Remove non-UTF8 characters from string fields.
The string type in the tombstone proto does not support non-UTF8
characters. Therefore, use the oct_encode function to encode the
abort_message field and message field from LogMessage.
Fix up stl includes, add ones that were missing and remove those not
being used.
Add new unit test to verify that the abort and log messages are
sanitized.
Bug: 279496937
Bug: 377940849
Bug: 378185483
Test: All unit tests pass.
Test: Ran pbtombstone on a crash with non-UTF8 characters and verified
Test: it processes properly after this change and fails before the change.
Change-Id: I3554d56caf9fcbfc410b4d554f6c3b4888b37e28
diff --git a/debuggerd/debuggerd_test.cpp b/debuggerd/debuggerd_test.cpp
index 13c8d70..5bdc946 100644
--- a/debuggerd/debuggerd_test.cpp
+++ b/debuggerd/debuggerd_test.cpp
@@ -3302,3 +3302,30 @@
ASSERT_MATCH(result, ":\\s*This line has a newline.");
ASSERT_MATCH(result, ":\\s*This is on the next line.");
}
+
+TEST_F(CrasherTest, log_with_non_utf8) {
+ StartProcess([]() { LOG(FATAL) << "Invalid UTF-8: \xA0\xB0\xC0\xD0 and some other data."; });
+
+ unique_fd output_fd;
+ StartIntercept(&output_fd);
+ FinishCrasher();
+ AssertDeath(SIGABRT);
+ int intercept_result;
+ FinishIntercept(&intercept_result);
+ ASSERT_EQ(1, intercept_result) << "tombstoned reported failure";
+
+ std::string result;
+ ConsumeFd(std::move(output_fd), &result);
+ // Verify the abort message is sanitized properly.
+ size_t pos = result.find(
+ "Abort message: 'Invalid UTF-8: "
+ "\x5C\x32\x34\x30\x5C\x32\x36\x30\x5C\x33\x30\x30\x5C\x33\x32\x30 and some other data.'");
+ EXPECT_TRUE(pos != std::string::npos) << "Couldn't find sanitized abort message: " << result;
+
+ // Make sure that the log message is sanitized properly too.
+ EXPECT_TRUE(
+ result.find("Invalid UTF-8: \x5C\x32\x34\x30\x5C\x32\x36\x30\x5C\x33\x30\x30\x5C\x33\x32\x30 "
+ "and some other data.",
+ pos + 30) != std::string::npos)
+ << "Couldn't find sanitized log message: " << result;
+}