CMSIS_5/CMSIS/Core_A/Include/cmsis_gcc.h

/**************************************************************************//**
 * @file     cmsis_gcc.h
 * @brief    CMSIS compiler GCC header file
 * @version  V1.3.3
 * @date     13. November 2022
 ******************************************************************************/
/*
 * Copyright (c) 2009-2022 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_GCC_H
#define __CMSIS_GCC_H

/* ignore some GCC warnings */
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wsign-conversion"
#pragma GCC diagnostic ignored "-Wconversion"
#pragma GCC diagnostic ignored "-Wunused-parameter"

/* Fallback for __has_builtin */
#ifndef __has_builtin
  #define __has_builtin(x) (0)
#endif

/* CMSIS compiler specific defines */
#ifndef   __ASM
  #define __ASM                                  __asm
#endif
#ifndef   __INLINE
  #define __INLINE                               inline
#endif
#ifndef   __FORCEINLINE
  #define __FORCEINLINE                          __attribute__((always_inline))
#endif
#ifndef   __STATIC_INLINE
  #define __STATIC_INLINE                        static inline
#endif
#ifndef   __STATIC_FORCEINLINE
  #define __STATIC_FORCEINLINE                   __attribute__((always_inline)) static inline
#endif
#ifndef   __NO_RETURN
  #define __NO_RETURN                            __attribute__((__noreturn__))
#endif
#ifndef   CMSIS_DEPRECATED
  #define CMSIS_DEPRECATED                       __attribute__((deprecated))
#endif
#ifndef   __USED
  #define __USED                                 __attribute__((used))
#endif
#ifndef   __WEAK
  #define __WEAK                                 __attribute__((weak))
#endif
#ifndef   __PACKED
  #define __PACKED                               __attribute__((packed, aligned(1)))
#endif
#ifndef   __PACKED_STRUCT
  #define __PACKED_STRUCT                        struct __attribute__((packed, aligned(1)))
#endif
#ifndef   __UNALIGNED_UINT16_WRITE
  #pragma GCC diagnostic push
  #pragma GCC diagnostic ignored "-Wpacked"
  #pragma GCC diagnostic ignored "-Wattributes"
  __PACKED_STRUCT T_UINT16_WRITE { uint16_t v; };
  #pragma GCC diagnostic pop
  #define __UNALIGNED_UINT16_WRITE(addr, val)    (void)((((struct T_UINT16_WRITE *)(void *)(addr))->v) = (val))
#endif
#ifndef   __UNALIGNED_UINT16_READ
  #pragma GCC diagnostic push
  #pragma GCC diagnostic ignored "-Wpacked"
  #pragma GCC diagnostic ignored "-Wattributes"
  __PACKED_STRUCT T_UINT16_READ { uint16_t v; };
  #pragma GCC diagnostic pop
  #define __UNALIGNED_UINT16_READ(addr)          (((const struct T_UINT16_READ *)(const void *)(addr))->v)
#endif
#ifndef   __UNALIGNED_UINT32_WRITE
  #pragma GCC diagnostic push
  #pragma GCC diagnostic ignored "-Wpacked"
  #pragma GCC diagnostic ignored "-Wattributes"
  __PACKED_STRUCT T_UINT32_WRITE { uint32_t v; };
  #pragma GCC diagnostic pop
  #define __UNALIGNED_UINT32_WRITE(addr, val)    (void)((((struct T_UINT32_WRITE *)(void *)(addr))->v) = (val))
#endif
#ifndef   __UNALIGNED_UINT32_READ
  #pragma GCC diagnostic push
  #pragma GCC diagnostic ignored "-Wpacked"
  #pragma GCC diagnostic ignored "-Wattributes"
  __PACKED_STRUCT T_UINT32_READ { uint32_t v; };
  #pragma GCC diagnostic pop
  #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
  #define __COMPILER_BARRIER()                   __ASM volatile("":::"memory")
#endif


/* ##########################  Core Instruction Access  ######################### */
/**
  \brief   No Operation
  \details No Operation does nothing. This instruction can be used for code alignment purposes.
 */
#define __NOP()                             __ASM volatile ("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()                             __ASM volatile ("wfi":::"memory")


/**
  \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()                             __ASM volatile ("wfe":::"memory")


/**
  \brief   Send Event
  \details Send Event is a hint instruction. It causes an event to be signaled to the CPU.
 */
#define __SEV()                             __ASM volatile ("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.
 */
__STATIC_FORCEINLINE  void __ISB(void)
{
  __ASM volatile ("isb 0xF":::"memory");
}


/**
  \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.
 */
__STATIC_FORCEINLINE  void __DSB(void)
{
  __ASM volatile ("dsb 0xF":::"memory");
}


/**
  \brief   Data Memory Barrier
  \details Ensures the apparent order of the explicit memory operations before
           and after the instruction, without ensuring their completion.
 */
__STATIC_FORCEINLINE  void __DMB(void)
{
  __ASM volatile ("dmb 0xF":::"memory");
}


/**
  \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
 */
__STATIC_FORCEINLINE  uint32_t __REV(uint32_t value)
{
#if (__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 5)
  return __builtin_bswap32(value);
#else
  uint32_t result;

  __ASM ("rev %0, %1" : "=r" (result) : "r" (value) );
  return result;
#endif
}


/**
  \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
 */
__STATIC_FORCEINLINE uint32_t __REV16(uint32_t value)
{
  uint32_t result;
  __ASM ("rev16 %0, %1" : "=r" (result) : "r" (value));
  return result;
}


/**
  \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
 */
__STATIC_FORCEINLINE  int16_t __REVSH(int16_t value)
{
#if (__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 8)
  return (int16_t)__builtin_bswap16(value);
#else
  int16_t result;

  __ASM ("revsh %0, %1" : "=r" (result) : "r" (value) );
  return result;
#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
 */
__STATIC_FORCEINLINE uint32_t __ROR(uint32_t op1, uint32_t op2)
{
  op2 %= 32U;
  if (op2 == 0U)
  {
    return op1;
  }
  return (op1 >> op2) | (op1 << (32U - op2));
}


/**
  \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)   __ASM volatile ("bkpt "#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
 */
__STATIC_FORCEINLINE  uint32_t __RBIT(uint32_t value)
{
  uint32_t result;
   __ASM ("rbit %0, %1" : "=r" (result) : "r" (value) );
  return result;
}


/**
  \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
 */
__STATIC_FORCEINLINE uint8_t __CLZ(uint32_t value)
{
  /* Even though __builtin_clz produces a CLZ instruction on ARM, formally
     __builtin_clz(0) is undefined behaviour, so handle this case specially.
     This guarantees ARM-compatible results if happening to compile on a non-ARM
     target, and ensures the compiler doesn't decide to activate any
     optimisations using the logic "value was passed to __builtin_clz, so it
     is non-zero".
     ARM GCC 7.3 and possibly earlier will optimise this test away, leaving a
     single CLZ instruction.
   */
  if (value == 0U)
  {
    return 32U;
  }
  return __builtin_clz(value);
}

/**
  \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)
 */
__STATIC_FORCEINLINE  uint8_t __LDREXB(volatile uint8_t *addr)
{
    uint32_t result;

#if (__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 8)
   __ASM volatile ("ldrexb %0, %1" : "=r" (result) : "Q" (*addr) );
#else
    /* Prior to GCC 4.8, "Q" will be expanded to [rx, #0] which is not
       accepted by assembler. So has to use following less efficient pattern.
    */
   __ASM volatile ("ldrexb %0, [%1]" : "=r" (result) : "r" (addr) : "memory" );
#endif
   return ((uint8_t) result);    /* Add explicit type cast here */
}


/**
  \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)
 */
__STATIC_FORCEINLINE  uint16_t __LDREXH(volatile uint16_t *addr)
{
    uint32_t result;

#if (__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 8)
   __ASM volatile ("ldrexh %0, %1" : "=r" (result) : "Q" (*addr) );
#else
    /* Prior to GCC 4.8, "Q" will be expanded to [rx, #0] which is not
       accepted by assembler. So has to use following less efficient pattern.
    */
   __ASM volatile ("ldrexh %0, [%1]" : "=r" (result) : "r" (addr) : "memory" );
#endif
   return ((uint16_t) result);    /* Add explicit type cast here */
}


/**
  \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)
 */
__STATIC_FORCEINLINE  uint32_t __LDREXW(volatile uint32_t *addr)
{
    uint32_t result;

   __ASM volatile ("ldrex %0, %1" : "=r" (result) : "Q" (*addr) );
   return(result);
}


/**
  \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
 */
__STATIC_FORCEINLINE  uint32_t __STREXB(uint8_t value, volatile uint8_t *addr)
{
   uint32_t result;

   __ASM volatile ("strexb %0, %2, %1" : "=&r" (result), "=Q" (*addr) : "r" ((uint32_t)value) );
   return(result);
}


/**
  \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
 */
__STATIC_FORCEINLINE  uint32_t __STREXH(uint16_t value, volatile uint16_t *addr)
{
   uint32_t result;

   __ASM volatile ("strexh %0, %2, %1" : "=&r" (result), "=Q" (*addr) : "r" ((uint32_t)value) );
   return(result);
}


/**
  \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
 */
__STATIC_FORCEINLINE  uint32_t __STREXW(uint32_t value, volatile uint32_t *addr)
{
   uint32_t result;

   __ASM volatile ("strex %0, %2, %1" : "=&r" (result), "=Q" (*addr) : "r" (value) );
   return(result);
}


/**
  \brief   Remove the exclusive lock
  \details Removes the exclusive lock which is created by LDREX.
 */
__STATIC_FORCEINLINE  void __CLREX(void)
{
  __ASM volatile ("clrex" ::: "memory");
}

/**
  \brief   Signed Saturate
  \details Saturates a signed value.
  \param [in]  ARG1  Value to be saturated
  \param [in]  ARG2  Bit position to saturate to (1..32)
  \return             Saturated value
 */
#define __SSAT(ARG1, ARG2) \
__extension__ \
({                          \
  int32_t __RES, __ARG1 = (ARG1); \
  __ASM volatile ("ssat %0, %1, %2" : "=r" (__RES) :  "I" (ARG2), "r" (__ARG1) : "cc" ); \
  __RES; \
 })


/**
  \brief   Unsigned Saturate
  \details Saturates an unsigned value.
  \param [in]  ARG1  Value to be saturated
  \param [in]  ARG2  Bit position to saturate to (0..31)
  \return             Saturated value
 */
#define __USAT(ARG1, ARG2) \
__extension__ \
({                          \
  uint32_t __RES, __ARG1 = (ARG1); \
  __ASM volatile ("usat %0, %1, %2" : "=r" (__RES) :  "I" (ARG2), "r" (__ARG1) : "cc" ); \
  __RES; \
 })

/* ###########################  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 special-purpose register PRIMASK.
           Can only be executed in Privileged modes.
 */
__STATIC_FORCEINLINE void __enable_irq(void)
{
  __ASM volatile ("cpsie i" : : : "memory");
}


/**
  \brief   Disable IRQ Interrupts
  \details Disables IRQ interrupts by setting special-purpose register PRIMASK.
  Can only be executed in Privileged modes.
 */
__STATIC_FORCEINLINE void __disable_irq(void)
{
  __ASM volatile ("cpsid i" : : : "memory");
}

/**
  \brief   Enable FIQ
  \details Enables FIQ interrupts by clearing special-purpose register FAULTMASK.
           Can only be executed in Privileged modes.
 */
__STATIC_FORCEINLINE void __enable_fault_irq(void)
{
  __ASM volatile ("cpsie f" : : : "memory");
}


/**
  \brief   Disable FIQ
  \details Disables FIQ interrupts by setting special-purpose register FAULTMASK.
           Can only be executed in Privileged modes.
 */
__STATIC_FORCEINLINE void __disable_fault_irq(void)
{
  __ASM volatile ("cpsid f" : : : "memory");
}

/**
  \brief   Get FPSCR
  \details Returns the current value of the Floating Point Status/Control register.
  \return               Floating Point Status/Control register value
 */
__STATIC_FORCEINLINE  uint32_t __get_FPSCR(void)
{
  #if ((defined (__FPU_PRESENT) && (__FPU_PRESENT == 1U)) && \
       (defined (__FPU_USED   ) && (__FPU_USED    == 1U))     )
  #if __has_builtin(__builtin_arm_get_fpscr) 
  // Re-enable using built-in when GCC has been fixed
  // || (__GNUC__ > 7) || (__GNUC__ == 7 && __GNUC_MINOR__ >= 2)
    /* see https://gcc.gnu.org/ml/gcc-patches/2017-04/msg00443.html */
    return __builtin_arm_get_fpscr();
  #else
    uint32_t result;

    __ASM volatile ("VMRS %0, fpscr" : "=r" (result) );
    return(result);
  #endif
  #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_FORCEINLINE void __set_FPSCR(uint32_t fpscr)
{
  #if ((defined (__FPU_PRESENT) && (__FPU_PRESENT == 1U)) && \
       (defined (__FPU_USED   ) && (__FPU_USED    == 1U))     )
  #if __has_builtin(__builtin_arm_set_fpscr)
  // Re-enable using built-in when GCC has been fixed
  // || (__GNUC__ > 7) || (__GNUC__ == 7 && __GNUC_MINOR__ >= 2)
    /* see https://gcc.gnu.org/ml/gcc-patches/2017-04/msg00443.html */
    __builtin_arm_set_fpscr(fpscr);
  #else
    __ASM volatile ("VMSR fpscr, %0" : : "r" (fpscr) : "vfpcc", "memory");
  #endif
  #else
    (void)fpscr;
  #endif
}


/*@} end of CMSIS_Core_RegAccFunctions */


/* ###################  Compiler specific Intrinsics  ########################### */
/** \defgroup CMSIS_SIMD_intrinsics CMSIS SIMD Intrinsics
  Access to dedicated SIMD instructions
  @{
*/

__STATIC_FORCEINLINE uint32_t __QADD8(uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM ("qadd8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}


__STATIC_FORCEINLINE uint32_t __QSUB8(uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM ("qsub8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}


__STATIC_FORCEINLINE uint32_t __QADD16(uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM ("qadd16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}

__STATIC_FORCEINLINE uint32_t __SHADD16(uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM ("shadd16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}

__STATIC_FORCEINLINE uint32_t __QSUB16(uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM ("qsub16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}

__STATIC_FORCEINLINE uint32_t __SHSUB16(uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM ("shsub16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}

__STATIC_FORCEINLINE uint32_t __QASX(uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM ("qasx %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}

__STATIC_FORCEINLINE uint32_t __SHASX(uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM ("shasx %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}

__STATIC_FORCEINLINE uint32_t __QSAX(uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM ("qsax %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}

__STATIC_FORCEINLINE uint32_t __SHSAX(uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM ("shsax %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}

__STATIC_FORCEINLINE uint32_t __SXTB16(uint32_t op1)
{
  uint32_t result;

  __ASM ("sxtb16 %0, %1" : "=r" (result) : "r" (op1));
  return(result);
}

__STATIC_FORCEINLINE uint32_t __SMUADX (uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM volatile ("smuadx %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}

__STATIC_FORCEINLINE uint32_t __SMLADX (uint32_t op1, uint32_t op2, uint32_t op3)
{
  uint32_t result;

  __ASM volatile ("smladx %0, %1, %2, %3" : "=r" (result) : "r" (op1), "r" (op2), "r" (op3) );
  return(result);
}

__STATIC_FORCEINLINE uint64_t __SMLALD (uint32_t op1, uint32_t op2, uint64_t acc)
{
  union llreg_u{
    uint32_t w32[2];
    uint64_t w64;
  } llr;
  llr.w64 = acc;

#ifndef __ARMEB__   /* Little endian */
  __ASM volatile ("smlald %0, %1, %2, %3" : "=r" (llr.w32[0]), "=r" (llr.w32[1]): "r" (op1), "r" (op2) , "0" (llr.w32[0]), "1" (llr.w32[1]) );
#else               /* Big endian */
  __ASM volatile ("smlald %0, %1, %2, %3" : "=r" (llr.w32[1]), "=r" (llr.w32[0]): "r" (op1), "r" (op2) , "0" (llr.w32[1]), "1" (llr.w32[0]) );
#endif

  return(llr.w64);
}

__STATIC_FORCEINLINE uint64_t __SMLALDX (uint32_t op1, uint32_t op2, uint64_t acc)
{
  union llreg_u{
    uint32_t w32[2];
    uint64_t w64;
  } llr;
  llr.w64 = acc;

#ifndef __ARMEB__   /* Little endian */
  __ASM volatile ("smlaldx %0, %1, %2, %3" : "=r" (llr.w32[0]), "=r" (llr.w32[1]): "r" (op1), "r" (op2) , "0" (llr.w32[0]), "1" (llr.w32[1]) );
#else               /* Big endian */
  __ASM volatile ("smlaldx %0, %1, %2, %3" : "=r" (llr.w32[1]), "=r" (llr.w32[0]): "r" (op1), "r" (op2) , "0" (llr.w32[1]), "1" (llr.w32[0]) );
#endif

  return(llr.w64);
}

__STATIC_FORCEINLINE uint32_t __SMUSD  (uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM volatile ("smusd %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}

__STATIC_FORCEINLINE uint32_t __SMUSDX (uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM volatile ("smusdx %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}

__STATIC_FORCEINLINE uint32_t __SMLSDX (uint32_t op1, uint32_t op2, uint32_t op3)
{
  uint32_t result;

  __ASM volatile ("smlsdx %0, %1, %2, %3" : "=r" (result) : "r" (op1), "r" (op2), "r" (op3) );
  return(result);
}

__STATIC_FORCEINLINE  int32_t __QADD( int32_t op1,  int32_t op2)
{
  int32_t result;

  __ASM volatile ("qadd %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}

__STATIC_FORCEINLINE  int32_t __QSUB( int32_t op1,  int32_t op2)
{
  int32_t result;

  __ASM volatile ("qsub %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}


__STATIC_FORCEINLINE uint32_t __SMUAD  (uint32_t op1, uint32_t op2)
{
  uint32_t result;

  __ASM volatile ("smuad %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
  return(result);
}



#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)  )

__STATIC_FORCEINLINE int32_t __SMMLA (int32_t op1, int32_t op2, int32_t op3)
{
 int32_t result;

 __ASM ("smmla %0, %1, %2, %3" : "=r" (result): "r"  (op1), "r" (op2), "r" (op3) );
 return(result);
}

__STATIC_FORCEINLINE uint32_t __SMLAD (uint32_t op1, uint32_t op2, uint32_t op3)
{
  uint32_t result;

  __ASM volatile ("smlad %0, %1, %2, %3" : "=r" (result) : "r" (op1), "r" (op2), "r" (op3) );
  return(result);
}

/*@} end of group CMSIS_SIMD_intrinsics */



/** \defgroup CMSIS_Core_intrinsics CMSIS Core Intrinsics
  Access to dedicated SIMD instructions
  @{
*/

/** \brief  Get CPSR Register
    \return               CPSR Register value
 */
__STATIC_FORCEINLINE uint32_t __get_CPSR(void)
{
  uint32_t result;
  __ASM volatile("MRS %0, cpsr" : "=r" (result) );
  return(result);
}

/** \brief  Set CPSR Register
    \param [in]    cpsr  CPSR value to set
 */
__STATIC_FORCEINLINE void __set_CPSR(uint32_t cpsr)
{
  __ASM volatile ("MSR cpsr, %0" : : "r" (cpsr) : "cc", "memory");
}

/** \brief  Get Mode
    \return                Processor Mode
 */
__STATIC_FORCEINLINE uint32_t __get_mode(void)
{
  return (__get_CPSR() & 0x1FU);
}

/** \brief  Set Mode
    \param [in]    mode  Mode value to set
 */
__STATIC_FORCEINLINE void __set_mode(uint32_t mode)
{
  __ASM volatile("MSR  cpsr_c, %0" : : "r" (mode) : "memory");
}

/** \brief  Get Stack Pointer
    \return Stack Pointer value
 */
__STATIC_FORCEINLINE uint32_t __get_SP(void)
{
  uint32_t result;
  __ASM volatile("MOV  %0, sp" : "=r" (result) : : "memory");
  return result;
}

/** \brief  Set Stack Pointer
    \param [in]    stack  Stack Pointer value to set
 */
__STATIC_FORCEINLINE void __set_SP(uint32_t stack)
{
  __ASM volatile("MOV  sp, %0" : : "r" (stack) : "memory");
}

/** \brief  Get USR/SYS Stack Pointer
    \return USR/SYS Stack Pointer value
 */
__STATIC_FORCEINLINE uint32_t __get_SP_usr(void)
{
  uint32_t cpsr = __get_CPSR();
  uint32_t result;
  __ASM volatile(
    "CPS     #0x1F  \n"
    "MOV     %0, sp   " : "=r"(result) : : "memory"
   );
  __set_CPSR(cpsr);
  __ISB();
  return result;
}

/** \brief  Set USR/SYS Stack Pointer
    \param [in]    topOfProcStack  USR/SYS Stack Pointer value to set
 */
__STATIC_FORCEINLINE void __set_SP_usr(uint32_t topOfProcStack)
{
  uint32_t cpsr = __get_CPSR();
  __ASM volatile(
    "CPS     #0x1F  \n"
    "MOV     sp, %0   " : : "r" (topOfProcStack) : "memory"
   );
  __set_CPSR(cpsr);
  __ISB();
}

/** \brief  Get FPEXC
    \return               Floating Point Exception Control register value
 */
__STATIC_FORCEINLINE uint32_t __get_FPEXC(void)
{
#if (__FPU_PRESENT == 1)
  uint32_t result;
  __ASM volatile("VMRS %0, fpexc" : "=r" (result) : : "memory");
  return(result);
#else
  return(0);
#endif
}

/** \brief  Set FPEXC
    \param [in]    fpexc  Floating Point Exception Control value to set
 */
__STATIC_FORCEINLINE void __set_FPEXC(uint32_t fpexc)
{
#if (__FPU_PRESENT == 1)
  __ASM volatile ("VMSR fpexc, %0" : : "r" (fpexc) : "memory");
#endif
}

/*
 * Include common core functions to access Coprocessor 15 registers
 */

#define __get_CP(cp, op1, Rt, CRn, CRm, op2) __ASM volatile("MRC p" # cp ", " # op1 ", %0, c" # CRn ", c" # CRm ", " # op2 : "=r" (Rt) : : "memory" )
#define __set_CP(cp, op1, Rt, CRn, CRm, op2) __ASM volatile("MCR p" # cp ", " # op1 ", %0, c" # CRn ", c" # CRm ", " # op2 : : "r" (Rt) : "memory" )
#define __get_CP64(cp, op1, Rt, CRm)         __ASM volatile("MRRC p" # cp ", " # op1 ", %Q0, %R0, c" # CRm  : "=r" (Rt) : : "memory" )
#define __set_CP64(cp, op1, Rt, CRm)         __ASM volatile("MCRR p" # cp ", " # op1 ", %Q0, %R0, c" # CRm  : : "r" (Rt) : "memory" )

#include "cmsis_cp15.h"

/** \brief  Enable Floating Point Unit

  Critical section, called from undef handler, so systick is disabled
 */
__STATIC_INLINE void __FPU_Enable(void)
{
  __ASM volatile(
    //Permit access to VFP/NEON, registers by modifying CPACR
    "        MRC     p15,0,R1,c1,c0,2  \n"
    "        ORR     R1,R1,#0x00F00000 \n"
    "        MCR     p15,0,R1,c1,c0,2  \n"

    //Ensure that subsequent instructions occur in the context of VFP/NEON access permitted
    "        ISB                       \n"

    //Enable VFP/NEON
    "        VMRS    R1,FPEXC          \n"
    "        ORR     R1,R1,#0x40000000 \n"
    "        VMSR    FPEXC,R1          \n"

    //Initialise VFP/NEON registers to 0
    "        MOV     R2,#0             \n"

    //Initialise D16 registers to 0
    "        VMOV    D0, R2,R2         \n"
    "        VMOV    D1, R2,R2         \n"
    "        VMOV    D2, R2,R2         \n"
    "        VMOV    D3, R2,R2         \n"
    "        VMOV    D4, R2,R2         \n"
    "        VMOV    D5, R2,R2         \n"
    "        VMOV    D6, R2,R2         \n"
    "        VMOV    D7, R2,R2         \n"
    "        VMOV    D8, R2,R2         \n"
    "        VMOV    D9, R2,R2         \n"
    "        VMOV    D10,R2,R2         \n"
    "        VMOV    D11,R2,R2         \n"
    "        VMOV    D12,R2,R2         \n"
    "        VMOV    D13,R2,R2         \n"
    "        VMOV    D14,R2,R2         \n"
    "        VMOV    D15,R2,R2         \n"

#if (defined(__ARM_NEON) && (__ARM_NEON == 1))
    //Initialise D32 registers to 0
    "        VMOV    D16,R2,R2         \n"
    "        VMOV    D17,R2,R2         \n"
    "        VMOV    D18,R2,R2         \n"
    "        VMOV    D19,R2,R2         \n"
    "        VMOV    D20,R2,R2         \n"
    "        VMOV    D21,R2,R2         \n"
    "        VMOV    D22,R2,R2         \n"
    "        VMOV    D23,R2,R2         \n"
    "        VMOV    D24,R2,R2         \n"
    "        VMOV    D25,R2,R2         \n"
    "        VMOV    D26,R2,R2         \n"
    "        VMOV    D27,R2,R2         \n"
    "        VMOV    D28,R2,R2         \n"
    "        VMOV    D29,R2,R2         \n"
    "        VMOV    D30,R2,R2         \n"
    "        VMOV    D31,R2,R2         \n"
#endif

    //Initialise FPSCR to a known state
    "        VMRS    R1,FPSCR          \n"
    "        LDR     R2,=0x00086060    \n" //Mask off all bits that do not have to be preserved. Non-preserved bits can/should be zero.
    "        AND     R1,R1,R2          \n"
    "        VMSR    FPSCR,R1            "
    : : : "cc", "r1", "r2"
  );
}

/*@} end of group CMSIS_Core_intrinsics */

#pragma GCC diagnostic pop

#endif /* __CMSIS_GCC_H */