[flang][OpenMP] Share DataSharingProcessing instance for simd loops

For `!$omp target ...` constructs, we need to share the DSP instance to prevent privatization for the same variable from happening more than once. This was partially the case already. However, we forgot to do so for `simd` variants of the construct. This commit fixes the issue.

This partially fixes: https://ontrack-internal.amd.com/browse/SWDEV-446525.

Author: ergawy

flang/lib/Lower/OpenMP/OpenMP.cpp

@@ -2032,12 +2032,14 @@ static void createWsloop(Fortran::lower::AbstractConverter &converter,
   }
 }
 
-static void createSimdWsloop(
-    Fortran::lower::AbstractConverter &converter,
-    Fortran::semantics::SemanticsContext &semaCtx,
-    Fortran::lower::pft::Evaluation &eval, llvm::omp::Directive ompDirective,
-    const Fortran::parser::OmpClauseList &beginClauseList,
-    const Fortran::parser::OmpClauseList *endClauseList, mlir::Location loc) {
+static void
+createSimdWsloop(Fortran::lower::AbstractConverter &converter,
+                 Fortran::semantics::SemanticsContext &semaCtx,
+                 Fortran::lower::pft::Evaluation &eval,
+                 llvm::omp::Directive ompDirective,
+                 const Fortran::parser::OmpClauseList &beginClauseList,
+                 const Fortran::parser::OmpClauseList *endClauseList,
+                 mlir::Location loc, DataSharingProcessor &dsp) {
   ClauseProcessor cp(converter, semaCtx, beginClauseList);
   cp.processTODO<clause::Aligned, clause::Allocate, clause::Linear,
                  clause::Safelen, clause::Simdlen, clause::Order>(loc,
@@ -2050,7 +2052,6 @@ static void createSimdWsloop(
   // When support for vectorization is enabled, then we need to add handling of
   // if clause. Currently if clause can be skipped because we always assume
   // SIMD length = 1.
-  DataSharingProcessor dsp(converter, semaCtx, beginClauseList, eval);
   createWsloop(converter, semaCtx, eval, ompDirective, beginClauseList,
                endClauseList, loc, dsp);
 }
@@ -2501,7 +2502,7 @@ static void genOMP(Fortran::lower::AbstractConverter &converter,
   if (llvm::omp::allDoSimdSet.test(ompDirective)) {
     // 2.9.3.2 Workshare SIMD construct
     createSimdWsloop(converter, semaCtx, eval, ompDirective, loopOpClauseList,
-                     endClauseList, currentLocation);
+                     endClauseList, currentLocation, dsp);
 
   } else if (llvm::omp::allSimdSet.test(ompDirective)) {
     // 2.9.3.1 SIMD construct

flang/test/Lower/OpenMP/target_private.f90

@@ -28,3 +28,36 @@ subroutine omp_target_private
 !CHECK-NEXT: }
 
 end subroutine omp_target_private
+
+!CHECK-LABEL: func.func @_QPomp_target_target_do_simd()
+subroutine omp_target_target_do_simd()
+    implicit none
+
+    real(8) :: var
+    integer(8) :: iv
+
+!$omp target teams distribute parallel do simd private(iv,var)
+    do iv=0,10
+        var = 3.14
+    end do
+!$omp end target teams distribute parallel do simd 
+
+!CHECK: omp.target trip_count
+!CHECK:       fir.alloca f64 {bindc_name = "var", pinned
+!CHECK:       omp.teams {
+!CHECK-NEXT:    omp.distribute {
+!CHECK-NEXT:      omp.parallel {
+!CHECK:             fir.alloca i64
+!CHECK:             omp.wsloop
+!CHECK:               omp.yield
+!CHECK-NEXT:        }
+!CHECK-NEXT:        omp.terminator
+!CHECK-NEXT:      }
+!CHECK-NEXT:      omp.terminator
+!CHECK-NEXT:    }
+!CHECK-NEXT:    omp.terminator
+!CHECK-NEXT:  }
+!CHECK-NEXT:  omp.terminator
+!CHECK-NEXT: }
+
+end subroutine omp_target_target_do_simd