private IndexRoot root;

        @Override
        protected byte getType() {
            return 0x77;
        }

        @Override
        protected int getSize() {
            return Block.INT_SIZE + Block.LONG_SIZE + (3 * Block.LONG_SIZE) * maxChildIndexEntries;
        }

        @Override
        public void read(DataInputStream instr) throws IOException {
            int count = instr.readInt();

						// at least one new coverage bit.
						c.queueForMinimization(result, keepCoverage)
					} else {
						// Update the coordinator's coverage mask and save the value.
						inputSize := len(result.entry.Data)
						entryNew, err := c.addCorpusEntries(true, result.entry)
						if err != nil {
							stop(err)
							break
						}
						if !entryNew {
							if shouldPrintDebugInfo() {

// Aleksey Sidorov <******@****.***>

#define YFER_SIZE (4*8)
#define SRND_SIZE (1*8)
#define INP_SIZE (1*8)

#define frame_YFER (0)
#define frame_SRND (frame_YFER + YFER_SIZE)
#define frame_INP (frame_SRND + SRND_SIZE)
#define frame_INPEND (frame_INP + INP_SIZE)

#define addm(p1, p2) \
	ADDQ p1, p2; \
	MOVQ p2, p1

#define COPY_YMM_AND_BSWAP(p1, p2, p3) \
	VMOVDQU p2, p1;    \

#define y0 R13
#define y1 R14
#define y2 R15
#define y3 DI

// Offsets
#define XFER_SIZE 2*64*4
#define INP_END_SIZE 8
#define INP_SIZE 8

#define _XFER 0
#define _INP_END _XFER + XFER_SIZE
#define _INP _INP_END + INP_END_SIZE
#define STACK_SIZE _INP + INP_SIZE

#define ROUND_AND_SCHED_N_0(disp, a, b, c, d, e, f, g, h, XDWORD0, XDWORD1, XDWORD2, XDWORD3) \

    // CHECK-SAME: batch_group_count = 1 : i64
    // CHECK-SAME: feature_group_count = 1 : i64
    // CHECK: return %[[RESULT]]
  %input_sizes = "tf.Const" () { value = dense<[100,28,28,1]> : tensor<4xi32> } : () -> tensor<4xi32>
  %result = "tf.Conv2DBackpropInput"(%input_sizes, %filter, %out_backprop) {
    data_format = "NHWC",
    dilations = [1, 1, 1, 1],
    explicit_paddings = [],
    padding = "VALID",

  }];

  let arguments = (ins
    Arg<TF_Int32Tensor, [{An integer vector representing the shape of `input`,
where `input` is a 4-D `[batch, height, width, channels]` tensor.}]>:$input_sizes,
    Arg<TensorOf<[TF_Bfloat16, TF_Float16, TF_Float32, TF_Float64, TF_Int32]>, [{4-D with shape
`[filter_height, filter_width, in_channels, out_channels]`.}]>:$filter,