[BugFix][OpenCL] Fix concat image impl when axis is not 1. test=develop (#4241)

* [BugFix][OpenCL] Fix concat image impl when concat axis is not 1

* fix code when axis == 1. test=develop

* fix illegal access when print debug info. test=develop

* fix typo

Author: zhaoyang-star

lite/kernels/opencl/concat_image_compute.cc

@@ -38,15 +38,13 @@ class ConcatComputeImage : public KernelLite<TARGET(kOpenCL),
   void PrepareForRun() override {
     auto& context = ctx_->As<OpenCLContext>();
     concat_param_ = param_.get_mutable<param_t>();
+    axis_ = concat_param_->axis;
+    if (-1 == axis_) {
+      axis_ = concat_param_->x[0]->dims().size() - 1;
+    }
 
     auto inputs = concat_param_->x;
-    auto axis_ = concat_param_->axis;
     auto output_tensor_dims = concat_param_->output->dims();
-    auto* axis_tensor = concat_param_->axis_tensor;
-    if (axis_tensor != nullptr) {
-      // auto* axis_tensor_data = axis_tensor->data<int>(TARGET(kARM));
-      // axis = axis_tensor_data[0];
-    }
 
     if (inputs.size() == 2) {
       kernel_func_name_ = "concat2";
@@ -100,8 +98,7 @@ class ConcatComputeImage : public KernelLite<TARGET(kOpenCL),
           width_ = output_tensor_dims[0];  // n
           flag_ = 2;
           break;
-        case 3:
-        case -1:                           // width
+        case 3:                            // width
           width_ = output_tensor_dims[1];  // c
           flag_ = 3;
           break;
@@ -113,17 +110,12 @@ class ConcatComputeImage : public KernelLite<TARGET(kOpenCL),
     auto input0_tensor_dims = inputs[0]->dims();
     for (int i = 1; i < inputs.size(); i++) {
       auto dims = inputs[i]->dims();
-      // auto flag = CHECK_EQ_OR_FALSE(input0_tensor_dims.size(), dims.size());
-      if (input0_tensor_dims.size() != dims.size()) {
-        printf("input shape must be same \n");
-        return;
-      }
+      CHECK(input0_tensor_dims.size() == dims.size())
+          << "All inputs must have the same axes!";
       for (int i = 0; i < dims.size(); i++) {
         if (i != axis_) {
-          if (input0_tensor_dims[i] != dims[i]) {
-            printf("input shape must be same \n");
-            return;
-          }
+          CHECK(input0_tensor_dims[i] == dims[i])
+              << "All inputs must have the same shape, except at concat axis!";
         }
       }
     }
@@ -151,29 +143,18 @@ class ConcatComputeImage : public KernelLite<TARGET(kOpenCL),
     VLOG(4) << "concat input shape:  ";
     for (size_t i = 0; i < inputs.size(); i++) {
       VLOG(4) << "inputs [" << i << "]"
-              << "[" << inputs[i]->dims().size() << "D]:"
-              << "   dims:" << inputs[i]->dims()[0] << " "
-              << inputs[i]->dims()[1] << " " << inputs[i]->dims()[2] << " "
-              << inputs[i]->dims()[3];
+              << "   dims:" << inputs[i]->dims();
     }
 
     VLOG(4) << "concat output shape:  ";
-    VLOG(4) << " out  dims:  "
-            << "[" << output_tensor_dims.size()
-            << "D]:" << output_tensor_dims[0] << " " << output_tensor_dims[1]
-            << " " << output_tensor_dims[2] << " " << output_tensor_dims[3];
+    VLOG(4) << " out  dims:  " << output_tensor_dims;
     VLOG(4) << "axis_: " << axis_;
     VLOG(4) << "flag_: " << flag_;
 
     VLOG(4) << TargetToStr(concat_param_->output->target());
-    VLOG(4) << "output_image_shape(w,h):" << output_image_shape["width"] << " "
+    VLOG(4) << "output_image_shape(w,h): " << output_image_shape["width"] << " "
             << output_image_shape["height"];
-    VLOG(4) << "output_tensor_dims[" << output_tensor_dims.size()
-            << "D]:" << output_tensor_dims[0] << " " << output_tensor_dims[1]
-            << " " << output_tensor_dims[2] << " " << output_tensor_dims[3]
-            << "output_tensor_dims[output_tensor_dims.size() - 1]"
-            << output_tensor_dims[output_tensor_dims.size() - 1];
-    VLOG(4) << "output_tensor_w: " << output_tensor_w << ", flag_: " << flag_;
+    VLOG(4) << "output_tensor_w: " << output_tensor_w;
     VLOG(4) << "width_:" << width_;
     VLOG(4) << "global_work_size: "
             << output_tensor_dims[output_tensor_dims.size() - 1] << "  "
@@ -433,6 +414,7 @@ class ConcatComputeImage : public KernelLite<TARGET(kOpenCL),
   }
 #endif
 
+ private:
   int axis_ = 1;
   int flag_ = 1;
   int width_ = 1;