Segfault with -H

[armintoepfer]

Hey Heng,

I have an example that leads to a segfault. We are running into many of those lately...even using HEAD.

minimap2 ref.fasta read.fasta -a -k 15 -w 5 -H
[M::mm_idx_gen::0.001*3.95] collected minimizers
[M::mm_idx_gen::0.002*2.96] sorted minimizers
@SQ	SN:ref	LN:2316
@PG	ID:minimap2	PN:minimap2	VN:2.14-r883	CL:minimap2 -a -k 15 -w 5 -H ref.fasta read.fasta
[M::main::0.002*2.89] loaded/built the index for 1 target sequence(s)
[M::mm_mapopt_update::0.002*2.77] mid_occ = 5
[M::mm_idx_stat] kmer size: 15; skip: 5; is_hpc: 1; #seq: 1
[M::mm_idx_stat::0.002*2.70] distinct minimizers: 507 (99.21% are singletons); average occurrences: 1.014; average spacing: 4.506
[1]    52247 segmentation fault  minimap2 ref.fasta read.fasta -a -k 15 -w 5 -H

I get to align.c:672

-> 672 			mm_seq_rev(rs - rs0, tseq);

(lldb) print rs
(int32_t) $13 = 75
(lldb) print rs0
(int32_t) $14 = 78

leading to an underflow.

Thank you,
Armin

[lh3]

Thanks. That is an oversight. -H creates quite a lot of corner cases.

[armintoepfer]

Much appreciated. Any chance to get a 2.15 cut soon, since the last one has been two months ago?

[lh3]

I am running routine tests. If the results are ok, I will cut a release today or tomorrow.