diff --git a/include/stdarg.h b/include/stdarg.h
index 06d592b..9fff0c0 100644
--- a/include/stdarg.h
+++ b/include/stdarg.h
@@ -30,7 +30,8 @@ void *__va_arg(__va_list_struct *ap, int arg_type, int size, int align);
 #else /* _WIN64 */
 typedef char *va_list;
 #define va_start(ap,last) __builtin_va_start(ap,last)
-#define va_arg(ap,type) (ap += 8, sizeof(type)<=8 ? *(type*)ap : **(type**)ap)
+#define va_arg(ap, t) ((sizeof(t) > 8 || (sizeof(t) & (sizeof(t) - 1))) \
+	? **(t **)((ap += 8) - 8) : *(t  *)((ap += 8) - 8))
 #define va_copy(dest, src) ((dest) = (src))
 #define va_end(ap)
 #endif
diff --git a/tccgen.c b/tccgen.c
index d7f8c0d..5d92e5f 100644
--- a/tccgen.c
+++ b/tccgen.c
@@ -4033,6 +4033,7 @@ ST_FUNC void unary(void)
                 tcc_error("__builtin_va_start expects a local variable");
             vtop->r &= ~(VT_LVAL | VT_REF);
             vtop->type = char_pointer_type;
+            vtop->c.i += 8;
             vstore();
         }
         break;
diff --git a/tccpp.c b/tccpp.c
index 024f9de..b8453da 100644
--- a/tccpp.c
+++ b/tccpp.c
@@ -464,9 +464,6 @@ ST_FUNC const char *get_tok_str(int v, CValue *cv)
     switch(v) {
     case TOK_CINT:
     case TOK_CUINT:
-        /* XXX: not quite exact, but only useful for testing */
-        sprintf(p, "%llu", (unsigned long long)cv->i);
-        break;
     case TOK_CLLONG:
     case TOK_CULLONG:
         /* XXX: not quite exact, but only useful for testing  */
diff --git a/tests/tcctest.c b/tests/tcctest.c
index e5ec783..6ec23a1 100644
--- a/tests/tcctest.c
+++ b/tests/tcctest.c
@@ -2116,6 +2116,14 @@ void stdarg_for_struct(struct myspace bob, ...)
     va_end(ap);
 }
 
+void stdarg_for_libc(const char *fmt, ...)
+{
+    va_list args;
+    va_start(args, fmt);
+    vprintf(fmt, args);
+    va_end(args);
+}
+
 void stdarg_test(void)
 {
     LONG_DOUBLE ld = 1234567891234LL;
@@ -2162,6 +2170,7 @@ void stdarg_test(void)
 
     bob.profile = 42;
     stdarg_for_struct(bob, bob, bob, bob.profile);
+    stdarg_for_libc("stdarg_for_libc: %s %.2f %d\n", "string", 1.23, 456);
 }
 
 void whitespace_test(void)
diff --git a/x86_64-gen.c b/x86_64-gen.c
index e313a5f..5a122d0 100644
--- a/x86_64-gen.c
+++ b/x86_64-gen.c
@@ -2228,7 +2228,6 @@ ST_FUNC void gen_vla_alloc(CType *type, int align) {
     vpush_global_sym(&func_old_type, TOK_alloca);
     vswap(); /* Move alloca ref past allocation size */
     gfunc_call(1);
-    vset(type, REG_IRET, 0);
 #else
     int r;
     r = gv(RC_INT); /* allocation size */