[Attributor] Simplify loads from constant globals

If a global is constant and the initializer is known we can simplify
loads from it as the value has to be the initializer.

Author: jdoerfert

llvm/lib/Transforms/IPO/Attributor.cpp

@@ -225,7 +225,9 @@ Constant *AA::getInitialValueForObj(Value &Obj, Type &Ty,
   if (isAllocationFn(&Obj, TLI))
     return getInitialValueOfAllocation(&cast<CallBase>(Obj), TLI, &Ty);
   auto *GV = dyn_cast<GlobalVariable>(&Obj);
-  if (!GV || !GV->hasLocalLinkage())
+  if (!GV)
+    return nullptr;
+  if (!GV->hasLocalLinkage() && !(GV->isConstant() && GV->hasInitializer()))
     return nullptr;
   if (!GV->hasInitializer())
     return UndefValue::get(&Ty);
@@ -365,7 +367,8 @@ static bool getPotentialCopiesOfMemoryValue(
       return false;
     }
     if (auto *GV = dyn_cast<GlobalVariable>(Obj))
-      if (!GV->hasLocalLinkage()) {
+      if (!GV->hasLocalLinkage() &&
+          !(GV->isConstant() && GV->hasInitializer())) {
         LLVM_DEBUG(dbgs() << "Underlying object is global with external "
                              "linkage, not supported yet: "
                           << *Obj << "\n";);
@@ -1022,6 +1025,8 @@ Optional<Constant *>
 Attributor::getAssumedConstant(const IRPosition &IRP,
                                const AbstractAttribute &AA,
                                bool &UsedAssumedInformation) {
+  if (auto *C = dyn_cast<Constant>(&IRP.getAssociatedValue()))
+    return C;
   // First check all callbacks provided by outside AAs. If any of them returns
   // a non-null value that is different from the associated value, or None, we
   // assume it's simpliied.

llvm/test/Transforms/Attributor/ArgumentPromotion/chained.ll

@@ -1,6 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-attributes --check-globals
-; RUN: opt -attributor -enable-new-pm=0 -attributor-manifest-internal  -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=2 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_NPM,NOT_CGSCC_OPM,NOT_TUNIT_NPM,IS__TUNIT____,IS________OPM,IS__TUNIT_OPM
-; RUN: opt -aa-pipeline=basic-aa -passes=attributor -attributor-manifest-internal  -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=2 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_OPM,NOT_CGSCC_NPM,NOT_TUNIT_OPM,IS__TUNIT____,IS________NPM,IS__TUNIT_NPM
+; RUN: opt -attributor -enable-new-pm=0 -attributor-manifest-internal  -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=4 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_NPM,NOT_CGSCC_OPM,NOT_TUNIT_NPM,IS__TUNIT____,IS________OPM,IS__TUNIT_OPM
+; RUN: opt -aa-pipeline=basic-aa -passes=attributor -attributor-manifest-internal  -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=4 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_OPM,NOT_CGSCC_NPM,NOT_TUNIT_OPM,IS__TUNIT____,IS________NPM,IS__TUNIT_NPM
 ; RUN: opt -attributor-cgscc -enable-new-pm=0 -attributor-manifest-internal  -attributor-annotate-decl-cs -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_TUNIT_NPM,NOT_TUNIT_OPM,NOT_CGSCC_NPM,IS__CGSCC____,IS________OPM,IS__CGSCC_OPM
 ; RUN: opt -aa-pipeline=basic-aa -passes=attributor-cgscc -attributor-manifest-internal  -attributor-annotate-decl-cs -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_TUNIT_NPM,NOT_TUNIT_OPM,NOT_CGSCC_OPM,IS__CGSCC____,IS________NPM,IS__CGSCC_NPM
 

llvm/test/Transforms/OpenMP/barrier_removal.ll

@@ -97,12 +97,11 @@ define void @neg_empty_2() {
 define void @pos_constant_loads() {
 ; CHECK-LABEL: define {{[^@]+}}@pos_constant_loads() {
 ; CHECK-NEXT:    [[ARG:%.*]] = load i32 addrspace(4)*, i32 addrspace(4)** addrspacecast (i32 addrspace(4)* addrspace(4)* @GPtr4 to i32 addrspace(4)**), align 8
-; CHECK-NEXT:    [[A:%.*]] = load i32, i32* @GC1, align 4
 ; CHECK-NEXT:    [[B:%.*]] = load i32, i32* addrspacecast (i32 addrspace(4)* @GC2 to i32*), align 4
 ; CHECK-NEXT:    [[ARGC:%.*]] = addrspacecast i32 addrspace(4)* [[ARG]] to i32*
 ; CHECK-NEXT:    [[C:%.*]] = load i32, i32* [[ARGC]], align 4
 ; CHECK-NEXT:    call void @aligned_barrier()
-; CHECK-NEXT:    [[D:%.*]] = add i32 [[A]], [[B]]
+; CHECK-NEXT:    [[D:%.*]] = add i32 42, [[B]]
 ; CHECK-NEXT:    [[E:%.*]] = add i32 [[D]], [[C]]
 ; CHECK-NEXT:    call void @useI32(i32 [[E]])
 ; CHECK-NEXT:    ret void