (IS_ERR(vma)) {
 				err = PTR_ERR(vma);