cfi-icall: Allow the jump table to be optionally made non-canonical.

The default behavior of Clang's indirect function call checker will replace
the address of each CFI-checked function in the output file's symbol table
with the address of a jump table entry which will pass CFI checks. We refer
to this as making the jump table `canonical`. This property allows code that
was not compiled with ``-fsanitize=cfi-icall`` to take a CFI-valid address
of a function, but it comes with a couple of caveats that are especially
relevant for users of cross-DSO CFI:

- There is a performance and code size overhead associated with each
  exported function, because each such function must have an associated
  jump table entry, which must be emitted even in the common case where the
  function is never address-taken anywhere in the program, and must be used
  even for direct calls between DSOs, in addition to the PLT overhead.

- There is no good way to take a CFI-valid address of a function written in
  assembly or a language not supported by Clang. The reason is that the code
  generator would need to insert a jump table in order to form a CFI-valid
  address for assembly functions, but there is no way in general for the
  code generator to determine the language of the function. This may be
  possible with LTO in the intra-DSO case, but in the cross-DSO case the only
  information available is the function declaration. One possible solution
  is to add a C wrapper for each assembly function, but these wrappers can
  present a significant maintenance burden for heavy users of assembly in
  addition to adding runtime overhead.

For these reasons, we provide the option of making the jump table non-canonical
with the flag ``-fno-sanitize-cfi-canonical-jump-tables``. When the jump
table is made non-canonical, symbol table entries point directly to the
function body. Any instances of a function's address being taken in C will
be replaced with a jump table address.

This scheme does have its own caveats, however. It does end up breaking
function address equality more aggressively than the default behavior,
especially in cross-DSO mode which normally preserves function address
equality entirely.

Furthermore, it is occasionally necessary for code not compiled with
``-fsanitize=cfi-icall`` to take a function address that is valid
for CFI. For example, this is necessary when a function's address
is taken by assembly code and then called by CFI-checking C code. The
``__attribute__((cfi_jump_table_canonical))`` attribute may be used to make
the jump table entry of a specific function canonical so that the external
code will end up taking a address for the function that will pass CFI checks.

Fixes PR41972.

Differential Revision: https://reviews.llvm.org/D65629

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@368495 91177308-0d34-0410-b5e6-96231b3b80d8

docs/ControlFlowIntegrity.rst

@@ -235,6 +235,54 @@ long as the qualifiers for the type they point to match. For example, ``char*``,
 ``-fsanitize-cfi-icall-generalize-pointers`` is not compatible with
 ``-fsanitize-cfi-cross-dso``.
 
+.. _cfi-canonical-jump-tables:
+
+``-fsanitize-cfi-canonical-jump-tables``
+----------------------------------------
+
+The default behavior of Clang's indirect function call checker will replace
+the address of each CFI-checked function in the output file's symbol table
+with the address of a jump table entry which will pass CFI checks. We refer
+to this as making the jump table `canonical`. This property allows code that
+was not compiled with ``-fsanitize=cfi-icall`` to take a CFI-valid address
+of a function, but it comes with a couple of caveats that are especially
+relevant for users of cross-DSO CFI:
+
+- There is a performance and code size overhead associated with each
+  exported function, because each such function must have an associated
+  jump table entry, which must be emitted even in the common case where the
+  function is never address-taken anywhere in the program, and must be used
+  even for direct calls between DSOs, in addition to the PLT overhead.
+
+- There is no good way to take a CFI-valid address of a function written in
+  assembly or a language not supported by Clang. The reason is that the code
+  generator would need to insert a jump table in order to form a CFI-valid
+  address for assembly functions, but there is no way in general for the
+  code generator to determine the language of the function. This may be
+  possible with LTO in the intra-DSO case, but in the cross-DSO case the only
+  information available is the function declaration. One possible solution
+  is to add a C wrapper for each assembly function, but these wrappers can
+  present a significant maintenance burden for heavy users of assembly in
+  addition to adding runtime overhead.
+
+For these reasons, we provide the option of making the jump table non-canonical
+with the flag ``-fno-sanitize-cfi-canonical-jump-tables``. When the jump
+table is made non-canonical, symbol table entries point directly to the
+function body. Any instances of a function's address being taken in C will
+be replaced with a jump table address.
+
+This scheme does have its own caveats, however. It does end up breaking
+function address equality more aggressively than the default behavior,
+especially in cross-DSO mode which normally preserves function address
+equality entirely.
+
+Furthermore, it is occasionally necessary for code not compiled with
+``-fsanitize=cfi-icall`` to take a function address that is valid
+for CFI. For example, this is necessary when a function's address
+is taken by assembly code and then called by CFI-checking C code. The
+``__attribute__((cfi_canonical_jump_table))`` attribute may be used to make
+the jump table entry of a specific function canonical so that the external
+code will end up taking a address for the function that will pass CFI checks.
 
 ``-fsanitize=cfi-icall`` and ``-fsanitize=function``
 ----------------------------------------------------

include/clang/Basic/Attr.td

@@ -2436,6 +2436,12 @@ def NoSanitizeSpecific : InheritableAttr {
   let ASTNode = 0;
 }
 
+def CFICanonicalJumpTable : InheritableAttr {
+  let Spellings = [Clang<"cfi_canonical_jump_table">];
+  let Subjects = SubjectList<[Function], ErrorDiag>;
+  let Documentation = [CFICanonicalJumpTableDocs];
+}
+
 // C/C++ Thread safety attributes (e.g. for deadlock, data race checking)
 // Not all of these attributes will be given a [[]] spelling. The attributes
 // which require access to function parameter names cannot use the [[]] spelling

include/clang/Basic/AttrDocs.td

@@ -2221,6 +2221,18 @@ to avoid false positives in other places.
   }];
 }
 
+def CFICanonicalJumpTableDocs : Documentation {
+  let Category = DocCatFunction;
+  let Heading = "cfi_canonical_jump_table";
+  let Content = [{
+.. _langext-cfi_canonical_jump_table:
+
+Use ``__attribute__((cfi_canonical_jump_table))`` on a function declaration to
+make the function's CFI jump table canonical. See :ref:`the CFI documentation
+<cfi-canonical-jump-tables>` for more details.
+  }];
+}
+
 def DocCatTypeSafety : DocumentationCategory<"Type Safety Checking"> {
   let Content = [{
 Clang supports additional attributes to enable checking type safety properties

include/clang/Basic/CodeGenOptions.def

@@ -195,6 +195,8 @@ CODEGENOPT(SanitizeMinimalRuntime, 1, 0) ///< Use "_minimal" sanitizer runtime f
                                          ///< diagnostics.
 CODEGENOPT(SanitizeCfiICallGeneralizePointers, 1, 0) ///< Generalize pointer types in
                                                      ///< CFI icall function signatures
+CODEGENOPT(SanitizeCfiCanonicalJumpTables, 1, 0) ///< Make jump table symbols canonical
+                                                 ///< instead of creating a local jump table.
 CODEGENOPT(SanitizeCoverageType, 2, 0) ///< Type of sanitizer coverage
                                        ///< instrumentation.
 CODEGENOPT(SanitizeCoverageIndirectCalls, 1, 0) ///< Enable sanitizer coverage

include/clang/Driver/Options.td

@@ -1061,6 +1061,13 @@ def fno_sanitize_cfi_cross_dso : Flag<["-"], "fno-sanitize-cfi-cross-dso">,
 def fsanitize_cfi_icall_generalize_pointers : Flag<["-"], "fsanitize-cfi-icall-generalize-pointers">,
                                               Group<f_clang_Group>,
                                               HelpText<"Generalize pointers in CFI indirect call type signature checks">;
+def fsanitize_cfi_canonical_jump_tables : Flag<["-"], "fsanitize-cfi-canonical-jump-tables">,
+                                          Group<f_clang_Group>,
+                                          HelpText<"Make the jump table addresses canonical in the symbol table">;
+def fno_sanitize_cfi_canonical_jump_tables : Flag<["-"], "fno-sanitize-cfi-canonical-jump-tables">,
+                                             Group<f_clang_Group>,
+                                             Flags<[CoreOption, DriverOption]>,
+                                             HelpText<"Do not make the jump table addresses canonical in the symbol table">;
 def fsanitize_stats : Flag<["-"], "fsanitize-stats">,
                               Group<f_clang_Group>,
                               HelpText<"Enable sanitizer statistics gathering.">;

include/clang/Driver/SanitizerArgs.h

@@ -32,6 +32,7 @@ class SanitizerArgs {
   bool MsanUseAfterDtor = true;
   bool CfiCrossDso = false;
   bool CfiICallGeneralizePointers = false;
+  bool CfiCanonicalJumpTables = false;
   int AsanFieldPadding = 0;
   bool SharedRuntime = false;
   bool AsanUseAfterScope = true;

lib/CodeGen/CodeGenFunction.cpp

@@ -762,6 +762,9 @@ void CodeGenFunction::StartFunction(GlobalDecl GD,
   if (CGM.getCodeGenOpts().ProfileSampleAccurate)
     Fn->addFnAttr("profile-sample-accurate");
 
+  if (D && D->hasAttr<CFICanonicalJumpTableAttr>())
+    Fn->addFnAttr("cfi-canonical-jump-table");
+
   if (getLangOpts().OpenCL) {
     // Add metadata for a kernel function.
     if (const FunctionDecl *FD = dyn_cast_or_null<FunctionDecl>(D))

lib/CodeGen/CodeGenModule.cpp

@@ -535,6 +535,12 @@ void CodeGenModule::Release() {
     getModule().addModuleFlag(llvm::Module::Override, "Cross-DSO CFI", 1);
   }
 
+  if (LangOpts.Sanitize.has(SanitizerKind::CFIICall)) {
+    getModule().addModuleFlag(llvm::Module::Override,
+                              "CFI Canonical Jump Tables",
+                              CodeGenOpts.SanitizeCfiCanonicalJumpTables);
+  }
+
   if (CodeGenOpts.CFProtectionReturn &&
       Target.checkCFProtectionReturnSupported(getDiags())) {
     // Indicate that we want to instrument return control flow protection.
@@ -1605,10 +1611,17 @@ void CodeGenModule::SetLLVMFunctionAttributesForDefinition(const Decl *D,
       F->setAlignment(2);
   }
 
-  // In the cross-dso CFI mode, we want !type attributes on definitions only.
-  if (CodeGenOpts.SanitizeCfiCrossDso)
-    if (auto *FD = dyn_cast<FunctionDecl>(D))
-      CreateFunctionTypeMetadataForIcall(FD, F);
+  // In the cross-dso CFI mode with canonical jump tables, we want !type
+  // attributes on definitions only.
+  if (CodeGenOpts.SanitizeCfiCrossDso &&
+      CodeGenOpts.SanitizeCfiCanonicalJumpTables) {
+    if (auto *FD = dyn_cast<FunctionDecl>(D)) {
+      // Skip available_externally functions. They won't be codegen'ed in the
+      // current module anyway.
+      if (getContext().GetGVALinkageForFunction(FD) != GVA_AvailableExternally)
+        CreateFunctionTypeMetadataForIcall(FD, F);
+    }
+  }
 
   // Emit type metadata on member functions for member function pointer checks.
   // These are only ever necessary on definitions; we're guaranteed that the
@@ -1765,14 +1778,6 @@ void CodeGenModule::CreateFunctionTypeMetadataForIcall(const FunctionDecl *FD,
   if (isa<CXXMethodDecl>(FD) && !cast<CXXMethodDecl>(FD)->isStatic())
     return;
 
-  // Additionally, if building with cross-DSO support...
-  if (CodeGenOpts.SanitizeCfiCrossDso) {
-    // Skip available_externally functions. They won't be codegen'ed in the
-    // current module anyway.
-    if (getContext().GetGVALinkageForFunction(FD) == GVA_AvailableExternally)
-      return;
-  }
-
   llvm::Metadata *MD = CreateMetadataIdentifierForType(FD->getType());
   F->addTypeMetadata(0, MD);
   F->addTypeMetadata(0, CreateMetadataIdentifierGeneralized(FD->getType()));
@@ -1849,8 +1854,11 @@ void CodeGenModule::SetFunctionAttributes(GlobalDecl GD, llvm::Function *F,
       F->setUnnamedAddr(llvm::GlobalValue::UnnamedAddr::Global);
 
   // Don't emit entries for function declarations in the cross-DSO mode. This
-  // is handled with better precision by the receiving DSO.
-  if (!CodeGenOpts.SanitizeCfiCrossDso)
+  // is handled with better precision by the receiving DSO. But if jump tables
+  // are non-canonical then we need type metadata in order to produce the local
+  // jump table.
+  if (!CodeGenOpts.SanitizeCfiCrossDso ||
+      !CodeGenOpts.SanitizeCfiCanonicalJumpTables)
     CreateFunctionTypeMetadataForIcall(FD, F);
 
   if (getLangOpts().OpenMP && FD->hasAttr<OMPDeclareSimdDeclAttr>())

lib/Driver/SanitizerArgs.cpp

@@ -636,6 +636,10 @@ SanitizerArgs::SanitizerArgs(const ToolChain &TC,
       D.Diag(diag::err_drv_argument_not_allowed_with)
           << "-fsanitize-cfi-cross-dso"
           << "-fsanitize-cfi-icall-generalize-pointers";
+
+    CfiCanonicalJumpTables =
+        Args.hasFlag(options::OPT_fsanitize_cfi_canonical_jump_tables,
+                     options::OPT_fno_sanitize_cfi_canonical_jump_tables, true);
   }
 
   Stats = Args.hasFlag(options::OPT_fsanitize_stats,
@@ -976,6 +980,9 @@ void SanitizerArgs::addArgs(const ToolChain &TC, const llvm::opt::ArgList &Args,
   if (CfiICallGeneralizePointers)
     CmdArgs.push_back("-fsanitize-cfi-icall-generalize-pointers");
 
+  if (CfiCanonicalJumpTables)
+    CmdArgs.push_back("-fsanitize-cfi-canonical-jump-tables");
+
   if (Stats)
     CmdArgs.push_back("-fsanitize-stats");
 

lib/Frontend/CompilerInvocation.cpp

@@ -1139,6 +1139,8 @@ static bool ParseCodeGenArgs(CodeGenOptions &Opts, ArgList &Args, InputKind IK,
   Opts.SanitizeCfiCrossDso = Args.hasArg(OPT_fsanitize_cfi_cross_dso);
   Opts.SanitizeCfiICallGeneralizePointers =
       Args.hasArg(OPT_fsanitize_cfi_icall_generalize_pointers);
+  Opts.SanitizeCfiCanonicalJumpTables =
+      Args.hasArg(OPT_fsanitize_cfi_canonical_jump_tables);
   Opts.SanitizeStats = Args.hasArg(OPT_fsanitize_stats);
   if (Arg *A = Args.getLastArg(
           OPT_fsanitize_address_poison_custom_array_cookie,

lib/Sema/SemaDeclAttr.cpp

@@ -7198,6 +7198,9 @@ static void ProcessDeclAttribute(Sema &S, Scope *scope, Decl *D,
     // Interacts with -fstack-protector options.
     handleSimpleAttribute<NoStackProtectorAttr>(S, D, AL);
     break;
+  case ParsedAttr::AT_CFICanonicalJumpTable:
+    handleSimpleAttribute<CFICanonicalJumpTableAttr>(S, D, AL);
+    break;
   case ParsedAttr::AT_StdCall:
   case ParsedAttr::AT_CDecl:
   case ParsedAttr::AT_FastCall:

test/CodeGen/cfi-icall-canonical-jump-tables.c

@@ -0,0 +1,24 @@
+// RUN: %clang_cc1 -triple x86_64-unknown-linux -fsanitize=cfi-icall -fsanitize-cfi-cross-dso -emit-llvm -o - %s | FileCheck --check-prefixes=CHECK,NOCANON %s
+// RUN: %clang_cc1 -triple x86_64-unknown-linux -fsanitize=cfi-icall -fsanitize-cfi-cross-dso -fsanitize-cfi-canonical-jump-tables -emit-llvm -o - %s | FileCheck --check-prefixes=CHECK,CANON %s
+
+void ext(void);
+
+// CHECK: define void @f({{.*}} [[ATTR1:#[0-9]+]]
+void f() {
+  ext();
+}
+
+// NOCANON: declare !type {{.*}} @ext()
+// CANON: declare void @ext()
+
+// CHECK: define void @g({{.*}} [[ATTR2:#[0-9]+]]
+__attribute__((cfi_canonical_jump_table)) void g() {}
+
+// CHECK: [[ATTR1]] = {
+// CHECK-NOT: "cfi-canonical-jump-table"
+// CHECK: }
+
+// CHECK: [[ATTR2]] = { {{.*}} "cfi-canonical-jump-table" {{.*}} }
+
+// NOCANON: !{i32 4, !"CFI Canonical Jump Tables", i32 0}
+// CANON: !{i32 4, !"CFI Canonical Jump Tables", i32 1}

test/CodeGen/cfi-icall-cross-dso.c

@@ -1,27 +1,27 @@
 // RUN: %clang_cc1 -triple x86_64-unknown-linux -O1 -fno-experimental-new-pass-manager \
 // RUN:   -fsanitize=cfi-icall -fsanitize-cfi-cross-dso \
-// RUN:   -emit-llvm -o - %s | FileCheck \
+// RUN:   -fsanitize-cfi-canonical-jump-tables -emit-llvm -o - %s | FileCheck \
 // RUN:       --check-prefix=CHECK --check-prefix=CHECK-DIAG \
 // RUN:       --check-prefix=ITANIUM --check-prefix=ITANIUM-DIAG \
 // RUN:       %s
 
 // RUN: %clang_cc1 -triple x86_64-unknown-linux -O1 -fno-experimental-new-pass-manager \
 // RUN:   -fsanitize=cfi-icall -fsanitize-cfi-cross-dso -fsanitize-trap=cfi-icall \
-// RUN:   -emit-llvm -o - %s | FileCheck \
+// RUN:   -fsanitize-cfi-canonical-jump-tables -emit-llvm -o - %s | FileCheck \
 // RUN:       --check-prefix=CHECK \
 // RUN:       --check-prefix=ITANIUM --check-prefix=ITANIUM-TRAP \
 // RUN:       %s
 
 // RUN: %clang_cc1 -triple x86_64-pc-windows-msvc -O1 -fno-experimental-new-pass-manager \
 // RUN:   -fsanitize=cfi-icall -fsanitize-cfi-cross-dso \
-// RUN:   -emit-llvm -o - %s | FileCheck \
+// RUN:   -fsanitize-cfi-canonical-jump-tables -emit-llvm -o - %s | FileCheck \
 // RUN:       --check-prefix=CHECK --check-prefix=CHECK-DIAG \
 // RUN:       --check-prefix=MS --check-prefix=MS-DIAG \
 // RUN:       %s
 
 // RUN: %clang_cc1 -triple x86_64-pc-windows-msvc -O1 -fno-experimental-new-pass-manager \
 // RUN:   -fsanitize=cfi-icall -fsanitize-cfi-cross-dso -fsanitize-trap=cfi-icall \
-// RUN:   -emit-llvm -o - %s | FileCheck \
+// RUN:   -fsanitize-cfi-canonical-jump-tables -emit-llvm -o - %s | FileCheck \
 // RUN:       --check-prefix=CHECK \
 // RUN:       --check-prefix=MS --check-prefix=MS-TRAP \
 // RUN:       %s

test/Driver/fsanitize.c

@@ -621,6 +621,12 @@
 // RUN: %clang -target x86_64-linux-gnu -fsanitize=cfi-icall -fsanitize-cfi-icall-generalize-pointers -fsanitize-cfi-cross-dso -fvisibility=hidden -flto -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-CFI-GENERALIZE-AND-CROSS-DSO
 // CHECK-CFI-GENERALIZE-AND-CROSS-DSO: error: invalid argument '-fsanitize-cfi-cross-dso' not allowed with '-fsanitize-cfi-icall-generalize-pointers'
 
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=cfi-icall -fsanitize-cfi-canonical-jump-tables -fvisibility=hidden -flto -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-CFI-CANONICAL-JUMP-TABLES
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=cfi-icall -fno-sanitize-cfi-canonical-jump-tables -fvisibility=hidden -flto -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-NO-CFI-CANONICAL-JUMP-TABLES
+// RUN: %clang -target x86_64-linux-gnu -fsanitize=cfi-icall -fvisibility=hidden -flto -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-CFI-CANONICAL-JUMP-TABLES
+// CHECK-CFI-CANONICAL-JUMP-TABLES: -fsanitize-cfi-canonical-jump-tables
+// CHECK-NO-CFI-CANONICAL-JUMP-TABLES-NOT: -fsanitize-cfi-canonical-jump-tables
+
 // RUN: %clang -target x86_64-linux-gnu -fsanitize=cfi -fsanitize-stats -flto -c %s -### 2>&1 | FileCheck %s --check-prefix=CHECK-CFI-STATS
 // CHECK-CFI-STATS: -fsanitize-stats
 

test/Misc/pragma-attribute-supported-attributes-list.test

@@ -21,6 +21,7 @@
 // CHECK-NEXT: Availability ((SubjectMatchRule_record, SubjectMatchRule_enum, SubjectMatchRule_enum_constant, SubjectMatchRule_field, SubjectMatchRule_function, SubjectMatchRule_namespace, SubjectMatchRule_objc_category, SubjectMatchRule_objc_implementation, SubjectMatchRule_objc_interface, SubjectMatchRule_objc_method, SubjectMatchRule_objc_property, SubjectMatchRule_objc_protocol, SubjectMatchRule_record, SubjectMatchRule_type_alias, SubjectMatchRule_variable))
 // CHECK-NEXT: CFAuditedTransfer (SubjectMatchRule_function)
 // CHECK-NEXT: CFConsumed (SubjectMatchRule_variable_is_parameter)
+// CHECK-NEXT: CFICanonicalJumpTable (SubjectMatchRule_function)
 // CHECK-NEXT: CFUnknownTransfer (SubjectMatchRule_function)
 // CHECK-NEXT: CPUDispatch (SubjectMatchRule_function)
 // CHECK-NEXT: CPUSpecific (SubjectMatchRule_function)

test/SemaCXX/attr-cfi-canonical-jump-table.cpp

@@ -0,0 +1,11 @@
+// RUN: %clang_cc1 -triple x86_64-unknown-linux -fsyntax-only -verify %s
+
+__attribute__((cfi_canonical_jump_table)) void fdecl();
+
+__attribute__((cfi_canonical_jump_table)) void f() {}
+
+struct S {
+  __attribute__((cfi_canonical_jump_table)) void f() {}
+};
+
+__attribute__((cfi_canonical_jump_table)) int i; // expected-error {{'cfi_canonical_jump_table' attribute only applies to functions}}