internal/cpu: align capability definitions for x86 with other architectures

Use constant masks and align the definition of isSet with
arm64 and ppc64x.

Change-Id: I0c6eae30da5e3ce797cde0dab4a39855d4d245d9
Reviewed-on: https://go-review.googlesource.com/94759
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

diff --git a/src/internal/cpu/cpu_x86.go b/src/internal/cpu/cpu_x86.go
index 239e7289008548aac2c00029c09f7ac519a6c9ee..17be6eed26c80cceccf37f45c5e8b37c99efa59a 100644 (file)
--- a/src/internal/cpu/cpu_x86.go
+++ b/src/internal/cpu/cpu_x86.go
@@ -14,6 +14,30 @@ func cpuid(eaxArg, ecxArg uint32) (eax, ebx, ecx, edx uint32)
 // xgetbv with ecx = 0 is implemented in cpu_x86.s.
 func xgetbv() (eax, edx uint32)
 
+const (
+       // edx bits
+       cpuid_SSE2 = 1 << 26
+
+       // ecx bits
+       cpuid_SSE3      = 1 << 0
+       cpuid_PCLMULQDQ = 1 << 1
+       cpuid_SSSE3     = 1 << 9
+       cpuid_FMA       = 1 << 12
+       cpuid_SSE41     = 1 << 19
+       cpuid_SSE42     = 1 << 20
+       cpuid_POPCNT    = 1 << 23
+       cpuid_AES       = 1 << 25
+       cpuid_OSXSAVE   = 1 << 27
+       cpuid_AVX       = 1 << 28
+
+       // ebx bits
+       cpuid_BMI1 = 1 << 3
+       cpuid_AVX2 = 1 << 5
+       cpuid_BMI2 = 1 << 8
+       cpuid_ERMS = 1 << 9
+       cpuid_ADX  = 1 << 19
+)
+
 func doinit() {
        maxID, _, _, _ := cpuid(0, 0)
 
@@ -22,40 +46,40 @@ func doinit() {
        }
 
        _, _, ecx1, edx1 := cpuid(1, 0)
-       X86.HasSSE2 = isSet(26, edx1)
-
-       X86.HasSSE3 = isSet(0, ecx1)
-       X86.HasPCLMULQDQ = isSet(1, ecx1)
-       X86.HasSSSE3 = isSet(9, ecx1)
-       X86.HasFMA = isSet(12, ecx1)
-       X86.HasSSE41 = isSet(19, ecx1)
-       X86.HasSSE42 = isSet(20, ecx1)
-       X86.HasPOPCNT = isSet(23, ecx1)
-       X86.HasAES = isSet(25, ecx1)
-       X86.HasOSXSAVE = isSet(27, ecx1)
+       X86.HasSSE2 = isSet(edx1, cpuid_SSE2)
+
+       X86.HasSSE3 = isSet(ecx1, cpuid_SSE3)
+       X86.HasPCLMULQDQ = isSet(ecx1, cpuid_PCLMULQDQ)
+       X86.HasSSSE3 = isSet(ecx1, cpuid_SSSE3)
+       X86.HasFMA = isSet(ecx1, cpuid_FMA)
+       X86.HasSSE41 = isSet(ecx1, cpuid_SSE41)
+       X86.HasSSE42 = isSet(ecx1, cpuid_SSE42)
+       X86.HasPOPCNT = isSet(ecx1, cpuid_POPCNT)
+       X86.HasAES = isSet(ecx1, cpuid_AES)
+       X86.HasOSXSAVE = isSet(ecx1, cpuid_OSXSAVE)
 
        osSupportsAVX := false
        // For XGETBV, OSXSAVE bit is required and sufficient.
        if X86.HasOSXSAVE {
                eax, _ := xgetbv()
                // Check if XMM and YMM registers have OS support.
-               osSupportsAVX = isSet(1, eax) && isSet(2, eax)
+               osSupportsAVX = isSet(eax, 1<<1) && isSet(eax, 1<<2)
        }
 
-       X86.HasAVX = isSet(28, ecx1) && osSupportsAVX
+       X86.HasAVX = isSet(ecx1, cpuid_AVX) && osSupportsAVX
 
        if maxID < 7 {
                return
        }
 
        _, ebx7, _, _ := cpuid(7, 0)
-       X86.HasBMI1 = isSet(3, ebx7)
-       X86.HasAVX2 = isSet(5, ebx7) && osSupportsAVX
-       X86.HasBMI2 = isSet(8, ebx7)
-       X86.HasERMS = isSet(9, ebx7)
-       X86.HasADX = isSet(19, ebx7)
+       X86.HasBMI1 = isSet(ebx7, cpuid_BMI1)
+       X86.HasAVX2 = isSet(ebx7, cpuid_AVX2) && osSupportsAVX
+       X86.HasBMI2 = isSet(ebx7, cpuid_BMI2)
+       X86.HasERMS = isSet(ebx7, cpuid_ERMS)
+       X86.HasADX = isSet(ebx7, cpuid_ADX)
 }
 
-func isSet(bitpos uint, value uint32) bool {
-       return value&(1<<bitpos) != 0
+func isSet(hwc uint32, value uint32) bool {
+       return hwc&value != 0
 }