ScopInfo: Check for possibly nested GEP in fixed-size delin

We currently only consider the first GEP when delinearizing access functions,
which makes us loose information about additional index expression offsets,
which results in our SCoP model to be incorrect. With this patch we now
compare the base pointers used to ensure we do not miss any additional offsets.
This fixes llvm.org/PR27195.

We may consider supporting nested GEP in our delinearization heuristics in
the future.

llvm-svn: 265379

1 files changed, 8 insertions, 0 deletions

diff --git a/polly/lib/Analysis/ScopInfo.cpp b/polly/lib/Analysis/ScopInfo.cpp
index 6dd63ce4d3d..0e17241c0b9 100644
--- a/polly/lib/Analysis/ScopInfo.cpp
+++ b/polly/lib/Analysis/ScopInfo.cpp
@@ -3985,6 +3985,14 @@ bool ScopInfo::buildAccessMultiDimFixed(
   std::tie(Subscripts, Sizes) = getIndexExpressionsFromGEP(GEP, *SE);
   auto *BasePtr = GEP->getOperand(0);
 
+  if (auto *BasePtrCast = dyn_cast<BitCastInst>(BasePtr))
+    BasePtr = BasePtrCast->getOperand(0);
+
+  // Check for identical base pointers to ensure that we do not miss index
+  // offsets that have been added before this GEP is applied.
+  if (BasePtr != BasePointer->getValue())
+    return false;
+
   std::vector<const SCEV *> SizesSCEV;
 
   for (auto *Subscript : Subscripts) {