split asm volatile back out (#1420)

src/rp2_common/hardware_divider/include/hardware/divider.h

@@ -85,7 +85,7 @@ static inline void hw_divider_wait_ready(void) {
     // we use one less register and instruction than gcc which uses a TST instruction
 
     uint32_t tmp; // allow compiler to pick scratch register
-    pico_default_asm (
+    pico_default_asm_volatile (
     "hw_divider_result_loop_%=:"
     "ldr %0, [%1, %2]\n\t"
     "lsrs %0, %0, #1\n\t"
@@ -296,7 +296,7 @@ static inline int32_t hw_divider_remainder_s32(int32_t a, int32_t b) {
  *  \ingroup hardware_divider
  */
 static inline void hw_divider_pause(void) {
-    pico_default_asm (
+    pico_default_asm_volatile(
     "b _1_%=\n"
     "_1_%=:\n"
     "b _2_%=\n"
@@ -305,7 +305,7 @@ static inline void hw_divider_pause(void) {
     "_3_%=:\n"
     "b _4_%=\n"
     "_4_%=:\n"
-    :: : );
+    :::);
 }
 
 /*! \brief Do a hardware unsigned HW divide, wait for result, return quotient

src/rp2_common/hardware_sync/include/hardware/sync.h

@@ -116,7 +116,7 @@ typedef volatile uint32_t spin_lock_t;
  */
 #if !__has_builtin(__sev)
 __force_inline static void __sev(void) {
-    pico_default_asm ("sev");
+    pico_default_asm_volatile ("sev");
 }
 #endif
 
@@ -128,7 +128,7 @@ __force_inline static void __sev(void) {
  */
 #if !__has_builtin(__wfe)
 __force_inline static void __wfe(void) {
-    pico_default_asm ("wfe");
+    pico_default_asm_volatile ("wfe");
 }
 #endif
 
@@ -139,7 +139,7 @@ __force_inline static void __wfe(void) {
  */
 #if !__has_builtin(__wfi)
 __force_inline static void __wfi(void) {
-    pico_default_asm ("wfi");
+    pico_default_asm_volatile("wfi");
 }
 #endif
 
@@ -150,7 +150,7 @@ __force_inline static void __wfi(void) {
  * instruction will be observed before any explicit access after the instruction.
  */
 __force_inline static void __dmb(void) {
-    pico_default_asm ("dmb" : : : "memory");
+    pico_default_asm_volatile("dmb" : : : "memory");
 }
 
 /*! \brief Insert a DSB instruction in to the code path.
@@ -161,7 +161,7 @@ __force_inline static void __dmb(void) {
  * accesses before this instruction complete.
  */
 __force_inline static void __dsb(void) {
-    pico_default_asm ("dsb" : : : "memory");
+    pico_default_asm_volatile("dsb" : : : "memory");
 }
 
 /*! \brief Insert a ISB instruction in to the code path.
@@ -172,7 +172,7 @@ __force_inline static void __dsb(void) {
  * the ISB instruction has been completed.
  */
 __force_inline static void __isb(void) {
-    pico_default_asm ("isb");
+    pico_default_asm_volatile("isb" ::: "memory");
 }
 
 /*! \brief Acquire a memory fence
@@ -213,8 +213,10 @@ __force_inline static void __mem_fence_release(void) {
  */
 __force_inline static uint32_t save_and_disable_interrupts(void) {
     uint32_t status;
-    pico_default_asm ("mrs %0, PRIMASK" : "=r" (status)::);
-    pico_default_asm ("cpsid i");
+    pico_default_asm_volatile(
+            "mrs %0, PRIMASK\n"
+            "cpsid i"
+            : "=r" (status) ::);
     return status;
 }
 
@@ -224,7 +226,7 @@ __force_inline static uint32_t save_and_disable_interrupts(void) {
  * \param status Previous interrupt status from save_and_disable_interrupts()
   */
 __force_inline static void restore_interrupts(uint32_t status) {
-    pico_default_asm ("msr PRIMASK,%0"::"r" (status) : );
+    pico_default_asm_volatile("msr PRIMASK,%0"::"r" (status) : );
 }
 
 /*! \brief Get HW Spinlock instance from number

src/rp2_common/pico_platform/include/pico/platform.h

@@ -335,7 +335,8 @@ extern "C" {
 #define MIN(a, b) ((b)>(a)?(a):(b))
 #endif
 
-#define pico_default_asm(...) __asm volatile (".syntax unified\n" __VA_ARGS__)
+#define pico_default_asm(...) __asm (".syntax unified\n" __VA_ARGS__)
+#define pico_default_asm_volatile(...) __asm volatile (".syntax unified\n" __VA_ARGS__)
 
 /*! \brief Execute a breakpoint instruction
  *  \ingroup pico_platform
@@ -357,7 +358,7 @@ static inline void __breakpoint(void) {
  * might - even above the memory store!)
  */
 __force_inline static void __compiler_memory_barrier(void) {
-    pico_default_asm ("" : : : "memory");
+    pico_default_asm_volatile ("" : : : "memory");
 }
 
 /*! \brief Macro for converting memory addresses to 32 bit addresses suitable for DMA
@@ -479,7 +480,7 @@ __force_inline static int32_t __mul_instruction(int32_t a, int32_t b) {
  */
 static __force_inline uint __get_current_exception(void) {
     uint exception;
-    pico_default_asm ("mrs %0, ipsr" : "=l" (exception));
+    pico_default_asm( "mrs %0, ipsr" : "=l" (exception));
     return exception;
 }
 
@@ -501,7 +502,7 @@ static __force_inline uint __get_current_exception(void) {
  * \param minimum_cycles the minimum number of system clock cycles to delay for
  */
 static inline void busy_wait_at_least_cycles(uint32_t minimum_cycles) {
-    pico_default_asm (
+    pico_default_asm_volatile(
         "1: subs %0, #3\n"
         "bcs 1b\n"
         : "+l" (minimum_cycles) : : "memory"