public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* [PATCH] tree-optimization/111891 - fix assert in vectorizable_simd_clone_call
@ 2023-10-20 12:13 Richard Biener
  0 siblings, 0 replies; only message in thread
From: Richard Biener @ 2023-10-20 12:13 UTC (permalink / raw)
  To: gcc-patches

The following fixes the assert in vectorizable_simd_clone_call to
assert we have a vector type during transform.  Whether we have
one during analysis depends on whether another SLP user decided
on the type of a constant/external already.  When we end up with
a mismatch in desire the updating will fail and make vectorization
fail.

Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.

	PR tree-optimization/111891
	* tree-vect-stmts.cc (vectorizable_simd_clone_call): Fix
	assert.

	* gfortran.dg/pr111891.f90: New testcase.
---
 gcc/testsuite/gfortran.dg/pr111891.f90 | 21 +++++++++++++++++++++
 gcc/tree-vect-stmts.cc                 |  8 +++++++-
 2 files changed, 28 insertions(+), 1 deletion(-)
 create mode 100644 gcc/testsuite/gfortran.dg/pr111891.f90

diff --git a/gcc/testsuite/gfortran.dg/pr111891.f90 b/gcc/testsuite/gfortran.dg/pr111891.f90
new file mode 100644
index 00000000000..1167ed60f4c
--- /dev/null
+++ b/gcc/testsuite/gfortran.dg/pr111891.f90
@@ -0,0 +1,21 @@
+! { dg-do compile }
+! { dg-options "-O2" }
+! { dg-additional-options "-mavx" { target avx } }
+
+!GCC$ builtin (powf) attributes simd (notinbranch) if('x86_64')
+
+PARAMETER (NX=3, G=1.4)
+DIMENSION T(NX,NX), P(NX,NX)
+INTEGER Apx
+COMMON P, T
+
+DO i = 1, 3
+  IF  (i < 0.0 ) THEN
+     P(Apx,i) = i**G
+     T(Apx,i) = i**G
+  ELSE
+     P(Apx,i) = 0
+     T(Apx,i) = 0
+  ENDIF
+ENDDO
+END
diff --git a/gcc/tree-vect-stmts.cc b/gcc/tree-vect-stmts.cc
index 8d9028654c9..337b24c51f9 100644
--- a/gcc/tree-vect-stmts.cc
+++ b/gcc/tree-vect-stmts.cc
@@ -4070,7 +4070,13 @@ vectorizable_simd_clone_call (vec_info *vinfo, stmt_vec_info stmt_info,
       if (thisarginfo.dt == vect_constant_def
 	  || thisarginfo.dt == vect_external_def)
 	{
-	  gcc_assert (vec_stmt || thisarginfo.vectype == NULL_TREE);
+	  /* With SLP we determine the vector type of constants/externals
+	     at analysis time, handling conflicts via
+	     vect_maybe_update_slp_op_vectype.  At transform time
+	     we have a vector type recorded for SLP.  */
+	  gcc_assert (!vec_stmt
+		      || !slp_node
+		      || thisarginfo.vectype != NULL_TREE);
 	  if (!vec_stmt)
 	    thisarginfo.vectype = get_vectype_for_scalar_type (vinfo,
 							       TREE_TYPE (op),
-- 
2.35.3

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2023-10-20 12:13 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-10-20 12:13 [PATCH] tree-optimization/111891 - fix assert in vectorizable_simd_clone_call Richard Biener

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).