init:初始框架

libraries/components/seekfree_assistant/seekfree_assistant.c

@@ -0,0 +1,441 @@
+/*********************************************************************************************************************
+* MSPM0G3507 Opensourec Library 即（MSPM0G3507 开源库）是一个基于官方 SDK 接口的第三方开源库
+* Copyright (c) 2022 SEEKFREE 逐飞科技
+* 
+* 本文件是 MSPM0G3507 开源库的一部分
+* 
+* MSPM0G3507 开源库 是免费软件
+* 您可以根据自由软件基金会发布的 GPL（GNU General Public License，即 GNU通用公共许可证）的条款
+* 即 GPL 的第3版（即 GPL3.0）或（您选择的）任何后来的版本，重新发布和/或修改它
+* 
+* 本开源库的发布是希望它能发挥作用，但并未对其作任何的保证
+* 甚至没有隐含的适销性或适合特定用途的保证
+* 更多细节请参见 GPL
+* 
+* 您应该在收到本开源库的同时收到一份 GPL 的副本
+* 如果没有，请参阅<https://www.gnu.org/licenses/>
+* 
+* 额外注明：
+* 本开源库使用 GPL3.0 开源许可证协议 以上许可申明为译文版本
+* 许可申明英文版在 libraries/doc 文件夹下的 GPL3_permission_statement.txt 文件中
+* 许可证副本在 libraries 文件夹下 即该文件夹下的 LICENSE 文件
+* 欢迎各位使用并传播本程序 但修改内容时必须保留逐飞科技的版权声明（即本声明）
+* 
+* 文件名称          seekfree_assistant
+* 公司名称          成都逐飞科技有限公司
+* 版本信息          查看 libraries/doc 文件夹内 version 文件 版本说明
+* 开发环境          MDK 5.37
+* 适用平台          MSPM0G3507
+* 店铺链接          https://seekfree.taobao.com/
+* 
+* 修改记录
+* 日期             作者             备注
+* 2024-1-11        SeekFree         first version
+********************************************************************************************************************/
+
+#include "zf_common_debug.h"
+
+#include "seekfree_assistant.h"
+
+
+extern uint32 seekfree_assistant_transfer       (const uint8 *buff, uint32 length);
+extern uint32 seekfree_assistant_receive        (uint8 *buff, uint32 length);
+
+#if (1 == SEEKFREE_ASSISTANT_SET_PARAMETR_ENABLE)
+#include "zf_common_fifo.h"
+static uint8        seekfree_assistant_buffer[SEEKFREE_ASSISTANT_BUFFER_SIZE];                                      // FIFO缓冲区
+static fifo_struct  seekfree_assistant_fifo =                                                                       // FIFO结构体
+{   
+    .buffer    = seekfree_assistant_buffer, 
+    .execution = FIFO_IDLE, 
+    .type      = FIFO_DATA_8BIT,    
+    .head      = 0, 
+    .end       = 0, 
+    .size      = SEEKFREE_ASSISTANT_BUFFER_SIZE,    
+    .max       = SEEKFREE_ASSISTANT_BUFFER_SIZE,    
+};  
+#endif
+
+static seekfree_assistant_camera_struct         seekfree_assistant_camera_data;                                     // 图像上位机协议数据
+static seekfree_assistant_camera_dot_struct     seekfree_assistant_camera_dot_data;                                 // 图像上位机打点协议数据
+static seekfree_assistant_camera_buffer_struct  seekfree_assistant_camera_buffer;                                   // 图像以及边界缓冲区信息
+
+seekfree_assistant_transfer_callback_function   seekfree_assistant_transfer_callback = seekfree_assistant_transfer; // 数据发送函数指针
+seekfree_assistant_receive_callback_function    seekfree_assistant_receive_callback  = seekfree_assistant_receive;  // 数据接收函数指针
+
+seekfree_assistant_oscilloscope_struct          seekfree_assistant_oscilloscope_data;                               // 虚拟示波器数据
+float   seekfree_assistant_parameter[SEEKFREE_ASSISTANT_SET_PARAMETR_COUNT] = {0};                                  // 保存接收到的参数
+vuint8  seekfree_assistant_parameter_update_flag[SEEKFREE_ASSISTANT_SET_PARAMETR_COUNT] = {0};                      // 参数更新标志位
+
+//-------------------------------------------------------------------------------------------------------------------
+// 函数简介     逐飞助手求和函数
+// 参数说明     *buffer         需要校验的数据地址
+// 参数说明     length          校验长度
+// 返回参数     uint8           和值
+// 使用示例
+//-------------------------------------------------------------------------------------------------------------------
+static uint8 seekfree_assistant_sum (uint8 *buffer, uint32 length)
+{
+    uint8 temp_sum = 0;
+
+    while(length--)
+    {
+        temp_sum += *buffer++;
+    }
+
+    return temp_sum;
+}
+
+//-------------------------------------------------------------------------------------------------------------------
+// 函数简介     逐飞助手 图像发送函数
+// 参数说明     camera_type     摄像头类型
+// 参数说明     *image_addr     图像首地址
+// 参数说明     boundary_num    图像中包含边界数量
+// 参数说明     width           图像宽度
+// 参数说明     height          图像高度
+// 返回参数     void
+// 使用示例
+//-------------------------------------------------------------------------------------------------------------------
+void seekfree_assistant_camera_data_send (seekfree_assistant_image_type_enum camera_type, void *image_addr, uint8 boundary_num, uint16 width, uint16 height)
+{
+    uint32 image_size = 0;
+
+    seekfree_assistant_camera_data.head           = SEEKFREE_ASSISTANT_SEND_HEAD;
+    seekfree_assistant_camera_data.function       = SEEKFREE_ASSISTANT_CAMERA_FUNCTION;
+    seekfree_assistant_camera_data.camera_type    = (camera_type << 5) | ((image_addr != NULL ? 0 : 1) << 4) | boundary_num;
+    // 写入包长度信息，仅包含协议部分
+    seekfree_assistant_camera_data.length         = sizeof(seekfree_assistant_camera_struct);
+    seekfree_assistant_camera_data.image_width    = width;
+    seekfree_assistant_camera_data.image_height   = height;
+
+    // 首先发送帧头、功能、摄像头类型、以及宽度高度等信息
+    seekfree_assistant_transfer_callback((const uint8 *)&seekfree_assistant_camera_data, sizeof(seekfree_assistant_camera_struct));
+
+    // 根据摄像头类型计算图像大小
+    switch(camera_type)
+    {
+        case SEEKFREE_ASSISTANT_OV7725_BIN:
+        {
+            image_size = width * height / 8;
+        }break;
+
+        case SEEKFREE_ASSISTANT_MT9V03X:
+        {
+            image_size = width * height;
+        }break;
+
+        case SEEKFREE_ASSISTANT_SCC8660:
+        {
+            image_size = width * height * 2;
+        }break;
+    }
+
+    // 发送图像数据
+    if(NULL != image_addr)
+    {
+        seekfree_assistant_transfer_callback(image_addr, image_size);
+    }
+}
+
+//-------------------------------------------------------------------------------------------------------------------
+// 函数简介     逐飞助手 图像边线绘制函数
+// 参数说明     *buffer     逐飞助手摄像头数据结构体
+// 返回参数     void
+// 使用示例
+//-------------------------------------------------------------------------------------------------------------------
+void seekfree_assistant_camera_dot_send (seekfree_assistant_camera_buffer_struct *buffer)
+{
+    uint8  i;
+    uint16 dot_bytes = 0;   // 点字节数量
+
+    dot_bytes = seekfree_assistant_camera_dot_data.dot_num;
+
+    if(seekfree_assistant_camera_dot_data.dot_type & (1 << 5))
+    {
+        dot_bytes *= 2;
+    }
+
+    // 首先发送帧头、功能、边界编号、坐标长度、点个数
+    seekfree_assistant_transfer_callback((const uint8 *)&seekfree_assistant_camera_dot_data, sizeof(seekfree_assistant_camera_dot_struct));
+
+    for(i=0; i < SEEKFREE_ASSISTANT_CAMERA_MAX_BOUNDARY; i++)
+    {
+        // 判断是否发送横坐标数据
+        if(NULL != buffer->boundary_x[i])
+        {
+            seekfree_assistant_transfer_callback((const uint8 *)buffer->boundary_x[i], dot_bytes);
+        }
+
+        // 判断是否发送纵坐标数据
+        if(NULL != buffer->boundary_y[i])
+        {
+            // 如果没有纵坐标数据，则表示每一行只有一个边界
+            // 指定了横纵坐标数据，这种方式可以实现同一行多个边界的情况，例如搜线算法能够搜索出回弯。
+            seekfree_assistant_transfer_callback((const uint8 *)buffer->boundary_y[i], dot_bytes);
+        }
+    }
+}
+
+//-------------------------------------------------------------------------------------------------------------------
+// 函数简介     逐飞助手 虚拟示波器发送函数
+// 参数说明     *seekfree_assistant_oscilloscope  示波器数据结构体
+// 返回参数     void
+// 使用示例     seekfree_assistant_oscilloscope_send(&seekfree_assistant_oscilloscope_data);
+//-------------------------------------------------------------------------------------------------------------------
+void seekfree_assistant_oscilloscope_send (seekfree_assistant_oscilloscope_struct *seekfree_assistant_oscilloscope)
+{
+    uint8 packet_size;
+
+    // 将高四位清空
+    seekfree_assistant_oscilloscope->channel_num &= 0x0f;
+
+    zf_assert(SEEKFREE_ASSISTANT_SET_OSCILLOSCOPE_COUNT >= seekfree_assistant_oscilloscope->channel_num);
+
+    // 帧头
+    seekfree_assistant_oscilloscope->head         = SEEKFREE_ASSISTANT_SEND_HEAD;
+
+    // 写入包长度信息
+    packet_size                         = sizeof(seekfree_assistant_oscilloscope_struct) - (SEEKFREE_ASSISTANT_SET_OSCILLOSCOPE_COUNT - seekfree_assistant_oscilloscope->channel_num) * 4;
+    seekfree_assistant_oscilloscope->length       = packet_size;
+
+    // 写入功能字与通道数量
+    seekfree_assistant_oscilloscope->channel_num |= SEEKFREE_ASSISTANT_CAMERA_OSCILLOSCOPE;
+
+    // 和校验计算
+    seekfree_assistant_oscilloscope->check_sum    = 0;
+    seekfree_assistant_oscilloscope->check_sum    = seekfree_assistant_sum((uint8 *)seekfree_assistant_oscilloscope, packet_size);
+
+    // 数据在调用本函数之前，由用户将需要发送的数据写入seekfree_assistant_oscilloscope_data.data[]
+
+    seekfree_assistant_transfer_callback((const uint8 *)seekfree_assistant_oscilloscope, packet_size);
+}
+
+//-------------------------------------------------------------------------------------------------------------------
+// 函数简介     逐飞助手图像信息配置函数
+// 参数说明     camera_type     图像类型
+// 参数说明     image_addr      图像地址    如果传递NULL参数则表示只发送边线信息到上位机
+// 参数说明     width           图像宽度
+// 参数说明     height          图像高度
+// 返回参数     void
+// 使用示例                     seekfree_assistant_camera_information_config(SEEKFREE_ASSISTANT_MT9V03X, mt9v03x_image[0], MT9V03X_W, MT9V03X_H);
+// 备注信息
+//-------------------------------------------------------------------------------------------------------------------
+void seekfree_assistant_camera_information_config (seekfree_assistant_image_type_enum camera_type, void *image_addr, uint16 width, uint16 height)
+{
+    seekfree_assistant_camera_dot_data.head       = SEEKFREE_ASSISTANT_SEND_HEAD;
+    seekfree_assistant_camera_dot_data.function   = SEEKFREE_ASSISTANT_CAMERA_DOT_FUNCTION;
+    // 写入包长度信息
+    seekfree_assistant_camera_dot_data.length     = sizeof(seekfree_assistant_camera_dot_struct);
+
+    seekfree_assistant_camera_buffer.camera_type  = camera_type;
+    seekfree_assistant_camera_buffer.image_addr   = image_addr;
+    seekfree_assistant_camera_buffer.width        = width;
+    seekfree_assistant_camera_buffer.height       = height;
+}
+
+//-------------------------------------------------------------------------------------------------------------------
+// 函数简介     逐飞助手图像边线发送配置函数
+// 参数说明     boundary_type   边界类型
+// 参数说明     dot_num         一条边界有多少个点
+// 参数说明     dot_x1          存放边线1横坐标的地址  如果传递NULL参数则表示不发送边线1
+// 参数说明     dot_x2          存放边线2横坐标的地址  如果传递NULL参数则表示不发送边线2
+// 参数说明     dot_x3          存放边线3横坐标的地址  如果传递NULL参数则表示不发送边线3
+// 参数说明     dot_y1          存放边线1纵坐标的地址  如果传递NULL参数则表示不发送边线1
+// 参数说明     dot_y2          存放边线2纵坐标的地址  如果传递NULL参数则表示不发送边线2
+// 参数说明     dot_y3          存放边线3纵坐标的地址  如果传递NULL参数则表示不发送边线3
+// 返回参数     void
+// 使用示例                     seekfree_assistant_camera_config(X_BOUNDARY, MT9V03X_H, x1_boundary, x2_boundary, x3_boundary, NULL, NULL, NULL);     // 图像发送时包含三条边线，边线只有横坐标
+// 使用示例                     seekfree_assistant_camera_config(Y_BOUNDARY, MT9V03X_W, NULL, NULL, NULL, y1_boundary, y2_boundary, y3_boundary);     // 图像发送时包含三条边线，边线只有纵坐标
+// 使用示例                     seekfree_assistant_camera_config(XY_BOUNDARY, 160, xy_x1_boundary, xy_x2_boundary, xy_x3_boundary, xy_y1_boundary, xy_y2_boundary, xy_y3_boundary);   // 图像发送时包含三条边线，边线包含横纵坐标
+// 备注信息
+//-------------------------------------------------------------------------------------------------------------------
+void seekfree_assistant_camera_boundary_config (seekfree_assistant_boundary_type_enum boundary_type, uint16 dot_num, void *dot_x1, void *dot_x2, void *dot_x3, void *dot_y1, void *dot_y2, void *dot_y3)
+{
+    uint8 i = 0;
+    uint8 boundary_num = 0;
+    uint8 boundary_data_type = 0;
+
+    // 检查图像发送缓冲区是否准备就绪, 调用此函数之前需要先调用seekfree_assistant_camera_config设置好图像信息
+    zf_assert(0 != seekfree_assistant_camera_buffer.camera_type);
+
+    seekfree_assistant_camera_dot_data.dot_num    = dot_num;
+    seekfree_assistant_camera_dot_data.valid_flag = 0;
+    for(i = 0; i < 3; i++)
+    {
+        seekfree_assistant_camera_buffer.boundary_x[i] = NULL;
+        seekfree_assistant_camera_buffer.boundary_y[i] = NULL;
+    }
+
+    switch(boundary_type)
+    {
+        case X_BOUNDARY:
+        {
+            if(NULL != dot_x1)
+            {
+                boundary_num++;
+                seekfree_assistant_camera_dot_data.valid_flag |= 1 << 0;
+                seekfree_assistant_camera_buffer.boundary_x[i++] = dot_x1;
+            }
+            if(NULL != dot_x2)
+            {
+                boundary_num++;
+                seekfree_assistant_camera_dot_data.valid_flag |= 1 << 1;
+                seekfree_assistant_camera_buffer.boundary_x[i++] = dot_x2;
+            }
+            if(NULL != dot_x3)
+            {
+                boundary_num++;
+                seekfree_assistant_camera_dot_data.valid_flag |= 1 << 2;
+                seekfree_assistant_camera_buffer.boundary_x[i++] = dot_x3;
+            }
+
+            if(255 < seekfree_assistant_camera_buffer.height)
+            {
+                boundary_data_type = 1;
+            }
+        }break;
+
+        case Y_BOUNDARY:
+        {
+            if(NULL != dot_y1)
+            {
+                boundary_num++;
+                seekfree_assistant_camera_dot_data.valid_flag |= 1 << 0;
+                seekfree_assistant_camera_buffer.boundary_y[i++] = dot_y1;
+            }
+            if(NULL != dot_y2)
+            {
+                boundary_num++;
+                seekfree_assistant_camera_dot_data.valid_flag |= 1 << 1;
+                seekfree_assistant_camera_buffer.boundary_y[i++] = dot_y2;
+            }
+            if(NULL != dot_y3)
+            {
+                boundary_num++;
+                seekfree_assistant_camera_dot_data.valid_flag |= 1 << 2;
+                seekfree_assistant_camera_buffer.boundary_y[i++] = dot_y3;
+            }
+
+            if(255 < seekfree_assistant_camera_buffer.width)
+            {
+                boundary_data_type = 1;
+            }
+        }break;
+
+        case XY_BOUNDARY:
+        {
+            if((NULL != dot_x1) && (NULL != dot_y1))
+            {
+                boundary_num++;
+                seekfree_assistant_camera_dot_data.valid_flag |= 1 << 0;
+                seekfree_assistant_camera_buffer.boundary_x[i]   = dot_x1;
+                seekfree_assistant_camera_buffer.boundary_y[i++] = dot_y1;
+            }
+            if((NULL != dot_x2) && (NULL != dot_y2))
+            {
+                boundary_num++;
+                seekfree_assistant_camera_dot_data.valid_flag |= 1 << 1;
+                seekfree_assistant_camera_buffer.boundary_x[i]   = dot_x2;
+                seekfree_assistant_camera_buffer.boundary_y[i++] = dot_y2;
+            }
+            if((NULL != dot_x3) && (NULL != dot_y3))
+            {
+                boundary_num++;
+                seekfree_assistant_camera_dot_data.valid_flag |= 1 << 2;
+                seekfree_assistant_camera_buffer.boundary_x[i]   = dot_x3;
+                seekfree_assistant_camera_buffer.boundary_y[i++] = dot_y3;
+            }
+
+            if((255 < seekfree_assistant_camera_buffer.width) || (255 < seekfree_assistant_camera_buffer.height))
+            {
+                boundary_data_type = 1;
+            }
+        }break;
+
+        case NO_BOUNDARY:break;
+    }
+
+    seekfree_assistant_camera_dot_data.dot_type   = (boundary_type << 6) | (boundary_data_type << 5) | boundary_num;
+}
+
+//-------------------------------------------------------------------------------------------------------------------
+// 函数简介     逐飞助手发送摄像头图像
+// 参数说明     void
+// 返回参数     void
+// 使用示例
+// 备注信息     在调用图像发送函数之前，请务必调用一次seekfree_assistant_camera_config函数，将对应的参数设置好
+//-------------------------------------------------------------------------------------------------------------------
+void seekfree_assistant_camera_send (void)
+{
+    // 检查图像发送缓冲区是否准备就绪
+    zf_assert(0 != seekfree_assistant_camera_buffer.camera_type);
+
+    seekfree_assistant_camera_data_send(seekfree_assistant_camera_buffer.camera_type, seekfree_assistant_camera_buffer.image_addr, seekfree_assistant_camera_dot_data.dot_type & 0x0f, seekfree_assistant_camera_buffer.width, seekfree_assistant_camera_buffer.height);
+
+    if(seekfree_assistant_camera_dot_data.dot_type & 0x0f)
+    {
+        seekfree_assistant_camera_dot_send(&seekfree_assistant_camera_buffer);
+    }
+}
+
+
+//-------------------------------------------------------------------------------------------------------------------
+// 函数简介     逐飞助手解析接收到的数据
+// 参数说明     void
+// 返回参数     void
+// 使用示例     函数只需要放到周期运行的PIT中断或者主循环即可
+//-------------------------------------------------------------------------------------------------------------------
+#if (1 == SEEKFREE_ASSISTANT_SET_PARAMETR_ENABLE)
+void seekfree_assistant_data_analysis (void)
+{
+    uint8  temp_sum;
+    uint32 read_length;
+    seekfree_assistant_parameter_struct *receive_packet;
+
+    // 这里使用uint32进行定义，目的是为了保证数组四字节对齐
+    uint32  temp_buffer[SEEKFREE_ASSISTANT_BUFFER_SIZE / 4];
+
+    // 尝试读取数据, 如果不是自定义的传输方式则从接收回调中读取数据
+    read_length = seekfree_assistant_receive_callback((uint8 *)temp_buffer, SEEKFREE_ASSISTANT_BUFFER_SIZE);
+    if(read_length)
+    {
+        // 将读取到的数据写入FIFO
+        fifo_write_buffer(&seekfree_assistant_fifo, (uint8 *)temp_buffer, read_length);
+    }
+
+    while(sizeof(seekfree_assistant_parameter_struct) <= fifo_used(&seekfree_assistant_fifo))
+    {
+        read_length = sizeof(seekfree_assistant_parameter_struct);
+        fifo_read_buffer(&seekfree_assistant_fifo, (uint8 *)temp_buffer, &read_length, FIFO_READ_ONLY);
+
+        if(SEEKFREE_ASSISTANT_RECEIVE_HEAD != ((uint8 *)temp_buffer)[0])
+        {
+            // 没有帧头则从FIFO中去掉第一个数据
+            read_length = 1;
+        }
+        else
+        {
+            // 找到帧头
+            receive_packet = (seekfree_assistant_parameter_struct *)temp_buffer;
+            temp_sum = receive_packet->check_sum;
+            receive_packet->check_sum = 0;
+            if(temp_sum == seekfree_assistant_sum((uint8 *)temp_buffer, sizeof(seekfree_assistant_parameter_struct)))
+            {
+                // 和校验成功保存数据
+                seekfree_assistant_parameter[receive_packet->channel - 1] = receive_packet->data;
+                seekfree_assistant_parameter_update_flag[receive_packet->channel - 1] = 1;
+            }
+            else
+            {
+                read_length = 1;
+            }
+        }
+
+        // 丢弃无需使用的数据
+        fifo_read_buffer(&seekfree_assistant_fifo, (uint8 *)temp_buffer, &read_length, FIFO_READ_AND_CLEAN);
+    }
+}
+#endif
+
+
+

libraries/components/seekfree_assistant/seekfree_assistant.h

@@ -0,0 +1,165 @@
+/*********************************************************************************************************************
+* MSPM0G3507 Opensourec Library 即（MSPM0G3507 开源库）是一个基于官方 SDK 接口的第三方开源库
+* Copyright (c) 2022 SEEKFREE 逐飞科技
+* 
+* 本文件是 MSPM0G3507 开源库的一部分
+* 
+* MSPM0G3507 开源库 是免费软件
+* 您可以根据自由软件基金会发布的 GPL（GNU General Public License，即 GNU通用公共许可证）的条款
+* 即 GPL 的第3版（即 GPL3.0）或（您选择的）任何后来的版本，重新发布和/或修改它
+* 
+* 本开源库的发布是希望它能发挥作用，但并未对其作任何的保证
+* 甚至没有隐含的适销性或适合特定用途的保证
+* 更多细节请参见 GPL
+* 
+* 您应该在收到本开源库的同时收到一份 GPL 的副本
+* 如果没有，请参阅<https://www.gnu.org/licenses/>
+* 
+* 额外注明：
+* 本开源库使用 GPL3.0 开源许可证协议 以上许可申明为译文版本
+* 许可申明英文版在 libraries/doc 文件夹下的 GPL3_permission_statement.txt 文件中
+* 许可证副本在 libraries 文件夹下 即该文件夹下的 LICENSE 文件
+* 欢迎各位使用并传播本程序 但修改内容时必须保留逐飞科技的版权声明（即本声明）
+* 
+* 文件名称          seekfree_assistant
+* 公司名称          成都逐飞科技有限公司
+* 版本信息          查看 libraries/doc 文件夹内 version 文件 版本说明
+* 开发环境          MDK 5.37
+* 适用平台          MSPM0G3507
+* 店铺链接          https://seekfree.taobao.com/
+* 
+* 修改记录
+* 日期             作者             备注
+* 2024-1-11        SeekFree         first version
+********************************************************************************************************************/
+#ifndef _seekfree_assistant_h_
+#define _seekfree_assistant_h_
+
+
+#include "zf_common_typedef.h"
+
+
+// 1：使能参数调节的功能  0：关闭参数调节的功能
+#define SEEKFREE_ASSISTANT_SET_PARAMETR_ENABLE      ( 1 )
+
+
+
+// 定义接收FIFO大小
+#define SEEKFREE_ASSISTANT_BUFFER_SIZE              ( 0x80 )
+    
+// 定义示波器的最大通道数量 
+#define SEEKFREE_ASSISTANT_SET_OSCILLOSCOPE_COUNT   ( 0x08 )
+    
+// 定义参数调试的最大通道数量    
+#define SEEKFREE_ASSISTANT_SET_PARAMETR_COUNT       ( 0x08 )
+    
+// 定义图像边线最大数量   
+#define SEEKFREE_ASSISTANT_CAMERA_MAX_BOUNDARY      ( 0x08 )
+    
+// 单片机往上位机发送的帧头 
+#define SEEKFREE_ASSISTANT_SEND_HEAD                ( 0xAA )
+    
+// 摄像头类 
+#define SEEKFREE_ASSISTANT_CAMERA_FUNCTION          ( 0x02 )
+#define SEEKFREE_ASSISTANT_CAMERA_DOT_FUNCTION      ( 0x03 )
+#define SEEKFREE_ASSISTANT_CAMERA_OSCILLOSCOPE      ( 0x10 )
+    
+// 上位机往单片机发送的帧头 
+#define SEEKFREE_ASSISTANT_RECEIVE_HEAD             ( 0x55 )
+    
+// 参数设置类    
+#define SEEKFREE_ASSISTANT_RECEIVE_SET_PARAMETER    ( 0x20 )
+
+
+// 摄像头类型枚举
+typedef enum
+{
+    // 按照摄像头型号定义
+    SEEKFREE_ASSISTANT_OV7725_BIN = 1,
+    SEEKFREE_ASSISTANT_MT9V03X,
+    SEEKFREE_ASSISTANT_SCC8660,
+
+    // 按照图像类型定义
+    SEEKFREE_ASSISTANT_BINARY = 1,
+    SEEKFREE_ASSISTANT_GRAY,
+    SEEKFREE_ASSISTANT_RGB565,
+}seekfree_assistant_image_type_enum;
+
+// 摄像头类型枚举
+typedef enum
+{
+    // 按照摄像头型号定义
+    X_BOUNDARY,     // 发送的图像中边界信息只包含X，也就是只有横坐标信息，纵坐标根据图像高度得到
+    Y_BOUNDARY,     // 发送的图像中边界信息只包含Y，也就是只有纵坐标信息，横坐标根据图像宽度得到，通常很少有这样的需求
+    XY_BOUNDARY,    // 发送的图像中边界信息包含X与Y，这样可以指定点在任意位置，就可以方便显示出回弯的效果
+    NO_BOUNDARY,    // 发送的图像中没有边线信息
+}seekfree_assistant_boundary_type_enum;
+
+typedef struct
+{
+    uint8 head;                                                 // 帧头
+    uint8 channel_num;                                          // 高四位为功能字  低四位为通道数量
+    uint8 check_sum;                                            // 和校验
+    uint8 length;                                               // 包长度
+    float data[SEEKFREE_ASSISTANT_SET_OSCILLOSCOPE_COUNT];      // 通道数据
+}seekfree_assistant_oscilloscope_struct;
+
+
+typedef struct
+{
+    uint8 head;                                                 // 帧头
+    uint8 function;                                             // 功能字
+    uint8 camera_type;                                          // 低四位表示边界数量 第四位表示是否有图像数据  例如0x13：其中3表示一副图像有三条边界（通常是左边界、中线、右边界）、1表示没有图像数据
+    uint8 length;                                               // 包长度（仅包含协议部分）
+    uint16 image_width;                                         // 图像宽度
+    uint16 image_height;                                        // 图像高度
+}seekfree_assistant_camera_struct;          
+            
+            
+typedef struct          
+{           
+    uint8 head;                                                 // 帧头
+    uint8 function;                                             // 功能字
+    uint8 dot_type;                                             // 点类型  BIT5：1：坐标是16位的 0：坐标是8位的    BIT7-BIT6：0：只有X坐标 1：只有Y坐标 2：X和Y坐标都有    BIT3-BIT0：边界数量
+    uint8 length;                                               // 包长度（仅包含协议部分）
+    uint16 dot_num;                                             // 画点数量
+    uint8  valid_flag;                                          // 通道标识
+    uint8  reserve;                                             // 保留
+}seekfree_assistant_camera_dot_struct;          
+            
+typedef struct          
+{           
+    void *image_addr;                                           // 摄像头地址
+    uint16 width;                                               // 图像宽度
+    uint16 height;                                              // 图像高度
+    seekfree_assistant_image_type_enum camera_type;             // 摄像头类型
+    void *boundary_x[SEEKFREE_ASSISTANT_CAMERA_MAX_BOUNDARY];   // 边界横坐标数组地址
+    void *boundary_y[SEEKFREE_ASSISTANT_CAMERA_MAX_BOUNDARY];   // 边界纵坐标数组地址
+}seekfree_assistant_camera_buffer_struct;
+
+typedef struct
+{
+    uint8 head;                                                 // 帧头
+    uint8 function;                                             // 功能字
+    uint8 channel;                                              // 通道
+    uint8 check_sum;                                            // 和校验
+    float data;                                                 // 数据
+}seekfree_assistant_parameter_struct;
+
+typedef uint32 (*seekfree_assistant_transfer_callback_function) (const uint8 *buff, uint32 length);
+typedef uint32 (*seekfree_assistant_receive_callback_function)  (uint8 *buff, uint32 length);
+
+extern seekfree_assistant_oscilloscope_struct                   seekfree_assistant_oscilloscope_data;                                               // 虚拟示波器数据
+extern float                                                    seekfree_assistant_parameter[SEEKFREE_ASSISTANT_SET_PARAMETR_COUNT];                // 保存接收到的参数
+extern vuint8                                                   seekfree_assistant_parameter_update_flag[SEEKFREE_ASSISTANT_SET_PARAMETR_COUNT];    // 参数更新标志位
+    
+    
+void    seekfree_assistant_oscilloscope_send                    (seekfree_assistant_oscilloscope_struct *seekfree_assistant_oscilloscope);
+void    seekfree_assistant_camera_information_config            (seekfree_assistant_image_type_enum camera_type, void *image_addr, uint16 width, uint16 height);
+void    seekfree_assistant_camera_boundary_config               (seekfree_assistant_boundary_type_enum boundary_type, uint16 dot_num, void *dot_x1, void *dot_x2, void *dot_x3, void *dot_y1, void *dot_y2, void *dot_y3);
+void    seekfree_assistant_camera_send                          (void);
+void    seekfree_assistant_data_analysis                        (void);
+
+
+
+#endif

libraries/components/seekfree_assistant/seekfree_assistant_interface.c

@@ -0,0 +1,129 @@
+/*********************************************************************************************************************
+* MSPM0G3507 Opensourec Library 即（MSPM0G3507 开源库）是一个基于官方 SDK 接口的第三方开源库
+* Copyright (c) 2022 SEEKFREE 逐飞科技
+* 
+* 本文件是 MSPM0G3507 开源库的一部分
+* 
+* MSPM0G3507 开源库 是免费软件
+* 您可以根据自由软件基金会发布的 GPL（GNU General Public License，即 GNU通用公共许可证）的条款
+* 即 GPL 的第3版（即 GPL3.0）或（您选择的）任何后来的版本，重新发布和/或修改它
+* 
+* 本开源库的发布是希望它能发挥作用，但并未对其作任何的保证
+* 甚至没有隐含的适销性或适合特定用途的保证
+* 更多细节请参见 GPL
+* 
+* 您应该在收到本开源库的同时收到一份 GPL 的副本
+* 如果没有，请参阅<https://www.gnu.org/licenses/>
+* 
+* 额外注明：
+* 本开源库使用 GPL3.0 开源许可证协议 以上许可申明为译文版本
+* 许可申明英文版在 libraries/doc 文件夹下的 GPL3_permission_statement.txt 文件中
+* 许可证副本在 libraries 文件夹下 即该文件夹下的 LICENSE 文件
+* 欢迎各位使用并传播本程序 但修改内容时必须保留逐飞科技的版权声明（即本声明）
+* 
+* 文件名称          seekfree_assistant_interface
+* 公司名称          成都逐飞科技有限公司
+* 版本信息          查看 libraries/doc 文件夹内 version 文件 版本说明
+* 开发环境          MDK 5.37
+* 适用平台          MSPM0G3507
+* 店铺链接          https://seekfree.taobao.com/
+* 
+* 修改记录
+* 日期             作者             备注
+* 2024-1-11        SeekFree         first version
+********************************************************************************************************************/
+
+#include "zf_common_typedef.h"
+#include "zf_common_fifo.h"
+#include "zf_common_debug.h"
+#include "zf_driver_uart.h"
+#include "zf_device_wireless_uart.h"
+#include "zf_device_bluetooth_ch9141.h"
+#include "zf_device_wifi_uart.h"
+#include "zf_device_wifi_spi.h"
+#include "seekfree_assistant.h"
+
+#include "seekfree_assistant_interface.h"
+
+
+extern seekfree_assistant_transfer_callback_function   seekfree_assistant_transfer_callback;    // 数据发送函数指针
+extern seekfree_assistant_receive_callback_function    seekfree_assistant_receive_callback;     // 数据接收函数指针
+
+
+//-------------------------------------------------------------------------------------------------------------------
+// 函数简介     逐飞助手发送函数
+// 参数说明     *buff           需要发送的数据地址
+// 参数说明     length          需要发送的长度
+// 返回参数     uint32          剩余未发送数据长度
+// 使用示例
+//-------------------------------------------------------------------------------------------------------------------
+ZF_WEAK uint32 seekfree_assistant_transfer (const uint8 *buff, uint32 length)
+{
+    
+    // 当选择自定义通讯方式时 需要自行完成数据发送功能
+    return length;
+}
+
+//-------------------------------------------------------------------------------------------------------------------
+// 函数简介     逐飞助手接收数据函数
+// 参数说明     *buff           需要接收的数据地址
+// 参数说明     length          要接收的数据最大长度
+// 返回参数     uint32          接收到的数据长度
+// 使用示例
+//-------------------------------------------------------------------------------------------------------------------
+ZF_WEAK uint32 seekfree_assistant_receive (uint8 *buff, uint32 length)
+{
+    // 当选择自定义通讯方式时 需要自行完成数据接收功能
+    return 0;
+}
+
+//-------------------------------------------------------------------------------------------------------------------
+// 函数简介     逐飞助手接口 初始化
+// 参数说明		transfer_device		逐飞助手接口结构体
+// 返回参数     void
+// 使用示例     seekfree_assistant_interface_init(SEEKFREE_ASSISTANT_WIFI_SPI); 使用高速WIFI SPI模块进行数据收发
+// 备注         需要自行调用设备的初始化，例如使用无线转串口进行数据的收发，则需要自行调用无线转串口的初始化，然后再调用seekfree_assistant_interface_init完成逐飞助手的接口初始化
+//-------------------------------------------------------------------------------------------------------------------
+ZF_WEAK void seekfree_assistant_interface_init (seekfree_assistant_transfer_device_enum transfer_device)
+{
+    switch(transfer_device)
+    {
+        case SEEKFREE_ASSISTANT_DEBUG_UART:
+        {
+            seekfree_assistant_transfer_callback = debug_send_buffer;
+            seekfree_assistant_receive_callback = debug_read_ring_buffer;
+        }break;
+        
+        case SEEKFREE_ASSISTANT_WIRELESS_UART:
+        {
+            seekfree_assistant_transfer_callback = wireless_uart_send_buffer;
+            seekfree_assistant_receive_callback = wireless_uart_read_buffer;
+        }break;
+        
+        case SEEKFREE_ASSISTANT_CH9141:
+        {
+            seekfree_assistant_transfer_callback = bluetooth_ch9141_send_buffer;
+            seekfree_assistant_receive_callback = bluetooth_ch9141_read_buffer;
+        }break;
+        
+        case SEEKFREE_ASSISTANT_WIFI_UART:
+        {
+            seekfree_assistant_transfer_callback = wifi_uart_send_buffer;
+            seekfree_assistant_receive_callback = wifi_uart_read_buffer;
+        }break;
+        
+        case SEEKFREE_ASSISTANT_WIFI_SPI:
+        {
+            seekfree_assistant_transfer_callback = wifi_spi_send_buffer;
+            seekfree_assistant_receive_callback = wifi_spi_read_buffer;
+        }break;
+        
+        case SEEKFREE_ASSISTANT_CUSTOM:
+        {         
+            // 根据自己的需求 自行实现seekfree_assistant_transfer与seekfree_assistant_receive函数，完成数据的收发
+            
+        }break;
+    }
+}
+
+

libraries/components/seekfree_assistant/seekfree_assistant_interface.h

@@ -0,0 +1,57 @@
+/*********************************************************************************************************************
+* MSPM0G3507 Opensourec Library 即（MSPM0G3507 开源库）是一个基于官方 SDK 接口的第三方开源库
+* Copyright (c) 2022 SEEKFREE 逐飞科技
+* 
+* 本文件是 MSPM0G3507 开源库的一部分
+* 
+* MSPM0G3507 开源库 是免费软件
+* 您可以根据自由软件基金会发布的 GPL（GNU General Public License，即 GNU通用公共许可证）的条款
+* 即 GPL 的第3版（即 GPL3.0）或（您选择的）任何后来的版本，重新发布和/或修改它
+* 
+* 本开源库的发布是希望它能发挥作用，但并未对其作任何的保证
+* 甚至没有隐含的适销性或适合特定用途的保证
+* 更多细节请参见 GPL
+* 
+* 您应该在收到本开源库的同时收到一份 GPL 的副本
+* 如果没有，请参阅<https://www.gnu.org/licenses/>
+* 
+* 额外注明：
+* 本开源库使用 GPL3.0 开源许可证协议 以上许可申明为译文版本
+* 许可申明英文版在 libraries/doc 文件夹下的 GPL3_permission_statement.txt 文件中
+* 许可证副本在 libraries 文件夹下 即该文件夹下的 LICENSE 文件
+* 欢迎各位使用并传播本程序 但修改内容时必须保留逐飞科技的版权声明（即本声明）
+* 
+* 文件名称          seekfree_assistant_interface
+* 公司名称          成都逐飞科技有限公司
+* 版本信息          查看 libraries/doc 文件夹内 version 文件 版本说明
+* 开发环境          MDK 5.37
+* 适用平台          MSPM0G3507
+* 店铺链接          https://seekfree.taobao.com/
+* 
+* 修改记录
+* 日期             作者             备注
+* 2024-1-11        SeekFree         first version
+********************************************************************************************************************/
+
+#ifndef _seekfree_assistant_interface_h_
+#define _seekfree_assistant_interface_h_
+
+
+
+// 数据发送设备枚举
+typedef enum
+{
+    SEEKFREE_ASSISTANT_DEBUG_UART,      // 调试串口    使用的串口由DEBUG_UART_INDEX宏定义指定
+    SEEKFREE_ASSISTANT_WIRELESS_UART,   // 无线转串口
+    SEEKFREE_ASSISTANT_CH9141,          // CH9141蓝牙
+    SEEKFREE_ASSISTANT_WIFI_UART,       // WIFI转串口
+    SEEKFREE_ASSISTANT_WIFI_SPI,        // 高速WIFI SPI
+    SEEKFREE_ASSISTANT_CUSTOM,          // 自定义通讯方式 需要自行seekfree_assistant_transfer与seekfree_assistant_receive函数
+}seekfree_assistant_transfer_device_enum;
+
+
+void    seekfree_assistant_interface_init   (seekfree_assistant_transfer_device_enum transfer_device);
+
+
+
+#endif

libraries/doc/version.txt

@@ -0,0 +1,49 @@
+
+V3.2.0
+	增加IMU660RC驱动，支持四元数与欧拉角的获取
+
+V3.1.0
+	删除无线串口驱动里的wireless_uart_send_image
+	
+V3.0.9
+	修复N处注释错的地方
+	
+V3.0.8
+	修正660RA默认模式选择、修改为SPI模式
+
+V3.0.7
+    修正 zf_driver_exti 中数组初值错误
+    将栈空间扩增至 4K byte 大小
+    修改工程设置 默认仅擦除使用部分 清空残留的调试变量
+
+V3.0.6
+    修复 EXTI 过高频率触发导致的跳转异常
+
+V3.0.5
+    修复陀螺仪IIC模式初始化失败问题
+    烧录代码模式默认Any
+
+V3.0.4
+    解决PIT20ms以下存在误差问题
+
+V3.0.3
+    增加zf_device_key文件
+    更新按键例程
+    增加舵机例程
+
+V3.0.3
+    修复 UART3 不生效的问题
+
+V3.0.3
+    更新编码器例程 修改测试代码增加注释与可调宏定义
+
+V3.0.2
+    修复pit注释错误
+    增加编码器例程说明
+    主板例程包含关系修改为默认统一格式
+
+V3.0.1
+    解决963模式设置错误问题
+
+V3.0.0
+    初始版本

libraries/sdk/third_party/CMSIS/Core/Include/cachel1_armv7.h

@@ -0,0 +1,411 @@
+/******************************************************************************
+ * @file     cachel1_armv7.h
+ * @brief    CMSIS Level 1 Cache API for Armv7-M and later
+ * @version  V1.0.0
+ * @date     03. March 2020
+ ******************************************************************************/
+/*
+ * Copyright (c) 2020 Arm Limited. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#if   defined ( __ICCARM__ )
+  #pragma system_include         /* treat file as system include file for MISRA check */
+#elif defined (__clang__)
+  #pragma clang system_header    /* treat file as system include file */
+#endif
+
+#ifndef ARM_CACHEL1_ARMV7_H
+#define ARM_CACHEL1_ARMV7_H
+
+/**
+  \ingroup  CMSIS_Core_FunctionInterface
+  \defgroup CMSIS_Core_CacheFunctions Cache Functions
+  \brief    Functions that configure Instruction and Data cache.
+  @{
+ */
+
+/* Cache Size ID Register Macros */
+#define CCSIDR_WAYS(x)         (((x) & SCB_CCSIDR_ASSOCIATIVITY_Msk) >> SCB_CCSIDR_ASSOCIATIVITY_Pos)
+#define CCSIDR_SETS(x)         (((x) & SCB_CCSIDR_NUMSETS_Msk      ) >> SCB_CCSIDR_NUMSETS_Pos      )
+
+#ifndef __SCB_DCACHE_LINE_SIZE
+#define __SCB_DCACHE_LINE_SIZE  32U /*!< Cortex-M7 cache line size is fixed to 32 bytes (8 words). See also register SCB_CCSIDR */
+#endif
+
+#ifndef __SCB_ICACHE_LINE_SIZE
+#define __SCB_ICACHE_LINE_SIZE  32U /*!< Cortex-M7 cache line size is fixed to 32 bytes (8 words). See also register SCB_CCSIDR */
+#endif 
+
+/**
+  \brief   Enable I-Cache
+  \details Turns on I-Cache
+  */
+__STATIC_FORCEINLINE void SCB_EnableICache (void)
+{
+  #if defined (__ICACHE_PRESENT) && (__ICACHE_PRESENT == 1U)
+    if (SCB->CCR & SCB_CCR_IC_Msk) return;  /* return if ICache is already enabled */
+
+    __DSB();
+    __ISB();
+    SCB->ICIALLU = 0UL;                     /* invalidate I-Cache */
+    __DSB();
+    __ISB();
+    SCB->CCR |=  (uint32_t)SCB_CCR_IC_Msk;  /* enable I-Cache */
+    __DSB();
+    __ISB();
+  #endif
+}
+
+
+/**
+  \brief   Disable I-Cache
+  \details Turns off I-Cache
+  */
+__STATIC_FORCEINLINE void SCB_DisableICache (void)
+{
+  #if defined (__ICACHE_PRESENT) && (__ICACHE_PRESENT == 1U)
+    __DSB();
+    __ISB();
+    SCB->CCR &= ~(uint32_t)SCB_CCR_IC_Msk;  /* disable I-Cache */
+    SCB->ICIALLU = 0UL;                     /* invalidate I-Cache */
+    __DSB();
+    __ISB();
+  #endif
+}
+
+
+/**
+  \brief   Invalidate I-Cache
+  \details Invalidates I-Cache
+  */
+__STATIC_FORCEINLINE void SCB_InvalidateICache (void)
+{
+  #if defined (__ICACHE_PRESENT) && (__ICACHE_PRESENT == 1U)
+    __DSB();
+    __ISB();
+    SCB->ICIALLU = 0UL;
+    __DSB();
+    __ISB();
+  #endif
+}
+
+
+/**
+  \brief   I-Cache Invalidate by address
+  \details Invalidates I-Cache for the given address.
+           I-Cache is invalidated starting from a 32 byte aligned address in 32 byte granularity.
+           I-Cache memory blocks which are part of given address + given size are invalidated.
+  \param[in]   addr    address
+  \param[in]   isize   size of memory block (in number of bytes)
+*/
+__STATIC_FORCEINLINE void SCB_InvalidateICache_by_Addr (void *addr, int32_t isize)
+{
+  #if defined (__ICACHE_PRESENT) && (__ICACHE_PRESENT == 1U)
+    if ( isize > 0 ) {
+       int32_t op_size = isize + (((uint32_t)addr) & (__SCB_ICACHE_LINE_SIZE - 1U));
+      uint32_t op_addr = (uint32_t)addr /* & ~(__SCB_ICACHE_LINE_SIZE - 1U) */;
+
+      __DSB();
+
+      do {
+        SCB->ICIMVAU = op_addr;             /* register accepts only 32byte aligned values, only bits 31..5 are valid */
+        op_addr += __SCB_ICACHE_LINE_SIZE;
+        op_size -= __SCB_ICACHE_LINE_SIZE;
+      } while ( op_size > 0 );
+
+      __DSB();
+      __ISB();
+    }
+  #endif
+}
+
+
+/**
+  \brief   Enable D-Cache
+  \details Turns on D-Cache
+  */
+__STATIC_FORCEINLINE void SCB_EnableDCache (void)
+{
+  #if defined (__DCACHE_PRESENT) && (__DCACHE_PRESENT == 1U)
+    uint32_t ccsidr;
+    uint32_t sets;
+    uint32_t ways;
+
+    if (SCB->CCR & SCB_CCR_DC_Msk) return;  /* return if DCache is already enabled */
+
+    SCB->CSSELR = 0U;                       /* select Level 1 data cache */
+    __DSB();
+
+    ccsidr = SCB->CCSIDR;
+
+                                            /* invalidate D-Cache */
+    sets = (uint32_t)(CCSIDR_SETS(ccsidr));
+    do {
+      ways = (uint32_t)(CCSIDR_WAYS(ccsidr));
+      do {
+        SCB->DCISW = (((sets << SCB_DCISW_SET_Pos) & SCB_DCISW_SET_Msk) |
+                      ((ways << SCB_DCISW_WAY_Pos) & SCB_DCISW_WAY_Msk)  );
+        #if defined ( __CC_ARM )
+          __schedule_barrier();
+        #endif
+      } while (ways-- != 0U);
+    } while(sets-- != 0U);
+    __DSB();
+
+    SCB->CCR |=  (uint32_t)SCB_CCR_DC_Msk;  /* enable D-Cache */
+
+    __DSB();
+    __ISB();
+  #endif
+}
+
+
+/**
+  \brief   Disable D-Cache
+  \details Turns off D-Cache
+  */
+__STATIC_FORCEINLINE void SCB_DisableDCache (void)
+{
+  #if defined (__DCACHE_PRESENT) && (__DCACHE_PRESENT == 1U)
+    uint32_t ccsidr;
+    uint32_t sets;
+    uint32_t ways;
+
+    SCB->CSSELR = 0U;                       /* select Level 1 data cache */
+    __DSB();
+
+    SCB->CCR &= ~(uint32_t)SCB_CCR_DC_Msk;  /* disable D-Cache */
+    __DSB();
+
+    ccsidr = SCB->CCSIDR;
+
+                                            /* clean & invalidate D-Cache */
+    sets = (uint32_t)(CCSIDR_SETS(ccsidr));
+    do {
+      ways = (uint32_t)(CCSIDR_WAYS(ccsidr));
+      do {
+        SCB->DCCISW = (((sets << SCB_DCCISW_SET_Pos) & SCB_DCCISW_SET_Msk) |
+                       ((ways << SCB_DCCISW_WAY_Pos) & SCB_DCCISW_WAY_Msk)  );
+        #if defined ( __CC_ARM )
+          __schedule_barrier();
+        #endif
+      } while (ways-- != 0U);
+    } while(sets-- != 0U);
+
+    __DSB();
+    __ISB();
+  #endif
+}
+
+
+/**
+  \brief   Invalidate D-Cache
+  \details Invalidates D-Cache
+  */
+__STATIC_FORCEINLINE void SCB_InvalidateDCache (void)
+{
+  #if defined (__DCACHE_PRESENT) && (__DCACHE_PRESENT == 1U)
+    uint32_t ccsidr;
+    uint32_t sets;
+    uint32_t ways;
+
+    SCB->CSSELR = 0U;                       /* select Level 1 data cache */
+    __DSB();
+
+    ccsidr = SCB->CCSIDR;
+
+                                            /* invalidate D-Cache */
+    sets = (uint32_t)(CCSIDR_SETS(ccsidr));
+    do {
+      ways = (uint32_t)(CCSIDR_WAYS(ccsidr));
+      do {
+        SCB->DCISW = (((sets << SCB_DCISW_SET_Pos) & SCB_DCISW_SET_Msk) |
+                      ((ways << SCB_DCISW_WAY_Pos) & SCB_DCISW_WAY_Msk)  );
+        #if defined ( __CC_ARM )
+          __schedule_barrier();
+        #endif
+      } while (ways-- != 0U);
+    } while(sets-- != 0U);
+
+    __DSB();
+    __ISB();
+  #endif
+}
+
+
+/**
+  \brief   Clean D-Cache
+  \details Cleans D-Cache
+  */
+__STATIC_FORCEINLINE void SCB_CleanDCache (void)
+{
+  #if defined (__DCACHE_PRESENT) && (__DCACHE_PRESENT == 1U)
+    uint32_t ccsidr;
+    uint32_t sets;
+    uint32_t ways;
+
+    SCB->CSSELR = 0U;                       /* select Level 1 data cache */
+    __DSB();
+
+    ccsidr = SCB->CCSIDR;
+
+                                            /* clean D-Cache */
+    sets = (uint32_t)(CCSIDR_SETS(ccsidr));
+    do {
+      ways = (uint32_t)(CCSIDR_WAYS(ccsidr));
+      do {
+        SCB->DCCSW = (((sets << SCB_DCCSW_SET_Pos) & SCB_DCCSW_SET_Msk) |
+                      ((ways << SCB_DCCSW_WAY_Pos) & SCB_DCCSW_WAY_Msk)  );
+        #if defined ( __CC_ARM )
+          __schedule_barrier();
+        #endif
+      } while (ways-- != 0U);
+    } while(sets-- != 0U);
+
+    __DSB();
+    __ISB();
+  #endif
+}
+
+
+/**
+  \brief   Clean & Invalidate D-Cache
+  \details Cleans and Invalidates D-Cache
+  */
+__STATIC_FORCEINLINE void SCB_CleanInvalidateDCache (void)
+{
+  #if defined (__DCACHE_PRESENT) && (__DCACHE_PRESENT == 1U)
+    uint32_t ccsidr;
+    uint32_t sets;
+    uint32_t ways;
+
+    SCB->CSSELR = 0U;                       /* select Level 1 data cache */
+    __DSB();
+
+    ccsidr = SCB->CCSIDR;
+
+                                            /* clean & invalidate D-Cache */
+    sets = (uint32_t)(CCSIDR_SETS(ccsidr));
+    do {
+      ways = (uint32_t)(CCSIDR_WAYS(ccsidr));
+      do {
+        SCB->DCCISW = (((sets << SCB_DCCISW_SET_Pos) & SCB_DCCISW_SET_Msk) |
+                       ((ways << SCB_DCCISW_WAY_Pos) & SCB_DCCISW_WAY_Msk)  );
+        #if defined ( __CC_ARM )
+          __schedule_barrier();
+        #endif
+      } while (ways-- != 0U);
+    } while(sets-- != 0U);
+
+    __DSB();
+    __ISB();
+  #endif
+}
+
+
+/**
+  \brief   D-Cache Invalidate by address
+  \details Invalidates D-Cache for the given address.
+           D-Cache is invalidated starting from a 32 byte aligned address in 32 byte granularity.
+           D-Cache memory blocks which are part of given address + given size are invalidated.
+  \param[in]   addr    address
+  \param[in]   dsize   size of memory block (in number of bytes)
+*/
+__STATIC_FORCEINLINE void SCB_InvalidateDCache_by_Addr (void *addr, int32_t dsize)
+{
+  #if defined (__DCACHE_PRESENT) && (__DCACHE_PRESENT == 1U)
+    if ( dsize > 0 ) { 
+       int32_t op_size = dsize + (((uint32_t)addr) & (__SCB_DCACHE_LINE_SIZE - 1U));
+      uint32_t op_addr = (uint32_t)addr /* & ~(__SCB_DCACHE_LINE_SIZE - 1U) */;
+    
+      __DSB();
+
+      do {
+        SCB->DCIMVAC = op_addr;             /* register accepts only 32byte aligned values, only bits 31..5 are valid */
+        op_addr += __SCB_DCACHE_LINE_SIZE;
+        op_size -= __SCB_DCACHE_LINE_SIZE;
+      } while ( op_size > 0 );
+
+      __DSB();
+      __ISB();
+    }
+  #endif
+}
+
+
+/**
+  \brief   D-Cache Clean by address
+  \details Cleans D-Cache for the given address
+           D-Cache is cleaned starting from a 32 byte aligned address in 32 byte granularity.
+           D-Cache memory blocks which are part of given address + given size are cleaned.
+  \param[in]   addr    address
+  \param[in]   dsize   size of memory block (in number of bytes)
+*/
+__STATIC_FORCEINLINE void SCB_CleanDCache_by_Addr (uint32_t *addr, int32_t dsize)
+{
+  #if defined (__DCACHE_PRESENT) && (__DCACHE_PRESENT == 1U)
+    if ( dsize > 0 ) { 
+       int32_t op_size = dsize + (((uint32_t)addr) & (__SCB_DCACHE_LINE_SIZE - 1U));
+      uint32_t op_addr = (uint32_t)addr /* & ~(__SCB_DCACHE_LINE_SIZE - 1U) */;
+    
+      __DSB();
+
+      do {
+        SCB->DCCMVAC = op_addr;             /* register accepts only 32byte aligned values, only bits 31..5 are valid */
+        op_addr += __SCB_DCACHE_LINE_SIZE;
+        op_size -= __SCB_DCACHE_LINE_SIZE;
+      } while ( op_size > 0 );
+
+      __DSB();
+      __ISB();
+    }
+  #endif
+}
+
+
+/**
+  \brief   D-Cache Clean and Invalidate by address
+  \details Cleans and invalidates D_Cache for the given address
+           D-Cache is cleaned and invalidated starting from a 32 byte aligned address in 32 byte granularity.
+           D-Cache memory blocks which are part of given address + given size are cleaned and invalidated.
+  \param[in]   addr    address (aligned to 32-byte boundary)
+  \param[in]   dsize   size of memory block (in number of bytes)
+*/
+__STATIC_FORCEINLINE void SCB_CleanInvalidateDCache_by_Addr (uint32_t *addr, int32_t dsize)
+{
+  #if defined (__DCACHE_PRESENT) && (__DCACHE_PRESENT == 1U)
+    if ( dsize > 0 ) { 
+       int32_t op_size = dsize + (((uint32_t)addr) & (__SCB_DCACHE_LINE_SIZE - 1U));
+      uint32_t op_addr = (uint32_t)addr /* & ~(__SCB_DCACHE_LINE_SIZE - 1U) */;
+    
+      __DSB();
+
+      do {
+        SCB->DCCIMVAC = op_addr;            /* register accepts only 32byte aligned values, only bits 31..5 are valid */
+        op_addr +=          __SCB_DCACHE_LINE_SIZE;
+        op_size -=          __SCB_DCACHE_LINE_SIZE;
+      } while ( op_size > 0 );
+
+      __DSB();
+      __ISB();
+    }
+  #endif
+}
+
+/*@} end of CMSIS_Core_CacheFunctions */
+
+#endif /* ARM_CACHEL1_ARMV7_H */

libraries/sdk/third_party/CMSIS/Core/Include/cmsis_armcc.h

@@ -0,0 +1,885 @@
+/**************************************************************************//**
+ * @file     cmsis_armcc.h
+ * @brief    CMSIS compiler ARMCC (Arm Compiler 5) header file
+ * @version  V5.2.1
+ * @date     26. March 2020
+ ******************************************************************************/
+/*
+ * Copyright (c) 2009-2020 Arm Limited. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef __CMSIS_ARMCC_H
+#define __CMSIS_ARMCC_H
+
+
+#if defined(__ARMCC_VERSION) && (__ARMCC_VERSION < 400677)
+  #error "Please use Arm Compiler Toolchain V4.0.677 or later!"
+#endif
+
+/* CMSIS compiler control architecture macros */
+#if ((defined (__TARGET_ARCH_6_M  ) && (__TARGET_ARCH_6_M   == 1)) || \
+     (defined (__TARGET_ARCH_6S_M ) && (__TARGET_ARCH_6S_M  == 1))   )
+  #define __ARM_ARCH_6M__           1
+#endif
+
+#if (defined (__TARGET_ARCH_7_M ) && (__TARGET_ARCH_7_M  == 1))
+  #define __ARM_ARCH_7M__           1
+#endif
+
+#if (defined (__TARGET_ARCH_7E_M) && (__TARGET_ARCH_7E_M == 1))
+  #define __ARM_ARCH_7EM__          1
+#endif
+
+  /* __ARM_ARCH_8M_BASE__  not applicable */
+  /* __ARM_ARCH_8M_MAIN__  not applicable */
+  /* __ARM_ARCH_8_1M_MAIN__  not applicable */
+
+/* CMSIS compiler control DSP macros */
+#if ((defined (__ARM_ARCH_7EM__) && (__ARM_ARCH_7EM__ == 1))     )
+  #define __ARM_FEATURE_DSP         1
+#endif
+
+/* CMSIS compiler specific defines */
+#ifndef   __ASM
+  #define __ASM                                  __asm
+#endif
+#ifndef   __INLINE
+  #define __INLINE                               __inline
+#endif
+#ifndef   __STATIC_INLINE
+  #define __STATIC_INLINE                        static __inline
+#endif
+#ifndef   __STATIC_FORCEINLINE                 
+  #define __STATIC_FORCEINLINE                   static __forceinline
+#endif           
+#ifndef   __NO_RETURN
+  #define __NO_RETURN                            __declspec(noreturn)
+#endif
+#ifndef   __USED
+  #define __USED                                 __attribute__((used))
+#endif
+#ifndef   __WEAK
+  #define __WEAK                                 __attribute__((weak))
+#endif
+#ifndef   __PACKED
+  #define __PACKED                               __attribute__((packed))
+#endif
+#ifndef   __PACKED_STRUCT
+  #define __PACKED_STRUCT                        __packed struct
+#endif
+#ifndef   __PACKED_UNION
+  #define __PACKED_UNION                         __packed union
+#endif
+#ifndef   __UNALIGNED_UINT32        /* deprecated */
+  #define __UNALIGNED_UINT32(x)                  (*((__packed uint32_t *)(x)))
+#endif
+#ifndef   __UNALIGNED_UINT16_WRITE
+  #define __UNALIGNED_UINT16_WRITE(addr, val)    ((*((__packed uint16_t *)(addr))) = (val))
+#endif
+#ifndef   __UNALIGNED_UINT16_READ
+  #define __UNALIGNED_UINT16_READ(addr)          (*((const __packed uint16_t *)(addr)))
+#endif
+#ifndef   __UNALIGNED_UINT32_WRITE
+  #define __UNALIGNED_UINT32_WRITE(addr, val)    ((*((__packed uint32_t *)(addr))) = (val))
+#endif
+#ifndef   __UNALIGNED_UINT32_READ
+  #define __UNALIGNED_UINT32_READ(addr)          (*((const __packed uint32_t *)(addr)))
+#endif
+#ifndef   __ALIGNED
+  #define __ALIGNED(x)                           __attribute__((aligned(x)))
+#endif
+#ifndef   __RESTRICT
+  #define __RESTRICT                             __restrict
+#endif
+#ifndef   __COMPILER_BARRIER
+  #define __COMPILER_BARRIER()                   __memory_changed()
+#endif
+
+/* #########################  Startup and Lowlevel Init  ######################## */
+
+#ifndef __PROGRAM_START
+#define __PROGRAM_START           __main
+#endif
+
+#ifndef __INITIAL_SP
+#define __INITIAL_SP              Image$$ARM_LIB_STACK$$ZI$$Limit
+#endif
+
+#ifndef __STACK_LIMIT
+#define __STACK_LIMIT             Image$$ARM_LIB_STACK$$ZI$$Base
+#endif
+
+#ifndef __VECTOR_TABLE
+#define __VECTOR_TABLE            __Vectors
+#endif
+
+#ifndef __VECTOR_TABLE_ATTRIBUTE
+#define __VECTOR_TABLE_ATTRIBUTE  __attribute__((used, section("RESET")))
+#endif
+
+/* ###########################  Core Function Access  ########################### */
+/** \ingroup  CMSIS_Core_FunctionInterface
+    \defgroup CMSIS_Core_RegAccFunctions CMSIS Core Register Access Functions
+  @{
+ */
+
+/**
+  \brief   Enable IRQ Interrupts
+  \details Enables IRQ interrupts by clearing the I-bit in the CPSR.
+           Can only be executed in Privileged modes.
+ */
+/* intrinsic void __enable_irq();     */
+
+
+/**
+  \brief   Disable IRQ Interrupts
+  \details Disables IRQ interrupts by setting the I-bit in the CPSR.
+           Can only be executed in Privileged modes.
+ */
+/* intrinsic void __disable_irq();    */
+
+/**
+  \brief   Get Control Register
+  \details Returns the content of the Control Register.
+  \return               Control Register value
+ */
+__STATIC_INLINE uint32_t __get_CONTROL(void)
+{
+  register uint32_t __regControl         __ASM("control");
+  return(__regControl);
+}
+
+
+/**
+  \brief   Set Control Register
+  \details Writes the given value to the Control Register.
+  \param [in]    control  Control Register value to set
+ */
+__STATIC_INLINE void __set_CONTROL(uint32_t control)
+{
+  register uint32_t __regControl         __ASM("control");
+  __regControl = control;
+}
+
+
+/**
+  \brief   Get IPSR Register
+  \details Returns the content of the IPSR Register.
+  \return               IPSR Register value
+ */
+__STATIC_INLINE uint32_t __get_IPSR(void)
+{
+  register uint32_t __regIPSR          __ASM("ipsr");
+  return(__regIPSR);
+}
+
+
+/**
+  \brief   Get APSR Register
+  \details Returns the content of the APSR Register.
+  \return               APSR Register value
+ */
+__STATIC_INLINE uint32_t __get_APSR(void)
+{
+  register uint32_t __regAPSR          __ASM("apsr");
+  return(__regAPSR);
+}
+
+
+/**
+  \brief   Get xPSR Register
+  \details Returns the content of the xPSR Register.
+  \return               xPSR Register value
+ */
+__STATIC_INLINE uint32_t __get_xPSR(void)
+{
+  register uint32_t __regXPSR          __ASM("xpsr");
+  return(__regXPSR);
+}
+
+
+/**
+  \brief   Get Process Stack Pointer
+  \details Returns the current value of the Process Stack Pointer (PSP).
+  \return               PSP Register value
+ */
+__STATIC_INLINE uint32_t __get_PSP(void)
+{
+  register uint32_t __regProcessStackPointer  __ASM("psp");
+  return(__regProcessStackPointer);
+}
+
+
+/**
+  \brief   Set Process Stack Pointer
+  \details Assigns the given value to the Process Stack Pointer (PSP).
+  \param [in]    topOfProcStack  Process Stack Pointer value to set
+ */
+__STATIC_INLINE void __set_PSP(uint32_t topOfProcStack)
+{
+  register uint32_t __regProcessStackPointer  __ASM("psp");
+  __regProcessStackPointer = topOfProcStack;
+}
+
+
+/**
+  \brief   Get Main Stack Pointer
+  \details Returns the current value of the Main Stack Pointer (MSP).
+  \return               MSP Register value
+ */
+__STATIC_INLINE uint32_t __get_MSP(void)
+{
+  register uint32_t __regMainStackPointer     __ASM("msp");
+  return(__regMainStackPointer);
+}
+
+
+/**
+  \brief   Set Main Stack Pointer
+  \details Assigns the given value to the Main Stack Pointer (MSP).
+  \param [in]    topOfMainStack  Main Stack Pointer value to set
+ */
+__STATIC_INLINE void __set_MSP(uint32_t topOfMainStack)
+{
+  register uint32_t __regMainStackPointer     __ASM("msp");
+  __regMainStackPointer = topOfMainStack;
+}
+
+
+/**
+  \brief   Get Priority Mask
+  \details Returns the current state of the priority mask bit from the Priority Mask Register.
+  \return               Priority Mask value
+ */
+__STATIC_INLINE uint32_t __get_PRIMASK(void)
+{
+  register uint32_t __regPriMask         __ASM("primask");
+  return(__regPriMask);
+}
+
+
+/**
+  \brief   Set Priority Mask
+  \details Assigns the given value to the Priority Mask Register.
+  \param [in]    priMask  Priority Mask
+ */
+__STATIC_INLINE void __set_PRIMASK(uint32_t priMask)
+{
+  register uint32_t __regPriMask         __ASM("primask");
+  __regPriMask = (priMask);
+}
+
+
+#if ((defined (__ARM_ARCH_7M__ ) && (__ARM_ARCH_7M__  == 1)) || \
+     (defined (__ARM_ARCH_7EM__) && (__ARM_ARCH_7EM__ == 1))     )
+
+/**
+  \brief   Enable FIQ
+  \details Enables FIQ interrupts by clearing the F-bit in the CPSR.
+           Can only be executed in Privileged modes.
+ */
+#define __enable_fault_irq                __enable_fiq
+
+
+/**
+  \brief   Disable FIQ
+  \details Disables FIQ interrupts by setting the F-bit in the CPSR.
+           Can only be executed in Privileged modes.
+ */
+#define __disable_fault_irq               __disable_fiq
+
+
+/**
+  \brief   Get Base Priority
+  \details Returns the current value of the Base Priority register.
+  \return               Base Priority register value
+ */
+__STATIC_INLINE uint32_t  __get_BASEPRI(void)
+{
+  register uint32_t __regBasePri         __ASM("basepri");
+  return(__regBasePri);
+}
+
+
+/**
+  \brief   Set Base Priority
+  \details Assigns the given value to the Base Priority register.
+  \param [in]    basePri  Base Priority value to set
+ */
+__STATIC_INLINE void __set_BASEPRI(uint32_t basePri)
+{
+  register uint32_t __regBasePri         __ASM("basepri");
+  __regBasePri = (basePri & 0xFFU);
+}
+
+
+/**
+  \brief   Set Base Priority with condition
+  \details Assigns the given value to the Base Priority register only if BASEPRI masking is disabled,
+           or the new value increases the BASEPRI priority level.
+  \param [in]    basePri  Base Priority value to set
+ */
+__STATIC_INLINE void __set_BASEPRI_MAX(uint32_t basePri)
+{
+  register uint32_t __regBasePriMax      __ASM("basepri_max");
+  __regBasePriMax = (basePri & 0xFFU);
+}
+
+
+/**
+  \brief   Get Fault Mask
+  \details Returns the current value of the Fault Mask register.
+  \return               Fault Mask register value
+ */
+__STATIC_INLINE uint32_t __get_FAULTMASK(void)
+{
+  register uint32_t __regFaultMask       __ASM("faultmask");
+  return(__regFaultMask);
+}
+
+
+/**
+  \brief   Set Fault Mask
+  \details Assigns the given value to the Fault Mask register.
+  \param [in]    faultMask  Fault Mask value to set
+ */
+__STATIC_INLINE void __set_FAULTMASK(uint32_t faultMask)
+{
+  register uint32_t __regFaultMask       __ASM("faultmask");
+  __regFaultMask = (faultMask & (uint32_t)1U);
+}
+
+#endif /* ((defined (__ARM_ARCH_7M__ ) && (__ARM_ARCH_7M__  == 1)) || \
+           (defined (__ARM_ARCH_7EM__) && (__ARM_ARCH_7EM__ == 1))     ) */
+
+
+/**
+  \brief   Get FPSCR
+  \details Returns the current value of the Floating Point Status/Control register.
+  \return               Floating Point Status/Control register value
+ */
+__STATIC_INLINE uint32_t __get_FPSCR(void)
+{
+#if ((defined (__FPU_PRESENT) && (__FPU_PRESENT == 1U)) && \
+     (defined (__FPU_USED   ) && (__FPU_USED    == 1U))     )
+  register uint32_t __regfpscr         __ASM("fpscr");
+  return(__regfpscr);
+#else
+   return(0U);
+#endif
+}
+
+
+/**
+  \brief   Set FPSCR
+  \details Assigns the given value to the Floating Point Status/Control register.
+  \param [in]    fpscr  Floating Point Status/Control value to set
+ */
+__STATIC_INLINE void __set_FPSCR(uint32_t fpscr)
+{
+#if ((defined (__FPU_PRESENT) && (__FPU_PRESENT == 1U)) && \
+     (defined (__FPU_USED   ) && (__FPU_USED    == 1U))     )
+  register uint32_t __regfpscr         __ASM("fpscr");
+  __regfpscr = (fpscr);
+#else
+  (void)fpscr;
+#endif
+}
+
+
+/*@} end of CMSIS_Core_RegAccFunctions */
+
+
+/* ##########################  Core Instruction Access  ######################### */
+/** \defgroup CMSIS_Core_InstructionInterface CMSIS Core Instruction Interface
+  Access to dedicated instructions
+  @{
+*/
+
+/**
+  \brief   No Operation
+  \details No Operation does nothing. This instruction can be used for code alignment purposes.
+ */
+#define __NOP                             __nop
+
+
+/**
+  \brief   Wait For Interrupt
+  \details Wait For Interrupt is a hint instruction that suspends execution until one of a number of events occurs.
+ */
+#define __WFI                             __wfi
+
+
+/**
+  \brief   Wait For Event
+  \details Wait For Event is a hint instruction that permits the processor to enter
+           a low-power state until one of a number of events occurs.
+ */
+#define __WFE                             __wfe
+
+
+/**
+  \brief   Send Event
+  \details Send Event is a hint instruction. It causes an event to be signaled to the CPU.
+ */
+#define __SEV                             __sev
+
+
+/**
+  \brief   Instruction Synchronization Barrier
+  \details Instruction Synchronization Barrier flushes the pipeline in the processor,
+           so that all instructions following the ISB are fetched from cache or memory,
+           after the instruction has been completed.
+ */
+#define __ISB()                           __isb(0xF)
+
+/**
+  \brief   Data Synchronization Barrier
+  \details Acts as a special kind of Data Memory Barrier.
+           It completes when all explicit memory accesses before this instruction complete.
+ */
+#define __DSB()                           __dsb(0xF)
+
+/**
+  \brief   Data Memory Barrier
+  \details Ensures the apparent order of the explicit memory operations before
+           and after the instruction, without ensuring their completion.
+ */
+#define __DMB()                           __dmb(0xF)
+
+                  
+/**
+  \brief   Reverse byte order (32 bit)
+  \details Reverses the byte order in unsigned integer value. For example, 0x12345678 becomes 0x78563412.
+  \param [in]    value  Value to reverse
+  \return               Reversed value
+ */
+#define __REV                             __rev
+
+
+/**
+  \brief   Reverse byte order (16 bit)
+  \details Reverses the byte order within each halfword of a word. For example, 0x12345678 becomes 0x34127856.
+  \param [in]    value  Value to reverse
+  \return               Reversed value
+ */
+#ifndef __NO_EMBEDDED_ASM
+__attribute__((section(".rev16_text"))) __STATIC_INLINE __ASM uint32_t __REV16(uint32_t value)
+{
+  rev16 r0, r0
+  bx lr
+}
+#endif
+
+
+/**
+  \brief   Reverse byte order (16 bit)
+  \details Reverses the byte order in a 16-bit value and returns the signed 16-bit result. For example, 0x0080 becomes 0x8000.
+  \param [in]    value  Value to reverse
+  \return               Reversed value
+ */
+#ifndef __NO_EMBEDDED_ASM
+__attribute__((section(".revsh_text"))) __STATIC_INLINE __ASM int16_t __REVSH(int16_t value)
+{
+  revsh r0, r0
+  bx lr
+}
+#endif
+
+
+/**
+  \brief   Rotate Right in unsigned value (32 bit)
+  \details Rotate Right (immediate) provides the value of the contents of a register rotated by a variable number of bits.
+  \param [in]    op1  Value to rotate
+  \param [in]    op2  Number of Bits to rotate
+  \return               Rotated value
+ */
+#define __ROR                             __ror
+
+
+/**
+  \brief   Breakpoint
+  \details Causes the processor to enter Debug state.
+           Debug tools can use this to investigate system state when the instruction at a particular address is reached.
+  \param [in]    value  is ignored by the processor.
+                 If required, a debugger can use it to store additional information about the breakpoint.
+ */
+#define __BKPT(value)                       __breakpoint(value)
+
+
+/**
+  \brief   Reverse bit order of value
+  \details Reverses the bit order of the given value.
+  \param [in]    value  Value to reverse
+  \return               Reversed value
+ */
+#if ((defined (__ARM_ARCH_7M__ ) && (__ARM_ARCH_7M__  == 1)) || \
+     (defined (__ARM_ARCH_7EM__) && (__ARM_ARCH_7EM__ == 1))     )
+  #define __RBIT                          __rbit
+#else
+__attribute__((always_inline)) __STATIC_INLINE uint32_t __RBIT(uint32_t value)
+{
+  uint32_t result;
+  uint32_t s = (4U /*sizeof(v)*/ * 8U) - 1U; /* extra shift needed at end */
+
+  result = value;                      /* r will be reversed bits of v; first get LSB of v */
+  for (value >>= 1U; value != 0U; value >>= 1U)
+  {
+    result <<= 1U;
+    result |= value & 1U;
+    s--;
+  }
+  result <<= s;                        /* shift when v's highest bits are zero */
+  return result;
+}
+#endif
+
+
+/**
+  \brief   Count leading zeros
+  \details Counts the number of leading zeros of a data value.
+  \param [in]  value  Value to count the leading zeros
+  \return             number of leading zeros in value
+ */
+#define __CLZ                             __clz
+
+
+#if ((defined (__ARM_ARCH_7M__ ) && (__ARM_ARCH_7M__  == 1)) || \
+     (defined (__ARM_ARCH_7EM__) && (__ARM_ARCH_7EM__ == 1))     )
+
+/**
+  \brief   LDR Exclusive (8 bit)
+  \details Executes a exclusive LDR instruction for 8 bit value.
+  \param [in]    ptr  Pointer to data
+  \return             value of type uint8_t at (*ptr)
+ */
+#if defined(__ARMCC_VERSION) && (__ARMCC_VERSION < 5060020)
+  #define __LDREXB(ptr)                                                        ((uint8_t ) __ldrex(ptr))
+#else
+  #define __LDREXB(ptr)          _Pragma("push") _Pragma("diag_suppress 3731") ((uint8_t ) __ldrex(ptr))  _Pragma("pop")
+#endif
+
+
+/**
+  \brief   LDR Exclusive (16 bit)
+  \details Executes a exclusive LDR instruction for 16 bit values.
+  \param [in]    ptr  Pointer to data
+  \return        value of type uint16_t at (*ptr)
+ */
+#if defined(__ARMCC_VERSION) && (__ARMCC_VERSION < 5060020)
+  #define __LDREXH(ptr)                                                        ((uint16_t) __ldrex(ptr))
+#else
+  #define __LDREXH(ptr)          _Pragma("push") _Pragma("diag_suppress 3731") ((uint16_t) __ldrex(ptr))  _Pragma("pop")
+#endif
+
+
+/**
+  \brief   LDR Exclusive (32 bit)
+  \details Executes a exclusive LDR instruction for 32 bit values.
+  \param [in]    ptr  Pointer to data
+  \return        value of type uint32_t at (*ptr)
+ */
+#if defined(__ARMCC_VERSION) && (__ARMCC_VERSION < 5060020)
+  #define __LDREXW(ptr)                                                        ((uint32_t ) __ldrex(ptr))
+#else
+  #define __LDREXW(ptr)          _Pragma("push") _Pragma("diag_suppress 3731") ((uint32_t ) __ldrex(ptr))  _Pragma("pop")
+#endif
+
+
+/**
+  \brief   STR Exclusive (8 bit)
+  \details Executes a exclusive STR instruction for 8 bit values.
+  \param [in]  value  Value to store
+  \param [in]    ptr  Pointer to location
+  \return          0  Function succeeded
+  \return          1  Function failed
+ */
+#if defined(__ARMCC_VERSION) && (__ARMCC_VERSION < 5060020)
+  #define __STREXB(value, ptr)                                                 __strex(value, ptr)
+#else
+  #define __STREXB(value, ptr)   _Pragma("push") _Pragma("diag_suppress 3731") __strex(value, ptr)        _Pragma("pop")
+#endif
+
+
+/**
+  \brief   STR Exclusive (16 bit)
+  \details Executes a exclusive STR instruction for 16 bit values.
+  \param [in]  value  Value to store
+  \param [in]    ptr  Pointer to location
+  \return          0  Function succeeded
+  \return          1  Function failed
+ */
+#if defined(__ARMCC_VERSION) && (__ARMCC_VERSION < 5060020)
+  #define __STREXH(value, ptr)                                                 __strex(value, ptr)
+#else
+  #define __STREXH(value, ptr)   _Pragma("push") _Pragma("diag_suppress 3731") __strex(value, ptr)        _Pragma("pop")
+#endif
+
+
+/**
+  \brief   STR Exclusive (32 bit)
+  \details Executes a exclusive STR instruction for 32 bit values.
+  \param [in]  value  Value to store
+  \param [in]    ptr  Pointer to location
+  \return          0  Function succeeded
+  \return          1  Function failed
+ */
+#if defined(__ARMCC_VERSION) && (__ARMCC_VERSION < 5060020)
+  #define __STREXW(value, ptr)                                                 __strex(value, ptr)
+#else
+  #define __STREXW(value, ptr)   _Pragma("push") _Pragma("diag_suppress 3731") __strex(value, ptr)        _Pragma("pop")
+#endif
+
+
+/**
+  \brief   Remove the exclusive lock
+  \details Removes the exclusive lock which is created by LDREX.
+ */
+#define __CLREX                           __clrex
+
+
+/**
+  \brief   Signed Saturate
+  \details Saturates a signed value.
+  \param [in]  value  Value to be saturated
+  \param [in]    sat  Bit position to saturate to (1..32)
+  \return             Saturated value
+ */
+#define __SSAT                            __ssat
+
+
+/**
+  \brief   Unsigned Saturate
+  \details Saturates an unsigned value.
+  \param [in]  value  Value to be saturated
+  \param [in]    sat  Bit position to saturate to (0..31)
+  \return             Saturated value
+ */
+#define __USAT                            __usat
+
+
+/**
+  \brief   Rotate Right with Extend (32 bit)
+  \details Moves each bit of a bitstring right by one bit.
+           The carry input is shifted in at the left end of the bitstring.
+  \param [in]    value  Value to rotate
+  \return               Rotated value
+ */
+#ifndef __NO_EMBEDDED_ASM
+__attribute__((section(".rrx_text"))) __STATIC_INLINE __ASM uint32_t __RRX(uint32_t value)
+{
+  rrx r0, r0
+  bx lr
+}
+#endif
+
+
+/**
+  \brief   LDRT Unprivileged (8 bit)
+  \details Executes a Unprivileged LDRT instruction for 8 bit value.
+  \param [in]    ptr  Pointer to data
+  \return             value of type uint8_t at (*ptr)
+ */
+#define __LDRBT(ptr)                      ((uint8_t )  __ldrt(ptr))
+
+
+/**
+  \brief   LDRT Unprivileged (16 bit)
+  \details Executes a Unprivileged LDRT instruction for 16 bit values.
+  \param [in]    ptr  Pointer to data
+  \return        value of type uint16_t at (*ptr)
+ */
+#define __LDRHT(ptr)                      ((uint16_t)  __ldrt(ptr))
+
+
+/**
+  \brief   LDRT Unprivileged (32 bit)
+  \details Executes a Unprivileged LDRT instruction for 32 bit values.
+  \param [in]    ptr  Pointer to data
+  \return        value of type uint32_t at (*ptr)
+ */
+#define __LDRT(ptr)                       ((uint32_t ) __ldrt(ptr))
+
+
+/**
+  \brief   STRT Unprivileged (8 bit)
+  \details Executes a Unprivileged STRT instruction for 8 bit values.
+  \param [in]  value  Value to store
+  \param [in]    ptr  Pointer to location
+ */
+#define __STRBT(value, ptr)               __strt(value, ptr)
+
+
+/**
+  \brief   STRT Unprivileged (16 bit)
+  \details Executes a Unprivileged STRT instruction for 16 bit values.
+  \param [in]  value  Value to store
+  \param [in]    ptr  Pointer to location
+ */
+#define __STRHT(value, ptr)               __strt(value, ptr)
+
+
+/**
+  \brief   STRT Unprivileged (32 bit)
+  \details Executes a Unprivileged STRT instruction for 32 bit values.
+  \param [in]  value  Value to store
+  \param [in]    ptr  Pointer to location
+ */
+#define __STRT(value, ptr)                __strt(value, ptr)
+
+#else  /* ((defined (__ARM_ARCH_7M__ ) && (__ARM_ARCH_7M__  == 1)) || \
+           (defined (__ARM_ARCH_7EM__) && (__ARM_ARCH_7EM__ == 1))     ) */
+
+/**
+  \brief   Signed Saturate
+  \details Saturates a signed value.
+  \param [in]  value  Value to be saturated
+  \param [in]    sat  Bit position to saturate to (1..32)
+  \return             Saturated value
+ */
+__attribute__((always_inline)) __STATIC_INLINE int32_t __SSAT(int32_t val, uint32_t sat)
+{
+  if ((sat >= 1U) && (sat <= 32U))
+  {
+    const int32_t max = (int32_t)((1U << (sat - 1U)) - 1U);
+    const int32_t min = -1 - max ;
+    if (val > max)
+    {
+      return max;
+    }
+    else if (val < min)
+    {
+      return min;
+    }
+  }
+  return val;
+}
+
+/**
+  \brief   Unsigned Saturate
+  \details Saturates an unsigned value.
+  \param [in]  value  Value to be saturated
+  \param [in]    sat  Bit position to saturate to (0..31)
+  \return             Saturated value
+ */
+__attribute__((always_inline)) __STATIC_INLINE uint32_t __USAT(int32_t val, uint32_t sat)
+{
+  if (sat <= 31U)
+  {
+    const uint32_t max = ((1U << sat) - 1U);
+    if (val > (int32_t)max)
+    {
+      return max;
+    }
+    else if (val < 0)
+    {
+      return 0U;
+    }
+  }
+  return (uint32_t)val;
+}
+
+#endif /* ((defined (__ARM_ARCH_7M__ ) && (__ARM_ARCH_7M__  == 1)) || \
+           (defined (__ARM_ARCH_7EM__) && (__ARM_ARCH_7EM__ == 1))     ) */
+
+/*@}*/ /* end of group CMSIS_Core_InstructionInterface */
+
+
+/* ###################  Compiler specific Intrinsics  ########################### */
+/** \defgroup CMSIS_SIMD_intrinsics CMSIS SIMD Intrinsics
+  Access to dedicated SIMD instructions
+  @{
+*/
+
+#if ((defined (__ARM_ARCH_7EM__) && (__ARM_ARCH_7EM__ == 1))     )
+
+#define __SADD8                           __sadd8
+#define __QADD8                           __qadd8
+#define __SHADD8                          __shadd8
+#define __UADD8                           __uadd8
+#define __UQADD8                          __uqadd8
+#define __UHADD8                          __uhadd8
+#define __SSUB8                           __ssub8
+#define __QSUB8                           __qsub8
+#define __SHSUB8                          __shsub8
+#define __USUB8                           __usub8
+#define __UQSUB8                          __uqsub8
+#define __UHSUB8                          __uhsub8
+#define __SADD16                          __sadd16
+#define __QADD16                          __qadd16
+#define __SHADD16                         __shadd16
+#define __UADD16                          __uadd16
+#define __UQADD16                         __uqadd16
+#define __UHADD16                         __uhadd16
+#define __SSUB16                          __ssub16
+#define __QSUB16                          __qsub16
+#define __SHSUB16                         __shsub16
+#define __USUB16                          __usub16
+#define __UQSUB16                         __uqsub16
+#define __UHSUB16                         __uhsub16
+#define __SASX                            __sasx
+#define __QASX                            __qasx
+#define __SHASX                           __shasx
+#define __UASX                            __uasx
+#define __UQASX                           __uqasx
+#define __UHASX                           __uhasx
+#define __SSAX                            __ssax
+#define __QSAX                            __qsax
+#define __SHSAX                           __shsax
+#define __USAX                            __usax
+#define __UQSAX                           __uqsax
+#define __UHSAX                           __uhsax
+#define __USAD8                           __usad8
+#define __USADA8                          __usada8
+#define __SSAT16                          __ssat16
+#define __USAT16                          __usat16
+#define __UXTB16                          __uxtb16
+#define __UXTAB16                         __uxtab16
+#define __SXTB16                          __sxtb16
+#define __SXTAB16                         __sxtab16
+#define __SMUAD                           __smuad
+#define __SMUADX                          __smuadx
+#define __SMLAD                           __smlad
+#define __SMLADX                          __smladx
+#define __SMLALD                          __smlald
+#define __SMLALDX                         __smlaldx
+#define __SMUSD                           __smusd
+#define __SMUSDX                          __smusdx
+#define __SMLSD                           __smlsd
+#define __SMLSDX                          __smlsdx
+#define __SMLSLD                          __smlsld
+#define __SMLSLDX                         __smlsldx
+#define __SEL                             __sel
+#define __QADD                            __qadd
+#define __QSUB                            __qsub
+
+#define __PKHBT(ARG1,ARG2,ARG3)          ( ((((uint32_t)(ARG1))          ) & 0x0000FFFFUL) |  \
+                                           ((((uint32_t)(ARG2)) << (ARG3)) & 0xFFFF0000UL)  )
+
+#define __PKHTB(ARG1,ARG2,ARG3)          ( ((((uint32_t)(ARG1))          ) & 0xFFFF0000UL) |  \
+                                           ((((uint32_t)(ARG2)) >> (ARG3)) & 0x0000FFFFUL)  )
+
+#define __SMMLA(ARG1,ARG2,ARG3)          ( (int32_t)((((int64_t)(ARG1) * (ARG2)) + \
+                                                      ((int64_t)(ARG3) << 32U)     ) >> 32U))
+
+#define __SXTB16_RORn(ARG1, ARG2)        __SXTB16(__ROR(ARG1, ARG2))
+
+#endif /* ((defined (__ARM_ARCH_7EM__) && (__ARM_ARCH_7EM__ == 1))     ) */
+/*@} end of group CMSIS_SIMD_intrinsics */
+
+
+#endif /* __CMSIS_ARMCC_H */

libraries/sdk/third_party/CMSIS/Core/Include/cmsis_compiler.h

@@ -0,0 +1,283 @@
+/**************************************************************************//**
+ * @file     cmsis_compiler.h
+ * @brief    CMSIS compiler generic header file
+ * @version  V5.1.0
+ * @date     09. October 2018
+ ******************************************************************************/
+/*
+ * Copyright (c) 2009-2018 Arm Limited. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef __CMSIS_COMPILER_H
+#define __CMSIS_COMPILER_H
+
+#include <stdint.h>
+
+/*
+ * Arm Compiler 4/5
+ */
+#if   defined ( __CC_ARM )
+  #include "cmsis_armcc.h"
+
+
+/*
+ * Arm Compiler 6.6 LTM (armclang)
+ */
+#elif defined (__ARMCC_VERSION) && (__ARMCC_VERSION >= 6010050) && (__ARMCC_VERSION < 6100100)
+  #include "cmsis_armclang_ltm.h"
+
+  /*
+ * Arm Compiler above 6.10.1 (armclang)
+ */
+#elif defined (__ARMCC_VERSION) && (__ARMCC_VERSION >= 6100100)
+  #include "cmsis_armclang.h"
+
+
+/*
+ * GNU Compiler
+ */
+#elif defined ( __GNUC__ )
+  #include "cmsis_gcc.h"
+
+
+/*
+ * IAR Compiler
+ */
+#elif defined ( __ICCARM__ )
+  #include <cmsis_iccarm.h>
+
+
+/*
+ * TI Arm Compiler
+ */
+#elif defined ( __TI_ARM__ )
+  #include <cmsis_ccs.h>
+
+  #ifndef   __ASM
+    #define __ASM                                  __asm
+  #endif
+  #ifndef   __INLINE
+    #define __INLINE                               inline
+  #endif
+  #ifndef   __STATIC_INLINE
+    #define __STATIC_INLINE                        static inline
+  #endif
+  #ifndef   __STATIC_FORCEINLINE
+    #define __STATIC_FORCEINLINE                   __STATIC_INLINE
+  #endif
+  #ifndef   __NO_RETURN
+    #define __NO_RETURN                            __attribute__((noreturn))
+  #endif
+  #ifndef   __USED
+    #define __USED                                 __attribute__((used))
+  #endif
+  #ifndef   __WEAK
+    #define __WEAK                                 __attribute__((weak))
+  #endif
+  #ifndef   __PACKED
+    #define __PACKED                               __attribute__((packed))
+  #endif
+  #ifndef   __PACKED_STRUCT
+    #define __PACKED_STRUCT                        struct __attribute__((packed))
+  #endif
+  #ifndef   __PACKED_UNION
+    #define __PACKED_UNION                         union __attribute__((packed))
+  #endif
+  #ifndef   __UNALIGNED_UINT32        /* deprecated */
+    struct __attribute__((packed)) T_UINT32 { uint32_t v; };
+    #define __UNALIGNED_UINT32(x)                  (((struct T_UINT32 *)(x))->v)
+  #endif
+  #ifndef   __UNALIGNED_UINT16_WRITE
+    __PACKED_STRUCT T_UINT16_WRITE { uint16_t v; };
+    #define __UNALIGNED_UINT16_WRITE(addr, val)    (void)((((struct T_UINT16_WRITE *)(void*)(addr))->v) = (val))
+  #endif
+  #ifndef   __UNALIGNED_UINT16_READ
+    __PACKED_STRUCT T_UINT16_READ { uint16_t v; };
+    #define __UNALIGNED_UINT16_READ(addr)          (((const struct T_UINT16_READ *)(const void *)(addr))->v)
+  #endif
+  #ifndef   __UNALIGNED_UINT32_WRITE
+    __PACKED_STRUCT T_UINT32_WRITE { uint32_t v; };
+    #define __UNALIGNED_UINT32_WRITE(addr, val)    (void)((((struct T_UINT32_WRITE *)(void *)(addr))->v) = (val))
+  #endif
+  #ifndef   __UNALIGNED_UINT32_READ
+    __PACKED_STRUCT T_UINT32_READ { uint32_t v; };
+    #define __UNALIGNED_UINT32_READ(addr)          (((const struct T_UINT32_READ *)(const void *)(addr))->v)
+  #endif
+  #ifndef   __ALIGNED
+    #define __ALIGNED(x)                           __attribute__((aligned(x)))
+  #endif
+  #ifndef   __RESTRICT
+    #define __RESTRICT                             __restrict
+  #endif
+  #ifndef   __COMPILER_BARRIER
+    #warning No compiler specific solution for __COMPILER_BARRIER. __COMPILER_BARRIER is ignored.
+    #define __COMPILER_BARRIER()                   (void)0
+  #endif
+
+
+/*
+ * TASKING Compiler
+ */
+#elif defined ( __TASKING__ )
+  /*
+   * The CMSIS functions have been implemented as intrinsics in the compiler.
+   * Please use "carm -?i" to get an up to date list of all intrinsics,
+   * Including the CMSIS ones.
+   */
+
+  #ifndef   __ASM
+    #define __ASM                                  __asm
+  #endif
+  #ifndef   __INLINE
+    #define __INLINE                               inline
+  #endif
+  #ifndef   __STATIC_INLINE
+    #define __STATIC_INLINE                        static inline
+  #endif
+  #ifndef   __STATIC_FORCEINLINE
+    #define __STATIC_FORCEINLINE                   __STATIC_INLINE
+  #endif
+  #ifndef   __NO_RETURN
+    #define __NO_RETURN                            __attribute__((noreturn))
+  #endif
+  #ifndef   __USED
+    #define __USED                                 __attribute__((used))
+  #endif
+  #ifndef   __WEAK
+    #define __WEAK                                 __attribute__((weak))
+  #endif
+  #ifndef   __PACKED
+    #define __PACKED                               __packed__
+  #endif
+  #ifndef   __PACKED_STRUCT
+    #define __PACKED_STRUCT                        struct __packed__
+  #endif
+  #ifndef   __PACKED_UNION
+    #define __PACKED_UNION                         union __packed__
+  #endif
+  #ifndef   __UNALIGNED_UINT32        /* deprecated */
+    struct __packed__ T_UINT32 { uint32_t v; };
+    #define __UNALIGNED_UINT32(x)                  (((struct T_UINT32 *)(x))->v)
+  #endif
+  #ifndef   __UNALIGNED_UINT16_WRITE
+    __PACKED_STRUCT T_UINT16_WRITE { uint16_t v; };
+    #define __UNALIGNED_UINT16_WRITE(addr, val)    (void)((((struct T_UINT16_WRITE *)(void *)(addr))->v) = (val))
+  #endif
+  #ifndef   __UNALIGNED_UINT16_READ
+    __PACKED_STRUCT T_UINT16_READ { uint16_t v; };
+    #define __UNALIGNED_UINT16_READ(addr)          (((const struct T_UINT16_READ *)(const void *)(addr))->v)
+  #endif
+  #ifndef   __UNALIGNED_UINT32_WRITE
+    __PACKED_STRUCT T_UINT32_WRITE { uint32_t v; };
+    #define __UNALIGNED_UINT32_WRITE(addr, val)    (void)((((struct T_UINT32_WRITE *)(void *)(addr))->v) = (val))
+  #endif
+  #ifndef   __UNALIGNED_UINT32_READ
+    __PACKED_STRUCT T_UINT32_READ { uint32_t v; };
+    #define __UNALIGNED_UINT32_READ(addr)          (((const struct T_UINT32_READ *)(const void *)(addr))->v)
+  #endif
+  #ifndef   __ALIGNED
+    #define __ALIGNED(x)              __align(x)
+  #endif
+  #ifndef   __RESTRICT
+    #warning No compiler specific solution for __RESTRICT. __RESTRICT is ignored.
+    #define __RESTRICT
+  #endif
+  #ifndef   __COMPILER_BARRIER
+    #warning No compiler specific solution for __COMPILER_BARRIER. __COMPILER_BARRIER is ignored.
+    #define __COMPILER_BARRIER()                   (void)0
+  #endif
+
+
+/*
+ * COSMIC Compiler
+ */
+#elif defined ( __CSMC__ )
+   #include <cmsis_csm.h>
+
+ #ifndef   __ASM
+    #define __ASM                                  _asm
+  #endif
+  #ifndef   __INLINE
+    #define __INLINE                               inline
+  #endif
+  #ifndef   __STATIC_INLINE
+    #define __STATIC_INLINE                        static inline
+  #endif
+  #ifndef   __STATIC_FORCEINLINE
+    #define __STATIC_FORCEINLINE                   __STATIC_INLINE
+  #endif
+  #ifndef   __NO_RETURN
+    // NO RETURN is automatically detected hence no warning here
+    #define __NO_RETURN
+  #endif
+  #ifndef   __USED
+    #warning No compiler specific solution for __USED. __USED is ignored.
+    #define __USED
+  #endif
+  #ifndef   __WEAK
+    #define __WEAK                                 __weak
+  #endif
+  #ifndef   __PACKED
+    #define __PACKED                               @packed
+  #endif
+  #ifndef   __PACKED_STRUCT
+    #define __PACKED_STRUCT                        @packed struct
+  #endif
+  #ifndef   __PACKED_UNION
+    #define __PACKED_UNION                         @packed union
+  #endif
+  #ifndef   __UNALIGNED_UINT32        /* deprecated */
+    @packed struct T_UINT32 { uint32_t v; };
+    #define __UNALIGNED_UINT32(x)                  (((struct T_UINT32 *)(x))->v)
+  #endif
+  #ifndef   __UNALIGNED_UINT16_WRITE
+    __PACKED_STRUCT T_UINT16_WRITE { uint16_t v; };
+    #define __UNALIGNED_UINT16_WRITE(addr, val)    (void)((((struct T_UINT16_WRITE *)(void *)(addr))->v) = (val))
+  #endif
+  #ifndef   __UNALIGNED_UINT16_READ
+    __PACKED_STRUCT T_UINT16_READ { uint16_t v; };
+    #define __UNALIGNED_UINT16_READ(addr)          (((const struct T_UINT16_READ *)(const void *)(addr))->v)
+  #endif
+  #ifndef   __UNALIGNED_UINT32_WRITE
+    __PACKED_STRUCT T_UINT32_WRITE { uint32_t v; };
+    #define __UNALIGNED_UINT32_WRITE(addr, val)    (void)((((struct T_UINT32_WRITE *)(void *)(addr))->v) = (val))
+  #endif
+  #ifndef   __UNALIGNED_UINT32_READ
+    __PACKED_STRUCT T_UINT32_READ { uint32_t v; };
+    #define __UNALIGNED_UINT32_READ(addr)          (((const struct T_UINT32_READ *)(const void *)(addr))->v)
+  #endif
+  #ifndef   __ALIGNED
+    #warning No compiler specific solution for __ALIGNED. __ALIGNED is ignored.
+    #define __ALIGNED(x)
+  #endif
+  #ifndef   __RESTRICT
+    #warning No compiler specific solution for __RESTRICT. __RESTRICT is ignored.
+    #define __RESTRICT
+  #endif
+  #ifndef   __COMPILER_BARRIER
+    #warning No compiler specific solution for __COMPILER_BARRIER. __COMPILER_BARRIER is ignored.
+    #define __COMPILER_BARRIER()                   (void)0
+  #endif
+
+
+#else
+  #error Unknown compiler.
+#endif
+
+
+#endif /* __CMSIS_COMPILER_H */
+

libraries/sdk/third_party/CMSIS/Core/Include/cmsis_iccarm.h

@@ -0,0 +1,968 @@
+/**************************************************************************//**
+ * @file     cmsis_iccarm.h
+ * @brief    CMSIS compiler ICCARM (IAR Compiler for Arm) header file
+ * @version  V5.2.0
+ * @date     28. January 2020
+ ******************************************************************************/
+
+//------------------------------------------------------------------------------
+//
+// Copyright (c) 2017-2019 IAR Systems
+// Copyright (c) 2017-2019 Arm Limited. All rights reserved. 
+//
+// SPDX-License-Identifier: Apache-2.0
+//
+// Licensed under the Apache License, Version 2.0 (the "License")
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+//
+//------------------------------------------------------------------------------
+
+
+#ifndef __CMSIS_ICCARM_H__
+#define __CMSIS_ICCARM_H__
+
+#ifndef __ICCARM__
+  #error This file should only be compiled by ICCARM
+#endif
+
+#pragma system_include
+
+#define __IAR_FT _Pragma("inline=forced") __intrinsic
+
+#if (__VER__ >= 8000000)
+  #define __ICCARM_V8 1
+#else
+  #define __ICCARM_V8 0
+#endif
+
+#ifndef __ALIGNED
+  #if __ICCARM_V8
+    #define __ALIGNED(x) __attribute__((aligned(x)))
+  #elif (__VER__ >= 7080000)
+    /* Needs IAR language extensions */
+    #define __ALIGNED(x) __attribute__((aligned(x)))
+  #else
+    #warning No compiler specific solution for __ALIGNED.__ALIGNED is ignored.
+    #define __ALIGNED(x)
+  #endif
+#endif
+
+
+/* Define compiler macros for CPU architecture, used in CMSIS 5.
+ */
+#if __ARM_ARCH_6M__ || __ARM_ARCH_7M__ || __ARM_ARCH_7EM__ || __ARM_ARCH_8M_BASE__ || __ARM_ARCH_8M_MAIN__
+/* Macros already defined */
+#else
+  #if defined(__ARM8M_MAINLINE__) || defined(__ARM8EM_MAINLINE__)
+    #define __ARM_ARCH_8M_MAIN__ 1
+  #elif defined(__ARM8M_BASELINE__)
+    #define __ARM_ARCH_8M_BASE__ 1
+  #elif defined(__ARM_ARCH_PROFILE) && __ARM_ARCH_PROFILE == 'M'
+    #if __ARM_ARCH == 6
+      #define __ARM_ARCH_6M__ 1
+    #elif __ARM_ARCH == 7
+      #if __ARM_FEATURE_DSP
+        #define __ARM_ARCH_7EM__ 1
+      #else
+        #define __ARM_ARCH_7M__ 1
+      #endif
+    #endif /* __ARM_ARCH */
+  #endif /* __ARM_ARCH_PROFILE == 'M' */
+#endif
+
+/* Alternativ core deduction for older ICCARM's */
+#if !defined(__ARM_ARCH_6M__) && !defined(__ARM_ARCH_7M__) && !defined(__ARM_ARCH_7EM__) && \
+    !defined(__ARM_ARCH_8M_BASE__) && !defined(__ARM_ARCH_8M_MAIN__)
+  #if defined(__ARM6M__) && (__CORE__ == __ARM6M__)
+    #define __ARM_ARCH_6M__ 1
+  #elif defined(__ARM7M__) && (__CORE__ == __ARM7M__)
+    #define __ARM_ARCH_7M__ 1
+  #elif defined(__ARM7EM__) && (__CORE__ == __ARM7EM__)
+    #define __ARM_ARCH_7EM__  1
+  #elif defined(__ARM8M_BASELINE__) && (__CORE == __ARM8M_BASELINE__)
+    #define __ARM_ARCH_8M_BASE__ 1
+  #elif defined(__ARM8M_MAINLINE__) && (__CORE == __ARM8M_MAINLINE__)
+    #define __ARM_ARCH_8M_MAIN__ 1
+  #elif defined(__ARM8EM_MAINLINE__) && (__CORE == __ARM8EM_MAINLINE__)
+    #define __ARM_ARCH_8M_MAIN__ 1
+  #else
+    #error "Unknown target."
+  #endif
+#endif
+
+
+
+#if defined(__ARM_ARCH_6M__) && __ARM_ARCH_6M__==1
+  #define __IAR_M0_FAMILY  1
+#elif defined(__ARM_ARCH_8M_BASE__) && __ARM_ARCH_8M_BASE__==1
+  #define __IAR_M0_FAMILY  1
+#else
+  #define __IAR_M0_FAMILY  0
+#endif
+
+
+#ifndef __ASM
+  #define __ASM __asm
+#endif
+
+#ifndef   __COMPILER_BARRIER
+  #define __COMPILER_BARRIER() __ASM volatile("":::"memory")
+#endif
+
+#ifndef __INLINE
+  #define __INLINE inline
+#endif
+
+#ifndef   __NO_RETURN
+  #if __ICCARM_V8
+    #define __NO_RETURN __attribute__((__noreturn__))
+  #else
+    #define __NO_RETURN _Pragma("object_attribute=__noreturn")
+  #endif
+#endif
+
+#ifndef   __PACKED
+  #if __ICCARM_V8
+    #define __PACKED __attribute__((packed, aligned(1)))
+  #else
+    /* Needs IAR language extensions */
+    #define __PACKED __packed
+  #endif
+#endif
+
+#ifndef   __PACKED_STRUCT
+  #if __ICCARM_V8
+    #define __PACKED_STRUCT struct __attribute__((packed, aligned(1)))
+  #else
+    /* Needs IAR language extensions */
+    #define __PACKED_STRUCT __packed struct
+  #endif
+#endif
+
+#ifndef   __PACKED_UNION
+  #if __ICCARM_V8
+    #define __PACKED_UNION union __attribute__((packed, aligned(1)))
+  #else
+    /* Needs IAR language extensions */
+    #define __PACKED_UNION __packed union
+  #endif
+#endif
+
+#ifndef   __RESTRICT
+  #if __ICCARM_V8
+    #define __RESTRICT            __restrict
+  #else
+    /* Needs IAR language extensions */
+    #define __RESTRICT            restrict
+  #endif
+#endif
+
+#ifndef   __STATIC_INLINE
+  #define __STATIC_INLINE       static inline
+#endif
+
+#ifndef   __FORCEINLINE
+  #define __FORCEINLINE         _Pragma("inline=forced")
+#endif
+
+#ifndef   __STATIC_FORCEINLINE
+  #define __STATIC_FORCEINLINE  __FORCEINLINE __STATIC_INLINE
+#endif
+
+#ifndef __UNALIGNED_UINT16_READ
+#pragma language=save
+#pragma language=extended
+__IAR_FT uint16_t __iar_uint16_read(void const *ptr)
+{
+  return *(__packed uint16_t*)(ptr);
+}
+#pragma language=restore
+#define __UNALIGNED_UINT16_READ(PTR) __iar_uint16_read(PTR)
+#endif
+
+
+#ifndef __UNALIGNED_UINT16_WRITE
+#pragma language=save
+#pragma language=extended
+__IAR_FT void __iar_uint16_write(void const *ptr, uint16_t val)
+{
+  *(__packed uint16_t*)(ptr) = val;;
+}
+#pragma language=restore
+#define __UNALIGNED_UINT16_WRITE(PTR,VAL) __iar_uint16_write(PTR,VAL)
+#endif
+
+#ifndef __UNALIGNED_UINT32_READ
+#pragma language=save
+#pragma language=extended
+__IAR_FT uint32_t __iar_uint32_read(void const *ptr)
+{
+  return *(__packed uint32_t*)(ptr);
+}
+#pragma language=restore
+#define __UNALIGNED_UINT32_READ(PTR) __iar_uint32_read(PTR)
+#endif
+
+#ifndef __UNALIGNED_UINT32_WRITE
+#pragma language=save
+#pragma language=extended
+__IAR_FT void __iar_uint32_write(void const *ptr, uint32_t val)
+{
+  *(__packed uint32_t*)(ptr) = val;;
+}
+#pragma language=restore
+#define __UNALIGNED_UINT32_WRITE(PTR,VAL) __iar_uint32_write(PTR,VAL)
+#endif
+
+#ifndef __UNALIGNED_UINT32   /* deprecated */
+#pragma language=save
+#pragma language=extended
+__packed struct  __iar_u32 { uint32_t v; };
+#pragma language=restore
+#define __UNALIGNED_UINT32(PTR) (((struct __iar_u32 *)(PTR))->v)
+#endif
+
+#ifndef   __USED
+  #if __ICCARM_V8
+    #define __USED __attribute__((used))
+  #else
+    #define __USED _Pragma("__root")
+  #endif
+#endif
+
+#ifndef   __WEAK
+  #if __ICCARM_V8
+    #define __WEAK __attribute__((weak))
+  #else
+    #define __WEAK _Pragma("__weak")
+  #endif
+#endif
+
+#ifndef __PROGRAM_START
+#define __PROGRAM_START           __iar_program_start
+#endif
+
+#ifndef __INITIAL_SP
+#define __INITIAL_SP              CSTACK$$Limit
+#endif
+
+#ifndef __STACK_LIMIT
+#define __STACK_LIMIT             CSTACK$$Base
+#endif
+
+#ifndef __VECTOR_TABLE
+#define __VECTOR_TABLE            __vector_table
+#endif
+
+#ifndef __VECTOR_TABLE_ATTRIBUTE
+#define __VECTOR_TABLE_ATTRIBUTE  @".intvec"
+#endif
+
+#ifndef __ICCARM_INTRINSICS_VERSION__
+  #define __ICCARM_INTRINSICS_VERSION__  0
+#endif
+
+#if __ICCARM_INTRINSICS_VERSION__ == 2
+
+  #if defined(__CLZ)
+    #undef __CLZ
+  #endif
+  #if defined(__REVSH)
+    #undef __REVSH
+  #endif
+  #if defined(__RBIT)
+    #undef __RBIT
+  #endif
+  #if defined(__SSAT)
+    #undef __SSAT
+  #endif
+  #if defined(__USAT)
+    #undef __USAT
+  #endif
+
+  #include "iccarm_builtin.h"
+
+  #define __disable_fault_irq __iar_builtin_disable_fiq
+  #define __disable_irq       __iar_builtin_disable_interrupt
+  #define __enable_fault_irq  __iar_builtin_enable_fiq
+  #define __enable_irq        __iar_builtin_enable_interrupt
+  #define __arm_rsr           __iar_builtin_rsr
+  #define __arm_wsr           __iar_builtin_wsr
+
+
+  #define __get_APSR()                (__arm_rsr("APSR"))
+  #define __get_BASEPRI()             (__arm_rsr("BASEPRI"))
+  #define __get_CONTROL()             (__arm_rsr("CONTROL"))
+  #define __get_FAULTMASK()           (__arm_rsr("FAULTMASK"))
+
+  #if ((defined (__FPU_PRESENT) && (__FPU_PRESENT == 1U)) && \
+       (defined (__FPU_USED   ) && (__FPU_USED    == 1U))     )
+    #define __get_FPSCR()             (__arm_rsr("FPSCR"))
+    #define __set_FPSCR(VALUE)        (__arm_wsr("FPSCR", (VALUE)))
+  #else
+    #define __get_FPSCR()             ( 0 )
+    #define __set_FPSCR(VALUE)        ((void)VALUE)
+  #endif
+
+  #define __get_IPSR()                (__arm_rsr("IPSR"))
+  #define __get_MSP()                 (__arm_rsr("MSP"))
+  #if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) && \
+       (!defined (__ARM_FEATURE_CMSE) || (__ARM_FEATURE_CMSE < 3)))
+    // without main extensions, the non-secure MSPLIM is RAZ/WI
+    #define __get_MSPLIM()            (0U)
+  #else
+    #define __get_MSPLIM()            (__arm_rsr("MSPLIM"))
+  #endif
+  #define __get_PRIMASK()             (__arm_rsr("PRIMASK"))
+  #define __get_PSP()                 (__arm_rsr("PSP"))
+
+  #if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) && \
+       (!defined (__ARM_FEATURE_CMSE) || (__ARM_FEATURE_CMSE < 3)))
+    // without main extensions, the non-secure PSPLIM is RAZ/WI
+    #define __get_PSPLIM()            (0U)
+  #else
+    #define __get_PSPLIM()            (__arm_rsr("PSPLIM"))
+  #endif
+
+  #define __get_xPSR()                (__arm_rsr("xPSR"))
+
+  #define __set_BASEPRI(VALUE)        (__arm_wsr("BASEPRI", (VALUE)))
+  #define __set_BASEPRI_MAX(VALUE)    (__arm_wsr("BASEPRI_MAX", (VALUE)))
+  #define __set_CONTROL(VALUE)        (__arm_wsr("CONTROL", (VALUE)))
+  #define __set_FAULTMASK(VALUE)      (__arm_wsr("FAULTMASK", (VALUE)))
+  #define __set_MSP(VALUE)            (__arm_wsr("MSP", (VALUE)))
+
+  #if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) && \
+       (!defined (__ARM_FEATURE_CMSE) || (__ARM_FEATURE_CMSE < 3)))
+    // without main extensions, the non-secure MSPLIM is RAZ/WI
+    #define __set_MSPLIM(VALUE)       ((void)(VALUE))
+  #else
+    #define __set_MSPLIM(VALUE)       (__arm_wsr("MSPLIM", (VALUE)))
+  #endif
+  #define __set_PRIMASK(VALUE)        (__arm_wsr("PRIMASK", (VALUE)))
+  #define __set_PSP(VALUE)            (__arm_wsr("PSP", (VALUE)))
+  #if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) && \
+       (!defined (__ARM_FEATURE_CMSE) || (__ARM_FEATURE_CMSE < 3)))
+    // without main extensions, the non-secure PSPLIM is RAZ/WI
+    #define __set_PSPLIM(VALUE)       ((void)(VALUE))
+  #else
+    #define __set_PSPLIM(VALUE)       (__arm_wsr("PSPLIM", (VALUE)))
+  #endif
+
+  #define __TZ_get_CONTROL_NS()       (__arm_rsr("CONTROL_NS"))
+  #define __TZ_set_CONTROL_NS(VALUE)  (__arm_wsr("CONTROL_NS", (VALUE)))
+  #define __TZ_get_PSP_NS()           (__arm_rsr("PSP_NS"))
+  #define __TZ_set_PSP_NS(VALUE)      (__arm_wsr("PSP_NS", (VALUE)))
+  #define __TZ_get_MSP_NS()           (__arm_rsr("MSP_NS"))
+  #define __TZ_set_MSP_NS(VALUE)      (__arm_wsr("MSP_NS", (VALUE)))
+  #define __TZ_get_SP_NS()            (__arm_rsr("SP_NS"))
+  #define __TZ_set_SP_NS(VALUE)       (__arm_wsr("SP_NS", (VALUE)))
+  #define __TZ_get_PRIMASK_NS()       (__arm_rsr("PRIMASK_NS"))
+  #define __TZ_set_PRIMASK_NS(VALUE)  (__arm_wsr("PRIMASK_NS", (VALUE)))
+  #define __TZ_get_BASEPRI_NS()       (__arm_rsr("BASEPRI_NS"))
+  #define __TZ_set_BASEPRI_NS(VALUE)  (__arm_wsr("BASEPRI_NS", (VALUE)))
+  #define __TZ_get_FAULTMASK_NS()     (__arm_rsr("FAULTMASK_NS"))
+  #define __TZ_set_FAULTMASK_NS(VALUE)(__arm_wsr("FAULTMASK_NS", (VALUE)))
+
+  #if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) && \
+       (!defined (__ARM_FEATURE_CMSE) || (__ARM_FEATURE_CMSE < 3)))
+    // without main extensions, the non-secure PSPLIM is RAZ/WI
+    #define __TZ_get_PSPLIM_NS()      (0U)
+    #define __TZ_set_PSPLIM_NS(VALUE) ((void)(VALUE))
+  #else
+    #define __TZ_get_PSPLIM_NS()      (__arm_rsr("PSPLIM_NS"))
+    #define __TZ_set_PSPLIM_NS(VALUE) (__arm_wsr("PSPLIM_NS", (VALUE)))
+  #endif
+
+  #define __TZ_get_MSPLIM_NS()        (__arm_rsr("MSPLIM_NS"))
+  #define __TZ_set_MSPLIM_NS(VALUE)   (__arm_wsr("MSPLIM_NS", (VALUE)))
+
+  #define __NOP     __iar_builtin_no_operation
+
+  #define __CLZ     __iar_builtin_CLZ
+  #define __CLREX   __iar_builtin_CLREX
+
+  #define __DMB     __iar_builtin_DMB
+  #define __DSB     __iar_builtin_DSB
+  #define __ISB     __iar_builtin_ISB
+
+  #define __LDREXB  __iar_builtin_LDREXB
+  #define __LDREXH  __iar_builtin_LDREXH
+  #define __LDREXW  __iar_builtin_LDREX
+
+  #define __RBIT    __iar_builtin_RBIT
+  #define __REV     __iar_builtin_REV
+  #define __REV16   __iar_builtin_REV16
+
+  __IAR_FT int16_t __REVSH(int16_t val)
+  {
+    return (int16_t) __iar_builtin_REVSH(val);
+  }
+
+  #define __ROR     __iar_builtin_ROR
+  #define __RRX     __iar_builtin_RRX
+
+  #define __SEV     __iar_builtin_SEV
+
+  #if !__IAR_M0_FAMILY
+    #define __SSAT    __iar_builtin_SSAT
+  #endif
+
+  #define __STREXB  __iar_builtin_STREXB
+  #define __STREXH  __iar_builtin_STREXH
+  #define __STREXW  __iar_builtin_STREX
+
+  #if !__IAR_M0_FAMILY
+    #define __USAT    __iar_builtin_USAT
+  #endif
+
+  #define __WFE     __iar_builtin_WFE
+  #define __WFI     __iar_builtin_WFI
+
+  #if __ARM_MEDIA__
+    #define __SADD8   __iar_builtin_SADD8
+    #define __QADD8   __iar_builtin_QADD8
+    #define __SHADD8  __iar_builtin_SHADD8
+    #define __UADD8   __iar_builtin_UADD8
+    #define __UQADD8  __iar_builtin_UQADD8
+    #define __UHADD8  __iar_builtin_UHADD8
+    #define __SSUB8   __iar_builtin_SSUB8
+    #define __QSUB8   __iar_builtin_QSUB8
+    #define __SHSUB8  __iar_builtin_SHSUB8
+    #define __USUB8   __iar_builtin_USUB8
+    #define __UQSUB8  __iar_builtin_UQSUB8
+    #define __UHSUB8  __iar_builtin_UHSUB8
+    #define __SADD16  __iar_builtin_SADD16
+    #define __QADD16  __iar_builtin_QADD16
+    #define __SHADD16 __iar_builtin_SHADD16
+    #define __UADD16  __iar_builtin_UADD16
+    #define __UQADD16 __iar_builtin_UQADD16
+    #define __UHADD16 __iar_builtin_UHADD16
+    #define __SSUB16  __iar_builtin_SSUB16
+    #define __QSUB16  __iar_builtin_QSUB16
+    #define __SHSUB16 __iar_builtin_SHSUB16
+    #define __USUB16  __iar_builtin_USUB16
+    #define __UQSUB16 __iar_builtin_UQSUB16
+    #define __UHSUB16 __iar_builtin_UHSUB16
+    #define __SASX    __iar_builtin_SASX
+    #define __QASX    __iar_builtin_QASX
+    #define __SHASX   __iar_builtin_SHASX
+    #define __UASX    __iar_builtin_UASX
+    #define __UQASX   __iar_builtin_UQASX
+    #define __UHASX   __iar_builtin_UHASX
+    #define __SSAX    __iar_builtin_SSAX
+    #define __QSAX    __iar_builtin_QSAX
+    #define __SHSAX   __iar_builtin_SHSAX
+    #define __USAX    __iar_builtin_USAX
+    #define __UQSAX   __iar_builtin_UQSAX
+    #define __UHSAX   __iar_builtin_UHSAX
+    #define __USAD8   __iar_builtin_USAD8
+    #define __USADA8  __iar_builtin_USADA8
+    #define __SSAT16  __iar_builtin_SSAT16
+    #define __USAT16  __iar_builtin_USAT16
+    #define __UXTB16  __iar_builtin_UXTB16
+    #define __UXTAB16 __iar_builtin_UXTAB16
+    #define __SXTB16  __iar_builtin_SXTB16
+    #define __SXTAB16 __iar_builtin_SXTAB16
+    #define __SMUAD   __iar_builtin_SMUAD
+    #define __SMUADX  __iar_builtin_SMUADX
+    #define __SMMLA   __iar_builtin_SMMLA
+    #define __SMLAD   __iar_builtin_SMLAD
+    #define __SMLADX  __iar_builtin_SMLADX
+    #define __SMLALD  __iar_builtin_SMLALD
+    #define __SMLALDX __iar_builtin_SMLALDX
+    #define __SMUSD   __iar_builtin_SMUSD
+    #define __SMUSDX  __iar_builtin_SMUSDX
+    #define __SMLSD   __iar_builtin_SMLSD
+    #define __SMLSDX  __iar_builtin_SMLSDX
+    #define __SMLSLD  __iar_builtin_SMLSLD
+    #define __SMLSLDX __iar_builtin_SMLSLDX
+    #define __SEL     __iar_builtin_SEL
+    #define __QADD    __iar_builtin_QADD
+    #define __QSUB    __iar_builtin_QSUB
+    #define __PKHBT   __iar_builtin_PKHBT
+    #define __PKHTB   __iar_builtin_PKHTB
+  #endif
+
+#else /* __ICCARM_INTRINSICS_VERSION__ == 2 */
+
+  #if __IAR_M0_FAMILY
+   /* Avoid clash between intrinsics.h and arm_math.h when compiling for Cortex-M0. */
+    #define __CLZ  __cmsis_iar_clz_not_active
+    #define __SSAT __cmsis_iar_ssat_not_active
+    #define __USAT __cmsis_iar_usat_not_active
+    #define __RBIT __cmsis_iar_rbit_not_active
+    #define __get_APSR  __cmsis_iar_get_APSR_not_active
+  #endif
+
+
+  #if (!((defined (__FPU_PRESENT) && (__FPU_PRESENT == 1U)) && \
+         (defined (__FPU_USED   ) && (__FPU_USED    == 1U))     ))
+    #define __get_FPSCR __cmsis_iar_get_FPSR_not_active
+    #define __set_FPSCR __cmsis_iar_set_FPSR_not_active
+  #endif
+
+  #ifdef __INTRINSICS_INCLUDED
+  #error intrinsics.h is already included previously!
+  #endif
+
+  #include <intrinsics.h>
+
+  #if __IAR_M0_FAMILY
+   /* Avoid clash between intrinsics.h and arm_math.h when compiling for Cortex-M0. */
+    #undef __CLZ
+    #undef __SSAT
+    #undef __USAT
+    #undef __RBIT
+    #undef __get_APSR
+
+    __STATIC_INLINE uint8_t __CLZ(uint32_t data)
+    {
+      if (data == 0U) { return 32U; }
+
+      uint32_t count = 0U;
+      uint32_t mask = 0x80000000U;
+
+      while ((data & mask) == 0U)
+      {
+        count += 1U;
+        mask = mask >> 1U;
+      }
+      return count;
+    }
+
+    __STATIC_INLINE uint32_t __RBIT(uint32_t v)
+    {
+      uint8_t sc = 31U;
+      uint32_t r = v;
+      for (v >>= 1U; v; v >>= 1U)
+      {
+        r <<= 1U;
+        r |= v & 1U;
+        sc--;
+      }
+      return (r << sc);
+    }
+
+    __STATIC_INLINE  uint32_t __get_APSR(void)
+    {
+      uint32_t res;
+      __asm("MRS      %0,APSR" : "=r" (res));
+      return res;
+    }
+
+  #endif
+
+  #if (!((defined (__FPU_PRESENT) && (__FPU_PRESENT == 1U)) && \
+         (defined (__FPU_USED   ) && (__FPU_USED    == 1U))     ))
+    #undef __get_FPSCR
+    #undef __set_FPSCR
+    #define __get_FPSCR()       (0)
+    #define __set_FPSCR(VALUE)  ((void)VALUE)
+  #endif
+
+  #pragma diag_suppress=Pe940
+  #pragma diag_suppress=Pe177
+
+  #define __enable_irq    __enable_interrupt
+  #define __disable_irq   __disable_interrupt
+  #define __NOP           __no_operation
+
+  #define __get_xPSR      __get_PSR
+
+  #if (!defined(__ARM_ARCH_6M__) || __ARM_ARCH_6M__==0)
+
+    __IAR_FT uint32_t __LDREXW(uint32_t volatile *ptr)
+    {
+      return __LDREX((unsigned long *)ptr);
+    }
+
+    __IAR_FT uint32_t __STREXW(uint32_t value, uint32_t volatile *ptr)
+    {
+      return __STREX(value, (unsigned long *)ptr);
+    }
+  #endif
+
+
+  /* __CORTEX_M is defined in core_cm0.h, core_cm3.h and core_cm4.h. */
+  #if (__CORTEX_M >= 0x03)
+
+    __IAR_FT uint32_t __RRX(uint32_t value)
+    {
+      uint32_t result;
+      __ASM volatile("RRX      %0, %1" : "=r"(result) : "r" (value));
+      return(result);
+    }
+
+    __IAR_FT void __set_BASEPRI_MAX(uint32_t value)
+    {
+      __asm volatile("MSR      BASEPRI_MAX,%0"::"r" (value));
+    }
+
+
+    #define __enable_fault_irq  __enable_fiq
+    #define __disable_fault_irq __disable_fiq
+
+
+  #endif /* (__CORTEX_M >= 0x03) */
+
+  __IAR_FT uint32_t __ROR(uint32_t op1, uint32_t op2)
+  {
+    return (op1 >> op2) | (op1 << ((sizeof(op1)*8)-op2));
+  }
+
+  #if ((defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) || \
+       (defined (__ARM_ARCH_8M_BASE__ ) && (__ARM_ARCH_8M_BASE__ == 1))    )
+
+   __IAR_FT uint32_t __get_MSPLIM(void)
+    {
+      uint32_t res;
+    #if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) && \
+         (!defined (__ARM_FEATURE_CMSE  ) || (__ARM_FEATURE_CMSE   < 3)))
+      // without main extensions, the non-secure MSPLIM is RAZ/WI
+      res = 0U;
+    #else
+      __asm volatile("MRS      %0,MSPLIM" : "=r" (res));
+    #endif
+      return res;
+    }
+
+    __IAR_FT void   __set_MSPLIM(uint32_t value)
+    {
+    #if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) && \
+         (!defined (__ARM_FEATURE_CMSE  ) || (__ARM_FEATURE_CMSE   < 3)))
+      // without main extensions, the non-secure MSPLIM is RAZ/WI
+      (void)value;
+    #else
+      __asm volatile("MSR      MSPLIM,%0" :: "r" (value));
+    #endif
+    }
+
+    __IAR_FT uint32_t __get_PSPLIM(void)
+    {
+      uint32_t res;
+    #if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) && \
+         (!defined (__ARM_FEATURE_CMSE  ) || (__ARM_FEATURE_CMSE   < 3)))
+      // without main extensions, the non-secure PSPLIM is RAZ/WI
+      res = 0U;
+    #else
+      __asm volatile("MRS      %0,PSPLIM" : "=r" (res));
+    #endif
+      return res;
+    }
+
+    __IAR_FT void   __set_PSPLIM(uint32_t value)
+    {
+    #if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) && \
+         (!defined (__ARM_FEATURE_CMSE  ) || (__ARM_FEATURE_CMSE   < 3)))
+      // without main extensions, the non-secure PSPLIM is RAZ/WI
+      (void)value;
+    #else
+      __asm volatile("MSR      PSPLIM,%0" :: "r" (value));
+    #endif
+    }
+
+    __IAR_FT uint32_t __TZ_get_CONTROL_NS(void)
+    {
+      uint32_t res;
+      __asm volatile("MRS      %0,CONTROL_NS" : "=r" (res));
+      return res;
+    }
+
+    __IAR_FT void   __TZ_set_CONTROL_NS(uint32_t value)
+    {
+      __asm volatile("MSR      CONTROL_NS,%0" :: "r" (value));
+    }
+
+    __IAR_FT uint32_t   __TZ_get_PSP_NS(void)
+    {
+      uint32_t res;
+      __asm volatile("MRS      %0,PSP_NS" : "=r" (res));
+      return res;
+    }
+
+    __IAR_FT void   __TZ_set_PSP_NS(uint32_t value)
+    {
+      __asm volatile("MSR      PSP_NS,%0" :: "r" (value));
+    }
+
+    __IAR_FT uint32_t   __TZ_get_MSP_NS(void)
+    {
+      uint32_t res;
+      __asm volatile("MRS      %0,MSP_NS" : "=r" (res));
+      return res;
+    }
+
+    __IAR_FT void   __TZ_set_MSP_NS(uint32_t value)
+    {
+      __asm volatile("MSR      MSP_NS,%0" :: "r" (value));
+    }
+
+    __IAR_FT uint32_t   __TZ_get_SP_NS(void)
+    {
+      uint32_t res;
+      __asm volatile("MRS      %0,SP_NS" : "=r" (res));
+      return res;
+    }
+    __IAR_FT void   __TZ_set_SP_NS(uint32_t value)
+    {
+      __asm volatile("MSR      SP_NS,%0" :: "r" (value));
+    }
+
+    __IAR_FT uint32_t   __TZ_get_PRIMASK_NS(void)
+    {
+      uint32_t res;
+      __asm volatile("MRS      %0,PRIMASK_NS" : "=r" (res));
+      return res;
+    }
+
+    __IAR_FT void   __TZ_set_PRIMASK_NS(uint32_t value)
+    {
+      __asm volatile("MSR      PRIMASK_NS,%0" :: "r" (value));
+    }
+
+    __IAR_FT uint32_t   __TZ_get_BASEPRI_NS(void)
+    {
+      uint32_t res;
+      __asm volatile("MRS      %0,BASEPRI_NS" : "=r" (res));
+      return res;
+    }
+
+    __IAR_FT void   __TZ_set_BASEPRI_NS(uint32_t value)
+    {
+      __asm volatile("MSR      BASEPRI_NS,%0" :: "r" (value));
+    }
+
+    __IAR_FT uint32_t   __TZ_get_FAULTMASK_NS(void)
+    {
+      uint32_t res;
+      __asm volatile("MRS      %0,FAULTMASK_NS" : "=r" (res));
+      return res;
+    }
+
+    __IAR_FT void   __TZ_set_FAULTMASK_NS(uint32_t value)
+    {
+      __asm volatile("MSR      FAULTMASK_NS,%0" :: "r" (value));
+    }
+
+    __IAR_FT uint32_t   __TZ_get_PSPLIM_NS(void)
+    {
+      uint32_t res;
+    #if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) && \
+         (!defined (__ARM_FEATURE_CMSE  ) || (__ARM_FEATURE_CMSE   < 3)))
+      // without main extensions, the non-secure PSPLIM is RAZ/WI
+      res = 0U;
+    #else
+      __asm volatile("MRS      %0,PSPLIM_NS" : "=r" (res));
+    #endif
+      return res;
+    }
+
+    __IAR_FT void   __TZ_set_PSPLIM_NS(uint32_t value)
+    {
+    #if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) && \
+         (!defined (__ARM_FEATURE_CMSE  ) || (__ARM_FEATURE_CMSE   < 3)))
+      // without main extensions, the non-secure PSPLIM is RAZ/WI
+      (void)value;
+    #else
+      __asm volatile("MSR      PSPLIM_NS,%0" :: "r" (value));
+    #endif
+    }
+
+    __IAR_FT uint32_t   __TZ_get_MSPLIM_NS(void)
+    {
+      uint32_t res;
+      __asm volatile("MRS      %0,MSPLIM_NS" : "=r" (res));
+      return res;
+    }
+
+    __IAR_FT void   __TZ_set_MSPLIM_NS(uint32_t value)
+    {
+      __asm volatile("MSR      MSPLIM_NS,%0" :: "r" (value));
+    }
+
+  #endif /* __ARM_ARCH_8M_MAIN__ or __ARM_ARCH_8M_BASE__ */
+
+#endif   /* __ICCARM_INTRINSICS_VERSION__ == 2 */
+
+#define __BKPT(value)    __asm volatile ("BKPT     %0" : : "i"(value))
+
+#if __IAR_M0_FAMILY
+  __STATIC_INLINE int32_t __SSAT(int32_t val, uint32_t sat)
+  {
+    if ((sat >= 1U) && (sat <= 32U))
+    {
+      const int32_t max = (int32_t)((1U << (sat - 1U)) - 1U);
+      const int32_t min = -1 - max ;
+      if (val > max)
+      {
+        return max;
+      }
+      else if (val < min)
+      {
+        return min;
+      }
+    }
+    return val;
+  }
+
+  __STATIC_INLINE uint32_t __USAT(int32_t val, uint32_t sat)
+  {
+    if (sat <= 31U)
+    {
+      const uint32_t max = ((1U << sat) - 1U);
+      if (val > (int32_t)max)
+      {
+        return max;
+      }
+      else if (val < 0)
+      {
+        return 0U;
+      }
+    }
+    return (uint32_t)val;
+  }
+#endif
+
+#if (__CORTEX_M >= 0x03)   /* __CORTEX_M is defined in core_cm0.h, core_cm3.h and core_cm4.h. */
+
+  __IAR_FT uint8_t __LDRBT(volatile uint8_t *addr)
+  {
+    uint32_t res;
+    __ASM volatile ("LDRBT %0, [%1]" : "=r" (res) : "r" (addr) : "memory");
+    return ((uint8_t)res);
+  }
+
+  __IAR_FT uint16_t __LDRHT(volatile uint16_t *addr)
+  {
+    uint32_t res;
+    __ASM volatile ("LDRHT %0, [%1]" : "=r" (res) : "r" (addr) : "memory");
+    return ((uint16_t)res);
+  }
+
+  __IAR_FT uint32_t __LDRT(volatile uint32_t *addr)
+  {
+    uint32_t res;
+    __ASM volatile ("LDRT %0, [%1]" : "=r" (res) : "r" (addr) : "memory");
+    return res;
+  }
+
+  __IAR_FT void __STRBT(uint8_t value, volatile uint8_t *addr)
+  {
+    __ASM volatile ("STRBT %1, [%0]" : : "r" (addr), "r" ((uint32_t)value) : "memory");
+  }
+
+  __IAR_FT void __STRHT(uint16_t value, volatile uint16_t *addr)
+  {
+    __ASM volatile ("STRHT %1, [%0]" : : "r" (addr), "r" ((uint32_t)value) : "memory");
+  }
+
+  __IAR_FT void __STRT(uint32_t value, volatile uint32_t *addr)
+  {
+    __ASM volatile ("STRT %1, [%0]" : : "r" (addr), "r" (value) : "memory");
+  }
+
+#endif /* (__CORTEX_M >= 0x03) */
+
+#if ((defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) || \
+     (defined (__ARM_ARCH_8M_BASE__ ) && (__ARM_ARCH_8M_BASE__ == 1))    )
+
+
+  __IAR_FT uint8_t __LDAB(volatile uint8_t *ptr)
+  {
+    uint32_t res;
+    __ASM volatile ("LDAB %0, [%1]" : "=r" (res) : "r" (ptr) : "memory");
+    return ((uint8_t)res);
+  }
+
+  __IAR_FT uint16_t __LDAH(volatile uint16_t *ptr)
+  {
+    uint32_t res;
+    __ASM volatile ("LDAH %0, [%1]" : "=r" (res) : "r" (ptr) : "memory");
+    return ((uint16_t)res);
+  }
+
+  __IAR_FT uint32_t __LDA(volatile uint32_t *ptr)
+  {
+    uint32_t res;
+    __ASM volatile ("LDA %0, [%1]" : "=r" (res) : "r" (ptr) : "memory");
+    return res;
+  }
+
+  __IAR_FT void __STLB(uint8_t value, volatile uint8_t *ptr)
+  {
+    __ASM volatile ("STLB %1, [%0]" :: "r" (ptr), "r" (value) : "memory");
+  }
+
+  __IAR_FT void __STLH(uint16_t value, volatile uint16_t *ptr)
+  {
+    __ASM volatile ("STLH %1, [%0]" :: "r" (ptr), "r" (value) : "memory");
+  }
+
+  __IAR_FT void __STL(uint32_t value, volatile uint32_t *ptr)
+  {
+    __ASM volatile ("STL %1, [%0]" :: "r" (ptr), "r" (value) : "memory");
+  }
+
+  __IAR_FT uint8_t __LDAEXB(volatile uint8_t *ptr)
+  {
+    uint32_t res;
+    __ASM volatile ("LDAEXB %0, [%1]" : "=r" (res) : "r" (ptr) : "memory");
+    return ((uint8_t)res);
+  }
+
+  __IAR_FT uint16_t __LDAEXH(volatile uint16_t *ptr)
+  {
+    uint32_t res;
+    __ASM volatile ("LDAEXH %0, [%1]" : "=r" (res) : "r" (ptr) : "memory");
+    return ((uint16_t)res);
+  }
+
+  __IAR_FT uint32_t __LDAEX(volatile uint32_t *ptr)
+  {
+    uint32_t res;
+    __ASM volatile ("LDAEX %0, [%1]" : "=r" (res) : "r" (ptr) : "memory");
+    return res;
+  }
+
+  __IAR_FT uint32_t __STLEXB(uint8_t value, volatile uint8_t *ptr)
+  {
+    uint32_t res;
+    __ASM volatile ("STLEXB %0, %2, [%1]" : "=r" (res) : "r" (ptr), "r" (value) : "memory");
+    return res;
+  }
+
+  __IAR_FT uint32_t __STLEXH(uint16_t value, volatile uint16_t *ptr)
+  {
+    uint32_t res;
+    __ASM volatile ("STLEXH %0, %2, [%1]" : "=r" (res) : "r" (ptr), "r" (value) : "memory");
+    return res;
+  }
+
+  __IAR_FT uint32_t __STLEX(uint32_t value, volatile uint32_t *ptr)
+  {
+    uint32_t res;
+    __ASM volatile ("STLEX %0, %2, [%1]" : "=r" (res) : "r" (ptr), "r" (value) : "memory");
+    return res;
+  }
+
+#endif /* __ARM_ARCH_8M_MAIN__ or __ARM_ARCH_8M_BASE__ */
+
+#undef __IAR_FT
+#undef __IAR_M0_FAMILY
+#undef __ICCARM_V8
+
+#pragma diag_default=Pe940
+#pragma diag_default=Pe177
+
+#define __SXTB16_RORn(ARG1, ARG2) __SXTB16(__ROR(ARG1, ARG2))
+
+#endif /* __CMSIS_ICCARM_H__ */

libraries/sdk/third_party/CMSIS/Core/Include/cmsis_version.h

@@ -0,0 +1,39 @@
+/**************************************************************************//**
+ * @file     cmsis_version.h
+ * @brief    CMSIS Core(M) Version definitions
+ * @version  V5.0.4
+ * @date     23. July 2019
+ ******************************************************************************/
+/*
+ * Copyright (c) 2009-2019 ARM Limited. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#if   defined ( __ICCARM__ )
+  #pragma system_include         /* treat file as system include file for MISRA check */
+#elif defined (__clang__)
+  #pragma clang system_header   /* treat file as system include file */
+#endif
+
+#ifndef __CMSIS_VERSION_H
+#define __CMSIS_VERSION_H
+
+/*  CMSIS Version definitions */
+#define __CM_CMSIS_VERSION_MAIN  ( 5U)                                      /*!< [31:16] CMSIS Core(M) main version */
+#define __CM_CMSIS_VERSION_SUB   ( 4U)                                      /*!< [15:0]  CMSIS Core(M) sub version */
+#define __CM_CMSIS_VERSION       ((__CM_CMSIS_VERSION_MAIN << 16U) | \
+                                   __CM_CMSIS_VERSION_SUB           )       /*!< CMSIS Core(M) version number */
+#endif

libraries/sdk/third_party/CMSIS/Core/Include/mpu_armv7.h

@@ -0,0 +1,275 @@
+/******************************************************************************
+ * @file     mpu_armv7.h
+ * @brief    CMSIS MPU API for Armv7-M MPU
+ * @version  V5.1.1
+ * @date     10. February 2020
+ ******************************************************************************/
+/*
+ * Copyright (c) 2017-2020 Arm Limited. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+ 
+#if   defined ( __ICCARM__ )
+  #pragma system_include         /* treat file as system include file for MISRA check */
+#elif defined (__clang__)
+  #pragma clang system_header    /* treat file as system include file */
+#endif
+ 
+#ifndef ARM_MPU_ARMV7_H
+#define ARM_MPU_ARMV7_H
+
+#define ARM_MPU_REGION_SIZE_32B      ((uint8_t)0x04U) ///!< MPU Region Size 32 Bytes
+#define ARM_MPU_REGION_SIZE_64B      ((uint8_t)0x05U) ///!< MPU Region Size 64 Bytes
+#define ARM_MPU_REGION_SIZE_128B     ((uint8_t)0x06U) ///!< MPU Region Size 128 Bytes
+#define ARM_MPU_REGION_SIZE_256B     ((uint8_t)0x07U) ///!< MPU Region Size 256 Bytes
+#define ARM_MPU_REGION_SIZE_512B     ((uint8_t)0x08U) ///!< MPU Region Size 512 Bytes
+#define ARM_MPU_REGION_SIZE_1KB      ((uint8_t)0x09U) ///!< MPU Region Size 1 KByte
+#define ARM_MPU_REGION_SIZE_2KB      ((uint8_t)0x0AU) ///!< MPU Region Size 2 KBytes
+#define ARM_MPU_REGION_SIZE_4KB      ((uint8_t)0x0BU) ///!< MPU Region Size 4 KBytes
+#define ARM_MPU_REGION_SIZE_8KB      ((uint8_t)0x0CU) ///!< MPU Region Size 8 KBytes
+#define ARM_MPU_REGION_SIZE_16KB     ((uint8_t)0x0DU) ///!< MPU Region Size 16 KBytes
+#define ARM_MPU_REGION_SIZE_32KB     ((uint8_t)0x0EU) ///!< MPU Region Size 32 KBytes
+#define ARM_MPU_REGION_SIZE_64KB     ((uint8_t)0x0FU) ///!< MPU Region Size 64 KBytes
+#define ARM_MPU_REGION_SIZE_128KB    ((uint8_t)0x10U) ///!< MPU Region Size 128 KBytes
+#define ARM_MPU_REGION_SIZE_256KB    ((uint8_t)0x11U) ///!< MPU Region Size 256 KBytes
+#define ARM_MPU_REGION_SIZE_512KB    ((uint8_t)0x12U) ///!< MPU Region Size 512 KBytes
+#define ARM_MPU_REGION_SIZE_1MB      ((uint8_t)0x13U) ///!< MPU Region Size 1 MByte
+#define ARM_MPU_REGION_SIZE_2MB      ((uint8_t)0x14U) ///!< MPU Region Size 2 MBytes
+#define ARM_MPU_REGION_SIZE_4MB      ((uint8_t)0x15U) ///!< MPU Region Size 4 MBytes
+#define ARM_MPU_REGION_SIZE_8MB      ((uint8_t)0x16U) ///!< MPU Region Size 8 MBytes
+#define ARM_MPU_REGION_SIZE_16MB     ((uint8_t)0x17U) ///!< MPU Region Size 16 MBytes
+#define ARM_MPU_REGION_SIZE_32MB     ((uint8_t)0x18U) ///!< MPU Region Size 32 MBytes
+#define ARM_MPU_REGION_SIZE_64MB     ((uint8_t)0x19U) ///!< MPU Region Size 64 MBytes
+#define ARM_MPU_REGION_SIZE_128MB    ((uint8_t)0x1AU) ///!< MPU Region Size 128 MBytes
+#define ARM_MPU_REGION_SIZE_256MB    ((uint8_t)0x1BU) ///!< MPU Region Size 256 MBytes
+#define ARM_MPU_REGION_SIZE_512MB    ((uint8_t)0x1CU) ///!< MPU Region Size 512 MBytes
+#define ARM_MPU_REGION_SIZE_1GB      ((uint8_t)0x1DU) ///!< MPU Region Size 1 GByte
+#define ARM_MPU_REGION_SIZE_2GB      ((uint8_t)0x1EU) ///!< MPU Region Size 2 GBytes
+#define ARM_MPU_REGION_SIZE_4GB      ((uint8_t)0x1FU) ///!< MPU Region Size 4 GBytes
+
+#define ARM_MPU_AP_NONE 0U ///!< MPU Access Permission no access
+#define ARM_MPU_AP_PRIV 1U ///!< MPU Access Permission privileged access only
+#define ARM_MPU_AP_URO  2U ///!< MPU Access Permission unprivileged access read-only
+#define ARM_MPU_AP_FULL 3U ///!< MPU Access Permission full access
+#define ARM_MPU_AP_PRO  5U ///!< MPU Access Permission privileged access read-only
+#define ARM_MPU_AP_RO   6U ///!< MPU Access Permission read-only access
+
+/** MPU Region Base Address Register Value
+*
+* \param Region The region to be configured, number 0 to 15.
+* \param BaseAddress The base address for the region.
+*/
+#define ARM_MPU_RBAR(Region, BaseAddress) \
+  (((BaseAddress) & MPU_RBAR_ADDR_Msk) |  \
+   ((Region) & MPU_RBAR_REGION_Msk)    |  \
+   (MPU_RBAR_VALID_Msk))
+
+/**
+* MPU Memory Access Attributes
+* 
+* \param TypeExtField      Type extension field, allows you to configure memory access type, for example strongly ordered, peripheral.
+* \param IsShareable       Region is shareable between multiple bus masters.
+* \param IsCacheable       Region is cacheable, i.e. its value may be kept in cache.
+* \param IsBufferable      Region is bufferable, i.e. using write-back caching. Cacheable but non-bufferable regions use write-through policy.
+*/  
+#define ARM_MPU_ACCESS_(TypeExtField, IsShareable, IsCacheable, IsBufferable)   \
+  ((((TypeExtField) << MPU_RASR_TEX_Pos) & MPU_RASR_TEX_Msk)                  | \
+   (((IsShareable)  << MPU_RASR_S_Pos)   & MPU_RASR_S_Msk)                    | \
+   (((IsCacheable)  << MPU_RASR_C_Pos)   & MPU_RASR_C_Msk)                    | \
+   (((IsBufferable) << MPU_RASR_B_Pos)   & MPU_RASR_B_Msk))
+
+/**
+* MPU Region Attribute and Size Register Value
+* 
+* \param DisableExec       Instruction access disable bit, 1= disable instruction fetches.
+* \param AccessPermission  Data access permissions, allows you to configure read/write access for User and Privileged mode.
+* \param AccessAttributes  Memory access attribution, see \ref ARM_MPU_ACCESS_.
+* \param SubRegionDisable  Sub-region disable field.
+* \param Size              Region size of the region to be configured, for example 4K, 8K.
+*/
+#define ARM_MPU_RASR_EX(DisableExec, AccessPermission, AccessAttributes, SubRegionDisable, Size)    \
+  ((((DisableExec)      << MPU_RASR_XN_Pos)   & MPU_RASR_XN_Msk)                                  | \
+   (((AccessPermission) << MPU_RASR_AP_Pos)   & MPU_RASR_AP_Msk)                                  | \
+   (((AccessAttributes) & (MPU_RASR_TEX_Msk | MPU_RASR_S_Msk | MPU_RASR_C_Msk | MPU_RASR_B_Msk))) | \
+   (((SubRegionDisable) << MPU_RASR_SRD_Pos)  & MPU_RASR_SRD_Msk)                                 | \
+   (((Size)             << MPU_RASR_SIZE_Pos) & MPU_RASR_SIZE_Msk)                                | \
+   (((MPU_RASR_ENABLE_Msk))))
+
+/**
+* MPU Region Attribute and Size Register Value
+* 
+* \param DisableExec       Instruction access disable bit, 1= disable instruction fetches.
+* \param AccessPermission  Data access permissions, allows you to configure read/write access for User and Privileged mode.
+* \param TypeExtField      Type extension field, allows you to configure memory access type, for example strongly ordered, peripheral.
+* \param IsShareable       Region is shareable between multiple bus masters.
+* \param IsCacheable       Region is cacheable, i.e. its value may be kept in cache.
+* \param IsBufferable      Region is bufferable, i.e. using write-back caching. Cacheable but non-bufferable regions use write-through policy.
+* \param SubRegionDisable  Sub-region disable field.
+* \param Size              Region size of the region to be configured, for example 4K, 8K.
+*/                         
+#define ARM_MPU_RASR(DisableExec, AccessPermission, TypeExtField, IsShareable, IsCacheable, IsBufferable, SubRegionDisable, Size) \
+  ARM_MPU_RASR_EX(DisableExec, AccessPermission, ARM_MPU_ACCESS_(TypeExtField, IsShareable, IsCacheable, IsBufferable), SubRegionDisable, Size)
+
+/**
+* MPU Memory Access Attribute for strongly ordered memory.
+*  - TEX: 000b
+*  - Shareable
+*  - Non-cacheable
+*  - Non-bufferable
+*/ 
+#define ARM_MPU_ACCESS_ORDERED ARM_MPU_ACCESS_(0U, 1U, 0U, 0U)
+
+/**
+* MPU Memory Access Attribute for device memory.
+*  - TEX: 000b (if shareable) or 010b (if non-shareable)
+*  - Shareable or non-shareable
+*  - Non-cacheable
+*  - Bufferable (if shareable) or non-bufferable (if non-shareable)
+*
+* \param IsShareable Configures the device memory as shareable or non-shareable.
+*/ 
+#define ARM_MPU_ACCESS_DEVICE(IsShareable) ((IsShareable) ? ARM_MPU_ACCESS_(0U, 1U, 0U, 1U) : ARM_MPU_ACCESS_(2U, 0U, 0U, 0U))
+
+/**
+* MPU Memory Access Attribute for normal memory.
+*  - TEX: 1BBb (reflecting outer cacheability rules)
+*  - Shareable or non-shareable
+*  - Cacheable or non-cacheable (reflecting inner cacheability rules)
+*  - Bufferable or non-bufferable (reflecting inner cacheability rules)
+*
+* \param OuterCp Configures the outer cache policy.
+* \param InnerCp Configures the inner cache policy.
+* \param IsShareable Configures the memory as shareable or non-shareable.
+*/ 
+#define ARM_MPU_ACCESS_NORMAL(OuterCp, InnerCp, IsShareable) ARM_MPU_ACCESS_((4U | (OuterCp)), IsShareable, ((InnerCp) >> 1U), ((InnerCp) & 1U))
+
+/**
+* MPU Memory Access Attribute non-cacheable policy.
+*/
+#define ARM_MPU_CACHEP_NOCACHE 0U
+
+/**
+* MPU Memory Access Attribute write-back, write and read allocate policy.
+*/
+#define ARM_MPU_CACHEP_WB_WRA 1U
+
+/**
+* MPU Memory Access Attribute write-through, no write allocate policy.
+*/
+#define ARM_MPU_CACHEP_WT_NWA 2U
+
+/**
+* MPU Memory Access Attribute write-back, no write allocate policy.
+*/
+#define ARM_MPU_CACHEP_WB_NWA 3U
+
+
+/**
+* Struct for a single MPU Region
+*/
+typedef struct {
+  uint32_t RBAR; //!< The region base address register value (RBAR)
+  uint32_t RASR; //!< The region attribute and size register value (RASR) \ref MPU_RASR
+} ARM_MPU_Region_t;
+    
+/** Enable the MPU.
+* \param MPU_Control Default access permissions for unconfigured regions.
+*/
+__STATIC_INLINE void ARM_MPU_Enable(uint32_t MPU_Control)
+{
+  __DMB();
+  MPU->CTRL = MPU_Control | MPU_CTRL_ENABLE_Msk;
+#ifdef SCB_SHCSR_MEMFAULTENA_Msk
+  SCB->SHCSR |= SCB_SHCSR_MEMFAULTENA_Msk;
+#endif
+  __DSB();
+  __ISB();
+}
+
+/** Disable the MPU.
+*/
+__STATIC_INLINE void ARM_MPU_Disable(void)
+{
+  __DMB();
+#ifdef SCB_SHCSR_MEMFAULTENA_Msk
+  SCB->SHCSR &= ~SCB_SHCSR_MEMFAULTENA_Msk;
+#endif
+  MPU->CTRL  &= ~MPU_CTRL_ENABLE_Msk;
+  __DSB();
+  __ISB();
+}
+
+/** Clear and disable the given MPU region.
+* \param rnr Region number to be cleared.
+*/
+__STATIC_INLINE void ARM_MPU_ClrRegion(uint32_t rnr)
+{
+  MPU->RNR = rnr;
+  MPU->RASR = 0U;
+}
+
+/** Configure an MPU region.
+* \param rbar Value for RBAR register.
+* \param rsar Value for RSAR register.
+*/   
+__STATIC_INLINE void ARM_MPU_SetRegion(uint32_t rbar, uint32_t rasr)
+{
+  MPU->RBAR = rbar;
+  MPU->RASR = rasr;
+}
+
+/** Configure the given MPU region.
+* \param rnr Region number to be configured.
+* \param rbar Value for RBAR register.
+* \param rsar Value for RSAR register.
+*/   
+__STATIC_INLINE void ARM_MPU_SetRegionEx(uint32_t rnr, uint32_t rbar, uint32_t rasr)
+{
+  MPU->RNR = rnr;
+  MPU->RBAR = rbar;
+  MPU->RASR = rasr;
+}
+
+/** Memcopy with strictly ordered memory access, e.g. for register targets.
+* \param dst Destination data is copied to.
+* \param src Source data is copied from.
+* \param len Amount of data words to be copied.
+*/
+__STATIC_INLINE void ARM_MPU_OrderedMemcpy(volatile uint32_t* dst, const uint32_t* __RESTRICT src, uint32_t len)
+{
+  uint32_t i;
+  for (i = 0U; i < len; ++i) 
+  {
+    dst[i] = src[i];
+  }
+}
+
+/** Load the given number of MPU regions from a table.
+* \param table Pointer to the MPU configuration table.
+* \param cnt Amount of regions to be configured.
+*/
+__STATIC_INLINE void ARM_MPU_Load(ARM_MPU_Region_t const* table, uint32_t cnt) 
+{
+  const uint32_t rowWordSize = sizeof(ARM_MPU_Region_t)/4U;
+  while (cnt > MPU_TYPE_RALIASES) {
+    ARM_MPU_OrderedMemcpy(&(MPU->RBAR), &(table->RBAR), MPU_TYPE_RALIASES*rowWordSize);
+    table += MPU_TYPE_RALIASES;
+    cnt -= MPU_TYPE_RALIASES;
+  }
+  ARM_MPU_OrderedMemcpy(&(MPU->RBAR), &(table->RBAR), cnt*rowWordSize);
+}
+
+#endif

libraries/sdk/third_party/CMSIS/Core/Include/mpu_armv8.h

@@ -0,0 +1,352 @@
+/******************************************************************************
+ * @file     mpu_armv8.h
+ * @brief    CMSIS MPU API for Armv8-M and Armv8.1-M MPU
+ * @version  V5.1.2
+ * @date     10. February 2020
+ ******************************************************************************/
+/*
+ * Copyright (c) 2017-2020 Arm Limited. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#if   defined ( __ICCARM__ )
+  #pragma system_include         /* treat file as system include file for MISRA check */
+#elif defined (__clang__)
+  #pragma clang system_header    /* treat file as system include file */
+#endif
+
+#ifndef ARM_MPU_ARMV8_H
+#define ARM_MPU_ARMV8_H
+
+/** \brief Attribute for device memory (outer only) */
+#define ARM_MPU_ATTR_DEVICE                           ( 0U )
+
+/** \brief Attribute for non-cacheable, normal memory */
+#define ARM_MPU_ATTR_NON_CACHEABLE                    ( 4U )
+
+/** \brief Attribute for normal memory (outer and inner)
+* \param NT Non-Transient: Set to 1 for non-transient data.
+* \param WB Write-Back: Set to 1 to use write-back update policy.
+* \param RA Read Allocation: Set to 1 to use cache allocation on read miss.
+* \param WA Write Allocation: Set to 1 to use cache allocation on write miss.
+*/
+#define ARM_MPU_ATTR_MEMORY_(NT, WB, RA, WA) \
+  ((((NT) & 1U) << 3U) | (((WB) & 1U) << 2U) | (((RA) & 1U) << 1U) | ((WA) & 1U))
+
+/** \brief Device memory type non Gathering, non Re-ordering, non Early Write Acknowledgement */
+#define ARM_MPU_ATTR_DEVICE_nGnRnE (0U)
+
+/** \brief Device memory type non Gathering, non Re-ordering, Early Write Acknowledgement */
+#define ARM_MPU_ATTR_DEVICE_nGnRE  (1U)
+
+/** \brief Device memory type non Gathering, Re-ordering, Early Write Acknowledgement */
+#define ARM_MPU_ATTR_DEVICE_nGRE   (2U)
+
+/** \brief Device memory type Gathering, Re-ordering, Early Write Acknowledgement */
+#define ARM_MPU_ATTR_DEVICE_GRE    (3U)
+
+/** \brief Memory Attribute
+* \param O Outer memory attributes
+* \param I O == ARM_MPU_ATTR_DEVICE: Device memory attributes, else: Inner memory attributes
+*/
+#define ARM_MPU_ATTR(O, I) ((((O) & 0xFU) << 4U) | ((((O) & 0xFU) != 0U) ? ((I) & 0xFU) : (((I) & 0x3U) << 2U)))
+
+/** \brief Normal memory non-shareable  */
+#define ARM_MPU_SH_NON   (0U)
+
+/** \brief Normal memory outer shareable  */
+#define ARM_MPU_SH_OUTER (2U)
+
+/** \brief Normal memory inner shareable  */
+#define ARM_MPU_SH_INNER (3U)
+
+/** \brief Memory access permissions
+* \param RO Read-Only: Set to 1 for read-only memory.
+* \param NP Non-Privileged: Set to 1 for non-privileged memory.
+*/
+#define ARM_MPU_AP_(RO, NP) ((((RO) & 1U) << 1U) | ((NP) & 1U))
+
+/** \brief Region Base Address Register value
+* \param BASE The base address bits [31:5] of a memory region. The value is zero extended. Effective address gets 32 byte aligned.
+* \param SH Defines the Shareability domain for this memory region.
+* \param RO Read-Only: Set to 1 for a read-only memory region.
+* \param NP Non-Privileged: Set to 1 for a non-privileged memory region.
+* \oaram XN eXecute Never: Set to 1 for a non-executable memory region.
+*/
+#define ARM_MPU_RBAR(BASE, SH, RO, NP, XN) \
+  (((BASE) & MPU_RBAR_BASE_Msk) | \
+  (((SH) << MPU_RBAR_SH_Pos) & MPU_RBAR_SH_Msk) | \
+  ((ARM_MPU_AP_(RO, NP) << MPU_RBAR_AP_Pos) & MPU_RBAR_AP_Msk) | \
+  (((XN) << MPU_RBAR_XN_Pos) & MPU_RBAR_XN_Msk))
+
+/** \brief Region Limit Address Register value
+* \param LIMIT The limit address bits [31:5] for this memory region. The value is one extended.
+* \param IDX The attribute index to be associated with this memory region.
+*/
+#define ARM_MPU_RLAR(LIMIT, IDX) \
+  (((LIMIT) & MPU_RLAR_LIMIT_Msk) | \
+  (((IDX) << MPU_RLAR_AttrIndx_Pos) & MPU_RLAR_AttrIndx_Msk) | \
+  (MPU_RLAR_EN_Msk))
+
+#if defined(MPU_RLAR_PXN_Pos)
+  
+/** \brief Region Limit Address Register with PXN value
+* \param LIMIT The limit address bits [31:5] for this memory region. The value is one extended.
+* \param PXN Privileged execute never. Defines whether code can be executed from this privileged region.
+* \param IDX The attribute index to be associated with this memory region.
+*/
+#define ARM_MPU_RLAR_PXN(LIMIT, PXN, IDX) \
+  (((LIMIT) & MPU_RLAR_LIMIT_Msk) | \
+  (((PXN) << MPU_RLAR_PXN_Pos) & MPU_RLAR_PXN_Msk) | \
+  (((IDX) << MPU_RLAR_AttrIndx_Pos) & MPU_RLAR_AttrIndx_Msk) | \
+  (MPU_RLAR_EN_Msk))
+  
+#endif
+
+/**
+* Struct for a single MPU Region
+*/
+typedef struct {
+  uint32_t RBAR;                   /*!< Region Base Address Register value */
+  uint32_t RLAR;                   /*!< Region Limit Address Register value */
+} ARM_MPU_Region_t;
+    
+/** Enable the MPU.
+* \param MPU_Control Default access permissions for unconfigured regions.
+*/
+__STATIC_INLINE void ARM_MPU_Enable(uint32_t MPU_Control)
+{
+  __DMB();
+  MPU->CTRL = MPU_Control | MPU_CTRL_ENABLE_Msk;
+#ifdef SCB_SHCSR_MEMFAULTENA_Msk
+  SCB->SHCSR |= SCB_SHCSR_MEMFAULTENA_Msk;
+#endif
+  __DSB();
+  __ISB();
+}
+
+/** Disable the MPU.
+*/
+__STATIC_INLINE void ARM_MPU_Disable(void)
+{
+  __DMB();
+#ifdef SCB_SHCSR_MEMFAULTENA_Msk
+  SCB->SHCSR &= ~SCB_SHCSR_MEMFAULTENA_Msk;
+#endif
+  MPU->CTRL  &= ~MPU_CTRL_ENABLE_Msk;
+  __DSB();
+  __ISB();
+}
+
+#ifdef MPU_NS
+/** Enable the Non-secure MPU.
+* \param MPU_Control Default access permissions for unconfigured regions.
+*/
+__STATIC_INLINE void ARM_MPU_Enable_NS(uint32_t MPU_Control)
+{
+  __DMB();
+  MPU_NS->CTRL = MPU_Control | MPU_CTRL_ENABLE_Msk;
+#ifdef SCB_SHCSR_MEMFAULTENA_Msk
+  SCB_NS->SHCSR |= SCB_SHCSR_MEMFAULTENA_Msk;
+#endif
+  __DSB();
+  __ISB();
+}
+
+/** Disable the Non-secure MPU.
+*/
+__STATIC_INLINE void ARM_MPU_Disable_NS(void)
+{
+  __DMB();
+#ifdef SCB_SHCSR_MEMFAULTENA_Msk
+  SCB_NS->SHCSR &= ~SCB_SHCSR_MEMFAULTENA_Msk;
+#endif
+  MPU_NS->CTRL  &= ~MPU_CTRL_ENABLE_Msk;
+  __DSB();
+  __ISB();
+}
+#endif
+
+/** Set the memory attribute encoding to the given MPU.
+* \param mpu Pointer to the MPU to be configured.
+* \param idx The attribute index to be set [0-7]
+* \param attr The attribute value to be set.
+*/
+__STATIC_INLINE void ARM_MPU_SetMemAttrEx(MPU_Type* mpu, uint8_t idx, uint8_t attr)
+{
+  const uint8_t reg = idx / 4U;
+  const uint32_t pos = ((idx % 4U) * 8U);
+  const uint32_t mask = 0xFFU << pos;
+  
+  if (reg >= (sizeof(mpu->MAIR) / sizeof(mpu->MAIR[0]))) {
+    return; // invalid index
+  }
+  
+  mpu->MAIR[reg] = ((mpu->MAIR[reg] & ~mask) | ((attr << pos) & mask));
+}
+
+/** Set the memory attribute encoding.
+* \param idx The attribute index to be set [0-7]
+* \param attr The attribute value to be set.
+*/
+__STATIC_INLINE void ARM_MPU_SetMemAttr(uint8_t idx, uint8_t attr)
+{
+  ARM_MPU_SetMemAttrEx(MPU, idx, attr);
+}
+
+#ifdef MPU_NS
+/** Set the memory attribute encoding to the Non-secure MPU.
+* \param idx The attribute index to be set [0-7]
+* \param attr The attribute value to be set.
+*/
+__STATIC_INLINE void ARM_MPU_SetMemAttr_NS(uint8_t idx, uint8_t attr)
+{
+  ARM_MPU_SetMemAttrEx(MPU_NS, idx, attr);
+}
+#endif
+
+/** Clear and disable the given MPU region of the given MPU.
+* \param mpu Pointer to MPU to be used.
+* \param rnr Region number to be cleared.
+*/
+__STATIC_INLINE void ARM_MPU_ClrRegionEx(MPU_Type* mpu, uint32_t rnr)
+{
+  mpu->RNR = rnr;
+  mpu->RLAR = 0U;
+}
+
+/** Clear and disable the given MPU region.
+* \param rnr Region number to be cleared.
+*/
+__STATIC_INLINE void ARM_MPU_ClrRegion(uint32_t rnr)
+{
+  ARM_MPU_ClrRegionEx(MPU, rnr);
+}
+
+#ifdef MPU_NS
+/** Clear and disable the given Non-secure MPU region.
+* \param rnr Region number to be cleared.
+*/
+__STATIC_INLINE void ARM_MPU_ClrRegion_NS(uint32_t rnr)
+{  
+  ARM_MPU_ClrRegionEx(MPU_NS, rnr);
+}
+#endif
+
+/** Configure the given MPU region of the given MPU.
+* \param mpu Pointer to MPU to be used.
+* \param rnr Region number to be configured.
+* \param rbar Value for RBAR register.
+* \param rlar Value for RLAR register.
+*/   
+__STATIC_INLINE void ARM_MPU_SetRegionEx(MPU_Type* mpu, uint32_t rnr, uint32_t rbar, uint32_t rlar)
+{
+  mpu->RNR = rnr;
+  mpu->RBAR = rbar;
+  mpu->RLAR = rlar;
+}
+
+/** Configure the given MPU region.
+* \param rnr Region number to be configured.
+* \param rbar Value for RBAR register.
+* \param rlar Value for RLAR register.
+*/   
+__STATIC_INLINE void ARM_MPU_SetRegion(uint32_t rnr, uint32_t rbar, uint32_t rlar)
+{
+  ARM_MPU_SetRegionEx(MPU, rnr, rbar, rlar);
+}
+
+#ifdef MPU_NS
+/** Configure the given Non-secure MPU region.
+* \param rnr Region number to be configured.
+* \param rbar Value for RBAR register.
+* \param rlar Value for RLAR register.
+*/   
+__STATIC_INLINE void ARM_MPU_SetRegion_NS(uint32_t rnr, uint32_t rbar, uint32_t rlar)
+{
+  ARM_MPU_SetRegionEx(MPU_NS, rnr, rbar, rlar);  
+}
+#endif
+
+/** Memcopy with strictly ordered memory access, e.g. for register targets.
+* \param dst Destination data is copied to.
+* \param src Source data is copied from.
+* \param len Amount of data words to be copied.
+*/
+__STATIC_INLINE void ARM_MPU_OrderedMemcpy(volatile uint32_t* dst, const uint32_t* __RESTRICT src, uint32_t len)
+{
+  uint32_t i;
+  for (i = 0U; i < len; ++i) 
+  {
+    dst[i] = src[i];
+  }
+}
+
+/** Load the given number of MPU regions from a table to the given MPU.
+* \param mpu Pointer to the MPU registers to be used.
+* \param rnr First region number to be configured.
+* \param table Pointer to the MPU configuration table.
+* \param cnt Amount of regions to be configured.
+*/
+__STATIC_INLINE void ARM_MPU_LoadEx(MPU_Type* mpu, uint32_t rnr, ARM_MPU_Region_t const* table, uint32_t cnt) 
+{
+  const uint32_t rowWordSize = sizeof(ARM_MPU_Region_t)/4U;
+  if (cnt == 1U) {
+    mpu->RNR = rnr;
+    ARM_MPU_OrderedMemcpy(&(mpu->RBAR), &(table->RBAR), rowWordSize);
+  } else {
+    uint32_t rnrBase   = rnr & ~(MPU_TYPE_RALIASES-1U);
+    uint32_t rnrOffset = rnr % MPU_TYPE_RALIASES;
+    
+    mpu->RNR = rnrBase;
+    while ((rnrOffset + cnt) > MPU_TYPE_RALIASES) {
+      uint32_t c = MPU_TYPE_RALIASES - rnrOffset;
+      ARM_MPU_OrderedMemcpy(&(mpu->RBAR)+(rnrOffset*2U), &(table->RBAR), c*rowWordSize);
+      table += c;
+      cnt -= c;
+      rnrOffset = 0U;
+      rnrBase += MPU_TYPE_RALIASES;
+      mpu->RNR = rnrBase;
+    }
+    
+    ARM_MPU_OrderedMemcpy(&(mpu->RBAR)+(rnrOffset*2U), &(table->RBAR), cnt*rowWordSize);
+  }
+}
+
+/** Load the given number of MPU regions from a table.
+* \param rnr First region number to be configured.
+* \param table Pointer to the MPU configuration table.
+* \param cnt Amount of regions to be configured.
+*/
+__STATIC_INLINE void ARM_MPU_Load(uint32_t rnr, ARM_MPU_Region_t const* table, uint32_t cnt) 
+{
+  ARM_MPU_LoadEx(MPU, rnr, table, cnt);
+}
+
+#ifdef MPU_NS
+/** Load the given number of MPU regions from a table to the Non-secure MPU.
+* \param rnr First region number to be configured.
+* \param table Pointer to the MPU configuration table.
+* \param cnt Amount of regions to be configured.
+*/
+__STATIC_INLINE void ARM_MPU_Load_NS(uint32_t rnr, ARM_MPU_Region_t const* table, uint32_t cnt) 
+{
+  ARM_MPU_LoadEx(MPU_NS, rnr, table, cnt);
+}
+#endif
+
+#endif
+

libraries/sdk/third_party/CMSIS/Core/Include/pmu_armv8.h

@@ -0,0 +1,337 @@
+/******************************************************************************
+ * @file     pmu_armv8.h
+ * @brief    CMSIS PMU API for Armv8.1-M PMU
+ * @version  V1.0.0
+ * @date     24. March 2020
+ ******************************************************************************/
+/*
+ * Copyright (c) 2020 Arm Limited. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#if   defined ( __ICCARM__ )
+  #pragma system_include         /* treat file as system include file for MISRA check */
+#elif defined (__clang__)
+  #pragma clang system_header    /* treat file as system include file */
+#endif
+
+#ifndef ARM_PMU_ARMV8_H
+#define ARM_PMU_ARMV8_H
+
+/**
+ * \brief PMU Events
+ * \note  See the Armv8.1-M Architecture Reference Manual for full details on these PMU events.
+ * */
+
+#define ARM_PMU_SW_INCR                              0x0000             /*!< Software update to the PMU_SWINC register, architecturally executed and condition code check pass */
+#define ARM_PMU_L1I_CACHE_REFILL                     0x0001             /*!< L1 I-Cache refill */
+#define ARM_PMU_L1D_CACHE_REFILL                     0x0003             /*!< L1 D-Cache refill */
+#define ARM_PMU_L1D_CACHE                            0x0004             /*!< L1 D-Cache access */
+#define ARM_PMU_LD_RETIRED                           0x0006             /*!< Memory-reading instruction architecturally executed and condition code check pass */
+#define ARM_PMU_ST_RETIRED                           0x0007             /*!< Memory-writing instruction architecturally executed and condition code check pass */
+#define ARM_PMU_INST_RETIRED                         0x0008             /*!< Instruction architecturally executed */
+#define ARM_PMU_EXC_TAKEN                            0x0009             /*!< Exception entry */
+#define ARM_PMU_EXC_RETURN                           0x000A             /*!< Exception return instruction architecturally executed and the condition code check pass */
+#define ARM_PMU_PC_WRITE_RETIRED                     0x000C             /*!< Software change to the Program Counter (PC). Instruction is architecturally executed and condition code check pass */
+#define ARM_PMU_BR_IMMED_RETIRED                     0x000D             /*!< Immediate branch architecturally executed */
+#define ARM_PMU_BR_RETURN_RETIRED                    0x000E             /*!< Function return instruction architecturally executed and the condition code check pass */
+#define ARM_PMU_UNALIGNED_LDST_RETIRED               0x000F             /*!< Unaligned memory memory-reading or memory-writing instruction architecturally executed and condition code check pass */
+#define ARM_PMU_BR_MIS_PRED                          0x0010             /*!< Mispredicted or not predicted branch speculatively executed */
+#define ARM_PMU_CPU_CYCLES                           0x0011             /*!< Cycle */
+#define ARM_PMU_BR_PRED                              0x0012             /*!< Predictable branch speculatively executed */
+#define ARM_PMU_MEM_ACCESS                           0x0013             /*!< Data memory access */
+#define ARM_PMU_L1I_CACHE                            0x0014             /*!< Level 1 instruction cache access */
+#define ARM_PMU_L1D_CACHE_WB                         0x0015             /*!< Level 1 data cache write-back */
+#define ARM_PMU_L2D_CACHE                            0x0016             /*!< Level 2 data cache access */
+#define ARM_PMU_L2D_CACHE_REFILL                     0x0017             /*!< Level 2 data cache refill */
+#define ARM_PMU_L2D_CACHE_WB                         0x0018             /*!< Level 2 data cache write-back */
+#define ARM_PMU_BUS_ACCESS                           0x0019             /*!< Bus access */
+#define ARM_PMU_MEMORY_ERROR                         0x001A             /*!< Local memory error */
+#define ARM_PMU_INST_SPEC                            0x001B             /*!< Instruction speculatively executed */
+#define ARM_PMU_BUS_CYCLES                           0x001D             /*!< Bus cycles */
+#define ARM_PMU_CHAIN                                0x001E             /*!< For an odd numbered counter, increment when an overflow occurs on the preceding even-numbered counter on the same PE */
+#define ARM_PMU_L1D_CACHE_ALLOCATE                   0x001F             /*!< Level 1 data cache allocation without refill */
+#define ARM_PMU_L2D_CACHE_ALLOCATE                   0x0020             /*!< Level 2 data cache allocation without refill */
+#define ARM_PMU_BR_RETIRED                           0x0021             /*!< Branch instruction architecturally executed */
+#define ARM_PMU_BR_MIS_PRED_RETIRED                  0x0022             /*!< Mispredicted branch instruction architecturally executed */
+#define ARM_PMU_STALL_FRONTEND                       0x0023             /*!< No operation issued because of the frontend */
+#define ARM_PMU_STALL_BACKEND                        0x0024             /*!< No operation issued because of the backend */
+#define ARM_PMU_L2I_CACHE                            0x0027             /*!< Level 2 instruction cache access */
+#define ARM_PMU_L2I_CACHE_REFILL                     0x0028             /*!< Level 2 instruction cache refill */
+#define ARM_PMU_L3D_CACHE_ALLOCATE                   0x0029             /*!< Level 3 data cache allocation without refill */
+#define ARM_PMU_L3D_CACHE_REFILL                     0x002A             /*!< Level 3 data cache refill */
+#define ARM_PMU_L3D_CACHE                            0x002B             /*!< Level 3 data cache access */
+#define ARM_PMU_L3D_CACHE_WB                         0x002C             /*!< Level 3 data cache write-back */
+#define ARM_PMU_LL_CACHE_RD                          0x0036             /*!< Last level data cache read */
+#define ARM_PMU_LL_CACHE_MISS_RD                     0x0037             /*!< Last level data cache read miss */
+#define ARM_PMU_L1D_CACHE_MISS_RD                    0x0039             /*!< Level 1 data cache read miss */
+#define ARM_PMU_OP_COMPLETE                          0x003A             /*!< Operation retired */
+#define ARM_PMU_OP_SPEC                              0x003B             /*!< Operation speculatively executed */
+#define ARM_PMU_STALL                                0x003C             /*!< Stall cycle for instruction or operation not sent for execution */
+#define ARM_PMU_STALL_OP_BACKEND                     0x003D             /*!< Stall cycle for instruction or operation not sent for execution due to pipeline backend */
+#define ARM_PMU_STALL_OP_FRONTEND                    0x003E             /*!< Stall cycle for instruction or operation not sent for execution due to pipeline frontend */
+#define ARM_PMU_STALL_OP                             0x003F             /*!< Instruction or operation slots not occupied each cycle */
+#define ARM_PMU_L1D_CACHE_RD                         0x0040             /*!< Level 1 data cache read */
+#define ARM_PMU_LE_RETIRED                           0x0100             /*!< Loop end instruction executed */
+#define ARM_PMU_LE_SPEC                              0x0101             /*!< Loop end instruction speculatively executed */
+#define ARM_PMU_BF_RETIRED                           0x0104             /*!< Branch future instruction architecturally executed and condition code check pass */
+#define ARM_PMU_BF_SPEC                              0x0105             /*!< Branch future instruction speculatively executed and condition code check pass */
+#define ARM_PMU_LE_CANCEL                            0x0108             /*!< Loop end instruction not taken */
+#define ARM_PMU_BF_CANCEL                            0x0109             /*!< Branch future instruction not taken */
+#define ARM_PMU_SE_CALL_S                            0x0114             /*!< Call to secure function, resulting in Security state change */
+#define ARM_PMU_SE_CALL_NS                           0x0115             /*!< Call to non-secure function, resulting in Security state change */
+#define ARM_PMU_DWT_CMPMATCH0                        0x0118             /*!< DWT comparator 0 match */
+#define ARM_PMU_DWT_CMPMATCH1                        0x0119             /*!< DWT comparator 1 match */
+#define ARM_PMU_DWT_CMPMATCH2                        0x011A             /*!< DWT comparator 2 match */
+#define ARM_PMU_DWT_CMPMATCH3                        0x011B             /*!< DWT comparator 3 match */
+#define ARM_PMU_MVE_INST_RETIRED                     0x0200             /*!< MVE instruction architecturally executed */
+#define ARM_PMU_MVE_INST_SPEC                        0x0201             /*!< MVE instruction speculatively executed */
+#define ARM_PMU_MVE_FP_RETIRED                       0x0204             /*!< MVE floating-point instruction architecturally executed */
+#define ARM_PMU_MVE_FP_SPEC                          0x0205             /*!< MVE floating-point instruction speculatively executed */
+#define ARM_PMU_MVE_FP_HP_RETIRED                    0x0208             /*!< MVE half-precision floating-point instruction architecturally executed */
+#define ARM_PMU_MVE_FP_HP_SPEC                       0x0209             /*!< MVE half-precision floating-point instruction speculatively executed */
+#define ARM_PMU_MVE_FP_SP_RETIRED                    0x020C             /*!< MVE single-precision floating-point instruction architecturally executed */
+#define ARM_PMU_MVE_FP_SP_SPEC                       0x020D             /*!< MVE single-precision floating-point instruction speculatively executed */
+#define ARM_PMU_MVE_FP_MAC_RETIRED                   0x0214             /*!< MVE floating-point multiply or multiply-accumulate instruction architecturally executed */
+#define ARM_PMU_MVE_FP_MAC_SPEC                      0x0215             /*!< MVE floating-point multiply or multiply-accumulate instruction speculatively executed */
+#define ARM_PMU_MVE_INT_RETIRED                      0x0224             /*!< MVE integer instruction architecturally executed */
+#define ARM_PMU_MVE_INT_SPEC                         0x0225             /*!< MVE integer instruction speculatively executed */
+#define ARM_PMU_MVE_INT_MAC_RETIRED                  0x0228             /*!< MVE multiply or multiply-accumulate instruction architecturally executed */
+#define ARM_PMU_MVE_INT_MAC_SPEC                     0x0229             /*!< MVE multiply or multiply-accumulate instruction speculatively executed */
+#define ARM_PMU_MVE_LDST_RETIRED                     0x0238             /*!< MVE load or store instruction architecturally executed */
+#define ARM_PMU_MVE_LDST_SPEC                        0x0239             /*!< MVE load or store instruction speculatively executed */
+#define ARM_PMU_MVE_LD_RETIRED                       0x023C             /*!< MVE load instruction architecturally executed */
+#define ARM_PMU_MVE_LD_SPEC                          0x023D             /*!< MVE load instruction speculatively executed */
+#define ARM_PMU_MVE_ST_RETIRED                       0x0240             /*!< MVE store instruction architecturally executed */
+#define ARM_PMU_MVE_ST_SPEC                          0x0241             /*!< MVE store instruction speculatively executed */
+#define ARM_PMU_MVE_LDST_CONTIG_RETIRED              0x0244             /*!< MVE contiguous load or store instruction architecturally executed */
+#define ARM_PMU_MVE_LDST_CONTIG_SPEC                 0x0245             /*!< MVE contiguous load or store instruction speculatively executed */
+#define ARM_PMU_MVE_LD_CONTIG_RETIRED                0x0248             /*!< MVE contiguous load instruction architecturally executed */
+#define ARM_PMU_MVE_LD_CONTIG_SPEC                   0x0249             /*!< MVE contiguous load instruction speculatively executed */
+#define ARM_PMU_MVE_ST_CONTIG_RETIRED                0x024C             /*!< MVE contiguous store instruction architecturally executed */
+#define ARM_PMU_MVE_ST_CONTIG_SPEC                   0x024D             /*!< MVE contiguous store instruction speculatively executed */
+#define ARM_PMU_MVE_LDST_NONCONTIG_RETIRED           0x0250             /*!< MVE non-contiguous load or store instruction architecturally executed */
+#define ARM_PMU_MVE_LDST_NONCONTIG_SPEC              0x0251             /*!< MVE non-contiguous load or store instruction speculatively executed */
+#define ARM_PMU_MVE_LD_NONCONTIG_RETIRED             0x0254             /*!< MVE non-contiguous load instruction architecturally executed */
+#define ARM_PMU_MVE_LD_NONCONTIG_SPEC                0x0255             /*!< MVE non-contiguous load instruction speculatively executed */
+#define ARM_PMU_MVE_ST_NONCONTIG_RETIRED             0x0258             /*!< MVE non-contiguous store instruction architecturally executed */
+#define ARM_PMU_MVE_ST_NONCONTIG_SPEC                0x0259             /*!< MVE non-contiguous store instruction speculatively executed */
+#define ARM_PMU_MVE_LDST_MULTI_RETIRED               0x025C             /*!< MVE memory instruction targeting multiple registers architecturally executed */
+#define ARM_PMU_MVE_LDST_MULTI_SPEC                  0x025D             /*!< MVE memory instruction targeting multiple registers speculatively executed */
+#define ARM_PMU_MVE_LD_MULTI_RETIRED                 0x0260             /*!< MVE memory load instruction targeting multiple registers architecturally executed */
+#define ARM_PMU_MVE_LD_MULTI_SPEC                    0x0261             /*!< MVE memory load instruction targeting multiple registers speculatively executed */
+#define ARM_PMU_MVE_ST_MULTI_RETIRED                 0x0261             /*!< MVE memory store instruction targeting multiple registers architecturally executed */
+#define ARM_PMU_MVE_ST_MULTI_SPEC                    0x0265             /*!< MVE memory store instruction targeting multiple registers speculatively executed */
+#define ARM_PMU_MVE_LDST_UNALIGNED_RETIRED           0x028C             /*!< MVE unaligned memory load or store instruction architecturally executed */
+#define ARM_PMU_MVE_LDST_UNALIGNED_SPEC              0x028D             /*!< MVE unaligned memory load or store instruction speculatively executed */
+#define ARM_PMU_MVE_LD_UNALIGNED_RETIRED             0x0290             /*!< MVE unaligned load instruction architecturally executed */
+#define ARM_PMU_MVE_LD_UNALIGNED_SPEC                0x0291             /*!< MVE unaligned load instruction speculatively executed */
+#define ARM_PMU_MVE_ST_UNALIGNED_RETIRED             0x0294             /*!< MVE unaligned store instruction architecturally executed */
+#define ARM_PMU_MVE_ST_UNALIGNED_SPEC                0x0295             /*!< MVE unaligned store instruction speculatively executed */
+#define ARM_PMU_MVE_LDST_UNALIGNED_NONCONTIG_RETIRED 0x0298             /*!< MVE unaligned noncontiguous load or store instruction architecturally executed */
+#define ARM_PMU_MVE_LDST_UNALIGNED_NONCONTIG_SPEC    0x0299             /*!< MVE unaligned noncontiguous load or store instruction speculatively executed */
+#define ARM_PMU_MVE_VREDUCE_RETIRED                  0x02A0             /*!< MVE vector reduction instruction architecturally executed */
+#define ARM_PMU_MVE_VREDUCE_SPEC                     0x02A1             /*!< MVE vector reduction instruction speculatively executed */
+#define ARM_PMU_MVE_VREDUCE_FP_RETIRED               0x02A4             /*!< MVE floating-point vector reduction instruction architecturally executed */
+#define ARM_PMU_MVE_VREDUCE_FP_SPEC                  0x02A5             /*!< MVE floating-point vector reduction instruction speculatively executed */
+#define ARM_PMU_MVE_VREDUCE_INT_RETIRED              0x02A8             /*!< MVE integer vector reduction instruction architecturally executed */
+#define ARM_PMU_MVE_VREDUCE_INT_SPEC                 0x02A9             /*!< MVE integer vector reduction instruction speculatively executed */
+#define ARM_PMU_MVE_PRED                             0x02B8             /*!< Cycles where one or more predicated beats architecturally executed */
+#define ARM_PMU_MVE_STALL                            0x02CC             /*!< Stall cycles caused by an MVE instruction */
+#define ARM_PMU_MVE_STALL_RESOURCE                   0x02CD             /*!< Stall cycles caused by an MVE instruction because of resource conflicts */
+#define ARM_PMU_MVE_STALL_RESOURCE_MEM               0x02CE             /*!< Stall cycles caused by an MVE instruction because of memory resource conflicts */
+#define ARM_PMU_MVE_STALL_RESOURCE_FP                0x02CF             /*!< Stall cycles caused by an MVE instruction because of floating-point resource conflicts */
+#define ARM_PMU_MVE_STALL_RESOURCE_INT               0x02D0             /*!< Stall cycles caused by an MVE instruction because of integer resource conflicts */
+#define ARM_PMU_MVE_STALL_BREAK                      0x02D3             /*!< Stall cycles caused by an MVE chain break */
+#define ARM_PMU_MVE_STALL_DEPENDENCY                 0x02D4             /*!< Stall cycles caused by MVE register dependency */
+#define ARM_PMU_ITCM_ACCESS                          0x4007             /*!< Instruction TCM access */
+#define ARM_PMU_DTCM_ACCESS                          0x4008             /*!< Data TCM access */
+#define ARM_PMU_TRCEXTOUT0                           0x4010             /*!< ETM external output 0 */
+#define ARM_PMU_TRCEXTOUT1                           0x4011             /*!< ETM external output 1 */
+#define ARM_PMU_TRCEXTOUT2                           0x4012             /*!< ETM external output 2 */
+#define ARM_PMU_TRCEXTOUT3                           0x4013             /*!< ETM external output 3 */
+#define ARM_PMU_CTI_TRIGOUT4                         0x4018             /*!< Cross-trigger Interface output trigger 4 */
+#define ARM_PMU_CTI_TRIGOUT5                         0x4019             /*!< Cross-trigger Interface output trigger 5 */
+#define ARM_PMU_CTI_TRIGOUT6                         0x401A             /*!< Cross-trigger Interface output trigger 6 */
+#define ARM_PMU_CTI_TRIGOUT7                         0x401B             /*!< Cross-trigger Interface output trigger 7 */
+
+/** \brief PMU Functions */
+
+__STATIC_INLINE void ARM_PMU_Enable(void);
+__STATIC_INLINE void ARM_PMU_Disable(void);
+
+__STATIC_INLINE void ARM_PMU_Set_EVTYPER(uint32_t num, uint32_t type);
+
+__STATIC_INLINE void ARM_PMU_CYCCNT_Reset(void);
+__STATIC_INLINE void ARM_PMU_EVCNTR_ALL_Reset(void);
+
+__STATIC_INLINE void ARM_PMU_CNTR_Enable(uint32_t mask);
+__STATIC_INLINE void ARM_PMU_CNTR_Disable(uint32_t mask);
+
+__STATIC_INLINE uint32_t ARM_PMU_Get_CCNTR(void);
+__STATIC_INLINE uint32_t ARM_PMU_Get_EVCNTR(uint32_t num);
+
+__STATIC_INLINE uint32_t ARM_PMU_Get_CNTR_OVS(void);
+__STATIC_INLINE void ARM_PMU_Set_CNTR_OVS(uint32_t mask);
+
+__STATIC_INLINE void ARM_PMU_Set_CNTR_IRQ_Enable(uint32_t mask);
+__STATIC_INLINE void ARM_PMU_Set_CNTR_IRQ_Disable(uint32_t mask);
+
+__STATIC_INLINE void ARM_PMU_CNTR_Increment(uint32_t mask);
+
+/** 
+  \brief   Enable the PMU
+*/
+__STATIC_INLINE void ARM_PMU_Enable(void) 
+{
+  PMU->CTRL |= PMU_CTRL_ENABLE_Msk;
+}
+
+/** 
+  \brief   Disable the PMU
+*/
+__STATIC_INLINE void ARM_PMU_Disable(void) 
+{
+  PMU->CTRL &= ~PMU_CTRL_ENABLE_Msk;
+}
+
+/** 
+  \brief   Set event to count for PMU eventer counter
+  \param [in]    num     Event counter (0-30) to configure
+  \param [in]    type    Event to count
+*/
+__STATIC_INLINE void ARM_PMU_Set_EVTYPER(uint32_t num, uint32_t type)
+{
+  PMU->EVTYPER[num] = type;
+}
+
+/** 
+  \brief  Reset cycle counter
+*/
+__STATIC_INLINE void ARM_PMU_CYCCNT_Reset(void)
+{
+  PMU->CTRL |= PMU_CTRL_CYCCNT_RESET_Msk;
+}
+
+/** 
+  \brief  Reset all event counters
+*/
+__STATIC_INLINE void ARM_PMU_EVCNTR_ALL_Reset(void)
+{
+  PMU->CTRL |= PMU_CTRL_EVENTCNT_RESET_Msk;
+}
+
+/** 
+  \brief  Enable counters 
+  \param [in]     mask    Counters to enable
+  \note   Enables one or more of the following:
+          - event counters (0-30)
+          - cycle counter
+*/
+__STATIC_INLINE void ARM_PMU_CNTR_Enable(uint32_t mask)
+{
+  PMU->CNTENSET = mask;
+}
+
+/** 
+  \brief  Disable counters
+  \param [in]     mask    Counters to enable
+  \note   Disables one or more of the following:
+          - event counters (0-30)
+          - cycle counter
+*/
+__STATIC_INLINE void ARM_PMU_CNTR_Disable(uint32_t mask)
+{
+  PMU->CNTENCLR = mask;
+}
+
+/** 
+  \brief  Read cycle counter
+  \return                 Cycle count
+*/
+__STATIC_INLINE uint32_t ARM_PMU_Get_CCNTR(void)
+{
+  return PMU->CCNTR;
+}
+
+/** 
+  \brief   Read event counter
+  \param [in]     num     Event counter (0-30) to read
+  \return                 Event count
+*/
+__STATIC_INLINE uint32_t ARM_PMU_Get_EVCNTR(uint32_t num)
+{
+  return PMU->EVCNTR[num];
+}
+
+/** 
+  \brief   Read counter overflow status
+  \return  Counter overflow status bits for the following:
+          - event counters (0-30)
+          - cycle counter
+*/
+__STATIC_INLINE uint32_t ARM_PMU_Get_CNTR_OVS(void)
+{
+  return PMU->OVSSET;	
+}
+
+/** 
+  \brief   Clear counter overflow status
+  \param [in]     mask    Counter overflow status bits to clear
+  \note    Clears overflow status bits for one or more of the following:
+           - event counters (0-30)
+           - cycle counter
+*/
+__STATIC_INLINE void ARM_PMU_Set_CNTR_OVS(uint32_t mask)
+{
+  PMU->OVSCLR = mask;
+}
+
+/** 
+  \brief   Enable counter overflow interrupt request 
+  \param [in]     mask    Counter overflow interrupt request bits to set
+  \note    Sets overflow interrupt request bits for one or more of the following:
+           - event counters (0-30)
+           - cycle counter
+*/
+__STATIC_INLINE void ARM_PMU_Set_CNTR_IRQ_Enable(uint32_t mask)
+{
+  PMU->INTENSET = mask;
+}
+
+/** 
+  \brief   Disable counter overflow interrupt request 
+  \param [in]     mask    Counter overflow interrupt request bits to clear
+  \note    Clears overflow interrupt request bits for one or more of the following:
+           - event counters (0-30)
+           - cycle counter
+*/
+__STATIC_INLINE void ARM_PMU_Set_CNTR_IRQ_Disable(uint32_t mask)
+{
+  PMU->INTENCLR = mask;
+}
+
+/** 
+  \brief   Software increment event counter 
+  \param [in]     mask    Counters to increment
+  \note    Software increment bits for one or more event counters (0-30)
+*/
+__STATIC_INLINE void ARM_PMU_CNTR_Increment(uint32_t mask)
+{
+  PMU->SWINC = mask;
+}
+
+#endif

libraries/sdk/third_party/CMSIS/Core/Include/tz_context.h

@@ -0,0 +1,70 @@
+/******************************************************************************
+ * @file     tz_context.h
+ * @brief    Context Management for Armv8-M TrustZone
+ * @version  V1.0.1
+ * @date     10. January 2018
+ ******************************************************************************/
+/*
+ * Copyright (c) 2017-2018 Arm Limited. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#if   defined ( __ICCARM__ )
+  #pragma system_include         /* treat file as system include file for MISRA check */
+#elif defined (__clang__)
+  #pragma clang system_header   /* treat file as system include file */
+#endif
+
+#ifndef TZ_CONTEXT_H
+#define TZ_CONTEXT_H
+ 
+#include <stdint.h>
+ 
+#ifndef TZ_MODULEID_T
+#define TZ_MODULEID_T
+/// \details Data type that identifies secure software modules called by a process.
+typedef uint32_t TZ_ModuleId_t;
+#endif
+ 
+/// \details TZ Memory ID identifies an allocated memory slot.
+typedef uint32_t TZ_MemoryId_t;
+  
+/// Initialize secure context memory system
+/// \return execution status (1: success, 0: error)
+uint32_t TZ_InitContextSystem_S (void);
+ 
+/// Allocate context memory for calling secure software modules in TrustZone
+/// \param[in]  module   identifies software modules called from non-secure mode
+/// \return value != 0 id TrustZone memory slot identifier
+/// \return value 0    no memory available or internal error
+TZ_MemoryId_t TZ_AllocModuleContext_S (TZ_ModuleId_t module);
+ 
+/// Free context memory that was previously allocated with \ref TZ_AllocModuleContext_S
+/// \param[in]  id  TrustZone memory slot identifier
+/// \return execution status (1: success, 0: error)
+uint32_t TZ_FreeModuleContext_S (TZ_MemoryId_t id);
+ 
+/// Load secure context (called on RTOS thread context switch)
+/// \param[in]  id  TrustZone memory slot identifier
+/// \return execution status (1: success, 0: error)
+uint32_t TZ_LoadContext_S (TZ_MemoryId_t id);
+ 
+/// Store secure context (called on RTOS thread context switch)
+/// \param[in]  id  TrustZone memory slot identifier
+/// \return execution status (1: success, 0: error)
+uint32_t TZ_StoreContext_S (TZ_MemoryId_t id);
+ 
+#endif  // TZ_CONTEXT_H

libraries/sdk/third_party/CMSIS/DSP/Include/arm_common_tables.h

@@ -0,0 +1,517 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_common_tables.h
+ * Description:  Extern declaration for common tables
+ *
+ * $Date:        27. January 2017
+ * $Revision:    V.1.5.1
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2017 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef _ARM_COMMON_TABLES_H
+#define _ARM_COMMON_TABLES_H
+
+#include "arm_math.h"
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_FFT_ALLOW_TABLES)
+  /* Double Precision Float CFFT twiddles */
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREV_1024)
+    extern const uint16_t armBitRevTable[1024];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F64_16)
+    extern const uint64_t twiddleCoefF64_16[32];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F64_32)
+    extern const uint64_t twiddleCoefF64_32[64];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F64_64)
+    extern const uint64_t twiddleCoefF64_64[128];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F64_128)
+    extern const uint64_t twiddleCoefF64_128[256];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F64_256)
+    extern const uint64_t twiddleCoefF64_256[512];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F64_512)
+    extern const uint64_t twiddleCoefF64_512[1024];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F64_1024)
+    extern const uint64_t twiddleCoefF64_1024[2048];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F64_2048)
+    extern const uint64_t twiddleCoefF64_2048[4096];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F64_4096)
+    extern const uint64_t twiddleCoefF64_4096[8192];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_16)
+    extern const float32_t twiddleCoef_16[32];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_32)
+    extern const float32_t twiddleCoef_32[64];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_64)
+    extern const float32_t twiddleCoef_64[128];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_128)
+    extern const float32_t twiddleCoef_128[256];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_256)
+    extern const float32_t twiddleCoef_256[512];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_512)
+    extern const float32_t twiddleCoef_512[1024];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_1024)
+    extern const float32_t twiddleCoef_1024[2048];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_2048)
+    extern const float32_t twiddleCoef_2048[4096];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_4096)
+    extern const float32_t twiddleCoef_4096[8192];
+    #define twiddleCoef twiddleCoef_4096
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_16)
+    extern const q31_t twiddleCoef_16_q31[24];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_32)
+    extern const q31_t twiddleCoef_32_q31[48];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_64)
+    extern const q31_t twiddleCoef_64_q31[96];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_128)
+    extern const q31_t twiddleCoef_128_q31[192];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_256)
+    extern const q31_t twiddleCoef_256_q31[384];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_512)
+    extern const q31_t twiddleCoef_512_q31[768];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_1024)
+    extern const q31_t twiddleCoef_1024_q31[1536];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_2048)
+    extern const q31_t twiddleCoef_2048_q31[3072];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_4096)
+    extern const q31_t twiddleCoef_4096_q31[6144];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_16)
+    extern const q15_t twiddleCoef_16_q15[24];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_32)
+    extern const q15_t twiddleCoef_32_q15[48];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_64)
+    extern const q15_t twiddleCoef_64_q15[96];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_128)
+    extern const q15_t twiddleCoef_128_q15[192];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_256)
+    extern const q15_t twiddleCoef_256_q15[384];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_512)
+    extern const q15_t twiddleCoef_512_q15[768];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_1024)
+    extern const q15_t twiddleCoef_1024_q15[1536];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_2048)
+    extern const q15_t twiddleCoef_2048_q15[3072];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_4096)
+    extern const q15_t twiddleCoef_4096_q15[6144];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  /* Double Precision Float RFFT twiddles */
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_32)
+    extern const uint64_t twiddleCoefF64_rfft_32[32];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_64)
+    extern const uint64_t twiddleCoefF64_rfft_64[64];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_128)
+    extern const uint64_t twiddleCoefF64_rfft_128[128];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_256)
+    extern const uint64_t twiddleCoefF64_rfft_256[256];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_512)
+    extern const uint64_t twiddleCoefF64_rfft_512[512];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_1024)
+    extern const uint64_t twiddleCoefF64_rfft_1024[1024];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_2048)
+    extern const uint64_t twiddleCoefF64_rfft_2048[2048];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_4096)
+    extern const uint64_t twiddleCoefF64_rfft_4096[4096];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_32)
+    extern const float32_t twiddleCoef_rfft_32[32];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_64)
+    extern const float32_t twiddleCoef_rfft_64[64];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_128)
+    extern const float32_t twiddleCoef_rfft_128[128];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_256)
+    extern const float32_t twiddleCoef_rfft_256[256];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_512)
+    extern const float32_t twiddleCoef_rfft_512[512];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_1024)
+    extern const float32_t twiddleCoef_rfft_1024[1024];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_2048)
+    extern const float32_t twiddleCoef_rfft_2048[2048];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_4096)
+    extern const float32_t twiddleCoef_rfft_4096[4096];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+
+  /* Double precision floating-point bit reversal tables */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT64_16)
+    #define ARMBITREVINDEXTABLEF64_16_TABLE_LENGTH ((uint16_t)12)
+    extern const uint16_t armBitRevIndexTableF64_16[ARMBITREVINDEXTABLEF64_16_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT64_32)
+    #define ARMBITREVINDEXTABLEF64_32_TABLE_LENGTH ((uint16_t)24)
+    extern const uint16_t armBitRevIndexTableF64_32[ARMBITREVINDEXTABLEF64_32_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT64_64)
+    #define ARMBITREVINDEXTABLEF64_64_TABLE_LENGTH ((uint16_t)56)
+    extern const uint16_t armBitRevIndexTableF64_64[ARMBITREVINDEXTABLEF64_64_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT64_128)
+    #define ARMBITREVINDEXTABLEF64_128_TABLE_LENGTH ((uint16_t)112)
+    extern const uint16_t armBitRevIndexTableF64_128[ARMBITREVINDEXTABLEF64_128_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT64_256)
+    #define ARMBITREVINDEXTABLEF64_256_TABLE_LENGTH ((uint16_t)240)
+    extern const uint16_t armBitRevIndexTableF64_256[ARMBITREVINDEXTABLEF64_256_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT64_512)
+    #define ARMBITREVINDEXTABLEF64_512_TABLE_LENGTH ((uint16_t)480)
+    extern const uint16_t armBitRevIndexTableF64_512[ARMBITREVINDEXTABLEF64_512_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT64_1024)
+    #define ARMBITREVINDEXTABLEF64_1024_TABLE_LENGTH ((uint16_t)992)
+    extern const uint16_t armBitRevIndexTableF64_1024[ARMBITREVINDEXTABLEF64_1024_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT64_2048)
+    #define ARMBITREVINDEXTABLEF64_2048_TABLE_LENGTH ((uint16_t)1984)
+    extern const uint16_t armBitRevIndexTableF64_2048[ARMBITREVINDEXTABLEF64_2048_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT64_4096)
+    #define ARMBITREVINDEXTABLEF64_4096_TABLE_LENGTH ((uint16_t)4032)
+    extern const uint16_t armBitRevIndexTableF64_4096[ARMBITREVINDEXTABLEF64_4096_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+  /* floating-point bit reversal tables */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT_16)
+    #define ARMBITREVINDEXTABLE_16_TABLE_LENGTH ((uint16_t)20)
+    extern const uint16_t armBitRevIndexTable16[ARMBITREVINDEXTABLE_16_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT_32)
+    #define ARMBITREVINDEXTABLE_32_TABLE_LENGTH ((uint16_t)48)
+    extern const uint16_t armBitRevIndexTable32[ARMBITREVINDEXTABLE_32_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT_64)
+    #define ARMBITREVINDEXTABLE_64_TABLE_LENGTH ((uint16_t)56)
+    extern const uint16_t armBitRevIndexTable64[ARMBITREVINDEXTABLE_64_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT_128)
+    #define ARMBITREVINDEXTABLE_128_TABLE_LENGTH ((uint16_t)208)
+    extern const uint16_t armBitRevIndexTable128[ARMBITREVINDEXTABLE_128_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT_256)
+    #define ARMBITREVINDEXTABLE_256_TABLE_LENGTH ((uint16_t)440)
+    extern const uint16_t armBitRevIndexTable256[ARMBITREVINDEXTABLE_256_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT_512)
+    #define ARMBITREVINDEXTABLE_512_TABLE_LENGTH ((uint16_t)448)
+    extern const uint16_t armBitRevIndexTable512[ARMBITREVINDEXTABLE_512_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT_1024)
+    #define ARMBITREVINDEXTABLE_1024_TABLE_LENGTH ((uint16_t)1800)
+    extern const uint16_t armBitRevIndexTable1024[ARMBITREVINDEXTABLE_1024_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT_2048)
+    #define ARMBITREVINDEXTABLE_2048_TABLE_LENGTH ((uint16_t)3808)
+    extern const uint16_t armBitRevIndexTable2048[ARMBITREVINDEXTABLE_2048_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FLT_4096)
+    #define ARMBITREVINDEXTABLE_4096_TABLE_LENGTH ((uint16_t)4032)
+    extern const uint16_t armBitRevIndexTable4096[ARMBITREVINDEXTABLE_4096_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+
+  /* fixed-point bit reversal tables */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FXT_16)
+    #define ARMBITREVINDEXTABLE_FIXED_16_TABLE_LENGTH ((uint16_t)12)
+    extern const uint16_t armBitRevIndexTable_fixed_16[ARMBITREVINDEXTABLE_FIXED_16_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FXT_32)
+    #define ARMBITREVINDEXTABLE_FIXED_32_TABLE_LENGTH ((uint16_t)24)
+    extern const uint16_t armBitRevIndexTable_fixed_32[ARMBITREVINDEXTABLE_FIXED_32_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FXT_64)
+    #define ARMBITREVINDEXTABLE_FIXED_64_TABLE_LENGTH ((uint16_t)56)
+    extern const uint16_t armBitRevIndexTable_fixed_64[ARMBITREVINDEXTABLE_FIXED_64_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FXT_128)
+    #define ARMBITREVINDEXTABLE_FIXED_128_TABLE_LENGTH ((uint16_t)112)
+    extern const uint16_t armBitRevIndexTable_fixed_128[ARMBITREVINDEXTABLE_FIXED_128_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FXT_256)
+    #define ARMBITREVINDEXTABLE_FIXED_256_TABLE_LENGTH ((uint16_t)240)
+    extern const uint16_t armBitRevIndexTable_fixed_256[ARMBITREVINDEXTABLE_FIXED_256_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FXT_512)
+    #define ARMBITREVINDEXTABLE_FIXED_512_TABLE_LENGTH ((uint16_t)480)
+    extern const uint16_t armBitRevIndexTable_fixed_512[ARMBITREVINDEXTABLE_FIXED_512_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FXT_1024)
+    #define ARMBITREVINDEXTABLE_FIXED_1024_TABLE_LENGTH ((uint16_t)992)
+    extern const uint16_t armBitRevIndexTable_fixed_1024[ARMBITREVINDEXTABLE_FIXED_1024_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FXT_2048)
+    #define ARMBITREVINDEXTABLE_FIXED_2048_TABLE_LENGTH ((uint16_t)1984)
+    extern const uint16_t armBitRevIndexTable_fixed_2048[ARMBITREVINDEXTABLE_FIXED_2048_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_BITREVIDX_FXT_4096)
+    #define ARMBITREVINDEXTABLE_FIXED_4096_TABLE_LENGTH ((uint16_t)4032)
+    extern const uint16_t armBitRevIndexTable_fixed_4096[ARMBITREVINDEXTABLE_FIXED_4096_TABLE_LENGTH];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_REALCOEF_F32)
+    extern const float32_t realCoefA[8192];
+    extern const float32_t realCoefB[8192];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_REALCOEF_Q31)
+    extern const q31_t realCoefAQ31[8192];
+    extern const q31_t realCoefBQ31[8192];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_REALCOEF_Q15)
+    extern const q15_t realCoefAQ15[8192];
+    extern const q15_t realCoefBQ15[8192];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_DCT4_F32_128)
+    extern const float32_t Weights_128[256];
+    extern const float32_t cos_factors_128[128];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_DCT4_F32_512)
+    extern const float32_t Weights_512[1024];
+    extern const float32_t cos_factors_512[512];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_DCT4_F32_2048)
+    extern const float32_t Weights_2048[4096];
+    extern const float32_t cos_factors_2048[2048];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_DCT4_F32_8192)
+    extern const float32_t Weights_8192[16384];
+    extern const float32_t cos_factors_8192[8192];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_DCT4_Q15_128)
+    extern const q15_t WeightsQ15_128[256];
+    extern const q15_t cos_factorsQ15_128[128];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_DCT4_Q15_512)
+    extern const q15_t WeightsQ15_512[1024];
+    extern const q15_t cos_factorsQ15_512[512];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_DCT4_Q15_2048)
+    extern const q15_t WeightsQ15_2048[4096];
+    extern const q15_t cos_factorsQ15_2048[2048];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_DCT4_Q15_8192)
+    extern const q15_t WeightsQ15_8192[16384];
+    extern const q15_t cos_factorsQ15_8192[8192];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_DCT4_Q31_128)
+    extern const q31_t WeightsQ31_128[256];
+    extern const q31_t cos_factorsQ31_128[128];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_DCT4_Q31_512)
+    extern const q31_t WeightsQ31_512[1024];
+    extern const q31_t cos_factorsQ31_512[512];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_DCT4_Q31_2048)
+    extern const q31_t WeightsQ31_2048[4096];
+    extern const q31_t cos_factorsQ31_2048[2048];
+  #endif
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_DCT4_Q31_8192)
+    extern const q31_t WeightsQ31_8192[16384];
+    extern const q31_t cos_factorsQ31_8192[8192];
+  #endif
+
+#endif /* if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_FFT_TABLES) */
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_FAST_ALLOW_TABLES)
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FAST_TABLES) || defined(ARM_TABLE_RECIP_Q15)
+    extern const q15_t armRecipTableQ15[64];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) defined(ARM_ALL_FAST_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FAST_TABLES) || defined(ARM_TABLE_RECIP_Q31)
+    extern const q31_t armRecipTableQ31[64];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) defined(ARM_ALL_FAST_TABLES) */
+
+  /* Tables for Fast Math Sine and Cosine */
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FAST_TABLES) || defined(ARM_TABLE_SIN_F32)
+    extern const float32_t sinTable_f32[FAST_MATH_TABLE_SIZE + 1];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) defined(ARM_ALL_FAST_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FAST_TABLES) || defined(ARM_TABLE_SIN_Q31)
+    extern const q31_t sinTable_q31[FAST_MATH_TABLE_SIZE + 1];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) defined(ARM_ALL_FAST_TABLES) */
+
+  #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FAST_TABLES) || defined(ARM_TABLE_SIN_Q15)
+    extern const q15_t sinTable_q15[FAST_MATH_TABLE_SIZE + 1];
+  #endif /* !defined(ARM_DSP_CONFIG_TABLES) defined(ARM_ALL_FAST_TABLES) */
+
+  #if defined(ARM_MATH_MVEI)
+     #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FAST_TABLES) || defined(ARM_TABLE_FAST_SQRT_Q31_MVE)
+       extern const q31_t sqrtTable_Q31[256];
+     #endif /* !defined(ARM_DSP_CONFIG_TABLES) defined(ARM_ALL_FAST_TABLES) */
+  #endif
+
+  #if defined(ARM_MATH_MVEI)
+     #if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FAST_TABLES) || defined(ARM_TABLE_FAST_SQRT_Q15_MVE)
+       extern const q15_t sqrtTable_Q15[256];
+     #endif /* !defined(ARM_DSP_CONFIG_TABLES) defined(ARM_ALL_FAST_TABLES) */
+  #endif
+
+#endif /* if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_FAST_TABLES) */
+
+#if (defined(ARM_MATH_MVEF) || defined(ARM_MATH_HELIUM)) && !defined(ARM_MATH_AUTOVECTORIZE)
+       extern const float32_t exp_tab[8];
+       extern const float32_t __logf_lut_f32[8];
+#endif /* (defined(ARM_MATH_MVEF) || defined(ARM_MATH_HELIUM)) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+#if (defined(ARM_MATH_MVEI) || defined(ARM_MATH_HELIUM))
+extern const unsigned char hwLUT[256];
+#endif /* (defined(ARM_MATH_MVEI) || defined(ARM_MATH_HELIUM)) */
+
+#endif /*  ARM_COMMON_TABLES_H */
+

libraries/sdk/third_party/CMSIS/DSP/Include/arm_const_structs.h

@@ -0,0 +1,76 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_const_structs.h
+ * Description:  Constant structs that are initialized for user convenience.
+ *               For example, some can be given as arguments to the arm_cfft_f32() function.
+ *
+ * $Date:        27. January 2017
+ * $Revision:    V.1.5.1
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2017 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef _ARM_CONST_STRUCTS_H
+#define _ARM_CONST_STRUCTS_H
+
+#include "arm_math.h"
+#include "arm_common_tables.h"
+
+   extern const arm_cfft_instance_f64 arm_cfft_sR_f64_len16;
+   extern const arm_cfft_instance_f64 arm_cfft_sR_f64_len32;
+   extern const arm_cfft_instance_f64 arm_cfft_sR_f64_len64;
+   extern const arm_cfft_instance_f64 arm_cfft_sR_f64_len128;
+   extern const arm_cfft_instance_f64 arm_cfft_sR_f64_len256;
+   extern const arm_cfft_instance_f64 arm_cfft_sR_f64_len512;
+   extern const arm_cfft_instance_f64 arm_cfft_sR_f64_len1024;
+   extern const arm_cfft_instance_f64 arm_cfft_sR_f64_len2048;
+   extern const arm_cfft_instance_f64 arm_cfft_sR_f64_len4096;
+
+   extern const arm_cfft_instance_f32 arm_cfft_sR_f32_len16;
+   extern const arm_cfft_instance_f32 arm_cfft_sR_f32_len32;
+   extern const arm_cfft_instance_f32 arm_cfft_sR_f32_len64;
+   extern const arm_cfft_instance_f32 arm_cfft_sR_f32_len128;
+   extern const arm_cfft_instance_f32 arm_cfft_sR_f32_len256;
+   extern const arm_cfft_instance_f32 arm_cfft_sR_f32_len512;
+   extern const arm_cfft_instance_f32 arm_cfft_sR_f32_len1024;
+   extern const arm_cfft_instance_f32 arm_cfft_sR_f32_len2048;
+   extern const arm_cfft_instance_f32 arm_cfft_sR_f32_len4096;
+
+   extern const arm_cfft_instance_q31 arm_cfft_sR_q31_len16;
+   extern const arm_cfft_instance_q31 arm_cfft_sR_q31_len32;
+   extern const arm_cfft_instance_q31 arm_cfft_sR_q31_len64;
+   extern const arm_cfft_instance_q31 arm_cfft_sR_q31_len128;
+   extern const arm_cfft_instance_q31 arm_cfft_sR_q31_len256;
+   extern const arm_cfft_instance_q31 arm_cfft_sR_q31_len512;
+   extern const arm_cfft_instance_q31 arm_cfft_sR_q31_len1024;
+   extern const arm_cfft_instance_q31 arm_cfft_sR_q31_len2048;
+   extern const arm_cfft_instance_q31 arm_cfft_sR_q31_len4096;
+
+   extern const arm_cfft_instance_q15 arm_cfft_sR_q15_len16;
+   extern const arm_cfft_instance_q15 arm_cfft_sR_q15_len32;
+   extern const arm_cfft_instance_q15 arm_cfft_sR_q15_len64;
+   extern const arm_cfft_instance_q15 arm_cfft_sR_q15_len128;
+   extern const arm_cfft_instance_q15 arm_cfft_sR_q15_len256;
+   extern const arm_cfft_instance_q15 arm_cfft_sR_q15_len512;
+   extern const arm_cfft_instance_q15 arm_cfft_sR_q15_len1024;
+   extern const arm_cfft_instance_q15 arm_cfft_sR_q15_len2048;
+   extern const arm_cfft_instance_q15 arm_cfft_sR_q15_len4096;
+
+#endif

libraries/sdk/third_party/CMSIS/DSP/Include/arm_helium_utils.h

@@ -0,0 +1,348 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_helium_utils.h
+ * Description:  Utility functions for Helium development
+ *
+ * $Date:        09. September 2019
+ * $Revision:    V.1.5.1
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef _ARM_UTILS_HELIUM_H_
+#define _ARM_UTILS_HELIUM_H_
+
+/***************************************
+
+Definitions available for MVEF and MVEI
+
+***************************************/
+#if defined (ARM_MATH_HELIUM) || defined(ARM_MATH_MVEF) || defined(ARM_MATH_MVEI)
+
+#define INACTIVELANE            0 /* inactive lane content */
+
+
+#endif /* defined (ARM_MATH_HELIUM) || defined(ARM_MATH_MVEF) || defined(ARM_MATH_MVEI) */
+
+/***************************************
+
+Definitions available for MVEF only
+
+***************************************/
+#if defined (ARM_MATH_HELIUM) || defined(ARM_MATH_MVEF)
+
+__STATIC_FORCEINLINE float32_t vecAddAcrossF32Mve(float32x4_t in)
+{
+    float32_t acc;
+
+    acc = vgetq_lane(in, 0) + vgetq_lane(in, 1) +
+          vgetq_lane(in, 2) + vgetq_lane(in, 3);
+
+    return acc;
+}
+
+/* newton initial guess */
+#define INVSQRT_MAGIC_F32           0x5f3759df
+
+#define INVSQRT_NEWTON_MVE_F32(invSqrt, xHalf, xStart)\
+{                                                     \
+    float32x4_t tmp;                                  \
+                                                      \
+    /* tmp = xhalf * x * x */                         \
+    tmp = vmulq(xStart, xStart);                      \
+    tmp = vmulq(tmp, xHalf);                          \
+    /* (1.5f - xhalf * x * x) */                      \
+    tmp = vsubq(vdupq_n_f32(1.5f), tmp);              \
+    /* x = x*(1.5f-xhalf*x*x); */                     \
+    invSqrt = vmulq(tmp, xStart);                     \
+}
+#endif /* defined (ARM_MATH_HELIUM) || defined(ARM_MATH_MVEF) */
+
+/***************************************
+
+Definitions available for MVEI only
+
+***************************************/
+#if defined (ARM_MATH_HELIUM) || defined(ARM_MATH_MVEI)
+
+
+#include "arm_common_tables.h"
+
+/* Following functions are used to transpose matrix in f32 and q31 cases */
+__STATIC_INLINE arm_status arm_mat_trans_32bit_2x2_mve(
+    uint32_t * pDataSrc,
+    uint32_t * pDataDest)
+{
+    static const uint32x4_t vecOffs = { 0, 2, 1, 3 };
+    /*
+     *
+     * | 0   1 |   =>  |  0   2 |
+     * | 2   3 |       |  1   3 |
+     *
+     */
+    uint32x4_t vecIn = vldrwq_u32((uint32_t const *)pDataSrc);
+    vstrwq_scatter_shifted_offset_u32(pDataDest, vecOffs, vecIn);
+
+    return (ARM_MATH_SUCCESS);
+}
+
+__STATIC_INLINE arm_status arm_mat_trans_32bit_3x3_mve(
+    uint32_t * pDataSrc,
+    uint32_t * pDataDest)
+{
+    const uint32x4_t vecOffs1 = { 0, 3, 6, 1};
+    const uint32x4_t vecOffs2 = { 4, 7, 2, 5};
+    /*
+     *
+     *  | 0   1   2 |       | 0   3   6 |  4 x 32 flattened version | 0   3   6   1 |
+     *  | 3   4   5 |   =>  | 1   4   7 |            =>             | 4   7   2   5 |
+     *  | 6   7   8 |       | 2   5   8 |       (row major)         | 8   .   .   . |
+     *
+     */
+    uint32x4_t vecIn1 = vldrwq_u32((uint32_t const *) pDataSrc);
+    uint32x4_t vecIn2 = vldrwq_u32((uint32_t const *) &pDataSrc[4]);
+
+    vstrwq_scatter_shifted_offset_u32(pDataDest, vecOffs1, vecIn1);
+    vstrwq_scatter_shifted_offset_u32(pDataDest, vecOffs2, vecIn2);
+
+    pDataDest[8] = pDataSrc[8];
+
+    return (ARM_MATH_SUCCESS);
+}
+
+__STATIC_INLINE arm_status arm_mat_trans_32bit_4x4_mve(uint32_t * pDataSrc, uint32_t * pDataDest)
+{
+    /*
+     * 4x4 Matrix transposition
+     * is 4 x de-interleave operation
+     *
+     * 0   1   2   3       0   4   8   12
+     * 4   5   6   7       1   5   9   13
+     * 8   9   10  11      2   6   10  14
+     * 12  13  14  15      3   7   11  15
+     */
+
+    uint32x4x4_t vecIn;
+
+    vecIn = vld4q((uint32_t const *) pDataSrc);
+    vstrwq(pDataDest, vecIn.val[0]);
+    pDataDest += 4;
+    vstrwq(pDataDest, vecIn.val[1]);
+    pDataDest += 4;
+    vstrwq(pDataDest, vecIn.val[2]);
+    pDataDest += 4;
+    vstrwq(pDataDest, vecIn.val[3]);
+
+    return (ARM_MATH_SUCCESS);
+}
+
+
+__STATIC_INLINE arm_status arm_mat_trans_32bit_generic_mve(
+    uint16_t    srcRows,
+    uint16_t    srcCols,
+    uint32_t  * pDataSrc,
+    uint32_t  * pDataDest)
+{
+    uint32x4_t vecOffs;
+    uint32_t  i;
+    uint32_t  blkCnt;
+    uint32_t const *pDataC;
+    uint32_t *pDataDestR;
+    uint32x4_t vecIn;
+
+    vecOffs = vidupq_u32((uint32_t)0, 1);
+    vecOffs = vecOffs * srcCols;
+
+    i = srcCols;
+    do
+    {
+        pDataC = (uint32_t const *) pDataSrc;
+        pDataDestR = pDataDest;
+
+        blkCnt = srcRows >> 2;
+        while (blkCnt > 0U)
+        {
+            vecIn = vldrwq_gather_shifted_offset_u32(pDataC, vecOffs);
+            vstrwq(pDataDestR, vecIn); 
+            pDataDestR += 4;
+            pDataC = pDataC + srcCols * 4;
+            /*
+             * Decrement the blockSize loop counter
+             */
+            blkCnt--;
+        }
+
+        /*
+         * tail
+         */
+        blkCnt = srcRows & 3;
+        if (blkCnt > 0U)
+        {
+            mve_pred16_t p0 = vctp32q(blkCnt);
+            vecIn = vldrwq_gather_shifted_offset_u32(pDataC, vecOffs);
+            vstrwq_p(pDataDestR, vecIn, p0);
+        }
+
+        pDataSrc += 1;
+        pDataDest += srcRows;
+    }
+    while (--i);
+
+    return (ARM_MATH_SUCCESS);
+}
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FAST_TABLES) || defined(ARM_TABLE_FAST_SQRT_Q31_MVE)
+__STATIC_INLINE q31x4_t FAST_VSQRT_Q31(q31x4_t vecIn)
+{
+    q63x2_t         vecTmpLL;
+    q31x4_t         vecTmp0, vecTmp1;
+    q31_t           scale;
+    q63_t           tmp64;
+    q31x4_t         vecNrm, vecDst, vecIdx, vecSignBits;
+
+
+    vecSignBits = vclsq(vecIn);
+    vecSignBits = vbicq(vecSignBits, 1);
+    /*
+     * in = in << no_of_sign_bits;
+     */
+    vecNrm = vshlq(vecIn, vecSignBits);
+    /*
+     * index = in >> 24;
+     */
+    vecIdx = vecNrm >> 24;
+    vecIdx = vecIdx << 1;
+
+    vecTmp0 = vldrwq_gather_shifted_offset_s32(sqrtTable_Q31, vecIdx);
+
+    vecIdx = vecIdx + 1;
+
+    vecTmp1 = vldrwq_gather_shifted_offset_s32(sqrtTable_Q31, vecIdx);
+
+    vecTmp1 = vqrdmulhq(vecTmp1, vecNrm);
+    vecTmp0 = vecTmp0 - vecTmp1;
+    vecTmp1 = vqrdmulhq(vecTmp0, vecTmp0);
+    vecTmp1 = vqrdmulhq(vecNrm, vecTmp1);
+    vecTmp1 = vdupq_n_s32(0x18000000) - vecTmp1;
+    vecTmp0 = vqrdmulhq(vecTmp0, vecTmp1);
+    vecTmpLL = vmullbq_int(vecNrm, vecTmp0);
+
+    /*
+     * scale elements 0, 2
+     */
+    scale = 26 + (vecSignBits[0] >> 1);
+    tmp64 = asrl(vecTmpLL[0], scale);
+    vecDst[0] = (q31_t) tmp64;
+
+    scale = 26 + (vecSignBits[2] >> 1);
+    tmp64 = asrl(vecTmpLL[1], scale);
+    vecDst[2] = (q31_t) tmp64;
+
+    vecTmpLL = vmulltq_int(vecNrm, vecTmp0);
+
+    /*
+     * scale elements 1, 3
+     */
+    scale = 26 + (vecSignBits[1] >> 1);
+    tmp64 = asrl(vecTmpLL[0], scale);
+    vecDst[1] = (q31_t) tmp64;
+
+    scale = 26 + (vecSignBits[3] >> 1);
+    tmp64 = asrl(vecTmpLL[1], scale);
+    vecDst[3] = (q31_t) tmp64;
+    /*
+     * set negative values to 0
+     */
+    vecDst = vdupq_m(vecDst, 0, vcmpltq_n_s32(vecIn, 0));
+
+    return vecDst;
+}
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FAST_TABLES) || defined(ARM_TABLE_FAST_SQRT_Q15_MVE)
+__STATIC_INLINE q15x8_t FAST_VSQRT_Q15(q15x8_t vecIn)
+{
+    q31x4_t         vecTmpLev, vecTmpLodd, vecSignL;
+    q15x8_t         vecTmp0, vecTmp1;
+    q15x8_t         vecNrm, vecDst, vecIdx, vecSignBits;
+
+    vecDst = vuninitializedq_s16();
+
+    vecSignBits = vclsq(vecIn);
+    vecSignBits = vbicq(vecSignBits, 1);
+    /*
+     * in = in << no_of_sign_bits;
+     */
+    vecNrm = vshlq(vecIn, vecSignBits);
+
+    vecIdx = vecNrm >> 8;
+    vecIdx = vecIdx << 1;
+
+    vecTmp0 = vldrhq_gather_shifted_offset_s16(sqrtTable_Q15, vecIdx);
+
+    vecIdx = vecIdx + 1;
+
+    vecTmp1 = vldrhq_gather_shifted_offset_s16(sqrtTable_Q15, vecIdx);
+
+    vecTmp1 = vqrdmulhq(vecTmp1, vecNrm);
+    vecTmp0 = vecTmp0 - vecTmp1;
+    vecTmp1 = vqrdmulhq(vecTmp0, vecTmp0);
+    vecTmp1 = vqrdmulhq(vecNrm, vecTmp1);
+    vecTmp1 = vdupq_n_s16(0x1800) - vecTmp1;
+    vecTmp0 = vqrdmulhq(vecTmp0, vecTmp1);
+
+    vecSignBits = vecSignBits >> 1;
+
+    vecTmpLev = vmullbq_int(vecNrm, vecTmp0);
+    vecTmpLodd = vmulltq_int(vecNrm, vecTmp0);
+
+    vecTmp0 = vecSignBits + 10;
+    /*
+     * negate sign to apply register based vshl
+     */
+    vecTmp0 = -vecTmp0;
+
+    /*
+     * shift even elements
+     */
+    vecSignL = vmovlbq(vecTmp0);
+    vecTmpLev = vshlq(vecTmpLev, vecSignL);
+    /*
+     * shift odd elements
+     */
+    vecSignL = vmovltq(vecTmp0);
+    vecTmpLodd = vshlq(vecTmpLodd, vecSignL);
+    /*
+     * merge and narrow odd and even parts
+     */
+    vecDst = vmovnbq_s32(vecDst, vecTmpLev);
+    vecDst = vmovntq_s32(vecDst, vecTmpLodd);
+    /*
+     * set negative values to 0
+     */
+    vecDst = vdupq_m(vecDst, 0, vcmpltq_n_s16(vecIn, 0));
+
+    return vecDst;
+}
+#endif
+
+#endif /* defined (ARM_MATH_HELIUM) || defined(ARM_MATH_MVEI) */
+
+#endif

libraries/sdk/third_party/CMSIS/DSP/Include/arm_mve_tables.h

@@ -0,0 +1,235 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_mve_tables.h
+ * Description:  common tables like fft twiddle factors, Bitreverse, reciprocal etc
+ *               used for MVE implementation only
+ *
+ * $Date:        08. January 2020
+ * $Revision:    V1.7.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2020 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+ #ifndef _ARM_MVE_TABLES_H
+ #define _ARM_MVE_TABLES_H
+
+ #include "arm_math.h"
+
+ 
+
+
+ 
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_FFT_ALLOW_TABLES)
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_16) || defined(ARM_TABLE_TWIDDLECOEF_F32_32)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_16_f32[2];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_16_f32[2];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_16_f32[2];
+extern float32_t rearranged_twiddle_stride1_16_f32[8];
+extern float32_t rearranged_twiddle_stride2_16_f32[8];
+extern float32_t rearranged_twiddle_stride3_16_f32[8];
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_64) || defined(ARM_TABLE_TWIDDLECOEF_F32_128)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_64_f32[3];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_64_f32[3];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_64_f32[3];
+extern float32_t rearranged_twiddle_stride1_64_f32[40];
+extern float32_t rearranged_twiddle_stride2_64_f32[40];
+extern float32_t rearranged_twiddle_stride3_64_f32[40];
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_256) || defined(ARM_TABLE_TWIDDLECOEF_F32_512)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_256_f32[4];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_256_f32[4];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_256_f32[4];
+extern float32_t rearranged_twiddle_stride1_256_f32[168];
+extern float32_t rearranged_twiddle_stride2_256_f32[168];
+extern float32_t rearranged_twiddle_stride3_256_f32[168];
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_1024) || defined(ARM_TABLE_TWIDDLECOEF_F32_2048)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_1024_f32[5];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_1024_f32[5];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_1024_f32[5];
+extern float32_t rearranged_twiddle_stride1_1024_f32[680];
+extern float32_t rearranged_twiddle_stride2_1024_f32[680];
+extern float32_t rearranged_twiddle_stride3_1024_f32[680];
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_F32_4096) || defined(ARM_TABLE_TWIDDLECOEF_F32_8192)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_4096_f32[6];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_4096_f32[6];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_4096_f32[6];
+extern float32_t rearranged_twiddle_stride1_4096_f32[2728];
+extern float32_t rearranged_twiddle_stride2_4096_f32[2728];
+extern float32_t rearranged_twiddle_stride3_4096_f32[2728];
+#endif
+
+
+#endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_FFT_ALLOW_TABLES) */
+
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+
+
+#if defined(ARM_MATH_MVEI) 
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_FFT_ALLOW_TABLES)
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_16) || defined(ARM_TABLE_TWIDDLECOEF_Q31_32)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_16_q31[2];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_16_q31[2];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_16_q31[2];
+extern q31_t rearranged_twiddle_stride1_16_q31[8];
+extern q31_t rearranged_twiddle_stride2_16_q31[8];
+extern q31_t rearranged_twiddle_stride3_16_q31[8];
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_64) || defined(ARM_TABLE_TWIDDLECOEF_Q31_128)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_64_q31[3];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_64_q31[3];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_64_q31[3];
+extern q31_t rearranged_twiddle_stride1_64_q31[40];
+extern q31_t rearranged_twiddle_stride2_64_q31[40];
+extern q31_t rearranged_twiddle_stride3_64_q31[40];
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_256) || defined(ARM_TABLE_TWIDDLECOEF_Q31_512)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_256_q31[4];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_256_q31[4];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_256_q31[4];
+extern q31_t rearranged_twiddle_stride1_256_q31[168];
+extern q31_t rearranged_twiddle_stride2_256_q31[168];
+extern q31_t rearranged_twiddle_stride3_256_q31[168];
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_1024) || defined(ARM_TABLE_TWIDDLECOEF_Q31_2048)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_1024_q31[5];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_1024_q31[5];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_1024_q31[5];
+extern q31_t rearranged_twiddle_stride1_1024_q31[680];
+extern q31_t rearranged_twiddle_stride2_1024_q31[680];
+extern q31_t rearranged_twiddle_stride3_1024_q31[680];
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q31_4096) || defined(ARM_TABLE_TWIDDLECOEF_Q31_8192)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_4096_q31[6];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_4096_q31[6];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_4096_q31[6];
+extern q31_t rearranged_twiddle_stride1_4096_q31[2728];
+extern q31_t rearranged_twiddle_stride2_4096_q31[2728];
+extern q31_t rearranged_twiddle_stride3_4096_q31[2728];
+#endif
+
+
+#endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_FFT_ALLOW_TABLES) */
+
+#endif /* defined(ARM_MATH_MVEI) */
+
+
+
+#if defined(ARM_MATH_MVEI) 
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_FFT_ALLOW_TABLES)
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_16) || defined(ARM_TABLE_TWIDDLECOEF_Q15_32)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_16_q15[2];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_16_q15[2];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_16_q15[2];
+extern q15_t rearranged_twiddle_stride1_16_q15[8];
+extern q15_t rearranged_twiddle_stride2_16_q15[8];
+extern q15_t rearranged_twiddle_stride3_16_q15[8];
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_64) || defined(ARM_TABLE_TWIDDLECOEF_Q15_128)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_64_q15[3];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_64_q15[3];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_64_q15[3];
+extern q15_t rearranged_twiddle_stride1_64_q15[40];
+extern q15_t rearranged_twiddle_stride2_64_q15[40];
+extern q15_t rearranged_twiddle_stride3_64_q15[40];
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_256) || defined(ARM_TABLE_TWIDDLECOEF_Q15_512)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_256_q15[4];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_256_q15[4];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_256_q15[4];
+extern q15_t rearranged_twiddle_stride1_256_q15[168];
+extern q15_t rearranged_twiddle_stride2_256_q15[168];
+extern q15_t rearranged_twiddle_stride3_256_q15[168];
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_1024) || defined(ARM_TABLE_TWIDDLECOEF_Q15_2048)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_1024_q15[5];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_1024_q15[5];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_1024_q15[5];
+extern q15_t rearranged_twiddle_stride1_1024_q15[680];
+extern q15_t rearranged_twiddle_stride2_1024_q15[680];
+extern q15_t rearranged_twiddle_stride3_1024_q15[680];
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || defined(ARM_TABLE_TWIDDLECOEF_Q15_4096) || defined(ARM_TABLE_TWIDDLECOEF_Q15_8192)
+
+extern uint32_t rearranged_twiddle_tab_stride1_arr_4096_q15[6];
+extern uint32_t rearranged_twiddle_tab_stride2_arr_4096_q15[6];
+extern uint32_t rearranged_twiddle_tab_stride3_arr_4096_q15[6];
+extern q15_t rearranged_twiddle_stride1_4096_q15[2728];
+extern q15_t rearranged_twiddle_stride2_4096_q15[2728];
+extern q15_t rearranged_twiddle_stride3_4096_q15[2728];
+#endif
+
+
+#endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_FFT_ALLOW_TABLES) */
+
+#endif /* defined(ARM_MATH_MVEI) */
+
+
+
+#if defined(ARM_MATH_MVEI) 
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_FFT_ALLOW_TABLES)
+
+
+#endif /* !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_FFT_ALLOW_TABLES) */
+
+#endif /* defined(ARM_MATH_MVEI) */
+
+
+
+#endif /*_ARM_MVE_TABLES_H*/
+

libraries/sdk/third_party/CMSIS/DSP/Include/arm_vec_math.h

@@ -0,0 +1,372 @@
+/******************************************************************************
+ * @file     arm_vec_math.h
+ * @brief    Public header file for CMSIS DSP Library
+ * @version  V1.7.0
+ * @date     15. October 2019
+ ******************************************************************************/
+/*
+ * Copyright (c) 2010-2019 Arm Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef _ARM_VEC_MATH_H
+#define _ARM_VEC_MATH_H
+
+#include "arm_math.h"
+#include "arm_common_tables.h"
+#include "arm_helium_utils.h"
+
+#ifdef   __cplusplus
+extern "C"
+{
+#endif
+
+#if (defined(ARM_MATH_MVEF) || defined(ARM_MATH_HELIUM)) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#define INV_NEWTON_INIT_F32         0x7EF127EA
+
+static const float32_t __logf_rng_f32=0.693147180f;
+
+
+/* fast inverse approximation (3x newton) */
+__STATIC_INLINE f32x4_t vrecip_medprec_f32(
+    f32x4_t x)
+{
+    q31x4_t         m;
+    f32x4_t         b;
+    any32x4_t       xinv;
+    f32x4_t         ax = vabsq(x);
+
+    xinv.f = ax;
+    m = 0x3F800000 - (xinv.i & 0x7F800000);
+    xinv.i = xinv.i + m;
+    xinv.f = 1.41176471f - 0.47058824f * xinv.f;
+    xinv.i = xinv.i + m;
+
+    b = 2.0f - xinv.f * ax;
+    xinv.f = xinv.f * b;
+
+    b = 2.0f - xinv.f * ax;
+    xinv.f = xinv.f * b;
+
+    b = 2.0f - xinv.f * ax;
+    xinv.f = xinv.f * b;
+
+    xinv.f = vdupq_m(xinv.f, INFINITY, vcmpeqq(x, 0.0f));
+    /*
+     * restore sign
+     */
+    xinv.f = vnegq_m(xinv.f, xinv.f, vcmpltq(x, 0.0f));
+
+    return xinv.f;
+}
+
+/* fast inverse approximation (4x newton) */
+__STATIC_INLINE f32x4_t vrecip_hiprec_f32(
+    f32x4_t x)
+{
+    q31x4_t         m;
+    f32x4_t         b;
+    any32x4_t       xinv;
+    f32x4_t         ax = vabsq(x);
+
+    xinv.f = ax;
+
+    m = 0x3F800000 - (xinv.i & 0x7F800000);
+    xinv.i = xinv.i + m;
+    xinv.f = 1.41176471f - 0.47058824f * xinv.f;
+    xinv.i = xinv.i + m;
+
+    b = 2.0f - xinv.f * ax;
+    xinv.f = xinv.f * b;
+
+    b = 2.0f - xinv.f * ax;
+    xinv.f = xinv.f * b;
+
+    b = 2.0f - xinv.f * ax;
+    xinv.f = xinv.f * b;
+
+    b = 2.0f - xinv.f * ax;
+    xinv.f = xinv.f * b;
+
+    xinv.f = vdupq_m(xinv.f, INFINITY, vcmpeqq(x, 0.0f));
+    /*
+     * restore sign
+     */
+    xinv.f = vnegq_m(xinv.f, xinv.f, vcmpltq(x, 0.0f));
+
+    return xinv.f;
+}
+
+__STATIC_INLINE f32x4_t vdiv_f32(
+    f32x4_t num, f32x4_t den)
+{
+    return vmulq(num, vrecip_hiprec_f32(den));
+}
+
+/**
+  @brief         Single-precision taylor dev.
+  @param[in]     x              f32 quad vector input
+  @param[in]     coeffs         f32 quad vector coeffs
+  @return        destination    f32 quad vector
+ */
+
+__STATIC_INLINE f32x4_t vtaylor_polyq_f32(
+        f32x4_t           x,
+        const float32_t * coeffs)
+{
+    f32x4_t         A = vfmasq(vdupq_n_f32(coeffs[4]), x, coeffs[0]);
+    f32x4_t         B = vfmasq(vdupq_n_f32(coeffs[6]), x, coeffs[2]);
+    f32x4_t         C = vfmasq(vdupq_n_f32(coeffs[5]), x, coeffs[1]);
+    f32x4_t         D = vfmasq(vdupq_n_f32(coeffs[7]), x, coeffs[3]);
+    f32x4_t         x2 = vmulq(x, x);
+    f32x4_t         x4 = vmulq(x2, x2);
+    f32x4_t         res = vfmaq(vfmaq_f32(A, B, x2), vfmaq_f32(C, D, x2), x4);
+
+    return res;
+}
+
+__STATIC_INLINE f32x4_t vmant_exp_f32(
+    f32x4_t     x,
+    int32x4_t * e)
+{
+    any32x4_t       r;
+    int32x4_t       n;
+
+    r.f = x;
+    n = r.i >> 23;
+    n = n - 127;
+    r.i = r.i - (n << 23);
+
+    *e = n;
+    return r.f;
+}
+
+
+__STATIC_INLINE f32x4_t vlogq_f32(f32x4_t vecIn)
+{
+    q31x4_t         vecExpUnBiased;
+    f32x4_t         vecTmpFlt0, vecTmpFlt1;
+    f32x4_t         vecAcc0, vecAcc1, vecAcc2, vecAcc3;
+    f32x4_t         vecExpUnBiasedFlt;
+
+    /*
+     * extract exponent
+     */
+    vecTmpFlt1 = vmant_exp_f32(vecIn, &vecExpUnBiased);
+
+    vecTmpFlt0 = vecTmpFlt1 * vecTmpFlt1;
+    /*
+     * a = (__logf_lut_f32[4] * r.f) + (__logf_lut_f32[0]);
+     */
+    vecAcc0 = vdupq_n_f32(__logf_lut_f32[0]);
+    vecAcc0 = vfmaq(vecAcc0, vecTmpFlt1, __logf_lut_f32[4]);
+    /*
+     * b = (__logf_lut_f32[6] * r.f) + (__logf_lut_f32[2]);
+     */
+    vecAcc1 = vdupq_n_f32(__logf_lut_f32[2]);
+    vecAcc1 = vfmaq(vecAcc1, vecTmpFlt1, __logf_lut_f32[6]);
+    /*
+     * c = (__logf_lut_f32[5] * r.f) + (__logf_lut_f32[1]);
+     */
+    vecAcc2 = vdupq_n_f32(__logf_lut_f32[1]);
+    vecAcc2 = vfmaq(vecAcc2, vecTmpFlt1, __logf_lut_f32[5]);
+    /*
+     * d = (__logf_lut_f32[7] * r.f) + (__logf_lut_f32[3]);
+     */
+    vecAcc3 = vdupq_n_f32(__logf_lut_f32[3]);
+    vecAcc3 = vfmaq(vecAcc3, vecTmpFlt1, __logf_lut_f32[7]);
+    /*
+     * a = a + b * xx;
+     */
+    vecAcc0 = vfmaq(vecAcc0, vecAcc1, vecTmpFlt0);
+    /*
+     * c = c + d * xx;
+     */
+    vecAcc2 = vfmaq(vecAcc2, vecAcc3, vecTmpFlt0);
+    /*
+     * xx = xx * xx;
+     */
+    vecTmpFlt0 = vecTmpFlt0 * vecTmpFlt0;
+    vecExpUnBiasedFlt = vcvtq_f32_s32(vecExpUnBiased);
+    /*
+     * r.f = a + c * xx;
+     */
+    vecAcc0 = vfmaq(vecAcc0, vecAcc2, vecTmpFlt0);
+    /*
+     * add exponent
+     * r.f = r.f + ((float32_t) m) * __logf_rng_f32;
+     */
+    vecAcc0 = vfmaq(vecAcc0, vecExpUnBiasedFlt, __logf_rng_f32);
+    // set log0 down to -inf
+    vecAcc0 = vdupq_m(vecAcc0, -INFINITY, vcmpeqq(vecIn, 0.0f));
+    return vecAcc0;
+}
+
+__STATIC_INLINE f32x4_t vexpq_f32(
+    f32x4_t x)
+{
+    // Perform range reduction [-log(2),log(2)]
+    int32x4_t       m = vcvtq_s32_f32(vmulq_n_f32(x, 1.4426950408f));
+    f32x4_t         val = vfmsq_f32(x, vcvtq_f32_s32(m), vdupq_n_f32(0.6931471805f));
+
+    // Polynomial Approximation
+    f32x4_t         poly = vtaylor_polyq_f32(val, exp_tab);
+
+    // Reconstruct
+    poly = (f32x4_t) (vqaddq_s32((q31x4_t) (poly), vqshlq_n_s32(m, 23)));
+
+    poly = vdupq_m(poly, 0.0f, vcmpltq_n_s32(m, -126));
+    return poly;
+}
+
+__STATIC_INLINE f32x4_t arm_vec_exponent_f32(f32x4_t x, int32_t nb)
+{
+    f32x4_t         r = x;
+    nb--;
+    while (nb > 0) {
+        r = vmulq(r, x);
+        nb--;
+    }
+    return (r);
+}
+
+__STATIC_INLINE f32x4_t vrecip_f32(f32x4_t vecIn)
+{
+    f32x4_t     vecSx, vecW, vecTmp;
+    any32x4_t   v;
+
+    vecSx = vabsq(vecIn);
+
+    v.f = vecIn;
+    v.i = vsubq(vdupq_n_s32(INV_NEWTON_INIT_F32), v.i);
+
+    vecW = vmulq(vecSx, v.f);
+
+    // v.f = v.f * (8 + w * (-28 + w * (56 + w * (-70 + w *(56 + w * (-28 + w * (8 - w)))))));
+    vecTmp = vsubq(vdupq_n_f32(8.0f), vecW);
+    vecTmp = vfmasq(vecW, vecTmp, -28.0f);
+    vecTmp = vfmasq(vecW, vecTmp, 56.0f);
+    vecTmp = vfmasq(vecW, vecTmp, -70.0f);
+    vecTmp = vfmasq(vecW, vecTmp, 56.0f);
+    vecTmp = vfmasq(vecW, vecTmp, -28.0f);
+    vecTmp = vfmasq(vecW, vecTmp, 8.0f);
+    v.f = vmulq(v.f,  vecTmp);
+
+    v.f = vdupq_m(v.f, INFINITY, vcmpeqq(vecIn, 0.0f));
+    /*
+     * restore sign
+     */
+    v.f = vnegq_m(v.f, v.f, vcmpltq(vecIn, 0.0f));
+    return v.f;
+}
+
+__STATIC_INLINE f32x4_t vtanhq_f32(
+    f32x4_t val)
+{
+    f32x4_t         x =
+        vminnmq_f32(vmaxnmq_f32(val, vdupq_n_f32(-10.f)), vdupq_n_f32(10.0f));
+    f32x4_t         exp2x = vexpq_f32(vmulq_n_f32(x, 2.f));
+    f32x4_t         num = vsubq_n_f32(exp2x, 1.f);
+    f32x4_t         den = vaddq_n_f32(exp2x, 1.f);
+    f32x4_t         tanh = vmulq_f32(num, vrecip_f32(den));
+    return tanh;
+}
+
+__STATIC_INLINE f32x4_t vpowq_f32(
+    f32x4_t val,
+    f32x4_t n)
+{
+    return vexpq_f32(vmulq_f32(n, vlogq_f32(val)));
+}
+
+#endif /* (defined(ARM_MATH_MVEF) || defined(ARM_MATH_HELIUM)) && !defined(ARM_MATH_AUTOVECTORIZE)*/
+
+#if (defined(ARM_MATH_MVEI) || defined(ARM_MATH_HELIUM))
+#endif /* (defined(ARM_MATH_MVEI) || defined(ARM_MATH_HELIUM)) */
+
+#if (defined(ARM_MATH_NEON) || defined(ARM_MATH_NEON_EXPERIMENTAL)) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#include "NEMath.h"
+/**
+ * @brief Vectorized integer exponentiation
+ * @param[in]    x           value
+ * @param[in]    nb          integer exponent >= 1
+ * @return x^nb
+ *
+ */
+__STATIC_INLINE  float32x4_t arm_vec_exponent_f32(float32x4_t x, int32_t nb)
+{
+    float32x4_t r = x;
+    nb --;
+    while(nb > 0)
+    {
+        r = vmulq_f32(r , x);
+        nb--;
+    }
+    return(r);
+}
+
+
+__STATIC_INLINE float32x4_t __arm_vec_sqrt_f32_neon(float32x4_t  x)
+{
+    float32x4_t x1 = vmaxq_f32(x, vdupq_n_f32(FLT_MIN));
+    float32x4_t e = vrsqrteq_f32(x1);
+    e = vmulq_f32(vrsqrtsq_f32(vmulq_f32(x1, e), e), e);
+    e = vmulq_f32(vrsqrtsq_f32(vmulq_f32(x1, e), e), e);
+    return vmulq_f32(x, e);
+}
+
+__STATIC_INLINE int16x8_t __arm_vec_sqrt_q15_neon(int16x8_t vec)
+{
+    float32x4_t tempF;
+    int32x4_t tempHI,tempLO;
+
+    tempLO = vmovl_s16(vget_low_s16(vec));
+    tempF = vcvtq_n_f32_s32(tempLO,15);
+    tempF = __arm_vec_sqrt_f32_neon(tempF);
+    tempLO = vcvtq_n_s32_f32(tempF,15);
+
+    tempHI = vmovl_s16(vget_high_s16(vec));
+    tempF = vcvtq_n_f32_s32(tempHI,15);
+    tempF = __arm_vec_sqrt_f32_neon(tempF);
+    tempHI = vcvtq_n_s32_f32(tempF,15);
+
+    return(vcombine_s16(vqmovn_s32(tempLO),vqmovn_s32(tempHI)));
+}
+
+__STATIC_INLINE int32x4_t __arm_vec_sqrt_q31_neon(int32x4_t vec)
+{
+  float32x4_t temp;
+
+  temp = vcvtq_n_f32_s32(vec,31);
+  temp = __arm_vec_sqrt_f32_neon(temp);
+  return(vcvtq_n_s32_f32(temp,31));
+}
+
+#endif /*  (defined(ARM_MATH_NEON) || defined(ARM_MATH_NEON_EXPERIMENTAL)) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+#ifdef   __cplusplus
+}
+#endif
+
+
+#endif /* _ARM_VEC_MATH_H */
+
+/**
+ *
+ * End of file.
+ */

libraries/sdk/third_party/CMSIS/DSP/PrivateInclude/arm_sorting.h

@@ -0,0 +1,200 @@
+/******************************************************************************
+ * @file     arm_sorting.h
+ * @brief    Private header file for CMSIS DSP Library
+ * @version  V1.7.0
+ * @date     2019
+ ******************************************************************************/
+/*
+ * Copyright (c) 2010-2019 Arm Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef _ARM_SORTING_H_
+#define _ARM_SORTING_H_
+
+#include "arm_math.h"
+
+#ifdef   __cplusplus
+extern "C"
+{
+#endif
+
+  /**
+   * @param[in]  S          points to an instance of the sorting structure.
+   * @param[in]  pSrc       points to the block of input data.
+   * @param[out] pDst       points to the block of output data.
+   * @param[in]  blockSize  number of samples to process.
+   */
+  void arm_bubble_sort_f32(
+    const arm_sort_instance_f32 * S, 
+          float32_t * pSrc, 
+          float32_t * pDst, 
+    uint32_t blockSize);
+
+   /**
+   * @param[in]  S          points to an instance of the sorting structure.
+   * @param[in]  pSrc       points to the block of input data.
+   * @param[out] pDst       points to the block of output data.
+   * @param[in]  blockSize  number of samples to process.
+   */
+  void arm_heap_sort_f32(
+    const arm_sort_instance_f32 * S, 
+          float32_t * pSrc, 
+          float32_t * pDst, 
+    uint32_t blockSize);
+
+  /**
+   * @param[in]  S          points to an instance of the sorting structure.
+   * @param[in]  pSrc       points to the block of input data.
+   * @param[out] pDst       points to the block of output data.
+   * @param[in]  blockSize  number of samples to process.
+   */
+  void arm_insertion_sort_f32(
+    const arm_sort_instance_f32 * S, 
+          float32_t *pSrc, 
+          float32_t* pDst, 
+    uint32_t blockSize);
+
+  /**
+   * @param[in]  S          points to an instance of the sorting structure.
+   * @param[in]  pSrc       points to the block of input data.
+   * @param[out] pDst       points to the block of output data
+   * @param[in]  blockSize  number of samples to process.
+   */
+  void arm_quick_sort_f32(
+    const arm_sort_instance_f32 * S, 
+          float32_t * pSrc, 
+          float32_t * pDst, 
+    uint32_t blockSize);
+
+  /**
+   * @param[in]  S          points to an instance of the sorting structure.
+   * @param[in]  pSrc       points to the block of input data.
+   * @param[out] pDst       points to the block of output data
+   * @param[in]  blockSize  number of samples to process.
+   */
+  void arm_selection_sort_f32(
+    const arm_sort_instance_f32 * S, 
+          float32_t * pSrc, 
+          float32_t * pDst, 
+    uint32_t blockSize);
+ 
+  /**
+   * @param[in]  S          points to an instance of the sorting structure.
+   * @param[in]  pSrc       points to the block of input data.
+   * @param[out] pDst       points to the block of output data
+   * @param[in]  blockSize  number of samples to process.
+   */
+  void arm_bitonic_sort_f32(
+    const arm_sort_instance_f32 * S,
+          float32_t * pSrc,
+          float32_t * pDst,
+          uint32_t blockSize);
+
+#if defined(ARM_MATH_NEON)
+
+#define vtrn256_128q(a, b)                   \
+do {                                         \
+	float32x4_t vtrn128_temp = a.val[1]; \
+	a.val[1] = b.val[0];                 \
+	b.val[0] = vtrn128_temp ;            \
+} while (0)
+
+#define vtrn128_64q(a, b)           \
+do {                                \
+	float32x2_t ab, cd, ef, gh; \
+	ab = vget_low_f32(a);	    \
+	ef = vget_low_f32(b);	    \
+	cd = vget_high_f32(a);	    \
+	gh = vget_high_f32(b);      \
+	a = vcombine_f32(ab, ef);   \
+	b = vcombine_f32(cd, gh);   \
+} while (0)
+
+#define vtrn256_64q(a, b)                  \
+do {                                       \
+	float32x2_t a_0, a_1, a_2, a_3;    \
+	float32x2_t b_0, b_1, b_2, b_3;    \
+	a_0 = vget_low_f32(a.val[0]);      \
+	a_1 = vget_high_f32(a.val[0]);     \
+	a_2 = vget_low_f32(a.val[1]);      \
+	a_3 = vget_high_f32(a.val[1]);     \
+	b_0 = vget_low_f32(b.val[0]);      \
+	b_1 = vget_high_f32(b.val[0]);     \
+	b_2 = vget_low_f32(b.val[1]);      \
+	b_3 = vget_high_f32(b.val[1]);     \
+	a.val[0] = vcombine_f32(a_0, b_0); \
+	a.val[1] = vcombine_f32(a_2, b_2); \
+	b.val[0] = vcombine_f32(a_1, b_1); \
+	b.val[1] = vcombine_f32(a_3, b_3); \
+} while (0) 
+
+#define vtrn128_32q(a, b)                               \
+do {                                                    \
+	float32x4x2_t vtrn32_tmp = vtrnq_f32((a), (b)); \
+	(a) = vtrn32_tmp.val[0];                        \
+	(b) = vtrn32_tmp.val[1];                        \
+} while (0)
+
+#define vtrn256_32q(a, b)               \
+do {                                    \
+	float32x4x2_t vtrn32_tmp_1 = vtrnq_f32((a.val[0]), (b.val[0])); \
+	float32x4x2_t vtrn32_tmp_2 = vtrnq_f32((a.val[1]), (b.val[1])); \
+	a.val[0] = vtrn32_tmp_1.val[0]; \
+	a.val[1] = vtrn32_tmp_2.val[0]; \
+	b.val[0] = vtrn32_tmp_1.val[1]; \
+	b.val[1] = vtrn32_tmp_2.val[1]; \
+} while (0) 
+
+#define vminmaxq(a, b)                    \
+	do {                              \
+	float32x4_t minmax_tmp = (a);     \
+	(a) = vminq_f32((a), (b));        \
+	(b) = vmaxq_f32(minmax_tmp, (b)); \
+} while (0)
+
+#define vminmax256q(a, b)                         \
+	do {                                      \
+	float32x4x2_t minmax256_tmp = (a);        \
+	a.val[0] = vminq_f32(a.val[0], b.val[0]); \
+	a.val[1] = vminq_f32(a.val[1], b.val[1]); \
+	b.val[0] = vmaxq_f32(minmax256_tmp.val[0], b.val[0]); \
+	b.val[1] = vmaxq_f32(minmax256_tmp.val[1], b.val[1]); \
+} while (0)
+
+#define vrev128q_f32(a) \
+        vcombine_f32(vrev64_f32(vget_high_f32(a)), vrev64_f32(vget_low_f32(a)))
+
+#define vrev256q_f32(a)     \
+	do {                \
+        float32x4_t rev_tmp = vcombine_f32(vrev64_f32(vget_high_f32(a.val[0])), vrev64_f32(vget_low_f32(a.val[0]))); \
+	a.val[0] = vcombine_f32(vrev64_f32(vget_high_f32(a.val[1])), vrev64_f32(vget_low_f32(a.val[1])));  \
+	a.val[1] = rev_tmp; \
+} while (0)
+
+#define vldrev128q_f32(a, p) \
+	do {                 \
+	a = vld1q_f32(p);    \
+	a = vrev128q_f32(a); \
+} while (0) 
+
+#endif /* ARM_MATH_NEON */
+
+#ifdef   __cplusplus
+}
+#endif
+
+#endif /* _ARM_SORTING_H */

libraries/sdk/third_party/CMSIS/DSP/PrivateInclude/arm_vec_fft.h

@@ -0,0 +1,58 @@
+/******************************************************************************
+ * @file     arm_vec_fft.h
+ * @brief    Private header file for CMSIS DSP Library
+ * @version  V1.7.0
+ * @date     07. January 2020
+ ******************************************************************************/
+/*
+ * Copyright (c) 2010-2020 Arm Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef _ARM_VEC_FFT_H_
+#define _ARM_VEC_FFT_H_
+
+#include "arm_math.h"
+#include "arm_helium_utils.h"
+
+#ifdef   __cplusplus
+extern "C"
+{
+#endif
+
+#if (defined(ARM_MATH_MVEF) || defined(ARM_MATH_MVEI) || defined(ARM_MATH_HELIUM)) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#define MVE_CMPLX_ADD_A_ixB(A, B)           vcaddq_rot90(A,B)
+#define MVE_CMPLX_SUB_A_ixB(A,B)            vcaddq_rot270(A,B)
+#define MVE_CMPLX_MULT_FLT_AxB(A,B)         vcmlaq_rot90(vcmulq(A, B), A, B)
+#define MVE_CMPLX_MULT_FLT_Conj_AxB(A,B)    vcmlaq_rot270(vcmulq(A, B), A, B)
+
+#define MVE_CMPLX_MULT_FX_AxB(A,B)          vqdmladhxq(vqdmlsdhq((__typeof(A))vuninitializedq_s32(), A, B), A, B);
+#define MVE_CMPLX_MULT_FX_AxConjB(A,B)      vqdmladhq(vqdmlsdhxq((__typeof(A))vuninitializedq_s32(), A, B), A, B);
+
+#define MVE_CMPLX_ADD_FX_A_ixB(A, B)        vhcaddq_rot90(A,B)
+#define MVE_CMPLX_SUB_FX_A_ixB(A,B)         vhcaddq_rot270(A,B)
+
+
+#endif /* (defined(ARM_MATH_MVEF) || defined(ARM_MATH_HELIUM)) && !defined(ARM_MATH_AUTOVECTORIZE)*/
+
+
+#ifdef   __cplusplus
+}
+#endif
+
+
+#endif /* _ARM_VEC_FFT_H_ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_abs_f32.c

@@ -0,0 +1,196 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_abs_f32.c
+ * Description:  Floating-point vector absolute value
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+#include <math.h>
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @defgroup BasicAbs Vector Absolute Value
+
+  Computes the absolute value of a vector on an element-by-element basis.
+
+  <pre>
+      pDst[n] = abs(pSrc[n]),   0 <= n < blockSize.
+  </pre>
+
+  The functions support in-place computation allowing the source and
+  destination pointers to reference the same memory buffer.
+  There are separate functions for floating-point, Q7, Q15, and Q31 data types.
+ */
+
+/**
+  @addtogroup BasicAbs
+  @{
+ */
+
+/**
+  @brief         Floating-point vector absolute value.
+  @param[in]     pSrc       points to the input vector
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#include "arm_helium_utils.h"
+
+void arm_abs_f32(
+  const float32_t * pSrc,
+        float32_t * pDst,
+        uint32_t blockSize)
+{
+    uint32_t blkCnt;                               /* Loop counter */
+    f32x4_t vec1;
+    f32x4_t res;
+
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        /* C = |A| */
+
+        /* Calculate absolute values and then store the results in the destination buffer. */
+        vec1 = vld1q(pSrc);
+        res = vabsq(vec1);
+        vst1q(pDst, res);
+
+        /* Increment pointers */
+        pSrc += 4;
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+
+    if (blkCnt > 0U)
+    {
+      /* C = |A| */
+      mve_pred16_t p0 = vctp32q(blkCnt);
+      vec1 = vld1q(pSrc);
+      vstrwq_p(pDst, vabsq(vec1), p0);
+    }
+
+}
+
+#else
+void arm_abs_f32(
+  const float32_t * pSrc,
+        float32_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    f32x4_t vec1;
+    f32x4_t res;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        /* C = |A| */
+
+    	/* Calculate absolute values and then store the results in the destination buffer. */
+        vec1 = vld1q_f32(pSrc);
+        res = vabsq_f32(vec1);
+        vst1q_f32(pDst, res);
+
+        /* Increment pointers */
+        pSrc += 4;
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+#else
+#if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = |A| */
+
+    /* Calculate absolute and store result in destination buffer. */
+    *pDst++ = fabsf(*pSrc++);
+
+    *pDst++ = fabsf(*pSrc++);
+
+    *pDst++ = fabsf(*pSrc++);
+
+    *pDst++ = fabsf(*pSrc++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined(ARM_MATH_NEON) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = |A| */
+
+    /* Calculate absolute and store result in destination buffer. */
+    *pDst++ = fabsf(*pSrc++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+/**
+  @} end of BasicAbs group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_abs_q15.c

@@ -0,0 +1,178 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_abs_q15.c
+ * Description:  Q15 vector absolute value
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicAbs
+  @{
+ */
+
+/**
+  @brief         Q15 vector absolute value.
+  @param[in]     pSrc       points to the input vector
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   The Q15 value -1 (0x8000) will be saturated to the maximum allowable positive value 0x7FFF.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_abs_q15(
+    const q15_t * pSrc,
+    q15_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q15x8_t vecSrc;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = |A|
+         * Calculate absolute and then store the results in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vst1q(pDst, vqabsq(vecSrc));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 8;
+        pDst += 8;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 7;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp16q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vstrhq_p(pDst, vqabsq(vecSrc), p0);
+    }
+}
+
+#else
+void arm_abs_q15(
+  const q15_t * pSrc,
+        q15_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q15_t in;                                      /* Temporary input variable */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = |A| */
+
+    /* Calculate absolute of input (if -1 then saturated to 0x7fff) and store result in destination buffer. */
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q15_t)__QSUB16(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == (q15_t) 0x8000) ? 0x7fff : -in);
+#endif
+
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q15_t)__QSUB16(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == (q15_t) 0x8000) ? 0x7fff : -in);
+#endif
+
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q15_t)__QSUB16(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == (q15_t) 0x8000) ? 0x7fff : -in);
+#endif
+
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q15_t)__QSUB16(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == (q15_t) 0x8000) ? 0x7fff : -in);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = |A| */
+
+    /* Calculate absolute of input (if -1 then saturated to 0x7fff) and store result in destination buffer. */
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q15_t)__QSUB16(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == (q15_t) 0x8000) ? 0x7fff : -in);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicAbs group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_abs_q31.c

@@ -0,0 +1,208 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_abs_q31.c
+ * Description:  Q31 vector absolute value
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicAbs
+  @{
+ */
+
+/**
+  @brief         Q31 vector absolute value.
+  @param[in]     pSrc       points to the input vector
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   The Q31 value -1 (0x80000000) will be saturated to the maximum allowable positive value 0x7FFFFFFF.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_abs_q31(
+    const q31_t * pSrc,
+    q31_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* Loop counters */
+    q31x4_t vecSrc;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2;
+
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = |A|
+         * Calculate absolute and then store the results in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vst1q(pDst, vqabsq(vecSrc));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * Advance vector source and destination pointers
+         */
+        pSrc += 4;
+        pDst += 4;
+    }
+    /*
+     * Tail
+     */
+    blkCnt = blockSize & 3;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vstrwq_p(pDst, vqabsq(vecSrc), p0);
+    }
+}
+
+#else
+void arm_abs_q31(
+  const q31_t * pSrc,
+        q31_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q31_t in;                                      /* Temporary variable */
+
+#if defined(ARM_MATH_NEON)
+    int32x4_t vec1;
+    int32x4_t res;
+
+    /* Compute 4 outputs at a time */  
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        /* C = |A| */
+        /* Calculate absolute and then store the results in the destination buffer. */
+
+        vec1 = vld1q_s32(pSrc);
+        res = vqabsq_s32(vec1);
+        vst1q_s32(pDst, res);
+
+        /* Increment pointers */
+        pSrc += 4;
+        pDst += 4;
+        
+        /* Decrement the blockSize loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+#else
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = |A| */
+
+    /* Calculate absolute of input (if -1 then saturated to 0x7fffffff) and store result in destination buffer. */
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q31_t)__QSUB(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == INT32_MIN) ? INT32_MAX : -in);
+#endif
+
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q31_t)__QSUB(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == INT32_MIN) ? INT32_MAX : -in);
+#endif
+
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q31_t)__QSUB(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == INT32_MIN) ? INT32_MAX : -in);
+#endif
+
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q31_t)__QSUB(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == INT32_MIN) ? INT32_MAX : -in);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined (ARM_MATH_NEON) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = |A| */
+
+    /* Calculate absolute of input (if -1 then saturated to 0x7fffffff) and store result in destination buffer. */
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q31_t)__QSUB(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == INT32_MIN) ? INT32_MAX : -in);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* #if defined (ARM_MATH_MVEI) */
+/**
+  @} end of BasicAbs group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_abs_q7.c

@@ -0,0 +1,180 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_abs_q7.c
+ * Description:  Q7 vector absolute value
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicAbs
+  @{
+ */
+
+/**
+  @brief         Q7 vector absolute value.
+  @param[in]     pSrc       points to the input vector
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Conditions for optimum performance
+                   Input and output buffers should be aligned by 32-bit
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   The Q7 value -1 (0x80) will be saturated to the maximum allowable positive value 0x7F.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_abs_q7(
+    const q7_t * pSrc,
+    q7_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q7x16_t vecSrc;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = |A|
+         * Calculate absolute and then store the results in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vst1q(pDst, vqabsq(vecSrc));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 16;
+        pDst += 16;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 0xF;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp8q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vstrbq_p(pDst, vqabsq(vecSrc), p0);
+    }
+}
+
+#else
+void arm_abs_q7(
+  const q7_t * pSrc,
+        q7_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q7_t in;                                       /* Temporary input variable */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = |A| */
+
+    /* Calculate absolute of input (if -1 then saturated to 0x7f) and store result in destination buffer. */
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q7_t)__QSUB8(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == (q7_t) 0x80) ? (q7_t) 0x7f : -in);
+#endif
+
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q7_t)__QSUB8(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == (q7_t) 0x80) ? (q7_t) 0x7f : -in);
+#endif
+
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q7_t)__QSUB8(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == (q7_t) 0x80) ? (q7_t) 0x7f : -in);
+#endif
+
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q7_t)__QSUB8(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == (q7_t) 0x80) ? (q7_t) 0x7f : -in);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = |A| */
+
+    /* Calculate absolute of input (if -1 then saturated to 0x7f) and store result in destination buffer. */
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (in > 0) ? in : (q7_t) __QSUB8(0, in);
+#else
+    *pDst++ = (in > 0) ? in : ((in == (q7_t) 0x80) ? (q7_t) 0x7f : -in);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicAbs group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_add_f32.c

@@ -0,0 +1,199 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_add_f32.c
+ * Description:  Floating-point vector addition
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @defgroup BasicAdd Vector Addition
+
+  Element-by-element addition of two vectors.
+
+  <pre>
+      pDst[n] = pSrcA[n] + pSrcB[n],   0 <= n < blockSize.
+  </pre>
+
+  There are separate functions for floating-point, Q7, Q15, and Q31 data types.
+ */
+
+/**
+  @addtogroup BasicAdd
+  @{
+ */
+
+/**
+  @brief         Floating-point vector addition.
+  @param[in]     pSrcA      points to first input vector
+  @param[in]     pSrcB      points to second input vector
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#include "arm_helium_utils.h"
+
+void arm_add_f32(
+  const float32_t * pSrcA,
+  const float32_t * pSrcB,
+        float32_t * pDst,
+        uint32_t blockSize)
+{
+    uint32_t blkCnt;                               /* Loop counter */
+
+    f32x4_t vec1;
+    f32x4_t vec2;
+    f32x4_t res;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        /* C = A + B */
+
+        /* Add and then store the results in the destination buffer. */
+        vec1 = vld1q(pSrcA);
+        vec2 = vld1q(pSrcB);
+        res = vaddq(vec1, vec2);
+        vst1q(pDst, res);
+
+        /* Increment pointers */
+        pSrcA += 4;
+        pSrcB += 4; 
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+    if (blkCnt > 0U)
+    {
+      /* C = A + B */
+      mve_pred16_t p0 = vctp32q(blkCnt);
+      vec1 = vld1q(pSrcA);
+      vec2 = vld1q(pSrcB);
+      vstrwq_p(pDst, vaddq(vec1,vec2), p0);
+    }
+
+}
+
+#else
+void arm_add_f32(
+  const float32_t * pSrcA,
+  const float32_t * pSrcB,
+        float32_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    f32x4_t vec1;
+    f32x4_t vec2;
+    f32x4_t res;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        /* C = A + B */
+
+    	/* Add and then store the results in the destination buffer. */
+        vec1 = vld1q_f32(pSrcA);
+        vec2 = vld1q_f32(pSrcB);
+        res = vaddq_f32(vec1, vec2);
+        vst1q_f32(pDst, res);
+
+        /* Increment pointers */
+        pSrcA += 4;
+        pSrcB += 4; 
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+#else
+#if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + B */
+
+    /* Add and store result in destination buffer. */
+    *pDst++ = (*pSrcA++) + (*pSrcB++);
+    *pDst++ = (*pSrcA++) + (*pSrcB++);
+    *pDst++ = (*pSrcA++) + (*pSrcB++);
+    *pDst++ = (*pSrcA++) + (*pSrcB++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined(ARM_MATH_NEON) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + B */
+
+    /* Add and store result in destination buffer. */
+    *pDst++ = (*pSrcA++) + (*pSrcB++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+/**
+  @} end of BasicAdd group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_add_q15.c

@@ -0,0 +1,176 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_add_q15.c
+ * Description:  Q15 vector addition
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicAdd
+  @{
+ */
+
+/**
+  @brief         Q15 vector addition.
+  @param[in]     pSrcA      points to the first input vector
+  @param[in]     pSrcB      points to the second input vector
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q15 range [0x8000 0x7FFF] are saturated.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_add_q15(
+    const q15_t * pSrcA,
+    const q15_t * pSrcB,
+    q15_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q15x8_t vecA;
+    q15x8_t vecB;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A + B
+         * Add and then store the results in the destination buffer.
+         */
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vst1q(pDst, vqaddq(vecA, vecB));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrcA  += 8;
+        pSrcB  += 8;
+        pDst   += 8;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 7;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp16q(blkCnt);
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vstrhq_p(pDst, vqaddq(vecA, vecB), p0);
+    }
+}
+
+#else
+void arm_add_q15(
+  const q15_t * pSrcA,
+  const q15_t * pSrcB,
+        q15_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+#if defined (ARM_MATH_DSP)
+  q31_t inA1, inA2;
+  q31_t inB1, inB2;
+#endif
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + B */
+
+#if defined (ARM_MATH_DSP)
+    /* read 2 times 2 samples at a time from sourceA */
+    inA1 = read_q15x2_ia ((q15_t **) &pSrcA);
+    inA2 = read_q15x2_ia ((q15_t **) &pSrcA);
+    /* read 2 times 2 samples at a time from sourceB */
+    inB1 = read_q15x2_ia ((q15_t **) &pSrcB);
+    inB2 = read_q15x2_ia ((q15_t **) &pSrcB);
+
+    /* Add and store 2 times 2 samples at a time */
+    write_q15x2_ia (&pDst, __QADD16(inA1, inB1));
+    write_q15x2_ia (&pDst, __QADD16(inA2, inB2));
+#else
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + B */
+
+    /* Add and store result in destination buffer. */
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (q15_t) __QADD16(*pSrcA++, *pSrcB++);
+#else
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+/**
+  @} end of BasicAdd group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_add_q31.c

@@ -0,0 +1,159 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_add_q31.c
+ * Description:  Q31 vector addition
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicAdd
+  @{
+ */
+
+/**
+  @brief         Q31 vector addition.
+  @param[in]     pSrcA      points to the first input vector
+  @param[in]     pSrcB      points to the second input vector
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q31 range [0x80000000 0x7FFFFFFF] are saturated.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_add_q31(
+  const q31_t * pSrcA,
+  const q31_t * pSrcB,
+        q31_t * pDst,
+        uint32_t blockSize)
+{
+    uint32_t blkCnt;   
+    q31x4_t vecA;
+    q31x4_t vecB;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A + B
+         * Add and then store the results in the destination buffer.
+         */
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vst1q(pDst, vqaddq(vecA, vecB));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrcA  += 4;
+        pSrcB  += 4;
+        pDst   += 4;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 3;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vstrwq_p(pDst, vqaddq(vecA, vecB), p0);
+    }
+}
+
+#else
+void arm_add_q31(
+  const q31_t * pSrcA,
+  const q31_t * pSrcB,
+        q31_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + B */
+
+    /* Add and store result in destination buffer. */
+    *pDst++ = __QADD(*pSrcA++, *pSrcB++);
+
+    *pDst++ = __QADD(*pSrcA++, *pSrcB++);
+
+    *pDst++ = __QADD(*pSrcA++, *pSrcB++);
+
+    *pDst++ = __QADD(*pSrcA++, *pSrcB++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + B */
+
+    /* Add and store result in destination buffer. */
+    *pDst++ = __QADD(*pSrcA++, *pSrcB++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+
+#endif /* defined(ARM_MATH_MVEI) */
+/**
+  @} end of BasicAdd group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_add_q7.c

@@ -0,0 +1,158 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_add_q7.c
+ * Description:  Q7 vector addition
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicAdd
+  @{
+ */
+
+/**
+  @brief         Q7 vector addition.
+  @param[in]     pSrcA      points to the first input vector
+  @param[in]     pSrcB      points to the second input vector
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q7 range [0x80 0x7F] are saturated.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_add_q7(
+    const q7_t * pSrcA,
+    const q7_t * pSrcB,
+    q7_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q7x16_t vecA;
+    q7x16_t vecB;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A + B
+         * Add and then store the results in the destination buffer.
+         */
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vst1q(pDst, vqaddq(vecA, vecB));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrcA  += 16;
+        pSrcB  += 16;
+        pDst   += 16;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 0xF;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp8q(blkCnt);
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vstrbq_p(pDst, vqaddq(vecA, vecB), p0);
+    }
+}
+#else
+void arm_add_q7(
+  const q7_t * pSrcA,
+  const q7_t * pSrcB,
+        q7_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + B */
+
+#if defined (ARM_MATH_DSP)
+    /* Add and store result in destination buffer (4 samples at a time). */
+    write_q7x4_ia (&pDst, __QADD8 (read_q7x4_ia ((q7_t **) &pSrcA), read_q7x4_ia ((q7_t **) &pSrcB)));
+#else
+    *pDst++ = (q7_t) __SSAT ((q15_t) *pSrcA++ + *pSrcB++, 8);
+    *pDst++ = (q7_t) __SSAT ((q15_t) *pSrcA++ + *pSrcB++, 8);
+    *pDst++ = (q7_t) __SSAT ((q15_t) *pSrcA++ + *pSrcB++, 8);
+    *pDst++ = (q7_t) __SSAT ((q15_t) *pSrcA++ + *pSrcB++, 8);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + B */
+
+    /* Add and store result in destination buffer. */
+    *pDst++ = (q7_t) __SSAT((q15_t) *pSrcA++ + *pSrcB++, 8);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+/**
+  @} end of BasicAdd group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_and_u16.c

@@ -0,0 +1,137 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_and_u16.c
+ * Description:  uint16_t bitwise AND
+ *
+ * $Date:        14 November 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @defgroup And Vector bitwise AND
+
+  Compute the logical bitwise AND.
+
+  There are separate functions for uint32_t, uint16_t, and uint7_t data types.
+ */
+
+/**
+  @addtogroup And
+  @{
+ */
+
+/**
+  @brief         Compute the logical bitwise AND of two fixed-point vectors.
+  @param[in]     pSrcA      points to input vector A
+  @param[in]     pSrcB      points to input vector B
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+void arm_and_u16(
+    const uint16_t * pSrcA,
+    const uint16_t * pSrcB,
+          uint16_t * pDst,
+          uint32_t blockSize)
+{
+    uint32_t blkCnt;      /* Loop counter */
+
+#if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
+    q15x8_t vecSrcA, vecSrcB;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3;
+
+    while (blkCnt > 0U)
+    {
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+
+        vst1q(pDst, vandq_u16(vecSrcA, vecSrcB) );
+
+        pSrcA += 8;
+        pSrcB += 8;
+        pDst  += 8;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 7;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp16q(blkCnt);
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+        vstrhq_p(pDst, vandq_u16(vecSrcA, vecSrcB), p0);
+    }
+#else
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    uint16x8_t vecA, vecB;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3U;
+
+    while (blkCnt > 0U)
+    {
+        vecA = vld1q_u16(pSrcA);
+        vecB = vld1q_u16(pSrcB);
+
+        vst1q_u16(pDst, vandq_u16(vecA, vecB) );
+
+        pSrcA += 8;
+        pSrcB += 8;
+        pDst  += 8;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 7;
+#else
+    /* Initialize blkCnt with number of samples */
+    blkCnt = blockSize;
+#endif
+
+    while (blkCnt > 0U)
+    {
+        *pDst++ = (*pSrcA++)&(*pSrcB++);
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+#endif /* if defined(ARM_MATH_MVEI) */
+}
+
+/**
+  @} end of And group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_and_u32.c

@@ -0,0 +1,129 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_and_u32.c
+ * Description:  uint32_t bitwise AND
+ *
+ * $Date:        14 November 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup And
+  @{
+ */
+
+/**
+  @brief         Compute the logical bitwise AND of two fixed-point vectors.
+  @param[in]     pSrcA      points to input vector A
+  @param[in]     pSrcB      points to input vector B
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+void arm_and_u32(
+    const uint32_t * pSrcA,
+    const uint32_t * pSrcB,
+          uint32_t * pDst,
+          uint32_t blockSize)
+{
+    uint32_t blkCnt;      /* Loop counter */
+
+#if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
+    q31x4_t vecSrcA, vecSrcB;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2;
+
+    while (blkCnt > 0U)
+    {
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+
+        vst1q(pDst, vandq_u32(vecSrcA, vecSrcB) );
+
+        pSrcA += 4;
+        pSrcB += 4;
+        pDst  += 4;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 3;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+        vstrwq_p(pDst, vandq_u32(vecSrcA, vecSrcB), p0);
+    }
+#else
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    uint32x4_t vecA, vecB;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        vecA = vld1q_u32(pSrcA);
+        vecB = vld1q_u32(pSrcB);
+
+        vst1q_u32(pDst, vandq_u32(vecA, vecB) );
+
+        pSrcA += 4;
+        pSrcB += 4;
+        pDst  += 4;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 3;
+#else
+    /* Initialize blkCnt with number of samples */
+    blkCnt = blockSize;
+#endif
+
+    while (blkCnt > 0U)
+    {
+        *pDst++ = (*pSrcA++)&(*pSrcB++);
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+#endif /* if defined(ARM_MATH_MVEI) */
+}
+
+/**
+  @} end of And group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_and_u8.c

@@ -0,0 +1,130 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_and_u8.c
+ * Description:  uint8_t bitwise AND
+ *
+ * $Date:        14 November 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+
+/**
+  @addtogroup And
+  @{
+ */
+
+/**
+  @brief         Compute the logical bitwise AND of two fixed-point vectors.
+  @param[in]     pSrcA      points to input vector A
+  @param[in]     pSrcB      points to input vector B
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+void arm_and_u8(
+    const uint8_t * pSrcA,
+    const uint8_t * pSrcB,
+          uint8_t * pDst,
+          uint32_t blockSize)
+{
+    uint32_t blkCnt;      /* Loop counter */
+
+#if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
+    q7x16_t vecSrcA, vecSrcB;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4;
+
+    while (blkCnt > 0U)
+    {
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+
+        vst1q(pDst, vandq_u8(vecSrcA, vecSrcB) );
+
+        pSrcA += 16;
+        pSrcB += 16;
+        pDst  += 16;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0xF;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp8q(blkCnt);
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+        vstrbq_p(pDst, vandq_u8(vecSrcA, vecSrcB), p0);
+    }
+#else
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    uint8x16_t vecA, vecB;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4U;
+
+    while (blkCnt > 0U)
+    {
+        vecA = vld1q_u8(pSrcA);
+        vecB = vld1q_u8(pSrcB);
+
+        vst1q_u8(pDst, vandq_u8(vecA, vecB) );
+
+        pSrcA += 16;
+        pSrcB += 16;
+        pDst  += 16;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0xF;
+#else
+    /* Initialize blkCnt with number of samples */
+    blkCnt = blockSize;
+#endif
+
+    while (blkCnt > 0U)
+    {
+        *pDst++ = (*pSrcA++)&(*pSrcB++);
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+#endif /* if defined(ARM_MATH_MVEI) */
+}
+
+/**
+  @} end of And group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_dot_prod_f32.c

@@ -0,0 +1,226 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_dot_prod_f32.c
+ * Description:  Floating-point dot product
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @defgroup BasicDotProd Vector Dot Product
+
+  Computes the dot product of two vectors.
+  The vectors are multiplied element-by-element and then summed.
+
+  <pre>
+      sum = pSrcA[0]*pSrcB[0] + pSrcA[1]*pSrcB[1] + ... + pSrcA[blockSize-1]*pSrcB[blockSize-1]
+  </pre>
+
+  There are separate functions for floating-point, Q7, Q15, and Q31 data types.
+ */
+
+/**
+  @addtogroup BasicDotProd
+  @{
+ */
+
+/**
+  @brief         Dot product of floating-point vectors.
+  @param[in]     pSrcA      points to the first input vector.
+  @param[in]     pSrcB      points to the second input vector.
+  @param[in]     blockSize  number of samples in each vector.
+  @param[out]    result     output result returned here.
+  @return        none
+ */
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#include "arm_helium_utils.h"
+
+
+void arm_dot_prod_f32(
+    const float32_t * pSrcA,
+    const float32_t * pSrcB,
+    uint32_t    blockSize,
+    float32_t * result)
+{
+    f32x4_t vecA, vecB;
+    f32x4_t vecSum;
+    uint32_t blkCnt; 
+    float32_t sum = 0.0f;  
+    vecSum = vdupq_n_f32(0.0f);
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1]
+         * Calculate dot product and then store the result in a temporary buffer.
+         * and advance vector source and destination pointers
+         */
+        vecA = vld1q(pSrcA);
+        pSrcA += 4;
+        
+        vecB = vld1q(pSrcB);
+        pSrcB += 4;
+
+        vecSum = vfmaq(vecSum, vecA, vecB);
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt --;
+    }
+
+
+    blkCnt = blockSize & 3;
+    if (blkCnt > 0U)
+    {
+        /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
+
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vecSum = vfmaq_m(vecSum, vecA, vecB, p0);
+    }
+
+    sum = vecAddAcrossF32Mve(vecSum);
+
+    /* Store result in destination buffer */
+    *result = sum;
+
+}
+
+#else
+
+void arm_dot_prod_f32(
+  const float32_t * pSrcA,
+  const float32_t * pSrcB,
+        uint32_t blockSize,
+        float32_t * result)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        float32_t sum = 0.0f;                          /* Temporary return variable */
+
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    f32x4_t vec1;
+    f32x4_t vec2;
+    f32x4_t accum = vdupq_n_f32(0);   
+    f32x2_t tmp = vdup_n_f32(0);    
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    vec1 = vld1q_f32(pSrcA);
+    vec2 = vld1q_f32(pSrcB);
+
+    while (blkCnt > 0U)
+    {
+        /* C = A[0]*B[0] + A[1]*B[1] + A[2]*B[2] + ... + A[blockSize-1]*B[blockSize-1] */
+        /* Calculate dot product and then store the result in a temporary buffer. */
+        
+	      accum = vmlaq_f32(accum, vec1, vec2);
+	
+        /* Increment pointers */
+        pSrcA += 4;
+        pSrcB += 4; 
+
+        vec1 = vld1q_f32(pSrcA);
+        vec2 = vld1q_f32(pSrcB);
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+    
+#if __aarch64__
+    sum = vpadds_f32(vpadd_f32(vget_low_f32(accum), vget_high_f32(accum)));
+#else
+    tmp = vpadd_f32(vget_low_f32(accum), vget_high_f32(accum));
+    sum = vget_lane_f32(tmp, 0) + vget_lane_f32(tmp, 1);
+
+#endif    
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+#else
+#if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  /* First part of the processing with loop unrolling. Compute 4 outputs at a time.
+   ** a second loop below computes the remaining 1 to 3 samples. */
+  while (blkCnt > 0U)
+  {
+    /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
+
+    /* Calculate dot product and store result in a temporary buffer. */
+    sum += (*pSrcA++) * (*pSrcB++);
+
+    sum += (*pSrcA++) * (*pSrcB++);
+
+    sum += (*pSrcA++) * (*pSrcB++);
+
+    sum += (*pSrcA++) * (*pSrcB++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined(ARM_MATH_NEON) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
+
+    /* Calculate dot product and store result in a temporary buffer. */
+    sum += (*pSrcA++) * (*pSrcB++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Store result in destination buffer */
+  *result = sum;
+}
+
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+/**
+  @} end of BasicDotProd group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_dot_prod_q15.c

@@ -0,0 +1,172 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_dot_prod_q15.c
+ * Description:  Q15 dot product
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicDotProd
+  @{
+ */
+
+/**
+  @brief         Dot product of Q15 vectors.
+  @param[in]     pSrcA      points to the first input vector
+  @param[in]     pSrcB      points to the second input vector
+  @param[in]     blockSize  number of samples in each vector
+  @param[out]    result     output result returned here
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The intermediate multiplications are in 1.15 x 1.15 = 2.30 format and these
+                   results are added to a 64-bit accumulator in 34.30 format.
+                   Nonsaturating additions are used and given that there are 33 guard bits in the accumulator
+                   there is no risk of overflow.
+                   The return result is in 34.30 format.
+ */
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_dot_prod_q15(
+    const q15_t * pSrcA,
+    const q15_t * pSrcB,
+    uint32_t blockSize,
+    q63_t * result)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q15x8_t vecA;
+    q15x8_t vecB;
+    q63_t     sum = 0LL;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1]
+         * Calculate dot product and then store the result in a temporary buffer.
+         */
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        sum = vmlaldavaq(sum, vecA, vecB);
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrcA += 8;
+        pSrcB += 8;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 7;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp16q(blkCnt);
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        sum = vmlaldavaq_p(sum, vecA, vecB, p0);
+    }
+
+    *result = sum;
+}
+
+#else
+void arm_dot_prod_q15(
+  const q15_t * pSrcA,
+  const q15_t * pSrcB,
+        uint32_t blockSize,
+        q63_t * result)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q63_t sum = 0;                                 /* Temporary return variable */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
+
+#if defined (ARM_MATH_DSP)
+    /* Calculate dot product and store result in a temporary buffer. */
+    sum = __SMLALD(read_q15x2_ia ((q15_t **) &pSrcA), read_q15x2_ia ((q15_t **) &pSrcB), sum);
+    sum = __SMLALD(read_q15x2_ia ((q15_t **) &pSrcA), read_q15x2_ia ((q15_t **) &pSrcB), sum);
+#else
+    sum += (q63_t)((q31_t) *pSrcA++ * *pSrcB++);
+    sum += (q63_t)((q31_t) *pSrcA++ * *pSrcB++);
+    sum += (q63_t)((q31_t) *pSrcA++ * *pSrcB++);
+    sum += (q63_t)((q31_t) *pSrcA++ * *pSrcB++);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
+
+    /* Calculate dot product and store result in a temporary buffer. */
+//#if defined (ARM_MATH_DSP)
+//    sum  = __SMLALD(*pSrcA++, *pSrcB++, sum);
+//#else
+    sum += (q63_t)((q31_t) *pSrcA++ * *pSrcB++);
+//#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Store result in destination buffer in 34.30 format */
+  *result = sum;
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicDotProd group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_dot_prod_q31.c

@@ -0,0 +1,174 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_dot_prod_q31.c
+ * Description:  Q31 dot product
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicDotProd
+  @{
+ */
+
+/**
+  @brief         Dot product of Q31 vectors.
+  @param[in]     pSrcA      points to the first input vector.
+  @param[in]     pSrcB      points to the second input vector.
+  @param[in]     blockSize  number of samples in each vector.
+  @param[out]    result     output result returned here.
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The intermediate multiplications are in 1.31 x 1.31 = 2.62 format and these
+                   are truncated to 2.48 format by discarding the lower 14 bits.
+                   The 2.48 result is then added without saturation to a 64-bit accumulator in 16.48 format.
+                   There are 15 guard bits in the accumulator and there is no risk of overflow as long as
+                   the length of the vectors is less than 2^16 elements.
+                   The return result is in 16.48 format.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_dot_prod_q31(
+    const q31_t * pSrcA,
+    const q31_t * pSrcB,
+    uint32_t blockSize,
+    q63_t * result)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q31x4_t vecA;
+    q31x4_t vecB;
+    q63_t     sum = 0LL;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1]
+         * Calculate dot product and then store the result in a temporary buffer.
+         */
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        sum = vrmlaldavhaq(sum, vecA, vecB);
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrcA += 4;
+        pSrcB += 4;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 3;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        sum = vrmlaldavhaq_p(sum, vecA, vecB, p0);
+    }
+
+    /*
+     * vrmlaldavhaq provides extra intermediate accumulator headroom.
+     * limiting the need of intermediate scaling
+     * Scalar variant uses 2.48 accu format by right shifting accumulators by 14.
+     * 16.48 output conversion is performed outside the loop by scaling accu. by 6
+     */
+    *result = asrl(sum, (14 - 8));
+}
+
+#else
+void arm_dot_prod_q31(
+  const q31_t * pSrcA,
+  const q31_t * pSrcB,
+        uint32_t blockSize,
+        q63_t * result)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q63_t sum = 0;                                 /* Temporary return variable */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
+
+    /* Calculate dot product and store result in a temporary buffer. */
+    sum += ((q63_t) *pSrcA++ * *pSrcB++) >> 14U;
+
+    sum += ((q63_t) *pSrcA++ * *pSrcB++) >> 14U;
+
+    sum += ((q63_t) *pSrcA++ * *pSrcB++) >> 14U;
+
+    sum += ((q63_t) *pSrcA++ * *pSrcB++) >> 14U;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
+
+    /* Calculate dot product and store result in a temporary buffer. */
+    sum += ((q63_t) *pSrcA++ * *pSrcB++) >> 14U;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Store result in destination buffer in 16.48 format */
+  *result = sum;
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicDotProd group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_dot_prod_q7.c

@@ -0,0 +1,191 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_dot_prod_q7.c
+ * Description:  Q7 dot product
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicDotProd
+  @{
+ */
+
+/**
+  @brief         Dot product of Q7 vectors.
+  @param[in]     pSrcA      points to the first input vector
+  @param[in]     pSrcB      points to the second input vector
+  @param[in]     blockSize  number of samples in each vector
+  @param[out]    result     output result returned here
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The intermediate multiplications are in 1.7 x 1.7 = 2.14 format and these
+                   results are added to an accumulator in 18.14 format.
+                   Nonsaturating additions are used and there is no danger of wrap around as long as
+                   the vectors are less than 2^18 elements long.
+                   The return result is in 18.14 format.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_dot_prod_q7(
+    const q7_t * pSrcA,
+    const q7_t * pSrcB,
+    uint32_t blockSize,
+    q31_t * result)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q7x16_t vecA;
+    q7x16_t vecB;
+    q31_t     sum = 0;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1]
+         * Calculate dot product and then store the result in a temporary buffer.
+         */
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        sum = vmladavaq(sum, vecA, vecB);
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrcA += 16;
+        pSrcB += 16;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 0xF;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp8q(blkCnt);
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        sum = vmladavaq_p(sum, vecA, vecB, p0);
+    }
+
+    *result = sum;
+}
+#else
+void arm_dot_prod_q7(
+  const q7_t * pSrcA,
+  const q7_t * pSrcB,
+        uint32_t blockSize,
+        q31_t * result)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q31_t sum = 0;                                 /* Temporary return variable */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+#if defined (ARM_MATH_DSP)
+  q31_t input1, input2;                          /* Temporary variables */
+  q31_t inA1, inA2, inB1, inB2;                  /* Temporary variables */
+#endif
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
+
+#if defined (ARM_MATH_DSP)
+    /* read 4 samples at a time from sourceA */
+    input1 = read_q7x4_ia ((q7_t **) &pSrcA);
+    /* read 4 samples at a time from sourceB */
+    input2 = read_q7x4_ia ((q7_t **) &pSrcB);
+
+    /* extract two q7_t samples to q15_t samples */
+    inA1 = __SXTB16(__ROR(input1, 8));
+    /* extract reminaing two samples */
+    inA2 = __SXTB16(input1);
+    /* extract two q7_t samples to q15_t samples */
+    inB1 = __SXTB16(__ROR(input2, 8));
+    /* extract reminaing two samples */
+    inB2 = __SXTB16(input2);
+
+    /* multiply and accumulate two samples at a time */
+    sum = __SMLAD(inA1, inB1, sum);
+    sum = __SMLAD(inA2, inB2, sum);
+#else
+    sum += (q31_t) ((q15_t) *pSrcA++ * *pSrcB++);
+    sum += (q31_t) ((q15_t) *pSrcA++ * *pSrcB++);
+    sum += (q31_t) ((q15_t) *pSrcA++ * *pSrcB++);
+    sum += (q31_t) ((q15_t) *pSrcA++ * *pSrcB++);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
+
+    /* Calculate dot product and store result in a temporary buffer. */
+//#if defined (ARM_MATH_DSP)
+//    sum  = __SMLAD(*pSrcA++, *pSrcB++, sum);
+//#else
+    sum += (q31_t) ((q15_t) *pSrcA++ * *pSrcB++);
+//#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Store result in destination buffer in 18.14 format */
+  *result = sum;
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicDotProd group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_mult_f32.c

@@ -0,0 +1,200 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_mult_f32.c
+ * Description:  Floating-point vector multiplication
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @defgroup BasicMult Vector Multiplication
+
+  Element-by-element multiplication of two vectors.
+
+  <pre>
+      pDst[n] = pSrcA[n] * pSrcB[n],   0 <= n < blockSize.
+  </pre>
+
+  There are separate functions for floating-point, Q7, Q15, and Q31 data types.
+ */
+
+/**
+  @addtogroup BasicMult
+  @{
+ */
+
+/**
+  @brief         Floating-point vector multiplication.
+  @param[in]     pSrcA      points to the first input vector.
+  @param[in]     pSrcB      points to the second input vector.
+  @param[out]    pDst       points to the output vector.
+  @param[in]     blockSize  number of samples in each vector.
+  @return        none
+ */
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#include "arm_helium_utils.h"
+
+void arm_mult_f32(
+  const float32_t * pSrcA,
+  const float32_t * pSrcB,
+        float32_t * pDst,
+        uint32_t blockSize)
+{
+    uint32_t blkCnt;                               /* Loop counter */
+
+    f32x4_t vec1;
+    f32x4_t vec2;
+    f32x4_t res;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+    while (blkCnt > 0U)
+    {
+        /* C = A + B */
+
+      /* Add and then store the results in the destination buffer. */
+        vec1 = vld1q(pSrcA);
+        vec2 = vld1q(pSrcB);
+        res = vmulq(vec1, vec2);
+        vst1q(pDst, res);
+
+        /* Increment pointers */
+        pSrcA += 4;
+        pSrcB += 4; 
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+    if (blkCnt > 0U)
+    {
+      /* C = A + B */
+      mve_pred16_t p0 = vctp32q(blkCnt);
+      vec1 = vld1q(pSrcA);
+      vec2 = vld1q(pSrcB);
+      vstrwq_p(pDst, vmulq(vec1,vec2), p0);
+    }
+
+}
+
+#else
+void arm_mult_f32(
+  const float32_t * pSrcA,
+  const float32_t * pSrcB,
+        float32_t * pDst,
+        uint32_t blockSize)
+{
+    uint32_t blkCnt;                               /* Loop counter */
+
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    f32x4_t vec1;
+    f32x4_t vec2;
+    f32x4_t res;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        /* C = A * B */
+
+    	/* Multiply the inputs and then store the results in the destination buffer. */
+        vec1 = vld1q_f32(pSrcA);
+        vec2 = vld1q_f32(pSrcB);
+        res = vmulq_f32(vec1, vec2);
+        vst1q_f32(pDst, res);
+
+        /* Increment pointers */
+        pSrcA += 4;
+        pSrcB += 4; 
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+#else
+#if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A * B */
+
+    /* Multiply inputs and store result in destination buffer. */
+    *pDst++ = (*pSrcA++) * (*pSrcB++);
+
+    *pDst++ = (*pSrcA++) * (*pSrcB++);
+
+    *pDst++ = (*pSrcA++) * (*pSrcB++);
+
+    *pDst++ = (*pSrcA++) * (*pSrcB++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined(ARM_MATH_NEON) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A * B */
+
+    /* Multiply input and store result in destination buffer. */
+    *pDst++ = (*pSrcA++) * (*pSrcB++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+/**
+  @} end of BasicMult group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_mult_q15.c

@@ -0,0 +1,192 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_mult_q15.c
+ * Description:  Q15 vector multiplication
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicMult
+  @{
+ */
+
+/**
+  @brief         Q15 vector multiplication
+  @param[in]     pSrcA      points to first input vector
+  @param[in]     pSrcB      points to second input vector
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q15 range [0x8000 0x7FFF] are saturated.
+ */
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_mult_q15(
+    const q15_t * pSrcA,
+    const q15_t * pSrcB,
+    q15_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q15x8_t vecA, vecB;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A * B
+         * Multiply the inputs and then store the results in the destination buffer.
+         */
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vst1q(pDst, vqdmulhq(vecA, vecB));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrcA  += 8;
+        pSrcB  += 8;
+        pDst   += 8;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 7;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp16q(blkCnt);
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vstrhq_p(pDst, vqdmulhq(vecA, vecB), p0);
+    }
+}
+
+#else
+void arm_mult_q15(
+  const q15_t * pSrcA,
+  const q15_t * pSrcB,
+        q15_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+#if defined (ARM_MATH_DSP)
+  q31_t inA1, inA2, inB1, inB2;                  /* Temporary input variables */
+  q15_t out1, out2, out3, out4;                  /* Temporary output variables */
+  q31_t mul1, mul2, mul3, mul4;                  /* Temporary variables */
+#endif
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A * B */
+
+#if defined (ARM_MATH_DSP)
+    /* read 2 samples at a time from sourceA */
+    inA1 = read_q15x2_ia ((q15_t **) &pSrcA);
+    /* read 2 samples at a time from sourceB */
+    inB1 = read_q15x2_ia ((q15_t **) &pSrcB);
+    /* read 2 samples at a time from sourceA */
+    inA2 = read_q15x2_ia ((q15_t **) &pSrcA);
+    /* read 2 samples at a time from sourceB */
+    inB2 = read_q15x2_ia ((q15_t **) &pSrcB);
+
+    /* multiply mul = sourceA * sourceB */
+    mul1 = (q31_t) ((q15_t) (inA1 >> 16) * (q15_t) (inB1 >> 16));
+    mul2 = (q31_t) ((q15_t) (inA1      ) * (q15_t) (inB1      ));
+    mul3 = (q31_t) ((q15_t) (inA2 >> 16) * (q15_t) (inB2 >> 16));
+    mul4 = (q31_t) ((q15_t) (inA2      ) * (q15_t) (inB2      ));
+
+    /* saturate result to 16 bit */
+    out1 = (q15_t) __SSAT(mul1 >> 15, 16);
+    out2 = (q15_t) __SSAT(mul2 >> 15, 16);
+    out3 = (q15_t) __SSAT(mul3 >> 15, 16);
+    out4 = (q15_t) __SSAT(mul4 >> 15, 16);
+
+    /* store result to destination */
+#ifndef ARM_MATH_BIG_ENDIAN
+    write_q15x2_ia (&pDst, __PKHBT(out2, out1, 16));
+    write_q15x2_ia (&pDst, __PKHBT(out4, out3, 16));
+#else
+    write_q15x2_ia (&pDst, __PKHBT(out1, out2, 16));
+    write_q15x2_ia (&pDst, __PKHBT(out3, out4, 16));
+#endif /* #ifndef ARM_MATH_BIG_ENDIAN */
+
+#else
+    *pDst++ = (q15_t) __SSAT((((q31_t) (*pSrcA++) * (*pSrcB++)) >> 15), 16);
+    *pDst++ = (q15_t) __SSAT((((q31_t) (*pSrcA++) * (*pSrcB++)) >> 15), 16);
+    *pDst++ = (q15_t) __SSAT((((q31_t) (*pSrcA++) * (*pSrcB++)) >> 15), 16);
+    *pDst++ = (q15_t) __SSAT((((q31_t) (*pSrcA++) * (*pSrcB++)) >> 15), 16);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A * B */
+
+    /* Multiply inputs and store result in destination buffer. */
+    *pDst++ = (q15_t) __SSAT((((q31_t) (*pSrcA++) * (*pSrcB++)) >> 15), 16);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicMult group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_mult_q31.c

@@ -0,0 +1,168 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_mult_q31.c
+ * Description:  Q31 vector multiplication
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicMult
+  @{
+ */
+
+/**
+  @brief         Q31 vector multiplication.
+  @param[in]     pSrcA      points to the first input vector.
+  @param[in]     pSrcB      points to the second input vector.
+  @param[out]    pDst       points to the output vector.
+  @param[in]     blockSize  number of samples in each vector.
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q31 range[0x80000000 0x7FFFFFFF] are saturated.
+ */
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_mult_q31(
+    const q31_t * pSrcA,
+    const q31_t * pSrcB,
+    q31_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q31x4_t vecA, vecB;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A * B
+         * Multiply the inputs and then store the results in the destination buffer.
+         */
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vst1q(pDst, vqdmulhq(vecA, vecB));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrcA  += 4;
+        pSrcB  += 4;
+        pDst   += 4;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 3;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vstrwq_p(pDst, vqdmulhq(vecA, vecB), p0);
+    }
+}
+
+#else
+void arm_mult_q31(
+  const q31_t * pSrcA,
+  const q31_t * pSrcB,
+        q31_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q31_t out;                                     /* Temporary output variable */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A * B */
+
+    /* Multiply inputs and store result in destination buffer. */
+    out = ((q63_t) *pSrcA++ * *pSrcB++) >> 32;
+    out = __SSAT(out, 31);
+    *pDst++ = out << 1U;
+
+    out = ((q63_t) *pSrcA++ * *pSrcB++) >> 32;
+    out = __SSAT(out, 31);
+    *pDst++ = out << 1U;
+
+    out = ((q63_t) *pSrcA++ * *pSrcB++) >> 32;
+    out = __SSAT(out, 31);
+    *pDst++ = out << 1U;
+
+    out = ((q63_t) *pSrcA++ * *pSrcB++) >> 32;
+    out = __SSAT(out, 31);
+    *pDst++ = out << 1U;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A * B */
+
+    /* Multiply inputs and store result in destination buffer. */
+    out = ((q63_t) *pSrcA++ * *pSrcB++) >> 32;
+    out = __SSAT(out, 31);
+    *pDst++ = out << 1U;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicMult group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_mult_q7.c

@@ -0,0 +1,168 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_mult_q7.c
+ * Description:  Q7 vector multiplication
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicMult
+  @{
+ */
+
+/**
+  @brief         Q7 vector multiplication
+  @param[in]     pSrcA      points to the first input vector
+  @param[in]     pSrcB      points to the second input vector
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q7 range [0x80 0x7F] are saturated.
+ */
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_mult_q7(
+    const q7_t * pSrcA,
+    const q7_t * pSrcB,
+    q7_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q7x16_t vecA, vecB;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A * B
+         * Multiply the inputs and then store the results in the destination buffer.
+         */
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vst1q(pDst, vqdmulhq(vecA, vecB));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrcA  += 16;
+        pSrcB  += 16;
+        pDst   += 16;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 0xF;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp8q(blkCnt);
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vstrbq_p(pDst, vqdmulhq(vecA, vecB), p0);
+    }
+}
+
+#else
+void arm_mult_q7(
+  const q7_t * pSrcA,
+  const q7_t * pSrcB,
+        q7_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+#if defined (ARM_MATH_DSP)
+  q7_t out1, out2, out3, out4;                   /* Temporary output variables */
+#endif
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A * B */
+
+#if defined (ARM_MATH_DSP)
+    /* Multiply inputs and store results in temporary variables */
+    out1 = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
+    out2 = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
+    out3 = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
+    out4 = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
+
+    /* Pack and store result in destination buffer (in single write) */
+    write_q7x4_ia (&pDst, __PACKq7(out1, out2, out3, out4));
+#else
+    *pDst++ = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
+    *pDst++ = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
+    *pDst++ = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
+    *pDst++ = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A * B */
+
+    /* Multiply input and store result in destination buffer. */
+    *pDst++ = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicMult group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_negate_f32.c

@@ -0,0 +1,192 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_negate_f32.c
+ * Description:  Negates floating-point vectors
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @defgroup BasicNegate Vector Negate
+
+  Negates the elements of a vector.
+
+  <pre>
+      pDst[n] = -pSrc[n],   0 <= n < blockSize.
+  </pre>
+
+  The functions support in-place computation allowing the source and
+  destination pointers to reference the same memory buffer.
+  There are separate functions for floating-point, Q7, Q15, and Q31 data types.
+ */
+
+/**
+  @addtogroup BasicNegate
+  @{
+ */
+
+/**
+  @brief         Negates the elements of a floating-point vector.
+  @param[in]     pSrc       points to input vector.
+  @param[out]    pDst       points to output vector.
+  @param[in]     blockSize  number of samples in each vector.
+  @return        none
+ */
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#include "arm_helium_utils.h"
+
+void arm_negate_f32(
+  const float32_t * pSrc,
+        float32_t * pDst,
+        uint32_t blockSize)
+{
+    uint32_t blkCnt;                               /* Loop counter */
+    f32x4_t vec1;
+    f32x4_t res;
+
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+    while (blkCnt > 0U)
+    {
+        /* C = |A| */
+
+        /* Calculate absolute values and then store the results in the destination buffer. */
+        vec1 = vld1q(pSrc);
+        res = vnegq(vec1);
+        vst1q(pDst, res);
+
+        /* Increment pointers */
+        pSrc += 4;
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+    if (blkCnt > 0U)
+    {
+      /* C = |A| */
+      mve_pred16_t p0 = vctp32q(blkCnt);
+      vec1 = vld1q((float32_t const *) pSrc);
+      vstrwq_p(pDst, vnegq(vec1), p0);
+    }
+
+}
+
+#else
+void arm_negate_f32(
+  const float32_t * pSrc,
+        float32_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined(ARM_MATH_NEON_EXPERIMENTAL) && !defined(ARM_MATH_AUTOVECTORIZE)
+    f32x4_t vec1;
+    f32x4_t res;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        /* C = -A */
+
+    	/* Negate and then store the results in the destination buffer. */
+        vec1 = vld1q_f32(pSrc);
+        res = vnegq_f32(vec1);
+        vst1q_f32(pDst, res);
+
+        /* Increment pointers */
+        pSrc += 4;
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+#else
+#if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = -A */
+
+    /* Negate and store result in destination buffer. */
+    *pDst++ = -*pSrc++;
+
+    *pDst++ = -*pSrc++;
+
+    *pDst++ = -*pSrc++;
+
+    *pDst++ = -*pSrc++;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined(ARM_MATH_NEON_EXPERIMENTAL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = -A */
+
+    /* Negate and store result in destination buffer. */
+    *pDst++ = -*pSrc++;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+/**
+  @} end of BasicNegate group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_negate_q15.c

@@ -0,0 +1,171 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_negate_q15.c
+ * Description:  Negates Q15 vectors
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicNegate
+  @{
+ */
+
+/**
+  @brief         Negates the elements of a Q15 vector.
+  @param[in]     pSrc       points to the input vector.
+  @param[out]    pDst       points to the output vector.
+  @param[in]     blockSize  number of samples in each vector.
+  @return        none
+
+  @par           Conditions for optimum performance
+                   Input and output buffers should be aligned by 32-bit
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   The Q15 value -1 (0x8000) is saturated to the maximum allowable positive value 0x7FFF.
+ */
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_negate_q15(
+    const q15_t  * pSrc,
+    q15_t  * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q15x8_t vecSrc;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = -A
+         * Negate and then store the results in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vst1q(pDst, vqnegq(vecSrc));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 8;
+        pDst += 8;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 7;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp16q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vstrhq_p(pDst, vqnegq(vecSrc), p0);
+    }
+}
+
+#else
+void arm_negate_q15(
+  const q15_t * pSrc,
+        q15_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q15_t in;                                      /* Temporary input variable */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+#if defined (ARM_MATH_DSP)
+  q31_t in1;                                    /* Temporary input variables */
+#endif
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = -A */
+
+#if defined (ARM_MATH_DSP)
+    /* Negate and store result in destination buffer (2 samples at a time). */
+    in1 = read_q15x2_ia ((q15_t **) &pSrc);
+    write_q15x2_ia (&pDst, __QSUB16(0, in1));
+
+    in1 = read_q15x2_ia ((q15_t **) &pSrc);
+    write_q15x2_ia (&pDst, __QSUB16(0, in1));
+#else
+    in = *pSrc++;
+    *pDst++ = (in == (q15_t) 0x8000) ? (q15_t) 0x7fff : -in;
+
+    in = *pSrc++;
+    *pDst++ = (in == (q15_t) 0x8000) ? (q15_t) 0x7fff : -in;
+
+    in = *pSrc++;
+    *pDst++ = (in == (q15_t) 0x8000) ? (q15_t) 0x7fff : -in;
+
+    in = *pSrc++;
+    *pDst++ = (in == (q15_t) 0x8000) ? (q15_t) 0x7fff : -in;
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = -A */
+
+    /* Negate and store result in destination buffer. */
+    in = *pSrc++;
+    *pDst++ = (in == (q15_t) 0x8000) ? (q15_t) 0x7fff : -in;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicNegate group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_negate_q31.c

@@ -0,0 +1,178 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_negate_q31.c
+ * Description:  Negates Q31 vectors
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicNegate
+  @{
+ */
+
+/**
+  @brief         Negates the elements of a Q31 vector.
+  @param[in]     pSrc       points to the input vector.
+  @param[out]    pDst       points to the output vector.
+  @param[in]     blockSize   number of samples in each vector.
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   The Q31 value -1 (0x80000000) is saturated to the maximum allowable positive value 0x7FFFFFFF.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_negate_q31(
+    const q31_t * pSrc,
+    q31_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q31x4_t vecSrc;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = -A
+         * Negate and then store the results in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vst1q(pDst, vqnegq(vecSrc));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 4;
+        pDst += 4;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 3;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vstrwq_p(pDst, vqnegq(vecSrc), p0);
+    }
+}
+
+#else
+void arm_negate_q31(
+  const q31_t * pSrc,
+        q31_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q31_t in;                                      /* Temporary input variable */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = -A */
+
+    /* Negate and store result in destination buffer. */
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QSUB(0, in);
+#else
+    *pDst++ = (in == INT32_MIN) ? INT32_MAX : -in;
+#endif
+
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QSUB(0, in);
+#else
+    *pDst++ = (in == INT32_MIN) ? INT32_MAX : -in;
+#endif
+
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QSUB(0, in);
+#else
+    *pDst++ = (in == INT32_MIN) ? INT32_MAX : -in;
+#endif
+
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QSUB(0, in);
+#else
+    *pDst++ = (in == INT32_MIN) ? INT32_MAX : -in;
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = -A */
+
+    /* Negate and store result in destination buffer. */
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QSUB(0, in);
+#else
+    *pDst++ = (in == INT32_MIN) ? INT32_MAX : -in;
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicNegate group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_negate_q7.c

@@ -0,0 +1,171 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_negate_q7.c
+ * Description:  Negates Q7 vectors
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicNegate
+  @{
+ */
+
+/**
+  @brief         Negates the elements of a Q7 vector.
+  @param[in]     pSrc       points to the input vector.
+  @param[out]    pDst       points to the output vector.
+  @param[in]     blockSize   number of samples in each vector.
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   The Q7 value -1 (0x80) is saturated to the maximum allowable positive value 0x7F.
+ */
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_negate_q7(
+    const q7_t   * pSrc,
+    q7_t   * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q7x16_t vecSrc;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = -A
+         * Negate and then store the results in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vst1q(pDst, vqnegq(vecSrc));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 16;
+        pDst += 16;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 0xF;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp8q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vstrbq_p(pDst, vqnegq(vecSrc), p0);
+    }
+}
+
+#else
+void arm_negate_q7(
+  const q7_t * pSrc,
+        q7_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q7_t in;                                       /* Temporary input variable */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+#if defined (ARM_MATH_DSP)
+  q31_t in1;                                    /* Temporary input variable */
+#endif
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = -A */
+
+#if defined (ARM_MATH_DSP)
+    /* Negate and store result in destination buffer (4 samples at a time). */
+    in1 = read_q7x4_ia ((q7_t **) &pSrc);
+    write_q7x4_ia (&pDst, __QSUB8(0, in1));
+#else
+    in = *pSrc++;
+    *pDst++ = (in == (q7_t) 0x80) ? (q7_t) 0x7f : -in;
+
+    in = *pSrc++;
+    *pDst++ = (in == (q7_t) 0x80) ? (q7_t) 0x7f : -in;
+
+    in = *pSrc++;
+    *pDst++ = (in == (q7_t) 0x80) ? (q7_t) 0x7f : -in;
+
+    in = *pSrc++;
+    *pDst++ = (in == (q7_t) 0x80) ? (q7_t) 0x7f : -in;
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = -A */
+
+    /* Negate and store result in destination buffer. */
+    in = *pSrc++;
+
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (q7_t) __QSUB8(0, in);
+#else
+    *pDst++ = (in == (q7_t) 0x80) ? (q7_t) 0x7f : -in;
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicNegate group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_not_u16.c

@@ -0,0 +1,130 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_not_u16.c
+ * Description:  uint16_t bitwise NOT
+ *
+ * $Date:        14 November 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @defgroup Not Vector bitwise NOT
+
+  Compute the logical bitwise NOT.
+
+  There are separate functions for uint32_t, uint16_t, and uint8_t data types.
+ */
+
+/**
+  @addtogroup Not
+  @{
+ */
+
+/**
+  @brief         Compute the logical bitwise NOT of a fixed-point vector.
+  @param[in]     pSrc       points to input vector 
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+void arm_not_u16(
+    const uint16_t * pSrc,
+          uint16_t * pDst,
+          uint32_t blockSize)
+{
+    uint32_t blkCnt;      /* Loop counter */
+
+#if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
+    q15x8_t vecSrc;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3;
+
+    while (blkCnt > 0U)
+    {
+        vecSrc = vld1q(pSrc);
+
+        vst1q(pDst, vmvnq_u16(vecSrc) );
+
+        pSrc += 8;
+        pDst += 8;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 7;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp16q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vstrhq_p(pDst, vmvnq_u16(vecSrc), p0);
+    }
+#else
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    uint16x8_t inV;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3U;
+
+    while (blkCnt > 0U)
+    {
+        inV = vld1q_u16(pSrc);
+
+        vst1q_u16(pDst, vmvnq_u16(inV) );
+
+        pSrc += 8;
+        pDst += 8;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 7;
+#else
+    /* Initialize blkCnt with number of samples */
+    blkCnt = blockSize;
+#endif
+
+    while (blkCnt > 0U)
+    {
+        *pDst++ = ~(*pSrc++);
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+#endif /* if defined(ARM_MATH_MVEI) */
+}
+
+/**
+  @} end of Not group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_not_u32.c

@@ -0,0 +1,122 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_not_u32.c
+ * Description:  uint32_t bitwise NOT
+ *
+ * $Date:        14 November 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup Not
+  @{
+ */
+
+/**
+  @brief         Compute the logical bitwise NOT of a fixed-point vector.
+  @param[in]     pSrc       points to input vector 
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+void arm_not_u32(
+    const uint32_t * pSrc,
+          uint32_t * pDst,
+          uint32_t blockSize)
+{
+    uint32_t blkCnt;      /* Loop counter */
+
+#if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
+    q31x4_t vecSrc;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 2;
+
+    while (blkCnt > 0U)
+    {
+        vecSrc = vld1q(pSrc);
+
+        vst1q(pDst, vmvnq_u32(vecSrc) );
+
+        pSrc += 4;
+        pDst += 4;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 3;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vstrwq_p(pDst, vmvnq_u32(vecSrc), p0);
+    }
+#else
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    uint32x4_t inV;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        inV = vld1q_u32(pSrc);
+
+        vst1q_u32(pDst, vmvnq_u32(inV) );
+
+        pSrc += 4;
+        pDst += 4;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 3;
+#else
+    /* Initialize blkCnt with number of samples */
+    blkCnt = blockSize;
+#endif
+
+    while (blkCnt > 0U)
+    {
+        *pDst++ = ~(*pSrc++);
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+#endif /* if defined(ARM_MATH_MVEI) */
+}
+
+/**
+  @} end of Not group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_not_u8.c

@@ -0,0 +1,122 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_not_u8.c
+ * Description:  uint8_t bitwise NOT
+ *
+ * $Date:        14 November 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup Not
+  @{
+ */
+
+/**
+  @brief         Compute the logical bitwise NOT of a fixed-point vector.
+  @param[in]     pSrc       points to input vector 
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+void arm_not_u8(
+    const uint8_t * pSrc,
+          uint8_t * pDst,
+          uint32_t blockSize)
+{
+    uint32_t blkCnt;      /* Loop counter */
+
+#if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
+    q7x16_t vecSrc;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4;
+
+    while (blkCnt > 0U)
+    {
+        vecSrc = vld1q(pSrc);
+
+        vst1q(pDst, vmvnq_u8(vecSrc) );
+
+        pSrc += 16;
+        pDst += 16;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0xF;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp8q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vstrbq_p(pDst, vmvnq_u8(vecSrc), p0);
+    }
+#else
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    uint8x16_t inV;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4U;
+
+    while (blkCnt > 0U)
+    {
+        inV = vld1q_u8(pSrc);
+
+        vst1q_u8(pDst, vmvnq_u8(inV) );
+
+        pSrc += 16;
+        pDst += 16;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0xF;
+#else
+    /* Initialize blkCnt with number of samples */
+    blkCnt = blockSize;
+#endif
+
+    while (blkCnt > 0U)
+    {
+        *pDst++ = ~(*pSrc++);
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+#endif /* if defined(ARM_MATH_MVEI) */
+}
+
+/**
+  @} end of Not group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_offset_f32.c

@@ -0,0 +1,196 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_offset_f32.c
+ * Description:  Floating-point vector offset
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @defgroup BasicOffset Vector Offset
+
+  Adds a constant offset to each element of a vector.
+
+  <pre>
+      pDst[n] = pSrc[n] + offset,   0 <= n < blockSize.
+  </pre>
+
+  The functions support in-place computation allowing the source and
+  destination pointers to reference the same memory buffer.
+  There are separate functions for floating-point, Q7, Q15, and Q31 data types.
+ */
+
+/**
+  @addtogroup BasicOffset
+  @{
+ */
+
+/**
+  @brief         Adds a constant offset to a floating-point vector.
+  @param[in]     pSrc       points to the input vector
+  @param[in]     offset     is the offset to be added
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#include "arm_helium_utils.h"
+
+void arm_offset_f32(
+  const float32_t * pSrc,
+        float32_t offset,
+        float32_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+    f32x4_t vec1;
+    f32x4_t res;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+    while (blkCnt > 0U)
+    {
+        /* C = A + offset */
+ 
+        /* Add offset and then store the results in the destination buffer. */
+        vec1 = vld1q(pSrc);
+        res = vaddq(vec1,offset);
+        vst1q(pDst, res);
+
+        /* Increment pointers */
+        pSrc += 4;
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vec1 = vld1q((float32_t const *) pSrc);
+        vstrwq_p(pDst, vaddq(vec1, offset), p0);
+    }
+
+
+}
+
+#else
+void arm_offset_f32(
+  const float32_t * pSrc,
+        float32_t offset,
+        float32_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined(ARM_MATH_NEON_EXPERIMENTAL) && !defined(ARM_MATH_AUTOVECTORIZE)
+    f32x4_t vec1;
+    f32x4_t res;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        /* C = A + offset */
+ 
+        /* Add offset and then store the results in the destination buffer. */
+        vec1 = vld1q_f32(pSrc);
+        res = vaddq_f32(vec1,vdupq_n_f32(offset));
+        vst1q_f32(pDst, res);
+
+        /* Increment pointers */
+        pSrc += 4;
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+#else
+#if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + offset */
+
+    /* Add offset and store result in destination buffer. */
+    *pDst++ = (*pSrc++) + offset;
+
+    *pDst++ = (*pSrc++) + offset;
+
+    *pDst++ = (*pSrc++) + offset;
+
+    *pDst++ = (*pSrc++) + offset;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined(ARM_MATH_NEON_EXPERIMENTAL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + offset */
+
+    /* Add offset and store result in destination buffer. */
+    *pDst++ = (*pSrc++) + offset;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+/**
+  @} end of BasicOffset group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_offset_q15.c

@@ -0,0 +1,168 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_offset_q15.c
+ * Description:  Q15 vector offset
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicOffset
+  @{
+ */
+
+/**
+  @brief         Adds a constant offset to a Q15 vector.
+  @param[in]     pSrc       points to the input vector
+  @param[in]     offset     is the offset to be added
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q15 range [0x8000 0x7FFF] are saturated.
+ */
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_offset_q15(
+    const q15_t * pSrc,
+    q15_t   offset,
+    q15_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q15x8_t vecSrc;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A + offset
+         * Add offset and then store the result in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vst1q(pDst, vqaddq(vecSrc, offset));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 8;
+        pDst += 8;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 7;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp16q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vstrhq_p(pDst, vqaddq(vecSrc, offset), p0);
+    }
+}
+
+
+#else
+void arm_offset_q15(
+  const q15_t * pSrc,
+        q15_t offset,
+        q15_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+#if defined (ARM_MATH_DSP)
+  q31_t offset_packed;                           /* Offset packed to 32 bit */
+
+  /* Offset is packed to 32 bit in order to use SIMD32 for addition */
+  offset_packed = __PKHBT(offset, offset, 16);
+#endif
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + offset */
+
+#if defined (ARM_MATH_DSP)
+    /* Add offset and store result in destination buffer (2 samples at a time). */
+    write_q15x2_ia (&pDst, __QADD16(read_q15x2_ia ((q15_t **) &pSrc), offset_packed));
+    write_q15x2_ia (&pDst, __QADD16(read_q15x2_ia ((q15_t **) &pSrc), offset_packed));
+#else
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrc++ + offset), 16);
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrc++ + offset), 16);
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrc++ + offset), 16);
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrc++ + offset), 16);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + offset */
+
+    /* Add offset and store result in destination buffer. */
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (q15_t) __QADD16(*pSrc++, offset);
+#else
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrc++ + offset), 16);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicOffset group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_offset_q31.c

@@ -0,0 +1,175 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_offset_q31.c
+ * Description:  Q31 vector offset
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicOffset
+  @{
+ */
+
+/**
+  @brief         Adds a constant offset to a Q31 vector.
+  @param[in]     pSrc       points to the input vector
+  @param[in]     offset     is the offset to be added
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q31 range [0x80000000 0x7FFFFFFF] are saturated.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_offset_q31(
+    const q31_t * pSrc,
+    q31_t   offset,
+    q31_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q31x4_t vecSrc;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A + offset
+         * Add offset and then store the result in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vst1q(pDst, vqaddq(vecSrc, offset));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 4;
+        pDst += 4;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 3;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vstrwq_p(pDst, vqaddq(vecSrc, offset), p0);
+    }
+}
+
+#else
+void arm_offset_q31(
+  const q31_t * pSrc,
+        q31_t offset,
+        q31_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + offset */
+
+    /* Add offset and store result in destination buffer. */
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QADD(*pSrc++, offset);
+#else
+    *pDst++ = (q31_t) clip_q63_to_q31((q63_t) * pSrc++ + offset);
+#endif
+
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QADD(*pSrc++, offset);
+#else
+    *pDst++ = (q31_t) clip_q63_to_q31((q63_t) * pSrc++ + offset);
+#endif
+
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QADD(*pSrc++, offset);
+#else
+    *pDst++ = (q31_t) clip_q63_to_q31((q63_t) * pSrc++ + offset);
+#endif
+
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QADD(*pSrc++, offset);
+#else
+    *pDst++ = (q31_t) clip_q63_to_q31((q63_t) * pSrc++ + offset);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + offset */
+
+    /* Add offset and store result in destination buffer. */
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QADD(*pSrc++, offset);
+#else
+    *pDst++ = (q31_t) clip_q63_to_q31((q63_t) * pSrc++ + offset);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicOffset group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_offset_q7.c

@@ -0,0 +1,162 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_offset_q7.c
+ * Description:  Q7 vector offset
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicOffset
+  @{
+ */
+
+/**
+  @brief         Adds a constant offset to a Q7 vector.
+  @param[in]     pSrc       points to the input vector
+  @param[in]     offset     is the offset to be added
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q7 range [0x80 0x7F] are saturated.
+ */
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_offset_q7(
+    const q7_t * pSrc,
+    q7_t   offset,
+    q7_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q7x16_t vecSrc;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A + offset
+         * Add offset and then store the result in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vst1q(pDst, vqaddq(vecSrc, offset));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 16;
+        pDst += 16;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 0xF;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp8q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vstrbq_p(pDst, vqaddq(vecSrc, offset), p0);
+    }
+}
+
+#else
+void arm_offset_q7(
+  const q7_t * pSrc,
+        q7_t offset,
+        q7_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+#if defined (ARM_MATH_DSP)
+  q31_t offset_packed;                           /* Offset packed to 32 bit */
+
+  /* Offset is packed to 32 bit in order to use SIMD32 for addition */
+  offset_packed = __PACKq7(offset, offset, offset, offset);
+#endif
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + offset */
+
+#if defined (ARM_MATH_DSP)
+    /* Add offset and store result in destination buffer (4 samples at a time). */
+    write_q7x4_ia (&pDst, __QADD8(read_q7x4_ia ((q7_t **) &pSrc), offset_packed));
+#else
+    *pDst++ = (q7_t) __SSAT(*pSrc++ + offset, 8);
+    *pDst++ = (q7_t) __SSAT(*pSrc++ + offset, 8);
+    *pDst++ = (q7_t) __SSAT(*pSrc++ + offset, 8);
+    *pDst++ = (q7_t) __SSAT(*pSrc++ + offset, 8);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A + offset */
+
+    /* Add offset and store result in destination buffer. */
+    *pDst++ = (q7_t) __SSAT((q15_t) *pSrc++ + offset, 8);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicOffset group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_or_u16.c

@@ -0,0 +1,137 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_or_u16.c
+ * Description:  uint16_t bitwise inclusive OR
+ *
+ * $Date:        14 November 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @defgroup Or Vector bitwise inclusive OR
+
+  Compute the logical bitwise OR.
+
+  There are separate functions for uint32_t, uint16_t, and uint8_t data types.
+ */
+
+/**
+  @addtogroup Or
+  @{
+ */
+
+/**
+  @brief         Compute the logical bitwise OR of two fixed-point vectors.
+  @param[in]     pSrcA      points to input vector A
+  @param[in]     pSrcB      points to input vector B
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+void arm_or_u16(
+    const uint16_t * pSrcA,
+    const uint16_t * pSrcB,
+          uint16_t * pDst,
+          uint32_t blockSize)
+{
+    uint32_t blkCnt;      /* Loop counter */
+
+#if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
+    q15x8_t vecSrcA, vecSrcB;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3;
+
+    while (blkCnt > 0U)
+    {
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+
+        vst1q(pDst, vorrq_u16(vecSrcA, vecSrcB) );
+
+        pSrcA += 8;
+        pSrcB += 8;
+        pDst  += 8;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 7;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp16q(blkCnt);
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+        vstrhq_p(pDst, vorrq_u16(vecSrcA, vecSrcB), p0);
+    }
+#else
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    uint16x8_t vecA, vecB;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3U;
+
+    while (blkCnt > 0U)
+    {
+        vecA = vld1q_u16(pSrcA);
+        vecB = vld1q_u16(pSrcB);
+
+        vst1q_u16(pDst, vorrq_u16(vecA, vecB) );
+
+        pSrcA += 8;
+        pSrcB += 8;
+        pDst  += 8;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 7;
+#else
+    /* Initialize blkCnt with number of samples */
+    blkCnt = blockSize;
+#endif
+
+    while (blkCnt > 0U)
+    {
+        *pDst++ = (*pSrcA++)|(*pSrcB++);
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+#endif /* if defined(ARM_MATH_MVEI) */
+}
+
+/**
+  @} end of Or group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_or_u32.c

@@ -0,0 +1,128 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_or_u32.c
+ * Description:  uint32_t bitwise inclusive OR
+ *
+ * $Date:        14 November 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup Or
+  @{
+ */
+
+/**
+  @brief         Compute the logical bitwise OR of two fixed-point vectors.
+  @param[in]     pSrcA      points to input vector A
+  @param[in]     pSrcB      points to input vector B
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+void arm_or_u32(
+    const uint32_t * pSrcA,
+    const uint32_t * pSrcB,
+          uint32_t * pDst,
+          uint32_t blockSize)
+{
+    uint32_t blkCnt;      /* Loop counter */
+
+#if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
+    q31x4_t vecSrcA, vecSrcB;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2;
+
+    while (blkCnt > 0U)
+    {
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+
+        vst1q(pDst, vorrq_u32(vecSrcA, vecSrcB) );
+
+        pSrcA += 4;
+        pSrcB += 4;
+        pDst  += 4;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 3;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+        vstrwq_p(pDst, vorrq_u32(vecSrcA, vecSrcB), p0);
+    }
+#else
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    uint32x4_t vecA, vecB;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        vecA = vld1q_u32(pSrcA);
+        vecB = vld1q_u32(pSrcB);
+
+        vst1q_u32(pDst, vorrq_u32(vecA, vecB) );
+
+        pSrcA += 4;
+        pSrcB += 4;
+        pDst  += 4;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 3;
+#else
+    /* Initialize blkCnt with number of samples */
+    blkCnt = blockSize;
+#endif
+
+    while (blkCnt > 0U)
+    {
+        *pDst++ = (*pSrcA++)|(*pSrcB++);
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+#endif /* if defined(ARM_MATH_MVEI) */
+}
+/**
+  @} end of Or group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_or_u8.c

@@ -0,0 +1,128 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_or_u8.c
+ * Description:  uint8_t bitwise inclusive OR
+ *
+ * $Date:        14 November 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup Or
+  @{
+ */
+
+/**
+  @brief         Compute the logical bitwise OR of two fixed-point vectors.
+  @param[in]     pSrcA      points to input vector A
+  @param[in]     pSrcB      points to input vector B
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+void arm_or_u8(
+    const uint8_t * pSrcA,
+    const uint8_t * pSrcB,
+          uint8_t * pDst,
+          uint32_t blockSize)
+{
+    uint32_t blkCnt;      /* Loop counter */
+
+#if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
+    q7x16_t vecSrcA, vecSrcB;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4;
+
+    while (blkCnt > 0U)
+    {
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+
+        vst1q(pDst, vorrq_u8(vecSrcA, vecSrcB) );
+
+        pSrcA += 16;
+        pSrcB += 16;
+        pDst  += 16;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0xF;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp8q(blkCnt);
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+        vstrbq_p(pDst, vorrq_u8(vecSrcA, vecSrcB), p0);
+    }
+#else
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    uint8x16_t vecA, vecB;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4U;
+
+    while (blkCnt > 0U)
+    {
+        vecA = vld1q_u8(pSrcA);
+        vecB = vld1q_u8(pSrcB);
+
+        vst1q_u8(pDst, vorrq_u8(vecA, vecB) );
+
+        pSrcA += 16;
+        pSrcB += 16;
+        pDst  += 16;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0xF;
+#else
+    /* Initialize blkCnt with number of samples */
+    blkCnt = blockSize;
+#endif
+
+    while (blkCnt > 0U)
+    {
+        *pDst++ = (*pSrcA++)|(*pSrcB++);
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+#endif /* if defined(ARM_MATH_MVEI) */
+}
+/**
+  @} end of Or group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_scale_f32.c

@@ -0,0 +1,216 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_scale_f32.c
+ * Description:  Multiplies a floating-point vector by a scalar
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @defgroup BasicScale Vector Scale
+
+  Multiply a vector by a scalar value.  For floating-point data, the algorithm used is:
+
+  <pre>
+      pDst[n] = pSrc[n] * scale,   0 <= n < blockSize.
+  </pre>
+
+  In the fixed-point Q7, Q15, and Q31 functions, <code>scale</code> is represented by
+  a fractional multiplication <code>scaleFract</code> and an arithmetic shift <code>shift</code>.
+  The shift allows the gain of the scaling operation to exceed 1.0.
+  The algorithm used with fixed-point data is:
+
+  <pre>
+      pDst[n] = (pSrc[n] * scaleFract) << shift,   0 <= n < blockSize.
+  </pre>
+
+  The overall scale factor applied to the fixed-point data is
+  <pre>
+      scale = scaleFract * 2^shift.
+  </pre>
+
+  The functions support in-place computation allowing the source and destination
+  pointers to reference the same memory buffer.
+ */
+
+/**
+  @addtogroup BasicScale
+  @{
+ */
+
+/**
+  @brief         Multiplies a floating-point vector by a scalar.
+  @param[in]     pSrc       points to the input vector
+  @param[in]     scale      scale factor to be applied
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#include "arm_helium_utils.h"
+
+void arm_scale_f32(
+  const float32_t * pSrc,
+        float32_t scale,
+        float32_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+    f32x4_t vec1;
+    f32x4_t res;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        /* C = A + offset */
+ 
+        /* Add offset and then store the results in the destination buffer. */
+        vec1 = vld1q(pSrc);
+        res = vmulq(vec1,scale);
+        vst1q(pDst, res);
+
+        /* Increment pointers */
+        pSrc += 4;
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vec1 = vld1q((float32_t const *) pSrc);
+        vstrwq_p(pDst, vmulq(vec1, scale), p0);
+    }
+
+
+}
+
+#else
+void arm_scale_f32(
+  const float32_t *pSrc,
+        float32_t scale,
+        float32_t *pDst,
+        uint32_t blockSize)
+{
+  uint32_t blkCnt;                               /* Loop counter */
+#if defined(ARM_MATH_NEON_EXPERIMENTAL)
+    f32x4_t vec1;
+    f32x4_t res;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        /* C = A * scale */
+
+    	/* Scale the input and then store the results in the destination buffer. */
+        vec1 = vld1q_f32(pSrc);
+        res = vmulq_f32(vec1, vdupq_n_f32(scale));
+        vst1q_f32(pDst, res);
+
+        /* Increment pointers */
+        pSrc += 4; 
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+#else
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    float32_t in1, in2, in3, in4;
+
+    /* C = A * scale */
+
+    /* Scale input and store result in destination buffer. */
+    in1 = (*pSrc++) * scale;
+
+    in2 = (*pSrc++) * scale;
+
+    in3 = (*pSrc++) * scale;
+
+    in4 = (*pSrc++) * scale;
+
+    *pDst++ = in1;
+    *pDst++ = in2;
+    *pDst++ = in3;
+    *pDst++ = in4;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined(ARM_MATH_NEON_EXPERIMENTAL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A * scale */
+
+    /* Scale input and store result in destination buffer. */
+    *pDst++ = (*pSrc++) * scale;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+/**
+  @} end of BasicScale group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_scale_q15.c

@@ -0,0 +1,201 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_scale_q15.c
+ * Description:  Multiplies a Q15 vector by a scalar
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicScale
+  @{
+ */
+
+/**
+  @brief         Multiplies a Q15 vector by a scalar.
+  @param[in]     pSrc       points to the input vector
+  @param[in]     scaleFract fractional portion of the scale value
+  @param[in]     shift      number of bits to shift the result by
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The input data <code>*pSrc</code> and <code>scaleFract</code> are in 1.15 format.
+                   These are multiplied to yield a 2.30 intermediate result and this is shifted with saturation to 1.15 format.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_scale_q15(
+    const q15_t * pSrc,
+    q15_t   scaleFract,
+    int8_t  shift,
+    q15_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q15x8_t vecSrc;
+    q15x8_t vecDst;
+
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3;
+
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A * scale
+         * Scale the input and then store the result in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vecDst = vmulhq(vecSrc, vdupq_n_s16(scaleFract));
+        vecDst = vqshlq_r(vecDst, shift + 1);
+        vst1q(pDst, vecDst);
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 8;
+        pDst += 8;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 7;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp16q(blkCnt);;
+        vecSrc = vld1q(pSrc);
+        vecDst = vmulhq(vecSrc, vdupq_n_s16(scaleFract));
+        vecDst = vqshlq_r(vecDst, shift + 1);
+        vstrhq_p(pDst, vecDst, p0);
+    }
+
+}
+
+
+#else
+void arm_scale_q15(
+  const q15_t *pSrc,
+        q15_t scaleFract,
+        int8_t shift,
+        q15_t *pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        int8_t kShift = 15 - shift;                    /* Shift to apply after scaling */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+#if defined (ARM_MATH_DSP)
+  q31_t inA1, inA2;
+  q31_t out1, out2, out3, out4;                  /* Temporary output variables */
+  q15_t in1, in2, in3, in4;                      /* Temporary input variables */
+#endif
+#endif
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A * scale */
+
+#if defined (ARM_MATH_DSP)
+    /* read 2 times 2 samples at a time from source */
+    inA1 = read_q15x2_ia ((q15_t **) &pSrc);
+    inA2 = read_q15x2_ia ((q15_t **) &pSrc);
+
+    /* Scale inputs and store result in temporary variables
+     * in single cycle by packing the outputs */
+    out1 = (q31_t) ((q15_t) (inA1 >> 16) * scaleFract);
+    out2 = (q31_t) ((q15_t) (inA1      ) * scaleFract);
+    out3 = (q31_t) ((q15_t) (inA2 >> 16) * scaleFract);
+    out4 = (q31_t) ((q15_t) (inA2      ) * scaleFract);
+
+    /* apply shifting */
+    out1 = out1 >> kShift;
+    out2 = out2 >> kShift;
+    out3 = out3 >> kShift;
+    out4 = out4 >> kShift;
+
+    /* saturate the output */
+    in1 = (q15_t) (__SSAT(out1, 16));
+    in2 = (q15_t) (__SSAT(out2, 16));
+    in3 = (q15_t) (__SSAT(out3, 16));
+    in4 = (q15_t) (__SSAT(out4, 16));
+
+    /* store result to destination */
+    write_q15x2_ia (&pDst, __PKHBT(in2, in1, 16));
+    write_q15x2_ia (&pDst, __PKHBT(in4, in3, 16));
+#else
+    *pDst++ = (q15_t) (__SSAT(((q31_t) *pSrc++ * scaleFract) >> kShift, 16));
+    *pDst++ = (q15_t) (__SSAT(((q31_t) *pSrc++ * scaleFract) >> kShift, 16));
+    *pDst++ = (q15_t) (__SSAT(((q31_t) *pSrc++ * scaleFract) >> kShift, 16));
+    *pDst++ = (q15_t) (__SSAT(((q31_t) *pSrc++ * scaleFract) >> kShift, 16));
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A * scale */
+
+    /* Scale input and store result in destination buffer. */
+    *pDst++ = (q15_t) (__SSAT(((q31_t) *pSrc++ * scaleFract) >> kShift, 16));
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicScale group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_scale_q31.c

@@ -0,0 +1,244 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_scale_q31.c
+ * Description:  Multiplies a Q31 vector by a scalar
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicScale
+  @{
+ */
+
+/**
+  @brief         Multiplies a Q31 vector by a scalar.
+  @param[in]     pSrc       points to the input vector
+  @param[in]     scaleFract fractional portion of the scale value
+  @param[in]     shift      number of bits to shift the result by
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The input data <code>*pSrc</code> and <code>scaleFract</code> are in 1.31 format.
+                   These are multiplied to yield a 2.62 intermediate result and this is shifted with saturation to 1.31 format.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_scale_q31(
+    const q31_t * pSrc,
+    q31_t   scaleFract,
+    int8_t  shift,
+    q31_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q31x4_t vecSrc;
+    q31x4_t vecDst;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A * scale
+         * Scale the input and then store the result in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vecDst = vmulhq(vecSrc, vdupq_n_s32(scaleFract));
+        vecDst = vqshlq_r(vecDst, shift + 1);
+        vst1q(pDst, vecDst);
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 4;
+        pDst += 4;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 3;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vecDst = vmulhq(vecSrc, vdupq_n_s32(scaleFract));
+        vecDst = vqshlq_r(vecDst, shift + 1);
+        vstrwq_p(pDst, vecDst, p0);
+    }
+}
+
+#else
+void arm_scale_q31(
+  const q31_t *pSrc,
+        q31_t scaleFract,
+        int8_t shift,
+        q31_t *pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q31_t in, out;                                 /* Temporary variables */
+        int8_t kShift = shift + 1;                     /* Shift to apply after scaling */
+        int8_t sign = (kShift & 0x80);
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  if (sign == 0U)
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A * scale */
+
+      /* Scale input and store result in destination buffer. */
+      in = *pSrc++;                                /* read input from source */
+      in = ((q63_t) in * scaleFract) >> 32;        /* multiply input with scaler value */
+      out = in << kShift;                          /* apply shifting */
+      if (in != (out >> kShift))                   /* saturate the result */
+        out = 0x7FFFFFFF ^ (in >> 31);
+      *pDst++ = out;                               /* Store result destination */
+
+      in = *pSrc++;
+      in = ((q63_t) in * scaleFract) >> 32;
+      out = in << kShift;
+      if (in != (out >> kShift))
+        out = 0x7FFFFFFF ^ (in >> 31);
+      *pDst++ = out;
+
+      in = *pSrc++;
+      in = ((q63_t) in * scaleFract) >> 32;
+      out = in << kShift;
+      if (in != (out >> kShift))
+        out = 0x7FFFFFFF ^ (in >> 31);
+      *pDst++ = out;
+
+      in = *pSrc++;
+      in = ((q63_t) in * scaleFract) >> 32;
+      out = in << kShift;
+      if (in != (out >> kShift))
+        out = 0x7FFFFFFF ^ (in >> 31);
+      *pDst++ = out;
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+  else
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A * scale */
+
+      /* Scale input and store result in destination buffer. */
+      in = *pSrc++;                                /* read four inputs from source */
+      in = ((q63_t) in * scaleFract) >> 32;        /* multiply input with scaler value */
+      out = in >> -kShift;                         /* apply shifting */
+      *pDst++ = out;                               /* Store result destination */
+
+      in = *pSrc++;
+      in = ((q63_t) in * scaleFract) >> 32;
+      out = in >> -kShift;
+      *pDst++ = out;
+
+      in = *pSrc++;
+      in = ((q63_t) in * scaleFract) >> 32;
+      out = in >> -kShift;
+      *pDst++ = out;
+
+      in = *pSrc++;
+      in = ((q63_t) in * scaleFract) >> 32;
+      out = in >> -kShift;
+      *pDst++ = out;
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  if (sign == 0U)
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A * scale */
+
+      /* Scale input and store result in destination buffer. */
+      in = *pSrc++;
+      in = ((q63_t) in * scaleFract) >> 32;
+      out = in << kShift;
+      if (in != (out >> kShift))
+          out = 0x7FFFFFFF ^ (in >> 31);
+      *pDst++ = out;
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+  else
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A * scale */
+
+      /* Scale input and store result in destination buffer. */
+      in = *pSrc++;
+      in = ((q63_t) in * scaleFract) >> 32;
+      out = in >> -kShift;
+      *pDst++ = out;
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicScale group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_scale_q7.c

@@ -0,0 +1,186 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_scale_q7.c
+ * Description:  Multiplies a Q7 vector by a scalar
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicScale
+  @{
+ */
+
+/**
+  @brief         Multiplies a Q7 vector by a scalar.
+  @param[in]     pSrc       points to the input vector
+  @param[in]     scaleFract fractional portion of the scale value
+  @param[in]     shift      number of bits to shift the result by
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The input data <code>*pSrc</code> and <code>scaleFract</code> are in 1.7 format.
+                   These are multiplied to yield a 2.14 intermediate result and this is shifted with saturation to 1.7 format.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+
+void arm_scale_q7(
+    const q7_t * pSrc,
+    q7_t   scaleFract,
+    int8_t  shift,
+    q7_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q7x16_t vecSrc;
+    q7x16_t vecDst;
+
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4;
+
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A * scale
+         * Scale the input and then store the result in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vecDst = vmulhq(vecSrc, vdupq_n_s8(scaleFract));
+        vecDst = vqshlq_r(vecDst, shift + 1);
+        vst1q(pDst, vecDst);
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 16;
+        pDst += 16;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 0xF;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp8q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vecDst = vmulhq(vecSrc, vdupq_n_s8(scaleFract));
+        vecDst = vqshlq_r(vecDst, shift + 1);
+        vstrbq_p(pDst, vecDst, p0);
+    }
+
+}
+
+#else
+void arm_scale_q7(
+  const q7_t * pSrc,
+        q7_t scaleFract,
+        int8_t shift,
+        q7_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        int8_t kShift = 7 - shift;                     /* Shift to apply after scaling */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+#if defined (ARM_MATH_DSP)
+  q7_t in1,  in2,  in3,  in4;                    /* Temporary input variables */
+  q7_t out1, out2, out3, out4;                   /* Temporary output variables */
+#endif
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A * scale */
+
+#if defined (ARM_MATH_DSP)
+    /* Reading 4 inputs from memory */
+    in1 = *pSrc++;
+    in2 = *pSrc++;
+    in3 = *pSrc++;
+    in4 = *pSrc++;
+
+    /* Scale inputs and store result in the temporary variable. */
+    out1 = (q7_t) (__SSAT(((in1) * scaleFract) >> kShift, 8));
+    out2 = (q7_t) (__SSAT(((in2) * scaleFract) >> kShift, 8));
+    out3 = (q7_t) (__SSAT(((in3) * scaleFract) >> kShift, 8));
+    out4 = (q7_t) (__SSAT(((in4) * scaleFract) >> kShift, 8));
+
+    /* Pack and store result in destination buffer (in single write) */
+    write_q7x4_ia (&pDst, __PACKq7(out1, out2, out3, out4));
+#else
+    *pDst++ = (q7_t) (__SSAT((((q15_t) *pSrc++ * scaleFract) >> kShift), 8));
+    *pDst++ = (q7_t) (__SSAT((((q15_t) *pSrc++ * scaleFract) >> kShift), 8));
+    *pDst++ = (q7_t) (__SSAT((((q15_t) *pSrc++ * scaleFract) >> kShift), 8));
+    *pDst++ = (q7_t) (__SSAT((((q15_t) *pSrc++ * scaleFract) >> kShift), 8));
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A * scale */
+
+    /* Scale input and store result in destination buffer. */
+    *pDst++ = (q7_t) (__SSAT((((q15_t) *pSrc++ * scaleFract) >> kShift), 8));
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicScale group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_shift_q15.c

@@ -0,0 +1,251 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_shift_q15.c
+ * Description:  Shifts the elements of a Q15 vector by a specified number of bits
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicShift
+  @{
+ */
+
+/**
+  @brief         Shifts the elements of a Q15 vector a specified number of bits
+  @param[in]     pSrc       points to the input vector
+  @param[in]     shiftBits  number of bits to shift.  A positive value shifts left; a negative value shifts right.
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q15 range [0x8000 0x7FFF] are saturated.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_shift_q15(
+    const q15_t * pSrc,
+    int8_t shiftBits,
+    q15_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q15x8_t vecSrc;
+    q15x8_t vecDst;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A (>> or <<) shiftBits
+         * Shift the input and then store the result in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vecDst = vqshlq_r(vecSrc, shiftBits);
+        vst1q(pDst, vecDst);
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 8;
+        pDst += 8;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 7;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp16q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vecDst = vqshlq_r(vecSrc, shiftBits);
+        vstrhq_p(pDst, vecDst, p0);
+    }
+}
+
+#else
+void arm_shift_q15(
+  const q15_t * pSrc,
+        int8_t shiftBits,
+        q15_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        uint8_t sign = (shiftBits & 0x80);             /* Sign of shiftBits */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+#if defined (ARM_MATH_DSP)
+  q15_t in1, in2;                                /* Temporary input variables */
+#endif
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  /* If the shift value is positive then do right shift else left shift */
+  if (sign == 0U)
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A << shiftBits */
+
+#if defined (ARM_MATH_DSP)
+      /* read 2 samples from source */
+      in1 = *pSrc++;
+      in2 = *pSrc++;
+
+      /* Shift the inputs and then store the results in the destination buffer. */
+#ifndef ARM_MATH_BIG_ENDIAN
+      write_q15x2_ia (&pDst, __PKHBT(__SSAT((in1 << shiftBits), 16),
+                                     __SSAT((in2 << shiftBits), 16), 16));
+#else
+      write_q15x2_ia (&pDst, __PKHBT(__SSAT((in2 << shiftBits), 16),
+                                      __SSAT((in1 << shiftBits), 16), 16));
+#endif /* #ifndef ARM_MATH_BIG_ENDIAN */
+
+      /* read 2 samples from source */
+      in1 = *pSrc++;
+      in2 = *pSrc++;
+
+#ifndef ARM_MATH_BIG_ENDIAN
+      write_q15x2_ia (&pDst, __PKHBT(__SSAT((in1 << shiftBits), 16),
+                                     __SSAT((in2 << shiftBits), 16), 16));
+#else
+      write_q15x2_ia (&pDst, __PKHBT(__SSAT((in2 << shiftBits), 16),
+                                     __SSAT((in1 << shiftBits), 16), 16));
+#endif /* #ifndef ARM_MATH_BIG_ENDIAN */
+
+#else
+      *pDst++ = __SSAT(((q31_t) *pSrc++ << shiftBits), 16);
+      *pDst++ = __SSAT(((q31_t) *pSrc++ << shiftBits), 16);
+      *pDst++ = __SSAT(((q31_t) *pSrc++ << shiftBits), 16);
+      *pDst++ = __SSAT(((q31_t) *pSrc++ << shiftBits), 16);
+#endif
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+  else
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A >> shiftBits */
+
+#if defined (ARM_MATH_DSP)
+      /* read 2 samples from source */
+      in1 = *pSrc++;
+      in2 = *pSrc++;
+
+      /* Shift the inputs and then store the results in the destination buffer. */
+#ifndef ARM_MATH_BIG_ENDIAN
+      write_q15x2_ia (&pDst, __PKHBT((in1 >> -shiftBits),
+                                     (in2 >> -shiftBits), 16));
+#else
+      write_q15x2_ia (&pDst, __PKHBT((in2 >> -shiftBits),
+                                     (in1 >> -shiftBits), 16));
+#endif /* #ifndef ARM_MATH_BIG_ENDIAN */
+
+      /* read 2 samples from source */
+      in1 = *pSrc++;
+      in2 = *pSrc++;
+
+#ifndef ARM_MATH_BIG_ENDIAN
+      write_q15x2_ia (&pDst, __PKHBT((in1 >> -shiftBits),
+                                     (in2 >> -shiftBits), 16));
+#else
+      write_q15x2_ia (&pDst, __PKHBT((in2 >> -shiftBits),
+                                     (in1 >> -shiftBits), 16));
+#endif /* #ifndef ARM_MATH_BIG_ENDIAN */
+
+#else
+      *pDst++ = (*pSrc++ >> -shiftBits);
+      *pDst++ = (*pSrc++ >> -shiftBits);
+      *pDst++ = (*pSrc++ >> -shiftBits);
+      *pDst++ = (*pSrc++ >> -shiftBits);
+#endif
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  /* If the shift value is positive then do right shift else left shift */
+  if (sign == 0U)
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A << shiftBits */
+
+      /* Shift input and store result in destination buffer. */
+      *pDst++ = __SSAT(((q31_t) *pSrc++ << shiftBits), 16);
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+  else
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A >> shiftBits */
+
+      /* Shift input and store result in destination buffer. */
+      *pDst++ = (*pSrc++ >> -shiftBits);
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicShift group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_shift_q31.c

@@ -0,0 +1,232 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_shift_q31.c
+ * Description:  Shifts the elements of a Q31 vector by a specified number of bits
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+/**
+  @defgroup BasicShift Vector Shift
+
+  Shifts the elements of a fixed-point vector by a specified number of bits.
+  There are separate functions for Q7, Q15, and Q31 data types.
+  The underlying algorithm used is:
+
+  <pre>
+      pDst[n] = pSrc[n] << shift,   0 <= n < blockSize.
+  </pre>
+
+  If <code>shift</code> is positive then the elements of the vector are shifted to the left.
+  If <code>shift</code> is negative then the elements of the vector are shifted to the right.
+
+  The functions support in-place computation allowing the source and destination
+  pointers to reference the same memory buffer.
+ */
+
+/**
+  @addtogroup BasicShift
+  @{
+ */
+
+/**
+  @brief         Shifts the elements of a Q31 vector a specified number of bits.
+  @param[in]     pSrc       points to the input vector
+  @param[in]     shiftBits  number of bits to shift.  A positive value shifts left; a negative value shifts right.
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in the vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q31 range [0x80000000 0x7FFFFFFF] are saturated.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_shift_q31(
+    const q31_t * pSrc,
+    int8_t shiftBits,
+    q31_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q31x4_t vecSrc;
+    q31x4_t vecDst;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A (>> or <<) shiftBits
+         * Shift the input and then store the result in the destination buffer.
+         */
+        vecSrc = vld1q((q31_t const *) pSrc);
+        vecDst = vqshlq_r(vecSrc, shiftBits);
+        vst1q(pDst, vecDst);
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 4;
+        pDst += 4;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 3;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecSrc = vld1q((q31_t const *) pSrc);
+        vecDst = vqshlq_r(vecSrc, shiftBits);
+        vstrwq_p(pDst, vecDst, p0);
+    }
+}
+
+
+#else
+void arm_shift_q31(
+  const q31_t * pSrc,
+        int8_t shiftBits,
+        q31_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        uint8_t sign = (shiftBits & 0x80);             /* Sign of shiftBits */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  q31_t in, out;                                 /* Temporary variables */
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  /* If the shift value is positive then do right shift else left shift */
+  if (sign == 0U)
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A << shiftBits */
+
+      /* Shift input and store result in destination buffer. */
+      in = *pSrc++;
+      out = in << shiftBits;
+      if (in != (out >> shiftBits))
+        out = 0x7FFFFFFF ^ (in >> 31);
+      *pDst++ = out;
+
+      in = *pSrc++;
+      out = in << shiftBits;
+      if (in != (out >> shiftBits))
+        out = 0x7FFFFFFF ^ (in >> 31);
+      *pDst++ = out;
+
+      in = *pSrc++;
+      out = in << shiftBits;
+      if (in != (out >> shiftBits))
+        out = 0x7FFFFFFF ^ (in >> 31);
+      *pDst++ = out;
+
+      in = *pSrc++;
+      out = in << shiftBits;
+      if (in != (out >> shiftBits))
+        out = 0x7FFFFFFF ^ (in >> 31);
+      *pDst++ = out;
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+  else
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A >> shiftBits */
+
+      /* Shift input and store results in destination buffer. */
+      *pDst++ = (*pSrc++ >> -shiftBits);
+      *pDst++ = (*pSrc++ >> -shiftBits);
+      *pDst++ = (*pSrc++ >> -shiftBits);
+      *pDst++ = (*pSrc++ >> -shiftBits);
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  /* If the shift value is positive then do right shift else left shift */
+  if (sign == 0U)
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A << shiftBits */
+
+      /* Shift input and store result in destination buffer. */
+      *pDst++ = clip_q63_to_q31((q63_t) *pSrc++ << shiftBits);
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+  else
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A >> shiftBits */
+
+      /* Shift input and store result in destination buffer. */
+      *pDst++ = (*pSrc++ >> -shiftBits);
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicShift group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_shift_q7.c

@@ -0,0 +1,225 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_shift_q7.c
+ * Description:  Processing function for the Q7 Shifting
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicShift
+  @{
+ */
+
+/**
+  @brief         Shifts the elements of a Q7 vector a specified number of bits
+  @param[in]     pSrc       points to the input vector
+  @param[in]     shiftBits  number of bits to shift.  A positive value shifts left; a negative value shifts right.
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           onditions for optimum performance
+                   Input and output buffers should be aligned by 32-bit
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q7 range [0x80 0x7F] are saturated.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_shift_q7(
+    const q7_t * pSrc,
+    int8_t shiftBits,
+    q7_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q7x16_t vecSrc;
+    q7x16_t vecDst;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A (>> or <<) shiftBits
+         * Shift the input and then store the result in the destination buffer.
+         */
+        vecSrc = vld1q(pSrc);
+        vecDst = vqshlq_r(vecSrc, shiftBits);
+        vst1q(pDst, vecDst);
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrc += 16;
+        pDst += 16;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 0xF;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp8q(blkCnt);
+        vecSrc = vld1q(pSrc);
+        vecDst = vqshlq_r(vecSrc, shiftBits);
+        vstrbq_p(pDst, vecDst, p0);
+    }
+}
+
+#else
+void arm_shift_q7(
+  const q7_t * pSrc,
+        int8_t shiftBits,
+        q7_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        uint8_t sign = (shiftBits & 0x80);             /* Sign of shiftBits */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+#if defined (ARM_MATH_DSP)
+  q7_t in1,  in2,  in3,  in4;                    /* Temporary input variables */
+#endif
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  /* If the shift value is positive then do right shift else left shift */
+  if (sign == 0U)
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A << shiftBits */
+
+#if defined (ARM_MATH_DSP)
+      /* Read 4 inputs */
+      in1 = *pSrc++;
+      in2 = *pSrc++;
+      in3 = *pSrc++;
+      in4 = *pSrc++;
+
+    /* Pack and store result in destination buffer (in single write) */
+      write_q7x4_ia (&pDst, __PACKq7(__SSAT((in1 << shiftBits), 8),
+                                     __SSAT((in2 << shiftBits), 8),
+                                     __SSAT((in3 << shiftBits), 8),
+                                     __SSAT((in4 << shiftBits), 8) ));
+#else
+      *pDst++ = (q7_t) __SSAT(((q15_t) *pSrc++ << shiftBits), 8);
+      *pDst++ = (q7_t) __SSAT(((q15_t) *pSrc++ << shiftBits), 8);
+      *pDst++ = (q7_t) __SSAT(((q15_t) *pSrc++ << shiftBits), 8);
+      *pDst++ = (q7_t) __SSAT(((q15_t) *pSrc++ << shiftBits), 8);
+#endif
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+  else
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A >> shiftBits */
+
+#if defined (ARM_MATH_DSP)
+      /* Read 4 inputs */
+      in1 = *pSrc++;
+      in2 = *pSrc++;
+      in3 = *pSrc++;
+      in4 = *pSrc++;
+
+    /* Pack and store result in destination buffer (in single write) */
+      write_q7x4_ia (&pDst, __PACKq7((in1 >> -shiftBits),
+                                     (in2 >> -shiftBits),
+                                     (in3 >> -shiftBits),
+                                     (in4 >> -shiftBits) ));
+#else
+      *pDst++ = (*pSrc++ >> -shiftBits);
+      *pDst++ = (*pSrc++ >> -shiftBits);
+      *pDst++ = (*pSrc++ >> -shiftBits);
+      *pDst++ = (*pSrc++ >> -shiftBits);
+#endif
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  /* If the shift value is positive then do right shift else left shift */
+  if (sign == 0U)
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A << shiftBits */
+
+      /* Shift input and store result in destination buffer. */
+      *pDst++ = (q7_t) __SSAT(((q15_t) *pSrc++ << shiftBits), 8);
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+  else
+  {
+    while (blkCnt > 0U)
+    {
+      /* C = A >> shiftBits */
+
+      /* Shift input and store result in destination buffer. */
+      *pDst++ = (*pSrc++ >> -shiftBits);
+
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicShift group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_sub_f32.c

@@ -0,0 +1,202 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_sub_f32.c
+ * Description:  Floating-point vector subtraction
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @defgroup BasicSub Vector Subtraction
+
+  Element-by-element subtraction of two vectors.
+
+  <pre>
+      pDst[n] = pSrcA[n] - pSrcB[n],   0 <= n < blockSize.
+  </pre>
+
+  There are separate functions for floating-point, Q7, Q15, and Q31 data types.
+ */
+
+/**
+  @addtogroup BasicSub
+  @{
+ */
+
+/**
+  @brief         Floating-point vector subtraction.
+  @param[in]     pSrcA      points to the first input vector
+  @param[in]     pSrcB      points to the second input vector
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#include "arm_helium_utils.h"
+
+void arm_sub_f32(
+  const float32_t * pSrcA,
+  const float32_t * pSrcB,
+        float32_t * pDst,
+        uint32_t blockSize)
+{
+    uint32_t blkCnt;                               /* Loop counter */
+
+    f32x4_t vec1;
+    f32x4_t vec2;
+    f32x4_t res;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        /* C = A + B */
+
+      /* Add and then store the results in the destination buffer. */
+        vec1 = vld1q(pSrcA);
+        vec2 = vld1q(pSrcB);
+        res = vsubq(vec1, vec2);
+        vst1q(pDst, res);
+
+        /* Increment pointers */
+        pSrcA += 4;
+        pSrcB += 4; 
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+    if (blkCnt > 0U)
+    {
+      /* C = A + B */
+      mve_pred16_t p0 = vctp32q(blkCnt);
+      vec1 = vld1q(pSrcA);
+      vec2 = vld1q(pSrcB);
+      vstrwq_p(pDst, vsubq(vec1,vec2), p0);
+    }
+
+}
+
+#else
+void arm_sub_f32(
+  const float32_t * pSrcA,
+  const float32_t * pSrcB,
+        float32_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    f32x4_t vec1;
+    f32x4_t vec2;
+    f32x4_t res;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        /* C = A - B */
+
+        /* Subtract and then store the results in the destination buffer. */
+        vec1 = vld1q_f32(pSrcA);
+        vec2 = vld1q_f32(pSrcB);
+        res = vsubq_f32(vec1, vec2);
+        vst1q_f32(pDst, res);
+
+        /* Increment pointers */
+        pSrcA += 4;
+        pSrcB += 4; 
+        pDst += 4;
+        
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0x3;
+
+#else
+#if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A - B */
+
+    /* Subtract and store result in destination buffer. */
+    *pDst++ = (*pSrcA++) - (*pSrcB++);
+
+    *pDst++ = (*pSrcA++) - (*pSrcB++);
+
+    *pDst++ = (*pSrcA++) - (*pSrcB++);
+
+    *pDst++ = (*pSrcA++) - (*pSrcB++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined(ARM_MATH_NEON) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A - B */
+
+    /* Subtract and store result in destination buffer. */
+    *pDst++ = (*pSrcA++) - (*pSrcB++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+/**
+  @} end of BasicSub group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_sub_q15.c

@@ -0,0 +1,178 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_sub_q15.c
+ * Description:  Q15 vector subtraction
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicSub
+  @{
+ */
+
+/**
+  @brief         Q15 vector subtraction.
+  @param[in]     pSrcA      points to the first input vector
+  @param[in]     pSrcB      points to the second input vector
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q15 range [0x8000 0x7FFF] are saturated.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_sub_q15(
+    const q15_t * pSrcA,
+    const q15_t * pSrcB,
+    q15_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q15x8_t vecA;
+    q15x8_t vecB;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A - B
+         * Subtract and then store the results in the destination buffer.
+         */
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vst1q(pDst, vqsubq(vecA, vecB));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrcA  += 8;
+        pSrcB  += 8;
+        pDst   += 8;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 7;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp16q(blkCnt);
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vstrhq_p(pDst, vqsubq(vecA, vecB), p0);
+    }
+}
+
+
+#else
+void arm_sub_q15(
+  const q15_t * pSrcA,
+  const q15_t * pSrcB,
+        q15_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+#if defined (ARM_MATH_DSP)
+  q31_t inA1, inA2;
+  q31_t inB1, inB2;
+#endif
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A - B */
+
+#if defined (ARM_MATH_DSP)
+    /* read 2 times 2 samples at a time from sourceA */
+    inA1 = read_q15x2_ia ((q15_t **) &pSrcA);
+    inA2 = read_q15x2_ia ((q15_t **) &pSrcA);
+    /* read 2 times 2 samples at a time from sourceB */
+    inB1 = read_q15x2_ia ((q15_t **) &pSrcB);
+    inB2 = read_q15x2_ia ((q15_t **) &pSrcB);
+
+    /* Subtract and store 2 times 2 samples at a time */
+    write_q15x2_ia (&pDst, __QSUB16(inA1, inB1));
+    write_q15x2_ia (&pDst, __QSUB16(inA2, inB2));
+#else
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ - *pSrcB++), 16);
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ - *pSrcB++), 16);
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ - *pSrcB++), 16);
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ - *pSrcB++), 16);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A - B */
+
+    /* Subtract and store result in destination buffer. */
+#if defined (ARM_MATH_DSP)
+    *pDst++ = (q15_t) __QSUB16(*pSrcA++, *pSrcB++);
+#else
+    *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ - *pSrcB++), 16);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicSub group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_sub_q31.c

@@ -0,0 +1,159 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_sub_q31.c
+ * Description:  Q31 vector subtraction
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicSub
+  @{
+ */
+
+/**
+  @brief         Q31 vector subtraction.
+  @param[in]     pSrcA      points to the first input vector
+  @param[in]     pSrcB      points to the second input vector
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q31 range [0x80000000 0x7FFFFFFF] are saturated.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_sub_q31(
+  const q31_t * pSrcA,
+  const q31_t * pSrcB,
+        q31_t * pDst,
+        uint32_t blockSize)
+{
+    uint32_t blkCnt;   
+    q31x4_t vecA;
+    q31x4_t vecB;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A + B
+         * Add and then store the results in the destination buffer.
+         */
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vst1q(pDst, vqsubq(vecA, vecB));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrcA  += 4;
+        pSrcB  += 4;
+        pDst   += 4;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 3;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vstrwq_p(pDst, vqsubq(vecA, vecB), p0);
+    }
+}
+
+#else
+void arm_sub_q31(
+  const q31_t * pSrcA,
+  const q31_t * pSrcB,
+        q31_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A - B */
+
+    /* Subtract and store result in destination buffer. */
+    *pDst++ = __QSUB(*pSrcA++, *pSrcB++);
+
+    *pDst++ = __QSUB(*pSrcA++, *pSrcB++);
+
+    *pDst++ = __QSUB(*pSrcA++, *pSrcB++);
+
+    *pDst++ = __QSUB(*pSrcA++, *pSrcB++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A - B */
+
+    /* Subtract and store result in destination buffer. */
+    *pDst++ = __QSUB(*pSrcA++, *pSrcB++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicSub group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_sub_q7.c

@@ -0,0 +1,158 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_sub_q7.c
+ * Description:  Q7 vector subtraction
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup BasicSub
+  @{
+ */
+
+/**
+  @brief         Q7 vector subtraction.
+  @param[in]     pSrcA      points to the first input vector
+  @param[in]     pSrcB      points to the second input vector
+  @param[out]    pDst       points to the output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q7 range [0x80 0x7F] will be saturated.
+ */
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_sub_q7(
+    const q7_t * pSrcA,
+    const q7_t * pSrcB,
+    q7_t * pDst,
+    uint32_t blockSize)
+{
+    uint32_t  blkCnt;           /* loop counters */
+    q7x16_t vecA;
+    q7x16_t vecB;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4;
+    while (blkCnt > 0U)
+    {
+        /*
+         * C = A - B
+         * Subtract and then store the results in the destination buffer.
+         */
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vst1q(pDst, vqsubq(vecA, vecB));
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        /*
+         * advance vector source and destination pointers
+         */
+        pSrcA  += 16;
+        pSrcB  += 16;
+        pDst   += 16;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 0xF;
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp8q(blkCnt);
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        vstrbq_p(pDst, vqsubq(vecA, vecB), p0);
+    }
+}
+#else
+void arm_sub_q7(
+  const q7_t * pSrcA,
+  const q7_t * pSrcB,
+        q7_t * pDst,
+        uint32_t blockSize)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = blockSize >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C = A - B */
+
+#if defined (ARM_MATH_DSP)
+    /* Subtract and store result in destination buffer (4 samples at a time). */
+    write_q7x4_ia (&pDst, __QSUB8(read_q7x4_ia ((q7_t **) &pSrcA), read_q7x4_ia ((q7_t **) &pSrcB)));
+#else
+    *pDst++ = (q7_t) __SSAT((q15_t) *pSrcA++ - *pSrcB++, 8);
+    *pDst++ = (q7_t) __SSAT((q15_t) *pSrcA++ - *pSrcB++, 8);
+    *pDst++ = (q7_t) __SSAT((q15_t) *pSrcA++ - *pSrcB++, 8);
+    *pDst++ = (q7_t) __SSAT((q15_t) *pSrcA++ - *pSrcB++, 8);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = blockSize % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = blockSize;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C = A - B */
+
+    /* Subtract and store result in destination buffer. */
+    *pDst++ = (q7_t) __SSAT((q15_t) *pSrcA++ - *pSrcB++, 8);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of BasicSub group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_xor_u16.c

@@ -0,0 +1,137 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_xor_u16.c
+ * Description:  uint16_t bitwise exclusive OR
+ *
+ * $Date:        14 November 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @defgroup Xor Vector bitwise exclusive OR
+
+  Compute the logical bitwise XOR.
+
+  There are separate functions for uint32_t, uint16_t, and uint8_t data types.
+ */
+
+/**
+  @addtogroup Xor
+  @{
+ */
+
+/**
+  @brief         Compute the logical bitwise XOR of two fixed-point vectors.
+  @param[in]     pSrcA      points to input vector A
+  @param[in]     pSrcB      points to input vector B
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+void arm_xor_u16(
+    const uint16_t * pSrcA,
+    const uint16_t * pSrcB,
+          uint16_t * pDst,
+          uint32_t blockSize)
+{
+    uint32_t blkCnt;      /* Loop counter */
+
+#if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
+    q15x8_t vecSrcA, vecSrcB;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3;
+
+    while (blkCnt > 0U)
+    {
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+
+        vst1q(pDst, veorq_u16(vecSrcA, vecSrcB) );
+
+        pSrcA += 8;
+        pSrcB += 8;
+        pDst  += 8;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 7;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp16q(blkCnt);
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+        vstrhq_p(pDst, veorq_u16(vecSrcA, vecSrcB), p0);
+    }
+#else
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    uint16x8_t vecA, vecB;
+
+    /* Compute 8 outputs at a time */
+    blkCnt = blockSize >> 3U;
+
+    while (blkCnt > 0U)
+    {
+        vecA = vld1q_u16(pSrcA);
+        vecB = vld1q_u16(pSrcB);
+
+        vst1q_u16(pDst, veorq_u16(vecA, vecB) );
+
+        pSrcA += 8;
+        pSrcB += 8;
+        pDst  += 8;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 7;
+#else
+    /* Initialize blkCnt with number of samples */
+    blkCnt = blockSize;
+#endif
+
+    while (blkCnt > 0U)
+    {
+        *pDst++ = (*pSrcA++)^(*pSrcB++);
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+#endif /* if defined(ARM_MATH_MVEI) */
+}
+
+/**
+  @} end of Xor group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_xor_u32.c

@@ -0,0 +1,129 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_xor_u32.c
+ * Description:  uint32_t bitwise exclusive OR
+ *
+ * $Date:        14 November 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup Xor
+  @{
+ */
+
+/**
+  @brief         Compute the logical bitwise XOR of two fixed-point vectors.
+  @param[in]     pSrcA      points to input vector A
+  @param[in]     pSrcB      points to input vector B
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+void arm_xor_u32(
+    const uint32_t * pSrcA,
+    const uint32_t * pSrcB,
+          uint32_t * pDst,
+          uint32_t blockSize)
+{
+    uint32_t blkCnt;      /* Loop counter */
+
+#if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
+    q31x4_t vecSrcA, vecSrcB;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2;
+
+    while (blkCnt > 0U)
+    {
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+
+        vst1q(pDst, veorq_u32(vecSrcA, vecSrcB) );
+
+        pSrcA += 4;
+        pSrcB += 4;
+        pDst  += 4;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 3;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp32q(blkCnt);
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+        vstrwq_p(pDst, veorq_u32(vecSrcA, vecSrcB), p0);
+    }
+#else
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    uint32x4_t vecA, vecB;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        vecA = vld1q_u32(pSrcA);
+        vecB = vld1q_u32(pSrcB);
+
+        vst1q_u32(pDst, veorq_u32(vecA, vecB) );
+
+        pSrcA += 4;
+        pSrcB += 4;
+        pDst  += 4;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 3;
+#else
+    /* Initialize blkCnt with number of samples */
+    blkCnt = blockSize;
+#endif
+
+    while (blkCnt > 0U)
+    {
+        *pDst++ = (*pSrcA++)^(*pSrcB++);
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+#endif /* if defined(ARM_MATH_MVEI) */
+}
+
+/**
+  @} end of Xor group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BasicMathFunctions/arm_xor_u8.c

@@ -0,0 +1,129 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_xor_u8.c
+ * Description:  uint8_t bitwise exclusive OR
+ *
+ * $Date:        14 November 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupMath
+ */
+
+/**
+  @addtogroup Xor
+  @{
+ */
+
+/**
+  @brief         Compute the logical bitwise XOR of two fixed-point vectors.
+  @param[in]     pSrcA      points to input vector A
+  @param[in]     pSrcB      points to input vector B
+  @param[out]    pDst       points to output vector
+  @param[in]     blockSize  number of samples in each vector
+  @return        none
+ */
+
+void arm_xor_u8(
+    const uint8_t * pSrcA,
+    const uint8_t * pSrcB,
+          uint8_t * pDst,
+          uint32_t blockSize)
+{
+    uint32_t blkCnt;      /* Loop counter */
+
+#if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
+    q7x16_t vecSrcA, vecSrcB;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4;
+
+    while (blkCnt > 0U)
+    {
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+
+        vst1q(pDst, veorq_u8(vecSrcA, vecSrcB) );
+
+        pSrcA += 16;
+        pSrcB += 16;
+        pDst  += 16;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0xF;
+
+    if (blkCnt > 0U)
+    {
+        mve_pred16_t p0 = vctp8q(blkCnt);
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+        vstrbq_p(pDst, veorq_u8(vecSrcA, vecSrcB), p0);
+    }
+#else
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    uint8x16_t vecA, vecB;
+
+    /* Compute 16 outputs at a time */
+    blkCnt = blockSize >> 4U;
+
+    while (blkCnt > 0U)
+    {
+        vecA = vld1q_u8(pSrcA);
+        vecB = vld1q_u8(pSrcB);
+
+        vst1q_u8(pDst, veorq_u8(vecA, vecB) );
+
+        pSrcA += 16;
+        pSrcB += 16;
+        pDst  += 16;
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 0xF;
+#else
+    /* Initialize blkCnt with number of samples */
+    blkCnt = blockSize;
+#endif
+
+    while (blkCnt > 0U)
+    {
+        *pDst++ = (*pSrcA++)^(*pSrcB++);
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+#endif /* if defined(ARM_MATH_MVEI) */
+}
+
+/**
+  @} end of Xor group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/BayesFunctions/arm_gaussian_naive_bayes_predict_f32.c

@@ -0,0 +1,397 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_naive_gaussian_bayes_predict_f32
+ * Description:  Naive Gaussian Bayesian Estimator
+ *
+ *
+ * Target Processor: Cortex-M and Cortex-A cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+#include <limits.h>
+#include <math.h>
+
+#define PI_F 3.1415926535897932384626433832795f
+#define DPI_F (2.0f*3.1415926535897932384626433832795f)
+
+/**
+ * @addtogroup groupBayes
+ * @{
+ */
+
+/**
+ * @brief Naive Gaussian Bayesian Estimator
+ *
+ * @param[in]  *S         points to a naive bayes instance structure
+ * @param[in]  *in        points to the elements of the input vector.
+ * @param[in]  *pBuffer   points to a buffer of length numberOfClasses
+ * @return The predicted class
+ *
+ * @par If the number of classes is big, MVE version will consume lot of
+ * stack since the log prior are computed on the stack.
+ *
+ */
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#include "arm_helium_utils.h"
+#include "arm_vec_math.h"
+
+uint32_t arm_gaussian_naive_bayes_predict_f32(const arm_gaussian_naive_bayes_instance_f32 *S, 
+   const float32_t * in, 
+   float32_t *pBuffer)
+{
+    uint32_t         nbClass;
+    const float32_t *pTheta = S->theta;
+    const float32_t *pSigma = S->sigma;
+    float32_t      *buffer = pBuffer;
+    const float32_t *pIn = in;
+    float32_t       result;
+    f32x4_t         vsigma;
+    float32_t       tmp;
+    f32x4_t         vacc1, vacc2;
+    uint32_t        index;
+    float32_t       logclassPriors[S->numberOfClasses];
+    float32_t      *pLogPrior = logclassPriors;
+
+    arm_vlog_f32((float32_t *) S->classPriors, logclassPriors, S->numberOfClasses);
+
+    pTheta = S->theta;
+    pSigma = S->sigma;
+
+    for (nbClass = 0; nbClass < S->numberOfClasses; nbClass++) {
+        pIn = in;
+
+        vacc1 = vdupq_n_f32(0);
+        vacc2 = vdupq_n_f32(0);
+
+        uint32_t         blkCnt =S->vectorDimension >> 2;
+        while (blkCnt > 0U) {
+            f32x4_t         vinvSigma, vtmp;
+
+            vsigma = vaddq_n_f32(vld1q(pSigma), S->epsilon);
+            vacc1 = vaddq(vacc1, vlogq_f32(vmulq_n_f32(vsigma, 2.0f * PI)));
+
+            vinvSigma = vrecip_medprec_f32(vsigma);
+
+            vtmp = vsubq(vld1q(pIn), vld1q(pTheta));
+            /* squaring */
+            vtmp = vmulq(vtmp, vtmp);
+
+            vacc2 = vfmaq(vacc2, vtmp, vinvSigma);
+
+            pIn += 4;
+            pTheta += 4;
+            pSigma += 4;
+            blkCnt--;
+        }
+
+        blkCnt = S->vectorDimension & 3;
+        if (blkCnt > 0U) {
+            mve_pred16_t    p0 = vctp32q(blkCnt);
+            f32x4_t         vinvSigma, vtmp;
+
+            vsigma = vaddq_n_f32(vld1q(pSigma), S->epsilon);
+            vacc1 =
+                vaddq_m_f32(vacc1, vacc1, vlogq_f32(vmulq_n_f32(vsigma, 2.0f * PI)), p0);
+
+            vinvSigma = vrecip_medprec_f32(vsigma);
+
+            vtmp = vsubq(vld1q(pIn), vld1q(pTheta));
+            /* squaring */
+            vtmp = vmulq(vtmp, vtmp);
+
+            vacc2 = vfmaq_m_f32(vacc2, vtmp, vinvSigma, p0);
+
+            pTheta += blkCnt;
+            pSigma += blkCnt;
+        }
+
+        tmp = -0.5f * vecAddAcrossF32Mve(vacc1);
+        tmp -= 0.5f * vecAddAcrossF32Mve(vacc2);
+
+        *buffer = tmp + *pLogPrior++;
+        buffer++;
+    }
+
+    arm_max_f32(pBuffer, S->numberOfClasses, &result, &index);
+
+    return (index);
+}
+
+#else
+
+#if defined(ARM_MATH_NEON)
+
+#include "NEMath.h"
+
+
+
+uint32_t arm_gaussian_naive_bayes_predict_f32(const arm_gaussian_naive_bayes_instance_f32 *S, 
+   const float32_t * in, 
+   float32_t *pBuffer)
+{
+    
+    const float32_t *pPrior = S->classPriors;
+
+    const float32_t *pTheta = S->theta;
+    const float32_t *pSigma = S->sigma;
+
+    const float32_t *pTheta1 = S->theta + S->vectorDimension;
+    const float32_t *pSigma1 = S->sigma + S->vectorDimension;
+
+    float32_t *buffer = pBuffer;
+    const float32_t *pIn=in;
+
+    float32_t result;
+    float32_t sigma,sigma1;
+    float32_t tmp,tmp1;
+    uint32_t index;
+    uint32_t vecBlkCnt;
+    uint32_t classBlkCnt;
+    float32x4_t epsilonV;
+    float32x4_t sigmaV,sigmaV1;
+    float32x4_t tmpV,tmpVb,tmpV1;
+    float32x2_t tmpV2;
+    float32x4_t thetaV,thetaV1;
+    float32x4_t inV;
+
+    epsilonV = vdupq_n_f32(S->epsilon);
+
+    classBlkCnt = S->numberOfClasses >> 1;
+    while(classBlkCnt > 0)
+    {
+
+        
+        pIn = in;
+
+        tmp = logf(*pPrior++);
+        tmp1 = logf(*pPrior++);
+        tmpV = vdupq_n_f32(0.0f);
+        tmpV1 = vdupq_n_f32(0.0f);
+
+        vecBlkCnt = S->vectorDimension >> 2;
+        while(vecBlkCnt > 0)
+        {
+           sigmaV = vld1q_f32(pSigma);
+           thetaV = vld1q_f32(pTheta);
+
+           sigmaV1 = vld1q_f32(pSigma1);
+           thetaV1 = vld1q_f32(pTheta1);
+
+           inV = vld1q_f32(pIn);
+
+           sigmaV = vaddq_f32(sigmaV, epsilonV);
+           sigmaV1 = vaddq_f32(sigmaV1, epsilonV);
+
+           tmpVb = vmulq_n_f32(sigmaV,DPI_F);
+           tmpVb = vlogq_f32(tmpVb);
+           tmpV = vmlsq_n_f32(tmpV,tmpVb,0.5f);
+
+           tmpVb = vmulq_n_f32(sigmaV1,DPI_F);
+           tmpVb = vlogq_f32(tmpVb);
+           tmpV1 = vmlsq_n_f32(tmpV1,tmpVb,0.5f);
+           
+           tmpVb = vsubq_f32(inV,thetaV);
+           tmpVb = vmulq_f32(tmpVb,tmpVb);
+           tmpVb = vmulq_f32(tmpVb, vinvq_f32(sigmaV));
+           tmpV = vmlsq_n_f32(tmpV,tmpVb,0.5f);
+
+           tmpVb = vsubq_f32(inV,thetaV1);
+           tmpVb = vmulq_f32(tmpVb,tmpVb);
+           tmpVb = vmulq_f32(tmpVb, vinvq_f32(sigmaV1));
+           tmpV1 = vmlsq_n_f32(tmpV1,tmpVb,0.5f);
+
+           pIn += 4;
+           pTheta += 4;
+           pSigma += 4;
+           pTheta1 += 4;
+           pSigma1 += 4;
+
+           vecBlkCnt--;
+        }
+        tmpV2 = vpadd_f32(vget_low_f32(tmpV),vget_high_f32(tmpV));
+        tmp += vget_lane_f32(tmpV2, 0) + vget_lane_f32(tmpV2, 1);
+         
+        tmpV2 = vpadd_f32(vget_low_f32(tmpV1),vget_high_f32(tmpV1));
+        tmp1 += vget_lane_f32(tmpV2, 0) + vget_lane_f32(tmpV2, 1);
+
+        vecBlkCnt = S->vectorDimension & 3;
+        while(vecBlkCnt > 0)
+        {
+           sigma = *pSigma + S->epsilon;
+           sigma1 = *pSigma1 + S->epsilon;
+
+           tmp -= 0.5f*logf(2.0f * PI_F * sigma);
+           tmp -= 0.5f*(*pIn - *pTheta) * (*pIn - *pTheta) / sigma;
+
+           tmp1 -= 0.5f*logf(2.0f * PI_F * sigma1);
+           tmp1 -= 0.5f*(*pIn - *pTheta1) * (*pIn - *pTheta1) / sigma1;
+
+           pIn++;
+           pTheta++;
+           pSigma++;
+           pTheta1++;
+           pSigma1++;
+           vecBlkCnt--;
+        }
+
+        *buffer++ = tmp;
+        *buffer++ = tmp1;
+
+        pSigma += S->vectorDimension;
+        pTheta += S->vectorDimension;
+        pSigma1 += S->vectorDimension;
+        pTheta1 += S->vectorDimension;
+        
+        classBlkCnt--;
+    }
+
+    classBlkCnt = S->numberOfClasses & 1;
+
+    while(classBlkCnt > 0)
+    {
+
+        
+        pIn = in;
+
+        tmp = logf(*pPrior++);
+        tmpV = vdupq_n_f32(0.0f);
+
+        vecBlkCnt = S->vectorDimension >> 2;
+        while(vecBlkCnt > 0)
+        {
+           sigmaV = vld1q_f32(pSigma);
+           thetaV = vld1q_f32(pTheta);
+           inV = vld1q_f32(pIn);
+
+           sigmaV = vaddq_f32(sigmaV, epsilonV);
+
+           tmpVb = vmulq_n_f32(sigmaV,DPI_F);
+           tmpVb = vlogq_f32(tmpVb);
+           tmpV = vmlsq_n_f32(tmpV,tmpVb,0.5f);
+           
+           tmpVb = vsubq_f32(inV,thetaV);
+           tmpVb = vmulq_f32(tmpVb,tmpVb);
+           tmpVb = vmulq_f32(tmpVb, vinvq_f32(sigmaV));
+           tmpV = vmlsq_n_f32(tmpV,tmpVb,0.5f);
+
+           pIn += 4;
+           pTheta += 4;
+           pSigma += 4;
+
+           vecBlkCnt--;
+        }
+        tmpV2 = vpadd_f32(vget_low_f32(tmpV),vget_high_f32(tmpV));
+        tmp += vget_lane_f32(tmpV2, 0) + vget_lane_f32(tmpV2, 1);
+
+        vecBlkCnt = S->vectorDimension & 3;
+        while(vecBlkCnt > 0)
+        {
+           sigma = *pSigma + S->epsilon;
+           tmp -= 0.5f*logf(2.0f * PI_F * sigma);
+           tmp -= 0.5f*(*pIn - *pTheta) * (*pIn - *pTheta) / sigma;
+
+           pIn++;
+           pTheta++;
+           pSigma++;
+           vecBlkCnt--;
+        }
+
+        *buffer++ = tmp;
+        
+        classBlkCnt--;
+    }
+
+    arm_max_f32(pBuffer,S->numberOfClasses,&result,&index);
+
+    return(index);
+}
+
+#else
+
+/**
+ * @brief Naive Gaussian Bayesian Estimator
+ *
+ * @param[in]  *S         points to a naive bayes instance structure
+ * @param[in]  *in        points to the elements of the input vector.
+ * @param[in]  *pBuffer   points to a buffer of length numberOfClasses
+ * @return The predicted class
+ *
+ */
+uint32_t arm_gaussian_naive_bayes_predict_f32(const arm_gaussian_naive_bayes_instance_f32 *S, 
+   const float32_t * in, 
+   float32_t *pBuffer)
+{
+    uint32_t nbClass;
+    uint32_t nbDim;
+    const float32_t *pPrior = S->classPriors;
+    const float32_t *pTheta = S->theta;
+    const float32_t *pSigma = S->sigma;
+    float32_t *buffer = pBuffer;
+    const float32_t *pIn=in;
+    float32_t result;
+    float32_t sigma;
+    float32_t tmp;
+    float32_t acc1,acc2;
+    uint32_t index;
+
+    pTheta=S->theta;
+    pSigma=S->sigma;
+
+    for(nbClass = 0; nbClass < S->numberOfClasses; nbClass++)
+    {
+
+        
+        pIn = in;
+
+        tmp = 0.0;
+        acc1 = 0.0f;
+        acc2 = 0.0f;
+        for(nbDim = 0; nbDim < S->vectorDimension; nbDim++)
+        {
+           sigma = *pSigma + S->epsilon;
+           acc1 += logf(2.0f * PI_F * sigma);
+           acc2 += (*pIn - *pTheta) * (*pIn - *pTheta) / sigma;
+
+           pIn++;
+           pTheta++;
+           pSigma++;
+        }
+
+        tmp = -0.5f * acc1;
+        tmp -= 0.5f * acc2;
+
+
+        *buffer = tmp + logf(*pPrior++);
+        buffer++;
+    }
+
+    arm_max_f32(pBuffer,S->numberOfClasses,&result,&index);
+
+    return(index);
+}
+
+#endif
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+/**
+ * @} end of groupBayes group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/CommonTables/arm_const_structs.c

@@ -0,0 +1,663 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_const_structs.c
+ * Description:  Constant structs that are initialized for user convenience.
+ *               For example, some can be given as arguments to the arm_cfft_f32() or arm_rfft_f32() functions.
+ *
+ * $Date:        27. January 2017
+ * $Revision:    V.1.5.1
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2017 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+#include "arm_const_structs.h"
+
+/*
+ALLOW TABLE is true when config table is enabled and the Tramsform folder is included 
+for compilation.
+*/
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_FFT_ALLOW_TABLES)
+
+/* Floating-point structs */
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_16) && defined(ARM_TABLE_BITREVIDX_FLT64_16))
+const arm_cfft_instance_f64 arm_cfft_sR_f64_len16 = {
+  16, (const float64_t *)twiddleCoefF64_16, armBitRevIndexTableF64_16, ARMBITREVINDEXTABLEF64_16_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_32) && defined(ARM_TABLE_BITREVIDX_FLT64_32))
+const arm_cfft_instance_f64 arm_cfft_sR_f64_len32 = {
+  32, (const float64_t *)twiddleCoefF64_32, armBitRevIndexTableF64_32, ARMBITREVINDEXTABLEF64_32_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_64) && defined(ARM_TABLE_BITREVIDX_FLT64_64))
+const arm_cfft_instance_f64 arm_cfft_sR_f64_len64 = {
+  64, (const float64_t *)twiddleCoefF64_64, armBitRevIndexTableF64_64, ARMBITREVINDEXTABLEF64_64_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_128) && defined(ARM_TABLE_BITREVIDX_FLT64_128))
+const arm_cfft_instance_f64 arm_cfft_sR_f64_len128 = {
+  128, (const float64_t *)twiddleCoefF64_128, armBitRevIndexTableF64_128, ARMBITREVINDEXTABLEF64_128_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_256) && defined(ARM_TABLE_BITREVIDX_FLT64_256))
+const arm_cfft_instance_f64 arm_cfft_sR_f64_len256 = {
+  256, (const float64_t *)twiddleCoefF64_256, armBitRevIndexTableF64_256, ARMBITREVINDEXTABLEF64_256_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_512) && defined(ARM_TABLE_BITREVIDX_FLT64_512))
+const arm_cfft_instance_f64 arm_cfft_sR_f64_len512 = {
+  512, (const float64_t *)twiddleCoefF64_512, armBitRevIndexTableF64_512, ARMBITREVINDEXTABLEF64_512_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_1024) && defined(ARM_TABLE_BITREVIDX_FLT64_1024))
+const arm_cfft_instance_f64 arm_cfft_sR_f64_len1024 = {
+  1024, (const float64_t *)twiddleCoefF64_1024, armBitRevIndexTableF64_1024, ARMBITREVINDEXTABLEF64_1024_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_2048) && defined(ARM_TABLE_BITREVIDX_FLT64_2048))
+const arm_cfft_instance_f64 arm_cfft_sR_f64_len2048 = {
+  2048, (const float64_t *)twiddleCoefF64_2048, armBitRevIndexTableF64_2048, ARMBITREVINDEXTABLEF64_2048_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_4096) && defined(ARM_TABLE_BITREVIDX_FLT64_4096))
+const arm_cfft_instance_f64 arm_cfft_sR_f64_len4096 = {
+  4096, (const float64_t *)twiddleCoefF64_4096, armBitRevIndexTableF64_4096, ARMBITREVINDEXTABLEF64_4096_TABLE_LENGTH
+};
+#endif
+
+/* Floating-point structs */
+#if !defined(ARM_MATH_MVEF) || defined(ARM_MATH_AUTOVECTORIZE)
+
+/* 
+
+Those structures cannot be used to initialize the MVE version of the FFT F32 instances.
+So they are not compiled when MVE is defined.
+
+For the MVE version, the new arm_cfft_init_f32 must be used.
+
+
+*/
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_16) && defined(ARM_TABLE_BITREVIDX_FLT_16))
+const arm_cfft_instance_f32 arm_cfft_sR_f32_len16 = {
+  16, twiddleCoef_16, armBitRevIndexTable16, ARMBITREVINDEXTABLE_16_TABLE_LENGTH
+};
+#endif
+
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_32) && defined(ARM_TABLE_BITREVIDX_FLT_32))
+const arm_cfft_instance_f32 arm_cfft_sR_f32_len32 = {
+  32, twiddleCoef_32, armBitRevIndexTable32, ARMBITREVINDEXTABLE_32_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_64) && defined(ARM_TABLE_BITREVIDX_FLT_64))
+const arm_cfft_instance_f32 arm_cfft_sR_f32_len64 = {
+  64, twiddleCoef_64, armBitRevIndexTable64, ARMBITREVINDEXTABLE_64_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_128) && defined(ARM_TABLE_BITREVIDX_FLT_128))
+const arm_cfft_instance_f32 arm_cfft_sR_f32_len128 = {
+  128, twiddleCoef_128, armBitRevIndexTable128, ARMBITREVINDEXTABLE_128_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_256) && defined(ARM_TABLE_BITREVIDX_FLT_256))
+const arm_cfft_instance_f32 arm_cfft_sR_f32_len256 = {
+  256, twiddleCoef_256, armBitRevIndexTable256, ARMBITREVINDEXTABLE_256_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_512) && defined(ARM_TABLE_BITREVIDX_FLT_512))
+const arm_cfft_instance_f32 arm_cfft_sR_f32_len512 = {
+  512, twiddleCoef_512, armBitRevIndexTable512, ARMBITREVINDEXTABLE_512_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_1024) && defined(ARM_TABLE_BITREVIDX_FLT_1024))
+const arm_cfft_instance_f32 arm_cfft_sR_f32_len1024 = {
+  1024, twiddleCoef_1024, armBitRevIndexTable1024, ARMBITREVINDEXTABLE_1024_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_2048) && defined(ARM_TABLE_BITREVIDX_FLT_2048))
+const arm_cfft_instance_f32 arm_cfft_sR_f32_len2048 = {
+  2048, twiddleCoef_2048, armBitRevIndexTable2048, ARMBITREVINDEXTABLE_2048_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_4096) && defined(ARM_TABLE_BITREVIDX_FLT_4096))
+const arm_cfft_instance_f32 arm_cfft_sR_f32_len4096 = {
+  4096, twiddleCoef_4096, armBitRevIndexTable4096, ARMBITREVINDEXTABLE_4096_TABLE_LENGTH
+};
+#endif
+
+#endif /* !defined(ARM_MATH_MVEF) || defined(ARM_MATH_AUTOVECTORIZE) */
+
+/* Fixed-point structs */
+
+#if !defined(ARM_MATH_MVEI)
+
+/* 
+
+Those structures cannot be used to initialize the MVE version of the FFT Q31 instances.
+So they are not compiled when MVE is defined.
+
+For the MVE version, the new arm_cfft_init_f32 must be used.
+
+
+*/
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q31_16) && defined(ARM_TABLE_BITREVIDX_FXT_16))
+const arm_cfft_instance_q31 arm_cfft_sR_q31_len16 = {
+  16, twiddleCoef_16_q31, armBitRevIndexTable_fixed_16, ARMBITREVINDEXTABLE_FIXED_16_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q31_32) && defined(ARM_TABLE_BITREVIDX_FXT_32))
+const arm_cfft_instance_q31 arm_cfft_sR_q31_len32 = {
+  32, twiddleCoef_32_q31, armBitRevIndexTable_fixed_32, ARMBITREVINDEXTABLE_FIXED_32_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q31_64) && defined(ARM_TABLE_BITREVIDX_FXT_64))
+const arm_cfft_instance_q31 arm_cfft_sR_q31_len64 = {
+  64, twiddleCoef_64_q31, armBitRevIndexTable_fixed_64, ARMBITREVINDEXTABLE_FIXED_64_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q31_128) && defined(ARM_TABLE_BITREVIDX_FXT_128))
+const arm_cfft_instance_q31 arm_cfft_sR_q31_len128 = {
+  128, twiddleCoef_128_q31, armBitRevIndexTable_fixed_128, ARMBITREVINDEXTABLE_FIXED_128_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q31_256) && defined(ARM_TABLE_BITREVIDX_FXT_256))
+const arm_cfft_instance_q31 arm_cfft_sR_q31_len256 = {
+  256, twiddleCoef_256_q31, armBitRevIndexTable_fixed_256, ARMBITREVINDEXTABLE_FIXED_256_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q31_512) && defined(ARM_TABLE_BITREVIDX_FXT_512))
+const arm_cfft_instance_q31 arm_cfft_sR_q31_len512 = {
+  512, twiddleCoef_512_q31, armBitRevIndexTable_fixed_512, ARMBITREVINDEXTABLE_FIXED_512_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q31_1024) && defined(ARM_TABLE_BITREVIDX_FXT_1024))
+const arm_cfft_instance_q31 arm_cfft_sR_q31_len1024 = {
+  1024, twiddleCoef_1024_q31, armBitRevIndexTable_fixed_1024, ARMBITREVINDEXTABLE_FIXED_1024_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q31_2048) && defined(ARM_TABLE_BITREVIDX_FXT_2048))
+const arm_cfft_instance_q31 arm_cfft_sR_q31_len2048 = {
+  2048, twiddleCoef_2048_q31, armBitRevIndexTable_fixed_2048, ARMBITREVINDEXTABLE_FIXED_2048_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q31_4096) && defined(ARM_TABLE_BITREVIDX_FXT_4096))
+const arm_cfft_instance_q31 arm_cfft_sR_q31_len4096 = {
+  4096, twiddleCoef_4096_q31, armBitRevIndexTable_fixed_4096, ARMBITREVINDEXTABLE_FIXED_4096_TABLE_LENGTH
+};
+#endif
+
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q15_16) && defined(ARM_TABLE_BITREVIDX_FXT_16))
+const arm_cfft_instance_q15 arm_cfft_sR_q15_len16 = {
+  16, twiddleCoef_16_q15, armBitRevIndexTable_fixed_16, ARMBITREVINDEXTABLE_FIXED_16_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q15_32) && defined(ARM_TABLE_BITREVIDX_FXT_32))
+const arm_cfft_instance_q15 arm_cfft_sR_q15_len32 = {
+  32, twiddleCoef_32_q15, armBitRevIndexTable_fixed_32, ARMBITREVINDEXTABLE_FIXED_32_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q15_64) && defined(ARM_TABLE_BITREVIDX_FXT_64))
+const arm_cfft_instance_q15 arm_cfft_sR_q15_len64 = {
+  64, twiddleCoef_64_q15, armBitRevIndexTable_fixed_64, ARMBITREVINDEXTABLE_FIXED_64_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q15_128) && defined(ARM_TABLE_BITREVIDX_FXT_128))
+const arm_cfft_instance_q15 arm_cfft_sR_q15_len128 = {
+  128, twiddleCoef_128_q15, armBitRevIndexTable_fixed_128, ARMBITREVINDEXTABLE_FIXED_128_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q15_256) && defined(ARM_TABLE_BITREVIDX_FXT_256))
+const arm_cfft_instance_q15 arm_cfft_sR_q15_len256 = {
+  256, twiddleCoef_256_q15, armBitRevIndexTable_fixed_256, ARMBITREVINDEXTABLE_FIXED_256_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q15_512) && defined(ARM_TABLE_BITREVIDX_FXT_512))
+const arm_cfft_instance_q15 arm_cfft_sR_q15_len512 = {
+  512, twiddleCoef_512_q15, armBitRevIndexTable_fixed_512, ARMBITREVINDEXTABLE_FIXED_512_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q15_1024) && defined(ARM_TABLE_BITREVIDX_FXT_1024))
+const arm_cfft_instance_q15 arm_cfft_sR_q15_len1024 = {
+  1024, twiddleCoef_1024_q15, armBitRevIndexTable_fixed_1024, ARMBITREVINDEXTABLE_FIXED_1024_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q15_2048) && defined(ARM_TABLE_BITREVIDX_FXT_2048))
+const arm_cfft_instance_q15 arm_cfft_sR_q15_len2048 = {
+  2048, twiddleCoef_2048_q15, armBitRevIndexTable_fixed_2048, ARMBITREVINDEXTABLE_FIXED_2048_TABLE_LENGTH
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_Q15_4096) && defined(ARM_TABLE_BITREVIDX_FXT_4096))
+const arm_cfft_instance_q15 arm_cfft_sR_q15_len4096 = {
+  4096, twiddleCoef_4096_q15, armBitRevIndexTable_fixed_4096, ARMBITREVINDEXTABLE_FIXED_4096_TABLE_LENGTH
+};
+#endif
+
+#endif /* !defined(ARM_MATH_MVEI) */
+
+/* Structure for real-value inputs */
+/* Double precision strucs */
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_32) && defined(ARM_TABLE_BITREVIDX_FLT64_32) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_32))
+const arm_rfft_fast_instance_f64 arm_rfft_fast_sR_f64_len32 = {
+  { 16, (const float64_t *)twiddleCoefF64_16, armBitRevIndexTableF64_16, ARMBITREVINDEXTABLEF64_16_TABLE_LENGTH },
+  32U,
+  (float64_t *)twiddleCoefF64_rfft_32
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_64) && defined(ARM_TABLE_BITREVIDX_FLT64_64) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_64))
+const arm_rfft_fast_instance_f64 arm_rfft_fast_sR_f64_len64 = {
+   { 32, (const float64_t *)twiddleCoefF64_32, armBitRevIndexTableF64_32, ARMBITREVINDEXTABLEF64_32_TABLE_LENGTH },
+  64U,
+  (float64_t *)twiddleCoefF64_rfft_64
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_128) && defined(ARM_TABLE_BITREVIDX_FLT64_128) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_128))
+const arm_rfft_fast_instance_f64 arm_rfft_fast_sR_f64_len128 = {
+  { 64, (const float64_t *)twiddleCoefF64_64, armBitRevIndexTableF64_64, ARMBITREVINDEXTABLEF64_64_TABLE_LENGTH },
+  128U,
+  (float64_t *)twiddleCoefF64_rfft_128
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_256) && defined(ARM_TABLE_BITREVIDX_FLT64_256) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_256))
+const arm_rfft_fast_instance_f64 arm_rfft_fast_sR_f64_len256 = {
+  { 128, (const float64_t *)twiddleCoefF64_128, armBitRevIndexTableF64_128, ARMBITREVINDEXTABLEF64_128_TABLE_LENGTH },
+  256U,
+  (float64_t *)twiddleCoefF64_rfft_256
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_512) && defined(ARM_TABLE_BITREVIDX_FLT64_512) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_512))
+const arm_rfft_fast_instance_f64 arm_rfft_fast_sR_f64_len512 = {
+  { 256, (const float64_t *)twiddleCoefF64_256, armBitRevIndexTableF64_256, ARMBITREVINDEXTABLEF64_256_TABLE_LENGTH },
+  512U,
+  (float64_t *)twiddleCoefF64_rfft_512
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_1024) && defined(ARM_TABLE_BITREVIDX_FLT64_1024) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_1024))
+const arm_rfft_fast_instance_f64 arm_rfft_fast_sR_f64_len1024 = {
+  { 512, (const float64_t *)twiddleCoefF64_512, armBitRevIndexTableF64_512, ARMBITREVINDEXTABLEF64_512_TABLE_LENGTH },
+  1024U,
+  (float64_t *)twiddleCoefF64_rfft_1024
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_2048) && defined(ARM_TABLE_BITREVIDX_FLT64_2048) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_2048))
+const arm_rfft_fast_instance_f64 arm_rfft_fast_sR_f64_len2048 = {
+  { 1024, (const float64_t *)twiddleCoefF64_1024, armBitRevIndexTableF64_1024, ARMBITREVINDEXTABLEF64_1024_TABLE_LENGTH },
+  2048U,
+  (float64_t *)twiddleCoefF64_rfft_2048
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F64_4096) && defined(ARM_TABLE_BITREVIDX_FLT64_4096) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F64_4096))
+const arm_rfft_fast_instance_f64 arm_rfft_fast_sR_f64_len4096 = {
+  { 2048, (const float64_t *)twiddleCoefF64_2048, armBitRevIndexTableF64_2048, ARMBITREVINDEXTABLEF64_2048_TABLE_LENGTH },
+  4096U,
+  (float64_t *)twiddleCoefF64_rfft_4096
+};
+#endif
+
+/* Floating-point structs */
+
+#if !defined(ARM_MATH_MVEF) || defined(ARM_MATH_AUTOVECTORIZE)
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_32) && defined(ARM_TABLE_BITREVIDX_FLT_32) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_32))
+const arm_rfft_fast_instance_f32 arm_rfft_fast_sR_f32_len32 = {
+  { 16, twiddleCoef_16, armBitRevIndexTable16, ARMBITREVINDEXTABLE_16_TABLE_LENGTH },
+  32U,
+  (float32_t *)twiddleCoef_rfft_32
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_64) && defined(ARM_TABLE_BITREVIDX_FLT_64) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_64))
+const arm_rfft_fast_instance_f32 arm_rfft_fast_sR_f32_len64 = {
+   { 32, twiddleCoef_32, armBitRevIndexTable32, ARMBITREVINDEXTABLE_32_TABLE_LENGTH },
+  64U,
+  (float32_t *)twiddleCoef_rfft_64
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_128) && defined(ARM_TABLE_BITREVIDX_FLT_128) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_128))
+const arm_rfft_fast_instance_f32 arm_rfft_fast_sR_f32_len128 = {
+  { 64, twiddleCoef_64, armBitRevIndexTable64, ARMBITREVINDEXTABLE_64_TABLE_LENGTH },
+  128U,
+  (float32_t *)twiddleCoef_rfft_128
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_256) && defined(ARM_TABLE_BITREVIDX_FLT_256) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_256))
+const arm_rfft_fast_instance_f32 arm_rfft_fast_sR_f32_len256 = {
+  { 128, twiddleCoef_128, armBitRevIndexTable128, ARMBITREVINDEXTABLE_128_TABLE_LENGTH },
+  256U,
+  (float32_t *)twiddleCoef_rfft_256
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_512) && defined(ARM_TABLE_BITREVIDX_FLT_512) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_512))
+const arm_rfft_fast_instance_f32 arm_rfft_fast_sR_f32_len512 = {
+  { 256, twiddleCoef_256, armBitRevIndexTable256, ARMBITREVINDEXTABLE_256_TABLE_LENGTH },
+  512U,
+  (float32_t *)twiddleCoef_rfft_512
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_1024) && defined(ARM_TABLE_BITREVIDX_FLT_1024) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_1024))
+const arm_rfft_fast_instance_f32 arm_rfft_fast_sR_f32_len1024 = {
+  { 512, twiddleCoef_512, armBitRevIndexTable512, ARMBITREVINDEXTABLE_512_TABLE_LENGTH },
+  1024U,
+  (float32_t *)twiddleCoef_rfft_1024
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_2048) && defined(ARM_TABLE_BITREVIDX_FLT_2048) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_2048))
+const arm_rfft_fast_instance_f32 arm_rfft_fast_sR_f32_len2048 = {
+  { 1024, twiddleCoef_1024, armBitRevIndexTable1024, ARMBITREVINDEXTABLE_1024_TABLE_LENGTH },
+  2048U,
+  (float32_t *)twiddleCoef_rfft_2048
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_TWIDDLECOEF_F32_4096) && defined(ARM_TABLE_BITREVIDX_FLT_4096) && defined(ARM_TABLE_TWIDDLECOEF_RFFT_F32_4096))
+const arm_rfft_fast_instance_f32 arm_rfft_fast_sR_f32_len4096 = {
+  { 2048, twiddleCoef_2048, armBitRevIndexTable2048, ARMBITREVINDEXTABLE_2048_TABLE_LENGTH },
+  4096U,
+  (float32_t *)twiddleCoef_rfft_4096
+};
+#endif
+
+#endif /* #if !defined(ARM_MATH_MVEF) || defined(ARM_MATH_AUTOVECTORIZE) */
+
+/* Fixed-point structs */
+/* q31_t */
+
+#if !defined(ARM_MATH_MVEI)
+
+/* 
+
+Those structures cannot be used to initialize the MVE version of the FFT Q31 instances.
+So they are not compiled when MVE is defined.
+
+For the MVE version, the new arm_cfft_init_f32 must be used.
+
+
+*/
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q31) && defined(ARM_TABLE_TWIDDLECOEF_Q31_16) && defined(ARM_TABLE_BITREVIDX_FXT_16))
+const arm_rfft_instance_q31 arm_rfft_sR_q31_len32 = {
+  32U,
+  0,
+  1,
+  256U,
+  (q31_t*)realCoefAQ31,
+  (q31_t*)realCoefBQ31,
+  &arm_cfft_sR_q31_len16
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q31) && defined(ARM_TABLE_TWIDDLECOEF_Q31_32) && defined(ARM_TABLE_BITREVIDX_FXT_32))
+const arm_rfft_instance_q31 arm_rfft_sR_q31_len64 = {
+  64U,
+  0,
+  1,
+  128U,
+  (q31_t*)realCoefAQ31,
+  (q31_t*)realCoefBQ31,
+  &arm_cfft_sR_q31_len32
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q31) && defined(ARM_TABLE_TWIDDLECOEF_Q31_64) && defined(ARM_TABLE_BITREVIDX_FXT_64))
+const arm_rfft_instance_q31 arm_rfft_sR_q31_len128 = {
+  128U,
+  0,
+  1,
+  64U,
+  (q31_t*)realCoefAQ31,
+  (q31_t*)realCoefBQ31,
+  &arm_cfft_sR_q31_len64
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q31) && defined(ARM_TABLE_TWIDDLECOEF_Q31_128) && defined(ARM_TABLE_BITREVIDX_FXT_128))
+const arm_rfft_instance_q31 arm_rfft_sR_q31_len256 = {
+  256U,
+  0,
+  1,
+  32U,
+  (q31_t*)realCoefAQ31,
+  (q31_t*)realCoefBQ31,
+  &arm_cfft_sR_q31_len128
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q31) && defined(ARM_TABLE_TWIDDLECOEF_Q31_256) && defined(ARM_TABLE_BITREVIDX_FXT_256))
+const arm_rfft_instance_q31 arm_rfft_sR_q31_len512 = {
+  512U,
+  0,
+  1,
+  16U,
+  (q31_t*)realCoefAQ31,
+  (q31_t*)realCoefBQ31,
+  &arm_cfft_sR_q31_len256
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q31) && defined(ARM_TABLE_TWIDDLECOEF_Q31_512) && defined(ARM_TABLE_BITREVIDX_FXT_512))
+const arm_rfft_instance_q31 arm_rfft_sR_q31_len1024 = {
+  1024U,
+  0,
+  1,
+  8U,
+  (q31_t*)realCoefAQ31,
+  (q31_t*)realCoefBQ31,
+  &arm_cfft_sR_q31_len512
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q31) && defined(ARM_TABLE_TWIDDLECOEF_Q31_1024) && defined(ARM_TABLE_BITREVIDX_FXT_1024))
+const arm_rfft_instance_q31 arm_rfft_sR_q31_len2048 = {
+  2048U,
+  0,
+  1,
+  4U,
+  (q31_t*)realCoefAQ31,
+  (q31_t*)realCoefBQ31,
+  &arm_cfft_sR_q31_len1024
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q31) && defined(ARM_TABLE_TWIDDLECOEF_Q31_2048) && defined(ARM_TABLE_BITREVIDX_FXT_2048))
+const arm_rfft_instance_q31 arm_rfft_sR_q31_len4096 = {
+  4096U,
+  0,
+  1,
+  2U,
+  (q31_t*)realCoefAQ31,
+  (q31_t*)realCoefBQ31,
+  &arm_cfft_sR_q31_len2048
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q31) && defined(ARM_TABLE_TWIDDLECOEF_Q31_4096) && defined(ARM_TABLE_BITREVIDX_FXT_4096))
+const arm_rfft_instance_q31 arm_rfft_sR_q31_len8192 = {
+  8192U,
+  0,
+  1,
+  1U,
+  (q31_t*)realCoefAQ31,
+  (q31_t*)realCoefBQ31,
+  &arm_cfft_sR_q31_len4096
+};
+#endif
+
+
+
+/* q15_t */
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q15) && defined(ARM_TABLE_TWIDDLECOEF_Q15_16) && defined(ARM_TABLE_BITREVIDX_FXT_16))
+const arm_rfft_instance_q15 arm_rfft_sR_q15_len32 = {
+  32U,
+  0,
+  1,
+  256U,
+  (q15_t*)realCoefAQ15,
+  (q15_t*)realCoefBQ15,
+  &arm_cfft_sR_q15_len16
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q15) && defined(ARM_TABLE_TWIDDLECOEF_Q15_32) && defined(ARM_TABLE_BITREVIDX_FXT_32))
+const arm_rfft_instance_q15 arm_rfft_sR_q15_len64 = {
+  64U,
+  0,
+  1,
+  128U,
+  (q15_t*)realCoefAQ15,
+  (q15_t*)realCoefBQ15,
+  &arm_cfft_sR_q15_len32
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q15) && defined(ARM_TABLE_TWIDDLECOEF_Q15_64) && defined(ARM_TABLE_BITREVIDX_FXT_64))
+const arm_rfft_instance_q15 arm_rfft_sR_q15_len128 = {
+  128U,
+  0,
+  1,
+  64U,
+  (q15_t*)realCoefAQ15,
+  (q15_t*)realCoefBQ15,
+  &arm_cfft_sR_q15_len64
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q15) && defined(ARM_TABLE_TWIDDLECOEF_Q15_128) && defined(ARM_TABLE_BITREVIDX_FXT_128))
+const arm_rfft_instance_q15 arm_rfft_sR_q15_len256 = {
+  256U,
+  0,
+  1,
+  32U,
+  (q15_t*)realCoefAQ15,
+  (q15_t*)realCoefBQ15,
+  &arm_cfft_sR_q15_len128
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q15) && defined(ARM_TABLE_TWIDDLECOEF_Q15_256) && defined(ARM_TABLE_BITREVIDX_FXT_256))
+const arm_rfft_instance_q15 arm_rfft_sR_q15_len512 = {
+  512U,
+  0,
+  1,
+  16U,
+  (q15_t*)realCoefAQ15,
+  (q15_t*)realCoefBQ15,
+  &arm_cfft_sR_q15_len256
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q15) && defined(ARM_TABLE_TWIDDLECOEF_Q15_512) && defined(ARM_TABLE_BITREVIDX_FXT_512))
+const arm_rfft_instance_q15 arm_rfft_sR_q15_len1024 = {
+  1024U,
+  0,
+  1,
+  8U,
+  (q15_t*)realCoefAQ15,
+  (q15_t*)realCoefBQ15,
+  &arm_cfft_sR_q15_len512
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q15) && defined(ARM_TABLE_TWIDDLECOEF_Q15_1024) && defined(ARM_TABLE_BITREVIDX_FXT_1024))
+const arm_rfft_instance_q15 arm_rfft_sR_q15_len2048 = {
+  2048U,
+  0,
+  1,
+  4U,
+  (q15_t*)realCoefAQ15,
+  (q15_t*)realCoefBQ15,
+  &arm_cfft_sR_q15_len1024
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q15) && defined(ARM_TABLE_TWIDDLECOEF_Q15_2048) && defined(ARM_TABLE_BITREVIDX_FXT_2048))
+const arm_rfft_instance_q15 arm_rfft_sR_q15_len4096 = {
+  4096U,
+  0,
+  1,
+  2U,
+  (q15_t*)realCoefAQ15,
+  (q15_t*)realCoefBQ15,
+  &arm_cfft_sR_q15_len2048
+};
+#endif
+
+#if !defined(ARM_DSP_CONFIG_TABLES) || defined(ARM_ALL_FFT_TABLES) || (defined(ARM_TABLE_REALCOEF_Q15) && defined(ARM_TABLE_TWIDDLECOEF_Q15_4096) && defined(ARM_TABLE_BITREVIDX_FXT_4096))
+const arm_rfft_instance_q15 arm_rfft_sR_q15_len8192 = {
+  8192U,
+  0,
+  1,
+  1U,
+  (q15_t*)realCoefAQ15,
+  (q15_t*)realCoefBQ15,
+  &arm_cfft_sR_q15_len4096
+};
+#endif
+
+#endif /* !defined(ARM_MATH_MVEI) */
+
+
+#endif

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_conj_f32.c

@@ -0,0 +1,213 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_conj_f32.c
+ * Description:  Floating-point complex conjugate
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @defgroup cmplx_conj Complex Conjugate
+
+  Conjugates the elements of a complex data vector.
+
+  The <code>pSrc</code> points to the source data and
+  <code>pDst</code> points to the destination data where the result should be written.
+  <code>numSamples</code> specifies the number of complex samples
+  and the data in each array is stored in an interleaved fashion
+  (real, imag, real, imag, ...).
+  Each array has a total of <code>2*numSamples</code> values.
+
+  The underlying algorithm is used:
+  <pre>
+  for (n = 0; n < numSamples; n++) {
+      pDst[(2*n)  ] =  pSrc[(2*n)  ];    // real part
+      pDst[(2*n)+1] = -pSrc[(2*n)+1];    // imag part
+  }
+  </pre>
+
+  There are separate functions for floating-point, Q15, and Q31 data types.
+ */
+
+/**
+  @addtogroup cmplx_conj
+  @{
+ */
+
+/**
+  @brief         Floating-point complex conjugate.
+  @param[in]     pSrc        points to the input vector
+  @param[out]    pDst        points to the output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+ */
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+void arm_cmplx_conj_f32(
+    const float32_t * pSrc,
+    float32_t * pDst,
+    uint32_t numSamples)
+{
+    static const float32_t cmplx_conj_sign[4] = { 1.0f, -1.0f, 1.0f, -1.0f };
+    uint32_t blockSize = numSamples * CMPLX_DIM;   /* loop counters */
+    uint32_t blkCnt;
+    f32x4_t vecSrc;
+    f32x4_t vecSign;
+
+    /*
+     * load sign vector
+     */
+    vecSign = *(f32x4_t *) cmplx_conj_sign;
+
+    /* Compute 4 real samples at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        vecSrc = vld1q(pSrc);
+        vst1q(pDst,vmulq(vecSrc, vecSign));
+        /*
+         * Decrement the blkCnt loop counter
+         * Advance vector source and destination pointers
+         */
+        pSrc += 4;
+        pDst += 4;
+        blkCnt--;
+    }
+
+     /* Tail */
+    blkCnt = (blockSize & 0x3) >> 1;
+
+    while (blkCnt > 0U)
+    {
+      /* C[0] + jC[1] = A[0]+ j(-1)A[1] */
+  
+      /* Calculate Complex Conjugate and store result in destination buffer. */
+      *pDst++ =  *pSrc++;
+      *pDst++ = -*pSrc++;
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+
+}
+
+#else
+void arm_cmplx_conj_f32(
+  const float32_t * pSrc,
+        float32_t * pDst,
+        uint32_t numSamples)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+   float32x4_t zero;
+   float32x4x2_t vec;
+
+   zero = vdupq_n_f32(0.0f);
+
+   /* Compute 4 outputs at a time */
+   blkCnt = numSamples >> 2U;
+
+   while (blkCnt > 0U)
+   {
+     /* C[0]+jC[1] = A[0]+(-1)*jA[1] */
+     /* Calculate Complex Conjugate and then store the results in the destination buffer. */
+     vec = vld2q_f32(pSrc);
+     vec.val[1] = vsubq_f32(zero,vec.val[1]);
+     vst2q_f32(pDst,vec);
+
+     /* Increment pointers */
+     pSrc += 8;
+     pDst += 8;
+        
+     /* Decrement the loop counter */
+     blkCnt--;
+   }
+
+   /* Tail */
+   blkCnt = numSamples & 0x3;
+
+#else
+#if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] + jC[1] = A[0]+ j(-1)A[1] */
+
+    /* Calculate Complex Conjugate and store result in destination buffer. */
+    *pDst++ =  *pSrc++;
+    *pDst++ = -*pSrc++;
+
+    *pDst++ =  *pSrc++;
+    *pDst++ = -*pSrc++;
+
+    *pDst++ =  *pSrc++;
+    *pDst++ = -*pSrc++;
+
+    *pDst++ =  *pSrc++;
+    *pDst++ = -*pSrc++;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined (ARM_MATH_NEON) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] + jC[1] = A[0]+ j(-1)A[1] */
+
+    /* Calculate Complex Conjugate and store result in destination buffer. */
+    *pDst++ =  *pSrc++;
+    *pDst++ = -*pSrc++;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+/**
+  @} end of cmplx_conj group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_conj_q15.c

@@ -0,0 +1,207 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_conj_q15.c
+ * Description:  Q15 complex conjugate
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @addtogroup cmplx_conj
+  @{
+ */
+
+/**
+  @brief         Q15 complex conjugate.
+  @param[in]     pSrc        points to the input vector
+  @param[out]    pDst        points to the output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   The Q15 value -1 (0x8000) is saturated to the maximum allowable positive value 0x7FFF.
+ */
+
+
+#if defined(ARM_MATH_MVEI)
+void arm_cmplx_conj_q15(
+  const q15_t * pSrc,
+        q15_t * pDst,
+        uint32_t numSamples)
+{
+    uint32_t blockSize = numSamples * CMPLX_DIM;   /* loop counters */
+    uint32_t blkCnt;
+    q31_t in1; 
+
+    q15x8x2_t vecSrc;
+    q15x8_t zero;
+
+    zero = vdupq_n_s16(0);
+
+    /* Compute 8 real samples at a time */
+    blkCnt = blockSize >> 4U;
+    while (blkCnt > 0U)
+    {
+        vecSrc = vld2q(pSrc);
+        vecSrc.val[1] = vqsubq(zero, vecSrc.val[1]);
+        vst2q(pDst,vecSrc);
+        /*
+         * Decrement the blkCnt loop counter
+         * Advance vector source and destination pointers
+         */
+        pSrc += 16;
+        pDst += 16;
+        blkCnt --;
+    }
+    
+     /* Tail */
+    blkCnt = (blockSize & 0xF) >> 1;
+
+    while (blkCnt > 0U)
+    {
+      /* C[0] + jC[1] = A[0]+ j(-1)A[1] */
+  
+      /* Calculate Complex Conjugate and store result in destination buffer. */
+      *pDst++ =  *pSrc++;
+      in1 = *pSrc++;
+      *pDst++ = __SSAT(-in1, 16);
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+}
+#else
+void arm_cmplx_conj_q15(
+  const q15_t * pSrc,
+        q15_t * pDst,
+        uint32_t numSamples)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q31_t in1;                                     /* Temporary input variable */
+
+#if defined (ARM_MATH_LOOPUNROLL) && defined (ARM_MATH_DSP)
+        q31_t in2, in3, in4;                           /* Temporary input variables */
+#endif
+
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] + jC[1] = A[0]+ j(-1)A[1] */
+
+    /* Calculate Complex Conjugate and store result in destination buffer. */
+
+    #if defined (ARM_MATH_DSP)
+    in1 = read_q15x2_ia ((q15_t **) &pSrc);
+    in2 = read_q15x2_ia ((q15_t **) &pSrc);
+    in3 = read_q15x2_ia ((q15_t **) &pSrc);
+    in4 = read_q15x2_ia ((q15_t **) &pSrc);
+
+#ifndef ARM_MATH_BIG_ENDIAN
+    in1 = __QASX(0, in1);
+    in2 = __QASX(0, in2);
+    in3 = __QASX(0, in3);
+    in4 = __QASX(0, in4);
+#else
+    in1 = __QSAX(0, in1);
+    in2 = __QSAX(0, in2);
+    in3 = __QSAX(0, in3);
+    in4 = __QSAX(0, in4);
+#endif /* #ifndef ARM_MATH_BIG_ENDIAN */
+
+    in1 = ((uint32_t) in1 >> 16) | ((uint32_t) in1 << 16);
+    in2 = ((uint32_t) in2 >> 16) | ((uint32_t) in2 << 16);
+    in3 = ((uint32_t) in3 >> 16) | ((uint32_t) in3 << 16);
+    in4 = ((uint32_t) in4 >> 16) | ((uint32_t) in4 << 16);
+
+    write_q15x2_ia (&pDst, in1);
+    write_q15x2_ia (&pDst, in2);
+    write_q15x2_ia (&pDst, in3);
+    write_q15x2_ia (&pDst, in4);
+#else
+    *pDst++ =  *pSrc++;
+    in1 = *pSrc++;
+    *pDst++ = (in1 == (q15_t) 0x8000) ? (q15_t) 0x7fff : -in1;
+
+    *pDst++ =  *pSrc++;
+    in1 = *pSrc++;
+    *pDst++ = (in1 == (q15_t) 0x8000) ? (q15_t) 0x7fff : -in1;
+
+    *pDst++ =  *pSrc++;
+    in1 = *pSrc++;
+    *pDst++ = (in1 == (q15_t) 0x8000) ? (q15_t) 0x7fff : -in1;
+
+    *pDst++ =  *pSrc++;
+    in1 = *pSrc++;
+    *pDst++ = (in1 == (q15_t) 0x8000) ? (q15_t) 0x7fff : -in1;
+
+#endif /* #if defined (ARM_MATH_DSP) */
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] + jC[1] = A[0]+ j(-1)A[1] */
+
+    /* Calculate Complex Conjugate and store result in destination buffer. */
+    *pDst++ =  *pSrc++;
+    in1 = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __SSAT(-in1, 16);
+#else
+    *pDst++ = (in1 == (q15_t) 0x8000) ? (q15_t) 0x7fff : -in1;
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of cmplx_conj group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_conj_q31.c

@@ -0,0 +1,193 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_conj_q31.c
+ * Description:  Q31 complex conjugate
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @addtogroup cmplx_conj
+  @{
+ */
+
+/**
+  @brief         Q31 complex conjugate.
+  @param[in]     pSrc        points to the input vector
+  @param[out]    pDst        points to the output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   The Q31 value -1 (0x80000000) is saturated to the maximum allowable positive value 0x7FFFFFFF.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+void arm_cmplx_conj_q31(
+  const q31_t * pSrc,
+        q31_t * pDst,
+        uint32_t numSamples)
+{
+
+    uint32_t blockSize = numSamples * CMPLX_DIM;   /* loop counters */
+    uint32_t blkCnt;
+    q31x4x2_t vecSrc;
+    q31_t in;                                      /* Temporary input variable */
+    q31x4_t zero;
+
+    zero = vdupq_n_s32(0);
+
+   
+    /* Compute 4 real samples at a time */
+    blkCnt = blockSize >> 3U;
+
+    while (blkCnt > 0U)
+    {
+
+        vecSrc = vld2q(pSrc);
+        vecSrc.val[1] = vqsubq(zero, vecSrc.val[1]);
+        vst2q(pDst,vecSrc);
+        /*
+         * Decrement the blkCnt loop counter
+         * Advance vector source and destination pointers
+         */
+        pSrc += 8;
+        pDst += 8;
+        blkCnt --;
+    }
+
+     /* Tail */
+    blkCnt = (blockSize & 0x7) >> 1;
+
+    while (blkCnt > 0U)
+    {
+      /* C[0] + jC[1] = A[0]+ j(-1)A[1] */
+  
+      /* Calculate Complex Conjugate and store result in destination buffer. */
+      *pDst++ =  *pSrc++;
+      in = *pSrc++;
+      *pDst++ = __QSUB(0, in);
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+
+
+}
+#else
+
+void arm_cmplx_conj_q31(
+  const q31_t * pSrc,
+        q31_t * pDst,
+        uint32_t numSamples)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q31_t in;                                      /* Temporary input variable */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] + jC[1] = A[0]+ j(-1)A[1] */
+
+    /* Calculate Complex Conjugate and store result in destination buffer. */
+    *pDst++ =  *pSrc++;
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QSUB(0, in);
+#else
+    *pDst++ = (in == INT32_MIN) ? INT32_MAX : -in;
+#endif
+
+    *pDst++ =  *pSrc++;
+    in =  *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QSUB(0, in);
+#else
+    *pDst++ = (in == INT32_MIN) ? INT32_MAX : -in;
+#endif
+
+    *pDst++ =  *pSrc++;
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QSUB(0, in);
+#else
+    *pDst++ = (in == INT32_MIN) ? INT32_MAX : -in;
+#endif
+
+    *pDst++ =  *pSrc++;
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QSUB(0, in);
+#else
+    *pDst++ = (in == INT32_MIN) ? INT32_MAX : -in;
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] + jC[1] = A[0]+ j(-1)A[1] */
+
+    /* Calculate Complex Conjugate and store result in destination buffer. */
+    *pDst++ =  *pSrc++;
+    in = *pSrc++;
+#if defined (ARM_MATH_DSP)
+    *pDst++ = __QSUB(0, in);
+#else
+    *pDst++ = (in == INT32_MIN) ? INT32_MAX : -in;
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of cmplx_conj group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_dot_prod_f32.c

@@ -0,0 +1,302 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_dot_prod_f32.c
+ * Description:  Floating-point complex dot product
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @defgroup cmplx_dot_prod Complex Dot Product
+
+  Computes the dot product of two complex vectors.
+  The vectors are multiplied element-by-element and then summed.
+
+  The <code>pSrcA</code> points to the first complex input vector and
+  <code>pSrcB</code> points to the second complex input vector.
+  <code>numSamples</code> specifies the number of complex samples
+  and the data in each array is stored in an interleaved fashion
+  (real, imag, real, imag, ...).
+  Each array has a total of <code>2*numSamples</code> values.
+
+  The underlying algorithm is used:
+
+  <pre>
+  realResult = 0;
+  imagResult = 0;
+  for (n = 0; n < numSamples; n++) {
+      realResult += pSrcA[(2*n)+0] * pSrcB[(2*n)+0] - pSrcA[(2*n)+1] * pSrcB[(2*n)+1];
+      imagResult += pSrcA[(2*n)+0] * pSrcB[(2*n)+1] + pSrcA[(2*n)+1] * pSrcB[(2*n)+0];
+  }
+  </pre>
+
+  There are separate functions for floating-point, Q15, and Q31 data types.
+ */
+
+/**
+  @addtogroup cmplx_dot_prod
+  @{
+ */
+
+/**
+  @brief         Floating-point complex dot product.
+  @param[in]     pSrcA       points to the first input vector
+  @param[in]     pSrcB       points to the second input vector
+  @param[in]     numSamples  number of samples in each vector
+  @param[out]    realResult  real part of the result returned here
+  @param[out]    imagResult  imaginary part of the result returned here
+  @return        none
+ */
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+void arm_cmplx_dot_prod_f32(
+    const float32_t * pSrcA,
+    const float32_t * pSrcB,
+    uint32_t numSamples,
+    float32_t * realResult,
+    float32_t * imagResult)
+{
+    uint32_t blockSize = numSamples * CMPLX_DIM;  /* loop counters */
+    uint32_t blkCnt;
+    float32_t real_sum, imag_sum;
+    f32x4_t vecSrcA, vecSrcB;
+    f32x4_t vec_acc = vdupq_n_f32(0.0f);
+    float32_t a0,b0,c0,d0;
+
+    /* Compute 2 complex samples at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+
+        vec_acc = vcmlaq(vec_acc, vecSrcA, vecSrcB);
+        vec_acc = vcmlaq_rot90(vec_acc, vecSrcA, vecSrcB);
+
+        /*
+         * Decrement the blkCnt loop counter
+         * Advance vector source and destination pointers
+         */
+        pSrcA += 4;
+        pSrcB += 4;
+        blkCnt--;
+    }
+
+
+    real_sum = vgetq_lane(vec_acc, 0) + vgetq_lane(vec_acc, 2);
+    imag_sum = vgetq_lane(vec_acc, 1) + vgetq_lane(vec_acc, 3);
+   
+    /* Tail */
+    blkCnt = (blockSize & 3) >> 1;
+
+    while (blkCnt > 0U)
+    {
+      a0 = *pSrcA++;
+      b0 = *pSrcA++;
+      c0 = *pSrcB++;
+      d0 = *pSrcB++;
+  
+      real_sum += a0 * c0;
+      imag_sum += a0 * d0;
+      real_sum -= b0 * d0;
+      imag_sum += b0 * c0;
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+
+
+    /*
+     * Store the real and imaginary results in the destination buffers
+     */
+    *realResult = real_sum;
+    *imagResult = imag_sum;
+}
+
+#else
+void arm_cmplx_dot_prod_f32(
+  const float32_t * pSrcA,
+  const float32_t * pSrcB,
+        uint32_t numSamples,
+        float32_t * realResult,
+        float32_t * imagResult)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        float32_t real_sum = 0.0f, imag_sum = 0.0f;    /* Temporary result variables */
+        float32_t a0,b0,c0,d0;
+
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    float32x4x2_t vec1,vec2,vec3,vec4;
+    float32x4_t accR,accI;
+    float32x2_t accum = vdup_n_f32(0);
+
+    accR = vdupq_n_f32(0.0f);
+    accI = vdupq_n_f32(0.0f);
+
+    /* Loop unrolling: Compute 8 outputs at a time */
+    blkCnt = numSamples >> 3U;
+
+    while (blkCnt > 0U)
+    {
+	/* C = (A[0]+jA[1])*(B[0]+jB[1]) + ...  */
+        /* Calculate dot product and then store the result in a temporary buffer. */
+
+	      vec1 = vld2q_f32(pSrcA);
+        vec2 = vld2q_f32(pSrcB);
+
+	/* Increment pointers */
+        pSrcA += 8;
+        pSrcB += 8;
+
+	/* Re{C} = Re{A}*Re{B} - Im{A}*Im{B} */
+        accR = vmlaq_f32(accR,vec1.val[0],vec2.val[0]);
+        accR = vmlsq_f32(accR,vec1.val[1],vec2.val[1]);
+
+	/* Im{C} = Re{A}*Im{B} + Im{A}*Re{B} */
+        accI = vmlaq_f32(accI,vec1.val[1],vec2.val[0]);
+        accI = vmlaq_f32(accI,vec1.val[0],vec2.val[1]);
+
+        vec3 = vld2q_f32(pSrcA);
+        vec4 = vld2q_f32(pSrcB);
+	
+	/* Increment pointers */
+        pSrcA += 8;
+        pSrcB += 8;
+
+	/* Re{C} = Re{A}*Re{B} - Im{A}*Im{B} */
+        accR = vmlaq_f32(accR,vec3.val[0],vec4.val[0]);
+        accR = vmlsq_f32(accR,vec3.val[1],vec4.val[1]);
+
+	/* Im{C} = Re{A}*Im{B} + Im{A}*Re{B} */
+        accI = vmlaq_f32(accI,vec3.val[1],vec4.val[0]);
+        accI = vmlaq_f32(accI,vec3.val[0],vec4.val[1]);
+
+        /* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    accum = vpadd_f32(vget_low_f32(accR), vget_high_f32(accR));
+    real_sum += vget_lane_f32(accum, 0) + vget_lane_f32(accum, 1);
+
+    accum = vpadd_f32(vget_low_f32(accI), vget_high_f32(accI));
+    imag_sum += vget_lane_f32(accum, 0) + vget_lane_f32(accum, 1);
+
+    /* Tail */
+    blkCnt = numSamples & 0x7;
+
+#else
+#if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += a0 * c0;
+    imag_sum += a0 * d0;
+    real_sum -= b0 * d0;
+    imag_sum += b0 * c0;
+
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += a0 * c0;
+    imag_sum += a0 * d0;
+    real_sum -= b0 * d0;
+    imag_sum += b0 * c0;
+
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += a0 * c0;
+    imag_sum += a0 * d0;
+    real_sum -= b0 * d0;
+    imag_sum += b0 * c0;
+
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += a0 * c0;
+    imag_sum += a0 * d0;
+    real_sum -= b0 * d0;
+    imag_sum += b0 * c0;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined(ARM_MATH_NEON) */
+
+  while (blkCnt > 0U)
+  {
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += a0 * c0;
+    imag_sum += a0 * d0;
+    real_sum -= b0 * d0;
+    imag_sum += b0 * c0;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Store real and imaginary result in destination buffer. */
+  *realResult = real_sum;
+  *imagResult = imag_sum;
+}
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+/**
+  @} end of cmplx_dot_prod group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_dot_prod_q15.c

@@ -0,0 +1,234 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_dot_prod_q15.c
+ * Description:  Processing function for the Q15 Complex Dot product
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @addtogroup cmplx_dot_prod
+  @{
+ */
+
+/**
+  @brief         Q15 complex dot product.
+  @param[in]     pSrcA       points to the first input vector
+  @param[in]     pSrcB       points to the second input vector
+  @param[in]     numSamples  number of samples in each vector
+  @param[out]    realResult  real part of the result returned here
+  @param[out]    imagResult  imaginary part of the result returned her
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function is implemented using an internal 64-bit accumulator.
+                   The intermediate 1.15 by 1.15 multiplications are performed with full precision and yield a 2.30 result.
+                   These are accumulated in a 64-bit accumulator with 34.30 precision.
+                   As a final step, the accumulators are converted to 8.24 format.
+                   The return results <code>realResult</code> and <code>imagResult</code> are in 8.24 format.
+ */
+
+#if defined(ARM_MATH_MVEI)
+void arm_cmplx_dot_prod_q15(
+  const q15_t * pSrcA,
+  const q15_t * pSrcB,
+        uint32_t numSamples,
+        q31_t * realResult,
+        q31_t * imagResult)
+{
+
+  uint32_t blockSize = numSamples * CMPLX_DIM;  /* loop counters */
+  uint32_t blkCnt;
+  q15_t a0,b0,c0,d0;
+
+  q63_t accReal = 0LL; q63_t accImag = 0LL;
+  q15x8_t vecSrcA, vecSrcB;
+
+
+
+  /* should give more freedom to generate stall free code */
+  vecSrcA = vld1q(pSrcA);
+  vecSrcB = vld1q(pSrcB);
+  pSrcA += 8;
+  pSrcB += 8;
+
+  /* Compute 4 complex samples at a time */
+  blkCnt = blockSize >> 3;
+  while (blkCnt > 0U) 
+  {
+      q15x8_t vecSrcC, vecSrcD;
+
+      accReal = vmlsldavaq(accReal, vecSrcA, vecSrcB);
+      vecSrcC = vld1q(pSrcA);
+      pSrcA += 8;
+
+      accImag = vmlaldavaxq(accImag, vecSrcA, vecSrcB);
+      vecSrcD = vld1q(pSrcB);
+      pSrcB += 8;
+
+      accReal = vmlsldavaq(accReal, vecSrcC, vecSrcD);
+      vecSrcA = vld1q(pSrcA);
+      pSrcA += 8;
+
+      accImag = vmlaldavaxq(accImag, vecSrcC, vecSrcD);
+      vecSrcB = vld1q(pSrcB);
+      pSrcB += 8;
+      /*
+       * Decrement the blockSize loop counter
+       */
+      blkCnt--;
+  }
+
+  /* Tail */
+  pSrcA -= 8;
+  pSrcB -= 8; 
+
+  blkCnt = (blockSize & 7) >> 1;
+  
+  while (blkCnt > 0U)
+  {
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    accReal += (q31_t)a0 * c0;
+    accImag += (q31_t)a0 * d0;
+    accReal -= (q31_t)b0 * d0;
+    accImag += (q31_t)b0 * c0;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Store real and imaginary result in 8.24 format  */
+  /* Convert real data in 34.30 to 8.24 by 6 right shifts */
+  *realResult = (q31_t) (accReal >> 6);
+  /* Convert imaginary data in 34.30 to 8.24 by 6 right shifts */
+  *imagResult = (q31_t) (accImag >> 6);
+}
+#else
+void arm_cmplx_dot_prod_q15(
+  const q15_t * pSrcA,
+  const q15_t * pSrcB,
+        uint32_t numSamples,
+        q31_t * realResult,
+        q31_t * imagResult)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q63_t real_sum = 0, imag_sum = 0;              /* Temporary result variables */
+        q15_t a0,b0,c0,d0;
+
+#if defined (ARM_MATH_LOOPUNROLL)
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += (q31_t)a0 * c0;
+    imag_sum += (q31_t)a0 * d0;
+    real_sum -= (q31_t)b0 * d0;
+    imag_sum += (q31_t)b0 * c0;
+
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += (q31_t)a0 * c0;
+    imag_sum += (q31_t)a0 * d0;
+    real_sum -= (q31_t)b0 * d0;
+    imag_sum += (q31_t)b0 * c0;
+
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += (q31_t)a0 * c0;
+    imag_sum += (q31_t)a0 * d0;
+    real_sum -= (q31_t)b0 * d0;
+    imag_sum += (q31_t)b0 * c0;
+
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += (q31_t)a0 * c0;
+    imag_sum += (q31_t)a0 * d0;
+    real_sum -= (q31_t)b0 * d0;
+    imag_sum += (q31_t)b0 * c0;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += (q31_t)a0 * c0;
+    imag_sum += (q31_t)a0 * d0;
+    real_sum -= (q31_t)b0 * d0;
+    imag_sum += (q31_t)b0 * c0;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Store real and imaginary result in 8.24 format  */
+  /* Convert real data in 34.30 to 8.24 by 6 right shifts */
+  *realResult = (q31_t) (real_sum >> 6);
+  /* Convert imaginary data in 34.30 to 8.24 by 6 right shifts */
+  *imagResult = (q31_t) (imag_sum >> 6);
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of cmplx_dot_prod group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_dot_prod_q31.c

@@ -0,0 +1,220 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_dot_prod_q31.c
+ * Description:  Q31 complex dot product
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @addtogroup cmplx_dot_prod
+  @{
+ */
+
+/**
+  @brief         Q31 complex dot product.
+  @param[in]     pSrcA       points to the first input vector
+  @param[in]     pSrcB       points to the second input vector
+  @param[in]     numSamples  number of samples in each vector
+  @param[out]    realResult  real part of the result returned here
+  @param[out]    imagResult  imaginary part of the result returned here
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function is implemented using an internal 64-bit accumulator.
+                   The intermediate 1.31 by 1.31 multiplications are performed with 64-bit precision and then shifted to 16.48 format.
+                   The internal real and imaginary accumulators are in 16.48 format and provide 15 guard bits.
+                   Additions are nonsaturating and no overflow will occur as long as <code>numSamples</code> is less than 32768.
+                   The return results <code>realResult</code> and <code>imagResult</code> are in 16.48 format.
+                   Input down scaling is not required.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+void arm_cmplx_dot_prod_q31(
+  const q31_t * pSrcA,
+  const q31_t * pSrcB,
+        uint32_t numSamples,
+        q63_t * realResult,
+        q63_t * imagResult)
+{
+    uint32_t blockSize = numSamples * CMPLX_DIM;  /* loop counters */
+    uint32_t blkCnt;
+    q31x4_t vecSrcA, vecSrcB;
+    q63_t accReal = 0LL; 
+    q63_t accImag = 0LL;
+
+    q31_t a0,b0,c0,d0;
+
+     /* Compute 2 complex samples at a time */
+    blkCnt = blockSize >> 2U;
+
+    while (blkCnt > 0U)
+    {        
+
+        vecSrcA = vld1q(pSrcA);
+        vecSrcB = vld1q(pSrcB);
+
+        accReal = vrmlsldavhaq(accReal, vecSrcA, vecSrcB);
+        accImag = vrmlaldavhaxq(accImag, vecSrcA, vecSrcB);
+
+        /*
+         * Decrement the blkCnt loop counter
+         * Advance vector source and destination pointers
+         */
+        pSrcA += 4;
+        pSrcB += 4;
+        blkCnt --;
+    }
+
+    accReal = asrl(accReal, (14 - 8));
+    accImag = asrl(accImag, (14 - 8));
+
+    /* Tail */
+    blkCnt = (blockSize & 3) >> 1;
+
+    while (blkCnt > 0U)
+    {
+      a0 = *pSrcA++;
+      b0 = *pSrcA++;
+      c0 = *pSrcB++;
+      d0 = *pSrcB++;
+  
+      accReal += ((q63_t)a0 * c0) >> 14;
+      accImag += ((q63_t)a0 * d0) >> 14;
+      accReal -= ((q63_t)b0 * d0) >> 14;
+      accImag += ((q63_t)b0 * c0) >> 14;
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+  
+    /* Store real and imaginary result in destination buffer. */
+    *realResult = accReal;
+    *imagResult = accImag;
+}
+
+#else
+void arm_cmplx_dot_prod_q31(
+  const q31_t * pSrcA,
+  const q31_t * pSrcB,
+        uint32_t numSamples,
+        q63_t * realResult,
+        q63_t * imagResult)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q63_t real_sum = 0, imag_sum = 0;              /* Temporary result variables */
+        q31_t a0,b0,c0,d0;
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += ((q63_t)a0 * c0) >> 14;
+    imag_sum += ((q63_t)a0 * d0) >> 14;
+    real_sum -= ((q63_t)b0 * d0) >> 14;
+    imag_sum += ((q63_t)b0 * c0) >> 14;
+
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += ((q63_t)a0 * c0) >> 14;
+    imag_sum += ((q63_t)a0 * d0) >> 14;
+    real_sum -= ((q63_t)b0 * d0) >> 14;
+    imag_sum += ((q63_t)b0 * c0) >> 14;
+
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += ((q63_t)a0 * c0) >> 14;
+    imag_sum += ((q63_t)a0 * d0) >> 14;
+    real_sum -= ((q63_t)b0 * d0) >> 14;
+    imag_sum += ((q63_t)b0 * c0) >> 14;
+
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += ((q63_t)a0 * c0) >> 14;
+    imag_sum += ((q63_t)a0 * d0) >> 14;
+    real_sum -= ((q63_t)b0 * d0) >> 14;
+    imag_sum += ((q63_t)b0 * c0) >> 14;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    a0 = *pSrcA++;
+    b0 = *pSrcA++;
+    c0 = *pSrcB++;
+    d0 = *pSrcB++;
+
+    real_sum += ((q63_t)a0 * c0) >> 14;
+    imag_sum += ((q63_t)a0 * d0) >> 14;
+    real_sum -= ((q63_t)b0 * d0) >> 14;
+    imag_sum += ((q63_t)b0 * c0) >> 14;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Store real and imaginary result in 16.48 format  */
+  *realResult = real_sum;
+  *imagResult = imag_sum;
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of cmplx_dot_prod group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_mag_f32.c

@@ -0,0 +1,273 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_mag_f32.c
+ * Description:  Floating-point complex magnitude
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @defgroup cmplx_mag Complex Magnitude
+
+  Computes the magnitude of the elements of a complex data vector.
+
+  The <code>pSrc</code> points to the source data and
+  <code>pDst</code> points to the where the result should be written.
+  <code>numSamples</code> specifies the number of complex samples
+  in the input array and the data is stored in an interleaved fashion
+  (real, imag, real, imag, ...).
+  The input array has a total of <code>2*numSamples</code> values;
+  the output array has a total of <code>numSamples</code> values.
+
+  The underlying algorithm is used:
+
+  <pre>
+  for (n = 0; n < numSamples; n++) {
+      pDst[n] = sqrt(pSrc[(2*n)+0]^2 + pSrc[(2*n)+1]^2);
+  }
+  </pre>
+
+  There are separate functions for floating-point, Q15, and Q31 data types.
+ */
+
+/**
+  @addtogroup cmplx_mag
+  @{
+ */
+
+/**
+  @brief         Floating-point complex magnitude.
+  @param[in]     pSrc        points to input vector
+  @param[out]    pDst        points to output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+ */
+
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+#include "arm_vec_math.h"
+#endif
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+#include "arm_helium_utils.h"
+
+
+void arm_cmplx_mag_f32(
+  const float32_t * pSrc,
+        float32_t * pDst,
+        uint32_t numSamples)
+{
+    int32_t blockSize = numSamples;  /* loop counters */
+    uint32_t  blkCnt;           /* loop counters */
+    f32x4x2_t vecSrc;
+    f32x4_t sum;
+    float32_t real, imag;                      /* Temporary variables to hold input values */
+
+    /* Compute 4 complex samples at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+        q31x4_t newtonStartVec;
+        f32x4_t sumHalf, invSqrt;
+
+        vecSrc = vld2q(pSrc);  
+        pSrc += 8;
+        sum = vmulq(vecSrc.val[0], vecSrc.val[0]);
+        sum = vfmaq(sum, vecSrc.val[1], vecSrc.val[1]);
+
+        /*
+         * inlined Fast SQRT using inverse SQRT newton-raphson method
+         */
+
+        /* compute initial value */
+        newtonStartVec = vdupq_n_s32(INVSQRT_MAGIC_F32) - vshrq((q31x4_t) sum, 1);
+        sumHalf = sum * 0.5f;
+        /*
+         * compute 3 x iterations
+         *
+         * The more iterations, the more accuracy.
+         * If you need to trade a bit of accuracy for more performance,
+         * you can comment out the 3rd use of the macro.
+         */
+        INVSQRT_NEWTON_MVE_F32(invSqrt, sumHalf, (f32x4_t) newtonStartVec);
+        INVSQRT_NEWTON_MVE_F32(invSqrt, sumHalf, invSqrt);
+        INVSQRT_NEWTON_MVE_F32(invSqrt, sumHalf, invSqrt);
+        /*
+         * set negative values to 0
+         */
+        invSqrt = vdupq_m(invSqrt, 0.0f, vcmpltq(invSqrt, 0.0f));
+        /*
+         * sqrt(x) = x * invSqrt(x)
+         */
+        sum = vmulq(sum, invSqrt);
+        vst1q(pDst, sum); 
+        pDst += 4;
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+    }
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 3;
+    while (blkCnt > 0U)
+    {
+      /* C[0] = sqrt(A[0] * A[0] + A[1] * A[1]) */
+  
+      real = *pSrc++;
+      imag = *pSrc++;
+  
+      /* store result in destination buffer. */
+      arm_sqrt_f32((real * real) + (imag * imag), pDst++);
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+}
+
+#else
+void arm_cmplx_mag_f32(
+  const float32_t * pSrc,
+        float32_t * pDst,
+        uint32_t numSamples)
+{
+  uint32_t blkCnt;                               /* loop counter */
+  float32_t real, imag;                      /* Temporary variables to hold input values */
+
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  float32x4x2_t vecA;
+  float32x4_t vRealA;
+  float32x4_t vImagA;
+  float32x4_t vMagSqA;
+
+  float32x4x2_t vecB;
+  float32x4_t vRealB;
+  float32x4_t vImagB;
+  float32x4_t vMagSqB;
+
+  /* Loop unrolling: Compute 8 outputs at a time */
+  blkCnt = numSamples >> 3;
+
+  while (blkCnt > 0U)
+  {
+    /* out = sqrt((real * real) + (imag * imag)) */
+
+    vecA = vld2q_f32(pSrc);
+    pSrc += 8;
+
+    vecB = vld2q_f32(pSrc);
+    pSrc += 8;
+
+    vRealA = vmulq_f32(vecA.val[0], vecA.val[0]);
+    vImagA = vmulq_f32(vecA.val[1], vecA.val[1]);
+    vMagSqA = vaddq_f32(vRealA, vImagA);
+
+    vRealB = vmulq_f32(vecB.val[0], vecB.val[0]);
+    vImagB = vmulq_f32(vecB.val[1], vecB.val[1]);
+    vMagSqB = vaddq_f32(vRealB, vImagB);
+
+    /* Store the result in the destination buffer. */
+    vst1q_f32(pDst, __arm_vec_sqrt_f32_neon(vMagSqA));
+    pDst += 4;
+
+    vst1q_f32(pDst, __arm_vec_sqrt_f32_neon(vMagSqB));
+    pDst += 4;
+
+    /* Decrement the loop counter */
+    blkCnt--;
+  }
+
+  blkCnt = numSamples & 7;
+
+#else
+
+#if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] = sqrt(A[0] * A[0] + A[1] * A[1]) */
+
+    real = *pSrc++;
+    imag = *pSrc++;
+
+    /* store result in destination buffer. */
+    arm_sqrt_f32((real * real) + (imag * imag), pDst++);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    arm_sqrt_f32((real * real) + (imag * imag), pDst++);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    arm_sqrt_f32((real * real) + (imag * imag), pDst++);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    arm_sqrt_f32((real * real) + (imag * imag), pDst++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined(ARM_MATH_NEON) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] = sqrt(A[0] * A[0] + A[1] * A[1]) */
+
+    real = *pSrc++;
+    imag = *pSrc++;
+
+    /* store result in destination buffer. */
+    arm_sqrt_f32((real * real) + (imag * imag), pDst++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+/**
+  @} end of cmplx_mag group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_mag_q15.c

@@ -0,0 +1,221 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_mag_q15.c
+ * Description:  Q15 complex magnitude
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @addtogroup cmplx_mag
+  @{
+ */
+
+/**
+  @brief         Q15 complex magnitude.
+  @param[in]     pSrc        points to input vector
+  @param[out]    pDst        points to output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function implements 1.15 by 1.15 multiplications and finally output is converted into 2.14 format.
+ */
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_cmplx_mag_q15(
+  const q15_t * pSrc,
+        q15_t * pDst,
+        uint32_t numSamples)
+{
+
+    int32_t blockSize = numSamples;  /* loop counters */
+    uint32_t  blkCnt;           /* loop counters */
+    q15x8x2_t vecSrc;
+    q15x8_t sum;
+    q31_t in;
+    q31_t acc0;
+
+    blkCnt = blockSize >> 3;
+    while (blkCnt > 0U)
+    {
+        vecSrc = vld2q(pSrc);  
+        pSrc += 16;
+        sum = vqaddq(vmulhq(vecSrc.val[0], vecSrc.val[0]),
+                     vmulhq(vecSrc.val[1], vecSrc.val[1]));
+
+        sum = vshrq(sum, 1);
+
+        sum = FAST_VSQRT_Q15(sum);
+
+        vst1q(pDst, sum); 
+        pDst += 8;
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+    }
+
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 7;
+
+    while (blkCnt > 0U)
+    {
+      /* C[0] = sqrt(A[0] * A[0] + A[1] * A[1]) */
+  
+      in = read_q15x2_ia ((q15_t **) &pSrc);
+      acc0 = __SMUAD(in, in);
+  
+      /* store result in 2.14 format in destination buffer. */
+      arm_sqrt_q15((q15_t) (acc0 >> 17), pDst++);
+  
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+}
+
+#else
+void arm_cmplx_mag_q15(
+  const q15_t * pSrc,
+        q15_t * pDst,
+        uint32_t numSamples)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined (ARM_MATH_DSP)
+        q31_t in;
+        q31_t acc0;                                    /* Accumulators */
+#else
+       q15_t real, imag;                              /* Temporary input variables */
+       q31_t acc0, acc1;                              /* Accumulators */
+#endif
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] = sqrt(A[0] * A[0] + A[1] * A[1]) */
+
+#if defined (ARM_MATH_DSP)
+    in = read_q15x2_ia ((q15_t **) &pSrc);
+    acc0 = __SMUAD(in, in);
+    /* store result in 2.14 format in destination buffer. */
+    arm_sqrt_q15((q15_t) (acc0 >> 17), pDst++);
+
+    in = read_q15x2_ia ((q15_t **) &pSrc);
+    acc0 = __SMUAD(in, in);
+    arm_sqrt_q15((q15_t) (acc0 >> 17), pDst++);
+
+    in = read_q15x2_ia ((q15_t **) &pSrc);
+    acc0 = __SMUAD(in, in);
+    arm_sqrt_q15((q15_t) (acc0 >> 17), pDst++);
+
+    in = read_q15x2_ia ((q15_t **) &pSrc);
+    acc0 = __SMUAD(in, in);
+    arm_sqrt_q15((q15_t) (acc0 >> 17), pDst++);
+#else
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = ((q31_t) real * real);
+    acc1 = ((q31_t) imag * imag);
+
+    /* store result in 2.14 format in destination buffer. */
+    arm_sqrt_q15((q15_t) (((q63_t) acc0 + acc1) >> 17), pDst++);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = ((q31_t) real * real);
+    acc1 = ((q31_t) imag * imag);
+    arm_sqrt_q15((q15_t) (((q63_t) acc0 + acc1) >> 17), pDst++);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = ((q31_t) real * real);
+    acc1 = ((q31_t) imag * imag);
+    arm_sqrt_q15((q15_t) (((q63_t) acc0 + acc1) >> 17), pDst++);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = ((q31_t) real * real);
+    acc1 = ((q31_t) imag * imag);
+    arm_sqrt_q15((q15_t) (((q63_t) acc0 + acc1) >> 17), pDst++);
+#endif /* #if defined (ARM_MATH_DSP) */
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] = sqrt(A[0] * A[0] + A[1] * A[1]) */
+
+#if defined (ARM_MATH_DSP)
+    in = read_q15x2_ia ((q15_t **) &pSrc);
+    acc0 = __SMUAD(in, in);
+
+    /* store result in 2.14 format in destination buffer. */
+    arm_sqrt_q15((q15_t) (acc0 >> 17), pDst++);
+#else
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = ((q31_t) real * real);
+    acc1 = ((q31_t) imag * imag);
+
+    /* store result in 2.14 format in destination buffer. */
+    arm_sqrt_q15((q15_t) (((q63_t) acc0 + acc1) >> 17), pDst++);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of cmplx_mag group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_mag_q31.c

@@ -0,0 +1,201 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_mag_q31.c
+ * Description:  Q31 complex magnitude
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @addtogroup cmplx_mag
+  @{
+ */
+
+/**
+  @brief         Q31 complex magnitude.
+  @param[in]     pSrc        points to input vector
+  @param[out]    pDst        points to output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function implements 1.31 by 1.31 multiplications and finally output is converted into 2.30 format.
+                   Input down scaling is not required.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+#include "arm_helium_utils.h"
+
+void arm_cmplx_mag_q31(
+  const q31_t * pSrc,
+        q31_t * pDst,
+        uint32_t numSamples)
+{
+    int32_t blockSize = numSamples;  /* loop counters */
+    uint32_t  blkCnt;           /* loop counters */
+
+    q31x4x2_t vecSrc;
+    q31x4_t sum;
+
+    q31_t real, imag;                              /* Temporary input variables */
+    q31_t acc0, acc1;                              /* Accumulators */
+
+    /* Compute 4 complex samples at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+        vecSrc = vld2q(pSrc);
+
+        sum = vqaddq(vmulhq(vecSrc.val[0], vecSrc.val[0]),
+                     vmulhq(vecSrc.val[1], vecSrc.val[1]));
+
+        sum = vshrq(sum, 1);
+
+        /*
+
+        This function is using a table. There are compilations flags to avoid
+        including this table (and in this case, arm_cmplx_maq_q31 must not
+        be built and linked.)
+
+        */
+        sum = FAST_VSQRT_Q31(sum);
+
+        vst1q(pDst, sum);
+
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+        pSrc += 8;
+        pDst += 4;
+    }
+
+    /*
+     * tail
+     */
+    blkCnt = blockSize & 3;
+    while (blkCnt > 0U)
+    {
+      /* C[0] = sqrt(A[0] * A[0] + A[1] * A[1]) */
+  
+      real = *pSrc++;
+      imag = *pSrc++;
+      acc0 = (q31_t) (((q63_t) real * real) >> 33);
+      acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
+  
+      /* store result in 2.30 format in destination buffer. */
+      arm_sqrt_q31(acc0 + acc1, pDst++);
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+}
+
+#else
+void arm_cmplx_mag_q31(
+  const q31_t * pSrc,
+        q31_t * pDst,
+        uint32_t numSamples)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q31_t real, imag;                              /* Temporary input variables */
+        q31_t acc0, acc1;                              /* Accumulators */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] = sqrt(A[0] * A[0] + A[1] * A[1]) */
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = (q31_t) (((q63_t) real * real) >> 33);
+    acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
+
+    /* store result in 2.30 format in destination buffer. */
+    arm_sqrt_q31(acc0 + acc1, pDst++);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = (q31_t) (((q63_t) real * real) >> 33);
+    acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
+    arm_sqrt_q31(acc0 + acc1, pDst++);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = (q31_t) (((q63_t) real * real) >> 33);
+    acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
+    arm_sqrt_q31(acc0 + acc1, pDst++);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = (q31_t) (((q63_t) real * real) >> 33);
+    acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
+    arm_sqrt_q31(acc0 + acc1, pDst++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] = sqrt(A[0] * A[0] + A[1] * A[1]) */
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = (q31_t) (((q63_t) real * real) >> 33);
+    acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
+
+    /* store result in 2.30 format in destination buffer. */
+    arm_sqrt_q31(acc0 + acc1, pDst++);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of cmplx_mag group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_mag_squared_f32.c

@@ -0,0 +1,235 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_mag_squared_f32.c
+ * Description:  Floating-point complex magnitude squared
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @defgroup cmplx_mag_squared Complex Magnitude Squared
+
+  Computes the magnitude squared of the elements of a complex data vector.
+
+  The <code>pSrc</code> points to the source data and
+  <code>pDst</code> points to the where the result should be written.
+  <code>numSamples</code> specifies the number of complex samples
+  in the input array and the data is stored in an interleaved fashion
+  (real, imag, real, imag, ...).
+  The input array has a total of <code>2*numSamples</code> values;
+  the output array has a total of <code>numSamples</code> values.
+
+  The underlying algorithm is used:
+
+  <pre>
+  for (n = 0; n < numSamples; n++) {
+      pDst[n] = pSrc[(2*n)+0]^2 + pSrc[(2*n)+1]^2;
+  }
+  </pre>
+
+  There are separate functions for floating-point, Q15, and Q31 data types.
+ */
+
+/**
+  @addtogroup cmplx_mag_squared
+  @{
+ */
+
+/**
+  @brief         Floating-point complex magnitude squared.
+  @param[in]     pSrc        points to input vector
+  @param[out]    pDst        points to output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+ */
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+void arm_cmplx_mag_squared_f32(
+  const float32_t * pSrc,
+        float32_t * pDst,
+        uint32_t numSamples)
+{
+    int32_t blockSize = numSamples;  /* loop counters */
+    uint32_t  blkCnt;           /* loop counters */
+    f32x4x2_t vecSrc;
+    f32x4_t sum;
+    float32_t real, imag;                          /* Temporary input variables */
+
+    /* Compute 4 complex samples at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+        vecSrc = vld2q(pSrc);
+        sum = vmulq(vecSrc.val[0], vecSrc.val[0]);
+        sum = vfmaq(sum, vecSrc.val[1], vecSrc.val[1]);
+        vst1q(pDst, sum);
+
+        pSrc += 8;
+        pDst += 4;
+        
+        /*
+         * Decrement the blockSize loop counter
+         */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 3;
+    while (blkCnt > 0U)
+    {
+      /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
+  
+      real = *pSrc++;
+      imag = *pSrc++;
+  
+      /* store result in destination buffer. */
+      *pDst++ = (real * real) + (imag * imag);
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+
+}
+
+#else
+void arm_cmplx_mag_squared_f32(
+  const float32_t * pSrc,
+        float32_t * pDst,
+        uint32_t numSamples)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        float32_t real, imag;                          /* Temporary input variables */
+
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+  float32x4x2_t vecA;
+  float32x4_t vRealA;
+  float32x4_t vImagA;
+  float32x4_t vMagSqA;
+
+  float32x4x2_t vecB;
+  float32x4_t vRealB;
+  float32x4_t vImagB;
+  float32x4_t vMagSqB;
+
+  /* Loop unrolling: Compute 8 outputs at a time */
+  blkCnt = numSamples >> 3;
+
+  while (blkCnt > 0U)
+  {
+    /* out = sqrt((real * real) + (imag * imag)) */
+
+    vecA = vld2q_f32(pSrc);
+    pSrc += 8;
+
+    vRealA = vmulq_f32(vecA.val[0], vecA.val[0]);
+    vImagA = vmulq_f32(vecA.val[1], vecA.val[1]);
+    vMagSqA = vaddq_f32(vRealA, vImagA);
+
+    vecB = vld2q_f32(pSrc);
+    pSrc += 8;
+
+    vRealB = vmulq_f32(vecB.val[0], vecB.val[0]);
+    vImagB = vmulq_f32(vecB.val[1], vecB.val[1]);
+    vMagSqB = vaddq_f32(vRealB, vImagB);
+
+    /* Store the result in the destination buffer. */
+    vst1q_f32(pDst, vMagSqA);
+    pDst += 4;
+
+    vst1q_f32(pDst, vMagSqB);
+    pDst += 4;
+
+    /* Decrement the loop counter */
+    blkCnt--;
+  }
+
+  blkCnt = numSamples & 7;
+
+#else
+#if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    *pDst++ = (real * real) + (imag * imag);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    *pDst++ = (real * real) + (imag * imag);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    *pDst++ = (real * real) + (imag * imag);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    *pDst++ = (real * real) + (imag * imag);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined(ARM_MATH_NEON) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
+
+    real = *pSrc++;
+    imag = *pSrc++;
+
+    /* store result in destination buffer. */
+    *pDst++ = (real * real) + (imag * imag);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+/**
+  @} end of cmplx_mag_squared group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_mag_squared_q15.c

@@ -0,0 +1,221 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_mag_squared_q15.c
+ * Description:  Q15 complex magnitude squared
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @addtogroup cmplx_mag_squared
+  @{
+ */
+
+/**
+  @brief         Q15 complex magnitude squared.
+  @param[in]     pSrc        points to input vector
+  @param[out]    pDst        points to output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function implements 1.15 by 1.15 multiplications and finally output is converted into 3.13 format.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+void arm_cmplx_mag_squared_q15(
+  const q15_t * pSrc,
+        q15_t * pDst,
+        uint32_t numSamples)
+{
+  int32_t blockSize = numSamples;  /* loop counters */
+  uint32_t  blkCnt;           /* loop counters */
+  q31_t in;
+  q31_t acc0;                                    /* Accumulators */
+  q15x8x2_t vecSrc;
+  q15x8_t vReal, vImag;
+  q15x8_t vMagSq;
+
+  
+  blkCnt = blockSize >> 3;
+  while (blkCnt > 0U)
+  {
+    vecSrc = vld2q(pSrc);
+    vReal = vmulhq(vecSrc.val[0], vecSrc.val[0]);
+    vImag = vmulhq(vecSrc.val[1], vecSrc.val[1]);
+    vMagSq = vqaddq(vReal, vImag);
+    vMagSq = vshrq(vMagSq, 1);
+
+    vst1q(pDst, vMagSq);
+
+    pSrc += 16;
+    pDst += 8;
+    /*
+     * Decrement the blkCnt loop counter
+     * Advance vector source and destination pointers
+     */
+    blkCnt --;
+  }
+
+  /*
+   * tail
+   */
+  blkCnt = blockSize & 7;
+  while (blkCnt > 0U)
+  {
+    /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
+
+    in = read_q15x2_ia ((q15_t **) &pSrc);
+    acc0 = __SMUAD(in, in);
+
+    /* store result in 3.13 format in destination buffer. */
+    *pDst++ = (q15_t) (acc0 >> 17);
+
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+
+#else
+void arm_cmplx_mag_squared_q15(
+  const q15_t * pSrc,
+        q15_t * pDst,
+        uint32_t numSamples)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+
+#if defined (ARM_MATH_DSP)
+        q31_t in;
+        q31_t acc0;                                    /* Accumulators */
+#else
+        q15_t real, imag;                              /* Temporary input variables */
+        q31_t acc0, acc1;                              /* Accumulators */
+#endif
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
+
+#if defined (ARM_MATH_DSP)
+    in = read_q15x2_ia ((q15_t **) &pSrc);
+    acc0 = __SMUAD(in, in);
+    /* store result in 3.13 format in destination buffer. */
+    *pDst++ = (q15_t) (acc0 >> 17);
+
+    in = read_q15x2_ia ((q15_t **) &pSrc);
+    acc0 = __SMUAD(in, in);
+    *pDst++ = (q15_t) (acc0 >> 17);
+
+    in = read_q15x2_ia ((q15_t **) &pSrc);
+    acc0 = __SMUAD(in, in);
+    *pDst++ = (q15_t) (acc0 >> 17);
+
+    in = read_q15x2_ia ((q15_t **) &pSrc);
+    acc0 = __SMUAD(in, in);
+    *pDst++ = (q15_t) (acc0 >> 17);
+#else
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = ((q31_t) real * real);
+    acc1 = ((q31_t) imag * imag);
+    /* store result in 3.13 format in destination buffer. */
+    *pDst++ = (q15_t) (((q63_t) acc0 + acc1) >> 17);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = ((q31_t) real * real);
+    acc1 = ((q31_t) imag * imag);
+    *pDst++ = (q15_t) (((q63_t) acc0 + acc1) >> 17);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = ((q31_t) real * real);
+    acc1 = ((q31_t) imag * imag);
+    *pDst++ = (q15_t) (((q63_t) acc0 + acc1) >> 17);
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = ((q31_t) real * real);
+    acc1 = ((q31_t) imag * imag);
+    *pDst++ = (q15_t) (((q63_t) acc0 + acc1) >> 17);
+#endif /* #if defined (ARM_MATH_DSP) */
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
+
+#if defined (ARM_MATH_DSP)
+    in = read_q15x2_ia ((q15_t **) &pSrc);
+    acc0 = __SMUAD(in, in);
+
+    /* store result in 3.13 format in destination buffer. */
+    *pDst++ = (q15_t) (acc0 >> 17);
+#else
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = ((q31_t) real * real);
+    acc1 = ((q31_t) imag * imag);
+
+    /* store result in 3.13 format in destination buffer. */
+    *pDst++ = (q15_t) (((q63_t) acc0 + acc1) >> 17);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of cmplx_mag_squared group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_mag_squared_q31.c

@@ -0,0 +1,187 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_mag_squared_q31.c
+ * Description:  Q31 complex magnitude squared
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @addtogroup cmplx_mag_squared
+  @{
+ */
+
+/**
+  @brief         Q31 complex magnitude squared.
+  @param[in]     pSrc        points to input vector
+  @param[out]    pDst        points to output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function implements 1.31 by 1.31 multiplications and finally output is converted into 3.29 format.
+                   Input down scaling is not required.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+void arm_cmplx_mag_squared_q31(
+  const q31_t * pSrc,
+        q31_t * pDst,
+        uint32_t numSamples)
+{
+    int32_t blockSize = numSamples;  /* loop counters */
+    uint32_t  blkCnt;           /* loop counters */
+    q31x4x2_t vecSrc;
+    q31x4_t vReal, vImag;
+    q31x4_t vMagSq;
+    q31_t real, imag;                              /* Temporary input variables */
+    q31_t acc0, acc1;                              /* Accumulators */
+
+    /* Compute 4 complex samples at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+        vecSrc = vld2q(pSrc);
+        vReal = vmulhq(vecSrc.val[0], vecSrc.val[0]);
+        vImag = vmulhq(vecSrc.val[1], vecSrc.val[1]);
+        vMagSq = vqaddq(vReal, vImag);
+        vMagSq = vshrq(vMagSq, 1);
+
+        vst1q(pDst, vMagSq);
+
+        pSrc += 8;
+        pDst += 4;
+        /*
+         * Decrement the blkCnt loop counter
+         * Advance vector source and destination pointers
+         */
+        blkCnt --;
+    } 
+
+    /* Tail */
+    blkCnt = blockSize & 3;
+    while (blkCnt > 0U)
+    {
+      /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
+  
+      real = *pSrc++;
+      imag = *pSrc++;
+      acc0 = (q31_t) (((q63_t) real * real) >> 33);
+      acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
+  
+      /* store result in 3.29 format in destination buffer. */
+      *pDst++ = acc0 + acc1;
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+}
+
+#else
+void arm_cmplx_mag_squared_q31(
+  const q31_t * pSrc,
+        q31_t * pDst,
+        uint32_t numSamples)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q31_t real, imag;                              /* Temporary input variables */
+        q31_t acc0, acc1;                              /* Accumulators */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = (q31_t) (((q63_t) real * real) >> 33);
+    acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
+    /* store the result in 3.29 format in the destination buffer. */
+    *pDst++ = acc0 + acc1;
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = (q31_t) (((q63_t) real * real) >> 33);
+    acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
+    *pDst++ = acc0 + acc1;
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = (q31_t) (((q63_t) real * real) >> 33);
+    acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
+    *pDst++ = acc0 + acc1;
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = (q31_t) (((q63_t) real * real) >> 33);
+    acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
+    *pDst++ = acc0 + acc1;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
+
+    real = *pSrc++;
+    imag = *pSrc++;
+    acc0 = (q31_t) (((q63_t) real * real) >> 33);
+    acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
+
+    /* store result in 3.29 format in destination buffer. */
+    *pDst++ = acc0 + acc1;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of cmplx_mag_squared group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_mult_cmplx_f32.c

@@ -0,0 +1,255 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_mult_cmplx_f32.c
+ * Description:  Floating-point complex-by-complex multiplication
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @defgroup CmplxByCmplxMult Complex-by-Complex Multiplication
+
+  Multiplies a complex vector by another complex vector and generates a complex result.
+  The data in the complex arrays is stored in an interleaved fashion
+  (real, imag, real, imag, ...).
+  The parameter <code>numSamples</code> represents the number of complex
+  samples processed.  The complex arrays have a total of <code>2*numSamples</code>
+  real values.
+
+  The underlying algorithm is used:
+
+  <pre>
+  for (n = 0; n < numSamples; n++) {
+      pDst[(2*n)+0] = pSrcA[(2*n)+0] * pSrcB[(2*n)+0] - pSrcA[(2*n)+1] * pSrcB[(2*n)+1];
+      pDst[(2*n)+1] = pSrcA[(2*n)+0] * pSrcB[(2*n)+1] + pSrcA[(2*n)+1] * pSrcB[(2*n)+0];
+  }
+  </pre>
+
+  There are separate functions for floating-point, Q15, and Q31 data types.
+ */
+
+/**
+  @addtogroup CmplxByCmplxMult
+  @{
+ */
+
+/**
+  @brief         Floating-point complex-by-complex multiplication.
+  @param[in]     pSrcA       points to first input vector
+  @param[in]     pSrcB       points to second input vector
+  @param[out]    pDst        points to output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+ */
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+void arm_cmplx_mult_cmplx_f32(
+  const float32_t * pSrcA,
+  const float32_t * pSrcB,
+        float32_t * pDst,
+        uint32_t numSamples)
+{
+    uint32_t blkCnt;           /* loop counters */
+    uint32_t blockSize = numSamples;  /* loop counters */
+    float32_t a, b, c, d;  /* Temporary variables to store real and imaginary values */
+
+    f32x4x2_t vecA;
+    f32x4x2_t vecB;
+    f32x4x2_t vecDst;
+
+    /* Compute 4 complex outputs at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+            vecA = vld2q(pSrcA);  // load & separate real/imag pSrcA (de-interleave 2)
+            vecB = vld2q(pSrcB);  // load & separate real/imag pSrcB
+            pSrcA += 8;
+            pSrcB += 8;
+
+            /* C[2 * i] = A[2 * i] * B[2 * i] - A[2 * i + 1] * B[2 * i + 1].  */
+            vecDst.val[0] = vmulq(vecA.val[0], vecB.val[0]);
+            vecDst.val[0] = vfmsq(vecDst.val[0],vecA.val[1], vecB.val[1]);
+            /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i].  */
+            vecDst.val[1] = vmulq(vecA.val[0], vecB.val[1]);
+            vecDst.val[1] = vfmaq(vecDst.val[1], vecA.val[1], vecB.val[0]);
+
+            vst2q(pDst, vecDst);
+            pDst += 8;
+
+            blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = blockSize & 3;
+    while (blkCnt > 0U)
+    {
+      /* C[2 * i    ] = A[2 * i] * B[2 * i    ] - A[2 * i + 1] * B[2 * i + 1]. */
+      /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i    ]. */
+  
+      a = *pSrcA++;
+      b = *pSrcA++;
+      c = *pSrcB++;
+      d = *pSrcB++;
+  
+      /* store result in destination buffer. */
+      *pDst++ = (a * c) - (b * d);
+      *pDst++ = (a * d) + (b * c);
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+
+}
+
+#else
+void arm_cmplx_mult_cmplx_f32(
+  const float32_t * pSrcA,
+  const float32_t * pSrcB,
+        float32_t * pDst,
+        uint32_t numSamples)
+{
+    uint32_t blkCnt;                               /* Loop counter */
+    float32_t a, b, c, d;  /* Temporary variables to store real and imaginary values */
+
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    float32x4x2_t va, vb;
+    float32x4x2_t outCplx;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = numSamples >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        va = vld2q_f32(pSrcA);  // load & separate real/imag pSrcA (de-interleave 2)
+        vb = vld2q_f32(pSrcB);  // load & separate real/imag pSrcB
+
+	/* Increment pointers */
+        pSrcA += 8;
+        pSrcB += 8;
+	
+	/* Re{C} = Re{A}*Re{B} - Im{A}*Im{B} */
+        outCplx.val[0] = vmulq_f32(va.val[0], vb.val[0]);
+        outCplx.val[0] = vmlsq_f32(outCplx.val[0], va.val[1], vb.val[1]);
+
+	/* Im{C} = Re{A}*Im{B} + Im{A}*Re{B} */
+        outCplx.val[1] = vmulq_f32(va.val[0], vb.val[1]);
+        outCplx.val[1] = vmlaq_f32(outCplx.val[1], va.val[1], vb.val[0]);
+
+        vst2q_f32(pDst, outCplx);
+
+	/* Increment pointer */
+        pDst += 8;
+
+	/* Decrement the loop counter */
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = numSamples & 3;
+
+#else
+#if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i] * B[2 * i    ] - A[2 * i + 1] * B[2 * i + 1]. */
+    /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i    ]. */
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+    /* store result in destination buffer. */
+    *pDst++ = (a * c) - (b * d);
+    *pDst++ = (a * d) + (b * c);
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+    *pDst++ = (a * c) - (b * d);
+    *pDst++ = (a * d) + (b * c);
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+    *pDst++ = (a * c) - (b * d);
+    *pDst++ = (a * d) + (b * c);
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+    *pDst++ = (a * c) - (b * d);
+    *pDst++ = (a * d) + (b * c);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined(ARM_MATH_NEON) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i] * B[2 * i    ] - A[2 * i + 1] * B[2 * i + 1]. */
+    /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i    ]. */
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+
+    /* store result in destination buffer. */
+    *pDst++ = (a * c) - (b * d);
+    *pDst++ = (a * d) + (b * c);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+/**
+  @} end of CmplxByCmplxMult group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_mult_cmplx_q15.c

@@ -0,0 +1,196 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_mult_cmplx_q15.c
+ * Description:  Q15 complex-by-complex multiplication
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @addtogroup CmplxByCmplxMult
+  @{
+ */
+
+/**
+  @brief         Q15 complex-by-complex multiplication.
+  @param[in]     pSrcA       points to first input vector
+  @param[in]     pSrcB       points to second input vector
+  @param[out]    pDst        points to output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function implements 1.15 by 1.15 multiplications and finally output is converted into 3.13 format.
+ */
+
+#if defined(ARM_MATH_MVEI)
+
+void arm_cmplx_mult_cmplx_q15(
+  const q15_t * pSrcA,
+  const q15_t * pSrcB,
+        q15_t * pDst,
+        uint32_t numSamples)
+{
+  uint32_t blkCnt;           /* loop counters */
+  uint32_t blockSize = numSamples * CMPLX_DIM;  /* loop counters */
+  q15_t a, b, c, d;  
+
+  q15x8_t vecA;
+  q15x8_t vecB;
+  q15x8_t vecDst;
+
+  blkCnt = blockSize >> 3;
+  while (blkCnt > 0U)
+  {
+      vecA = vld1q(pSrcA);
+      vecB = vld1q(pSrcB);
+      /* C[2 * i] = A[2 * i] * B[2 * i] - A[2 * i + 1] * B[2 * i + 1].  */
+      vecDst = vqdmlsdhq_s16(vuninitializedq_s16(), vecA, vecB);
+      /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i].  */
+      vecDst = vqdmladhxq_s16(vecDst, vecA, vecB);
+
+      vecDst = vshrq(vecDst, 2);
+
+      vst1q(pDst, vecDst);
+
+      blkCnt --;
+      pSrcA += 8;
+      pSrcB += 8;
+      pDst += 8;
+  };
+
+  /*
+   * tail
+   */
+  blkCnt = (blockSize & 7) >> 1;
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i] * B[2 * i    ] - A[2 * i + 1] * B[2 * i + 1]. */
+    /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i    ]. */
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+
+    /* store result in 3.13 format in destination buffer. */
+    *pDst++ = (q15_t) ( (((q31_t) a * c) >> 17) - (((q31_t) b * d) >> 17) );
+    *pDst++ = (q15_t) ( (((q31_t) a * d) >> 17) + (((q31_t) b * c) >> 17) );
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+}
+#else
+void arm_cmplx_mult_cmplx_q15(
+  const q15_t * pSrcA,
+  const q15_t * pSrcB,
+        q15_t * pDst,
+        uint32_t numSamples)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q15_t a, b, c, d;                              /* Temporary variables */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i] * B[2 * i    ] - A[2 * i + 1] * B[2 * i + 1]. */
+    /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i    ]. */
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+    /* store result in 3.13 format in destination buffer. */
+    *pDst++ = (q15_t) ( (((q31_t) a * c) >> 17) - (((q31_t) b * d) >> 17) );
+    *pDst++ = (q15_t) ( (((q31_t) a * d) >> 17) + (((q31_t) b * c) >> 17) );
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+    *pDst++ = (q15_t) ( (((q31_t) a * c) >> 17) - (((q31_t) b * d) >> 17) );
+    *pDst++ = (q15_t) ( (((q31_t) a * d) >> 17) + (((q31_t) b * c) >> 17) );
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+    *pDst++ = (q15_t) ( (((q31_t) a * c) >> 17) - (((q31_t) b * d) >> 17) );
+    *pDst++ = (q15_t) ( (((q31_t) a * d) >> 17) + (((q31_t) b * c) >> 17) );
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+    *pDst++ = (q15_t) ( (((q31_t) a * c) >> 17) - (((q31_t) b * d) >> 17) );
+    *pDst++ = (q15_t) ( (((q31_t) a * d) >> 17) + (((q31_t) b * c) >> 17) );
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i] * B[2 * i    ] - A[2 * i + 1] * B[2 * i + 1]. */
+    /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i    ]. */
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+
+    /* store result in 3.13 format in destination buffer. */
+    *pDst++ = (q15_t) ( (((q31_t) a * c) >> 17) - (((q31_t) b * d) >> 17) );
+    *pDst++ = (q15_t) ( (((q31_t) a * d) >> 17) + (((q31_t) b * c) >> 17) );
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of CmplxByCmplxMult group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_mult_cmplx_q31.c

@@ -0,0 +1,194 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_mult_cmplx_q31.c
+ * Description:  Q31 complex-by-complex multiplication
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @addtogroup CmplxByCmplxMult
+  @{
+ */
+
+/**
+  @brief         Q31 complex-by-complex multiplication.
+  @param[in]     pSrcA       points to first input vector
+  @param[in]     pSrcB       points to second input vector
+  @param[out]    pDst        points to output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function implements 1.31 by 1.31 multiplications and finally output is converted into 3.29 format.
+                   Input down scaling is not required.
+ */
+
+#if defined(ARM_MATH_MVEI)
+void arm_cmplx_mult_cmplx_q31(
+  const q31_t * pSrcA,
+  const q31_t * pSrcB,
+        q31_t * pDst,
+        uint32_t numSamples)
+{
+
+    uint32_t blkCnt;           /* loop counters */
+    uint32_t blockSize = numSamples * CMPLX_DIM;  /* loop counters */
+    q31x4_t vecA;
+    q31x4_t vecB;
+    q31x4_t vecDst;
+    q31_t a, b, c, d;                              /* Temporary variables */
+
+    /* Compute 2 complex outputs at a time */
+    blkCnt = blockSize >> 2;
+    while (blkCnt > 0U)
+    {
+
+        vecA = vld1q(pSrcA);
+        vecB = vld1q(pSrcB);
+        /* C[2 * i] = A[2 * i] * B[2 * i] - A[2 * i + 1] * B[2 * i + 1].  */
+        vecDst = vqdmlsdhq(vuninitializedq_s32(),vecA, vecB);
+        /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i].  */
+        vecDst = vqdmladhxq(vecDst, vecA, vecB);
+
+        vecDst = vshrq(vecDst, 2);
+        vst1q(pDst, vecDst);
+
+        blkCnt --;
+        pSrcA += 4;
+        pSrcB += 4;
+        pDst += 4;
+    };
+
+    blkCnt = (blockSize & 3) >> 1;
+    while (blkCnt > 0U)
+    {
+      /* C[2 * i    ] = A[2 * i] * B[2 * i    ] - A[2 * i + 1] * B[2 * i + 1]. */
+      /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i    ]. */
+  
+      a = *pSrcA++;
+      b = *pSrcA++;
+      c = *pSrcB++;
+      d = *pSrcB++;
+  
+      /* store result in 3.29 format in destination buffer. */
+      *pDst++ = (q31_t) ( (((q63_t) a * c) >> 33) - (((q63_t) b * d) >> 33) );
+      *pDst++ = (q31_t) ( (((q63_t) a * d) >> 33) + (((q63_t) b * c) >> 33) );
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+}
+#else
+void arm_cmplx_mult_cmplx_q31(
+  const q31_t * pSrcA,
+  const q31_t * pSrcB,
+        q31_t * pDst,
+        uint32_t numSamples)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q31_t a, b, c, d;                              /* Temporary variables */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i] * B[2 * i    ] - A[2 * i + 1] * B[2 * i + 1]. */
+    /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i    ]. */
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+    /* store result in 3.29 format in destination buffer. */
+    *pDst++ = (q31_t) ( (((q63_t) a * c) >> 33) - (((q63_t) b * d) >> 33) );
+    *pDst++ = (q31_t) ( (((q63_t) a * d) >> 33) + (((q63_t) b * c) >> 33) );
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+    *pDst++ = (q31_t) ( (((q63_t) a * c) >> 33) - (((q63_t) b * d) >> 33) );
+    *pDst++ = (q31_t) ( (((q63_t) a * d) >> 33) + (((q63_t) b * c) >> 33) );
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+    *pDst++ = (q31_t) ( (((q63_t) a * c) >> 33) - (((q63_t) b * d) >> 33) );
+    *pDst++ = (q31_t) ( (((q63_t) a * d) >> 33) + (((q63_t) b * c) >> 33) );
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+    *pDst++ = (q31_t) ( (((q63_t) a * c) >> 33) - (((q63_t) b * d) >> 33) );
+    *pDst++ = (q31_t) ( (((q63_t) a * d) >> 33) + (((q63_t) b * c) >> 33) );
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i] * B[2 * i    ] - A[2 * i + 1] * B[2 * i + 1]. */
+    /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i    ]. */
+
+    a = *pSrcA++;
+    b = *pSrcA++;
+    c = *pSrcB++;
+    d = *pSrcB++;
+
+    /* store result in 3.29 format in destination buffer. */
+    *pDst++ = (q31_t) ( (((q63_t) a * c) >> 33) - (((q63_t) b * d) >> 33) );
+    *pDst++ = (q31_t) ( (((q63_t) a * d) >> 33) + (((q63_t) b * c) >> 33) );
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of CmplxByCmplxMult group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_mult_real_f32.c

@@ -0,0 +1,224 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_mult_real_f32.c
+ * Description:  Floating-point complex by real multiplication
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @defgroup CmplxByRealMult Complex-by-Real Multiplication
+
+  Multiplies a complex vector by a real vector and generates a complex result.
+  The data in the complex arrays is stored in an interleaved fashion
+  (real, imag, real, imag, ...).
+  The parameter <code>numSamples</code> represents the number of complex
+  samples processed.  The complex arrays have a total of <code>2*numSamples</code>
+  real values while the real array has a total of <code>numSamples</code>
+  real values.
+
+  The underlying algorithm is used:
+
+  <pre>
+  for (n = 0; n < numSamples; n++) {
+      pCmplxDst[(2*n)+0] = pSrcCmplx[(2*n)+0] * pSrcReal[n];
+      pCmplxDst[(2*n)+1] = pSrcCmplx[(2*n)+1] * pSrcReal[n];
+  }
+  </pre>
+
+  There are separate functions for floating-point, Q15, and Q31 data types.
+ */
+
+/**
+  @addtogroup CmplxByRealMult
+  @{
+ */
+
+/**
+  @brief         Floating-point complex-by-real multiplication.
+  @param[in]     pSrcCmplx   points to complex input vector
+  @param[in]     pSrcReal    points to real input vector
+  @param[out]    pCmplxDst   points to complex output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+ */
+
+#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+void arm_cmplx_mult_real_f32(
+  const float32_t * pSrcCmplx,
+  const float32_t * pSrcReal,
+        float32_t * pCmplxDst,
+        uint32_t numSamples)
+{
+    const static uint32_t stride_cmplx_x_real_32[4] = { 0, 0, 1, 1 };
+
+    uint32_t blockSizeC = numSamples * CMPLX_DIM;   /* loop counters */
+    uint32_t blkCnt;
+    f32x4_t rVec;
+    f32x4_t cmplxVec;
+    f32x4_t dstVec;
+    uint32x4_t strideVec;
+    float32_t in;  
+
+
+    /* stride vector for pairs of real generation */
+    strideVec = vld1q(stride_cmplx_x_real_32);
+
+    /* Compute 4 complex outputs at a time */
+    blkCnt = blockSizeC >> 2;
+    while (blkCnt > 0U) 
+    {
+        cmplxVec = vld1q(pSrcCmplx);
+        rVec = vldrwq_gather_shifted_offset_f32(pSrcReal, strideVec);
+        dstVec = vmulq(cmplxVec, rVec);
+        vst1q(pCmplxDst, dstVec);
+
+        pSrcReal += 2;
+        pSrcCmplx += 4;
+        pCmplxDst += 4;
+        blkCnt--;
+    }
+
+    blkCnt = (blockSizeC & 3) >> 1; 
+    while (blkCnt > 0U)
+    {
+      /* C[2 * i    ] = A[2 * i    ] * B[i]. */
+      /* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
+  
+      in = *pSrcReal++;
+      /* store result in destination buffer. */
+      *pCmplxDst++ = *pSrcCmplx++ * in;
+      *pCmplxDst++ = *pSrcCmplx++ * in;
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+}
+
+#else
+void arm_cmplx_mult_real_f32(
+  const float32_t * pSrcCmplx,
+  const float32_t * pSrcReal,
+        float32_t * pCmplxDst,
+        uint32_t numSamples)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        float32_t in;                                  /* Temporary variable */
+
+#if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
+    float32x4_t r;
+    float32x4x2_t ab,outCplx;
+
+    /* Compute 4 outputs at a time */
+    blkCnt = numSamples >> 2U;
+
+    while (blkCnt > 0U)
+    {
+        ab = vld2q_f32(pSrcCmplx);  // load & separate real/imag pSrcA (de-interleave 2)
+        r = vld1q_f32(pSrcReal);  // load & separate real/imag pSrcB
+
+	/* Increment pointers */
+        pSrcCmplx += 8;
+        pSrcReal += 4;
+
+        outCplx.val[0] = vmulq_f32(ab.val[0], r);
+        outCplx.val[1] = vmulq_f32(ab.val[1], r);
+
+        vst2q_f32(pCmplxDst, outCplx);
+        pCmplxDst += 8;
+
+        blkCnt--;
+    }
+
+    /* Tail */
+    blkCnt = numSamples & 3;
+#else
+#if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i    ] * B[i]. */
+    /* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
+
+    in = *pSrcReal++;
+    /* store result in destination buffer. */
+    *pCmplxDst++ = *pSrcCmplx++ * in;
+    *pCmplxDst++ = *pSrcCmplx++ * in;
+
+    in = *pSrcReal++;
+    *pCmplxDst++ = *pSrcCmplx++ * in;
+    *pCmplxDst++ = *pSrcCmplx++ * in;
+
+    in = *pSrcReal++;
+    *pCmplxDst++ = *pSrcCmplx++ * in;
+    *pCmplxDst++ = *pSrcCmplx++ * in;
+
+    in = *pSrcReal++;
+    *pCmplxDst++ = *pSrcCmplx++* in;
+    *pCmplxDst++ = *pSrcCmplx++ * in;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+#endif /* #if defined(ARM_MATH_NEON) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i    ] * B[i]. */
+    /* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
+
+    in = *pSrcReal++;
+    /* store result in destination buffer. */
+    *pCmplxDst++ = *pSrcCmplx++ * in;
+    *pCmplxDst++ = *pSrcCmplx++ * in;
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
+
+/**
+  @} end of CmplxByRealMult group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_mult_real_q15.c

@@ -0,0 +1,238 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_mult_real_q15.c
+ * Description:  Q15 complex by real multiplication
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @addtogroup CmplxByRealMult
+  @{
+ */
+
+/**
+  @brief         Q15 complex-by-real multiplication.
+  @param[in]     pSrcCmplx   points to complex input vector
+  @param[in]     pSrcReal    points to real input vector
+  @param[out]    pCmplxDst   points to complex output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q15 range [0x8000 0x7FFF] are saturated.
+ */
+#if defined(ARM_MATH_MVEI)
+
+void arm_cmplx_mult_real_q15(
+  const q15_t * pSrcCmplx,
+  const q15_t * pSrcReal,
+        q15_t * pCmplxDst,
+        uint32_t numSamples)
+{
+  const static uint16_t stride_cmplx_x_real_16[8] = {
+      0, 0, 1, 1, 2, 2, 3, 3
+      };
+  q15x8_t rVec;
+  q15x8_t cmplxVec;
+  q15x8_t dstVec;
+  uint16x8_t strideVec;
+  uint32_t blockSizeC = numSamples * CMPLX_DIM;   /* loop counters */
+  uint32_t blkCnt;
+  q15_t in;  
+
+  /*
+  * stride vector for pairs of real generation
+  */
+  strideVec = vld1q(stride_cmplx_x_real_16);
+
+  blkCnt = blockSizeC >> 3;
+
+  while (blkCnt > 0U) 
+  {
+    cmplxVec = vld1q(pSrcCmplx);
+    rVec = vldrhq_gather_shifted_offset_s16(pSrcReal, strideVec);
+    dstVec = vqdmulhq(cmplxVec, rVec);
+    vst1q(pCmplxDst, dstVec);
+
+    pSrcReal += 4;
+    pSrcCmplx += 8;
+    pCmplxDst += 8;
+    blkCnt --;
+  }
+
+  /* Tail */
+  blkCnt = (blockSizeC & 7) >> 1;
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i    ] * B[i]. */
+    /* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
+
+    in = *pSrcReal++;
+    /* store the result in the destination buffer. */
+    *pCmplxDst++ = (q15_t) __SSAT((((q31_t) *pSrcCmplx++ * in) >> 15), 16);
+    *pCmplxDst++ = (q15_t) __SSAT((((q31_t) *pSrcCmplx++ * in) >> 15), 16);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+}
+#else
+void arm_cmplx_mult_real_q15(
+  const q15_t * pSrcCmplx,
+  const q15_t * pSrcReal,
+        q15_t * pCmplxDst,
+        uint32_t numSamples)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q15_t in;                                      /* Temporary variable */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+#if defined (ARM_MATH_DSP)
+        q31_t inA1, inA2;                              /* Temporary variables to hold input data */
+        q31_t inB1;                                    /* Temporary variables to hold input data */
+        q15_t out1, out2, out3, out4;                  /* Temporary variables to hold output data */
+        q31_t mul1, mul2, mul3, mul4;                  /* Temporary variables to hold intermediate data */
+#endif
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i    ] * B[i]. */
+    /* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
+
+#if defined (ARM_MATH_DSP)
+    /* read 2 complex numbers both real and imaginary from complex input buffer */
+    inA1 = read_q15x2_ia ((q15_t **) &pSrcCmplx);
+    inA2 = read_q15x2_ia ((q15_t **) &pSrcCmplx);
+    /* read 2 real values at a time from real input buffer */
+    inB1 = read_q15x2_ia ((q15_t **) &pSrcReal);
+
+    /* multiply complex number with real numbers */
+#ifndef ARM_MATH_BIG_ENDIAN
+    mul1 = (q31_t) ((q15_t) (inA1)       * (q15_t) (inB1));
+    mul2 = (q31_t) ((q15_t) (inA1 >> 16) * (q15_t) (inB1));
+    mul3 = (q31_t) ((q15_t) (inA2)       * (q15_t) (inB1 >> 16));
+    mul4 = (q31_t) ((q15_t) (inA2 >> 16) * (q15_t) (inB1 >> 16));
+#else
+    mul2 = (q31_t) ((q15_t) (inA1 >> 16) * (q15_t) (inB1 >> 16));
+    mul1 = (q31_t) ((q15_t) inA1         * (q15_t) (inB1 >> 16));
+    mul4 = (q31_t) ((q15_t) (inA2 >> 16) * (q15_t) inB1);
+    mul3 = (q31_t) ((q15_t) inA2         * (q15_t) inB1);
+#endif /* #ifndef ARM_MATH_BIG_ENDIAN */
+
+    /* saturate the result */
+    out1 = (q15_t) __SSAT(mul1 >> 15U, 16);
+    out2 = (q15_t) __SSAT(mul2 >> 15U, 16);
+    out3 = (q15_t) __SSAT(mul3 >> 15U, 16);
+    out4 = (q15_t) __SSAT(mul4 >> 15U, 16);
+
+    /* pack real and imaginary outputs and store them to destination */
+    write_q15x2_ia (&pCmplxDst, __PKHBT(out1, out2, 16));
+    write_q15x2_ia (&pCmplxDst, __PKHBT(out3, out4, 16));
+
+    inA1 = read_q15x2_ia ((q15_t **) &pSrcCmplx);
+    inA2 = read_q15x2_ia ((q15_t **) &pSrcCmplx);
+    inB1 = read_q15x2_ia ((q15_t **) &pSrcReal);
+
+#ifndef ARM_MATH_BIG_ENDIAN
+    mul1 = (q31_t) ((q15_t) (inA1)       * (q15_t) (inB1));
+    mul2 = (q31_t) ((q15_t) (inA1 >> 16) * (q15_t) (inB1));
+    mul3 = (q31_t) ((q15_t) (inA2)       * (q15_t) (inB1 >> 16));
+    mul4 = (q31_t) ((q15_t) (inA2 >> 16) * (q15_t) (inB1 >> 16));
+#else
+    mul2 = (q31_t) ((q15_t) (inA1 >> 16) * (q15_t) (inB1 >> 16));
+    mul1 = (q31_t) ((q15_t) inA1         * (q15_t) (inB1 >> 16));
+    mul4 = (q31_t) ((q15_t) (inA2 >> 16) * (q15_t) inB1);
+    mul3 = (q31_t) ((q15_t) inA2 * (q15_t) inB1);
+#endif /* #ifndef ARM_MATH_BIG_ENDIAN */
+
+    out1 = (q15_t) __SSAT(mul1 >> 15U, 16);
+    out2 = (q15_t) __SSAT(mul2 >> 15U, 16);
+    out3 = (q15_t) __SSAT(mul3 >> 15U, 16);
+    out4 = (q15_t) __SSAT(mul4 >> 15U, 16);
+
+    write_q15x2_ia (&pCmplxDst, __PKHBT(out1, out2, 16));
+    write_q15x2_ia (&pCmplxDst, __PKHBT(out3, out4, 16));
+#else
+    in = *pSrcReal++;
+    *pCmplxDst++ = (q15_t) __SSAT((((q31_t) *pSrcCmplx++ * in) >> 15), 16);
+    *pCmplxDst++ = (q15_t) __SSAT((((q31_t) *pSrcCmplx++ * in) >> 15), 16);
+
+    in = *pSrcReal++;
+    *pCmplxDst++ = (q15_t) __SSAT((((q31_t) *pSrcCmplx++ * in) >> 15), 16);
+    *pCmplxDst++ = (q15_t) __SSAT((((q31_t) *pSrcCmplx++ * in) >> 15), 16);
+
+    in = *pSrcReal++;
+    *pCmplxDst++ = (q15_t) __SSAT((((q31_t) *pSrcCmplx++ * in) >> 15), 16);
+    *pCmplxDst++ = (q15_t) __SSAT((((q31_t) *pSrcCmplx++ * in) >> 15), 16);
+
+    in = *pSrcReal++;
+    *pCmplxDst++ = (q15_t) __SSAT((((q31_t) *pSrcCmplx++ * in) >> 15), 16);
+    *pCmplxDst++ = (q15_t) __SSAT((((q31_t) *pSrcCmplx++ * in) >> 15), 16);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i    ] * B[i]. */
+    /* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
+
+    in = *pSrcReal++;
+    /* store the result in the destination buffer. */
+    *pCmplxDst++ = (q15_t) __SSAT((((q31_t) *pSrcCmplx++ * in) >> 15), 16);
+    *pCmplxDst++ = (q15_t) __SSAT((((q31_t) *pSrcCmplx++ * in) >> 15), 16);
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of CmplxByRealMult group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_mult_real_q31.c

@@ -0,0 +1,204 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_cmplx_mult_real_q31.c
+ * Description:  Q31 complex by real multiplication
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @ingroup groupCmplxMath
+ */
+
+/**
+  @addtogroup CmplxByRealMult
+  @{
+ */
+
+/**
+  @brief         Q31 complex-by-real multiplication.
+  @param[in]     pSrcCmplx   points to complex input vector
+  @param[in]     pSrcReal    points to real input vector
+  @param[out]    pCmplxDst   points to complex output vector
+  @param[in]     numSamples  number of samples in each vector
+  @return        none
+
+  @par           Scaling and Overflow Behavior
+                   The function uses saturating arithmetic.
+                   Results outside of the allowable Q31 range[0x80000000 0x7FFFFFFF] are saturated.
+ */
+
+#if defined(ARM_MATH_MVEI)
+void arm_cmplx_mult_real_q31(
+  const q31_t * pSrcCmplx,
+  const q31_t * pSrcReal,
+        q31_t * pCmplxDst,
+        uint32_t numSamples)
+{
+
+    const static uint32_t stride_cmplx_x_real_32[4] = {
+        0, 0, 1, 1
+    };
+    q31x4_t rVec;
+    q31x4_t cmplxVec;
+    q31x4_t dstVec;
+    uint32x4_t strideVec;
+    uint32_t blockSizeC = numSamples * CMPLX_DIM;   /* loop counters */
+    uint32_t blkCnt;
+    q31_t in;
+
+    /*
+     * stride vector for pairs of real generation
+     */
+    strideVec = vld1q(stride_cmplx_x_real_32);
+
+    /* Compute 4 complex outputs at a time */
+    blkCnt = blockSizeC >> 2;
+    while (blkCnt > 0U) 
+    {
+        cmplxVec = vld1q(pSrcCmplx);
+        rVec = vldrwq_gather_shifted_offset_s32(pSrcReal, strideVec);
+        dstVec = vqdmulhq(cmplxVec, rVec);
+        vst1q(pCmplxDst, dstVec);
+
+        pSrcReal += 2;
+        pSrcCmplx += 4;
+        pCmplxDst += 4;
+        blkCnt --;
+    }
+
+    blkCnt = (blockSizeC & 3) >> 1; 
+    while (blkCnt > 0U)
+    {
+      /* C[2 * i    ] = A[2 * i    ] * B[i]. */
+      /* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
+  
+      in = *pSrcReal++;
+      /* store saturated result in 1.31 format to destination buffer */
+      *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
+      *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
+  
+      /* Decrement loop counter */
+      blkCnt--;
+    }
+}
+#else
+void arm_cmplx_mult_real_q31(
+  const q31_t * pSrcCmplx,
+  const q31_t * pSrcReal,
+        q31_t * pCmplxDst,
+        uint32_t numSamples)
+{
+        uint32_t blkCnt;                               /* Loop counter */
+        q31_t in;                                      /* Temporary variable */
+
+#if defined (ARM_MATH_LOOPUNROLL)
+
+  /* Loop unrolling: Compute 4 outputs at a time */
+  blkCnt = numSamples >> 2U;
+
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i    ] * B[i]. */
+    /* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
+
+    in = *pSrcReal++;
+#if defined (ARM_MATH_DSP)
+    /* store saturated result in 1.31 format to destination buffer */
+    *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
+    *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
+#else
+    /* store result in destination buffer. */
+    *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
+    *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
+#endif
+
+    in = *pSrcReal++;
+#if defined (ARM_MATH_DSP)
+    *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
+    *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
+#else
+    *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
+    *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
+#endif
+
+    in = *pSrcReal++;
+#if defined (ARM_MATH_DSP)
+    *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
+    *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
+#else
+    *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
+    *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
+#endif
+
+    in = *pSrcReal++;
+#if defined (ARM_MATH_DSP)
+    *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
+    *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
+#else
+    *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
+    *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+  /* Loop unrolling: Compute remaining outputs */
+  blkCnt = numSamples % 0x4U;
+
+#else
+
+  /* Initialize blkCnt with number of samples */
+  blkCnt = numSamples;
+
+#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
+
+  while (blkCnt > 0U)
+  {
+    /* C[2 * i    ] = A[2 * i    ] * B[i]. */
+    /* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
+
+    in = *pSrcReal++;
+#if defined (ARM_MATH_DSP)
+    /* store saturated result in 1.31 format to destination buffer */
+    *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
+    *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
+#else
+    /* store result in destination buffer. */
+    *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
+    *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
+#endif
+
+    /* Decrement loop counter */
+    blkCnt--;
+  }
+
+}
+#endif /* defined(ARM_MATH_MVEI) */
+
+/**
+  @} end of CmplxByRealMult group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ControllerFunctions/arm_pid_init_f32.c

@@ -0,0 +1,75 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_pid_init_f32.c
+ * Description:  Floating-point PID Control initialization function
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @addtogroup PID
+  @{
+ */
+
+/**
+  @brief         Initialization function for the floating-point PID Control.
+  @param[in,out] S               points to an instance of the PID structure
+  @param[in]     resetStateFlag
+                   - value = 0: no change in state
+                   - value = 1: reset state
+  @return        none
+
+  @par           Details
+                   The <code>resetStateFlag</code> specifies whether to set state to zero or not. \n
+                   The function computes the structure fields: <code>A0</code>, <code>A1</code> <code>A2</code>
+                   using the proportional gain( \c Kp), integral gain( \c Ki) and derivative gain( \c Kd)
+                   also sets the state variables to all zeros.
+ */
+
+void arm_pid_init_f32(
+  arm_pid_instance_f32 * S,
+  int32_t resetStateFlag)
+{
+  /* Derived coefficient A0 */
+  S->A0 = S->Kp + S->Ki + S->Kd;
+
+  /* Derived coefficient A1 */
+  S->A1 = (-S->Kp) - ((float32_t) 2.0f * S->Kd);
+
+  /* Derived coefficient A2 */
+  S->A2 = S->Kd;
+
+  /* Check whether state needs reset or not */
+  if (resetStateFlag)
+  {
+    /* Reset state to zero, The size will be always 3 samples */
+    memset(S->state, 0, 3U * sizeof(float32_t));
+  }
+
+}
+
+/**
+  @} end of PID group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ControllerFunctions/arm_pid_init_q15.c

@@ -0,0 +1,95 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_pid_init_q15.c
+ * Description:  Q15 PID Control initialization function
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @addtogroup PID
+  @{
+ */
+
+/**
+  @brief         Initialization function for the Q15 PID Control.
+  @param[in,out] S               points to an instance of the Q15 PID structure
+  @param[in]     resetStateFlag
+                   - value = 0: no change in state
+                   - value = 1: reset state
+  @return        none
+
+  @par           Details
+                   The <code>resetStateFlag</code> specifies whether to set state to zero or not. \n
+                   The function computes the structure fields: <code>A0</code>, <code>A1</code> <code>A2</code>
+                   using the proportional gain( \c Kp), integral gain( \c Ki) and derivative gain( \c Kd)
+                   also sets the state variables to all zeros.
+ */
+
+void arm_pid_init_q15(
+  arm_pid_instance_q15 * S,
+  int32_t resetStateFlag)
+{
+
+#if defined (ARM_MATH_DSP)
+
+  /* Derived coefficient A0 */
+  S->A0 = __QADD16(__QADD16(S->Kp, S->Ki), S->Kd);
+
+  /* Derived coefficients and pack into A1 */
+
+#ifndef  ARM_MATH_BIG_ENDIAN
+  S->A1 = __PKHBT(-__QADD16(__QADD16(S->Kd, S->Kd), S->Kp), S->Kd, 16);
+#else
+  S->A1 = __PKHBT(S->Kd, -__QADD16(__QADD16(S->Kd, S->Kd), S->Kp), 16);
+#endif
+
+#else
+
+  q31_t temp;                                    /* to store the sum */
+
+  /* Derived coefficient A0 */
+  temp = S->Kp + S->Ki + S->Kd;
+  S->A0 = (q15_t) __SSAT(temp, 16);
+
+  /* Derived coefficients and pack into A1 */
+  temp = -(S->Kd + S->Kd + S->Kp);
+  S->A1 = (q15_t) __SSAT(temp, 16);
+  S->A2 = S->Kd;
+
+#endif /* #if defined (ARM_MATH_DSP) */
+
+  /* Check whether state needs reset or not */
+  if (resetStateFlag)
+  {
+    /* Reset state to zero, The size will be always 3 samples */
+    memset(S->state, 0, 3U * sizeof(q15_t));
+  }
+
+}
+
+/**
+  @} end of PID group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ControllerFunctions/arm_pid_init_q31.c

@@ -0,0 +1,92 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_pid_init_q31.c
+ * Description:  Q31 PID Control initialization function
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @addtogroup PID
+  @{
+ */
+
+/**
+  @brief         Initialization function for the Q31 PID Control.
+  @param[in,out] S               points to an instance of the Q31 PID structure
+  @param[in]     resetStateFlag
+                   - value = 0: no change in state
+                   - value = 1: reset state
+  @return        none
+
+  @par           Details
+                   The <code>resetStateFlag</code> specifies whether to set state to zero or not. \n
+                   The function computes the structure fields: <code>A0</code>, <code>A1</code> <code>A2</code>
+                   using the proportional gain( \c Kp), integral gain( \c Ki) and derivative gain( \c Kd)
+                   also sets the state variables to all zeros.
+ */
+
+void arm_pid_init_q31(
+  arm_pid_instance_q31 * S,
+  int32_t resetStateFlag)
+{
+
+#if defined (ARM_MATH_DSP)
+
+  /* Derived coefficient A0 */
+  S->A0 = __QADD(__QADD(S->Kp, S->Ki), S->Kd);
+
+  /* Derived coefficient A1 */
+  S->A1 = -__QADD(__QADD(S->Kd, S->Kd), S->Kp);
+
+#else
+
+  q31_t temp;                                    /* to store the sum */
+
+  /* Derived coefficient A0 */
+  temp = clip_q63_to_q31((q63_t) S->Kp + S->Ki);
+  S->A0 = clip_q63_to_q31((q63_t) temp + S->Kd);
+
+  /* Derived coefficient A1 */
+  temp = clip_q63_to_q31((q63_t) S->Kd + S->Kd);
+  S->A1 = -clip_q63_to_q31((q63_t) temp + S->Kp);
+
+#endif /* #if defined (ARM_MATH_DSP) */
+
+  /* Derived coefficient A2 */
+  S->A2 = S->Kd;
+
+  /* Check whether state needs reset or not */
+  if (resetStateFlag)
+  {
+    /* Reset state to zero, The size will be always 3 samples */
+    memset(S->state, 0, 3U * sizeof(q31_t));
+  }
+
+}
+
+/**
+  @} end of PID group
+ */

libraries/sdk/third_party/CMSIS/DSP/Source/ControllerFunctions/arm_pid_reset_f32.c

@@ -0,0 +1,54 @@
+/* ----------------------------------------------------------------------
+ * Project:      CMSIS DSP Library
+ * Title:        arm_pid_reset_f32.c
+ * Description:  Floating-point PID Control reset function
+ *
+ * $Date:        18. March 2019
+ * $Revision:    V1.6.0
+ *
+ * Target Processor: Cortex-M cores
+ * -------------------------------------------------------------------- */
+/*
+ * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved.
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the License); you may
+ * not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an AS IS BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "arm_math.h"
+
+/**
+  @addtogroup PID
+  @{
+ */
+
+/**
+  @brief         Reset function for the floating-point PID Control.
+  @param[in,out] S  points to an instance of the floating-point PID structure
+  @return        none
+
+  @par           Details
+                   The function resets the state buffer to zeros.
+ */
+
+void arm_pid_reset_f32(
+  arm_pid_instance_f32 * S)
+{
+  /* Reset state to zero, The size will be always 3 samples */
+  memset(S->state, 0, 3U * sizeof(float32_t));
+}
+
+/**
+  @} end of PID group
+ */