nrf/gccollect: Use the SP register instead of MSP.

Using the current stack pointer directly saves 8 bytes of code.
We need the *current* register anyway for GC (which is always MSP).

ports/nrf/gccollect.c

@@ -31,18 +31,18 @@
 #include "py/gc.h"
 #include "gccollect.h"
 
-static inline uint32_t get_msp(void)
+static inline uintptr_t get_sp(void) {
-{
+    uintptr_t result;
-	register uint32_t result;
+    __asm__ ("mov %0, sp\n" : "=r" (result) );
-	__asm volatile ("MRS %0, msp\n" : "=r" (result) );
+    return result;
-	return(result);
 }
 
 void gc_collect(void) {
     // start the GC
     gc_collect_start();
 
-    mp_uint_t sp = get_msp(); // Get stack pointer
+    // Get stack pointer
+    uintptr_t sp = get_sp();
 
     // trace the stack, including the registers (since they live on the stack in this function)
     gc_collect_root((void**)sp, ((uint32_t)&_ram_end - sp) / sizeof(uint32_t));