Skip to content

A lightweight WebAssembly runtime that is fast, secure, and standards-compliant

winch(aarch64): v128 unary ops (#13938)

Daniel Macoveiaa1da38b8d8d
diff --git a/tests/disas/winch/aarch64/v128_ops/f32x4_abs.wat b/tests/disas/winch/aarch64/v128_ops/f32x4_abs.wat
new file mode 100644
index 0000000000..77f550da20
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f32x4_abs.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f32x4.abs (v128.const f32x4 -1.5 -0 nan -inf)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; fabs v0.4s, v0.4s
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: .byte 0x00, 0x00, 0xc0, 0xbf
+;; 84: .byte 0x00, 0x00, 0x00, 0x80
+;; 88: .byte 0x00, 0x00, 0xc0, 0x7f
+;; 8c: .byte 0x00, 0x00, 0x80, 0xff
diff --git a/tests/disas/winch/aarch64/v128_ops/f32x4_ceil.wat b/tests/disas/winch/aarch64/v128_ops/f32x4_ceil.wat
new file mode 100644
index 0000000000..9dd6097e85
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f32x4_ceil.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f32x4.ceil (v128.const f32x4 1.5 -2.5 nan -inf)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; frintp v0.4s, v0.4s
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: .byte 0x00, 0x00, 0xc0, 0x3f
+;; 84: .byte 0x00, 0x00, 0x20, 0xc0
+;; 88: .byte 0x00, 0x00, 0xc0, 0x7f
+;; 8c: .byte 0x00, 0x00, 0x80, 0xff
diff --git a/tests/disas/winch/aarch64/v128_ops/f32x4_floor.wat b/tests/disas/winch/aarch64/v128_ops/f32x4_floor.wat
new file mode 100644
index 0000000000..634c4e9ac0
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f32x4_floor.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f32x4.floor (v128.const f32x4 1.5 -2.5 nan -inf)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; frintm v0.4s, v0.4s
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: .byte 0x00, 0x00, 0xc0, 0x3f
+;; 84: .byte 0x00, 0x00, 0x20, 0xc0
+;; 88: .byte 0x00, 0x00, 0xc0, 0x7f
+;; 8c: .byte 0x00, 0x00, 0x80, 0xff
diff --git a/tests/disas/winch/aarch64/v128_ops/f32x4_nearest.wat b/tests/disas/winch/aarch64/v128_ops/f32x4_nearest.wat
new file mode 100644
index 0000000000..a6c952035f
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f32x4_nearest.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f32x4.nearest (v128.const f32x4 1.5 -2.5 nan -inf)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; frintn v0.4s, v0.4s
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: .byte 0x00, 0x00, 0xc0, 0x3f
+;; 84: .byte 0x00, 0x00, 0x20, 0xc0
+;; 88: .byte 0x00, 0x00, 0xc0, 0x7f
+;; 8c: .byte 0x00, 0x00, 0x80, 0xff
diff --git a/tests/disas/winch/aarch64/v128_ops/f32x4_neg.wat b/tests/disas/winch/aarch64/v128_ops/f32x4_neg.wat
new file mode 100644
index 0000000000..3da3c4d1d6
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f32x4_neg.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f32x4.neg (v128.const f32x4 1.5 -2.5 0 -0)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; fneg v0.4s, v0.4s
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: .byte 0x00, 0x00, 0xc0, 0x3f
+;; 84: .byte 0x00, 0x00, 0x20, 0xc0
+;; 88: udf #0
+;; 8c: .byte 0x00, 0x00, 0x00, 0x80
diff --git a/tests/disas/winch/aarch64/v128_ops/f32x4_sqrt.wat b/tests/disas/winch/aarch64/v128_ops/f32x4_sqrt.wat
new file mode 100644
index 0000000000..1c1b1c9e7f
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f32x4_sqrt.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f32x4.sqrt (v128.const f32x4 1.5 -2.5 nan -inf)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; fsqrt v0.4s, v0.4s
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: .byte 0x00, 0x00, 0xc0, 0x3f
+;; 84: .byte 0x00, 0x00, 0x20, 0xc0
+;; 88: .byte 0x00, 0x00, 0xc0, 0x7f
+;; 8c: .byte 0x00, 0x00, 0x80, 0xff
diff --git a/tests/disas/winch/aarch64/v128_ops/f32x4_trunc.wat b/tests/disas/winch/aarch64/v128_ops/f32x4_trunc.wat
new file mode 100644
index 0000000000..9e6db062b6
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f32x4_trunc.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f32x4.trunc (v128.const f32x4 1.5 -2.5 nan -inf)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; frintz v0.4s, v0.4s
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: .byte 0x00, 0x00, 0xc0, 0x3f
+;; 84: .byte 0x00, 0x00, 0x20, 0xc0
+;; 88: .byte 0x00, 0x00, 0xc0, 0x7f
+;; 8c: .byte 0x00, 0x00, 0x80, 0xff
diff --git a/tests/disas/winch/aarch64/v128_ops/f64x2_abs.wat b/tests/disas/winch/aarch64/v128_ops/f64x2_abs.wat
new file mode 100644
index 0000000000..ff12921634
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f64x2_abs.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f64x2.abs (v128.const f64x2 -0 nan)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; fabs v0.2d, v0.2d
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: udf #0
+;; 84: .byte 0x00, 0x00, 0x00, 0x80
+;; 88: udf #0
+;; 8c: .byte 0x00, 0x00, 0xf8, 0x7f
diff --git a/tests/disas/winch/aarch64/v128_ops/f64x2_ceil.wat b/tests/disas/winch/aarch64/v128_ops/f64x2_ceil.wat
new file mode 100644
index 0000000000..190ab3c9ca
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f64x2_ceil.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f64x2.ceil (v128.const f64x2 -1.5 2.5)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; frintp v0.2d, v0.2d
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: udf #0
+;; 84: .byte 0x00, 0x00, 0xf8, 0xbf
+;; 88: udf #0
+;; 8c: .byte 0x00, 0x00, 0x04, 0x40
diff --git a/tests/disas/winch/aarch64/v128_ops/f64x2_floor.wat b/tests/disas/winch/aarch64/v128_ops/f64x2_floor.wat
new file mode 100644
index 0000000000..22f6eb7861
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f64x2_floor.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f64x2.floor (v128.const f64x2 -1.5 2.5)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; frintm v0.2d, v0.2d
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: udf #0
+;; 84: .byte 0x00, 0x00, 0xf8, 0xbf
+;; 88: udf #0
+;; 8c: .byte 0x00, 0x00, 0x04, 0x40
diff --git a/tests/disas/winch/aarch64/v128_ops/f64x2_nearest.wat b/tests/disas/winch/aarch64/v128_ops/f64x2_nearest.wat
new file mode 100644
index 0000000000..c3066ad0f9
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f64x2_nearest.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f64x2.nearest (v128.const f64x2 -1.5 2.5)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; frintn v0.2d, v0.2d
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: udf #0
+;; 84: .byte 0x00, 0x00, 0xf8, 0xbf
+;; 88: udf #0
+;; 8c: .byte 0x00, 0x00, 0x04, 0x40
diff --git a/tests/disas/winch/aarch64/v128_ops/f64x2_neg.wat b/tests/disas/winch/aarch64/v128_ops/f64x2_neg.wat
new file mode 100644
index 0000000000..a3153c76de
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f64x2_neg.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f64x2.neg (v128.const f64x2 1.5 -0)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; fneg v0.2d, v0.2d
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: udf #0
+;; 84: .byte 0x00, 0x00, 0xf8, 0x3f
+;; 88: udf #0
+;; 8c: .byte 0x00, 0x00, 0x00, 0x80
diff --git a/tests/disas/winch/aarch64/v128_ops/f64x2_sqrt.wat b/tests/disas/winch/aarch64/v128_ops/f64x2_sqrt.wat
new file mode 100644
index 0000000000..09facf778f
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f64x2_sqrt.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f64x2.sqrt (v128.const f64x2 -1.5 2.5)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; fsqrt v0.2d, v0.2d
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: udf #0
+;; 84: .byte 0x00, 0x00, 0xf8, 0xbf
+;; 88: udf #0
+;; 8c: .byte 0x00, 0x00, 0x04, 0x40
diff --git a/tests/disas/winch/aarch64/v128_ops/f64x2_trunc.wat b/tests/disas/winch/aarch64/v128_ops/f64x2_trunc.wat
new file mode 100644
index 0000000000..eb939081d7
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/f64x2_trunc.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (f64x2.trunc (v128.const f64x2 -1.5 2.5)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; frintz v0.2d, v0.2d
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: udf #0
+;; 84: .byte 0x00, 0x00, 0xf8, 0xbf
+;; 88: udf #0
+;; 8c: .byte 0x00, 0x00, 0x04, 0x40
diff --git a/tests/disas/winch/aarch64/v128_ops/i16x8_abs.wat b/tests/disas/winch/aarch64/v128_ops/i16x8_abs.wat
new file mode 100644
index 0000000000..34fc19e22c
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/i16x8_abs.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (i16x8.abs (v128.const i16x8 0x8000 -1 1 0 12345 -12345 32767 255)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; abs v0.8h, v0.8h
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: .byte 0x00, 0x80, 0xff, 0xff
+;; 84: udf #1
+;; 88: .byte 0x39, 0x30, 0xc7, 0xcf
+;; 8c: .byte 0xff, 0x7f, 0xff, 0x00
diff --git a/tests/disas/winch/aarch64/v128_ops/i16x8_all_true.wat b/tests/disas/winch/aarch64/v128_ops/i16x8_all_true.wat
new file mode 100644
index 0000000000..89b83e6e5e
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/i16x8_all_true.wat
@@ -0,0 +1,39 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (func (export "run") (result i32)
+ (i16x8.all_true (v128.const i16x8 1 2 3 4 5 6 7 0))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x6c
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x70
+;; uminv h31, v0.8h
+;; mov x0, v31.d[0]
+;; cmp x0, #0
+;; cset x0, ne
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 6c: udf #0xc11f
+;; 70: .byte 0x01, 0x00, 0x02, 0x00
+;; 74: .byte 0x03, 0x00, 0x04, 0x00
+;; 78: .byte 0x05, 0x00, 0x06, 0x00
+;; 7c: udf #7
diff --git a/tests/disas/winch/aarch64/v128_ops/i16x8_neg.wat b/tests/disas/winch/aarch64/v128_ops/i16x8_neg.wat
new file mode 100644
index 0000000000..368e8a95c3
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/i16x8_neg.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (i16x8.neg (v128.const i16x8 1 -2 3 -4 5 -6 7 -8)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; neg v0.8h, v0.8h
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: .byte 0x01, 0x00, 0xfe, 0xff
+;; 84: .byte 0x03, 0x00, 0xfc, 0xff
+;; 88: .byte 0x05, 0x00, 0xfa, 0xff
+;; 8c: .byte 0x07, 0x00, 0xf8, 0xff
diff --git a/tests/disas/winch/aarch64/v128_ops/i32x4_abs.wat b/tests/disas/winch/aarch64/v128_ops/i32x4_abs.wat
new file mode 100644
index 0000000000..d17ff928e4
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/i32x4_abs.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (i32x4.abs (v128.const i32x4 0x80000000 -1 1 0)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; abs v0.4s, v0.4s
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: .byte 0x00, 0x00, 0x00, 0x80
+;; 84: .byte 0xff, 0xff, 0xff, 0xff
+;; 88: udf #1
+;; 8c: udf #0
diff --git a/tests/disas/winch/aarch64/v128_ops/i32x4_all_true.wat b/tests/disas/winch/aarch64/v128_ops/i32x4_all_true.wat
new file mode 100644
index 0000000000..b36f79f958
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/i32x4_all_true.wat
@@ -0,0 +1,39 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (func (export "run") (result i32)
+ (i32x4.all_true (v128.const i32x4 1 2 3 4))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x6c
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x70
+;; uminv s31, v0.4s
+;; mov x0, v31.d[0]
+;; cmp x0, #0
+;; cset x0, ne
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 6c: udf #0xc11f
+;; 70: udf #1
+;; 74: udf #2
+;; 78: udf #3
+;; 7c: udf #4
diff --git a/tests/disas/winch/aarch64/v128_ops/i32x4_neg.wat b/tests/disas/winch/aarch64/v128_ops/i32x4_neg.wat
new file mode 100644
index 0000000000..6f015551f0
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/i32x4_neg.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (i32x4.neg (v128.const i32x4 1 -2 3 0x80000000)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; neg v0.4s, v0.4s
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: udf #1
+;; 84: .byte 0xfe, 0xff, 0xff, 0xff
+;; 88: udf #3
+;; 8c: .byte 0x00, 0x00, 0x00, 0x80
diff --git a/tests/disas/winch/aarch64/v128_ops/i64x2_abs.wat b/tests/disas/winch/aarch64/v128_ops/i64x2_abs.wat
new file mode 100644
index 0000000000..39c0fcd270
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/i64x2_abs.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (i64x2.abs (v128.const i64x2 0x8000000000000000 -1)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; abs v0.2d, v0.2d
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: udf #0
+;; 84: .byte 0x00, 0x00, 0x00, 0x80
+;; 88: .byte 0xff, 0xff, 0xff, 0xff
+;; 8c: .byte 0xff, 0xff, 0xff, 0xff
diff --git a/tests/disas/winch/aarch64/v128_ops/i64x2_all_true.wat b/tests/disas/winch/aarch64/v128_ops/i64x2_all_true.wat
new file mode 100644
index 0000000000..7c6a531c9a
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/i64x2_all_true.wat
@@ -0,0 +1,39 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (func (export "run") (result i32)
+ (i64x2.all_true (v128.const i64x2 1 0))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x6c
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x70
+;; cmeq v31.2d, v0.2d, #0
+;; addp v31.2d, v31.2d, v31.2d
+;; fcmp d31, d31
+;; cset x0, eq
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 6c: udf #0xc11f
+;; 70: udf #1
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
diff --git a/tests/disas/winch/aarch64/v128_ops/i64x2_neg.wat b/tests/disas/winch/aarch64/v128_ops/i64x2_neg.wat
new file mode 100644
index 0000000000..07799119dc
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/i64x2_neg.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (i64x2.neg (v128.const i64x2 1 -2)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; neg v0.2d, v0.2d
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: udf #1
+;; 84: udf #0
+;; 88: .byte 0xfe, 0xff, 0xff, 0xff
+;; 8c: .byte 0xff, 0xff, 0xff, 0xff
diff --git a/tests/disas/winch/aarch64/v128_ops/i8x16_abs.wat b/tests/disas/winch/aarch64/v128_ops/i8x16_abs.wat
new file mode 100644
index 0000000000..d8d68db004
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/i8x16_abs.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (i8x16.abs (v128.const i8x16 128 255 1 0 2 254 127 3 4 5 6 7 8 9 10 11)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; abs v0.16b, v0.16b
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: .byte 0x80, 0xff, 0x01, 0x00
+;; 84: .byte 0x02, 0xfe, 0x7f, 0x03
+;; 88: .byte 0x04, 0x05, 0x06, 0x07
+;; 8c: add w8, w8, w10, lsl #2
diff --git a/tests/disas/winch/aarch64/v128_ops/i8x16_all_true.wat b/tests/disas/winch/aarch64/v128_ops/i8x16_all_true.wat
new file mode 100644
index 0000000000..6c41d40af6
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/i8x16_all_true.wat
@@ -0,0 +1,39 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (func (export "run") (result i32)
+ (i8x16.all_true (v128.const i8x16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x6c
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x70
+;; uminv b31, v0.16b
+;; mov x0, v31.d[0]
+;; cmp x0, #0
+;; cset x0, ne
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 6c: udf #0xc11f
+;; 70: subr z1.b, p0/m, z1.b, z16.b
+;; 74: stxrb w7, w5, [x16]
+;; 78: .byte 0x09, 0x0a, 0x0b, 0x0c
+;; 7c: adr x13, #0x1e23c
diff --git a/tests/disas/winch/aarch64/v128_ops/i8x16_neg.wat b/tests/disas/winch/aarch64/v128_ops/i8x16_neg.wat
new file mode 100644
index 0000000000..0186f258d3
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/i8x16_neg.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (i8x16.neg (v128.const i8x16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; neg v0.16b, v0.16b
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: subr z1.b, p0/m, z1.b, z16.b
+;; 84: stxrb w7, w5, [x16]
+;; 88: .byte 0x09, 0x0a, 0x0b, 0x0c
+;; 8c: adr x13, #0x1e24c
diff --git a/tests/disas/winch/aarch64/v128_ops/i8x16_popcnt.wat b/tests/disas/winch/aarch64/v128_ops/i8x16_popcnt.wat
new file mode 100644
index 0000000000..11eab6e8ff
--- /dev/null
+++ b/tests/disas/winch/aarch64/v128_ops/i8x16_popcnt.wat
@@ -0,0 +1,45 @@
+;;! target = "aarch64"
+;;! test = "winch"
+
+(module
+ (memory 1)
+ (func (export "run")
+ (v128.store (i32.const 0)
+ (i8x16.popcnt (v128.const i8x16 0 1 3 7 15 31 63 127 255 128 64 170 85 204 51 2)))))
+;; wasm[0]::function[0]:
+;; stp x29, x30, [sp, #-0x10]!
+;; mov x29, sp
+;; str x28, [sp, #-0x10]!
+;; mov x28, sp
+;; ldur x16, [x0, #8]
+;; ldur x16, [x16, #0x18]
+;; mov x17, #0
+;; movk x17, #0x10
+;; add x16, x16, x17
+;; cmp sp, x16
+;; b.lo #0x70
+;; 2c: mov x9, x0
+;; sub x28, x28, #0x10
+;; mov sp, x28
+;; stur x0, [x28, #8]
+;; stur x1, [x28]
+;; ldr q0, #0x80
+;; cnt v0.16b, v0.16b
+;; mov x0, #0
+;; ldur x1, [x9, #0x38]
+;; add x1, x1, w0, uxtw
+;; stur q0, [x1]
+;; add x28, x28, #0x10
+;; mov sp, x28
+;; mov sp, x28
+;; ldr x28, [sp], #0x10
+;; ldp x29, x30, [sp], #0x10
+;; ret
+;; 70: udf #0xc11f
+;; 74: udf #0
+;; 78: udf #0
+;; 7c: udf #0
+;; 80: .byte 0x00, 0x01, 0x03, 0x07
+;; 84: .byte 0x0f, 0x1f, 0x3f, 0x7f
+;; 88: orr xzr, x7, x0, lsr #32
+;; 8c: .byte 0x55, 0xcc, 0x33, 0x02
diff --git a/winch/codegen/src/isa/aarch64/asm.rs b/winch/codegen/src/isa/aarch64/asm.rs
index 43459e4c54..431df7b15b 100644
--- a/winch/codegen/src/isa/aarch64/asm.rs
+++ b/winch/codegen/src/isa/aarch64/asm.rs
@@ -497,6 +497,16 @@ impl Assembler {
});
}
+ /// Vector reduction across lanes
+ pub fn vec_lanes(&mut self, op: VecLanesOp, rn: Reg, rd: WritableReg, size: VectorSize) {
+ self.emit(Inst::VecLanes {
+ op,
+ rd: rd.map(Into::into),
+ rn: rn.into(),
+ size,
+ });
+ }
+
/// Vector ALU op
pub fn vec_rrr(
&mut self,
diff --git a/winch/codegen/src/isa/aarch64/masm.rs b/winch/codegen/src/isa/aarch64/masm.rs
index c4779f9f2b..1727bb87a0 100644
--- a/winch/codegen/src/isa/aarch64/masm.rs
+++ b/winch/codegen/src/isa/aarch64/masm.rs
@@ -34,7 +34,7 @@ use cranelift_codegen::{
isa::aarch64,
isa::aarch64::inst::{
self, Cond, ExtendOp, Imm12, ImmLogic, ImmShift, SImm7Scaled, SImm9, ScalarSize,
- VecALUModOp, VecALUOp, VecExtendOp, VecMisc2, VecRRNarrowOp, VectorSize,
+ VecALUModOp, VecALUOp, VecExtendOp, VecLanesOp, VecMisc2, VecRRNarrowOp, VectorSize,
},
settings,
};
@@ -1760,12 +1760,30 @@ impl Masm for MacroAssembler {
Err(format_err!(CodeGenError::unimplemented_masm_instruction()))
}
- fn v128_abs(&mut self, _src: Reg, _dst: WritableReg, _kind: V128AbsKind) -> Result<()> {
- bail!(CodeGenError::unimplemented_masm_instruction())
+ fn v128_abs(&mut self, src: Reg, dst: WritableReg, kind: V128AbsKind) -> Result<()> {
+ let (misc_op, size) = match kind {
+ V128AbsKind::I8x16 => (VecMisc2::Abs, VectorSize::Size8x16),
+ V128AbsKind::I16x8 => (VecMisc2::Abs, VectorSize::Size16x8),
+ V128AbsKind::I32x4 => (VecMisc2::Abs, VectorSize::Size32x4),
+ V128AbsKind::I64x2 => (VecMisc2::Abs, VectorSize::Size64x2),
+ V128AbsKind::F32x4 => (VecMisc2::Fabs, VectorSize::Size32x4),
+ V128AbsKind::F64x2 => (VecMisc2::Fabs, VectorSize::Size64x2),
+ };
+ self.asm.vec_misc(misc_op, src, dst, size);
+ Ok(())
}
- fn v128_neg(&mut self, _op: WritableReg, _kind: V128NegKind) -> Result<()> {
- Err(format_err!(CodeGenError::unimplemented_masm_instruction()))
+ fn v128_neg(&mut self, op: WritableReg, kind: V128NegKind) -> Result<()> {
+ let (misc_op, size) = match kind {
+ V128NegKind::I8x16 => (VecMisc2::Neg, VectorSize::Size8x16),
+ V128NegKind::I16x8 => (VecMisc2::Neg, VectorSize::Size16x8),
+ V128NegKind::I32x4 => (VecMisc2::Neg, VectorSize::Size32x4),
+ V128NegKind::I64x2 => (VecMisc2::Neg, VectorSize::Size64x2),
+ V128NegKind::F32x4 => (VecMisc2::Fneg, VectorSize::Size32x4),
+ V128NegKind::F64x2 => (VecMisc2::Fneg, VectorSize::Size64x2),
+ };
+ self.asm.vec_misc(misc_op, op.to_reg(), op, size);
+ Ok(())
}
fn v128_shift(
@@ -1787,8 +1805,43 @@ impl Masm for MacroAssembler {
bail!(CodeGenError::unimplemented_masm_instruction())
}
- fn v128_all_true(&mut self, _src: Reg, _dst: WritableReg, _size: OperandSize) -> Result<()> {
- bail!(CodeGenError::unimplemented_masm_instruction())
+ fn v128_all_true(&mut self, src: Reg, dst: WritableReg, size: OperandSize) -> Result<()> {
+ match size {
+ OperandSize::S8 | OperandSize::S16 | OperandSize::S32 => {
+ self.with_scratch::<FloatScratch, _>(|masm, tmp| {
+ masm.asm.vec_lanes(
+ VecLanesOp::Uminv,
+ src,
+ tmp.writable(),
+ VectorSize::from_lane_size(size.into(), true),
+ );
+ masm.asm.mov_from_vec(tmp.inner(), dst, 0, OperandSize::S64);
+ });
+ self.asm.subs_ir(
+ Imm12::maybe_from_u64(0).unwrap(),
+ dst.to_reg(),
+ OperandSize::S64,
+ );
+ self.asm.cset(dst, Cond::Ne);
+ }
+ OperandSize::S64 => {
+ self.with_scratch::<FloatScratch, _>(|masm, tmp| {
+ masm.asm
+ .vec_misc(VecMisc2::Cmeq0, src, tmp.writable(), VectorSize::Size64x2);
+ masm.asm.vec_rrr(
+ VecALUOp::Addp,
+ tmp.inner(),
+ tmp.inner(),
+ tmp.writable(),
+ VectorSize::Size64x2,
+ );
+ masm.asm.fcmp(tmp.inner(), tmp.inner(), OperandSize::S64);
+ });
+ self.asm.cset(dst, Cond::Eq);
+ }
+ OperandSize::S128 => bail!(CodeGenError::unexpected_operand_size()),
+ }
+ Ok(())
}
fn v128_bitmask(&mut self, _src: Reg, _dst: WritableReg, _size: OperandSize) -> Result<()> {
@@ -1797,10 +1850,23 @@ impl Masm for MacroAssembler {
fn v128_trunc(
&mut self,
- _context: &mut CodeGenContext<Emission>,
- _kind: V128TruncKind,
+ context: &mut CodeGenContext<Emission>,
+ kind: V128TruncKind,
) -> Result<()> {
- bail!(CodeGenError::unimplemented_masm_instruction())
+ match kind {
+ V128TruncKind::F32x4 | V128TruncKind::F64x2 => {
+ let reg = writable!(context.pop_to_reg(self, None)?.reg);
+ self.asm.vec_misc(
+ VecMisc2::Frintz,
+ reg.to_reg(),
+ reg,
+ VectorSize::from_lane_size(kind.dst_lane_size().into(), true),
+ );
+ context.stack.push(TypedReg::v128(reg.to_reg()).into());
+ Ok(())
+ }
+ _ => bail!(CodeGenError::unimplemented_masm_instruction()),
+ }
}
fn v128_min(
@@ -1844,8 +1910,12 @@ impl Masm for MacroAssembler {
bail!(CodeGenError::unimplemented_masm_instruction())
}
- fn v128_popcnt(&mut self, _context: &mut CodeGenContext<Emission>) -> Result<()> {
- bail!(CodeGenError::unimplemented_masm_instruction())
+ fn v128_popcnt(&mut self, context: &mut CodeGenContext<Emission>) -> Result<()> {
+ let reg = writable!(context.pop_to_reg(self, None)?.reg);
+ self.asm
+ .vec_misc(VecMisc2::Cnt, reg.to_reg(), reg, VectorSize::Size8x16);
+ context.stack.push(TypedReg::v128(reg.to_reg()).into());
+ Ok(())
}
fn v128_avgr(
@@ -1868,20 +1938,44 @@ impl Masm for MacroAssembler {
bail!(CodeGenError::unimplemented_masm_instruction())
}
- fn v128_sqrt(&mut self, _src: Reg, _dst: WritableReg, _size: OperandSize) -> Result<()> {
- bail!(CodeGenError::unimplemented_masm_instruction())
+ fn v128_sqrt(&mut self, src: Reg, dst: WritableReg, size: OperandSize) -> Result<()> {
+ self.asm.vec_misc(
+ VecMisc2::Fsqrt,
+ src,
+ dst,
+ VectorSize::from_lane_size(size.into(), true),
+ );
+ Ok(())
}
- fn v128_ceil(&mut self, _src: Reg, _dst: WritableReg, _size: OperandSize) -> Result<()> {
- bail!(CodeGenError::unimplemented_masm_instruction())
+ fn v128_ceil(&mut self, src: Reg, dst: WritableReg, size: OperandSize) -> Result<()> {
+ self.asm.vec_misc(
+ VecMisc2::Frintp,
+ src,
+ dst,
+ VectorSize::from_lane_size(size.into(), true),
+ );
+ Ok(())
}
- fn v128_floor(&mut self, _src: Reg, _dst: WritableReg, _size: OperandSize) -> Result<()> {
- bail!(CodeGenError::unimplemented_masm_instruction())
+ fn v128_floor(&mut self, src: Reg, dst: WritableReg, size: OperandSize) -> Result<()> {
+ self.asm.vec_misc(
+ VecMisc2::Frintm,
+ src,
+ dst,
+ VectorSize::from_lane_size(size.into(), true),
+ );
+ Ok(())
}
- fn v128_nearest(&mut self, _src: Reg, _dst: WritableReg, _size: OperandSize) -> Result<()> {
- bail!(CodeGenError::unimplemented_masm_instruction())
+ fn v128_nearest(&mut self, src: Reg, dst: WritableReg, size: OperandSize) -> Result<()> {
+ self.asm.vec_misc(
+ VecMisc2::Frintn,
+ src,
+ dst,
+ VectorSize::from_lane_size(size.into(), true),
+ );
+ Ok(())
}
fn v128_pmin(