Add the -fsanitize=shadow-call-stack flag

[vlad.tsyrklevich]

Summary

Add support for the -fsanitize=shadow-call-stack flag which causes clang
to add ShadowCallStack attribute to functions compiled with that flag
enabled.

[pcc]

LGTM

Please add a test for the driver functionality.

[vlad.tsyrklevich]

• Add Driver tests

[kcc]

[didn't look at the code yet, just at the docs]

Please add a docs section describing how to handle leaf functions.
If they are not handled yet, no need to change the implementation in these pathches -- ok to do it later.

docs/ShadowCallStack.rst

15

prologue/epilogue?
(it's your native tongue, not mine, though)

21

Provide short comparison with RFG (more instructions, less memory, same racy attack)

39

link to wikipedia maybe?

42

... due to return branch predictor (or some such)

48

Say something about attacks that first try to discover the secret location of the shadow call stack.
side channels, thread spaying, whatever you have.

75

Please add a section that shows the assembly for the following example:

int foo() {
   return bar() + 1;
}

[kcc]

please also add a short comparison with Intel CET.

[vlad.tsyrklevich]

• Address Kostya's documentation feedback

[kcc]

LGTM modulo prolog vs prlogue and epilog vs epilogue

https://en.wiktionary.org/wiki/epilog says these are alternative spellings, so up to you.

docs/ShadowCallStack.rst

15

PTAL

[vlad.tsyrklevich]

docs/ShadowCallStack.rst

15

Forgot to submit this comment: It's used both ways across LLVM but I chose to go with this one just because that's how the PrologEpilogInserter pass wrote it.

[vlad.tsyrklevich]

• Small test, doc updates

[MaskRay]

cfe/trunk/test/Driver/sanitizer-ld.c

563 ↗

(On Diff #140869)

sanitizer-ld.c is for linker option tests. This merely checks incompatibility of compile flags and fsanitize.c may be more suitable.

docs/ShadowCallStack.rst

+===============
+ShadowCallStack
+===============
+
+.. contents::
+   :local:
+
+Introduction
+============
+
+ShadowCallStack is an **experimental** instrumentation pass, currently only
+implemented for x86_64, that protects programs against return address
+overwrites (e.g. stack buffer overflows.) It works by saving a function's return
+address to a separately allocated 'shadow call stack' in the function prolog and
+checking the return address on the stack against the shadow call stack in the
+function epilog.
+
+Comparison
+----------
+
+To optimize for memory consumption and cache locality, the shadow call stack
+stores an index followed by an array of return addresses. This is in contrast
+to other schemes, like :doc:`SafeStack`, that mirror the entire stack and
+trade-off consuming more memory for shorter function prologs and epilogs with
+fewer memory accesses. Similarly, `Return Flow Guard`_ consumes more memory with
+shorter function prologs and epilogs than ShadowCallStack but suffers from the
+same race conditions (see `Security`_). Intel `Control-flow Enforcement Technology`_
+(CET) is a proposed hardware extension that would add native support to
+use a shadow stack to store/check return addresses at call/return time. It
+would not suffer from race conditions at calls and returns and not incur the
+overhead of function instrumentation, but it does require operating system
+support.
+
+.. _`Return Flow Guard`: https://xlab.tencent.com/en/2016/11/02/return-flow-guard/
+.. _`Control-flow Enforcement Technology`: https://software.intel.com/sites/default/files/managed/4d/2a/control-flow-enforcement-technology-preview.pdf
+
+Compatibility
+-------------
+
+ShadowCallStack currently only supports x86_64. A runtime is not currently
+provided in compiler-rt so one must be provided by the compiled application.
+
+Security
+========
+
+ShadowCallStack is intended to be a stronger alternative to
+``-fstack-protector``. It protects from non-linear overflows and arbitrary
+memory writes to the return address slot; however, similarly to
+``-fstack-protector`` this protection suffers from race conditions because of
+the call-return semantics on x86_64. There is a short race between the call
+instruction and the first instruction in the function that reads the return
+address where an attacker could overwrite the return address and bypass
+ShadowCallStack. Similarly, there is a time-of-check-to-time-of-use race in the
+function prolog where an attacker could overwrite the return address after it
+has been checked and before it has been returned to. Modifying the call-return
+semantics to fix this on x86_64 would incur an unacceptable performance overhead
+due to return branch prediction.
+
+The instrumentation makes use of the ``gs`` segment register to reference the
+shadow call stack meaning that references to the shadow call stack do not have
+to be written to memory. This makes it possible to implement a runtime that is
+avoids exposing the address of the shadow call stack to attackers that can read
+arbitrary memory. However, attackers could still try to exploit side channels
+exposed by the operating system `[1]`_ `[2]`_ or processor `[3]`_ to discover
+the address of the shadow call stack.
+
+.. _`[1]`: https://eyalitkin.wordpress.com/2017/09/01/cartography-lighting-up-the-shadows/
+.. _`[2]`: https://www.blackhat.com/docs/eu-16/materials/eu-16-Goktas-Bypassing-Clangs-SafeStack.pdf
+.. _`[3]`: https://www.vusec.net/projects/anc/
+
+Usage
+=====
+
+To enable ShadowCallStack, just pass ``-fsanitize=shadow-call-stack`` flag to
+both compile and link command lines.
+
+Low-level API
+-------------
+
+``__has_feature(shadow_call_stack)``
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+In some cases one may need to execute different code depending on whether
+ShadowCallStack is enabled. The macro ``__has_feature(shadow_call_stack)`` can
+be used for this purpose.
+
+.. code-block:: c
+
+    #if defined(__has_feature)
+    #  if __has_feature(shadow_call_stack)
+    // code that builds only under ShadowCallStack
+    #  endif
+    #endif
+
+``__attribute__((no_sanitize("shadow-call-stack")))``
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Use ``__attribute__((no_sanitize("shadow-call-stack")))`` on a function
+declaration to specify that the shadow call stack instrumentation should not be
+applied to that function, even if enabled globally.
+
+Example
+=======
+
+The following example code:
+
+.. code-block:: c++
+
+    int foo() {
+      return bar() + 1;
+    }
+
+Generates the following x86_64 assembly when compiled with ``-O2``:
+
+.. code-block:: gas
+
+    push   %rax
+    callq  foo
+    add    $0x1,%eax
+    pop    %rcx
+    retq
+
+Adding ``-fsanitize=shadow-call-stack`` would output the following:
+
+.. code-block:: gas
+
+    mov    (%rsp),%r10
+    xor    %r11,%r11
+    addq   $0x8,%gs:(%r11)
+    mov    %gs:(%r11),%r11
+    mov    %r10,%gs:(%r11)
+    push   %rax
+    callq  foo
+    add    $0x1,%eax
+    pop    %rcx
+    xor    %r11,%r11
+    mov    %gs:(%r11),%r10
+    mov    %gs:(%r10),%r10
+    subq   $0x8,%gs:(%r11)
+    cmp    %r10,(%rsp)
+    jne    trap
+    retq
+
+    trap:
+    ud2

docs/index.rst

.. toctree::
    DataFlowSanitizer
    LeakSanitizer
    SanitizerCoverage
    SanitizerStats
    SanitizerSpecialCaseList
    ControlFlowIntegrity
    LTOVisibility
    SafeStack
+   ShadowCallStack
    SourceBasedCodeCoverage
    Modules
    MSVCCompatibility
    OpenMPSupport
    ThinLTO
    CommandGuide/index
    FAQ
 

include/clang/Basic/Sanitizers.def

 SANITIZER("cfi-vcall", CFIVCall)
 SANITIZER_GROUP("cfi", CFI,
                 CFIDerivedCast | CFIICall | CFIUnrelatedCast | CFINVCall |
                 CFIVCall)
 
 // Safe Stack
 SANITIZER("safe-stack", SafeStack)
 
+// Shadow Call Stack
+SANITIZER("shadow-call-stack", ShadowCallStack)
+
 // -fsanitize=undefined includes all the sanitizers which have low overhead, no
 // ABI or address space layout implications, and only catch undefined behavior.
 SANITIZER_GROUP("undefined", Undefined,
                 Alignment | Bool | Builtin | ArrayBounds | Enum |
                     FloatCastOverflow | FloatDivideByZero |
                     IntegerDivideByZero | NonnullAttribute | Null | ObjectSize |
                     PointerOverflow | Return | ReturnsNonnullAttribute | Shift |
                     SignedIntegerOverflow | Unreachable | VLABound | Function |

lib/CodeGen/CGDeclCXX.cpp

llvm::Function *CodeGenModule::CreateGlobalInitOrDestructFunction(
   if (getLangOpts().Sanitize.has(SanitizerKind::Memory) &&
       !isInSanitizerBlacklist(SanitizerKind::Memory, Fn, Loc))
     Fn->addFnAttr(llvm::Attribute::SanitizeMemory);
 
   if (getLangOpts().Sanitize.has(SanitizerKind::SafeStack) &&
       !isInSanitizerBlacklist(SanitizerKind::SafeStack, Fn, Loc))
     Fn->addFnAttr(llvm::Attribute::SafeStack);
 
+  if (getLangOpts().Sanitize.has(SanitizerKind::ShadowCallStack) &&
+      !isInSanitizerBlacklist(SanitizerKind::ShadowCallStack, Fn, Loc))
+    Fn->addFnAttr(llvm::Attribute::ShadowCallStack);
+
   return Fn;
 }
 
 /// Create a global pointer to a function that will initialize a global
 /// variable.  The user has requested that this pointer be emitted in a specific
 /// section.
 void CodeGenModule::EmitPointerToInitFunc(const VarDecl *D,
                                           llvm::GlobalVariable *GV,

lib/CodeGen/CodeGenFunction.cpp

#undef SANITIZER
   if (SanOpts.hasOneOf(SanitizerKind::HWAddress))
     Fn->addFnAttr(llvm::Attribute::SanitizeHWAddress);
   if (SanOpts.has(SanitizerKind::Thread))
     Fn->addFnAttr(llvm::Attribute::SanitizeThread);
   if (SanOpts.has(SanitizerKind::Memory))
     Fn->addFnAttr(llvm::Attribute::SanitizeMemory);
   if (SanOpts.has(SanitizerKind::SafeStack))
     Fn->addFnAttr(llvm::Attribute::SafeStack);
+  if (SanOpts.has(SanitizerKind::ShadowCallStack))
+    Fn->addFnAttr(llvm::Attribute::ShadowCallStack);
 
   // Ignore TSan memory acesses from within ObjC/ObjC++ dealloc, initialize,
   // .cxx_destruct, __destroy_helper_block_ and all of their calees at run time.
   if (SanOpts.has(SanitizerKind::Thread)) {
     if (const auto *OMD = dyn_cast_or_null<ObjCMethodDecl>(D)) {
       IdentifierInfo *II = OMD->getSelector().getIdentifierInfoForSlot(0);
       if (OMD->getMethodFamily() == OMF_dealloc ||
           OMD->getMethodFamily() == OMF_initialize ||

lib/Driver/SanitizerArgs.cpp

std::pair<SanitizerMask, SanitizerMask> IncompatibleGroups[] = {
       std::make_pair(Leak, Thread | Memory),
       std::make_pair(KernelAddress, Address | Leak | Thread | Memory),
       std::make_pair(HWAddress, Address | Thread | Memory | KernelAddress),
       std::make_pair(Efficiency, Address | HWAddress | Leak | Thread | Memory |
                                      KernelAddress),
       std::make_pair(Scudo, Address | HWAddress | Leak | Thread | Memory |
                                 KernelAddress | Efficiency),
       std::make_pair(SafeStack, Address | HWAddress | Leak | Thread | Memory |
-                                    KernelAddress | Efficiency)};
+                                    KernelAddress | Efficiency),
+      std::make_pair(ShadowCallStack, Address | HWAddress | Leak | Thread |
+                                          Memory | KernelAddress | Efficiency |
+                                          SafeStack)};
 
   // Enable toolchain specific default sanitizers if not explicitly disabled.
   SanitizerMask Default = TC.getDefaultSanitizers() & ~AllRemove;
 
   // Disable default sanitizers that are incompatible with explicitly requested
   // ones.
   for (auto G : IncompatibleGroups) {
     SanitizerMask Group = G.first;

lib/Driver/ToolChain.cpp

SanitizerMask Res = (Undefined & ~Vptr & ~Function) | (CFI & ~CFIICall) |
                       LocalBounds;
   if (getTriple().getArch() == llvm::Triple::x86 ||
       getTriple().getArch() == llvm::Triple::x86_64 ||
       getTriple().getArch() == llvm::Triple::arm ||
       getTriple().getArch() == llvm::Triple::aarch64 ||
       getTriple().getArch() == llvm::Triple::wasm32 ||
       getTriple().getArch() == llvm::Triple::wasm64)
     Res |= CFIICall;
+  if (getTriple().getArch() == llvm::Triple::x86_64)
+    Res |= ShadowCallStack;
   return Res;
 }
 
 void ToolChain::AddCudaIncludeArgs(const ArgList &DriverArgs,
                                    ArgStringList &CC1Args) const {}
 
 void ToolChain::AddIAMCUIncludeArgs(const ArgList &DriverArgs,
                                     ArgStringList &CC1Args) const {}

lib/Lex/PPMacroExpansion.cpp

return llvm::StringSwitch<bool>(Feature)
       .Case("is_sealed", LangOpts.CPlusPlus && LangOpts.MicrosoftExt)
       .Case("is_trivial", LangOpts.CPlusPlus)
       .Case("is_trivially_assignable", LangOpts.CPlusPlus)
       .Case("is_trivially_constructible", LangOpts.CPlusPlus)
       .Case("is_trivially_copyable", LangOpts.CPlusPlus)
       .Case("is_union", LangOpts.CPlusPlus)
       .Case("modules", LangOpts.Modules)
       .Case("safe_stack", LangOpts.Sanitize.has(SanitizerKind::SafeStack))
+      .Case("shadow_call_stack",
+            LangOpts.Sanitize.has(SanitizerKind::ShadowCallStack))
       .Case("tls", PP.getTargetInfo().isTLSSupported())
       .Case("underlying_type", LangOpts.CPlusPlus)
       .Default(false);
 }
 
 /// HasExtension - Return true if we recognize and implement the feature
 /// specified by the identifier, either as an extension or a standard language
 /// feature.

test/CodeGen/shadowcallstack-attr.c

+// RUN: %clang_cc1 -triple x86_64-linux-unknown -emit-llvm -o - %s -fsanitize=shadow-call-stack | FileCheck %s
+
+__attribute__((no_sanitize("shadow-call-stack")))
+int foo(int *a) { return *a; }
+
+ int bar(int *a) { return *a; }
+
+// CHECK: define i32 @foo(i32* %a) #[[FOO_ATTR:[0-9]+]] {
+// CHECK: define i32 @bar(i32* %a) #[[BAR_ATTR:[0-9]+]] {
+
+// CHECK-NOT: attributes #[[FOO_ATTR]] = { {{.*}}shadowcallstack{{.*}} }
+// CHECK: attributes #[[BAR_ATTR]] = { {{.*}}shadowcallstack{{.*}} }

test/Driver/sanitizer-ld.c

 // CHECK-SAFESTACK-LINUX: "{{(.*[^-.0-9A-Z_a-z])?}}ld{{(.exe)?}}"
 // CHECK-SAFESTACK-LINUX-NOT: "-lc"
 // CHECK-SAFESTACK-LINUX-NOT: whole-archive
 // CHECK-SAFESTACK-LINUX: libclang_rt.safestack-x86_64.a"
 // CHECK-SAFESTACK-LINUX: "-u" "__safestack_init"
 // CHECK-SAFESTACK-LINUX: "-lpthread"
 // CHECK-SAFESTACK-LINUX: "-ldl"
 
+// RUN: %clang -fsanitize=shadow-call-stack %s -### -o %t.o 2>&1 \
+// RUN:     -target x86_64-unknown-linux -fuse-ld=ld \
+// RUN:   | FileCheck --check-prefix=CHECK-SHADOWCALLSTACK-LINUX-X86-64 %s
+// CHECK-SHADOWCALLSTACK-LINUX-X86-64-NOT: error:
+
+// RUN: %clang -fsanitize=shadow-call-stack %s -### -o %t.o 2>&1 \
+// RUN:     -target x86-unknown-linux -fuse-ld=ld \
+// RUN:   | FileCheck --check-prefix=CHECK-SHADOWCALLSTACK-LINUX-X86 %s
+// CHECK-SHADOWCALLSTACK-LINUX-X86: error: unsupported option '-fsanitize=shadow-call-stack' for target 'x86-unknown-linux'
+
+// RUN: %clang -fsanitize=shadow-call-stack %s -### -o %t.o 2>&1 \
+// RUN:     -fsanitize=safe-stack -target x86_64-unknown-linux -fuse-ld=ld \
+// RUN:   | FileCheck --check-prefix=CHECK-SHADOWCALLSTACK-SAFESTACK %s
+// CHECK-SHADOWCALLSTACK-SAFESTACK: error: invalid argument '-fsanitize=shadow-call-stack' not allowed with '-fsanitize=safe-stack'
+
 // RUN: %clang -fsanitize=cfi -fsanitize-stats %s -### -o %t.o 2>&1 \
 // RUN:     -target x86_64-unknown-linux -fuse-ld=ld \
 // RUN:     --sysroot=%S/Inputs/basic_linux_tree \
 // RUN:   | FileCheck --check-prefix=CHECK-CFI-STATS-LINUX %s
 // CHECK-CFI-STATS-LINUX: "{{.*}}ld{{(.exe)?}}"
 // CHECK-CFI-STATS-LINUX: "--whole-archive" "{{[^"]*}}libclang_rt.stats_client-x86_64.a" "--no-whole-archive"
 // CHECK-CFI-STATS-LINUX-NOT: "--whole-archive"
 // CHECK-CFI-STATS-LINUX: "{{[^"]*}}libclang_rt.stats-x86_64.a"