From: liuhongt <hongtao.liu@intel.com>
Date: Fri, 14 Nov 2025 03:59:08 +0000 (-0800)
Subject: Also handle vptestnm + and15/and3 to just vptestnm.
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=594dc80c8e49cb65f77a13c201a0bea9423329ec;p=thirdparty%2Fgcc.git

Also handle vptestnm + and15/and3 to just vptestnm.

r16-1298-gcdfa5fe03512f7 optimizes vpcmp + and15/and3 to vpcmp when VF
is 2 or 4. vptestnm is a variant of vpcmpeq which accepts
nonimm_or_0_operand. The patch handles that.

gcc/ChangeLog:

	PR target/103750
	* config/i386/sse.md (*<avx512>_eq<mode>3_and15): New
	define_insn.
	(*avx512vl_eqv2di_and3): Ditto.
	* config/i386/i386.md (*ior<mode>_ccz_1): Fix the typo in the
	comments above.

gcc/testsuite/ChangeLog:

	* gcc.target/i386/avx512vl-pr103750-2.c: New test.
---

diff --git a/gcc/config/i386/i386.md b/gcc/config/i386/i386.md
index b925a037b2d..6af7dcfcdd3 100644
--- a/gcc/config/i386/i386.md
+++ b/gcc/config/i386/i386.md
@@ -14203,7 +14203,7 @@
    (set_attr "isa" "*,apx_ndd")
    (set_attr "mode" "SI")])
 
-;; It must be put before *<code><mode>_3, the blow one.
+;; It must be put before *<code><mode>_3, the one below.
 (define_insn "*ior<mode>_ccz_1"
   [(set (reg:CCZ FLAGS_REG)
 	(compare:CCZ
diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md
index 7d91585b05d..8b90845260a 100644
--- a/gcc/config/i386/sse.md
+++ b/gcc/config/i386/sse.md
@@ -4653,6 +4653,9 @@
 	   UNSPEC_PCMP))]
   "operands[4] = GEN_INT (INTVAL (operands[3]) ^ 4);")
 
+(define_int_iterator UNSPEC_PCMP_ITER
+  [UNSPEC_PCMP UNSPEC_UNSIGNED_PCMP])
+
 (define_insn "*<avx512>_cmp<mode>3_and15"
   [(set (match_operand:QI 0 "register_operand" "=k")
 	(and:QI
@@ -4685,6 +4688,23 @@
    (set_attr "prefix" "evex")
    (set_attr "mode" "<sseinsnmode>")])
 
+(define_insn "*<avx512>_eq<mode>3_and15"
+  [(set (match_operand:QI 0 "register_operand" "=k, k")
+	(and:QI
+	  (unspec:QI
+	    [(match_operand:VI48_AVX512VL_4 1 "nonimm_or_0_operand" "%v, v")
+	     (match_operand:VI48_AVX512VL_4 2 "nonimm_or_0_operand" "vm, C")
+	     (const_int 0)]
+	    UNSPEC_PCMP_ITER)
+	  (const_int 15)))]
+  "TARGET_AVX512F && !(MEM_P (operands[1]) && MEM_P (operands[2]))"
+  "@
+   vpcmpeq<ssemodesuffix>\t{%2, %1, %0|%0, %1, %2}
+   vptestnm<ssemodesuffix>\t{%1, %1, %0|%0, %1, %1}"
+  [(set_attr "type" "ssecmp")
+   (set_attr "prefix" "evex")
+   (set_attr "mode" "<sseinsnmode>")])
+
 (define_insn "*<avx512>_cmp<mode>3_and3"
   [(set (match_operand:QI 0 "register_operand" "=k")
 	(and:QI
@@ -4717,6 +4737,23 @@
    (set_attr "prefix" "evex")
    (set_attr "mode" "TI")])
 
+(define_insn "*avx512vl_eqv2di_and3"
+  [(set (match_operand:QI 0 "register_operand" "=k, k")
+	(and:QI
+	  (unspec:QI
+	    [(match_operand:V2DI 1 "nonimm_or_0_operand" "%v, v")
+	     (match_operand:V2DI 2 "nonimm_or_0_operand" "vm, C")
+	     (const_int 0)]
+	    UNSPEC_PCMP_ITER)
+	  (const_int 3)))]
+  "TARGET_AVX512F && !(MEM_P (operands[1]) && MEM_P (operands[2]))"
+  "@
+   vpcmpeqq\t{%2, %1, %0|%0, %1, %2}
+   vptestnmq\t{%1, %1, %0|%0, %1, %1}"
+  [(set_attr "type" "ssecmp")
+   (set_attr "prefix" "evex")
+   (set_attr "mode" "TI")])
+
 (define_insn "<avx512>_cmp<mode>3<mask_scalar_merge_name>"
   [(set (match_operand:<avx512fmaskmode> 0 "register_operand" "=k")
 	(unspec:<avx512fmaskmode>
@@ -4790,9 +4827,6 @@
    (set_attr "prefix" "evex")
    (set_attr "mode" "<VI12_AVX512VL:sseinsnmode>")])
 
-(define_int_iterator UNSPEC_PCMP_ITER
-  [UNSPEC_PCMP UNSPEC_UNSIGNED_PCMP])
-
 (define_insn_and_split "*<avx512>_cmp<mode>3"
   [(set (match_operand:<avx512fmaskmode> 0 "register_operand")
 	(not:<avx512fmaskmode>
diff --git a/gcc/testsuite/gcc.target/i386/avx512vl-pr103750-2.c b/gcc/testsuite/gcc.target/i386/avx512vl-pr103750-2.c
new file mode 100644
index 00000000000..7c6e77b79e3
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/avx512vl-pr103750-2.c
@@ -0,0 +1,13 @@
+/* { dg-do compile } */
+/* { dg-options "-march=x86-64-v4 -mprefer-vector-width=128 -O3" } */
+/* { dg-final { scan-assembler "kortest" } } */
+/* { dg-final { scan-assembler-not "kmov" } } */
+
+int
+foo (int *__restrict a)
+{
+  for (int i = 0; i != 100; i++)
+    if (a[i] == 0)
+      return 1;
+  return 0;
+}