[Debian-med-packaging] gmap tests segfault on i386

[Andreas Tille]

Hi Alex,

On Sun, Oct 26, 2014 at 01:02:47PM +0100, Alexander Mestiashvili wrote:
> Hi Andreas,
> 
> I've tried to build gmap again on my i386 virtualbox machine, but
> one of the tests segfaults.
> There are no buildlogs for this package, so I can not check if it
> worked on i386 in the past.
> Also I think because the package operates with huge datasets, it is
> not likely than anybody uses it on i386.
> So how should we proceed ? upload to unstable anyway ?

I see two options:

   1. Upload to amd64 (or to 64bit arches) only
   2. Patch the test suite to skip this test for i386 and document
      this in README.test

> I've submitted a bug report to upstream, but it seems we do not have
> time to wait for the answer for inclusion in jessy.

I'd suggest to choose the option you regard as simple enough to
implement until today evening.

Kind regards

      Andreas.

> Thanks,
> Alex
> 
> make  check-TESTS
> make[1]: Entering directory '/tmp/gmap/tests'
> GMAP version 2014-10-22 called with args: /tmp/gmap/src/gmap -A -g
> /tmp/gmap/tests/ss.chr17test /tmp/gmap/tests/ss.her2
> Checking compiler assumptions for popcnt: 6B8B4567 __builtin_clz=1
> __builtin_ctz=0 __builtin_popcount=17
> Checking compiler assumptions for SSE2: 6B8B4567 327B23C6 xor=59F066A1
> Processed 1 queries in 0.44 seconds (2.27 queries/sec)
> PASS: align.test
> Opening file /tmp/gmap/tests/ss.chr17test
>   Processed short contigs (<1000000 nt): .
> ============================================================
> Contig mapping information has been written to file coords.chr17test.
> You should look at this file, and edit it if necessary
> If everything is okay, you should proceed by running
>     make -f Makefile.chr17test gmapdb
> ============================================================
> PASS: coords1.test
> Sorting chromosomes in chrom order.  To turn off or sort other ways,
> use the -s flag.
> Checking compiler assumptions for popcnt: 6B8B4567 __builtin_clz=1
> __builtin_ctz=0 __builtin_popcount=17
> Checking compiler assumptions for SSE2: 6B8B4567 327B23C6 xor=59F066A1
> Creating files in directory ./chr17test
> Running /tmp/gmap/src/fa_coords    -o ./chr17test.coords -f
> ./chr17test.sources
> Opening file /tmp/gmap/tests/ss.chr17test
>   Processed short contigs (<1000000 nt): .
> ============================================================
> Contig mapping information has been written to file ./chr17test.coords.
> You should look at this file, and edit it if necessary
> If everything is okay, you should proceed by running
>     make gmapdb
> ============================================================
> Running /tmp/gmap/src/gmap_process  -c ./chr17test.coords -f
> ./chr17test.sources | /tmp/gmap/src/gmapindex  -d chr17test -D
> ./chr17test -A
> Reading coordinates from file ./chr17test.coords
> Logging contig chr17test at chr17test:1..200000 in genome chr17test
> Total genomic length = 200000 bp
> Chromosome chr17test has universal coordinates 1..0
> Writing IIT file header information...coordinates require 4 bytes
> each...done
> Processing null division/chromosome...sorting...writing...done (1 intervals)
> Writing IIT file footer information...done
> 
> Segmentation fault
> 
> /tmp/gmap/src/gmap_process  -c ./chr17test.coords -f
> ./chr17test.sources | /tmp/gmap/src/gmapindex  -d chr17test -D
> ./chr17test -A  failed with return code 35584 at
> /tmp/gmap/util/gmap_build line 269.
> diff: ./chr17test/chr17test.genomecomp: No such file or directory
> Difference found between chr17test.genomecomp and setup.genomecomp.ok
> FAIL: setup1.test
> Finished reading FASTA file -- total entries: 3
> Maximum coordinate: 40000
> Total label length: 3 + 0 separators
> Total annotation length: 80 + 0 separators
> Saw 0 distinct divisions/chromosomes
> Saw 3 distinct tags/types
> No divs/chromosomes provided, so storing as IIT version 1
> Writing IIT file header information...coordinates require 4 bytes
> each...done
> Processing null division/chromosome...sorting...writing...done (3 intervals)
> Writing IIT file footer information...done
> 
> 
> 
> 
> 
> 
> 
> _______________________________________________
> Debian-med-packaging mailing list
> Debian-med-packaging at lists.alioth.debian.org
> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/debian-med-packaging
> 

-- 
http://fam-tille.de