#14 mrsfast: malloc(): memory corruption
This started occurring after running mrsfast sequentially many times with a script and now occurs more frequently.
*** glibc detected *** mrsfast: malloc(): memory corruption: 0x0000000002bfdf20 ***
======= Backtrace: =========
[0x428274]
[0x42a992]
[0x42c200]
[0x406a5a]
[0x4003bb]
[0x4195ab]
[0x4001b9]
======= Memory map: ========
00400000-004bb000 r-xp 00000000 fe:01 165711 /usr/local/bin/mrsfast
006ba000-006bc000 rw-p 000ba000 fe:01 165711 /usr/local/bin/mrsfast
006bc000-00782000 rw-p 006bc000 00:00 0
0086c000-02c20000 rw-p 0086c000 00:00 0 [heap]
7facb8000000-7facb8023000 rw-p 7facb8000000 00:00 0
7facb8023000-7facbc000000 ---p 7facb8023000 00:00 0
7facbe8cc000-7facc2f6b000 rw-p 7facbe8cc000 00:00 0
7fffdf4c5000-7fffdf4da000 rw-p 7ffffffe9000 00:00 0 [stack]
7fffdf5d7000-7fffdf5d8000 r-xp 7fffdf5d7000 00:00 0 [vdso]
ffffffffff600000-ffffffffff601000 r-xp 00000000 00:00 0 [vsyscall]
*** glibc detected *** mrsfast: free(): invalid next size (fast): 0x0000000003424f30 ***
======= Backtrace: =========
[0x428274]
[0x42a284]
[0x405588]
[0x4058a6]
[0x400b8a]
[0x4195ab]
[0x4001b9]
======= Memory map: ========
00400000-004bb000 r-xp 00000000 fe:01 165711 /usr/local/bin/mrsfast
006ba000-006bc000 rw-p 000ba000 fe:01 165711 /usr/local/bin/mrsfast
006bc000-00782000 rw-p 006bc000 00:00 0
01616000-4e58d000 rw-p 01616000 00:00 0 [heap]
7febf0000000-7febf0023000 rw-p 7febf0000000 00:00 0
7febf0023000-7febf4000000 ---p 7febf0023000 00:00 0
7febf7eef000-7febfdb6f000 rw-p 7febf7eef000 00:00 0
7fec0452f000-7fec11d6b000 rw-p 7fec0452f000 00:00 0
7fff1fe94000-7fff1fea9000 rw-p 7ffffffe9000 00:00 0 [stack]
7fff1ffff000-7fff20000000 r-xp 7fff1ffff000 00:00 0 [vdso]
ffffffffff600000-ffffffffff601000 r-xp 00000000 00:00 0 [vsyscall]
This happens when I run mrsfast as a thread in Perl.
This was probably happening because the read length was too long. I shortened the read length and it looks good.
Hi
I don't know anything about Perl threads, but does it also crash without the Perl wrappper? Also what was the original read length, and was the read length uniform among all reads?
this is mrsfast category. I assume it was fixed in 2.4.0.4, but I am leaving this to fhach to decide to close or not
Would you please provide the command you used? It would be also useful if you can provide me with your read length, # of the reads and reference genome that you used.
I am not sure of the read length, I was downloading 1000 Genomes Project sequence files and running using mrsfast. At the time of the error, I wasn't splitting the reads into 36 base chunks. After splitting the reads into 36 base chunks, the error didn't occur anymore. I am not sure of which files from the 1000 Genomes Project were causing the error since I was downloading and deleting in my script. I think this may have occurred because of the long sequence read length and also I was running mrsfast alignments using Perl threads. I have seen other bug reports from different applications where there are some memory errors after running threads for many hours.
here is the command I was using. My "--search" file was multiple fasta sequences each of 200 bases length.
mrsfast --search my_seqs.fa --seq query_file -o alignment_outfile