Fix int8 softmax bugs

mindspore/lite/src/runtime/kernel/arm/int8/softmax_int8.cc

@@ -15,6 +15,7 @@
  */
 
 #include "src/runtime/kernel/arm/int8/softmax_int8.h"
+#include <limits>
 #include "src/runtime/kernel/arm/nnacl/int8/softmax_int8.h"
 #include "schema/model_generated.h"
 #include "src/runtime/runtime_api.h"
@@ -44,6 +45,8 @@ int SoftmaxInt8CPUKernel::Init() {
   auto out_quant_args = out_tensor->GetQuantParams();
   quant_params_.out_quant_arg_.scale_ = out_quant_args.front().scale;
   quant_params_.out_quant_arg_.zp_ = out_quant_args.front().zeroPoint;
+  quant_params_.output_activation_min_ = std::numeric_limits<int8_t>::min();
+  quant_params_.output_activation_max_ = std::numeric_limits<int8_t>::max();
 
   if (!InferShapeDone()) {
     return RET_OK;
@@ -95,12 +98,10 @@ int SoftmaxInt8CPUKernel::DoSoftmax(int task_id) {
 
   int stride = UP_DIV(outter_size, thread_count_);
   int count = MSMIN(stride, outter_size - stride * task_id);
+  int stride_size = stride * task_id * inner_size;
 
-  input_ptr += stride * task_id * inner_size;
-  output_ptr += stride * task_id * inner_size;
-  exp_data_ += stride * task_id * inner_size;
-
-  auto error_code = Int8Softmax(input_ptr, output_ptr, count, exp_data_, sum_data_, quant_params_, softmax_param_);
+  auto error_code = SoftmaxInt8(input_ptr + stride_size, output_ptr + stride_size, count, exp_data_ + stride_size,
+                                sum_data_, quant_params_, softmax_param_);
   if (error_code != RET_OK) {
     MS_LOG(ERROR) << "DoSoftmax error task_id[" << task_id << "] error_code[" << error_code << "]";
     return RET_ERROR;

mindspore/lite/src/runtime/kernel/arm/int8/softmax_int8.h

@@ -37,8 +37,8 @@ class SoftmaxInt8CPUKernel : public SoftmaxBaseCPUKernel {
 
  private:
   void FreeTmpBuffer();
-  float *sum_data_;
-  float *exp_data_;
+  float *sum_data_ = nullptr;
+  float *exp_data_ = nullptr;
   SoftmaxQuantArg quant_params_;
 };
 }  // namespace mindspore::kernel

mindspore/lite/src/runtime/kernel/arm/nnacl/int8/softmax_int8.c

@@ -17,7 +17,7 @@
 #include "nnacl/int8/softmax_int8.h"
 #include <math.h>
 
-int Int8Softmax(const int8_t *input_ptr, int8_t *output_ptr, int count, float *exp_data, float *sum_data,
+int SoftmaxInt8(const int8_t *input_ptr, int8_t *output_ptr, int count, float *exp_data, float *sum_data,
                 SoftmaxQuantArg quant_param, SoftmaxParameter *parameter) {
   int32_t axis = parameter->axis_;
   int n_dim = parameter->n_dim_;
@@ -48,7 +48,8 @@ int Int8Softmax(const int8_t *input_ptr, int8_t *output_ptr, int count, float *e
         int inner_offset = axis_offset + i;
         float real_output = exp_data[inner_offset] / sum_data[i];
         int32_t output_scaled = round(real_output / output_scale) + output_zp;
-        output_ptr[inner_offset] = MSMAX(CHAR_MIN, MSMIN(CHAR_MAX, output_scaled));
+        output_ptr[inner_offset] =
+          MSMAX(quant_param.output_activation_min_, MSMIN(quant_param.output_activation_max_, output_scaled));
       }
     }
   }

mindspore/lite/src/runtime/kernel/arm/nnacl/int8/softmax_int8.h

@@ -24,7 +24,7 @@
 #ifdef __cplusplus
 extern "C" {
 #endif
-int Int8Softmax(const int8_t *input_ptr, int8_t *output_ptr, int count, float *exp_data, float *sum_data,
+int SoftmaxInt8(const int8_t *input_ptr, int8_t *output_ptr, int count, float *exp_data, float *sum_data,
                 SoftmaxQuantArg quant_param, SoftmaxParameter *parameter);
 #ifdef __cplusplus
 }

mindspore/lite/src/runtime/kernel/arm/nnacl/quantization/quantize.h

@@ -169,6 +169,11 @@ typedef struct SplitQuantArg {
 typedef struct SoftmaxQuantArg {
   QuantArg in_quant_args_;
   QuantArg out_quant_arg_;
+  int output_activation_min_;
+  int output_activation_max_;
+  int output_multiplier_;
+  int shift_left_;
+  int shift_right_;
 } SoftmaxQuantArg;
 
 typedef struct ReshapeQuantArg {