SIMD integer abs and bitmask instructions (#2703)

Adds full support for the {i8x16,i16x8,i32x4}.abs instructions merged
to the SIMD proposal in https://github.com/WebAssembly/simd/pull/128
as well as the {i8x16,i16x8,i32x4}.bitmask instructions proposed in
https://github.com/WebAssembly/simd/pull/201.

1 files changed, 17 insertions, 1 deletions

diff --git a/test/spec/simd.wast b/test/spec/simd.wast
index d64789f3e..90bf1a581 100644
--- a/test/spec/simd.wast
+++ b/test/spec/simd.wast
@@ -103,9 +103,11 @@
  (func (export "v128.bitselect") (param $0 v128) (param $1 v128) (param $2 v128) (result v128)
    (v128.bitselect (local.get $0) (local.get $1) (local.get $2))
  )
+ (func (export "i8x16.abs") (param $0 v128) (result v128) (i8x16.abs (local.get $0)))
  (func (export "i8x16.neg") (param $0 v128) (result v128) (i8x16.neg (local.get $0)))
  (func (export "i8x16.any_true") (param $0 v128) (result i32) (i8x16.any_true (local.get $0)))
  (func (export "i8x16.all_true") (param $0 v128) (result i32) (i8x16.all_true (local.get $0)))
+ (func (export "i8x16.bitmask") (param $0 v128) (result i32) (i8x16.bitmask (local.get $0)))
  (func (export "i8x16.shl") (param $0 v128) (param $1 i32) (result v128) (i8x16.shl (local.get $0) (local.get $1)))
  (func (export "i8x16.shr_s") (param $0 v128) (param $1 i32) (result v128) (i8x16.shr_s (local.get $0) (local.get $1)))
  (func (export "i8x16.shr_u") (param $0 v128) (param $1 i32) (result v128) (i8x16.shr_u (local.get $0) (local.get $1)))
@@ -121,9 +123,11 @@
  (func (export "i8x16.max_s") (param $0 v128) (param $1 v128) (result v128) (i8x16.max_s (local.get $0) (local.get $1)))
  (func (export "i8x16.max_u") (param $0 v128) (param $1 v128) (result v128) (i8x16.max_u (local.get $0) (local.get $1)))
  (func (export "i8x16.avgr_u") (param $0 v128) (param $1 v128) (result v128) (i8x16.avgr_u (local.get $0) (local.get $1)))
+ (func (export "i16x8.abs") (param $0 v128) (result v128) (i16x8.abs (local.get $0)))
  (func (export "i16x8.neg") (param $0 v128) (result v128) (i16x8.neg (local.get $0)))
  (func (export "i16x8.any_true") (param $0 v128) (result i32) (i16x8.any_true (local.get $0)))
  (func (export "i16x8.all_true") (param $0 v128) (result i32) (i16x8.all_true (local.get $0)))
+ (func (export "i16x8.bitmask") (param $0 v128) (result i32) (i16x8.bitmask (local.get $0)))
  (func (export "i16x8.shl") (param $0 v128) (param $1 i32) (result v128) (i16x8.shl (local.get $0) (local.get $1)))
  (func (export "i16x8.shr_s") (param $0 v128) (param $1 i32) (result v128) (i16x8.shr_s (local.get $0) (local.get $1)))
  (func (export "i16x8.shr_u") (param $0 v128) (param $1 i32) (result v128) (i16x8.shr_u (local.get $0) (local.get $1)))
@@ -139,9 +143,11 @@
  (func (export "i16x8.max_s") (param $0 v128) (param $1 v128) (result v128) (i16x8.max_s (local.get $0) (local.get $1)))
  (func (export "i16x8.max_u") (param $0 v128) (param $1 v128) (result v128) (i16x8.max_u (local.get $0) (local.get $1)))
  (func (export "i16x8.avgr_u") (param $0 v128) (param $1 v128) (result v128) (i16x8.avgr_u (local.get $0) (local.get $1)))
+ (func (export "i32x4.abs") (param $0 v128) (result v128) (i32x4.abs (local.get $0)))
  (func (export "i32x4.neg") (param $0 v128) (result v128) (i32x4.neg (local.get $0)))
  (func (export "i32x4.any_true") (param $0 v128) (result i32) (i32x4.any_true (local.get $0)))
  (func (export "i32x4.all_true") (param $0 v128) (result i32) (i32x4.all_true (local.get $0)))
+ (func (export "i32x4.bitmask") (param $0 v128) (result i32) (i32x4.bitmask (local.get $0)))
  (func (export "i32x4.shl") (param $0 v128) (param $1 i32) (result v128) (i32x4.shl (local.get $0) (local.get $1)))
  (func (export "i32x4.shr_s") (param $0 v128) (param $1 i32) (result v128) (i32x4.shr_s (local.get $0) (local.get $1)))
  (func (export "i32x4.shr_u") (param $0 v128) (param $1 i32) (result v128) (i32x4.shr_u (local.get $0) (local.get $1)))
@@ -490,6 +496,9 @@
 )
 
 ;; i8x16 arithmetic
+(assert_return (invoke "i8x16.abs" (v128.const i8x16 0 1 42 -3 -56 127 -128 -126 0 -1 -42 3 56 -127 -128 126))
+  (v128.const i8x16 0 1 42 3 56 127 -128 126 0 1 42 3 56 127 -128 126)
+)
 (assert_return (invoke "i8x16.neg" (v128.const i32x4 0 1 42 -3 -56 127 -128 -126 0 -1 -42 3 56 -127 -128 126))
   (v128.const i32x4 0 -1 -42 3 56 -127 -128 126 0 1 42 -3 -56 127 -128 -126)
 )
@@ -501,6 +510,7 @@
 (assert_return (invoke "i8x16.all_true" (v128.const i32x4 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0)) (i32.const 0))
 (assert_return (invoke "i8x16.all_true" (v128.const i32x4 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1)) (i32.const 0))
 (assert_return (invoke "i8x16.all_true" (v128.const i32x4 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1)) (i32.const 1))
+(assert_return (invoke "i8x16.bitmask" (v128.const i8x16 -1 0 1 -128 127 -127 0 128 -1 0 1 -128 127 -127 0 128)) (i32.const 43433))
 (assert_return (invoke "i8x16.shl" (v128.const i32x4 0 1 2 4 8 16 32 64 -128 3 6 12 24 48 96 -64) (i32.const 1))
   (v128.const i32x4 0 2 4 8 16 32 64 -128 0 6 12 24 48 96 -64 -128)
 )
@@ -605,6 +615,9 @@
 )
 
 ;; i16x8 arithmetic
+(assert_return (invoke "i16x8.abs" (v128.const i16x8 0 1 42 -3 -56 32767 -32768 32766))
+  (v128.const i16x8 0 1 42 3 56 32767 -32768 32766)
+)
 (assert_return (invoke "i16x8.neg" (v128.const i32x4 0 1 42 -3 -56 32767 -32768 32766))
   (v128.const i32x4 0 -1 -42 3 56 -32767 -32768 -32766)
 )
@@ -616,6 +629,7 @@
 (assert_return (invoke "i16x8.all_true" (v128.const i32x4 0 0 1 0 0 0 0 0)) (i32.const 0))
 (assert_return (invoke "i16x8.all_true" (v128.const i32x4 1 1 1 1 1 0 1 1)) (i32.const 0))
 (assert_return (invoke "i16x8.all_true" (v128.const i32x4 1 1 1 1 1 1 1 1)) (i32.const 1))
+(assert_return (invoke "i16x8.bitmask" (v128.const i16x8 -1 0 1 -32768 32767 -32767 0 32768)) (i32.const 169))
 (assert_return (invoke "i16x8.shl" (v128.const i32x4 0 8 16 128 256 2048 4096 -32768) (i32.const 1)) (v128.const i32x4 0 16 32 256 512 4096 8192 0))
 (assert_return (invoke "i16x8.shl" (v128.const i32x4 0 8 16 128 256 2048 4096 -32768) (i32.const 16)) (v128.const i32x4 0 8 16 128 256 2048 4096 -32768))
 (assert_return (invoke "i16x8.shr_u" (v128.const i32x4 0 8 16 128 256 2048 4096 -32768) (i32.const 1)) (v128.const i32x4 0 4 8 64 128 1024 2048 16384))
@@ -708,7 +722,8 @@
 )
 
 ;; i32x4 arithmetic
-(assert_return (invoke "i32x4.neg" (v128.const i32x4 0 1 0x80000000 0x7fffffff)) (v128.const i32x4 0 -1 0x80000000 0x80000001))
+(assert_return (invoke "i32x4.abs" (v128.const i32x4 0 1 0x80000000 0x80000001)) (v128.const i32x4 0 1 0x80000000 0x7fffffff))
+(assert_return (invoke "i32x4.neg" (v128.const i32x4 0 1 0x80000000 0x80000001)) (v128.const i32x4 0 -1 0x80000000 0x7fffffff))
 (assert_return (invoke "i32x4.any_true" (v128.const i32x4 0 0 0 0)) (i32.const 0))
 (assert_return (invoke "i32x4.any_true" (v128.const i32x4 0 0 1 0)) (i32.const 1))
 (assert_return (invoke "i32x4.any_true" (v128.const i32x4 1 0 1 1)) (i32.const 1))
@@ -717,6 +732,7 @@
 (assert_return (invoke "i32x4.all_true" (v128.const i32x4 0 0 1 0)) (i32.const 0))
 (assert_return (invoke "i32x4.all_true" (v128.const i32x4 1 0 1 1)) (i32.const 0))
 (assert_return (invoke "i32x4.all_true" (v128.const i32x4 1 1 1 1)) (i32.const 1))
+(assert_return (invoke "i32x4.bitmask" (v128.const i32x4 -1 0 -128 127)) (i32.const 5))
 (assert_return (invoke "i32x4.shl" (v128.const i32x4 1 0x40000000 0x80000000 -1) (i32.const 1)) (v128.const i32x4 2 0x80000000 0 -2))
 (assert_return (invoke "i32x4.shl" (v128.const i32x4 1 0x40000000 0x80000000 -1) (i32.const 32)) (v128.const i32x4 1 0x40000000 0x80000000 -1))
 (assert_return (invoke "i32x4.shr_s" (v128.const i32x4 1 0x40000000 0x80000000 -1) (i32.const 1)) (v128.const i32x4 0 0x20000000 0xc0000000 -1))