[Debian-med-packaging] Bowtie (Was: py2removal packages)

[Andreas Tille]

Hi Alex,
[moving to med-devel list ...]

On Thu, Sep 05, 2019 at 10:49:08AM +0200, Alex Mestiashvili wrote:
> thank you for updating the packages (multipletau,subread,bowtie2,...)
> and sorry that I haven't updated them in time, I was just about to do
> that, but you were faster :)

I was picking the easy ones. ;-) There are other harder ones left.
Bowtie2 is making me some headache.  The test suite is failing.  It
becomes even worse when trying to build on a stretch machine:



./bowtie2 -x .dummy_index .simple_tests.fq.gz | samtools view -b -o .test.bam
1 reads; of these:
  1 (100.00%) were unpaired; of these:
    1 (100.00%) aligned 0 times
    0 (0.00%) aligned exactly 1 time
    0 (0.00%) aligned >1 times
0.00% overall alignment rate
./bowtie2 --sanitized    --quiet -a --reads-per-batch 4 -x .simple_tests.tmp -b .test.bam
=================================================================
==24891==ERROR: AddressSanitizer: heap-use-after-free on address 0x61600000fc80 at pc 0x556383c3133e bp 0x7f558f4f2b90 sp 0x7f558f4f2b88
READ of size 4 at 0x61600000fc80 thread T1
    #0 0x556383c3133d in ferror_unlocked /usr/include/x86_64-linux-gnu/bits/stdio.h:134
    #1 0x556383c3133d in BAMPatternSource::nextBGZFBlockFromFile(BAMPatternSource::BGZF&) pat.cpp:1251
    #2 0x556383c85c10 in BAMPatternSource::nextBatch(PerThreadReadBuf&, bool, bool) pat.cpp:1316
    #3 0x556383c20f9c in DualPatternComposer::nextBatch(PerThreadReadBuf&) pat.cpp:229
    #4 0x556383c51497 in PatternSourcePerThread::nextBatch() pat.h:1175
    #5 0x556383c51497 in PatternSourcePerThread::nextReadPair() pat.cpp:155
    #6 0x556383a7e99f in multiseedSearchWorker bt2_search.cpp:3142
    #7 0x556383aa571b in tbb::internal::thread_closure_1<void (*)(void*), void*>::start_routine(void*) /usr/include/tbb/tbb_thread.h:92
    #8 0x7f5594c0e4a3 in start_thread (/lib/x86_64-linux-gnu/libpthread.so.0+0x74a3)
    #9 0x7f5592f5ad0e in __clone (/lib/x86_64-linux-gnu/libc.so.6+0xe8d0e)

0x61600000fc80 is located 0 bytes inside of 552-byte region [0x61600000fc80,0x61600000fea8)
freed by thread T1 here:
==24891==AddressSanitizer CHECK failed: ../../../../src/libsanitizer/asan/asan_allocator.cc:665 "((res.trace)) != (0)" (0x0, 0x0)
    #0 0x7f559510aebd  (/usr/lib/x86_64-linux-gnu/libasan.so.3+0xcaebd)
    #1 0x7f5595110b13 in __sanitizer::CheckFailed(char const*, int, char const*, unsigned long long, unsigned long long) (/usr/lib/x86_64-linux-gnu/libasan.so.3+0xd0b13)
    #2 0x7f559505fb9c  (/usr/lib/x86_64-linux-gnu/libasan.so.3+0x1fb9c)
    #3 0x7f55950609ce  (/usr/lib/x86_64-linux-gnu/libasan.so.3+0x209ce)
    #4 0x7f5595106951  (/usr/lib/x86_64-linux-gnu/libasan.so.3+0xc6951)
    #5 0x7f559510a376  (/usr/lib/x86_64-linux-gnu/libasan.so.3+0xca376)
    #6 0x7f559510b307 in __asan_report_load4 (/usr/lib/x86_64-linux-gnu/libasan.so.3+0xcb307)
    #7 0x556383c3133d in ferror_unlocked /usr/include/x86_64-linux-gnu/bits/stdio.h:134
    #8 0x556383c3133d in BAMPatternSource::nextBGZFBlockFromFile(BAMPatternSource::BGZF&) pat.cpp:1251
    #9 0x556383c85c10 in BAMPatternSource::nextBatch(PerThreadReadBuf&, bool, bool) pat.cpp:1316
    #10 0x556383c20f9c in DualPatternComposer::nextBatch(PerThreadReadBuf&) pat.cpp:229
    #11 0x556383c51497 in PatternSourcePerThread::nextBatch() pat.h:1175
    #12 0x556383c51497 in PatternSourcePerThread::nextReadPair() pat.cpp:155
    #13 0x556383a7e99f in multiseedSearchWorker bt2_search.cpp:3142
    #14 0x556383aa571b in tbb::internal::thread_closure_1<void (*)(void*), void*>::start_routine(void*) /usr/include/tbb/tbb_thread.h:92
    #15 0x7f5594c0e4a3 in start_thread (/lib/x86_64-linux-gnu/libpthread.so.0+0x74a3)
    #16 0x7f5592f5ad0e in __clone (/lib/x86_64-linux-gnu/libc.so.6+0xe8d0e)

(ERR): bowtie2-align exited with value 1


I'd be super happy if somebody could dedicate some time into it (not
only the issue under stretch but it is a clear signal that the code
has memory allocation issues).

I suspect these issues are also in Bowtie (version 1).  In a simple
chroot it segfaults when just calling it with --help option.

Kind regards

        Andreas.

-- 
http://fam-tille.de