cmsis: Update includes to those of CMSIS v5.6.0

Update CMSIS includes to those of the latest CMSIS
release (v5.6.0) to get the support for ARMv8.1-M
Mainline architecture.

Change-Id: Ifc29a0539989f08d2fb805b72294431b26b82cda
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
diff --git a/platform/ext/cmsis/cmsis_armclang.h b/platform/ext/cmsis/cmsis_armclang.h
index c024ff0..e917f35 100644
--- a/platform/ext/cmsis/cmsis_armclang.h
+++ b/platform/ext/cmsis/cmsis_armclang.h
@@ -1,11 +1,11 @@
 /**************************************************************************//**
  * @file     cmsis_armclang.h
- * @brief    CMSIS compiler ARMCLANG (ARM compiler V6) header file
- * @version  V5.0.3
- * @date     27. March 2017
+ * @brief    CMSIS compiler armclang (Arm Compiler 6) header file
+ * @version  V5.2.0
+ * @date     08. May 2019
  ******************************************************************************/
 /*
- * Copyright (c) 2009-2017 ARM Limited. All rights reserved.
+ * Copyright (c) 2009-2019 Arm Limited. All rights reserved.
  *
  * SPDX-License-Identifier: Apache-2.0
  *
@@ -22,13 +22,15 @@
  * limitations under the License.
  */
 
-//lint -esym(9058, IRQn)   disable MISRA 2012 Rule 2.4 for IRQn
+/*lint -esym(9058, IRQn)*/ /* disable MISRA 2012 Rule 2.4 for IRQn */
 
 #ifndef __CMSIS_ARMCLANG_H
 #define __CMSIS_ARMCLANG_H
 
+#pragma clang system_header   /* treat file as system include file */
+
 #ifndef __ARM_COMPAT_H
-#include <arm_compat.h>    /* Compatibility header for ARM Compiler 5 intrinsics */
+#include <arm_compat.h>    /* Compatibility header for Arm Compiler 5 intrinsics */
 #endif
 
 /* CMSIS compiler specific defines */
@@ -41,8 +43,11 @@
 #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))
+  #define __NO_RETURN                            __attribute__((__noreturn__))
 #endif
 #ifndef   __USED
   #define __USED                                 __attribute__((used))
@@ -56,10 +61,13 @@
 #ifndef   __PACKED_STRUCT
   #define __PACKED_STRUCT                        struct __attribute__((packed, aligned(1)))
 #endif
+#ifndef   __PACKED_UNION
+  #define __PACKED_UNION                         union __attribute__((packed, aligned(1)))
+#endif
 #ifndef   __UNALIGNED_UINT32        /* deprecated */
   #pragma clang diagnostic push
   #pragma clang diagnostic ignored "-Wpacked"
-//lint -esym(9058, T_UINT32)  disable MISRA 2012 Rule 2.4 for T_UINT32
+/*lint -esym(9058, T_UINT32)*/ /* disable MISRA 2012 Rule 2.4 for T_UINT32 */
   struct __attribute__((packed)) T_UINT32 { uint32_t v; };
   #pragma clang diagnostic pop
   #define __UNALIGNED_UINT32(x)                  (((struct T_UINT32 *)(x))->v)
@@ -67,7 +75,7 @@
 #ifndef   __UNALIGNED_UINT16_WRITE
   #pragma clang diagnostic push
   #pragma clang diagnostic ignored "-Wpacked"
-//lint -esym(9058, T_UINT16_WRITE)  disable MISRA 2012 Rule 2.4 for T_UINT16_WRITE
+/*lint -esym(9058, T_UINT16_WRITE)*/ /* disable MISRA 2012 Rule 2.4 for T_UINT16_WRITE */
   __PACKED_STRUCT T_UINT16_WRITE { uint16_t v; };
   #pragma clang diagnostic pop
   #define __UNALIGNED_UINT16_WRITE(addr, val)    (void)((((struct T_UINT16_WRITE *)(void *)(addr))->v) = (val))
@@ -75,7 +83,7 @@
 #ifndef   __UNALIGNED_UINT16_READ
   #pragma clang diagnostic push
   #pragma clang diagnostic ignored "-Wpacked"
-//lint -esym(9058, T_UINT16_READ)  disable MISRA 2012 Rule 2.4 for T_UINT16_READ
+/*lint -esym(9058, T_UINT16_READ)*/ /* disable MISRA 2012 Rule 2.4 for T_UINT16_READ */
   __PACKED_STRUCT T_UINT16_READ { uint16_t v; };
   #pragma clang diagnostic pop
   #define __UNALIGNED_UINT16_READ(addr)          (((const struct T_UINT16_READ *)(const void *)(addr))->v)
@@ -83,7 +91,7 @@
 #ifndef   __UNALIGNED_UINT32_WRITE
   #pragma clang diagnostic push
   #pragma clang diagnostic ignored "-Wpacked"
-//lint -esym(9058, T_UINT32_WRITE)  disable MISRA 2012 Rule 2.4 for T_UINT32_WRITE
+/*lint -esym(9058, T_UINT32_WRITE)*/ /* disable MISRA 2012 Rule 2.4 for T_UINT32_WRITE */
   __PACKED_STRUCT T_UINT32_WRITE { uint32_t v; };
   #pragma clang diagnostic pop
   #define __UNALIGNED_UINT32_WRITE(addr, val)    (void)((((struct T_UINT32_WRITE *)(void *)(addr))->v) = (val))
@@ -91,7 +99,7 @@
 #ifndef   __UNALIGNED_UINT32_READ
   #pragma clang diagnostic push
   #pragma clang diagnostic ignored "-Wpacked"
-//lint -esym(9058, T_UINT32_READ)  disable MISRA 2012 Rule 2.4 for T_UINT32_READ
+/*lint -esym(9058, T_UINT32_READ)*/ /* disable MISRA 2012 Rule 2.4 for T_UINT32_READ */
   __PACKED_STRUCT T_UINT32_READ { uint32_t v; };
   #pragma clang diagnostic pop
   #define __UNALIGNED_UINT32_READ(addr)          (((const struct T_UINT32_READ *)(const void *)(addr))->v)
@@ -99,7 +107,34 @@
 #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
 
+/* #########################  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
@@ -128,7 +163,7 @@
   \details Returns the content of the Control Register.
   \return               Control Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __get_CONTROL(void)
+__STATIC_FORCEINLINE uint32_t __get_CONTROL(void)
 {
   uint32_t result;
 
@@ -143,7 +178,7 @@
   \details Returns the content of the non-secure Control Register when in secure mode.
   \return               non-secure Control Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __TZ_get_CONTROL_NS(void)
+__STATIC_FORCEINLINE uint32_t __TZ_get_CONTROL_NS(void)
 {
   uint32_t result;
 
@@ -158,7 +193,7 @@
   \details Writes the given value to the Control Register.
   \param [in]    control  Control Register value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __set_CONTROL(uint32_t control)
+__STATIC_FORCEINLINE void __set_CONTROL(uint32_t control)
 {
   __ASM volatile ("MSR control, %0" : : "r" (control) : "memory");
 }
@@ -170,7 +205,7 @@
   \details Writes the given value to the non-secure Control Register when in secure state.
   \param [in]    control  Control Register value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __TZ_set_CONTROL_NS(uint32_t control)
+__STATIC_FORCEINLINE void __TZ_set_CONTROL_NS(uint32_t control)
 {
   __ASM volatile ("MSR control_ns, %0" : : "r" (control) : "memory");
 }
@@ -182,7 +217,7 @@
   \details Returns the content of the IPSR Register.
   \return               IPSR Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __get_IPSR(void)
+__STATIC_FORCEINLINE uint32_t __get_IPSR(void)
 {
   uint32_t result;
 
@@ -196,7 +231,7 @@
   \details Returns the content of the APSR Register.
   \return               APSR Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __get_APSR(void)
+__STATIC_FORCEINLINE uint32_t __get_APSR(void)
 {
   uint32_t result;
 
@@ -210,7 +245,7 @@
   \details Returns the content of the xPSR Register.
   \return               xPSR Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __get_xPSR(void)
+__STATIC_FORCEINLINE uint32_t __get_xPSR(void)
 {
   uint32_t result;
 
@@ -224,9 +259,9 @@
   \details Returns the current value of the Process Stack Pointer (PSP).
   \return               PSP Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __get_PSP(void)
+__STATIC_FORCEINLINE uint32_t __get_PSP(void)
 {
-  register uint32_t result;
+  uint32_t result;
 
   __ASM volatile ("MRS %0, psp"  : "=r" (result) );
   return(result);
@@ -239,9 +274,9 @@
   \details Returns the current value of the non-secure Process Stack Pointer (PSP) when in secure state.
   \return               PSP Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __TZ_get_PSP_NS(void)
+__STATIC_FORCEINLINE uint32_t __TZ_get_PSP_NS(void)
 {
-  register uint32_t result;
+  uint32_t result;
 
   __ASM volatile ("MRS %0, psp_ns"  : "=r" (result) );
   return(result);
@@ -254,7 +289,7 @@
   \details Assigns the given value to the Process Stack Pointer (PSP).
   \param [in]    topOfProcStack  Process Stack Pointer value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __set_PSP(uint32_t topOfProcStack)
+__STATIC_FORCEINLINE void __set_PSP(uint32_t topOfProcStack)
 {
   __ASM volatile ("MSR psp, %0" : : "r" (topOfProcStack) : );
 }
@@ -266,7 +301,7 @@
   \details Assigns the given value to the non-secure Process Stack Pointer (PSP) when in secure state.
   \param [in]    topOfProcStack  Process Stack Pointer value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __TZ_set_PSP_NS(uint32_t topOfProcStack)
+__STATIC_FORCEINLINE void __TZ_set_PSP_NS(uint32_t topOfProcStack)
 {
   __ASM volatile ("MSR psp_ns, %0" : : "r" (topOfProcStack) : );
 }
@@ -278,9 +313,9 @@
   \details Returns the current value of the Main Stack Pointer (MSP).
   \return               MSP Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __get_MSP(void)
+__STATIC_FORCEINLINE uint32_t __get_MSP(void)
 {
-  register uint32_t result;
+  uint32_t result;
 
   __ASM volatile ("MRS %0, msp" : "=r" (result) );
   return(result);
@@ -293,9 +328,9 @@
   \details Returns the current value of the non-secure Main Stack Pointer (MSP) when in secure state.
   \return               MSP Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __TZ_get_MSP_NS(void)
+__STATIC_FORCEINLINE uint32_t __TZ_get_MSP_NS(void)
 {
-  register uint32_t result;
+  uint32_t result;
 
   __ASM volatile ("MRS %0, msp_ns" : "=r" (result) );
   return(result);
@@ -308,7 +343,7 @@
   \details Assigns the given value to the Main Stack Pointer (MSP).
   \param [in]    topOfMainStack  Main Stack Pointer value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __set_MSP(uint32_t topOfMainStack)
+__STATIC_FORCEINLINE void __set_MSP(uint32_t topOfMainStack)
 {
   __ASM volatile ("MSR msp, %0" : : "r" (topOfMainStack) : );
 }
@@ -320,7 +355,7 @@
   \details Assigns the given value to the non-secure Main Stack Pointer (MSP) when in secure state.
   \param [in]    topOfMainStack  Main Stack Pointer value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __TZ_set_MSP_NS(uint32_t topOfMainStack)
+__STATIC_FORCEINLINE void __TZ_set_MSP_NS(uint32_t topOfMainStack)
 {
   __ASM volatile ("MSR msp_ns, %0" : : "r" (topOfMainStack) : );
 }
@@ -333,9 +368,9 @@
   \details Returns the current value of the non-secure Stack Pointer (SP) when in secure state.
   \return               SP Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __TZ_get_SP_NS(void)
+__STATIC_FORCEINLINE uint32_t __TZ_get_SP_NS(void)
 {
-  register uint32_t result;
+  uint32_t result;
 
   __ASM volatile ("MRS %0, sp_ns" : "=r" (result) );
   return(result);
@@ -347,7 +382,7 @@
   \details Assigns the given value to the non-secure Stack Pointer (SP) when in secure state.
   \param [in]    topOfStack  Stack Pointer value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __TZ_set_SP_NS(uint32_t topOfStack)
+__STATIC_FORCEINLINE void __TZ_set_SP_NS(uint32_t topOfStack)
 {
   __ASM volatile ("MSR sp_ns, %0" : : "r" (topOfStack) : );
 }
@@ -359,7 +394,7 @@
   \details Returns the current state of the priority mask bit from the Priority Mask Register.
   \return               Priority Mask value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __get_PRIMASK(void)
+__STATIC_FORCEINLINE uint32_t __get_PRIMASK(void)
 {
   uint32_t result;
 
@@ -374,7 +409,7 @@
   \details Returns the current state of the non-secure priority mask bit from the Priority Mask Register when in secure state.
   \return               Priority Mask value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __TZ_get_PRIMASK_NS(void)
+__STATIC_FORCEINLINE uint32_t __TZ_get_PRIMASK_NS(void)
 {
   uint32_t result;
 
@@ -389,7 +424,7 @@
   \details Assigns the given value to the Priority Mask Register.
   \param [in]    priMask  Priority Mask
  */
-__attribute__((always_inline)) __STATIC_INLINE void __set_PRIMASK(uint32_t priMask)
+__STATIC_FORCEINLINE void __set_PRIMASK(uint32_t priMask)
 {
   __ASM volatile ("MSR primask, %0" : : "r" (priMask) : "memory");
 }
@@ -401,7 +436,7 @@
   \details Assigns the given value to the non-secure Priority Mask Register when in secure state.
   \param [in]    priMask  Priority Mask
  */
-__attribute__((always_inline)) __STATIC_INLINE void __TZ_set_PRIMASK_NS(uint32_t priMask)
+__STATIC_FORCEINLINE void __TZ_set_PRIMASK_NS(uint32_t priMask)
 {
   __ASM volatile ("MSR primask_ns, %0" : : "r" (priMask) : "memory");
 }
@@ -432,7 +467,7 @@
   \details Returns the current value of the Base Priority register.
   \return               Base Priority register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __get_BASEPRI(void)
+__STATIC_FORCEINLINE uint32_t __get_BASEPRI(void)
 {
   uint32_t result;
 
@@ -447,7 +482,7 @@
   \details Returns the current value of the non-secure Base Priority register when in secure state.
   \return               Base Priority register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __TZ_get_BASEPRI_NS(void)
+__STATIC_FORCEINLINE uint32_t __TZ_get_BASEPRI_NS(void)
 {
   uint32_t result;
 
@@ -462,7 +497,7 @@
   \details Assigns the given value to the Base Priority register.
   \param [in]    basePri  Base Priority value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __set_BASEPRI(uint32_t basePri)
+__STATIC_FORCEINLINE void __set_BASEPRI(uint32_t basePri)
 {
   __ASM volatile ("MSR basepri, %0" : : "r" (basePri) : "memory");
 }
@@ -474,7 +509,7 @@
   \details Assigns the given value to the non-secure Base Priority register when in secure state.
   \param [in]    basePri  Base Priority value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __TZ_set_BASEPRI_NS(uint32_t basePri)
+__STATIC_FORCEINLINE void __TZ_set_BASEPRI_NS(uint32_t basePri)
 {
   __ASM volatile ("MSR basepri_ns, %0" : : "r" (basePri) : "memory");
 }
@@ -487,7 +522,7 @@
            or the new value increases the BASEPRI priority level.
   \param [in]    basePri  Base Priority value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __set_BASEPRI_MAX(uint32_t basePri)
+__STATIC_FORCEINLINE void __set_BASEPRI_MAX(uint32_t basePri)
 {
   __ASM volatile ("MSR basepri_max, %0" : : "r" (basePri) : "memory");
 }
@@ -498,7 +533,7 @@
   \details Returns the current value of the Fault Mask register.
   \return               Fault Mask register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __get_FAULTMASK(void)
+__STATIC_FORCEINLINE uint32_t __get_FAULTMASK(void)
 {
   uint32_t result;
 
@@ -513,7 +548,7 @@
   \details Returns the current value of the non-secure Fault Mask register when in secure state.
   \return               Fault Mask register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __TZ_get_FAULTMASK_NS(void)
+__STATIC_FORCEINLINE uint32_t __TZ_get_FAULTMASK_NS(void)
 {
   uint32_t result;
 
@@ -528,7 +563,7 @@
   \details Assigns the given value to the Fault Mask register.
   \param [in]    faultMask  Fault Mask value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __set_FAULTMASK(uint32_t faultMask)
+__STATIC_FORCEINLINE void __set_FAULTMASK(uint32_t faultMask)
 {
   __ASM volatile ("MSR faultmask, %0" : : "r" (faultMask) : "memory");
 }
@@ -540,7 +575,7 @@
   \details Assigns the given value to the non-secure Fault Mask register when in secure state.
   \param [in]    faultMask  Fault Mask value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __TZ_set_FAULTMASK_NS(uint32_t faultMask)
+__STATIC_FORCEINLINE void __TZ_set_FAULTMASK_NS(uint32_t faultMask)
 {
   __ASM volatile ("MSR faultmask_ns, %0" : : "r" (faultMask) : "memory");
 }
@@ -556,162 +591,204 @@
 
 /**
   \brief   Get Process Stack Pointer Limit
+  Devices without ARMv8-M Main Extensions (i.e. Cortex-M23) lack the non-secure
+  Stack Pointer Limit register hence zero is returned always in non-secure
+  mode.
+  
   \details Returns the current value of the Process Stack Pointer Limit (PSPLIM).
   \return               PSPLIM Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __get_PSPLIM(void)
+__STATIC_FORCEINLINE uint32_t __get_PSPLIM(void)
 {
-  register uint32_t result;
-
+#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
+  return 0U;
+#else
+  uint32_t result;
   __ASM volatile ("MRS %0, psplim"  : "=r" (result) );
-  return(result);
+  return result;
+#endif
 }
 
-
-#if ((defined (__ARM_FEATURE_CMSE  ) && (__ARM_FEATURE_CMSE   == 3)) && \
-     (defined (__ARM_ARCH_8M_MAIN__) && (__ARM_ARCH_8M_MAIN__ == 1))    )
+#if (defined (__ARM_FEATURE_CMSE) && (__ARM_FEATURE_CMSE == 3))
 /**
   \brief   Get Process Stack Pointer Limit (non-secure)
+  Devices without ARMv8-M Main Extensions (i.e. Cortex-M23) lack the non-secure
+  Stack Pointer Limit register hence zero is returned always in non-secure
+  mode.
+
   \details Returns the current value of the non-secure Process Stack Pointer Limit (PSPLIM) when in secure state.
   \return               PSPLIM Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __TZ_get_PSPLIM_NS(void)
+__STATIC_FORCEINLINE uint32_t __TZ_get_PSPLIM_NS(void)
 {
-  register uint32_t result;
-
+#if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)))
+  // without main extensions, the non-secure PSPLIM is RAZ/WI
+  return 0U;
+#else
+  uint32_t result;
   __ASM volatile ("MRS %0, psplim_ns"  : "=r" (result) );
-  return(result);
+  return result;
+#endif
 }
 #endif
 
 
 /**
   \brief   Set Process Stack Pointer Limit
+  Devices without ARMv8-M Main Extensions (i.e. Cortex-M23) lack the non-secure
+  Stack Pointer Limit register hence the write is silently ignored in non-secure
+  mode.
+  
   \details Assigns the given value to the Process Stack Pointer Limit (PSPLIM).
   \param [in]    ProcStackPtrLimit  Process Stack Pointer Limit value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __set_PSPLIM(uint32_t ProcStackPtrLimit)
+__STATIC_FORCEINLINE void __set_PSPLIM(uint32_t ProcStackPtrLimit)
 {
+#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)ProcStackPtrLimit;
+#else
   __ASM volatile ("MSR psplim, %0" : : "r" (ProcStackPtrLimit));
+#endif
 }
 
 
-#if ((defined (__ARM_FEATURE_CMSE  ) && (__ARM_FEATURE_CMSE   == 3)) && \
-     (defined (__ARM_ARCH_8M_MAIN__) && (__ARM_ARCH_8M_MAIN__ == 1))    )
+#if (defined (__ARM_FEATURE_CMSE  ) && (__ARM_FEATURE_CMSE   == 3))
 /**
   \brief   Set Process Stack Pointer (non-secure)
+  Devices without ARMv8-M Main Extensions (i.e. Cortex-M23) lack the non-secure
+  Stack Pointer Limit register hence the write is silently ignored in non-secure
+  mode.
+
   \details Assigns the given value to the non-secure Process Stack Pointer Limit (PSPLIM) when in secure state.
   \param [in]    ProcStackPtrLimit  Process Stack Pointer Limit value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __TZ_set_PSPLIM_NS(uint32_t ProcStackPtrLimit)
+__STATIC_FORCEINLINE void __TZ_set_PSPLIM_NS(uint32_t ProcStackPtrLimit)
 {
+#if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)))
+  // without main extensions, the non-secure PSPLIM is RAZ/WI
+  (void)ProcStackPtrLimit;
+#else
   __ASM volatile ("MSR psplim_ns, %0\n" : : "r" (ProcStackPtrLimit));
+#endif
 }
 #endif
 
 
 /**
   \brief   Get Main Stack Pointer Limit
+  Devices without ARMv8-M Main Extensions (i.e. Cortex-M23) lack the non-secure
+  Stack Pointer Limit register hence zero is returned always.
+
   \details Returns the current value of the Main Stack Pointer Limit (MSPLIM).
   \return               MSPLIM Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __get_MSPLIM(void)
+__STATIC_FORCEINLINE uint32_t __get_MSPLIM(void)
 {
-  register uint32_t result;
-
+#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
+  return 0U;
+#else
+  uint32_t result;
   __ASM volatile ("MRS %0, msplim" : "=r" (result) );
-
-  return(result);
+  return result;
+#endif
 }
 
 
-#if ((defined (__ARM_FEATURE_CMSE  ) && (__ARM_FEATURE_CMSE   == 3)) && \
-     (defined (__ARM_ARCH_8M_MAIN__) && (__ARM_ARCH_8M_MAIN__ == 1))    )
+#if (defined (__ARM_FEATURE_CMSE  ) && (__ARM_FEATURE_CMSE   == 3))
 /**
   \brief   Get Main Stack Pointer Limit (non-secure)
+  Devices without ARMv8-M Main Extensions (i.e. Cortex-M23) lack the non-secure
+  Stack Pointer Limit register hence zero is returned always.
+
   \details Returns the current value of the non-secure Main Stack Pointer Limit(MSPLIM) when in secure state.
   \return               MSPLIM Register value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __TZ_get_MSPLIM_NS(void)
+__STATIC_FORCEINLINE uint32_t __TZ_get_MSPLIM_NS(void)
 {
-  register uint32_t result;
-
+#if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)))
+  // without main extensions, the non-secure MSPLIM is RAZ/WI
+  return 0U;
+#else
+  uint32_t result;
   __ASM volatile ("MRS %0, msplim_ns" : "=r" (result) );
-  return(result);
+  return result;
+#endif
 }
 #endif
 
 
 /**
   \brief   Set Main Stack Pointer Limit
+  Devices without ARMv8-M Main Extensions (i.e. Cortex-M23) lack the non-secure
+  Stack Pointer Limit register hence the write is silently ignored.
+
   \details Assigns the given value to the Main Stack Pointer Limit (MSPLIM).
   \param [in]    MainStackPtrLimit  Main Stack Pointer Limit value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __set_MSPLIM(uint32_t MainStackPtrLimit)
+__STATIC_FORCEINLINE void __set_MSPLIM(uint32_t MainStackPtrLimit)
 {
+#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)MainStackPtrLimit;
+#else
   __ASM volatile ("MSR msplim, %0" : : "r" (MainStackPtrLimit));
+#endif
 }
 
 
-#if ((defined (__ARM_FEATURE_CMSE  ) && (__ARM_FEATURE_CMSE   == 3)) && \
-     (defined (__ARM_ARCH_8M_MAIN__) && (__ARM_ARCH_8M_MAIN__ == 1))    )
+#if (defined (__ARM_FEATURE_CMSE  ) && (__ARM_FEATURE_CMSE   == 3))
 /**
   \brief   Set Main Stack Pointer Limit (non-secure)
+  Devices without ARMv8-M Main Extensions (i.e. Cortex-M23) lack the non-secure
+  Stack Pointer Limit register hence the write is silently ignored.
+
   \details Assigns the given value to the non-secure Main Stack Pointer Limit (MSPLIM) when in secure state.
   \param [in]    MainStackPtrLimit  Main Stack Pointer value to set
  */
-__attribute__((always_inline)) __STATIC_INLINE void __TZ_set_MSPLIM_NS(uint32_t MainStackPtrLimit)
+__STATIC_FORCEINLINE void __TZ_set_MSPLIM_NS(uint32_t MainStackPtrLimit)
 {
+#if (!(defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)))
+  // without main extensions, the non-secure MSPLIM is RAZ/WI
+  (void)MainStackPtrLimit;
+#else
   __ASM volatile ("MSR msplim_ns, %0" : : "r" (MainStackPtrLimit));
+#endif
 }
 #endif
 
 #endif /* ((defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1)) || \
            (defined (__ARM_ARCH_8M_BASE__ ) && (__ARM_ARCH_8M_BASE__ == 1))    ) */
 
-
-#if ((defined (__ARM_ARCH_7EM__     ) && (__ARM_ARCH_7EM__     == 1)) || \
-     (defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1))    )
-
 /**
   \brief   Get FPSCR
   \details Returns the current value of the Floating Point Status/Control register.
   \return               Floating Point Status/Control register value
  */
-/* #define __get_FPSCR      __builtin_arm_get_fpscr */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __get_FPSCR(void)
-{
 #if ((defined (__FPU_PRESENT) && (__FPU_PRESENT == 1U)) && \
      (defined (__FPU_USED   ) && (__FPU_USED    == 1U))     )
-  uint32_t result;
-
-  __ASM volatile ("VMRS %0, fpscr" : "=r" (result) );
-  return(result);
+#define __get_FPSCR      (uint32_t)__builtin_arm_get_fpscr
 #else
-  return(0U);
+#define __get_FPSCR()      ((uint32_t)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
  */
-/* #define __set_FPSCR      __builtin_arm_set_fpscr */
-__attribute__((always_inline)) __STATIC_INLINE void __set_FPSCR(uint32_t fpscr)
-{
 #if ((defined (__FPU_PRESENT) && (__FPU_PRESENT == 1U)) && \
      (defined (__FPU_USED   ) && (__FPU_USED    == 1U))     )
-  __ASM volatile ("VMSR fpscr, %0" : : "r" (fpscr) : "memory");
+#define __set_FPSCR      __builtin_arm_set_fpscr
 #else
-  (void)fpscr;
+#define __set_FPSCR(x)      ((void)(x))
 #endif
-}
-
-#endif /* ((defined (__ARM_ARCH_7EM__     ) && (__ARM_ARCH_7EM__     == 1)) || \
-           (defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1))    ) */
-
 
 
 /*@} end of CMSIS_Core_RegAccFunctions */
@@ -728,9 +805,11 @@
  * Otherwise, use general registers, specified by constraint "r" */
 #if defined (__thumb__) && !defined (__thumb2__)
 #define __CMSIS_GCC_OUT_REG(r) "=l" (r)
+#define __CMSIS_GCC_RW_REG(r) "+l" (r)
 #define __CMSIS_GCC_USE_REG(r) "l" (r)
 #else
 #define __CMSIS_GCC_OUT_REG(r) "=r" (r)
+#define __CMSIS_GCC_RW_REG(r) "+r" (r)
 #define __CMSIS_GCC_USE_REG(r) "r" (r)
 #endif
 
@@ -768,14 +847,14 @@
            so that all instructions following the ISB are fetched from cache or memory,
            after the instruction has been completed.
  */
-#define __ISB()        __builtin_arm_isb(0xF);
+#define __ISB()        __builtin_arm_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()        __builtin_arm_dsb(0xF);
+#define __DSB()        __builtin_arm_dsb(0xF)
 
 
 /**
@@ -783,50 +862,34 @@
   \details Ensures the apparent order of the explicit memory operations before
            and after the instruction, without ensuring their completion.
  */
-#define __DMB()        __builtin_arm_dmb(0xF);
+#define __DMB()        __builtin_arm_dmb(0xF)
 
 
 /**
   \brief   Reverse byte order (32 bit)
-  \details Reverses the byte order in integer value.
+  \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          __builtin_bswap32
+#define __REV(value)   __builtin_bswap32(value)
 
 
 /**
   \brief   Reverse byte order (16 bit)
-  \details Reverses the byte order in two unsigned short values.
+  \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
  */
-#define __REV16          __builtin_bswap16                /* ToDo ARMCLANG: check if __builtin_bswap16 could be used */
-#if 0
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __REV16(uint32_t value)
-{
-  uint32_t result;
-
-  __ASM volatile ("rev16 %0, %1" : __CMSIS_GCC_OUT_REG (result) : __CMSIS_GCC_USE_REG (value) );
-  return(result);
-}
-#endif
+#define __REV16(value) __ROR(__REV(value), 16)
 
 
 /**
-  \brief   Reverse byte order in signed short value
-  \details Reverses the byte order in a signed short value with sign extension to integer.
+  \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
  */
-                                                          /* ToDo ARMCLANG: check if __builtin_bswap16 could be used */
-__attribute__((always_inline)) __STATIC_INLINE int32_t __REVSH(int32_t value)
-{
-  int32_t result;
-
-  __ASM volatile ("revsh %0, %1" : __CMSIS_GCC_OUT_REG (result) : __CMSIS_GCC_USE_REG (value) );
-  return(result);
-}
+#define __REVSH(value) (int16_t)__builtin_bswap16(value)
 
 
 /**
@@ -836,8 +899,13 @@
   \param [in]    op2  Number of Bits to rotate
   \return               Rotated value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __ROR(uint32_t op1, uint32_t op2)
+__STATIC_FORCEINLINE uint32_t __ROR(uint32_t op1, uint32_t op2)
 {
+  op2 %= 32U;
+  if (op2 == 0U)
+  {
+    return op1;
+  }
   return (op1 >> op2) | (op1 << (32U - op2));
 }
 
@@ -849,7 +917,7 @@
   \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)
+#define __BKPT(value)     __ASM volatile ("bkpt "#value)
 
 
 /**
@@ -858,30 +926,7 @@
   \param [in]    value  Value to reverse
   \return               Reversed value
  */
-                                                          /* ToDo ARMCLANG: check if __builtin_arm_rbit is supported */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __RBIT(uint32_t value)
-{
-  uint32_t result;
-
-#if ((defined (__ARM_ARCH_7M__      ) && (__ARM_ARCH_7M__      == 1)) || \
-     (defined (__ARM_ARCH_7EM__     ) && (__ARM_ARCH_7EM__     == 1)) || \
-     (defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1))    )
-   __ASM volatile ("rbit %0, %1" : "=r" (result) : "r" (value) );
-#else
-  int32_t s = (4 /*sizeof(v)*/ * 8) - 1; /* extra shift needed at end */
-
-  result = value;                      /* r will be reversed bits of v; first get LSB of v */
-  for (value >>= 1U; value; value >>= 1U)
-  {
-    result <<= 1U;
-    result |= value & 1U;
-    s--;
-  }
-  result <<= s;                        /* shift when v's highest bits are zero */
-#endif
-  return(result);
-}
-
+#define __RBIT            __builtin_arm_rbit
 
 /**
   \brief   Count leading zeros
@@ -889,7 +934,23 @@
   \param [in]  value  Value to count the leading zeros
   \return             number of leading zeros in value
  */
-#define __CLZ             __builtin_clz
+__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 Compiler 6.10 and possibly earlier will optimise this test away, leaving a
+     single CLZ instruction.
+   */
+  if (value == 0U)
+  {
+    return 32U;
+  }
+  return __builtin_clz(value);
+}
 
 
 #if ((defined (__ARM_ARCH_7M__      ) && (__ARM_ARCH_7M__      == 1)) || \
@@ -971,6 +1032,7 @@
 #if ((defined (__ARM_ARCH_7M__      ) && (__ARM_ARCH_7M__      == 1)) || \
      (defined (__ARM_ARCH_7EM__     ) && (__ARM_ARCH_7EM__     == 1)) || \
      (defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1))    )
+
 /**
   \brief   Signed Saturate
   \details Saturates a signed value.
@@ -998,7 +1060,7 @@
   \param [in]    value  Value to rotate
   \return               Rotated value
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __RRX(uint32_t value)
+__STATIC_FORCEINLINE uint32_t __RRX(uint32_t value)
 {
   uint32_t result;
 
@@ -1013,7 +1075,7 @@
   \param [in]    ptr  Pointer to data
   \return             value of type uint8_t at (*ptr)
  */
-__attribute__((always_inline)) __STATIC_INLINE uint8_t __LDRBT(volatile uint8_t *ptr)
+__STATIC_FORCEINLINE uint8_t __LDRBT(volatile uint8_t *ptr)
 {
   uint32_t result;
 
@@ -1028,7 +1090,7 @@
   \param [in]    ptr  Pointer to data
   \return        value of type uint16_t at (*ptr)
  */
-__attribute__((always_inline)) __STATIC_INLINE uint16_t __LDRHT(volatile uint16_t *ptr)
+__STATIC_FORCEINLINE uint16_t __LDRHT(volatile uint16_t *ptr)
 {
   uint32_t result;
 
@@ -1043,7 +1105,7 @@
   \param [in]    ptr  Pointer to data
   \return        value of type uint32_t at (*ptr)
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __LDRT(volatile uint32_t *ptr)
+__STATIC_FORCEINLINE uint32_t __LDRT(volatile uint32_t *ptr)
 {
   uint32_t result;
 
@@ -1058,7 +1120,7 @@
   \param [in]  value  Value to store
   \param [in]    ptr  Pointer to location
  */
-__attribute__((always_inline)) __STATIC_INLINE void __STRBT(uint8_t value, volatile uint8_t *ptr)
+__STATIC_FORCEINLINE void __STRBT(uint8_t value, volatile uint8_t *ptr)
 {
   __ASM volatile ("strbt %1, %0" : "=Q" (*ptr) : "r" ((uint32_t)value) );
 }
@@ -1070,7 +1132,7 @@
   \param [in]  value  Value to store
   \param [in]    ptr  Pointer to location
  */
-__attribute__((always_inline)) __STATIC_INLINE void __STRHT(uint16_t value, volatile uint16_t *ptr)
+__STATIC_FORCEINLINE void __STRHT(uint16_t value, volatile uint16_t *ptr)
 {
   __ASM volatile ("strht %1, %0" : "=Q" (*ptr) : "r" ((uint32_t)value) );
 }
@@ -1082,11 +1144,64 @@
   \param [in]  value  Value to store
   \param [in]    ptr  Pointer to location
  */
-__attribute__((always_inline)) __STATIC_INLINE void __STRT(uint32_t value, volatile uint32_t *ptr)
+__STATIC_FORCEINLINE void __STRT(uint32_t value, volatile uint32_t *ptr)
 {
   __ASM volatile ("strt %1, %0" : "=Q" (*ptr) : "r" (value) );
 }
 
+#else  /* ((defined (__ARM_ARCH_7M__      ) && (__ARM_ARCH_7M__      == 1)) || \
+           (defined (__ARM_ARCH_7EM__     ) && (__ARM_ARCH_7EM__     == 1)) || \
+           (defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 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
+ */
+__STATIC_FORCEINLINE 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
+ */
+__STATIC_FORCEINLINE 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)) || \
            (defined (__ARM_ARCH_8M_MAIN__ ) && (__ARM_ARCH_8M_MAIN__ == 1))    ) */
@@ -1100,7 +1215,7 @@
   \param [in]    ptr  Pointer to data
   \return             value of type uint8_t at (*ptr)
  */
-__attribute__((always_inline)) __STATIC_INLINE uint8_t __LDAB(volatile uint8_t *ptr)
+__STATIC_FORCEINLINE uint8_t __LDAB(volatile uint8_t *ptr)
 {
   uint32_t result;
 
@@ -1115,7 +1230,7 @@
   \param [in]    ptr  Pointer to data
   \return        value of type uint16_t at (*ptr)
  */
-__attribute__((always_inline)) __STATIC_INLINE uint16_t __LDAH(volatile uint16_t *ptr)
+__STATIC_FORCEINLINE uint16_t __LDAH(volatile uint16_t *ptr)
 {
   uint32_t result;
 
@@ -1130,7 +1245,7 @@
   \param [in]    ptr  Pointer to data
   \return        value of type uint32_t at (*ptr)
  */
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __LDA(volatile uint32_t *ptr)
+__STATIC_FORCEINLINE uint32_t __LDA(volatile uint32_t *ptr)
 {
   uint32_t result;
 
@@ -1145,7 +1260,7 @@
   \param [in]  value  Value to store
   \param [in]    ptr  Pointer to location
  */
-__attribute__((always_inline)) __STATIC_INLINE void __STLB(uint8_t value, volatile uint8_t *ptr)
+__STATIC_FORCEINLINE void __STLB(uint8_t value, volatile uint8_t *ptr)
 {
   __ASM volatile ("stlb %1, %0" : "=Q" (*ptr) : "r" ((uint32_t)value) );
 }
@@ -1157,7 +1272,7 @@
   \param [in]  value  Value to store
   \param [in]    ptr  Pointer to location
  */
-__attribute__((always_inline)) __STATIC_INLINE void __STLH(uint16_t value, volatile uint16_t *ptr)
+__STATIC_FORCEINLINE void __STLH(uint16_t value, volatile uint16_t *ptr)
 {
   __ASM volatile ("stlh %1, %0" : "=Q" (*ptr) : "r" ((uint32_t)value) );
 }
@@ -1169,7 +1284,7 @@
   \param [in]  value  Value to store
   \param [in]    ptr  Pointer to location
  */
-__attribute__((always_inline)) __STATIC_INLINE void __STL(uint32_t value, volatile uint32_t *ptr)
+__STATIC_FORCEINLINE void __STL(uint32_t value, volatile uint32_t *ptr)
 {
   __ASM volatile ("stl %1, %0" : "=Q" (*ptr) : "r" ((uint32_t)value) );
 }
@@ -1248,532 +1363,65 @@
 
 #if (defined (__ARM_FEATURE_DSP) && (__ARM_FEATURE_DSP == 1))
 
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SADD8(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("sadd8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __QADD8(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("qadd8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SHADD8(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("shadd8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UADD8(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uadd8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UQADD8(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uqadd8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UHADD8(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uhadd8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SSUB8(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("ssub8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __QSUB8(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("qsub8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SHSUB8(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("shsub8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __USUB8(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("usub8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UQSUB8(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uqsub8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UHSUB8(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uhsub8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SADD16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("sadd16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __QADD16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("qadd16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SHADD16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("shadd16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UADD16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uadd16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UQADD16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uqadd16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UHADD16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uhadd16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SSUB16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("ssub16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __QSUB16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("qsub16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SHSUB16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("shsub16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __USUB16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("usub16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UQSUB16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uqsub16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UHSUB16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uhsub16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SASX(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("sasx %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __QASX(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("qasx %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SHASX(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("shasx %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UASX(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uasx %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UQASX(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uqasx %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UHASX(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uhasx %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SSAX(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("ssax %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __QSAX(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("qsax %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SHSAX(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("shsax %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __USAX(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("usax %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UQSAX(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uqsax %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UHSAX(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uhsax %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __USAD8(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("usad8 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __USADA8(uint32_t op1, uint32_t op2, uint32_t op3)
-{
-  uint32_t result;
-
-  __ASM volatile ("usada8 %0, %1, %2, %3" : "=r" (result) : "r" (op1), "r" (op2), "r" (op3) );
-  return(result);
-}
-
-#define __SSAT16(ARG1,ARG2) \
-({                          \
-  int32_t __RES, __ARG1 = (ARG1); \
-  __ASM ("ssat16 %0, %1, %2" : "=r" (__RES) :  "I" (ARG2), "r" (__ARG1) ); \
-  __RES; \
- })
-
-#define __USAT16(ARG1,ARG2) \
-({                          \
-  uint32_t __RES, __ARG1 = (ARG1); \
-  __ASM ("usat16 %0, %1, %2" : "=r" (__RES) :  "I" (ARG2), "r" (__ARG1) ); \
-  __RES; \
- })
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UXTB16(uint32_t op1)
-{
-  uint32_t result;
-
-  __ASM volatile ("uxtb16 %0, %1" : "=r" (result) : "r" (op1));
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __UXTAB16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("uxtab16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SXTB16(uint32_t op1)
-{
-  uint32_t result;
-
-  __ASM volatile ("sxtb16 %0, %1" : "=r" (result) : "r" (op1));
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SXTAB16(uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("sxtab16 %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE 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);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE 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);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE 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);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE 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);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE 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);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE 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);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE 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);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE 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);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SMLSD (uint32_t op1, uint32_t op2, uint32_t op3)
-{
-  uint32_t result;
-
-  __ASM volatile ("smlsd %0, %1, %2, %3" : "=r" (result) : "r" (op1), "r" (op2), "r" (op3) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE 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);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint64_t __SMLSLD (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 ("smlsld %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 ("smlsld %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);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint64_t __SMLSLDX (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 ("smlsldx %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 ("smlsldx %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);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE uint32_t __SEL  (uint32_t op1, uint32_t op2)
-{
-  uint32_t result;
-
-  __ASM volatile ("sel %0, %1, %2" : "=r" (result) : "r" (op1), "r" (op2) );
-  return(result);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE  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);
-}
-
-__attribute__((always_inline)) __STATIC_INLINE  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);
-}
-
-#if 0
-#define __PKHBT(ARG1,ARG2,ARG3) \
-({                          \
-  uint32_t __RES, __ARG1 = (ARG1), __ARG2 = (ARG2); \
-  __ASM ("pkhbt %0, %1, %2, lsl %3" : "=r" (__RES) :  "r" (__ARG1), "r" (__ARG2), "I" (ARG3)  ); \
-  __RES; \
- })
-
-#define __PKHTB(ARG1,ARG2,ARG3) \
-({                          \
-  uint32_t __RES, __ARG1 = (ARG1), __ARG2 = (ARG2); \
-  if (ARG3 == 0) \
-    __ASM ("pkhtb %0, %1, %2" : "=r" (__RES) :  "r" (__ARG1), "r" (__ARG2)  ); \
-  else \
-    __ASM ("pkhtb %0, %1, %2, asr %3" : "=r" (__RES) :  "r" (__ARG1), "r" (__ARG2), "I" (ARG3)  ); \
-  __RES; \
- })
-#endif
+#define     __SADD8                 __builtin_arm_sadd8
+#define     __QADD8                 __builtin_arm_qadd8
+#define     __SHADD8                __builtin_arm_shadd8
+#define     __UADD8                 __builtin_arm_uadd8
+#define     __UQADD8                __builtin_arm_uqadd8
+#define     __UHADD8                __builtin_arm_uhadd8
+#define     __SSUB8                 __builtin_arm_ssub8
+#define     __QSUB8                 __builtin_arm_qsub8
+#define     __SHSUB8                __builtin_arm_shsub8
+#define     __USUB8                 __builtin_arm_usub8
+#define     __UQSUB8                __builtin_arm_uqsub8
+#define     __UHSUB8                __builtin_arm_uhsub8
+#define     __SADD16                __builtin_arm_sadd16
+#define     __QADD16                __builtin_arm_qadd16
+#define     __SHADD16               __builtin_arm_shadd16
+#define     __UADD16                __builtin_arm_uadd16
+#define     __UQADD16               __builtin_arm_uqadd16
+#define     __UHADD16               __builtin_arm_uhadd16
+#define     __SSUB16                __builtin_arm_ssub16
+#define     __QSUB16                __builtin_arm_qsub16
+#define     __SHSUB16               __builtin_arm_shsub16
+#define     __USUB16                __builtin_arm_usub16
+#define     __UQSUB16               __builtin_arm_uqsub16
+#define     __UHSUB16               __builtin_arm_uhsub16
+#define     __SASX                  __builtin_arm_sasx
+#define     __QASX                  __builtin_arm_qasx
+#define     __SHASX                 __builtin_arm_shasx
+#define     __UASX                  __builtin_arm_uasx
+#define     __UQASX                 __builtin_arm_uqasx
+#define     __UHASX                 __builtin_arm_uhasx
+#define     __SSAX                  __builtin_arm_ssax
+#define     __QSAX                  __builtin_arm_qsax
+#define     __SHSAX                 __builtin_arm_shsax
+#define     __USAX                  __builtin_arm_usax
+#define     __UQSAX                 __builtin_arm_uqsax
+#define     __UHSAX                 __builtin_arm_uhsax
+#define     __USAD8                 __builtin_arm_usad8
+#define     __USADA8                __builtin_arm_usada8
+#define     __SSAT16                __builtin_arm_ssat16
+#define     __USAT16                __builtin_arm_usat16
+#define     __UXTB16                __builtin_arm_uxtb16
+#define     __UXTAB16               __builtin_arm_uxtab16
+#define     __SXTB16                __builtin_arm_sxtb16
+#define     __SXTAB16               __builtin_arm_sxtab16
+#define     __SMUAD                 __builtin_arm_smuad
+#define     __SMUADX                __builtin_arm_smuadx
+#define     __SMLAD                 __builtin_arm_smlad
+#define     __SMLADX                __builtin_arm_smladx
+#define     __SMLALD                __builtin_arm_smlald
+#define     __SMLALDX               __builtin_arm_smlaldx
+#define     __SMUSD                 __builtin_arm_smusd
+#define     __SMUSDX                __builtin_arm_smusdx
+#define     __SMLSD                 __builtin_arm_smlsd
+#define     __SMLSDX                __builtin_arm_smlsdx
+#define     __SMLSLD                __builtin_arm_smlsld
+#define     __SMLSLDX               __builtin_arm_smlsldx
+#define     __SEL                   __builtin_arm_sel
+#define     __QADD                  __builtin_arm_qadd
+#define     __QSUB                  __builtin_arm_qsub
 
 #define __PKHBT(ARG1,ARG2,ARG3)          ( ((((uint32_t)(ARG1))          ) & 0x0000FFFFUL) |  \
                                            ((((uint32_t)(ARG2)) << (ARG3)) & 0xFFFF0000UL)  )
@@ -1781,7 +1429,7 @@
 #define __PKHTB(ARG1,ARG2,ARG3)          ( ((((uint32_t)(ARG1))          ) & 0xFFFF0000UL) |  \
                                            ((((uint32_t)(ARG2)) >> (ARG3)) & 0x0000FFFFUL)  )
 
-__attribute__((always_inline)) __STATIC_INLINE int32_t __SMMLA (int32_t op1, int32_t op2, int32_t op3)
+__STATIC_FORCEINLINE int32_t __SMMLA (int32_t op1, int32_t op2, int32_t op3)
 {
   int32_t result;