[llvm] [WebAssembly] Fold any/alltrue (setcc x, 0, eq/ne) to [not] any/alltrue x (PR #144741)

[Luke Lau via llvm-commits]

================
@@ -0,0 +1,136 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc < %s -verify-machineinstrs -disable-wasm-fallthrough-return-opt -wasm-disable-explicit-locals -wasm-keep-registers -mattr=+simd128 | FileCheck %s
+
+target triple = "wasm64"
+
+define i32 @all_true_16_i8(<16 x i8> %v) {
+; CHECK-LABEL: all_true_16_i8:
+; CHECK:         .functype all_true_16_i8 (v128) -> (i32)
+; CHECK-NEXT:  # %bb.0:
+; CHECK-NEXT:    i8x16.all_true $push0=, $0
+; CHECK-NEXT:    return $pop0
+  %1 = icmp eq <16 x i8> %v, zeroinitializer
+  %2 = bitcast <16 x i1> %1 to i16
+  %3 = icmp eq i16 %2, 0
+  %conv3 = zext i1 %3 to i32
+  ret i32 %conv3
+}
+
+
+define i32 @all_true_4_i32(<4 x i32> %v) {
+; CHECK-LABEL: all_true_4_i32:
+; CHECK:         .functype all_true_4_i32 (v128) -> (i32)
+; CHECK-NEXT:  # %bb.0:
+; CHECK-NEXT:    i32x4.all_true $push0=, $0
+; CHECK-NEXT:    return $pop0
+  %1 = icmp eq <4 x i32> %v, zeroinitializer
+  %2 = bitcast <4 x i1> %1 to i4
+  %3 = icmp eq i4 %2, 0
+  %conv3 = zext i1 %3 to i32
+  ret i32 %conv3
+}
+
+
+define i32 @all_true_8_i16(<8 x i16> %v) {
+; CHECK-LABEL: all_true_8_i16:
+; CHECK:         .functype all_true_8_i16 (v128) -> (i32)
+; CHECK-NEXT:  # %bb.0:
+; CHECK-NEXT:    i16x8.all_true $push0=, $0
+; CHECK-NEXT:    return $pop0
+  %1 = icmp eq <8 x i16> %v, zeroinitializer
+  %2 = bitcast <8 x i1> %1 to i8
+  %3 = icmp eq i8 %2, 0
+  %conv3 = zext i1 %3 to i32
+  ret i32 %conv3
+}
+
+
+define i32 @all_true_2_i64(<2 x i64> %v) {
+; CHECK-LABEL: all_true_2_i64:
+; CHECK:         .functype all_true_2_i64 (v128) -> (i32)
+; CHECK-NEXT:  # %bb.0:
+; CHECK-NEXT:    i64x2.all_true $push0=, $0
+; CHECK-NEXT:    return $pop0
+  %1 = icmp eq <2 x i64> %v, zeroinitializer
+  %2 = bitcast <2 x i1> %1 to i2
+  %3 = icmp eq i2 %2, 0
+  %conv3 = zext i1 %3 to i32
+  ret i32 %conv3
+}
+
+
+define i32 @all_true_4_i64(<4 x i64> %v) {
+; CHECK-LABEL: all_true_4_i64:
+; CHECK:         .functype all_true_4_i64 (v128, v128) -> (i32)
+; CHECK-NEXT:  # %bb.0:
+; CHECK-NEXT:    v128.const $push9=, 0, 0
+; CHECK-NEXT:    local.tee $push8=, $2=, $pop9
+; CHECK-NEXT:    i64x2.eq $push1=, $0, $pop8
+; CHECK-NEXT:    i64x2.eq $push0=, $1, $2
+; CHECK-NEXT:    i8x16.shuffle $push2=, $pop1, $pop0, 0, 1, 2, 3, 8, 9, 10, 11, 16, 17, 18, 19, 24, 25, 26, 27
+; CHECK-NEXT:    v128.any_true $push3=, $pop2
+; CHECK-NEXT:    i32.const $push4=, -1
+; CHECK-NEXT:    i32.xor $push5=, $pop3, $pop4
+; CHECK-NEXT:    i32.const $push6=, 1
+; CHECK-NEXT:    i32.and $push7=, $pop5, $pop6
+; CHECK-NEXT:    return $pop7
+  %1 = icmp eq <4 x i64> %v, zeroinitializer
+  %2 = bitcast <4 x i1> %1 to i4
+  %3 = icmp eq i4 %2, 0
+  %conv3 = zext i1 %3 to i32
+  ret i32 %conv3
+}
+
+
+; setcc (iN (bitcast (set_cc (vNi1 X), 0, ne)), 0, ne
+;   => any_true (set_cc (X), 0, ne)
+;   => any_true (X)
+define i32 @any_true_1_4_i32(<4 x i32> %v) {
+; CHECK-LABEL: any_true_1_4_i32:
+; CHECK:         .functype any_true_1_4_i32 (v128) -> (i32)
+; CHECK-NEXT:  # %bb.0:
+; CHECK-NEXT:    v128.any_true $push0=, $0
+; CHECK-NEXT:    return $pop0
+  %1 = icmp ne <4 x i32> %v, zeroinitializer
+  %2 = bitcast <4 x i1> %1 to i4
+  %3 = icmp ne i4 %2, 0
+  %conv3 = zext i1 %3 to i32
+  ret i32 %conv3
+}
+
+; setcc (iN (bitcast (set_cc (vNi1 X), 0, eq)), -1, ne
+;   => not all_true (set_cc (X), 0, eq)
+;   => not all_true (set_cc (X), 0, eq)
----------------
lukel97 wrote:

Looks like a duplicate comment line?

```suggestion
```

https://github.com/llvm/llvm-project/pull/144741