Just a guess: Maybe the escape analysis cannot prove that it does not escape the function?
The unsafe pointer probably makes it nearly impossible for the compiler to tell. Perhaps? The pointer does get passed to a.Encrypt(), which could store it beyond the lifetime of the Precomp() call. I cannot tell myself.. https://appliedgo.com/blog/how-to-do-escape-analysis On Tuesday, October 31, 2023 at 3:02:48 PM UTC Nugraha wrote: > Hi, I was wondering why is `var f [d * 4]byte` allocated with > `runtime.newobject`, as we can see the size is already known at compile > time because `d` is a constant value, by doing some propagation we can > end up having `[16]byte`. > > TIA > > > 42 . . } > > 43 . . } > > 44 . . > > 45 . . type ffx struct{} > > 46 . . > > 47 2MB 2MB func (ffx) PrecompF(a cipher.Block, h int, g string, e int) > (res [16]byte) { > > 48 . . const d = 4 > > 49 1MB 1MB var f [d * 4]byte > > . . 5aa4e0: LEAQ 0x1ec59(IP), AX ffx.go:49 > > 1MB 1MB 5aa4e7: CALL runtime.newobject(SB) ffx.go:49 > > . . 5aa4ec: MOVQ AX, 0x60(SP) ffx.go:49 > > > > 50 . . fv := (*[d]uint32)(unsafe.Pointer(&f[0])) > > 51 . . c := len(g) > > 52 . . b := 10 > > 53 . . > > 54 . . fv[0] |= 0x01_02_01_00 > > 55 . . fv[0] |= uint32((e >> 16) & 255) > > 56 . . > > 57 . . fv[1] |= uint32((e >> 8) & 255 << 24) > > 58 . . fv[1] |= uint32((e & 255) << 16) > > 59 . . fv[1] |= uint32(b << 8) > > 60 . . fv[1] |= uint32((h / 2) & 255) > > 61 . . > > 62 . . fv[2] = uint32(h) > > 63 . . fv[3] = uint32(c) > > 64 . . > > 65 . . endianSwaps(fv) > > 66 . . a.Encrypt(res[:], f[:]) > > 67 . . return > > 68 . . } > > 69 . . > > 70 . . func (ffx) Precompb(c int, g int) int { > > 71 . . var e int64 = int64(g) / 2 > > 72 . . var a int = 0 > > -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/e92b9227-ea08-4a02-9004-8bcf6f561f65n%40googlegroups.com.
