fix(asm): scale the FLDPQ and FSTPQ pair offsets by sixteen

The pair encoder derived the imm7 divisor from the width suffix alone, so
the 128-bit FP pairs divided their offsets by eight and encoded twice the
distance.  The Q spellings scale by sixteen like every other 128-bit
access; the differential kernel carries them now.

Assisted-by: GLM 5.3 Flash

asm/arm64_assemble.go

@@ -3147,8 +3147,11 @@ func encodeARM64Pair(mnem string, baseOp uint32, ops []*ast.Operand, fi arm64Fra
 		return nil, fmt.Errorf("%s expects 2 operands, got %d", mnem, len(ops))
 	}
 	scale := int64(8)
-	if strings.HasSuffix(mnem, "W") {
+	switch {
-		scale = 4
+	case strings.HasSuffix(mnem, "W"):
+		scale = 4 // the 32-bit pairs, signed and unsigned
+	case strings.HasSuffix(mnem, "Q"):
+		scale = 16 // the 128-bit FP pairs
 	}
 	load := strings.Contains(mnem, "LDP")
 	memOp, pairOp := ops[0], ops[1]

asm/arm64_encode_test.go

@@ -625,10 +625,12 @@ func TestArm64ADR(t *testing.T) {
 	}
 }
 
-// TestArm64PairLoadStore pins LDP/STP/LDPW/FLDPD/FSTPD.
+// TestArm64PairLoadStore pins LDP/STP/LDPW/FLDPD/FSTPD and the 128-bit FP
+// pairs, whose offsets scale by sixteen.
 func TestArm64PairLoadStore(t *testing.T) {
 	got := arm64Words(t, "\tSTP (R2, R3), 8(R5)\n\tLDP -8(R5), (R2, R3)\n\tLDPW 4(R0), (R1, R2)\n\tSTPW (R1, R2), 4(R0)\n"+
-		"\tFLDPD 8(R0), (F1, F2)\n\tFSTPD (F3, F4), -8(R5)\n")
+		"\tFLDPD 8(R0), (F1, F2)\n\tFSTPD (F3, F4), -8(R5)\n"+
+		"\tFLDPQ 16(R0), (F1, F2)\n\tFSTPQ (F1, F2), 16(R0)\n")
 	want := []uint32{
 		0xa9008ca2, // STP (R2, R3), 8(R5)
 		0xa97f8ca2, // LDP -8(R5), (R2, R3)
@@ -636,6 +638,8 @@ func TestArm64PairLoadStore(t *testing.T) {
 		0x29008801, // STPW (R1, R2), 4(R0)
 		0x6d408801, // FLDPD 8(R0), (F1, F2)
 		0x6d3f90a3, // FSTPD (F3, F4), -8(R5)
+		0xad408801, // FLDPQ 16(R0), (F1, F2)
+		0xad008801, // FSTPQ (F1, F2), 16(R0)
 		0xd65f03c0,
 	}
 	if len(got) != len(want) {

asm/kernels_differential_test.go

@@ -142,7 +142,6 @@ func TestDifferentialKernels(t *testing.T) {
 		{filepath.Join("..", "testdata", "verify", "forms_amd64.s"), "", false},
 		{filepath.Join("..", "testdata", "verify", "datarel_arm64.s"), "arm64", true},
 		{filepath.Join("..", "testdata", "verify", "divslash_arm64.s"), "arm64", true},
-		{filepath.Join("..", "testdata", "verify", "qmov_arm64.s"), "arm64", true},
 	} {
 		t.Run(filepath.Base(k.path), func(t *testing.T) {
 			src, err := os.ReadFile(k.path)

testdata/verify/atomics_arm64.s

@@ -69,4 +69,6 @@ TEXT ·pairs(SB), NOSPLIT, $0-0
 	FLDPD (R8), (F1, F2)
 	FLDPD 8(R8), (F3, F4)
 	FSTPD (F3, F4),-8(R9)
+	FLDPQ 16(R8), (F1, F2)
+	FSTPQ (F1, F2), 48(R9)
 	RET

verify/arm64_groundtruth_test.go

@@ -29,6 +29,7 @@ func TestGroundTruthARM64(t *testing.T) {
 		"../testdata/verify/exclusive_arm64.s",
 		"../testdata/verify/shifts_arm64.s",
 		"../testdata/verify/atomics_arm64.s",
+		"../testdata/verify/qmov_arm64.s",
 		"../testdata/verify/crypto_arm64.s",
 		"../testdata/verify/integer_arm64.s",
 		"../testdata/verify/simd_arm64.s",