[cli/trampolines]: Fix aarch64-apple-darwin trampoline ASM syntax

It turns out that aarch64 assembly syntax uses `;` as the comment
character, not as the statement separator.  So we need to polyfill
that.  We also set an alignment of 4 bytes (`2^2`) which is required on
Apple targets, but is also a good idea on most other aarch64 machines.

(cherry picked from commit cdb0833)

Author: staticfloat

cli/Makefile

@@ -54,7 +54,11 @@ $(BUILDDIR)/loader_exe.o : $(SRCDIR)/loader_exe.c $(HEADERS) $(JULIAHOME)/VERSIO
 $(BUILDDIR)/loader_exe.dbg.obj : $(SRCDIR)/loader_exe.c $(HEADERS) $(JULIAHOME)/VERSION
 	@$(call PRINT_CC, $(CC) $(DEBUGFLAGS) $(LOADER_CFLAGS) -c $< -o $@)
 $(BUILDDIR)/loader_trampolines.o : $(SRCDIR)/trampolines/trampolines_$(ARCH).S
-	@$(call PRINT_CC, $(CC) $(DEBUGFLAGS) $(LOADER_CFLAGS) $< -c -o $@)
+	@$(call PRINT_CC, $(CC) $(SHIPFLAGS) $(LOADER_CFLAGS) $< -c -o $@)
+
+# Debugging target to help us see what kind of code is being generated for our trampolines
+dump-trampolines: $(SRCDIR)/trampolines/trampolines_$(ARCH).S
+	$(CC) $(SHIPFLAGS) $(LOADER_CFLAGS) $< -S | sed -E 's/ ((%%)|;) /\n/g' | sed -E 's/.global/\n.global/g'
 
 DIRS = $(build_bindir) $(build_libdir)
 $(DIRS):

cli/trampolines/trampolines_aarch64.S

@@ -1,12 +1,27 @@
 #include "../../src/jl_exported_funcs.inc"
 
+// On macOS, we need to prepend underscores on symbols
+#if defined(__APPLE__) && defined(__MACH__)
+#define CNAME(x)    _##x
+#define PAGE(x)     x##@PAGE
+#define PAGEOFF(x)  x##@PAGEOFF
+#define SEP         %%
+#else
+#define CNAME(x)    x
+#define PAGE(x)     x
+#define PAGEOFF(x)  #:lo12:##x
+#define SEP         ;
+#endif
+
 #define XX(name) \
-.global name; \
-.cfi_startproc; \
-name##:; \
-    adrp x0, name##_addr; \
-    ldr x0, [x0, #:lo12:name##_addr]; \
-    br x0; \
-.cfi_endproc; \
+.global CNAME(name) SEP \
+.cfi_startproc SEP \
+.p2align    2 SEP \
+CNAME(name)##: SEP \
+    adrp x0, PAGE(CNAME(name##_addr)) SEP \
+    ldr x0, [x0, PAGEOFF(CNAME(name##_addr))] SEP \
+    br x0 SEP \
+.cfi_endproc SEP \
+
 JL_EXPORTED_FUNCS(XX)
 #undef XX