From 1aa71af09350b9ff4d2fad88a440b682545682ec Mon Sep 17 00:00:00 2001
From: liuhongt <hongtao.liu@intel.com>
Date: Tue, 11 Aug 2020 11:05:40 +0800
Subject: [PATCH] Refine expander
 vec_unpacku_float_hi_v16si/vec_unpacku_float_lo_v16si

gcc/
	PR target/96551
	* config/i386/sse.md (vec_unpacku_float_hi_v16si): For vector
	compare to integer mask, don't use gen_rtx_LT, use
	ix86_expand_mask_vec_cmp instead.
	(vec_unpacku_float_hi_v16si): Ditto.

gcc/testsuite
	* gcc.target/i386/avx512f-pr96551-1.c: New test.
	* gcc.target/i386/avx512f-pr96551-2.c: New test.
---
 gcc/config/i386/sse.md                        |  4 +--
 .../gcc.target/i386/avx512f-pr96551-1.c       | 18 ++++++++++
 .../gcc.target/i386/avx512f-pr96551-2.c       | 33 +++++++++++++++++++
 3 files changed, 53 insertions(+), 2 deletions(-)
 create mode 100644 gcc/testsuite/gcc.target/i386/avx512f-pr96551-1.c
 create mode 100644 gcc/testsuite/gcc.target/i386/avx512f-pr96551-2.c

diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md
index 5e26e9c5aa2..8250325e1a3 100644
--- a/gcc/config/i386/sse.md
+++ b/gcc/config/i386/sse.md
@@ -7026,7 +7026,7 @@
 
   emit_insn (gen_vec_extract_hi_v16si (tmp[3], operands[1]));
   emit_insn (gen_floatv8siv8df2 (tmp[2], tmp[3]));
-  emit_insn (gen_rtx_SET (k, gen_rtx_LT (QImode, tmp[2], tmp[0])));
+  ix86_expand_mask_vec_cmp (k, LT, tmp[2], tmp[0]);
   emit_insn (gen_addv8df3_mask (tmp[2], tmp[2], tmp[1], tmp[2], k));
   emit_move_insn (operands[0], tmp[2]);
   DONE;
@@ -7073,7 +7073,7 @@
   k = gen_reg_rtx (QImode);
 
   emit_insn (gen_avx512f_cvtdq2pd512_2 (tmp[2], operands[1]));
-  emit_insn (gen_rtx_SET (k, gen_rtx_LT (QImode, tmp[2], tmp[0])));
+  ix86_expand_mask_vec_cmp (k, LT, tmp[2], tmp[0]);
   emit_insn (gen_addv8df3_mask (tmp[2], tmp[2], tmp[1], tmp[2], k));
   emit_move_insn (operands[0], tmp[2]);
   DONE;
diff --git a/gcc/testsuite/gcc.target/i386/avx512f-pr96551-1.c b/gcc/testsuite/gcc.target/i386/avx512f-pr96551-1.c
new file mode 100644
index 00000000000..598bb6e85f3
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/avx512f-pr96551-1.c
@@ -0,0 +1,18 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -mavx512f -mprefer-vector-width=512" } */
+
+unsigned int a[256];
+double b[256];
+
+void
+__attribute__ ((noipa, optimize ("tree-vectorize")))
+foo(void)
+{
+  int i;
+
+  for (i=0; i<256; ++i)
+    b[i] = a[i];
+}
+
+/* { dg-final { scan-assembler "vcvtdq2pd\[^\n\]*zmm" } } */
+
diff --git a/gcc/testsuite/gcc.target/i386/avx512f-pr96551-2.c b/gcc/testsuite/gcc.target/i386/avx512f-pr96551-2.c
new file mode 100644
index 00000000000..f8bb1da7966
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/avx512f-pr96551-2.c
@@ -0,0 +1,33 @@
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f -mprefer-vector-width=512" } */
+/* { dg-require-effective-target avx512f } */
+
+#ifndef CHECK
+#define CHECK "avx512f-helper.h"
+#endif
+
+#include CHECK
+
+#ifndef TEST
+#define TEST test_512
+#endif
+
+#include "avx512f-pr96551-1.c"
+
+void
+test_512 (void)
+{
+  double exp[256];
+  for (int i = 0; i != 256; i++)
+    {
+      a[i] = i * i + 3 * i + 13;
+      exp[i] = a[i];
+      b[i] = 0;
+    }
+
+  foo ();
+
+  for (int i = 0; i != 256; i++)
+    if (exp[i] != b[i])
+      abort ();
+}
-- 
2.30.2