Fix the stack-protector test for x86/x86-64.

Built for fugu this was working fine, but built for generic x86/x86-64
the compiler was (a) optimizing out all the stack writes and (b) inserting
enough padding on x86-64 for the canary to be safely out of the way.

While here, let's tighten up this test so we test that it's sufficient to
only overwrite the buffer by one byte...

Bug: http://b/27815668
Change-Id: I80a646de4b30fd5c78df20fdaa7e3eb163585caf

diff --git a/tests/stack_protector_test_helper.cpp b/tests/stack_protector_test_helper.cpp
index 34f3c77..53a5e05 100644 (file)
--- a/tests/stack_protector_test_helper.cpp
+++ b/tests/stack_protector_test_helper.cpp
@@ -19,6 +19,13 @@ __attribute__((noinline)) void modify_stack_protector_test() {
   char buf[128];
   // We can't use memset here because it's fortified, and we want to test
   // the line of defense *after* that.
-  char* p = buf;
-  while ((p - buf) < static_cast<int>(sizeof(buf) + sizeof(void*))) *p++ = '\0';
+  // Without volatile, the generic x86/x86-64 targets don't write to the stack.
+  volatile char* p = buf;
+  int size = static_cast<int>(sizeof(buf) + 1);
+#if __x86_64__
+  // The generic x86-64 target leaves an 8-byte gap between `buf` and the stack guard.
+  // We only need to corrupt one byte though.
+  size += 8;
+#endif
+  while ((p - buf) < size) *p++ = '\0';
 }