#pragma GCC target("avx")
#pragma GCC optimize("O3")
#pragma GCC optimize("unroll-loops")
#include <bits/extc++.h>
using namespace std;
using ld = long double; 
const vector<int> dx = {0, 0, 1, -1};
const vector<int> dy = {1, -1, 0, 0};
#define vec vector
#define int long long
#define double long double
//cout<<setprecision(10)<<fixed<<..
#define rep(i, n) for (int i = 0; i < (int)(n); i++)
#define repp(i, x, n) for (int i = x; i < (int)(n); i++)
#define pii pair<int,int>
#define pq priority_queue
#define all(V) begin(V),end(V)
#define printpair(p) cout<<p.first<<" "<<p.second<<"\n"
#define tple tuple<int,int,int>
template<class T, class U> inline bool chmax(T &a, const U &b) { if (a < b) { a = b; return true; } return false; }
template<class T, class U> inline bool chmin(T &a, const U &b) { if (a > b) { a = b; return true; } return false; }
#define mend(a,b) G[a].push_back(b)

vec<int> daiku(vec<vec<int>> &G,int s,int n){
    pq<pii,vec<pii>,greater<pii>> Q;
    vec<bool> done(n+1,false);
    vec<int>  dist(n+1,1e9);
    Q.push({0,s});
    dist[s]=0;
    while(Q.size()){
        int v=Q.top().second;
        Q.pop();
        if(done[v]){continue;}
        done[v]=true;
        for(int nv:G[v]){
            if(dist[nv]>dist[v]+1){
                dist[nv]=dist[v]+1;
                Q.push({dist[nv],nv});
            }
        }
    }
    return dist;
}

signed main(){
    ios::sync_with_stdio(false);
    cin.tie(nullptr);

    int n,m;cin>>n>>m;
    vec G(n+1,vec<int>());
    rep(i,m){
        int a,b;cin>>a>>b;
        G[a].push_back(b);
    }
    vec<int> one_to_n1=daiku(G,1,n);
    vec<int> n1_to_n=daiku(G,n-1,n);
    vec<int> n_to_1=daiku(G,n,n);
    int inf=1e9;
    int ans=one_to_n1[n-1]+n1_to_n[n]+n_to_1[1];
    vec<int> one_to_n=daiku(G,1,n);
    vec<int> n_to_n1=daiku(G,n,n);
    vec<int> n1_to_1=daiku(G,n-1,n);
    chmin(ans,one_to_n[n]+n_to_n1[n-1]+n1_to_1[1]);
    if(ans>=inf){
        cout<<-1<<"\n";return 0;
    }
    cout<<ans<<"\n";
    

}