diff --git a/chain.c b/chain.c
index 102907e..45801b7 100644
--- a/chain.c
+++ b/chain.c
@@ -69,11 +69,15 @@ int mm_chain_dp(int match_len, int max_dist, int bw, int max_skip, int min_sc, i
 		if (t[i] == 0 && f[i] >= min_sc)
 			u[n_u++] = (uint64_t)f[i] << 32 | i;
 	radix_sort_64(u, u + n_u);
+	for (i = 0; i < n_u>>1; ++i) { // reverse, s.t. the highest scoring chain is the first
+		uint64_t t = u[i];
+		u[i] = u[n_u - 1], u[n_u - 1] = t;
+	}
 
 	// backtrack
 	memset(t, 0, n * 4);
 	v = (int32_t*)kmalloc(km, n * 4);
-	for (i = n_u - 1, n_v = k = 0; i >= 0; --i) { // starting from the highest score
+	for (i = n_v = k = 0; i < n_u; ++i) { // starting from the highest score
 		int32_t n_v0 = n_v;
 		j = (int32_t)u[i];
 		do {
diff --git a/index.c b/index.c
index ab9d63d..e89c2f2 100644
--- a/index.c
+++ b/index.c
@@ -63,6 +63,7 @@ void mm_idx_stat(const mm_idx_t *mi)
 {
 	int i, n = 0, n1 = 0;
 	uint64_t sum = 0, len = 0;
+	fprintf(stderr, "[M::%s] kmer size: %d; skip: %d; is_HPC: %d; #seq: %d\n", __func__, mi->k, mi->w, mi->is_hpc, mi->n_seq);
 	for (i = 0; i < mi->n_seq; ++i)
 		len += mi->seq[i].len;
 	for (i = 0; i < 1<<mi->b; ++i)
diff --git a/main.c b/main.c
index 662eb6a..27f36a6 100644
--- a/main.c
+++ b/main.c
@@ -94,11 +94,11 @@ int main(int argc, char *argv[])
 	if (w < 0) w = (int)(.6666667 * k + .499);
 
 	if (argc == optind) {
-		fprintf(stderr, "Usage: minimap2 [options] <target.fa> [query.fa] [...]\n");
+		fprintf(stderr, "Usage: minimap2 [options] <target.fa>|<target.idx> [query.fa] [...]\n");
 		fprintf(stderr, "Options:\n");
 		fprintf(stderr, "  Indexing:\n");
 		fprintf(stderr, "    -H         use homopolymer-compressed k-mer\n");
-		fprintf(stderr, "    -k INT     k-mer size [%d]\n", k);
+		fprintf(stderr, "    -k INT     k-mer size (no larger than 28) [%d]\n", k);
 		fprintf(stderr, "    -w INT     minizer window size [{-k}*2/3]\n");
 		fprintf(stderr, "    -I NUM     split index for every ~NUM input bases [4G]\n");
 		fprintf(stderr, "    -d FILE    dump index to FILE []\n");