From: Jakub Jelinek <jakub@redhat.com>
Date: Fri, 30 Jun 2017 14:52:24 +0000 (+0200)
Subject: re PR target/81225 (ICE with -mavx512ifma -O3 -ffloat-store)
X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=e855bdc04db3b2ff010d6d490dc9a4bee5f08b3b;p=gcc.git

re PR target/81225 (ICE with -mavx512ifma -O3 -ffloat-store)

	PR target/81225
	* config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For
	V8FI, V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
	of nonimmediate_operand and <store_mask_constraint> instead of m for
	the input operand.  For V8FI iterator, always split if input is a MEM.
	For V16FI and V8SF_256 iterators, don't test if both operands are MEM
	if <mask_applied>.  For VI4F_256 iterator, use <store_mask_predicate>
	instead of register_operand and <store_mask_constraint> instead of v for
        the input operand.  Make sure both operands aren't MEMs for if not
	<mask_applied>.

	* gcc.target/i386/pr81225.c: New test.

From-SVN: r249844
---

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 84a93992292..be0b4bc98de 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,16 @@
+2017-06-30  Jakub Jelinek  <jakub@redhat.com>
+
+	PR target/81225
+	* config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For
+	V8FI, V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
+	of nonimmediate_operand and <store_mask_constraint> instead of m for
+	the input operand.  For V8FI iterator, always split if input is a MEM.
+	For V16FI and V8SF_256 iterators, don't test if both operands are MEM
+	if <mask_applied>.  For VI4F_256 iterator, use <store_mask_predicate>
+	instead of register_operand and <store_mask_constraint> instead of v for
+        the input operand.  Make sure both operands aren't MEMs for if not
+	<mask_applied>.
+
 2017-06-30  Sylvestre Ledru  <sylvestre@debian.org>
 
 	* lto-wrapper.c (copy_file) Close both file descriptors before
diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md
index f61ae2bb927..5cf6c928bfb 100644
--- a/gcc/config/i386/sse.md
+++ b/gcc/config/i386/sse.md
@@ -7359,13 +7359,13 @@
 (define_insn "vec_extract_lo_<mode><mask_name>"
   [(set (match_operand:<ssehalfvecmode> 0 "<store_mask_predicate>" "=<store_mask_constraint>,v")
 	(vec_select:<ssehalfvecmode>
-	  (match_operand:V8FI 1 "nonimmediate_operand" "v,m")
+	  (match_operand:V8FI 1 "<store_mask_predicate>" "v,<store_mask_constraint>")
 	  (parallel [(const_int 0) (const_int 1)
             (const_int 2) (const_int 3)])))]
   "TARGET_AVX512F
    && (<mask_applied> || !(MEM_P (operands[0]) && MEM_P (operands[1])))"
 {
-  if (<mask_applied> || !TARGET_AVX512VL)
+  if (<mask_applied> || (!TARGET_AVX512VL && !MEM_P (operands[1])))
     return "vextract<shuffletype>64x4\t{$0x0, %1, %0<mask_operand2>|%0<mask_operand2>, %1, 0x0}";
   else
     return "#";
@@ -7515,14 +7515,15 @@
 (define_insn "vec_extract_lo_<mode><mask_name>"
   [(set (match_operand:<ssehalfvecmode> 0 "nonimmediate_operand" "=v,m")
 	(vec_select:<ssehalfvecmode>
-	  (match_operand:V16FI 1 "nonimmediate_operand" "vm,v")
+	  (match_operand:V16FI 1 "<store_mask_predicate>"
+				 "<store_mask_constraint>,v")
 	  (parallel [(const_int 0) (const_int 1)
                      (const_int 2) (const_int 3)
                      (const_int 4) (const_int 5)
                      (const_int 6) (const_int 7)])))]
   "TARGET_AVX512F
    && <mask_mode512bit_condition>
-   && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
+   && (<mask_applied> || !(MEM_P (operands[0]) && MEM_P (operands[1])))"
 {
   if (<mask_applied>)
     return "vextract<shuffletype>32x8\t{$0x0, %1, %0<mask_operand2>|%0<mask_operand2>, %1, 0x0}";
@@ -7546,11 +7547,12 @@
 (define_insn "vec_extract_lo_<mode><mask_name>"
   [(set (match_operand:<ssehalfvecmode> 0 "<store_mask_predicate>" "=v,m")
 	(vec_select:<ssehalfvecmode>
-	  (match_operand:VI8F_256 1 "nonimmediate_operand" "vm,v")
+	  (match_operand:VI8F_256 1 "<store_mask_predicate>"
+				    "<store_mask_constraint>,v")
 	  (parallel [(const_int 0) (const_int 1)])))]
   "TARGET_AVX
    && <mask_avx512vl_condition> && <mask_avx512dq_condition>
-   && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
+   && (<mask_applied> || !(MEM_P (operands[0]) && MEM_P (operands[1])))"
 {
   if (<mask_applied>)
     return "vextract<shuffletype>64x2\t{$0x0, %1, %0%{%3%}|%0%{%3%}, %1, 0x0}";
@@ -7610,12 +7612,16 @@
   "operands[1] = gen_lowpart (<ssehalfvecmode>mode, operands[1]);")
 
 (define_insn "vec_extract_lo_<mode><mask_name>"
-  [(set (match_operand:<ssehalfvecmode> 0 "<store_mask_predicate>" "=<store_mask_constraint>")
+  [(set (match_operand:<ssehalfvecmode> 0 "<store_mask_predicate>"
+					  "=<store_mask_constraint>,v")
 	(vec_select:<ssehalfvecmode>
-	  (match_operand:VI4F_256 1 "register_operand" "v")
+	  (match_operand:VI4F_256 1 "<store_mask_predicate>"
+				    "v,<store_mask_constraint>")
 	  (parallel [(const_int 0) (const_int 1)
 		     (const_int 2) (const_int 3)])))]
-  "TARGET_AVX && <mask_avx512vl_condition> && <mask_avx512dq_condition>"
+  "TARGET_AVX
+   && <mask_avx512vl_condition> && <mask_avx512dq_condition>
+   && (<mask_applied> || !(MEM_P (operands[0]) && MEM_P (operands[1])))"
 {
   if (<mask_applied>)
     return "vextract<shuffletype>32x4\t{$0x0, %1, %0<mask_operand2>|%0<mask_operand2>, %1, 0x0}";
diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog
index 2637ecc6262..f42ea73f704 100644
--- a/gcc/testsuite/ChangeLog
+++ b/gcc/testsuite/ChangeLog
@@ -1,3 +1,8 @@
+2017-06-30  Jakub Jelinek  <jakub@redhat.com>
+
+	PR target/81225
+	* gcc.target/i386/pr81225.c: New test.
+
 2017-06-30  Nathan Sidwell  <nathan@acm.org>
 
 	* g++.dg/concepts/memfun-err.C: Adjust diagnostics.
diff --git a/gcc/testsuite/gcc.target/i386/pr81225.c b/gcc/testsuite/gcc.target/i386/pr81225.c
new file mode 100644
index 00000000000..db95e941c14
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/pr81225.c
@@ -0,0 +1,14 @@
+/* PR target/81225 */
+/* { dg-do compile } */
+/* { dg-options "-mavx512ifma -O3 -ffloat-store" } */
+
+long a[24];
+float b[4], c[24];
+int d;
+
+void
+foo ()
+{
+  for (d = 0; d < 24; d++)
+    c[d] = (float) d ? : b[a[d]];
+}