[flang-commits] [flang] [llvm] [mlir] [flang][OpenMP] Enable tiling (PR #143715)

[Sergio Afonso via flang-commits]

================
@@ -1660,6 +1687,20 @@ genLoopNestClauses(lower::AbstractConverter &converter,
     cp.processCollapse(loc, eval, clauseOps, iv);
 
   clauseOps.loopInclusive = converter.getFirOpBuilder().getUnitAttr();
+
+  fir::FirOpBuilder &firOpBuilder = converter.getFirOpBuilder();
+  for (auto &clause : clauses)
+    if (clause.id == llvm::omp::Clause::OMPC_collapse) {
+      const auto &collapse = std::get<clause::Collapse>(clause.u);
+      int64_t collapseValue = evaluate::ToInt64(collapse.v).value();
+      clauseOps.numCollapse = firOpBuilder.getI64IntegerAttr(collapseValue);
+    }
+
+  llvm::SmallVector<int64_t> sizeValues;
+  auto *ompCons{eval.getIf<parser::OpenMPConstruct>()};
+  collectTileSizesFromOpenMPConstruct(ompCons, sizeValues, semaCtx);
+  if (sizeValues.size() > 0)
+    clauseOps.tileSizes = sizeValues;
----------------
skatrak wrote:

This should be moved to a new `ClauseProcessor::processTile` method and guarded against re-evaluation if it was host-evaluated, the same way it is currently done for `collapse` at the top of this same method. I suppose this won't currently cause an issue as long as tile sizes are constant, but it will break as soon as we introduce variables and we evaluate them in the host.

https://github.com/llvm/llvm-project/pull/143715