Try not to cause a MemoryError when raising an exception during nterrupt handling.

Step 1 fixes #732

py/gc.c

@@ -172,6 +172,10 @@ void gc_unlock(void) {
     gc_lock_depth--;
 }
 
+bool gc_is_locked(void) {
+    return gc_lock_depth != 0;
+}
+
 #define VERIFY_PTR(ptr) ( \
         (ptr & (BYTES_PER_BLOCK - 1)) == 0          /* must be aligned on a block */ \
         && ptr >= (machine_uint_t)gc_pool_start     /* must be above start of pool */ \

py/gc.h

@@ -30,6 +30,7 @@ void gc_init(void *start, void *end);
 // They can be used to prevent the GC from allocating/freeing.
 void gc_lock(void);
 void gc_unlock(void);
+bool gc_is_locked(void);
 
 // A given port must implement gc_collect by using the other collect functions.
 void gc_collect(void);

py/objexcept.c

@@ -37,6 +37,7 @@
 #include "objtype.h"
 #include "runtime.h"
 #include "runtime0.h"
+#include "gc.h"
 
 typedef struct _mp_obj_exception_t {
     mp_obj_base_t base;
@@ -335,6 +336,10 @@ void mp_obj_exception_clear_traceback(mp_obj_t self_in) {
 }
 
 void mp_obj_exception_add_traceback(mp_obj_t self_in, qstr file, machine_uint_t line, qstr block) {
+    if (gc_is_locked()) {
+        // We can't allocate memory, so don't bother to try
+        return;
+    }
     GET_NATIVE_EXCEPTION(self, self_in);
 
     // for traceback, we are just using the list object for convenience, it's not really a list of Python objects