[SelectionDAG] Don't promote the alignment of allocas beyond the stack alignment.

allocas in LLVM IR have a specified alignment. When that alignment is
specified, the alloca has at least that alignment at runtime.

If the specified type of the alloca has a higher preferred alignment,
SelectionDAG currently ignores that specified alignment, and increases
the alignment. It does this even if it would trigger stack realignment.
I don't think this makes sense, so this patch changes that.

I was looking into this for SVE in particular: for SVE, overaligning
vscale'ed types is extra expensive because it requires realigning the
stack multiple times, or using dynamic allocation. (This currently isn't
implemented.)

I updated the expected assembly for a couple tests; in particular, for
arg-copy-elide.ll, the optimization in question does not increase the
alignment the way SelectionDAG normally would. For the rest, I just
increased the specified alignment on the allocas to match what
SelectionDAG was inferring.

Differential Revision: https://reviews.llvm.org/D79532

1 files changed, 13 insertions, 1 deletions

diff --git a/llvm/lib/CodeGen/SelectionDAG/FunctionLoweringInfo.cpp b/llvm/lib/CodeGen/SelectionDAG/FunctionLoweringInfo.cpp
index 3f302d5..c482125 100644
--- a/llvm/lib/CodeGen/SelectionDAG/FunctionLoweringInfo.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/FunctionLoweringInfo.cpp
@@ -134,8 +134,20 @@ void FunctionLoweringInfo::set(const Function &fn, MachineFunction &mf,
     for (const Instruction &I : BB) {
       if (const AllocaInst *AI = dyn_cast<AllocaInst>(&I)) {
         Type *Ty = AI->getAllocatedType();
+        Align TyPrefAlign = MF->getDataLayout().getPrefTypeAlign(Ty);
+        // The "specified" alignment is the alignment written on the alloca,
+        // or the preferred alignment of the type if none is specified.
+        //
+        // (Unspecified alignment on allocas will be going away soon.)
+        Align SpecifiedAlign = AI->getAlign() ? *AI->getAlign() : TyPrefAlign;
+
+        // If the preferred alignment of the type is higher than the specified
+        // alignment of the alloca, promote the alignment, as long as it doesn't
+        // require realigning the stack.
+        //
+        // FIXME: Do we really want to second-guess the IR in isel?
         Align Alignment =
-            max(MF->getDataLayout().getPrefTypeAlign(Ty), AI->getAlign());
+            std::max(std::min(TyPrefAlign, StackAlign), SpecifiedAlign);
 
         // Static allocas can be folded into the initial stack frame
         // adjustment. For targets that don't realign the stack, don't