Add missing atomic libcall support.

[jyknight]

Summary

Support for emitting libcalls for atomic_fetch_nand and
atomic_{add,sub,and,or,xor,nand}_fetch was missing; add it, and some
test cases.

[majnemer]

lib/CodeGen/CGAtomic.cpp

811–815

Do any of the tests cover this part of the change?

[jyknight]

lib/CodeGen/CGAtomic.cpp

811–815

None of the new ones do, but also note that this part is not a change in behavior, just changing from "default:" to listing all the options not listed above, so that missing a case emits a warning.

These calls are at least mostly covered by other tests, although I did not audit them to ensure 100% coverage.

[jyknight]

ping.

[jyknight]

ping

[compnerd]

lib/CodeGen/CGAtomic.cpp

788

Can you make this a bit more useful by indicating where it should have been handled (Id assume Sema).

811–815

There are existing tests for the C11 atomics.

846

Similar.

docs/LanguageExtensions.rst

 * ``__c11_atomic_fetch_or``
 * ``__c11_atomic_fetch_xor``
 
 The macros ``__ATOMIC_RELAXED``, ``__ATOMIC_CONSUME``, ``__ATOMIC_ACQUIRE``,
 ``__ATOMIC_RELEASE``, ``__ATOMIC_ACQ_REL``, and ``__ATOMIC_SEQ_CST`` are
 provided, with values corresponding to the enumerators of C11's
 ``memory_order`` enumeration.
 
+(Note that Clang additionally provides GCC-compatible ``__atomic_*``
+builtins)
+
 Low-level ARM exclusive memory builtins
 ---------------------------------------
 
 Clang provides overloaded builtins giving direct access to the three key ARM
 instructions for implementing atomic operations.
 
 .. code-block:: c
 

lib/CodeGen/CGAtomic.cpp

auto GetDest = [&] {
     }
     return Dest;
   };
 
   // Use a library call.  See: http://gcc.gnu.org/wiki/Atomic/GCCMM/LIbrary .
   if (UseLibcall) {
     bool UseOptimizedLibcall = false;
     switch (E->getOp()) {
+    case AtomicExpr::AO__c11_atomic_init:
+      llvm_unreachable("Already handled!");
+
     case AtomicExpr::AO__c11_atomic_fetch_add:
     case AtomicExpr::AO__atomic_fetch_add:
     case AtomicExpr::AO__c11_atomic_fetch_and:
     case AtomicExpr::AO__atomic_fetch_and:
     case AtomicExpr::AO__c11_atomic_fetch_or:
     case AtomicExpr::AO__atomic_fetch_or:
+    case AtomicExpr::AO__atomic_fetch_nand:
     case AtomicExpr::AO__c11_atomic_fetch_sub:
     case AtomicExpr::AO__atomic_fetch_sub:
     case AtomicExpr::AO__c11_atomic_fetch_xor:
     case AtomicExpr::AO__atomic_fetch_xor:
+    case AtomicExpr::AO__atomic_add_fetch:
+    case AtomicExpr::AO__atomic_and_fetch:
+    case AtomicExpr::AO__atomic_nand_fetch:
+    case AtomicExpr::AO__atomic_or_fetch:
+    case AtomicExpr::AO__atomic_sub_fetch:
+    case AtomicExpr::AO__atomic_xor_fetch:
       // For these, only library calls for certain sizes exist.
       UseOptimizedLibcall = true;
       break;
-    default:
+
+    case AtomicExpr::AO__c11_atomic_load:
+    case AtomicExpr::AO__c11_atomic_store:
+    case AtomicExpr::AO__c11_atomic_exchange:
+    case AtomicExpr::AO__c11_atomic_compare_exchange_weak:
+    case AtomicExpr::AO__c11_atomic_compare_exchange_strong:
+    case AtomicExpr::AO__atomic_load_n:
+    case AtomicExpr::AO__atomic_load:
+    case AtomicExpr::AO__atomic_store_n:
+    case AtomicExpr::AO__atomic_store:
+    case AtomicExpr::AO__atomic_exchange_n:
+    case AtomicExpr::AO__atomic_exchange:
+    case AtomicExpr::AO__atomic_compare_exchange_n:
+    case AtomicExpr::AO__atomic_compare_exchange:
       // Only use optimized library calls for sizes for which they exist.
       if (Size == 1 || Size == 2 || Size == 4 || Size == 8)
         UseOptimizedLibcall = true;
       break;
     }
 
     CallArgList Args;
     if (!UseOptimizedLibcall) {
       // For non-optimized library calls, the size is the first parameter
       Args.add(RValue::get(llvm::ConstantInt::get(SizeTy, Size)),
                getContext().getSizeType());
     }
     // Atomic address is the first or second parameter
     Args.add(RValue::get(EmitCastToVoidPtr(Ptr)), getContext().VoidPtrTy);
 
     std::string LibCallName;
     QualType LoweredMemTy =
       MemTy->isPointerType() ? getContext().getIntPtrType() : MemTy;
     QualType RetTy;
     bool HaveRetTy = false;
     switch (E->getOp()) {
+    case AtomicExpr::AO__c11_atomic_init:
+      llvm_unreachable("Already handled!");
+
     // There is only one libcall for compare an exchange, because there is no
     // optimisation benefit possible from a libcall version of a weak compare
     // and exchange.
     // bool __atomic_compare_exchange(size_t size, void *mem, void *expected,
     //                                void *desired, int success, int failure)
     // bool __atomic_compare_exchange_N(T *mem, T *expected, T desired,
     //                                  int success, int failure)
     case AtomicExpr::AO__c11_atomic_compare_exchange_weak:
case AtomicExpr::AO__atomic_fetch_sub:
       break;
     // T __atomic_fetch_xor_N(T *mem, T val, int order)
     case AtomicExpr::AO__c11_atomic_fetch_xor:
     case AtomicExpr::AO__atomic_fetch_xor:
       LibCallName = "__atomic_fetch_xor";
       AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1, MemTy,
                         E->getExprLoc(), sizeChars);
       break;
-    default: return EmitUnsupportedRValue(E, "atomic library call");
+    // T __atomic_fetch_nand_N(T *mem, T val, int order)
+    case AtomicExpr::AO__atomic_fetch_nand:
+      LibCallName = "__atomic_fetch_nand";
+      AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1, MemTy,
+                        E->getExprLoc(), sizeChars);
+      break;
+
+    // T __atomic_add_fetch_N(T *mem, T val, int order)
+    case AtomicExpr::AO__atomic_add_fetch:
+      LibCallName = "__atomic_add_fetch";
+      AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1, LoweredMemTy,
+                        E->getExprLoc(), sizeChars);
+      break;
+    // T __atomic_and_fetch_N(T *mem, T val, int order)
+    case AtomicExpr::AO__atomic_and_fetch:
+      LibCallName = "__atomic_and_fetch";
+      AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1, MemTy,
+                        E->getExprLoc(), sizeChars);
+      break;
+    // T __atomic_or_fetch_N(T *mem, T val, int order)
+    case AtomicExpr::AO__atomic_or_fetch:
+      LibCallName = "__atomic_or_fetch";
+      AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1, MemTy,
+                        E->getExprLoc(), sizeChars);
+      break;
+    // T __atomic_sub_fetch_N(T *mem, T val, int order)
+    case AtomicExpr::AO__atomic_sub_fetch:
+      LibCallName = "__atomic_sub_fetch";
+      AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1, LoweredMemTy,
+                        E->getExprLoc(), sizeChars);
+      break;
+    // T __atomic_xor_fetch_N(T *mem, T val, int order)
+    case AtomicExpr::AO__atomic_xor_fetch:
+      LibCallName = "__atomic_xor_fetch";
+      AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1, MemTy,
+                        E->getExprLoc(), sizeChars);
+      break;
+    // T __atomic_nand_fetch_N(T *mem, T val, int order)
+    case AtomicExpr::AO__atomic_nand_fetch:
+      LibCallName = "__atomic_nand_fetch";
+      AddDirectArgument(*this, Args, UseOptimizedLibcall, Val1, MemTy,
+                        E->getExprLoc(), sizeChars);
+      break;
     }
 
     // Optimized functions have the size in their name.
     if (UseOptimizedLibcall)
       LibCallName += "_" + llvm::utostr(Size);
     // By default, assume we return a value of the atomic type.
     if (!HaveRetTy) {
       if (UseOptimizedLibcall) {

test/CodeGen/atomic-ops-libcall.c

 }
 
 int *fp2a(int **p) {
   // CHECK: @fp2a
   // CHECK: {{%[^ ]*}} = tail call i32 @__atomic_fetch_sub_4(i8* {{%[0-9]+}}, i32 4, i32 0)
   // Note, the GNU builtins do not multiply by sizeof(T)!
   return __atomic_fetch_sub(p, 4, memory_order_relaxed);
 }
+
+int test_atomic_fetch_add(int *p) {
+  // CHECK: test_atomic_fetch_add
+  // CHECK: {{%[^ ]*}} = tail call i32 @__atomic_fetch_add_4(i8* {{%[0-9]+}}, i32 55, i32 5)
+  return __atomic_fetch_add(p, 55, memory_order_seq_cst);
+}
+
+int test_atomic_fetch_sub(int *p) {
+  // CHECK: test_atomic_fetch_sub
+  // CHECK: {{%[^ ]*}} = tail call i32 @__atomic_fetch_sub_4(i8* {{%[0-9]+}}, i32 55, i32 5)
+  return __atomic_fetch_sub(p, 55, memory_order_seq_cst);
+}
+
+int test_atomic_fetch_and(int *p) {
+  // CHECK: test_atomic_fetch_and
+  // CHECK: {{%[^ ]*}} = tail call i32 @__atomic_fetch_and_4(i8* {{%[0-9]+}}, i32 55, i32 5)
+  return __atomic_fetch_and(p, 55, memory_order_seq_cst);
+}
+
+int test_atomic_fetch_or(int *p) {
+  // CHECK: test_atomic_fetch_or
+  // CHECK: {{%[^ ]*}} = tail call i32 @__atomic_fetch_or_4(i8* {{%[0-9]+}}, i32 55, i32 5)
+  return __atomic_fetch_or(p, 55, memory_order_seq_cst);
+}
+
+int test_atomic_fetch_xor(int *p) {
+  // CHECK: test_atomic_fetch_xor
+  // CHECK: {{%[^ ]*}} = tail call i32 @__atomic_fetch_xor_4(i8* {{%[0-9]+}}, i32 55, i32 5)
+  return __atomic_fetch_xor(p, 55, memory_order_seq_cst);
+}
+
+int test_atomic_fetch_nand(int *p) {
+  // CHECK: test_atomic_fetch_nand
+  // CHECK: {{%[^ ]*}} = tail call i32 @__atomic_fetch_nand_4(i8* {{%[0-9]+}}, i32 55, i32 5)
+  return __atomic_fetch_nand(p, 55, memory_order_seq_cst);
+}
+
+int test_atomic_add_fetch(int *p) {
+  // CHECK: test_atomic_add_fetch
+  // CHECK: {{%[^ ]*}} = tail call i32 @__atomic_add_fetch_4(i8* {{%[0-9]+}}, i32 55, i32 5)
+  return __atomic_add_fetch(p, 55, memory_order_seq_cst);
+}
+
+int test_atomic_sub_fetch(int *p) {
+  // CHECK: test_atomic_sub_fetch
+  // CHECK: {{%[^ ]*}} = tail call i32 @__atomic_sub_fetch_4(i8* {{%[0-9]+}}, i32 55, i32 5)
+  return __atomic_sub_fetch(p, 55, memory_order_seq_cst);
+}
+
+int test_atomic_and_fetch(int *p) {
+  // CHECK: test_atomic_and_fetch
+  // CHECK: {{%[^ ]*}} = tail call i32 @__atomic_and_fetch_4(i8* {{%[0-9]+}}, i32 55, i32 5)
+  return __atomic_and_fetch(p, 55, memory_order_seq_cst);
+}
+
+int test_atomic_or_fetch(int *p) {
+  // CHECK: test_atomic_or_fetch
+  // CHECK: {{%[^ ]*}} = tail call i32 @__atomic_or_fetch_4(i8* {{%[0-9]+}}, i32 55, i32 5)
+  return __atomic_or_fetch(p, 55, memory_order_seq_cst);
+}
+
+int test_atomic_xor_fetch(int *p) {
+  // CHECK: test_atomic_xor_fetch
+  // CHECK: {{%[^ ]*}} = tail call i32 @__atomic_xor_fetch_4(i8* {{%[0-9]+}}, i32 55, i32 5)
+  return __atomic_xor_fetch(p, 55, memory_order_seq_cst);
+}
+
+int test_atomic_nand_fetch(int *p) {
+  // CHECK: test_atomic_nand_fetch
+  // CHECK: {{%[^ ]*}} = tail call i32 @__atomic_nand_fetch_4(i8* {{%[0-9]+}}, i32 55, i32 5)
+  return __atomic_nand_fetch(p, 55, memory_order_seq_cst);
+}