[MLIR][Affine] Fix bug and MSAN issue in affine loop utils

[bondhugula]

Summary

Fix bug and MSAN issue in affine loop utils introduced by
d25e022cd19b83c22a6022edb78c4b97a5fc1b49 (D146495). While on it,
fix/clean up issues in immediately surrounding code.

[bondhugula]

Update commit summary.

[springerm]

mlir/lib/Dialect/Affine/Utils/LoopUtils.cpp

2187

should this be =?

[bondhugula]

Fix output arg initialization on the error paths as well.

[bondhugula]

mlir/lib/Dialect/Affine/Utils/LoopUtils.cpp

2187

Thanks for spotting - it should be that!

[bondhugula]

Fix typo spotted by reviewer.

mlir/lib/Dialect/Affine/Utils/LoopUtils.cpp

static LogicalResult generateCopy(
     Block::iterator *nBegin, Block::iterator *nEnd) {
   *nBegin = begin;
   *nEnd = end;
 
   func::FuncOp f = begin->getParentOfType<func::FuncOp>();
   OpBuilder topBuilder(f.getBody());
   Value zeroIndex = topBuilder.create<arith::ConstantIndexOp>(f.getLoc(), 0);
 
+  *sizeInBytes = 0;
+
   if (begin == end)
     return success();
 
   // Is the copy out point at the end of the block where we are doing
   // explicit copying.
   bool isCopyOutAtEndOfBlock = (end == copyOutPlacementStart);
 
   // Copies for read regions are going to be inserted at 'begin'.
std::optional<int64_t> numElements = region.getConstantBoundingSizeAndShape(
       &fastBufferShape, &lbs, &lbDivisors);
   if (!numElements) {
     LLVM_DEBUG(llvm::dbgs() << "Non-constant region size not supported\n");
     return failure();
   }
 
   if (*numElements == 0) {
     LLVM_DEBUG(llvm::dbgs() << "Nothing to copy\n");
-    *sizeInBytes = 0;
     return success();
   }
 
   SmallVector<AffineMap, 4> lbMaps(rank), ubMaps(rank);
   for (unsigned i = 0; i < rank; ++i)
     region.getLowerAndUpperBound(i, lbMaps[i], ubMaps[i]);
 
   const FlatAffineValueConstraints *cst = region.getConstraints();
if (!existingBuf) {
     // operation.
     fastMemRef =
         prologue.create<memref::AllocOp>(loc, fastMemRefType).getResult();
     // Record it.
     fastBufferMap[memref] = fastMemRef;
     // fastMemRefType is a constant shaped memref.
     auto maySizeInBytes = getIntOrFloatMemRefSizeInBytes(fastMemRefType);
     // We don't account for things of unknown size.
-    if (!maySizeInBytes)
-      maySizeInBytes = 0;
+    *sizeInBytes = maySizeInBytes ? *maySizeInBytes : 0;
 
     LLVM_DEBUG(emitRemarkForBlock(*block)
                << "Creating fast buffer of type " << fastMemRefType
                << " and size " << llvm::divideCeil(*sizeInBytes, 1024)
                << " KiB\n");
   } else {
     // Reuse the one already created.
     fastMemRef = fastBufferMap[memref];
-    *sizeInBytes = 0;
   }
 
   auto numElementsSSA = top.create<arith::ConstantIndexOp>(loc, *numElements);
 
   Value dmaStride = nullptr;
   Value numEltPerDmaStride = nullptr;
   if (copyOptions.generateDma) {
     SmallVector<StrideInfo, 4> dmaStrideInfos;
if (!ret) {
     return failure();
   }
 
   // For a range of operations, a note will be emitted at the caller.
   AffineForOp forOp;
   if (llvm::DebugFlag && (forOp = dyn_cast<AffineForOp>(&*begin))) {
     LLVM_DEBUG(forOp.emitRemark()
                << llvm::divideCeil(totalCopyBuffersSizeInBytes, 1024)
-               << " KiB of copy buffers in fast memory space for this block\n");
+               << " KiB of copy buffers in fast memory space for this block");
   }
 
   if (totalCopyBuffersSizeInBytes > copyOptions.fastMemCapacityBytes) {
-    StringRef str = "Total size of all copy buffers' for this block "
-                    "exceeds fast memory capacity\n";
-    block->getParentOp()->emitWarning(str);
+    block->getParentOp()->emitWarning(
+        "total size of all copy buffers' for this block exceeds fast memory "
+        "capacity");
   }
 
   return success();
 }
 
 // A convenience version of affineDataCopyGenerate for all ops in the body of
 // an AffineForOp.
 LogicalResult mlir::affineDataCopyGenerate(AffineForOp forOp,