#include <atcoder/all>

#include <bits/stdc++.h>

using namespace std;
#define SZ(x) (int) (x).size()
#define REP(i, n) for(int i = 0; i < (n); i++)
#define FOR(i, a, b) for(auto i = (a); i < (b); i++)
#define For(i, a, b, c) \
	for(auto i = (a); i != (b); i += (c))
#define REPR(i, n) for(auto i = (n) -1; i >= 0; i--)
#define ALL(s) (s).begin(), (s).end()
#define so(V) sort(ALL(V))
#define rev(V) reverse(ALL(V))
#define uni(v) v.erase(unique(ALL(v)), (v).end())
#define eb emplace_back

typedef unsigned long long ull;
typedef long long ll;
typedef vector<int> vi;
typedef vector<ll> vll;
typedef vector<bool> vb;
typedef vector<vi> vvi;
typedef vector<vll> vvll;
typedef pair<int, int> PI;
typedef pair<ll, ll> PL;
const double EPS = 1e-6;
const int MOD = 1e9 + 7;
const int INF = (1 << 30);
const ll LINF = 1e18;
const double math_PI = acos(-1);

template<typename T>
vector<T> make_v(size_t a) {
	return vector<T>(a);
}

template<typename T, typename... Ts>
auto make_v(size_t a, Ts... ts) {
	return vector<decltype(make_v<T>(ts...))>(
		a, make_v<T>(ts...));
}

template<typename T, typename V>
typename enable_if<is_class<T>::value == 0>::type fill_v(
	T& t, const V& v) {
	t = v;
}

template<typename T, typename V>
typename enable_if<is_class<T>::value != 0>::type fill_v(
	T& t, const V& v) {
	for(auto& e: t) fill_v(e, v);
}

template<class T>
bool chmax(T& a, const T& b) {
	if(a < b) {
		a = b;
		return true;
	}
	return false;
}

template<class T>
bool chmin(T& a, const T& b) {
	if(a > b) {
		a = b;
		return true;
	}
	return false;
}

template<typename S, typename T>
istream& operator>>(istream& is, pair<S, T>& p) {
	cin >> p.first >> p.second;
	return is;
}

template<typename T>
istream& operator>>(istream& is, vector<T>& vec) {
	for(T& x: vec) is >> x;
	return is;
}

template<typename T>
string join(vector<T>& vec, string splitter) {
	stringstream ss;
	REP(i, SZ(vec)) {
		if(i != 0) ss << splitter;
		ss << vec[i];
	}
	return ss.str();
}

template<typename T>
ostream& operator<<(ostream& os, vector<T>& vec) {
	os << join(vec, " ");
	return os;
}
using namespace atcoder;

using mint = modint998244353;
struct S {
	mint value;
	int size;
};
using F = ll;

S op(S a, S b) {
	return { a.value + b.value, a.size + b.size };
}
S e() {
	return { mint(0), 0 };
}
S mapping(F f, S x) {
	return { x.value + f * x.size, x.size };
}
F composition(F f, F g) {
	return f + g;
}
F id() {
	return 0;
}

int main() {
	cin.tie(nullptr);
	ios::sync_with_stdio(false);
	int N, M;
	cin >> N >> M;
	vector<mint> fact(N + 2);
	fact[0] = mint(1);
	fact[1] = mint(1);
	FOR(i, 2, N + 2) {
		fact[i] = fact[i - 1] * i;
	}
	vi A(N, -1);
	set<int> SE;
	REP(i, N) SE.insert(i);
	REP(i, M) {
		int P, K;
		cin >> P >> K;
		P--;
		K--;
		A[K] = P;
		SE.erase(P);
	}
	fenwick_tree<ll> fw(N);
	mint ans = 0;
	REP(i, N) {
		if(A[i] != -1) {
			ans += fw.sum(A[i] + 1, N);
			fw.add(A[i], 1);
		}
	}
	REP(i, N) {
		fw.add(i, -fw.sum(i, i + 1));
	}
	int rest = N - M;
	ans *= fact[rest];
	if(rest >= 2)
		ans += (mint(rest) * mint(rest - 1) / 2).pow(2)
			* fact[rest - 2];
	lazy_segtree<S, op, e, F, mapping, composition, id> seg(
		vector<S>(rest, { mint(0), 1 }));
	vi R;
	for(auto s: SE) {
		R.push_back(s);
	}
	if(rest >= 1) {
		REP(i, N) {
			if(A[i] == -1) {
				ans += seg.all_prod().value
					* fact[rest - 1];
			} else {
				auto it = upper_bound(ALL(R), A[i]);
				seg.apply(0, it - R.begin(), 1);
			}
		}
		REP(i, rest) {
			seg.set(i, { mint(0), 1 });
		}
		REPR(i, N) {
			if(A[i] == -1) {
				ans += seg.all_prod().value
					* fact[rest - 1];
			} else {
				auto it = lower_bound(ALL(R), A[i]);
				seg.apply(it - R.begin(), rest, 1);
			}
		}
	}
	cout << ans.val() << endl;
	return 0;
}