* [committed] openmp: Avoid ICEs with declare simd; declare simd inbranch [PR93555]
@ 2020-02-05 10:39 Jakub Jelinek
0 siblings, 0 replies; only message in thread
From: Jakub Jelinek @ 2020-02-05 10:39 UTC (permalink / raw)
To: gcc-patches
Hi!
The testcases ICE because when processing the declare simd inbranch,
we don't create the i == 0 clone as it already exists, which means
clone_info->nargs is not adjusted, but we then rely on it being adjusted
when trying other clones.
Bootstrapped/regtested on x86_64-linux and i686-linux, committed to trunk.
2020-02-05 Jakub Jelinek <jakub@redhat.com>
PR middle-end/93555
* omp-simd-clone.c (expand_simd_clones): If simd_clone_mangle or
simd_clone_create failed when i == 0, adjust clone->nargs by
clone->inbranch.
* c-c++-common/gomp/pr93555-1.c: New test.
* c-c++-common/gomp/pr93555-2.c: New test.
* gfortran.dg/gomp/pr93555.f90: New test.
--- gcc/omp-simd-clone.c.jj 2020-01-12 11:54:36.688409260 +0100
+++ gcc/omp-simd-clone.c 2020-02-04 18:08:11.271036096 +0100
@@ -1713,14 +1713,22 @@ expand_simd_clones (struct cgraph_node *
already. */
tree id = simd_clone_mangle (node, clone);
if (id == NULL_TREE)
- continue;
+ {
+ if (i == 0)
+ clone->nargs += clone->inbranch;
+ continue;
+ }
/* Only when we are sure we want to create the clone actually
clone the function (or definitions) or create another
extern FUNCTION_DECL (for prototypes without definitions). */
struct cgraph_node *n = simd_clone_create (node);
if (n == NULL)
- continue;
+ {
+ if (i == 0)
+ clone->nargs += clone->inbranch;
+ continue;
+ }
n->simdclone = clone;
clone->origin = node;
--- gcc/testsuite/c-c++-common/gomp/pr93555-1.c.jj 2020-02-04 22:28:54.557506688 +0100
+++ gcc/testsuite/c-c++-common/gomp/pr93555-1.c 2020-02-04 22:30:18.575246428 +0100
@@ -0,0 +1,18 @@
+/* PR middle-end/93555 */
+/* { dg-do compile } */
+
+#pragma omp declare simd
+#pragma omp declare simd inbranch
+int
+foo (int x)
+{
+ return x;
+}
+
+#pragma omp declare simd inbranch
+#pragma omp declare simd
+int
+bar (int x)
+{
+ return x;
+}
--- gcc/testsuite/c-c++-common/gomp/pr93555-2.c.jj 2020-02-04 22:28:57.581461328 +0100
+++ gcc/testsuite/c-c++-common/gomp/pr93555-2.c 2020-02-04 22:30:25.712139374 +0100
@@ -0,0 +1,16 @@
+/* PR middle-end/93555 */
+/* { dg-do compile } */
+
+#pragma omp declare simd
+#pragma omp declare simd inbranch
+void
+foo (void)
+{
+}
+
+#pragma omp declare simd inbranch
+#pragma omp declare simd
+void
+bar (void)
+{
+}
--- gcc/testsuite/gfortran.dg/gomp/pr93555.f90.jj 2020-02-04 22:31:27.578211386 +0100
+++ gcc/testsuite/gfortran.dg/gomp/pr93555.f90 2020-02-04 22:31:18.267351050 +0100
@@ -0,0 +1,11 @@
+! PR middle-end/93555
+! { dg-do compile }
+
+subroutine foo
+ !$omp declare simd(foo)
+ !$omp declare simd(foo) inbranch
+end
+subroutine bar
+ !$omp declare simd(bar) inbranch
+ !$omp declare simd(bar)
+end
Jakub
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2020-02-05 10:39 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-02-05 10:39 [committed] openmp: Avoid ICEs with declare simd; declare simd inbranch [PR93555] Jakub Jelinek
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).