Extend the uwtable attribute with unwind table kind

We have the `clang -cc1` command-line option `-funwind-tables=1|2` and
the codegen option `VALUE_CODEGENOPT(UnwindTables, 2, 0) ///< Unwind
tables (1) or asynchronous unwind tables (2)`. However, this is
encoded in LLVM IR by the presence or the absence of the `uwtable`
attribute, i.e.  we lose the information whether to generate want just
some unwind tables or asynchronous unwind tables.

Asynchronous unwind tables take more space in the runtime image, I'd
estimate something like 80-90% more, as the difference is adding
roughly the same number of CFI directives as for prologues, only a bit
simpler (e.g. `.cfi_offset reg, off` vs. `.cfi_restore reg`). Or even
more, if you consider tail duplication of epilogue blocks.
Asynchronous unwind tables could also restrict code generation to
having only a finite number of frame pointer adjustments (an example
of *not* having a finite number of `SP` adjustments is on AArch64 when
untagging the stack (MTE) in some cases the compiler can modify `SP`
in a loop).
Having the CFI precise up to an instruction generally also means one
cannot bundle together CFI instructions once the prologue is done,
they need to be interspersed with ordinary instructions, which means
extra `DW_CFA_advance_loc` commands, further increasing the unwind
tables size.

That is to say, async unwind tables impose a non-negligible overhead,
yet for the most common use cases (like C++ exceptions), they are not
even needed.

This patch extends the `uwtable` attribute with an optional
value:
      -  `uwtable` (default to `async`)
      -  `uwtable(sync)`, synchronous unwind tables
      -  `uwtable(async)`, asynchronous (instruction precise) unwind tables

Reviewed By: MaskRay

Differential Revision: https://reviews.llvm.org/D114543

Author: momchil-velikov

clang/lib/CodeGen/CGExpr.cpp

@@ -3188,7 +3188,7 @@ static void emitCheckHandlerCall(CodeGenFunction &CGF,
     B.addAttribute(llvm::Attribute::NoReturn)
         .addAttribute(llvm::Attribute::NoUnwind);
   }
-  B.addAttribute(llvm::Attribute::UWTable);
+  B.addUWTableAttr(llvm::UWTableKind::Default);
 
   llvm::FunctionCallee Fn = CGF.CGM.CreateRuntimeFunction(
       FnType, FnName,

clang/lib/CodeGen/CodeGenModule.cpp

@@ -828,7 +828,7 @@ void CodeGenModule::Release() {
   if (CodeGenOpts.NoPLT)
     getModule().setRtLibUseGOT();
   if (CodeGenOpts.UnwindTables)
-    getModule().setUwtable();
+    getModule().setUwtable(llvm::UWTableKind(CodeGenOpts.UnwindTables));
 
   switch (CodeGenOpts.getFramePointer()) {
   case CodeGenOptions::FramePointerKind::None:
@@ -1839,7 +1839,7 @@ void CodeGenModule::SetLLVMFunctionAttributesForDefinition(const Decl *D,
   llvm::AttrBuilder B(F->getContext());
 
   if (CodeGenOpts.UnwindTables)
-    B.addAttribute(llvm::Attribute::UWTable);
+    B.addUWTableAttr(llvm::UWTableKind(CodeGenOpts.UnwindTables));
 
   if (CodeGenOpts.StackClashProtector)
     B.addAttribute("probe-stack", "inline-asm");

clang/test/CodeGen/asan-globals.cpp

@@ -48,7 +48,7 @@ void func() {
 // RUN: %clang_cc1 -emit-llvm -fsanitize=address -funwind-tables=2 -o - %s | FileCheck %s --check-prefixes=UWTABLE
 // UWTABLE: define internal void @asan.module_dtor() #[[#ATTR:]] {
 // UWTABLE: attributes #[[#ATTR]] = { nounwind uwtable }
-// UWTABLE: ![[#]] = !{i32 7, !"uwtable", i32 1}
+// UWTABLE: ![[#]] = !{i32 7, !"uwtable", i32 2}
 
 // CHECK: !llvm.asan.globals = !{![[EXTRA_GLOBAL:[0-9]+]], ![[GLOBAL:[0-9]+]], ![[DYN_INIT_GLOBAL:[0-9]+]], ![[ATTR_GLOBAL:[0-9]+]], ![[IGNORELISTED_GLOBAL:[0-9]+]], ![[SECTIONED_GLOBAL:[0-9]+]], ![[SPECIAL_GLOBAL:[0-9]+]], ![[STATIC_VAR:[0-9]+]], ![[LITERAL:[0-9]+]]}
 // CHECK: ![[EXTRA_GLOBAL]] = !{{{.*}} ![[EXTRA_GLOBAL_LOC:[0-9]+]], !"extra_global", i1 false, i1 false}

clang/test/CodeGen/uwtable-attr.c

@@ -0,0 +1,30 @@
+// Test that function and modules attributes react on the command-line options,
+// it does not state the current behaviour makes sense in all cases (it does not).
+
+// RUN: %clang -S -emit-llvm -o - %s                                                    | FileCheck %s -check-prefixes=CHECK,DEFAULT
+// RUN: %clang -S -emit-llvm -o - %s -funwind-tables    -fno-asynchronous-unwind-tables | FileCheck %s -check-prefixes=CHECK,TABLES
+// RUN: %clang -S -emit-llvm -o - %s -fno-unwind-tables -fno-asynchronous-unwind-tables | FileCheck %s -check-prefixes=CHECK,NO_TABLES
+
+// RUN: %clang -S -emit-llvm -o - -x c++ %s                                                                     | FileCheck %s -check-prefixes=CHECK,DEFAULT
+// RUN: %clang -S -emit-llvm -o - -x c++ %s                  -funwind-tables    -fno-asynchronous-unwind-tables | FileCheck %s -check-prefixes=CHECK,TABLES
+// RUN: %clang -S -emit-llvm -o - -x c++ %s  -fno-exceptions -fno-unwind-tables -fno-asynchronous-unwind-tables | FileCheck %s -check-prefixes=CHECK,NO_TABLES
+
+#ifdef __cplusplus
+extern "C" void g(void);
+struct S { ~S(); };
+extern "C" int f() { S s; g(); return 0;};
+#else
+void g(void);
+int f() { g(); return 0; };
+#endif
+
+// CHECK: define {{.*}} @f() #[[#F:]]
+// CHECK: declare {{.*}} @g() #[[#]]
+
+// DEFAULT: attributes #[[#F]] = { {{.*}} uwtable{{ }}{{.*}} }
+// DEFAULT: ![[#]] = !{i32 7, !"uwtable", i32 2}
+
+// TABLES: attributes #[[#F]] = { {{.*}} uwtable(sync){{.*}} }
+// TABLES: ![[#]] = !{i32 7, !"uwtable", i32 1}
+
+// NO_TABLES-NOT: uwtable

llvm/bindings/go/llvm/ir_test.go

@@ -83,7 +83,6 @@ func TestAttributes(t *testing.T) {
 		"sspstrong",
 		"sanitize_thread",
 		"sanitize_memory",
-		"uwtable",
 		"zeroext",
 		"cold",
 		"nocf_check",

llvm/docs/LangRef.rst

@@ -2108,12 +2108,15 @@ example:
     function with a tail call. The prototype of a thunk should not be used for
     optimization purposes. The caller is expected to cast the thunk prototype to
     match the thunk target prototype.
-``uwtable``
+``uwtable[(sync|async)]``
     This attribute indicates that the ABI being targeted requires that
     an unwind table entry be produced for this function even if we can
     show that no exceptions passes by it. This is normally the case for
     the ELF x86-64 abi, but it can be disabled for some compilation
-    units.
+    units. The optional parameter describes what kind of unwind tables
+    to generate: ``sync`` for normal unwind tables, ``async`` for asynchronous
+    (instruction precise) unwind tables. Without the parameter, the attribute
+    ``uwtable`` is equivalent to ``uwtable(async)``.
 ``nocf_check``
     This attribute indicates that no control-flow check will be performed on
     the attributed entity. It disables -fcf-protection=<> for a specific
@@ -7215,8 +7218,9 @@ functions is small.
 - "frame-pointer": **Max**. The value can be 0, 1, or 2. A synthesized function
   will get the "frame-pointer" function attribute, with value being "none",
   "non-leaf", or "all", respectively.
-- "uwtable": **Max**. The value can be 0 or 1. If the value is 1, a synthesized
-  function will get the ``uwtable`` function attribute.
+- "uwtable": **Max**. The value can be 0, 1, or 2. If the value is 1, a synthesized
+  function will get the ``uwtable(sync)`` function attribute, if the value is 2,
+  a synthesized function will get the ``uwtable(async)`` function attribute.
 
 Objective-C Garbage Collection Module Flags Metadata
 ----------------------------------------------------

llvm/include/llvm/AsmParser/LLParser.h

@@ -263,6 +263,7 @@ namespace llvm {
     bool parseOptionalAlignment(MaybeAlign &Alignment,
                                 bool AllowParens = false);
     bool parseOptionalDerefAttrBytes(lltok::Kind AttrKind, uint64_t &Bytes);
+    bool parseOptionalUWTableKind(UWTableKind &Kind);
     bool parseScopeAndOrdering(bool IsAtomic, SyncScope::ID &SSID,
                                AtomicOrdering &Ordering);
     bool parseScope(SyncScope::ID &SSID);

llvm/include/llvm/AsmParser/LLToken.h

@@ -252,6 +252,8 @@ enum Kind {
   kw_immarg,
   kw_byref,
   kw_mustprogress,
+  kw_sync,
+  kw_async,
 
   kw_type,
   kw_opaque,

llvm/include/llvm/IR/Attributes.h

@@ -22,6 +22,7 @@
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Config/llvm-config.h"
 #include "llvm/Support/Alignment.h"
+#include "llvm/Support/CodeGen.h"
 #include "llvm/Support/PointerLikeTypeTraits.h"
 #include <bitset>
 #include <cassert>
@@ -130,6 +131,7 @@ class Attribute {
   static Attribute getWithByRefType(LLVMContext &Context, Type *Ty);
   static Attribute getWithPreallocatedType(LLVMContext &Context, Type *Ty);
   static Attribute getWithInAllocaType(LLVMContext &Context, Type *Ty);
+  static Attribute getWithUWTableKind(LLVMContext &Context, UWTableKind Kind);
 
   /// For a typed attribute, return the equivalent attribute with the type
   /// changed to \p ReplacementTy.
@@ -223,6 +225,9 @@ class Attribute {
   /// unknown.
   Optional<unsigned> getVScaleRangeMax() const;
 
+  // Returns the unwind table kind.
+  UWTableKind getUWTableKind() const;
+
   /// The Attribute is converted to a string of equivalent mnemonic. This
   /// is, presumably, for writing out the mnemonics for the assembly writer.
   std::string getAsString(bool InAttrGrp = false) const;
@@ -353,6 +358,7 @@ class AttributeSet {
   std::pair<unsigned, Optional<unsigned>> getAllocSizeArgs() const;
   unsigned getVScaleRangeMin() const;
   Optional<unsigned> getVScaleRangeMax() const;
+  UWTableKind getUWTableKind() const;
   std::string getAsString(bool InAttrGrp = false) const;
 
   /// Return true if this attribute set belongs to the LLVMContext.
@@ -841,6 +847,9 @@ class AttributeList {
   /// arg.
   uint64_t getParamDereferenceableOrNullBytes(unsigned ArgNo) const;
 
+  /// Get the unwind table kind requested for the function.
+  UWTableKind getUWTableKind() const;
+
   /// Return the attributes at the index as a string.
   std::string getAsString(unsigned Index, bool InAttrGrp = false) const;
 
@@ -1190,6 +1199,10 @@ class AttrBuilder {
   /// Attribute.getIntValue().
   AttrBuilder &addVScaleRangeAttrFromRawRepr(uint64_t RawVScaleRangeRepr);
 
+  /// This turns the unwind table kind into the form used internally in
+  /// Attribute.
+  AttrBuilder &addUWTableAttr(UWTableKind Kind);
+
   ArrayRef<Attribute> attrs() const { return Attrs; }
 
   bool operator==(const AttrBuilder &B) const;

llvm/include/llvm/IR/Attributes.td

@@ -273,7 +273,7 @@ def SwiftSelf : EnumAttr<"swiftself", [ParamAttr]>;
 def SwiftAsync : EnumAttr<"swiftasync", [ParamAttr]>;
 
 /// Function must be in a unwind table.
-def UWTable : EnumAttr<"uwtable", [FnAttr]>;
+def UWTable : IntAttr<"uwtable", [FnAttr]>;
 
 /// Minimum/Maximum vscale value for function.
 def VScaleRange : IntAttr<"vscale_range", [FnAttr]>;