Hi!
When offloading to nvptx is enabled, scan_omp_simd duplicates the simd
region including its clauses and body using inliner's
copy_gimple_seq_and_replace_locals. That works nicely for decls, remaps
only those that are seen in the nested bind expr vars (i.e. local variables)
and doesn't remap other vars. But for SSA_NAMEs it remaps them always, doesn't
know if their def stmt is outside of the simd (then it better shouldn't be
remapped)
or inside of it (then it should) and without cfg/dominators that is pretty hard
to figure out (well, we could walk the region twice, once note SSA_NAMEs defined
by each stmt seen there and once do the remapping of only those visited
SSA_NAMEs).
This patch uses a simpler way, disables temporarily into_ssa for the clauses and
body of each simd region; we already disable into_ssa e.g. in
parallel/target/task
etc. regions through push_gimplify_context () but for simd we don't push
any gimplification context and appart from into_ssa I think we don't need it.
Bootstrapped/regtested on x86_64-linux and i686-linux, committed to trunk.
2022-03-03 Jakub Jelinek <ja...@redhat.com>
PR middle-end/104757
* gimplify.cc (gimplify_omp_loop): Call gimplify_expr rather than
gimplify_omp_for.
(gimplify_expr) <case OMP_SIMD>: Temporarily disable
gimplify_ctxp->into_ssa around call to gimplify_omp_for.
* gfortran.dg/gomp/pr104757.f90: New test.
* gcc.dg/gomp/pr104757.c: New test.
--- gcc/gimplify.cc.jj 2022-02-14 13:14:39.058128337 +0100
+++ gcc/gimplify.cc 2022-03-02 19:31:46.675210108 +0100
@@ -13725,7 +13725,7 @@ gimplify_omp_loop (tree *expr_p, gimple_
*pc = NULL_TREE;
*expr_p = t;
}
- return gimplify_omp_for (expr_p, pre_p);
+ return gimplify_expr (expr_p, pre_p, NULL, is_gimple_stmt, fb_none);
}
@@ -15479,8 +15479,19 @@ gimplify_expr (tree *expr_p, gimple_seq
ret = GS_ALL_DONE;
break;
- case OMP_FOR:
case OMP_SIMD:
+ {
+ /* Temporarily disable into_ssa, as scan_omp_simd
+ which calls copy_gimple_seq_and_replace_locals can't deal
+ with SSA_NAMEs defined outside of the body properly. */
+ bool saved_into_ssa = gimplify_ctxp->into_ssa;
+ gimplify_ctxp->into_ssa = false;
+ ret = gimplify_omp_for (expr_p, pre_p);
+ gimplify_ctxp->into_ssa = saved_into_ssa;
+ break;
+ }
+
+ case OMP_FOR:
case OMP_DISTRIBUTE:
case OMP_TASKLOOP:
case OACC_LOOP:
--- gcc/testsuite/gfortran.dg/gomp/pr104757.f90.jj 2022-03-02
18:35:00.606549074 +0100
+++ gcc/testsuite/gfortran.dg/gomp/pr104757.f90 2022-03-02 18:34:42.028804663
+0100
@@ -0,0 +1,19 @@
+! PR middle-end/104757
+! { dg-do compile }
+! { dg-options "-O -fopenmp" }
+
+module pr104757
+ implicit none (external, type)
+ integer :: ll
+ !$omp declare target (ll)
+contains
+ subroutine foo (i1)
+ !$omp declare target (foo)
+ logical :: i1
+ integer :: i
+ !$omp distribute simd if(i1)
+ do i = 1, 64
+ ll = ll + 1
+ end do
+ end
+end module
--- gcc/testsuite/gcc.dg/gomp/pr104757.c.jj 2022-03-02 19:33:52.637485113
+0100
+++ gcc/testsuite/gcc.dg/gomp/pr104757.c 2022-03-02 19:34:09.365256034
+0100
@@ -0,0 +1,14 @@
+/* PR middle-end/104757 */
+/* { dg-do compile } */
+/* { dg-options "-O2 -fopenmp" } */
+
+#pragma omp declare target
+void
+foo (int x, int y, int *z)
+{
+ int j = 0;
+ #pragma omp simd linear(j:x + y)
+ for (int i = 0; i < 64; i++)
+ j += x + y;
+}
+#pragma omp end declare target
Jakub
