#include <iostream>
#include <algorithm>
#include <array>
#include <bitset>
#include <map>
#include <numeric>
#include <queue>
#include <set>
#include <string>
#include <tuple>
#include <bit>
#include <chrono>
#include <functional>
#include <iomanip>
#include <utility>
#include <type_traits>
#include <cassert>
#include <cctype>
#include <cmath>
#include <cstring>
#include <ctime>
#include <limits>
#include <ranges>
#include <concepts>

#define TE template <typename T>
#define TES template <typename T, typename ...S>
#define Z auto
#define ep emplace_back
#define eb emplace
#define fi first
#define se second
#define bg begin
#define ed end
#define all(x) bg(x), ed(x)

#define ov(a, b, c, d, e, ...) e
#define FO1(a) for (int _ = 0; _ < (a); ++_)
#define FO2(i, a) for (int i = 0; i < (a); ++i)
#define FO3(i, a, b) for (int i = (a); i < (b); ++i)
#define FO4(i, a, b, c) for (int i = (a); i < (b); i += (c))
#define FOR(...) ov(__VA_ARGS__, FO4, FO3, FO2, FO1)(__VA_ARGS__)
#define FF1(a) for (int _ = (a) - 1; _ >= 0; --_)
#define FF2(i, a) for (int i = (a) - 1; i >= 0; --i)
#define FF3(i, a, b) for (int i = (b) - 1; i >= (a); --i)
#define FF4(i, a, b, c) for (int i = (b) - 1; i >= (a); i -= (c))
#define FOR_R(...) ov(__VA_ARGS__, FF4, FF3, FF2, FF1)(__VA_ARGS__)
#define FOR_subset(t, s) for (int t = (s); t > -1; t = (t == 0 ? -1 : (t - 1) & s))

#define sort ranges::sort

using namespace std;

TE using vc = vector<T>;
TE using vvc = vc<vc<T>>;
TE using T1 = tuple<T>;
TE using T2 = tuple<T, T>;
TE using T3 = tuple<T, T, T>;
TE using T4 = tuple<T, T, T, T>;
TE using max_heap = priority_queue<T>;
TE using min_heap = priority_queue<T, vc<T>, greater<T>>;
using u8 = unsigned char; using uint = unsigned int; using ll = long long;      using ull = unsigned long long;
using ld = long double;   using i128 = __int128;     using u128 = __uint128_t;  using f128 = __float128;
using u16 = uint16_t;
using PII = pair<int, int>;   using PLL = pair<ll, ll>;

#ifdef YRSD
constexpr bool dbg = 1;
#else
constexpr bool dbg = 0;
#endif

istream &operator>>(istream &I, i128 &x) {
  static string s;
  I >> s;
  int f = s[0] == '-';
  x = 0;
  const int N = (int)s.size();
  FOR(i, f, N) x = x * 10 + s[i] - '0';
  if (f) x = -x;
  return I;
}
ostream &operator<<(ostream &O, i128 x) {
  static string s;
  s.clear();
  bool f = x < 0;
  if (f) x = -x;
  while (x) s += '0' + x % 10, x /= 10;
  if (s.empty()) s += '0';
  if (f) s += '-';
  reverse(all(s));
  return O << s;
}
istream &operator>>(istream &I, f128 &x) {
  static string s;
  I >> s, x = stold(s);
  return I;
}
ostream &operator<<(ostream &O, const f128 x) { return O << ld(x); }
template <typename... S>
istream &operator>>(istream &I, tuple<S...> &t) {
  return apply([&I](Z &...s) { ((I >> s), ...); }, t), I;
}
template <typename T, typename U>
istream &operator>>(istream &I, pair<T, U> &x) {
  return I >> x.fi >> x.se;
}
template <typename T, typename U>
ostream &operator<<(ostream &O, const pair<T, U> &x) {
  return O << x.fi << ' ' << x.se;
}
TE requires requires(T &c) { begin(c); end(c); } and 
                          (not is_same_v<decay_t<T>, string>)
istream &operator>>(istream &I, T &c) {
  for (Z &e : c) I >> e;
  return I;
}
TE requires requires(const T &c) { begin(c); end(c); } and 
  (not is_same_v<decay_t<T>, const char*>) and 
  (not is_same_v<decay_t<T>, string>) and 
  (not is_array_v<remove_reference_t<T>> or 
   not is_same_v<remove_extent_t<remove_reference_t<T>>, char>)
ostream &operator<<(ostream &O, const T &a) {
  if (a.empty()) return O;
  Z i = a.begin();
  O << *i++;
  for (; i != a.end(); ++i) O << ' ' << *i;
  return O;
}
void IN() {}
TE void IN(T &x, Z &...s) { cin >> x, IN(s...); }
void print() { cout << '\n'; }
TES void print(T &&x, S &&...y) {
  cout << x;
  if constexpr (sizeof...(S)) cout << ' ';
  print(forward<S>(y)...);
}
void put() {}
TES void put(T &&x, S &&...y) {
  cout << x;
  put(forward<S>(y)...);
}

#define INT(...)  int    __VA_ARGS__; IN(__VA_ARGS__)
#define UINT(...) uint   __VA_ARGS__; IN(__VA_ARGS__)
#define LL(...)   ll     __VA_ARGS__; IN(__VA_ARGS__)
#define ULL(...)  ull    __VA_ARGS__; IN(__VA_ARGS__)
#define I128(...) i128   __VA_ARGS__; IN(__VA_ARGS__)
#define STR(...)  string __VA_ARGS__; IN(__VA_ARGS__)
#define CH(...)   char   __VA_ARGS__; IN(__VA_ARGS__)
#define REAL(...) re     __VA_ARGS__; IN(__VA_ARGS__)
#define VEC(T, a, n) vc<T> a(n); IN(a)

void YES(bool o = 1) { print(o ? "YES" : "NO"); }
void Yes(bool o = 1) { print(o ? "Yes" : "No"); }
void yes(bool o = 1) { print(o ? "yes" : "no"); }
void NO(bool o = 1) { YES(not o); }
void No(bool o = 1) { Yes(not o); }
void no(bool o = 1) { yes(not o); }
void ALICE(bool o = 1) { print(o ? "ALICE" : "BOB"); }
void Alice(bool o = 1) { print(o ? "Alice" : "Bob"); }
void alice(bool o = 1) { print(o ? "alice" : "bob"); }
void BOB(bool o = 1) { ALICE(not o); }
void Bob(bool o = 1) { Alice(not o); }
void bob(bool o = 1) { alice(not o); }
void POSSIBLE(bool o = 1) { print(o ? "POSSIBLE" : "IMPOSSIBLE"); }
void Possible(bool o = 1) { print(o ? "Possible" : "Impossible"); }
void possible(bool o = 1) { print(o ? "possible" : "impossible"); }
void IMPOSSIBLE(bool o = 1) { POSSIBLE(not o); }
void Impossible(bool o = 1) { Possible(not o); }
void impossible(bool o = 1) { possible(not o); }
void TAK(bool o = 1) { print(o ? "TAK" : "NIE"); }
void NIE(bool o = 1) { TAK(not o); }

#if (__cplusplus >= 202002L)
#include <numbers>
constexpr ld pi = numbers::pi_v<ld>;
#endif
TE constexpr T inf = numeric_limits<T>::max();
template <> constexpr i128 inf<i128> = i128(inf<ll>) * 2'000'000'000'000'000'000;
template <typename T, typename U>
constexpr pair<T, U> inf<pair<T, U>> = {inf<T>, inf<U>};

TE constexpr static inline int pc(T x) { return popcount(make_unsigned_t<T>(x)); }
constexpr static inline ll si(const Z &a) { return a.size(); }

void reverse(Z &a) { reverse(all(a)); }

void unique(Z &a) {
  sort(a);
  a.erase(unique(all(a)), a.end());
}
TE vc<int> inverse(const vc<T> &a) {
  int N = si(a);
  vc<int> b(N, -1);
  FOR(i, N) if (a[i] != -1) b[a[i]] = i;
  return b;
}

Z QMAX(const Z &a) { return *max_element(all(a)); }
Z QMIN(const Z &a) { return *min_element(all(a)); }
TE Z QMAX(T l, T r) { return *max_element(l, r); }
TE Z QMIN(T l, T r) { return *min_element(l, r); }
constexpr bool chmax(Z &a, const Z &b) { return (a < b ? a = b, 1 : 0); }
constexpr bool chmin(Z &a, const Z &b) { return (a > b ? a = b, 1 : 0); }

vc<int> argsort(const Z &a) {
  vc<int> I(si(a));
  iota(all(I), 0);
  sort(I, [&](int i, int k) { return a[i] < a[k] or (a[i] == a[k] and i < k); });
  return I;
}
TE vc<T> rearrange(const vc<T> &a, const vc<int> &I) {
  int N = si(I);
  vc<T> b(N);
  FOR(i, N) b[i] = a[I[i]];
  return b;
}
template <int of = 1, typename T> 
vc<T> pre_sum(const vc<T> &a) {
  int N = si(a);
  vc<T> c(N + 1);
  FOR(i, N) c[i + 1] = c[i] + a[i];
  if (of == 0) c.erase(c.begin());
  return c;
}

TE constexpr static int topbit(T x) {
  if (x == 0) return - 1;
  if constexpr (sizeof(T) <= 4) return 31 - __builtin_clz(x);
  else return 63 - __builtin_clzll(x);
}
TE constexpr static int lowbit(T x) {
  if (x == 0) return -1;
  if constexpr (sizeof(T) <= 4) return __builtin_ctz(x);
  else return __builtin_ctzll(x);
}

TE constexpr T floor(T x, T y) { return x / y - (x % y and (x ^ y) < 0); }
TE constexpr T ceil(T x, T y) { return floor(x + y - 1, y); }
TE constexpr T bmod(T x, T y) { return x - floor(x, y) * y; }
TE constexpr pair<T, T> divmod(T x, T y) {
  T q = floor(x, y);
  return pair{q, x - q * y};
}
TE T SUM(const Z &v) { return accumulate(all(v), T()); }
TE T SUM(Z l, Z r) { return accumulate(l, r, T()); }
int lb(const Z &a, Z x) { return lower_bound(all(a), x) - a.begin(); }
TE int lb(T l, T r, Z x) { return lower_bound(l, r, x) - l; }
int ub(const Z &a, Z x) { return upper_bound(all(a), x) - a.begin(); }
TE int ub(T l, T r, Z x) { return upper_bound(l, r, x) - l; }

template <bool ck = 1>
ll bina(Z f, ll l, ll r) {
  if (ck) assert(f(l));
  while (abs(l - r) > 1) {
    ll x = (r + l) >> 1;
    (f(x) ? l : r) = x;
  }
  return l;
}
TE T bina_real(Z f, T l, T r, int c = 100) {
  while (c--) {
    T x = (l + r) / 2;
    (f(x) ? l : r) = x;
  }
  return (l + r) / 2;
}

TE T pop(vc<T> &a) {
  T x = a.back();
  a.pop_back();
  return x;
}
TE T pop(max_heap<T> &q) {
  T x = q.top();
  q.pop();
  return x;
}
TE T pop(min_heap<T> &q) {
  T x = q.top();
  q.pop();
  return x;
}
char pop(string &s) {
  char x = s.back();
  s.pop_back();
  return x;
}
void setp(int x) { cout << fixed << setprecision(x); }

TE inline void sh(vc<T> &a, int N, T b = {}) {
  a.resize(N, b);
}

#define c constexpr
template <int mod>
struct mint_t {
  using T = mint_t;
  static c uint m = mod;
  uint x;

  c inline uint val() const { return x; }

  c mint_t() : x(0) {}
  TE requires(is_unsigned_v<T>) mint_t(T x) : x(x % m) {}
  mint_t(u128 x) : x(x % m) {}
  TE requires(is_signed_v<T>) mint_t(T x) : x((x %= mod) < 0 ? x + mod : x) {}
  mint_t(i128 x) : x((x %= mod) < 0 ? x + mod : x) {}

  c T &operator+=(T p) {
    if ((x += p.x) >= m) x -= m;
    return *this;
  }
  c T &operator-=(T p) {
    if ((x += m - p.x) >= m) x -= m;
    return *this;
  }
  c T operator+(T p) const { return T(*this) += p; }
  c T operator-(T p) const { return T(*this) -= p; }

  c T &operator*=(T p) {
    x = ull(x) * p.x % m;
    return *this;
  }
  c T operator*(T p) const { return T(*this) *= p; }

  c T &operator/=(T p) { return *this *= p.inv(); }
  c T operator/(T p) const { return T(*this) /= p; }

  c T operator-() const { return T::gen(x ? mod - x : 0); }

  c T inv() const {
    int a = x, b = mod, x = 1, y = 0;
    while (b > 0) {
      int t = a / b;
      swap(a -= t * b, b);
      swap(x -= t * y, y);
    }
    return T(x);
  }

  
  c T pow(ll k) const {
    if (k < 0) return inv().pow(-k);
    T s(1), a(x);
    for (; k; k >>= 1, a *= a) {
      if (k & 1) s *= a;
    }
    return s;
  }

  c bool operator<(T p) const { return x < p.x; }
  c bool operator==(T p) const { return x == p.x; }
  c bool operator!=(T p) const { return x != p.x; }

  static c T gen(uint x) {
    T s;
    s.x = x;
    return s;
  }

  friend istream &operator>>(istream &cin, T &p) {
    ll t;
    cin >> t;
    p = t;
    return cin;
  }
  
  friend ostream &operator<<(ostream &cout, T p) { return cout << p.x; }

  static c int get_mod() { return mod; }

  static c PII ntt_info() {
    if (mod == 167772161) return {25, 17};
    if (mod == 469762049) return {26, 30};
    if (mod == 754974721) return {24, 362};
    if (mod == 998244353) return {23, 31};
    if (mod == 120586241) return {20, 74066978};
    if (mod == 880803841) return {23, 211};
    if (mod == 943718401) return {22, 663003469};
    if (mod == 1004535809) return {21, 582313106};
    if (mod == 1012924417) return {21, 368093570};
    return {-1, -1};
  }
  
  static c bool can_ntt() { return ntt_info().fi != -1; }
};
#undef c

using M99 = mint_t<998244353>;
using M17 = mint_t<1000000007>;
using M11 = M17;

#ifdef FIO
template <int mod>
void rd(mint_t<mod> &x) {
  LL(y);
  x = y;
}
template <int mod>
void wt(mint_t<mod> x) {
  wt(x.x);
}
#endif

TE requires (T::commute) 
struct bit_t : T {
  using X = T::X;
  using T::op, T::inv, T::unit;

  int N;
  vc<X> a;
  
  bit_t() {}
  bit_t(int N) { build(N); }
  bit_t(int N, Z f) { build(N, f); }
  bit_t(const vc<X> &a) { build(a); }

  void build(int M) { N = M, a.assign(N, unit()); }

  void build(const vc<X> &a) {
    build(si(a), [&](int i) { return a[i]; });
  }

  void build(int M, Z f) {
    N = M;
    a.resize(N);
    FOR(i, N) a[i] = f(i);
    FOR(i, 1, N + 1) {
      int k = i + (i & -i);
      if (k <= N) a[k - 1] = op(a[i - 1], a[k - 1]);
    }
  }

  void multiply(int i, X x) {
    for (; i < N; i |= i + 1) a[i] = op(a[i], x);
  }

  inline X prod(int i) {
    assert(i >= 0 and i <= N);
    X s = unit();
    for (; i > 0; i -= i & -i) s = op(s, a[i - 1]);
    return s;
  }

  X prod(int l, int r) {
    assert(l <= r);
    return op(prod(r), inv(prod(l)));
  }

  X prod_all() { return prod(N); }

  void set(int i, X x) { multiply(i, op(inv(get(i)), x)); }

  X get(int i) { return prod(i, i + 1); }

  vc<X> get_all() {
    vc<X> s(N);
    FOR(i, N) s[i] = get(i);
    return s;
  }

  int max_right(Z f, int L = 0) {
    assert(f(unit()));
    X s = unit();
    int i = L, k = [&]() -> int {
      while (1) {
        if (i & 1) s = op(s, inv(a[i - 1])), --i;
        if (i == 0) return topbit(N) + 1;
        int k = lowbit(i) - 1;
        if (i + (1 << k) > N) return k;
        X t = op(s, a[i + (1 << k) - 1]);
        if (not f(t)) return k;
        s = op(s, inv(a[i - 1])), i -= i & -i;
      }
    }();
    while (k) {
      --k;
      if (i + (1 << k) - 1 < N) {
        X t = op(s, a[i + (1 << k) - 1]);
        if (f(t)) i += (1 << k), s = t;
      }
    }
    return i;
  }

  int kth(X k, int L = 0) {
    return max_right([&](X x) { return x <= k; }, L);
  }

  int min_left(Z f, int R) {
    assert(f(unit()));
    X s = unit();
    int i = R, k = 0;
    while (i > 0 and f(s)) {
      s = op(s, a[i - 1]);
      k = lowbit(i);
      i -= i & -i;
    }
    if (f(s)) return assert(i == 0), 0;
    while (k) {
      --k;
      X t = op(s, inv(a[i + (1 << k) - 1]));
      if (not f(t)) i += (1 << k), s = t;
    }
    return i + 1;
  }
};

TE struct monoid_add {
  using X = T;
  static constexpr inline X op(const X &x, const X &y) { return x + y; }
  static constexpr inline X inv(const X &x) { return -x; }
  static constexpr inline X pow(const X &x, ll n) { return X(n) * x; }
  static constexpr inline X unit() { return X(0); }
  static constexpr bool commute = 1;
};

template <typename mint>
mint count_inv_of_permutation(const vc<int> &p, ll K) {
  int N = si(p);
  bit_t<monoid_add<ll>> bit(N + 1);
  vc<ll> a;
  for (int x : p) bit.multiply(x, 1), a.ep(x - bit.prod(x + 1));
  reverse(a);
  i128 in = i128(K) + 10, re = 1, fa = 1;
  mint s;
  Z ke = [&](i128 n, i128 mod, i128 x) -> mint {
    mint s = -x * i128(x - 1) / 2;
    n += x;
    x = 0;
    Z [q, r] = divmod<i128>(n, mod);
    s += mint(q) * (mod - 1) * mod / 2;
    s += r * (r - 1) / 2;
    return s;
  };
  Z ca = [&](i128 N, i128 mod, i128 re, i128 x, i128 res) -> mint {
    mint s;
    i128 n = min(N, res);
    N -= n;
    s += mint(x) * n;
    x = (x + 1) % mod;
    Z [q, r] = divmod(N, re);
    s += ke(q, mod, x) * re;
    x = (x + q) % mod;
    s += x * r;
    return s;
  };
  FOR(i, N) {
    ll x = a[i];
    s += ca(K, i + 1, fa, x, re);
    if (re < in) re += fa * (i - x);
    if (fa < in) fa *= (i + 1);
  }
  return s;
}

using mint = M11;
void Yorisou() {
  LL(N, K);
  VEC(int, p ,N);
  print(count_inv_of_permutation<mint>(p, K));
}
constexpr int tests = 0, fl = 0, DB = 10;
int main() {
  cin.tie(0)->sync_with_stdio(0);
  int T = 1;
  if (fl) cerr.tie(0);
  if (tests and not fl) IN(T);
  for (int i = 0; i < T or fl; ++i) {
    Yorisou();
    if (fl and i % DB == 0) cerr << "Case: " << i << '\n';
  }
  return 0;
}