cmd/compile,cmd/internal/obj/riscv,runtime: use Duff's devices on riscv64

Implement runtime.duffzero and runtime.duffcopy for riscv64.
Use obj.ADUFFZERO/obj.ADUFFCOPY for medium size, word aligned
zeroing/moving.

Change-Id: I42ec622055630c94cb77e286d8d33dbe7c9f846c
Reviewed-on: https://go-review.googlesource.com/c/go/+/237797
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Zhang <cherryyz@google.com>

1 files changed, 28 insertions, 0 deletions

diff --git a/src/runtime/mkduff.go b/src/runtime/mkduff.go
index 8859ed68cc..6ddf0256e9 100644
--- a/src/runtime/mkduff.go
+++ b/src/runtime/mkduff.go
@@ -38,6 +38,7 @@ func main() {
 	gen("arm64", notags, zeroARM64, copyARM64)
 	gen("ppc64x", tagsPPC64x, zeroPPC64x, copyPPC64x)
 	gen("mips64x", tagsMIPS64x, zeroMIPS64x, copyMIPS64x)
+	gen("riscv64", notags, zeroRISCV64, copyRISCV64)
 }
 
 func gen(arch string, tags, zero, copy func(io.Writer)) {
@@ -227,3 +228,30 @@ func copyMIPS64x(w io.Writer) {
 	}
 	fmt.Fprintln(w, "\tRET")
 }
+
+func zeroRISCV64(w io.Writer) {
+	// ZERO: always zero
+	// X10: ptr to memory to be zeroed
+	// X10 is updated as a side effect.
+	fmt.Fprintln(w, "TEXT runtime·duffzero(SB), NOSPLIT|NOFRAME, $0-0")
+	for i := 0; i < 128; i++ {
+		fmt.Fprintln(w, "\tMOV\tZERO, (X10)")
+		fmt.Fprintln(w, "\tADD\t$8, X10")
+	}
+	fmt.Fprintln(w, "\tRET")
+}
+
+func copyRISCV64(w io.Writer) {
+	// X10: ptr to source memory
+	// X11: ptr to destination memory
+	// X10 and X11 are updated as a side effect
+	fmt.Fprintln(w, "TEXT runtime·duffcopy(SB), NOSPLIT|NOFRAME, $0-0")
+	for i := 0; i < 128; i++ {
+		fmt.Fprintln(w, "\tMOV\t(X10), X31")
+		fmt.Fprintln(w, "\tADD\t$8, X10")
+		fmt.Fprintln(w, "\tMOV\tX31, (X11)")
+		fmt.Fprintln(w, "\tADD\t$8, X11")
+		fmt.Fprintln(w)
+	}
+	fmt.Fprintln(w, "\tRET")
+}