[clang] Define ptrauth_string_discriminator builtin.

[ahmedbougacha]

This exposes the ABI-stable hash function that allows computing a 16-bit discriminator from a constant string.

This allows manually matching the implicit string discriminators computed in the ABI (e.g., from mangled names for vtable pointer/entry signing), as well as enabling the use of interesting discriminators when manually annotating specific pointers with the __ptrauth qualifier.

[llvmbot]

@llvm/pr-subscribers-backend-aarch64
@llvm/pr-subscribers-backend-x86

@llvm/pr-subscribers-clang

Author: Ahmed Bougacha (ahmedbougacha)

Changes

This exposes the ABI-stable hash function that allows computing a 16-bit discriminator from a constant string.

This allows manually matching the implicit string discriminators computed in the ABI (e.g., from mangled names for vtable pointer/entry signing), as well as enabling the use of interesting discriminators when manually annotating specific pointers with the __ptrauth qualifier.

---

Full diff: https://github.com/llvm/llvm-project/pull/93903.diff

7 Files Affected:

• (modified) clang/include/clang/Basic/Builtins.td (+6)
• (modified) clang/include/clang/Basic/DiagnosticSemaKinds.td (+2)
• (modified) clang/lib/AST/ExprConstant.cpp (+7)
• (modified) clang/lib/Headers/ptrauth.h (+17)
• (modified) clang/lib/Sema/SemaChecking.cpp (+20)
• (modified) clang/test/CodeGen/ptrauth-intrinsics.c (+13)
• (modified) clang/test/Sema/ptrauth-intrinsics-macro.c (+5)

diff --git a/clang/include/clang/Basic/Builtins.td b/clang/include/clang/Basic/Builtins.td
index 11982af3fa609..836697632a3bc 100644
--- a/clang/include/clang/Basic/Builtins.td
+++ b/clang/include/clang/Basic/Builtins.td
@@ -4411,6 +4411,12 @@ def PtrauthAuth : Builtin {
   let Prototype = "void*(void*,int,void*)";
 }
 
+def PtrauthStringDiscriminator : Builtin {
+  let Spellings = ["__builtin_ptrauth_string_discriminator"];
+  let Attributes = [NoThrow, Const, Constexpr];
+  let Prototype = "size_t(char const*)";
+}
+
 // OpenCL v2.0 s6.13.16, s9.17.3.5 - Pipe functions.
 // We need the generic prototype, since the packet type could be anything.
 def ReadPipe : OCLPipeLangBuiltin {
diff --git a/clang/include/clang/Basic/DiagnosticSemaKinds.td b/clang/include/clang/Basic/DiagnosticSemaKinds.td
index f15cba63624ea..64add46248c69 100644
--- a/clang/include/clang/Basic/DiagnosticSemaKinds.td
+++ b/clang/include/clang/Basic/DiagnosticSemaKinds.td
@@ -928,6 +928,8 @@ def warn_ptrauth_sign_null_pointer :
 def warn_ptrauth_auth_null_pointer :
   Warning<"authenticating a null pointer will almost certainly trap">,
   InGroup<PtrAuthNullPointers>;
+def err_ptrauth_string_not_literal : Error<
+  "argument must be a string literal%select{| of char type}0">;
 
 /// main()
 // static main() is not an error in C, just in C++.
diff --git a/clang/lib/AST/ExprConstant.cpp b/clang/lib/AST/ExprConstant.cpp
index f1aa19e4409e1..eafecfb5fe5b1 100644
--- a/clang/lib/AST/ExprConstant.cpp
+++ b/clang/lib/AST/ExprConstant.cpp
@@ -58,6 +58,7 @@
 #include "llvm/ADT/StringExtras.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/SaveAndRestore.h"
+#include "llvm/Support/SipHash.h"
 #include "llvm/Support/TimeProfiler.h"
 #include "llvm/Support/raw_ostream.h"
 #include <cstring>
@@ -12583,6 +12584,12 @@ bool IntExprEvaluator::VisitBuiltinCallExpr(const CallExpr *E,
   case Builtin::BI__builtin_expect_with_probability:
     return Visit(E->getArg(0));
 
+  case Builtin::BI__builtin_ptrauth_string_discriminator: {
+    auto literal = cast<StringLiteral>(E->getArg(0)->IgnoreParenImpCasts());
+    auto result = getPointerAuthStableSipHash16(literal->getString());
+    return Success(result, E);
+  }
+
   case Builtin::BI__builtin_ffs:
   case Builtin::BI__builtin_ffsl:
   case Builtin::BI__builtin_ffsll: {
diff --git a/clang/lib/Headers/ptrauth.h b/clang/lib/Headers/ptrauth.h
index 036665d75a91b..3e58af1802084 100644
--- a/clang/lib/Headers/ptrauth.h
+++ b/clang/lib/Headers/ptrauth.h
@@ -135,6 +135,17 @@ typedef __UINTPTR_TYPE__ ptrauth_generic_signature_t;
 #define ptrauth_auth_data(__value, __old_key, __old_data)                      \
   __builtin_ptrauth_auth(__value, __old_key, __old_data)
 
+/* Compute a constant discriminator from the given string.
+
+   The result can be used as the second argument to
+   ptrauth_blend_discriminator or the third argument to the
+   __ptrauth qualifier.  It has type size_t.
+
+   The argument must be a string literal.
+   A call to this function is an integer constant expression. */
+#define ptrauth_string_discriminator(__string)                                 \
+  __builtin_ptrauth_string_discriminator(__string)
+
 /* Compute a signature for the given pair of pointer-sized values.
    The order of the arguments is significant.
 
@@ -196,6 +207,12 @@ typedef __UINTPTR_TYPE__ ptrauth_generic_signature_t;
     __value;                                                                   \
   })
 
+#define ptrauth_string_discriminator(__string)                                 \
+  ({                                                                           \
+    (void)__string;                                                            \
+    ((ptrauth_extra_data_t)0);                                                 \
+  })
+
 #define ptrauth_sign_generic_data(__value, __data)                             \
   ({                                                                           \
     (void)__value;                                                             \
diff --git a/clang/lib/Sema/SemaChecking.cpp b/clang/lib/Sema/SemaChecking.cpp
index c3251f3cc9d81..f6012ef4b3601 100644
--- a/clang/lib/Sema/SemaChecking.cpp
+++ b/clang/lib/Sema/SemaChecking.cpp
@@ -2156,6 +2156,24 @@ static ExprResult PointerAuthAuthAndResign(Sema &S, CallExpr *Call) {
   return Call;
 }
 
+static ExprResult PointerAuthStringDiscriminator(Sema &S, CallExpr *call) {
+  if (checkPointerAuthEnabled(S, call)) return ExprError();
+
+  // We've already performed normal call type-checking.
+  Expr *arg = call->getArgs()[0]->IgnoreParenImpCasts();
+
+  // Operand must be an ordinary or UTF-8 string literal.
+  auto literal = dyn_cast<StringLiteral>(arg);
+  if (!literal || literal->getCharByteWidth() != 1) {
+    S.Diag(arg->getExprLoc(), diag::err_ptrauth_string_not_literal)
+      << (literal ? 1 : 0)
+      << arg->getSourceRange();
+    return ExprError();
+  }
+
+  return call;
+}
+
 static ExprResult BuiltinLaunder(Sema &S, CallExpr *TheCall) {
   if (S.checkArgCount(TheCall, 1))
     return ExprError();
@@ -2922,6 +2940,8 @@ Sema::CheckBuiltinFunctionCall(FunctionDecl *FDecl, unsigned BuiltinID,
     return PointerAuthSignGenericData(*this, TheCall);
   case Builtin::BI__builtin_ptrauth_auth_and_resign:
     return PointerAuthAuthAndResign(*this, TheCall);
+  case Builtin::BI__builtin_ptrauth_string_discriminator:
+    return PointerAuthStringDiscriminator(*this, TheCall);
   // OpenCL v2.0, s6.13.16 - Pipe functions
   case Builtin::BIread_pipe:
   case Builtin::BIwrite_pipe:
diff --git a/clang/test/CodeGen/ptrauth-intrinsics.c b/clang/test/CodeGen/ptrauth-intrinsics.c
index 17f28dddb3801..d602325da8b17 100644
--- a/clang/test/CodeGen/ptrauth-intrinsics.c
+++ b/clang/test/CodeGen/ptrauth-intrinsics.c
@@ -71,3 +71,16 @@ void test_sign_generic_data() {
   // CHECK-NEXT: store i64 [[RESULT]], ptr @signature,
   signature = __builtin_ptrauth_sign_generic_data(fnptr, ptr_discriminator);
 }
+
+// CHECK-LABEL: define void @test_string_discriminator()
+void test_string_discriminator() {
+  // CHECK:      [[X:%.*]] = alloca i32
+
+  // Check a couple of random discriminators used by Swift.
+
+  // CHECK:      store i32 58298, ptr [[X]],
+  int x = __builtin_ptrauth_string_discriminator("InitializeWithCopy");
+
+  // CHECK:      store i32 9112, ptr [[X]],
+  x = __builtin_ptrauth_string_discriminator("DestroyArray");
+}
diff --git a/clang/test/Sema/ptrauth-intrinsics-macro.c b/clang/test/Sema/ptrauth-intrinsics-macro.c
index 07d6374045145..540f2846b7ce1 100644
--- a/clang/test/Sema/ptrauth-intrinsics-macro.c
+++ b/clang/test/Sema/ptrauth-intrinsics-macro.c
@@ -32,3 +32,8 @@ void test(int *dp, int value) {
   int t2 = ptrauth_sign_generic_data(dp, 0);
   (void)t2;
 }
+
+void test_string_discriminator(int *dp) {
+  ptrauth_extra_data_t t0 = ptrauth_string_discriminator("string");
+  (void)t0;
+}

[asl]

See mostly code style nits above

[kovdan01]

The changes themself in terms of functionality look OK to me - but I want someone else with deeper understanding of the context to take a look before this gets merged.

[kovdan01]

LGTM, but I want someone else with deeper understanding of the context to take a look before this gets merged.

Dismiss "changes requested" status

[kovdan01]

@ahmedbougacha After the latest force-push, the PR seems to contain SipHash-related changes from PRs 93902, 94393 and 94394. Could you please limit the changes only to those which are actually being intended to be merged as a part of this PR and resolve merge conflicts with the base branch?

I'd like to take one more look at this since there were updates since my latest feedback, but with different changes mixed, it's a bit messy.

[asl]

@ahmedbougacha Will you please rebase?

[asl]

Thanks!