Silent error when running extractHAIRS

[nh13]

I am encountering a silent error in extractHAIRS. I run extractHAIRS and see that it stops output fragments part way through the input file. The exit code returned by extractHAIRS is 0!

I ran it through gdb and found the following:

#0  0x00007fff9c2d4002 in __pthread_kill () from /usr/lib/system/libsystem_kernel.dylib
#1  0x00007fff969c15c5 in pthread_kill () from /usr/lib/system/libsystem_pthread.dylib
#2  0x00007fff965446e7 in abort () from /usr/lib/system/libsystem_c.dylib
#3  0x00007fff9ab17e66 in szone_error () from /usr/lib/system/libsystem_malloc.dylib
#4  0x00007fff9ab0e0f9 in small_malloc_from_free_list () from /usr/lib/system/libsystem_malloc.dylib
#5  0x00007fff9ab0ab74 in szone_malloc_should_clear () from /usr/lib/system/libsystem_malloc.dylib
#6  0x00007fff9ab0a5b1 in malloc_zone_malloc () from /usr/lib/system/libsystem_malloc.dylib
#7  0x00007fff9ab090dc in malloc () from /usr/lib/system/libsystem_malloc.dylib
#8  0x00007fff9159b224 in _objc_opt_rw_data () from /usr/lib/libz.1.dylib
#9  0x00007fff5fbfe520 in ?? ()
#10 0x00000000fffffff1 in ?? ()
#11 0x00007fff5fbfe4f0 in ?? ()
#12 0x00007fff91596f05 in _objc_opt_rw_data () from /usr/lib/libz.1.dylib
#13 0x00000001c3d02c00 in ?? ()
#14 0x00000001c3d12c12 in ?? ()
#15 0x0000000102d5b4a0 in ?? ()
#16 0x0000000000000b02 in ?? ()
#17 0x00007fff5fbfe5e0 in ?? ()
#18 0x000000010000d4b8 in inflate_block (fp=0x7fff7c4a1000, block_length=<optimized out>) at bgzf.c:300
#19 bgzf_read_block (fp=0x7fff7c4a1000) at bgzf.c:534
Backtrace stopped: previous frame identical to this frame (corrupt stack?)

I updated both the samtools and htslib submodules to master and re-ran extractHAIRS and got:

reading sorted bamfile ../../path/to/BAM
[E::bgzf_read] bgzf_read_block error -1 after 27499 of 107505 bytes

The reads in the SAM file are very long so my suspicion would be with the underlying htslib/samtools code, but I would be happy to share my test case privately.

[nh13]

I wrote a little tool in samtools to read the input BAM and it completed successfully. My guess is there is a memory leak or something is overwriting memory in extractHairs.

[nh13]

It looks like if I increase this from 1000 to 10000 in extracthairs.c like in fosmidbam_hairs.c the problem goes away.

Ideally, the # of alleles per-read shouldn't be fixed, but instead be increased (use realloc) when more space is needed. @pjedge could you take a shot at this?

[nh13]

@vibansal I am not sure if this is a problem on HapCut(1) but I thought I'd make you aware.