mirror of
https://github.com/golang/go
synced 2024-11-11 19:21:37 -07:00
test/codegen: enable more tests for ppc64/ppc64le
Adding cases for ppc64,ppc64le to the codegen tests where appropriate. Change-Id: Idf8cbe88a4ab4406a4ef1ea777bd15a58b68f3ed Reviewed-on: https://go-review.googlesource.com/c/142557 Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
This commit is contained in:
Lynn Boger
parent
4b36e129f8
commit
39fa301bdc
@ -48,12 +48,16 @@ func Pow2Muls(n1, n2 int) (int, int) {
|
||||
// 386:"SHLL\t[$]5",-"IMULL"
|
||||
// arm:"SLL\t[$]5",-"MUL"
|
||||
// arm64:"LSL\t[$]5",-"MUL"
|
||||
// ppc64:"SLD\t[$]5",-"MUL"
|
||||
// ppc64le:"SLD\t[$]5",-"MUL"
|
||||
a := n1 * 32
|
||||
|
||||
// amd64:"SHLQ\t[$]6",-"IMULQ"
|
||||
// 386:"SHLL\t[$]6",-"IMULL"
|
||||
// arm:"SLL\t[$]6",-"MUL"
|
||||
// arm64:`NEG\sR[0-9]+<<6,\sR[0-9]+`,-`LSL`,-`MUL`
|
||||
// ppc64:"SLD\t[$]6","NEG\\sR[0-9]+,\\sR[0-9]+",-"MUL"
|
||||
// ppc64le:"SLD\t[$]6","NEG\\sR[0-9]+,\\sR[0-9]+",-"MUL"
|
||||
b := -64 * n2
|
||||
|
||||
return a, b
|
||||
@ -117,12 +121,16 @@ func Pow2Divs(n1 uint, n2 int) (uint, int) {
|
||||
// amd64:"SHRQ\t[$]5",-"DIVQ"
|
||||
// arm:"SRL\t[$]5",-".*udiv"
|
||||
// arm64:"LSR\t[$]5",-"UDIV"
|
||||
// ppc64:"SRD"
|
||||
// ppc64le:"SRD"
|
||||
a := n1 / 32 // unsigned
|
||||
|
||||
// amd64:"SARQ\t[$]6",-"IDIVQ"
|
||||
// 386:"SARL\t[$]6",-"IDIVL"
|
||||
// arm:"SRA\t[$]6",-".*udiv"
|
||||
// arm64:"ASR\t[$]6",-"SDIV"
|
||||
// ppc64:"SRAD"
|
||||
// ppc64le:"SRAD"
|
||||
b := n2 / 64 // signed
|
||||
|
||||
return a, b
|
||||
@ -149,6 +157,8 @@ func Pow2Mods(n1 uint, n2 int) (uint, int) {
|
||||
// amd64:"ANDQ\t[$]31",-"DIVQ"
|
||||
// arm:"AND\t[$]31",-".*udiv"
|
||||
// arm64:"AND\t[$]31",-"UDIV"
|
||||
// ppc64:"ANDCC\t[$]31"
|
||||
// ppc64le:"ANDCC\t[$]31"
|
||||
a := n1 % 32 // unsigned
|
||||
|
||||
// 386:-"IDIVL"
|
||||
@ -177,36 +187,48 @@ func ConstMods(n1 uint, n2 int) (uint, int) {
|
||||
func LenDiv1(a []int) int {
|
||||
// 386:"SHRL\t[$]10"
|
||||
// amd64:"SHRQ\t[$]10"
|
||||
// ppc64:"SRD"\t[$]10"
|
||||
// ppc64le:"SRD"\t[$]10"
|
||||
return len(a) / 1024
|
||||
}
|
||||
|
||||
func LenDiv2(s string) int {
|
||||
// 386:"SHRL\t[$]11"
|
||||
// amd64:"SHRQ\t[$]11"
|
||||
// ppc64:"SRD\t[$]11"
|
||||
// ppc64le:"SRD\t[$]11"
|
||||
return len(s) / (4097 >> 1)
|
||||
}
|
||||
|
||||
func LenMod1(a []int) int {
|
||||
// 386:"ANDL\t[$]1023"
|
||||
// amd64:"ANDQ\t[$]1023"
|
||||
// ppc64:"ANDCC\t[$]1023"
|
||||
// ppc64le:"ANDCC\t[$]1023"
|
||||
return len(a) % 1024
|
||||
}
|
||||
|
||||
func LenMod2(s string) int {
|
||||
// 386:"ANDL\t[$]2047"
|
||||
// amd64:"ANDQ\t[$]2047"
|
||||
// ppc64:"ANDCC\t[$]2047"
|
||||
// ppc64le:"ANDCC\t[$]2047"
|
||||
return len(s) % (4097 >> 1)
|
||||
}
|
||||
|
||||
func CapDiv(a []int) int {
|
||||
// 386:"SHRL\t[$]12"
|
||||
// amd64:"SHRQ\t[$]12"
|
||||
// ppc64:"SRD\t[$]12"
|
||||
// ppc64le:"SRD\t[$]12"
|
||||
return cap(a) / ((1 << 11) + 2048)
|
||||
}
|
||||
|
||||
func CapMod(a []int) int {
|
||||
// 386:"ANDL\t[$]4095"
|
||||
// amd64:"ANDQ\t[$]4095"
|
||||
// ppc64:"ANDCC\t[$]4095"
|
||||
// ppc64le:"ANDCC\t[$]4095"
|
||||
return cap(a) % ((1 << 11) + 2048)
|
||||
}
|
||||
|
||||
|
||||
@ -36,6 +36,7 @@ func CompareString2(s string) bool {
|
||||
func CompareString3(s string) bool {
|
||||
// amd64:`CMPQ\t\(.*\), [A-Z]`
|
||||
// arm64:-`CMPW\t`
|
||||
// ppc64:-`CMPW\t`
|
||||
// ppc64le:-`CMPW\t`
|
||||
// s390x:-`CMPW\t`
|
||||
return s == "xxxxxxxx"
|
||||
|
||||
@ -16,6 +16,8 @@ func movesmall4() {
|
||||
// amd64:-".*memmove"
|
||||
// arm:-".*memmove"
|
||||
// arm64:-".*memmove"
|
||||
// ppc64:-".*memmove"
|
||||
// ppc64le:-".*memmove"
|
||||
copy(x[1:], x[:])
|
||||
}
|
||||
|
||||
@ -24,6 +26,8 @@ func movesmall7() {
|
||||
// 386:-".*memmove"
|
||||
// amd64:-".*memmove"
|
||||
// arm64:-".*memmove"
|
||||
// ppc64:-".*memmove"
|
||||
// ppc64le:-".*memmove"
|
||||
copy(x[1:], x[:])
|
||||
}
|
||||
|
||||
@ -63,6 +67,7 @@ func moveDisjointNoOverlap(a *[256]byte) {
|
||||
|
||||
func ptrEqual() {
|
||||
// amd64:-"JEQ",-"JNE"
|
||||
// ppc64:-"BEQ",-"BNE"
|
||||
// ppc64le:-"BEQ",-"BNE"
|
||||
// s390x:-"BEQ",-"BNE"
|
||||
copy(x[:], x[:])
|
||||
@ -70,6 +75,7 @@ func ptrEqual() {
|
||||
|
||||
func ptrOneOffset() {
|
||||
// amd64:-"JEQ",-"JNE"
|
||||
// ppc64:-"BEQ",-"BNE"
|
||||
// ppc64le:-"BEQ",-"BNE"
|
||||
// s390x:-"BEQ",-"BNE"
|
||||
copy(x[1:], x[:])
|
||||
@ -77,6 +83,7 @@ func ptrOneOffset() {
|
||||
|
||||
func ptrBothOffset() {
|
||||
// amd64:-"JEQ",-"JNE"
|
||||
// ppc64:-"BEQ",-"BNE"
|
||||
// ppc64le:-"BEQ",-"BNE"
|
||||
// s390x:-"BEQ",-"BNE"
|
||||
copy(x[1:], x[2:])
|
||||
|
||||
@ -22,6 +22,8 @@ func Mul2(f float64) float64 {
|
||||
// amd64:"ADDSD",-"MULSD"
|
||||
// arm/7:"ADDD",-"MULD"
|
||||
// arm64:"FADDD",-"FMULD"
|
||||
// ppc64:"FADD",-"FMUL"
|
||||
// ppc64le:"FADD",-"FMUL"
|
||||
return f * 2.0
|
||||
}
|
||||
|
||||
@ -31,6 +33,8 @@ func DivPow2(f1, f2, f3 float64) (float64, float64, float64) {
|
||||
// amd64:"MULSD",-"DIVSD"
|
||||
// arm/7:"MULD",-"DIVD"
|
||||
// arm64:"FMULD",-"FDIVD"
|
||||
// ppc64:"FMUL",-"FDIV"
|
||||
// ppc64le:"FMUL",-"FDIV"
|
||||
x := f1 / 16.0
|
||||
|
||||
// 386/sse2:"MULSD",-"DIVSD"
|
||||
@ -38,6 +42,8 @@ func DivPow2(f1, f2, f3 float64) (float64, float64, float64) {
|
||||
// amd64:"MULSD",-"DIVSD"
|
||||
// arm/7:"MULD",-"DIVD"
|
||||
// arm64:"FMULD",-"FDIVD"
|
||||
// ppc64:"FMUL",-"FDIVD"
|
||||
// ppc64le:"FMUL",-"FDIVD"
|
||||
y := f2 / 0.125
|
||||
|
||||
// 386/sse2:"ADDSD",-"DIVSD",-"MULSD"
|
||||
@ -45,6 +51,8 @@ func DivPow2(f1, f2, f3 float64) (float64, float64, float64) {
|
||||
// amd64:"ADDSD",-"DIVSD",-"MULSD"
|
||||
// arm/7:"ADDD",-"MULD",-"DIVD"
|
||||
// arm64:"FADDD",-"FMULD",-"FDIVD"
|
||||
// ppc64:"FADD",-"FMUL",-"FDIV"
|
||||
// ppc64le:"FADD",-"FMUL",-"FDIV"
|
||||
z := f3 / 0.5
|
||||
|
||||
return x, y, z
|
||||
@ -71,6 +79,7 @@ func indexStore(b0 []float64, b1 float64, idx int) {
|
||||
|
||||
func FusedAdd32(x, y, z float32) float32 {
|
||||
// s390x:"FMADDS\t"
|
||||
// ppc64:"FMADDS\t"
|
||||
// ppc64le:"FMADDS\t"
|
||||
// arm64:"FMADDS"
|
||||
return x*y + z
|
||||
@ -78,6 +87,7 @@ func FusedAdd32(x, y, z float32) float32 {
|
||||
|
||||
func FusedSub32_a(x, y, z float32) float32 {
|
||||
// s390x:"FMSUBS\t"
|
||||
// ppc64:"FMSUBS\t"
|
||||
// ppc64le:"FMSUBS\t"
|
||||
return x*y - z
|
||||
}
|
||||
@ -89,6 +99,7 @@ func FusedSub32_b(x, y, z float32) float32 {
|
||||
|
||||
func FusedAdd64(x, y, z float64) float64 {
|
||||
// s390x:"FMADD\t"
|
||||
// ppc64:"FMADD\t"
|
||||
// ppc64le:"FMADD\t"
|
||||
// arm64:"FMADDD"
|
||||
return x*y + z
|
||||
@ -96,6 +107,7 @@ func FusedAdd64(x, y, z float64) float64 {
|
||||
|
||||
func FusedSub64_a(x, y, z float64) float64 {
|
||||
// s390x:"FMSUB\t"
|
||||
// ppc64:"FMSUB\t"
|
||||
// ppc64le:"FMSUB\t"
|
||||
return x*y - z
|
||||
}
|
||||
|
||||
@ -13,21 +13,25 @@ var sink64 [8]float64
|
||||
func approx(x float64) {
|
||||
// s390x:"FIDBR\t[$]6"
|
||||
// arm64:"FRINTPD"
|
||||
// ppc64:"FRIP"
|
||||
// ppc64le:"FRIP"
|
||||
sink64[0] = math.Ceil(x)
|
||||
|
||||
// s390x:"FIDBR\t[$]7"
|
||||
// arm64:"FRINTMD"
|
||||
// ppc64:"FRIM"
|
||||
// ppc64le:"FRIM"
|
||||
sink64[1] = math.Floor(x)
|
||||
|
||||
// s390x:"FIDBR\t[$]1"
|
||||
// arm64:"FRINTAD"
|
||||
// ppc64:"FRIN"
|
||||
// ppc64le:"FRIN"
|
||||
sink64[2] = math.Round(x)
|
||||
|
||||
// s390x:"FIDBR\t[$]5"
|
||||
// arm64:"FRINTZD"
|
||||
// ppc64:"FRIZ"
|
||||
// ppc64le:"FRIZ"
|
||||
sink64[3] = math.Trunc(x)
|
||||
|
||||
@ -51,11 +55,13 @@ func abs(x, y float64) {
|
||||
// amd64:"BTRQ\t[$]63"
|
||||
// arm64:"FABSD\t"
|
||||
// s390x:"LPDFR\t",-"MOVD\t" (no integer load/store)
|
||||
// ppc64:"FABS\t"
|
||||
// ppc64le:"FABS\t"
|
||||
sink64[0] = math.Abs(x)
|
||||
|
||||
// amd64:"BTRQ\t[$]63","PXOR" (TODO: this should be BTSQ)
|
||||
// s390x:"LNDFR\t",-"MOVD\t" (no integer load/store)
|
||||
// ppc64:"FNABS\t"
|
||||
// ppc64le:"FNABS\t"
|
||||
sink64[1] = -math.Abs(y)
|
||||
}
|
||||
@ -70,11 +76,13 @@ func abs32(x float32) float32 {
|
||||
func copysign(a, b, c float64) {
|
||||
// amd64:"BTRQ\t[$]63","SHRQ\t[$]63","SHLQ\t[$]63","ORQ"
|
||||
// s390x:"CPSDR",-"MOVD" (no integer load/store)
|
||||
// ppc64:"FCPSGN"
|
||||
// ppc64le:"FCPSGN"
|
||||
sink64[0] = math.Copysign(a, b)
|
||||
|
||||
// amd64:"BTSQ\t[$]63"
|
||||
// s390x:"LNDFR\t",-"MOVD\t" (no integer load/store)
|
||||
// ppc64:"FCPSGN"
|
||||
// ppc64le:"FCPSGN"
|
||||
// arm64:"ORR", -"AND"
|
||||
sink64[1] = math.Copysign(c, -1)
|
||||
@ -86,6 +94,7 @@ func copysign(a, b, c float64) {
|
||||
|
||||
// amd64:-"SHLQ\t[$]1",-"SHRQ\t[$]1","SHRQ\t[$]63","SHLQ\t[$]63","ORQ"
|
||||
// s390x:"CPSDR\t",-"MOVD\t" (no integer load/store)
|
||||
// ppc64:"FCPSGN"
|
||||
// ppc64le:"FCPSGN"
|
||||
sink64[3] = math.Copysign(-1, c)
|
||||
}
|
||||
@ -93,6 +102,8 @@ func copysign(a, b, c float64) {
|
||||
func fromFloat64(f64 float64) uint64 {
|
||||
// amd64:"MOVQ\tX.*, [^X].*"
|
||||
// arm64:"FMOVD\tF.*, R.*"
|
||||
// ppc64:"MFVSRD"
|
||||
// ppc64le:"MFVSRD"
|
||||
return math.Float64bits(f64+1) + 1
|
||||
}
|
||||
|
||||
@ -105,6 +116,8 @@ func fromFloat32(f32 float32) uint32 {
|
||||
func toFloat64(u64 uint64) float64 {
|
||||
// amd64:"MOVQ\t[^X].*, X.*"
|
||||
// arm64:"FMOVD\tR.*, F.*"
|
||||
// ppc64:"MTVSRD"
|
||||
// ppc64le:"MTVSRD"
|
||||
return math.Float64frombits(u64+1) + 1
|
||||
}
|
||||
|
||||
@ -135,6 +148,7 @@ func constantCheck32() bool {
|
||||
func constantConvert32(x float32) float32 {
|
||||
// amd64:"MOVSS\t[$]f32.3f800000\\(SB\\)"
|
||||
// s390x:"FMOVS\t[$]f32.3f800000\\(SB\\)"
|
||||
// ppc64:"FMOVS\t[$]f32.3f800000\\(SB\\)"
|
||||
// ppc64le:"FMOVS\t[$]f32.3f800000\\(SB\\)"
|
||||
// arm64:"FMOVS\t[$]\\(1.0\\)"
|
||||
if x > math.Float32frombits(0x3f800000) {
|
||||
@ -146,6 +160,7 @@ func constantConvert32(x float32) float32 {
|
||||
func constantConvertInt32(x uint32) uint32 {
|
||||
// amd64:-"MOVSS"
|
||||
// s390x:-"FMOVS"
|
||||
// ppc64:-"FMOVS"
|
||||
// ppc64le:-"FMOVS"
|
||||
// arm64:-"FMOVS"
|
||||
if x > math.Float32bits(1) {
|
||||
|
||||
@ -105,6 +105,7 @@ func OnesCount(n uint) int {
|
||||
// arm64:"VCNT","VUADDLV"
|
||||
// s390x:"POPCNT"
|
||||
// ppc64:"POPCNTD"
|
||||
// ppc64le:"POPCNTD"
|
||||
return bits.OnesCount(n)
|
||||
}
|
||||
|
||||
@ -113,6 +114,7 @@ func OnesCount64(n uint64) int {
|
||||
// arm64:"VCNT","VUADDLV"
|
||||
// s390x:"POPCNT"
|
||||
// ppc64:"POPCNTD"
|
||||
// ppc64le:"POPCNTD"
|
||||
return bits.OnesCount64(n)
|
||||
}
|
||||
|
||||
@ -121,6 +123,7 @@ func OnesCount32(n uint32) int {
|
||||
// arm64:"VCNT","VUADDLV"
|
||||
// s390x:"POPCNT"
|
||||
// ppc64:"POPCNTW"
|
||||
// ppc64le:"POPCNTW"
|
||||
return bits.OnesCount32(n)
|
||||
}
|
||||
|
||||
@ -129,12 +132,14 @@ func OnesCount16(n uint16) int {
|
||||
// arm64:"VCNT","VUADDLV"
|
||||
// s390x:"POPCNT"
|
||||
// ppc64:"POPCNTW"
|
||||
// ppc64le:"POPCNTW"
|
||||
return bits.OnesCount16(n)
|
||||
}
|
||||
|
||||
func OnesCount8(n uint8) int {
|
||||
// s390x:"POPCNT"
|
||||
// ppc64:"POPCNTB"
|
||||
// ppc64le:"POPCNTB"
|
||||
return bits.OnesCount8(n)
|
||||
}
|
||||
|
||||
@ -176,6 +181,7 @@ func RotateLeft64(n uint64) uint64 {
|
||||
// amd64:"ROLQ"
|
||||
// arm64:"ROR"
|
||||
// ppc64:"ROTL"
|
||||
// ppc64le:"ROTL"
|
||||
// s390x:"RLLG"
|
||||
return bits.RotateLeft64(n, 37)
|
||||
}
|
||||
@ -184,6 +190,7 @@ func RotateLeft32(n uint32) uint32 {
|
||||
// amd64:"ROLL" 386:"ROLL"
|
||||
// arm64:"RORW"
|
||||
// ppc64:"ROTLW"
|
||||
// ppc64le:"ROTLW"
|
||||
// s390x:"RLL"
|
||||
return bits.RotateLeft32(n, 9)
|
||||
}
|
||||
@ -202,6 +209,7 @@ func RotateLeftVariable(n uint, m int) uint {
|
||||
// amd64:"ROLQ"
|
||||
// arm64:"ROR"
|
||||
// ppc64:"ROTL"
|
||||
// ppc64le:"ROTL"
|
||||
// s390x:"RLLG"
|
||||
return bits.RotateLeft(n, m)
|
||||
}
|
||||
@ -210,6 +218,7 @@ func RotateLeftVariable64(n uint64, m int) uint64 {
|
||||
// amd64:"ROLQ"
|
||||
// arm64:"ROR"
|
||||
// ppc64:"ROTL"
|
||||
// ppc64le:"ROTL"
|
||||
// s390x:"RLLG"
|
||||
return bits.RotateLeft64(n, m)
|
||||
}
|
||||
@ -218,6 +227,7 @@ func RotateLeftVariable32(n uint32, m int) uint32 {
|
||||
// amd64:"ROLL"
|
||||
// arm64:"RORW"
|
||||
// ppc64:"ROTLW"
|
||||
// ppc64le:"ROTLW"
|
||||
// s390x:"RLL"
|
||||
return bits.RotateLeft32(n, m)
|
||||
}
|
||||
@ -230,6 +240,7 @@ func TrailingZeros(n uint) int {
|
||||
// amd64:"BSFQ","MOVL\t\\$64","CMOVQEQ"
|
||||
// s390x:"FLOGR"
|
||||
// ppc64:"ANDN","POPCNTD"
|
||||
// ppc64le:"ANDN","POPCNTD"
|
||||
return bits.TrailingZeros(n)
|
||||
}
|
||||
|
||||
@ -237,6 +248,7 @@ func TrailingZeros64(n uint64) int {
|
||||
// amd64:"BSFQ","MOVL\t\\$64","CMOVQEQ"
|
||||
// s390x:"FLOGR"
|
||||
// ppc64:"ANDN","POPCNTD"
|
||||
// ppc64le:"ANDN","POPCNTD"
|
||||
return bits.TrailingZeros64(n)
|
||||
}
|
||||
|
||||
@ -244,6 +256,7 @@ func TrailingZeros32(n uint32) int {
|
||||
// amd64:"BTSQ\\t\\$32","BSFQ"
|
||||
// s390x:"FLOGR","MOVWZ"
|
||||
// ppc64:"ANDN","POPCNTW"
|
||||
// ppc64le:"ANDN","POPCNTW"
|
||||
return bits.TrailingZeros32(n)
|
||||
}
|
||||
|
||||
@ -251,6 +264,7 @@ func TrailingZeros16(n uint16) int {
|
||||
// amd64:"BSFL","BTSL\\t\\$16"
|
||||
// s390x:"FLOGR","OR\t\\$65536"
|
||||
// ppc64:"POPCNTD","OR\\t\\$65536"
|
||||
// ppc64le:"POPCNTD","OR\\t\\$65536"
|
||||
return bits.TrailingZeros16(n)
|
||||
}
|
||||
|
||||
@ -319,13 +333,15 @@ func IterateBits8(n uint8) int {
|
||||
func Mul(x, y uint) (hi, lo uint) {
|
||||
// amd64:"MULQ"
|
||||
// arm64:"UMULH","MUL"
|
||||
// ppc64: "MULHDU", "MULLD"
|
||||
// ppc64:"MULHDU","MULLD"
|
||||
// ppc64le:"MULHDU","MULLD"
|
||||
return bits.Mul(x, y)
|
||||
}
|
||||
|
||||
func Mul64(x, y uint64) (hi, lo uint64) {
|
||||
// amd64:"MULQ"
|
||||
// arm64:"UMULH","MUL"
|
||||
// ppc64: "MULHDU", "MULLD"
|
||||
// ppc64:"MULHDU","MULLD"
|
||||
// ppc64le:"MULHDU","MULLD"
|
||||
return bits.Mul64(x, y)
|
||||
}
|
||||
|
||||
@ -21,30 +21,38 @@ var val8 [8]uint8
|
||||
|
||||
func set16(x8 int8, u8 uint8, y8 int8, z8 uint8) {
|
||||
// Truncate not needed, load does sign/zero extend
|
||||
// ppc64:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
sval16[0] = int16(x8)
|
||||
|
||||
// ppc64:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
val16[0] = uint16(u8)
|
||||
|
||||
// AND not needed due to size
|
||||
// ppc64:-"ANDCC"
|
||||
// ppc64le:-"ANDCC"
|
||||
sval16[1] = 255 & int16(x8+y8)
|
||||
|
||||
// ppc64:-"ANDCC"
|
||||
// ppc64le:-"ANDCC"
|
||||
val16[1] = 255 & uint16(u8+z8)
|
||||
|
||||
}
|
||||
func shiftidx(x8 int8, u8 uint8, x16 int16, u16 uint16, x32 int32, u32 uint32) {
|
||||
// ppc64:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
sval16[0] = int16(val16[x8>>1])
|
||||
|
||||
// ppc64:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
val16[0] = uint16(sval16[u8>>2])
|
||||
|
||||
// ppc64:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
sval16[1] = int16(val16[x16>>1])
|
||||
|
||||
// ppc64:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
val16[1] = uint16(sval16[u16>>2])
|
||||
|
||||
@ -53,87 +61,109 @@ func shiftidx(x8 int8, u8 uint8, x16 int16, u16 uint16, x32 int32, u32 uint32) {
|
||||
func setnox(x8 int8, u8 uint8, y8 int8, z8 uint8, x16 int16, u16 uint16, x32 int32, u32 uint32) {
|
||||
// Truncate not needed due to sign/zero extension on load
|
||||
|
||||
// ppc64:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
sval16[0] = int16(x8)
|
||||
|
||||
// ppc64:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
val16[0] = uint16(u8)
|
||||
|
||||
// AND not needed due to size
|
||||
// ppc64:-"ANDCC"
|
||||
// ppc64le:-"ANDCC"
|
||||
sval16[1] = 255 & int16(x8+y8)
|
||||
|
||||
// ppc64:-"ANDCC"
|
||||
// ppc64le:-"ANDCC"
|
||||
val16[1] = 255 & uint16(u8+z8)
|
||||
|
||||
// ppc64:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
sval32[0] = int32(x8)
|
||||
|
||||
// ppc64:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
sval32[1] = int32(x16)
|
||||
|
||||
//ppc64:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
//ppc64le:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
val32[0] = uint32(u8)
|
||||
|
||||
// ppc64:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
val32[1] = uint32(u16)
|
||||
|
||||
// ppc64:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
sval64[0] = int64(x8)
|
||||
|
||||
// ppc64:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
sval64[1] = int64(x16)
|
||||
|
||||
// ppc64:-"MOVW\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVW\tR\\d+,\\sR\\d+"
|
||||
sval64[2] = int64(x32)
|
||||
|
||||
//ppc64:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
//ppc64le:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
val64[0] = uint64(u8)
|
||||
|
||||
// ppc64:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
val64[1] = uint64(u16)
|
||||
|
||||
// ppc64:-"MOVWZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVWZ\tR\\d+,\\sR\\d+"
|
||||
val64[2] = uint64(u32)
|
||||
}
|
||||
|
||||
func cmp16(x8 int8, u8 uint8, x32 int32, u32 uint32, x64 int64, u64 uint64) bool {
|
||||
// ppc64:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
if int16(x8) == sval16[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
if uint16(u8) == val16[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
if uint16(u32>>16) == val16[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
if uint16(u64>>48) == val16[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// Verify the truncates are using the correct sign.
|
||||
// ppc64:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
if int16(x32) == sval16[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
if uint16(u32) == val16[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
if int16(x64) == sval16[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
if uint16(u64) == val16[0] {
|
||||
return true
|
||||
@ -143,32 +173,38 @@ func cmp16(x8 int8, u8 uint8, x32 int32, u32 uint32, x64 int64, u64 uint64) bool
|
||||
}
|
||||
|
||||
func cmp32(x8 int8, u8 uint8, x16 int16, u16 uint16, x64 int64, u64 uint64) bool {
|
||||
// ppc64:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
if int32(x8) == sval32[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
if uint32(u8) == val32[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
if int32(x16) == sval32[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
if uint32(u16) == val32[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// Verify the truncates are using the correct sign.
|
||||
// ppc64:-"MOVWZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVWZ\tR\\d+,\\sR\\d+"
|
||||
if int32(x64) == sval32[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVW\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVW\tR\\d+,\\sR\\d+"
|
||||
if uint32(u64) == val32[0] {
|
||||
return true
|
||||
@ -179,31 +215,37 @@ func cmp32(x8 int8, u8 uint8, x16 int16, u16 uint16, x64 int64, u64 uint64) bool
|
||||
|
||||
|
||||
func cmp64(x8 int8, u8 uint8, x16 int16, u16 uint16, x32 int32, u32 uint32) bool {
|
||||
// ppc64:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVB\tR\\d+,\\sR\\d+"
|
||||
if int64(x8) == sval64[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVBZ\tR\\d+,\\sR\\d+"
|
||||
if uint64(u8) == val64[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVH\tR\\d+,\\sR\\d+"
|
||||
if int64(x16) == sval64[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVHZ\tR\\d+,\\sR\\d+"
|
||||
if uint64(u16) == val64[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVW\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVW\tR\\d+,\\sR\\d+"
|
||||
if int64(x32) == sval64[0] {
|
||||
return true
|
||||
}
|
||||
|
||||
// ppc64:-"MOVWZ\tR\\d+,\\sR\\d+"
|
||||
// ppc64le:-"MOVWZ\tR\\d+,\\sR\\d+"
|
||||
if uint64(u32) == val64[0] {
|
||||
return true
|
||||
|
||||
@ -16,18 +16,21 @@ func rot64(x uint64) uint64 {
|
||||
// amd64:"ROLQ\t[$]7"
|
||||
// arm64:"ROR\t[$]57"
|
||||
// s390x:"RLLG\t[$]7"
|
||||
// ppc64:"ROTL\t[$]7"
|
||||
// ppc64le:"ROTL\t[$]7"
|
||||
a += x<<7 | x>>57
|
||||
|
||||
// amd64:"ROLQ\t[$]8"
|
||||
// arm64:"ROR\t[$]56"
|
||||
// s390x:"RLLG\t[$]8"
|
||||
// ppc64:"ROTL\t[$]8"
|
||||
// ppc64le:"ROTL\t[$]8"
|
||||
a += x<<8 + x>>56
|
||||
|
||||
// amd64:"ROLQ\t[$]9"
|
||||
// arm64:"ROR\t[$]55"
|
||||
// s390x:"RLLG\t[$]9"
|
||||
// ppc64:"ROTL\t[$]9"
|
||||
// ppc64le:"ROTL\t[$]9"
|
||||
a += x<<9 ^ x>>55
|
||||
|
||||
@ -41,6 +44,7 @@ func rot32(x uint32) uint32 {
|
||||
// arm:"MOVW\tR\\d+@>25"
|
||||
// arm64:"RORW\t[$]25"
|
||||
// s390x:"RLL\t[$]7"
|
||||
// ppc64:"ROTLW\t[$]7"
|
||||
// ppc64le:"ROTLW\t[$]7"
|
||||
a += x<<7 | x>>25
|
||||
|
||||
@ -48,6 +52,7 @@ func rot32(x uint32) uint32 {
|
||||
// arm:"MOVW\tR\\d+@>24"
|
||||
// arm64:"RORW\t[$]24"
|
||||
// s390x:"RLL\t[$]8"
|
||||
// ppc64:"ROTLW\t[$]8"
|
||||
// ppc64le:"ROTLW\t[$]8"
|
||||
a += x<<8 + x>>24
|
||||
|
||||
@ -55,6 +60,7 @@ func rot32(x uint32) uint32 {
|
||||
// arm:"MOVW\tR\\d+@>23"
|
||||
// arm64:"RORW\t[$]23"
|
||||
// s390x:"RLL\t[$]9"
|
||||
// ppc64:"ROTLW\t[$]9"
|
||||
// ppc64le:"ROTLW\t[$]9"
|
||||
a += x<<9 ^ x>>23
|
||||
|
||||
@ -101,6 +107,7 @@ func rot64nc(x uint64, z uint) uint64 {
|
||||
z &= 63
|
||||
|
||||
// amd64:"ROLQ"
|
||||
// ppc64:"ROTL"
|
||||
// ppc64le:"ROTL"
|
||||
a += x<<z | x>>(64-z)
|
||||
|
||||
@ -116,6 +123,7 @@ func rot32nc(x uint32, z uint) uint32 {
|
||||
z &= 31
|
||||
|
||||
// amd64:"ROLL"
|
||||
// ppc64:"ROTLW"
|
||||
// ppc64le:"ROTLW"
|
||||
a += x<<z | x>>(32-z)
|
||||
|
||||
|
||||
@ -18,6 +18,7 @@ import "runtime"
|
||||
// arm:"TEXT\t.*, [$]-4-"
|
||||
// arm64:"TEXT\t.*, [$]0-"
|
||||
// mips:"TEXT\t.*, [$]-4-"
|
||||
// ppc64:"TEXT\t.*, [$]0-"
|
||||
// ppc64le:"TEXT\t.*, [$]0-"
|
||||
// s390x:"TEXT\t.*, [$]0-"
|
||||
func StackStore() int {
|
||||
@ -37,6 +38,7 @@ type T struct {
|
||||
// arm:"TEXT\t.*, [$]0-" (spills return address)
|
||||
// arm64:"TEXT\t.*, [$]0-"
|
||||
// mips:"TEXT\t.*, [$]-4-"
|
||||
// ppc64:"TEXT\t.*, [$]0-"
|
||||
// ppc64le:"TEXT\t.*, [$]0-"
|
||||
// s390x:"TEXT\t.*, [$]0-"
|
||||
func ZeroLargeStruct(x *T) {
|
||||
@ -51,6 +53,7 @@ func ZeroLargeStruct(x *T) {
|
||||
// amd64:"TEXT\t.*, [$]0-"
|
||||
// arm:"TEXT\t.*, [$]0-" (spills return address)
|
||||
// arm64:"TEXT\t.*, [$]0-"
|
||||
// ppc64:"TEXT\t.*, [$]0-"
|
||||
// ppc64le:"TEXT\t.*, [$]0-"
|
||||
// s390x:"TEXT\t.*, [$]0-"
|
||||
// Note: that 386 currently has to spill a register.
|
||||
@ -65,6 +68,7 @@ func KeepWanted(t *T) {
|
||||
// - arm & mips fail due to softfloat calls
|
||||
// amd64:"TEXT\t.*, [$]0-"
|
||||
// arm64:"TEXT\t.*, [$]0-"
|
||||
// ppc64:"TEXT\t.*, [$]0-"
|
||||
// ppc64le:"TEXT\t.*, [$]0-"
|
||||
// s390x:"TEXT\t.*, [$]0-"
|
||||
func ArrayAdd64(a, b [4]float64) [4]float64 {
|
||||
@ -78,6 +82,7 @@ func ArrayAdd64(a, b [4]float64) [4]float64 {
|
||||
// arm:"TEXT\t.*, [$]0-" (spills return address)
|
||||
// arm64:"TEXT\t.*, [$]0-"
|
||||
// mips:"TEXT\t.*, [$]-4-"
|
||||
// ppc64:"TEXT\t.*, [$]0-"
|
||||
// ppc64le:"TEXT\t.*, [$]0-"
|
||||
// s390x:"TEXT\t.*, [$]0-"
|
||||
func ArrayInit(i, j int) [4]int {
|
||||
|
||||
Loading…
Reference in New Issue
Block a user