cmd/compile: store combine on amd64

Fixes golang#54120

src/cmd/compile/internal/ssa/gen/AMD64.rules

@@ -2043,15 +2043,18 @@
   && clobber(x)
   => (MOVQstore [i] {s} p0 w0 mem)
 
-(MOVBstore [7] {s} p1 (SHRQconst [56] w)
-  x1:(MOVWstore [5] {s} p1 (SHRQconst [40] w)
-  x2:(MOVLstore [1] {s} p1 (SHRQconst [8] w)
-  x3:(MOVBstore [0] {s} p1 w mem))))
+(MOVBstore  [c3] {s}  p3 (SHRQconst [56]  w)
+  x1:(MOVWstore [c2]  {s} p2  (SHRQconst [40] w)
+  x2:(MOVLstore [c1] {s} p1  (SHRQconst [8] w)
+  x3:(MOVBstore [c0] {s} p0  w mem))))
   && x1.Uses == 1
   && x2.Uses == 1
   && x3.Uses == 1
+  && sequentialAddresses(p0, p1, int64(1 + c0 - c1))
+  && sequentialAddresses(p0, p2, int64(5 + c0 - c2))
+  && sequentialAddresses(p0, p3, int64(7 + c0 - c3))
   && clobber(x1, x2, x3)
-  => (MOVQstore {s} p1 w mem)
+  => (MOVQstore [c0] {s} p0 w mem)
 
 (MOVBstore [i] {s} p
   x1:(MOVBload [j] {s2} p2 mem)

src/cmd/compile/internal/ssa/rewrite.go

@@ -1764,6 +1764,9 @@ func read64(sym interface{}, off int64, byteorder binary.ByteOrder) uint64 {
 
 // sequentialAddresses reports true if it can prove that x + n == y
 func sequentialAddresses(x, y *Value, n int64) bool {
+	if x == y && n == 0 {
+		return true
+	}
 	if x.Op == Op386ADDL && y.Op == Op386LEAL1 && y.AuxInt == n && y.Aux == nil &&
 		(x.Args[0] == y.Args[0] && x.Args[1] == y.Args[1] ||
 			x.Args[0] == y.Args[1] && x.Args[1] == y.Args[0]) {

test/codegen/memcombine.go

@@ -369,6 +369,14 @@ func store_le64_idx(b []byte, x uint64, idx int) {
 	binary.LittleEndian.PutUint64(b[idx:], x)
 }
 
+func store_le64_idx2(dst []byte, d, length, offset int) []byte {
+	a := dst[d : d+length]
+	b := dst[d-offset:]
+	// amd64:`MOVQ\s.*\(.*\)\(.*\*1\)$`,-`SHR.`
+	binary.LittleEndian.PutUint64(a, binary.LittleEndian.Uint64(b))
+	return dst
+}
+
 func store_le64_load(b []byte, x *[8]byte) {
 	_ = b[8]
 	// amd64:-`MOV[BWL]`