[llvm] [llvm-exegesis][AArch64] Check for PAC keys before disabling them (PR #138643)

Anatoly Trosinenko via llvm-commits llvm-commits at lists.llvm.org
Wed Jun 4 09:48:04 PDT 2025


================
@@ -151,6 +205,72 @@ class ExegesisAArch64Target : public ExegesisTarget {
     // Function return is a pseudo-instruction that needs to be expanded
     PM.add(createAArch64ExpandPseudoPass());
   }
+
+#if defined(__aarch64__) && defined(__linux__)
+  // Converts variadic arguments to `long` and passes zeros for the unused
+  // arg2-arg5, as tested by the Linux kernel.
+  static long prctl_wrapper(int op, long arg2 = 0, long arg3 = 0) {
+    return prctl(op, arg2, arg3, /*arg4=*/0L, /*arg5=*/0L);
+  }
+#endif
+
+  const char *getIgnoredOpcodeReasonOrNull(const LLVMState &State,
+                                           unsigned Opcode) const override {
+    if (const char *Reason =
+            ExegesisTarget::getIgnoredOpcodeReasonOrNull(State, Opcode))
+      return Reason;
+
+    if (isPointerAuth(Opcode)) {
+#if defined(__aarch64__) && defined(__linux__)
+      // Only proceed with PAC key control if explicitly requested
+      if (!AArch64DisablePacControl) {
+        // For some systems with existing PAC keys set, it is better to
+        // check the existing state of the key before setting it.
+        // If the CPU implements FEAT_FPAC,
+        // authentication instructions almost certainly crash when being
+        // benchmarked, so disable all the keys by default. On the other hand,
+        // disabling the keys at run-time can probably crash llvm-exegesis at
+        // some later point, depending on how it was built. For that reason, the
+        // user may pass --aarch64-disable-pac-control in case
----------------
atrosinenko wrote:

_If the option is renamed, this line has to be updated as well._

https://github.com/llvm/llvm-project/pull/138643


More information about the llvm-commits mailing list