fix(crashtracking): Redaction on macOS aarch64 register-to-memory mapping

Author: bric3

dd-java-agent/agent-crashtracking/src/main/java/datadog/crashtracking/parsers/HotspotCrashLogParser.java

@@ -650,7 +650,7 @@ private static State nextThreadSectionState(String line, boolean previousLineBla
     if (line.contains("P R O C E S S")) {
       return State.PROCESS;
     }
-    if (previousLineBlank && !line.contains("=") && SUBSECTION_TITLE.matcher(line).matches()) {
+    if (previousLineBlank && SUBSECTION_TITLE.matcher(line).matches()) {
       return State.STACKTRACE;
     }
     return null;

dd-java-agent/agent-crashtracking/src/main/java/datadog/crashtracking/parsers/RedactUtils.java

@@ -38,9 +38,11 @@ public final class RedactUtils {
   // 'in 'class'' clause in {method} descriptor entries
   private static final Pattern METHOD_IN_CLASS = Pattern.compile("( in ')([^']+)(')");
 
-  // Library path after <offset 0x...> in /path/to/lib.so
+  // Library path in two formats produced by os::print_location():
+  //   <offset 0x...> in /path/to/lib.so at 0x...       (no dladdr symbol)
+  //   symbol+offset in /path/to/lib.so at 0x...         (dladdr resolved a symbol name)
   private static final Pattern LIBRARY_PATH =
-      Pattern.compile("(<offset 0x[0-9a-fA-F]+> in )(/\\S+)");
+      Pattern.compile("((?:<[^>]+>|\\S+\\+\\S+)\\s+in\\s+)(/\\S+)");
 
   // Dotted class name followed by an OOP reference: "com.company.Type"{0x...}
   // This specifically identifies the inline string value of a java.lang.Class 'name' field
@@ -115,9 +117,9 @@ static String redactMethodClass(String line) {
   }
 
   /**
-   * Redacts all but the parent directory and filename from a library path in the line: <code>
-   * &lt;offset 0x...&gt; in /path/to/dir/lib.so</code> to <code>
-   * &lt;offset 0x...&gt; in /redacted/redacted/dir/lib.so</code>
+   * Redacts all but the parent directory and filename from a library path. Handles both
+   * <code>&lt;offset 0x...&gt; in /path/to/dir/lib.so</code> and <code>symbol+0 in
+   * /path/to/dir/lib.so</code> to <code>... in /redacted/redacted/dir/lib.so</code>
    */
   static String redactLibraryPath(String line) {
     return replaceAll(LIBRARY_PATH, line, m -> m.group(1) + redactPath(m.group(2)));

dd-java-agent/agent-crashtracking/src/test/java/datadog/crashtracking/parsers/HotspotCrashLogParserTest.java

@@ -13,6 +13,7 @@
 import java.io.InputStream;
 import java.io.InputStreamReader;
 import java.nio.charset.StandardCharsets;
+import java.util.Map;
 import java.util.Objects;
 import java.util.UUID;
 import java.util.stream.Collectors;
@@ -68,6 +69,12 @@ public void testRegisterParsingMacosAarch64() throws Exception {
     assertEquals("0x000000010f8ac794", crashLog.experimental.ucontext.get("pc"));
     assertEquals("0x0000000060001000", crashLog.experimental.ucontext.get("cpsr"));
 
+    // "Top of Stack: (sp=0x...)" contains "=" — verify the parser stops at it and doesn't
+    // absorb its hex-dump content into the last register mapping entry (sp).
+    assertThat(crashLog.experimental.registerToMemoryMapping)
+        .extractingByKey("sp", STRING)
+        .doesNotContain("Top of Stack:");
+
     assertNotNull(crashLog.experimental.runtimeArgs);
     assertTrue(crashLog.experimental.runtimeArgs.contains("--enable-native-access=ALL-UNNAMED"));
     assertTrue(crashLog.experimental.runtimeArgs.contains("--add-modules=ALL-DEFAULT"));
@@ -76,6 +83,50 @@ public void testRegisterParsingMacosAarch64() throws Exception {
             .anyMatch(arg -> arg.contains("SourceLauncher") || arg.endsWith("CrashTest.java")));
   }
 
+  /**
+   * Verifies the register-to-memory mapping section for the macOS aarch64 sample:
+   * representative values, library path redaction, and that "Top of Stack:" / "Instructions:"
+   * subsections are not absorbed into register values.
+   */
+  @Test
+  public void testRegisterToMemoryMappingMacosAarch64() throws Exception {
+    CrashLog crashLog =
+        new HotspotCrashLogParser()
+            .parse(
+                UUID.randomUUID().toString(), readFileAsString("sample-crash-macos-aarch64.txt"));
+
+    Map<String, String> mapping = crashLog.experimental.registerToMemoryMapping;
+
+    // Representative single-line entries
+    assertThat(mapping)
+        .containsEntry("x0", "0x0000000000000c55 is an unknown value")
+        .containsEntry("x2", "0x0 is null")
+        .containsEntry("x28", "0x0000000100a153f0 is a thread");
+
+    // Library path (symbol+offset format) — path must be redacted, keeping only last 2 segments
+    // /usr/lib/system/libsystem_pthread.dylib → 2 redacted ("usr","lib") + "system/lib..."
+    assertThat(mapping)
+        .extractingByKey("x16", STRING)
+        .isEqualTo(
+            "0x0000000182d709d0: pthread_jit_write_protect_np+0 in /redacted/redacted/system/libsystem_pthread.dylib at 0x0000000182d69000");
+    // /Users/USER/.local/share/mise/installs/java/25.0.2/lib/server/libjvm.dylib → 9 redacted + "server/libjvm.dylib"
+    assertThat(mapping)
+        .extractingByKey("x21", STRING)
+        .isEqualTo(
+            "0x0000000106c1ccc0: _ZN19TemplateInterpreter13_active_tableE+0 in /redacted/redacted/redacted/redacted/redacted/redacted/redacted/redacted/redacted/server/libjvm.dylib at 0x0000000105efc000");
+
+    // "Top of Stack: (sp=0x...)" and "Instructions: (pc=0x...)" must not leak into register values
+    assertThat(mapping).doesNotContainKey("Top of Stack");
+    assertThat(mapping)
+        .allSatisfy(
+            (k, v) -> assertThat(v).doesNotContain("Top of Stack:", "Instructions:"));
+
+    // sp is the last register before "Top of Stack:" — its value must be clean
+    assertThat(mapping)
+        .extractingByKey("sp", STRING)
+        .isEqualTo("0x000000016feee0f0 is pointing into the stack for thread: 0x0000000100a153f0");
+  }
+
   /** Linux aarch64 uses uppercase register names: R0-R30 */
   @Test
   public void testRegisterParsingLinuxAarch64() throws Exception {

dd-java-agent/agent-crashtracking/src/test/java/datadog/crashtracking/parsers/RedactUtilsTest.java

@@ -124,14 +124,43 @@ void testRedactMethodClass_keepsKnownPackage() {
   }
 
   @Test
-  void testRedactLibraryPath_redactsIntermediateSegments() {
+  void testRedactLibraryPath_offsetFormat() {
     assertThat(
             RedactUtils.redactLibraryPath(
                 "0x0000ffff9efa1650: <offset 0x0000000000e01650> in /opt/company/lib/server/app.so at 0x0000ffff9e1a0000"))
         .isEqualTo(
             "0x0000ffff9efa1650: <offset 0x0000000000e01650> in /redacted/redacted/redacted/server/app.so at 0x0000ffff9e1a0000");
   }
 
+  @Test
+  void testRedactLibraryPath_symbolOffsetFormat() {
+    // macOS/Linux: dladdr resolved a C++ mangled symbol — "symbol+offset in /path at 0x..."
+    assertThat(
+            RedactUtils.redactLibraryPath(
+                "0x0000000106c1ccc0: _ZN19TemplateInterpreter13_active_tableE+0 in /Users/USER/.local/share/mise/installs/java/25.0.2/lib/server/libjvm.dylib at 0x0000000105efc000"))
+        .isEqualTo(
+            "0x0000000106c1ccc0: _ZN19TemplateInterpreter13_active_tableE+0 in /redacted/redacted/redacted/redacted/redacted/redacted/redacted/redacted/redacted/server/libjvm.dylib at 0x0000000105efc000");
+  }
+
+  @Test
+  void testRedactLibraryPath_cSymbolFormat() {
+    // macOS: C symbol "symbol+0 in /usr/lib/system/lib.dylib at 0x..."
+    assertThat(
+            RedactUtils.redactLibraryPath(
+                "0x0000000182d709d0: pthread_jit_write_protect_np+0 in /usr/lib/system/libsystem_pthread.dylib at 0x0000000182d69000"))
+        .isEqualTo(
+            "0x0000000182d709d0: pthread_jit_write_protect_np+0 in /redacted/redacted/system/libsystem_pthread.dylib at 0x0000000182d69000");
+  }
+
+  @Test
+  void testRedactLibraryPath_doesNotMatchInterpreterCodelet() {
+    // "code_begin+1776 in an Interpreter codelet" — "an" doesn't start with "/" so no match
+    assertThat(
+            RedactUtils.redactLibraryPath(
+                "0x0000000116d0c970 is at code_begin+1776 in an Interpreter codelet"))
+        .isEqualTo("0x0000000116d0c970 is at code_begin+1776 in an Interpreter codelet");
+  }
+
   @Test
   void testRedactLibraryPath_leavesUnrelatedLinesUnchanged() {
     assertThat(RedactUtils.redactLibraryPath("0x00007f37a16e2590 is an unknown value"))