From 04fb2c838d935783db207ec1dfb1e7835599ca53 Mon Sep 17 00:00:00 2001
From: Florian Mayer <fmayer@google.com>
Date: Mon, 9 Jan 2023 21:46:07 +0000
Subject: [PATCH] Add comment about out-of-bounds being on purpose

Change-Id: I97e334f0555cce80eeab6567905054d807af3c91
---
 init/test_upgrade_mte/mte_upgrade_test_helper.cpp | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/init/test_upgrade_mte/mte_upgrade_test_helper.cpp b/init/test_upgrade_mte/mte_upgrade_test_helper.cpp
index 3188337ce..6728cc6d0 100644
--- a/init/test_upgrade_mte/mte_upgrade_test_helper.cpp
+++ b/init/test_upgrade_mte/mte_upgrade_test_helper.cpp
@@ -60,6 +60,10 @@ int main(int argc, char** argv) {
         if (prctl(PR_SET_TAGGED_ADDR_CTRL, res & ~PR_MTE_TCF_SYNC, 0, 0, 0) == -1) abort();
     }
     std::unique_ptr<volatile char[]> f(new char[1]);
+    // This out-of-bounds is on purpose: we are testing MTE, which is designed to turn
+    // out-of-bound errors into segfaults.
+    // This binary gets run by src/com/android/tests/init/MteUpgradeTest.java, which
+    // asserts that it crashes as expected.
     f[17] = 'x';
     char buf[1];
     read(1, buf, 1);