diff --git a/drivers/media/platform/rockchip/cif/capture.c b/drivers/media/platform/rockchip/cif/capture.c
index 2e2ccab3de4c..481498974537 100644
--- a/drivers/media/platform/rockchip/cif/capture.c
+++ b/drivers/media/platform/rockchip/cif/capture.c
@@ -4704,7 +4704,12 @@ static int rkcif_queue_setup(struct vb2_queue *queue,
 
 	for (i = 0; i < cif_fmt->mplanes; i++) {
 		const struct v4l2_plane_pix_format *plane_fmt;
-		int h = round_up(height, MEMORY_ALIGN_ROUND_UP_HEIGHT);
+		int h;
+
+		if (dev->chip_id >= CHIP_RK3588_CIF && dev->terminal_sensor.hdmi_input_en)
+			h = height;
+		else
+			h = round_up(height, MEMORY_ALIGN_ROUND_UP_HEIGHT);
 
 		plane_fmt = &pixm->plane_fmt[i];
 		sizes[i] = plane_fmt->sizeimage / height * h;
@@ -6111,6 +6116,13 @@ int rkcif_update_sensor_info(struct rkcif_stream *stream)
 		} else {
 			terminal_sensor->dsi_mode = 0;
 		}
+		if (v4l2_subdev_call(terminal_sensor->sd, core, ioctl, RKMODULE_GET_HDMI_MODE,
+					&terminal_sensor->hdmi_input_en)) {
+			v4l2_dbg(1, rkcif_debug, &stream->cifdev->v4l2_dev,
+				"%s: get terminal %s hdmiin, default!\n",
+				__func__, terminal_sensor->sd->name);
+			terminal_sensor->hdmi_input_en = 0;
+		}
 	} else {
 		v4l2_err(&stream->cifdev->v4l2_dev,
 			 "%s: stream[%d] get remote terminal sensor failed!\n",
diff --git a/drivers/media/platform/rockchip/cif/dev.h b/drivers/media/platform/rockchip/cif/dev.h
index 3219f7752b8a..a15e578549b6 100644
--- a/drivers/media/platform/rockchip/cif/dev.h
+++ b/drivers/media/platform/rockchip/cif/dev.h
@@ -227,6 +227,7 @@ struct rkcif_sensor_info {
 	struct v4l2_subdev_selection selection;
 	int dsi_input_en;
 	int dsi_mode;
+	int hdmi_input_en;
 };
 
 enum cif_fmt_type {