+/// TryToOptimizeStoreOfMallocToGlobal - This function is called when we see a
+/// pointer global variable with a single value stored it that is a malloc or
+/// cast of malloc.
+static bool TryToOptimizeStoreOfMallocToGlobal(GlobalVariable *GV,
+ CallInst *CI,
+ BitCastInst *BCI,
+ Module::global_iterator &GVI,
+ TargetData *TD,
+ LLVMContext &Context) {
+ // If we can't figure out the type being malloced, then we can't optimize.
+ const Type *AllocTy = getMallocAllocatedType(CI);
+ assert(AllocTy);
+
+ // If this is a malloc of an abstract type, don't touch it.
+ if (!AllocTy->isSized())
+ return false;
+
+ // We can't optimize this global unless all uses of it are *known* to be
+ // of the malloc value, not of the null initializer value (consider a use
+ // that compares the global's value against zero to see if the malloc has
+ // been reached). To do this, we check to see if all uses of the global
+ // would trap if the global were null: this proves that they must all
+ // happen after the malloc.
+ if (!AllUsesOfLoadedValueWillTrapIfNull(GV))
+ return false;
+
+ // We can't optimize this if the malloc itself is used in a complex way,
+ // for example, being stored into multiple globals. This allows the
+ // malloc to be stored into the specified global, loaded setcc'd, and
+ // GEP'd. These are all things we could transform to using the global
+ // for.
+ {
+ SmallPtrSet<PHINode*, 8> PHIs;
+ if (!ValueIsOnlyUsedLocallyOrStoredToOneGlobal(BCI, GV, PHIs))
+ return false;
+ }
+
+ // If we have a global that is only initialized with a fixed size malloc,
+ // transform the program to use global memory instead of malloc'd memory.
+ // This eliminates dynamic allocation, avoids an indirection accessing the
+ // data, and exposes the resultant global to further GlobalOpt.
+ Value *NElems = getMallocArraySize(CI, Context, TD);
+ // We cannot optimize the malloc if we cannot determine malloc array size.
+ if (NElems) {
+ if (ConstantInt *NElements = dyn_cast<ConstantInt>(NElems))
+ // Restrict this transformation to only working on small allocations
+ // (2048 bytes currently), as we don't want to introduce a 16M global or
+ // something.
+ if (TD &&
+ NElements->getZExtValue() * TD->getTypeAllocSize(AllocTy) < 2048) {
+ GVI = OptimizeGlobalAddressOfMalloc(GV, CI, BCI, Context, TD);
+ return true;
+ }
+
+ // If the allocation is an array of structures, consider transforming this
+ // into multiple malloc'd arrays, one for each field. This is basically
+ // SRoA for malloc'd memory.
+
+ // If this is an allocation of a fixed size array of structs, analyze as a
+ // variable size array. malloc [100 x struct],1 -> malloc struct, 100
+ if (!isArrayMalloc(CI, Context, TD))
+ if (const ArrayType *AT = dyn_cast<ArrayType>(AllocTy))
+ AllocTy = AT->getElementType();
+
+ if (const StructType *AllocSTy = dyn_cast<StructType>(AllocTy)) {
+ // This the structure has an unreasonable number of fields, leave it
+ // alone.
+ if (AllocSTy->getNumElements() <= 16 && AllocSTy->getNumElements() != 0 &&
+ AllGlobalLoadUsesSimpleEnoughForHeapSRA(GV, BCI)) {
+
+ // If this is a fixed size array, transform the Malloc to be an alloc of
+ // structs. malloc [100 x struct],1 -> malloc struct, 100
+ if (const ArrayType *AT =
+ dyn_cast<ArrayType>(getMallocAllocatedType(CI))) {
+ Value* NumElements = ConstantInt::get(Type::getInt32Ty(Context),
+ AT->getNumElements());
+ Value* NewMI = CallInst::CreateMalloc(CI, TD->getIntPtrType(Context),
+ AllocSTy, NumElements,
+ BCI->getName());
+ Value *Cast = new BitCastInst(NewMI, getMallocType(CI), "tmp", CI);
+ BCI->replaceAllUsesWith(Cast);
+ BCI->eraseFromParent();
+ CI->eraseFromParent();
+ BCI = cast<BitCastInst>(NewMI);
+ CI = extractMallocCallFromBitCast(NewMI);
+ }
+
+ GVI = PerformHeapAllocSRoA(GV, CI, BCI, Context, TD);
+ return true;
+ }
+ }
+ }
+
+ return false;
+}