[PATCH PR96195] aarch64: ICE during GIMPLE pass:vect

["yangyang (ET)" <yangyang305@huawei.com>]

[-- Attachment #1: Type: text/plain, Size: 1223 bytes --]

Hi, 

	This is a simple fix for PR96195.

	For the test case, GCC generates the following gimple statement in pass_vect:

	  vect__21.16_58 = zp.simdclone.2 (vect_vec_iv_.15_56);

	The mode of vect__21.16_58 is VNx2SI while the mode of zp.simdclone.2 (vect_vec_iv_.15_56) is V4SI, resulting in the crash.

	In vectorizable_simd_clone_call, type compatibility is handled based on the number of elements and the type compatibility of elements, which is not enough. 
	This patch add VIEW_CONVERT_EXPRs if the arguments types and return type of simd clone function are distinct with the vectype of stmt.

	Added one testcase for this. Bootstrap and tested on both aarch64 and x86 Linux platform, no new regression witnessed.

	Ok for trunk?

Thanks,
Yang Yang


+2020-07-15  Yang Yang  <yangyang305@huawei.com>
+
+       PR tree-optimization/96195
+       * tree-vect-stmts.c (vectorizable_simd_clone_call): Add
+       VIEW_CONVERT_EXPRs if the arguments types and return type
+       of simd clone function are distinct with the vectype of stmt.
+

+2020-07-15  Yang Yang  <yangyang305@huawei.com>
+
+       PR tree-optimization/96195
+       * gcc.target/aarch64/sve/pr96195.c: New test.
+

[-- Attachment #2: pr96195-v1.patch --]
[-- Type: application/octet-stream, Size: 3503 bytes --]

From 4dd96316aacffd11c3a0da8fce87d8cc295e6366 Mon Sep 17 00:00:00 2001
From: y00520163 <yangyang305@huawei.com>
Date: Wed, 15 Jul 2020 19:53:52 +0800
Subject: [PATCH] vect: Fix an ICE in vectorizable_simd_clone_call

In vectorizable_simd_clone_call, type compatibility is handled based on
the number of elements and the type compatibility of elements, which is
not enough. This patch add VIEW_CONVERT_EXPRs if the arguments types
and return type of simd clone function are distinct with the vectype of
stmt.

2020-07-15  Yang Yang <yangyang305@huawei.com>

gcc/ChangeLog:

	* tree-vect-stmts.c (vectorizable_simd_clone_call): Add
	VIEW_CONVERT_EXPRs if the arguments types and return type
	of simd clone function are distinct with the vectype of stmt.

gcc/testsuite/ChangeLog:

	* gcc.target/aarch64/sve/pr96195.c: New test.
---
 .../gcc.target/aarch64/sve/pr96195.c          | 17 +++++++++++++
 gcc/tree-vect-stmts.c                         | 25 ++++++++++++++++---
 2 files changed, 39 insertions(+), 3 deletions(-)
 create mode 100644 gcc/testsuite/gcc.target/aarch64/sve/pr96195.c

diff --git a/gcc/testsuite/gcc.target/aarch64/sve/pr96195.c b/gcc/testsuite/gcc.target/aarch64/sve/pr96195.c
new file mode 100644
index 00000000000..d879efda5c8
--- /dev/null
+++ b/gcc/testsuite/gcc.target/aarch64/sve/pr96195.c
@@ -0,0 +1,17 @@
+/* { dg-do compile } */
+/* { dg-options "-O1 -fopenmp-simd -ftree-vectorize -msve-vector-bits=128" } */
+
+int by;
+
+#pragma omp declare simd
+int
+zp (int);
+
+void
+qh (int oh)
+{
+#pragma omp simd
+  for (by = 0; by < oh; ++by)
+    by = zp (by);
+}
+
diff --git a/gcc/tree-vect-stmts.c b/gcc/tree-vect-stmts.c
index 6730cae8085..8301494fa41 100644
--- a/gcc/tree-vect-stmts.c
+++ b/gcc/tree-vect-stmts.c
@@ -4108,7 +4108,20 @@ vectorizable_simd_clone_call (vec_info *vinfo, stmt_vec_info stmt_info,
 						  vec_oprnd0);
 			}
 		      if (k == 1)
-			vargs.safe_push (vec_oprnd0);
+			if (!useless_type_conversion_p (TREE_TYPE (vec_oprnd0),
+						       atype))
+			  {
+			    vec_oprnd0
+			      = build1 (VIEW_CONVERT_EXPR, atype, vec_oprnd0);
+			    gassign *new_stmt
+			      = gimple_build_assign (make_ssa_name (atype),
+						       vec_oprnd0);
+			    vect_finish_stmt_generation (vinfo, stmt_info,
+							 new_stmt, gsi);
+			    vargs.safe_push (gimple_assign_lhs (new_stmt));
+			  }
+			else
+			  vargs.safe_push (vec_oprnd0);
 		      else
 			{
 			  vec_oprnd0 = build_constructor (atype, ctor_elts);
@@ -4204,8 +4217,7 @@ vectorizable_simd_clone_call (vec_info *vinfo, stmt_vec_info stmt_info,
 	  gcc_assert (ratype || simd_clone_subparts (rtype) == nunits);
 	  if (ratype)
 	    new_temp = create_tmp_var (ratype);
-	  else if (simd_clone_subparts (vectype)
-		   == simd_clone_subparts (rtype))
+	  else if (useless_type_conversion_p (vectype, rtype))
 	    new_temp = make_ssa_name (vec_dest, new_call);
 	  else
 	    new_temp = make_ssa_name (rtype, new_call);
@@ -4293,6 +4305,13 @@ vectorizable_simd_clone_call (vec_info *vinfo, stmt_vec_info stmt_info,
 	      vect_finish_stmt_generation (vinfo, stmt_info, new_stmt, gsi);
 	      vect_clobber_variable (vinfo, stmt_info, gsi, new_temp);
 	    }
+	  else if (!useless_type_conversion_p (vectype, rtype))
+	    {
+	      vec_oprnd0 = build1 (VIEW_CONVERT_EXPR, vectype, new_temp);
+	      new_stmt
+		= gimple_build_assign (make_ssa_name (vec_dest), vec_oprnd0);
+	      vect_finish_stmt_generation (vinfo, stmt_info, new_stmt, gsi);
+	    }
 	}
 
       if (j == 0)
-- 
2.19.1