search

bioperl / bioperl-live-redmine

Legacy tickets migrated from the OBF Redmine issue tracker: http://redmine.open-bio.org
0 stars 0 forks source link

Segmentation Fault in Bio::Tools::dpAlign::pairwise_alignment_score #33

Open cjfields opened 9 years ago

cjfields commented 9 years ago

Author Name: Florent Angly (@fangly) Original Redmine Issue: 2384, https://redmine.open-bio.org/issues/2384 Original Date: 2007-10-18 Original Assignee: Bioperl Guts

Some sequences produce an segmentation fault in the external dpAlign C code. I have isolated 2 sequences that produce the bug to happen. If the query sequence (that makes the profile) and target sequence are interverted, the score is calculated without problems.

cjfields commented 9 years ago

Original Redmine Comment Author Name: Florent Angly Original Date: 2007-10-18T16:54:09Z

Created an attachment (id=775) Script to reproduce the crash

cjfields commented 9 years ago

Original Redmine Comment Author Name: Florent Angly Original Date: 2007-10-18T16:54:44Z

Created an attachment (id=776) Backtrace produced by GDB

cjfields commented 9 years ago

Original Redmine Comment Author Name: Florent Angly Original Date: 2007-10-18T18:57:39Z

Created an attachment (id=777) Script to reproduce the crash

cjfields commented 9 years ago

Original Redmine Comment Author Name: Chris Fields Original Date: 2007-10-23T12:46:52Z

Florent, this works for me on my MacBook (Mac OS X, Tiger, perl 5.8.6), though I do get a warning (though the sequences are obviously there):

——————————- WARNING ——————————- MSG: Replacing one sequence [ABC|9944760/1-104]

Getting score for ABC|9944760 -> ABC|9986984 = 300 Getting score for ABC|9986984 -> ABC|9944760 = 303

This may be a platform-related issue. Could you give a bit more information on os, perl version, etc.?

cjfields commented 9 years ago

Original Redmine Comment Author Name: Chris Fields Original Date: 2007-10-23T12:58:41Z

A ‘blocker’ indicates that bioperl dev can’t proceed. This bug has no impact on the other BioPerl modules, so I’m demoting severity down to ‘normal’.

cjfields commented 9 years ago

Original Redmine Comment Author Name: Florent Angly Original Date: 2007-10-23T14:38:23Z

Yes, sorry, I shouldn’t touch stuff that I don’t know.

Regarding the Segfault, note that I had altered the Perl Makefile.PL to enable compiling with debugging symbols to see all the debugging details in GDB.

The really weird thing is that using the one sequence as target and the other as target works but interverting generates the crash.

Attached are the specs of my system. I am not sure what else to put. Just let me know if you need more.

Attached also is the script to reproduce the segmentation fault. There was a small typo that generated the warning. It doesn’t change anything regarding the segfault anyways…

cjfields commented 9 years ago

Original Redmine Comment Author Name: Florent Angly Original Date: 2007-10-23T14:39:31Z

Created an attachment (id=788) Script to reproduce the crash

cjfields commented 9 years ago

Original Redmine Comment Author Name: Florent Angly Original Date: 2007-10-23T14:40:04Z

Created an attachment (id=789) My system specs

cjfields commented 9 years ago

Original Redmine Comment Author Name: Chris Fields Original Date: 2007-10-23T15:13:54Z

Odd; still works for me (w/o the error now):

Getting score for ABC|9944760 -> ABC|9986984 = 300 Getting score for ABC|9986984 -> ABC|9944760 = 303

Not sure what’s going on, but it def. seems platform-dependent. I’ll have to leave it as open for now and let someone else run tests with a similar platform.

cjfields commented 9 years ago

Original Redmine Comment Author Name: Chris Fields Original Date: 2008-02-10T20:57:31Z

Possibly related to bug 1837 (all part of libsw).

cjfields commented 9 years ago

Original Redmine Comment Author Name: Chris Fields Original Date: 2008-02-10T21:00:34Z

(In reply to comment #10)

Possibly related to bug 1837 (all part of libsw).

Oops, should be bug 1857!

cjfields commented 9 years ago

Original Redmine Comment Author Name: Chris Fields Original Date: 2008-02-12T10:58:27Z

Pushing to the 1.7 release (bioperl-ext overhaul).

cjfields commented 9 years ago

Original Redmine Comment Author Name: Dave Messina Original Date: 2008-11-27T14:03:28Z

Comment already indicated this should be pushed to 1.7 — updating milestone to reflect that.

cjfields commented 9 years ago

Original Redmine Comment Author Name: Chris Fields Original Date: 2008-11-27T15:15:05Z

(In reply to comment #13)

Comment already indicated this should be pushed to 1.7 — updating milestone to reflect that.

It’s possible this was fixed with the last bioperl-ext patch a few days ago in svn r15030:

http://code.open-bio.org/svnweb/index.cgi/bioperl/browse/bioperl-ext/trunk/Bio/Ext/Align?rev=15030

It just needs further testing.

cjfields commented 9 years ago

Original Redmine Comment Author Name: Florent Angly Original Date: 2009-01-03T05:32:19Z

It still crashes on my side. The debug message is slightly different though (attached). That’s on Linux 2.6.27, with Perl 5.10.0 and today’s Bioperl svn.

cjfields commented 9 years ago

Original Redmine Comment Author Name: Florent Angly Original Date: 2009-01-03T05:35:06Z

Created an attachment (id=1183) GDB backtrace 2

cjfields commented 9 years ago

Original Redmine Comment Author Name: Chris Fields Original Date: 2010-08-21T09:46:49Z

(In reply to comment #15)

It still crashes on my side. The debug message is slightly different though (attached). That’s on Linux 2.6.27, with Perl 5.10.0 and today’s Bioperl svn.

I’m able to somewhat replicate this using both OS X 10.6 and Ubuntu 9.10. I can start looking into it, but it’s not top-priority on the list of bugs to be fixed b/c of the status of bioperl-ext.