Bug 524493 - [abrt] prelinking causes octave to crash
Summary: [abrt] prelinking causes octave to crash
Keywords:
Status: CLOSED WORKSFORME
Alias: None
Product: Fedora
Classification: Fedora
Component: octave
Version: 13
Hardware: x86_64
OS: Linux
low
medium
Target Milestone: ---
Assignee: Alex Lancaster
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard: abrt_hash:d4c82c0f32f11cd241b058cb111...
: 526833 540795 (view as bug list)
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2009-09-20 19:39 UTC by Georg Wild
Modified: 2011-06-02 18:03 UTC (History)
8 users (show)

Fixed In Version: octave-3.2.4-2.fc13
Clone Of:
Environment:
Last Closed: 2011-06-02 17:51:01 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)
File: backtrace (1.42 KB, text/plain)
2009-09-20 19:39 UTC, Georg Wild
no flags Details
octave version from /usr/bin, which does not work (28.15 KB, application/octet-stream)
2009-10-07 07:53 UTC, Georg Wild
no flags Details
octave version from package, which works (5.53 KB, application/octet-stream)
2009-10-07 07:53 UTC, Georg Wild
no flags Details
/tmp/log from prelink (119.22 KB, application/octet-stream)
2009-10-10 10:32 UTC, Georg Wild
no flags Details
always use the correct version in the prelink workaround (2.19 KB, patch)
2010-02-28 11:20 UTC, Michal Schmidt
no flags Details | Diff

Description Georg Wild 2009-09-20 19:39:33 UTC
abrt detected a crash.


How to reproduce
-----
1.
2.
3.


Additional information
======


Attached files
----
backtrace

cmdline
-----
octave 


component
-----
octave


executable
-----
/usr/bin/octave-3.2.2


kernel
-----
2.6.31-33.fc12.x86_64


package
-----
octave-6:3.2.2-4.fc12


reason
-----
Process was terminated by signal 11

Comment 1 Georg Wild 2009-09-20 19:39:36 UTC
Created attachment 361834 [details]
File: backtrace

Comment 2 Alex Lancaster 2009-09-21 00:46:27 UTC
Can you please describe what triggered the crash.  Did it crash immediately upon execution of the octave command-line?

Comment 3 Georg Wild 2009-09-21 10:17:37 UTC
Yes, it happens all time I start octave and even if I change user!

Example:
[georg@asusnb octave]$ octave
Segmentation fault (core dumped)

Comment 4 Rakesh Pandit 2009-09-21 10:58:00 UTC
I am not able to reproduce it.

[rakesh@test-box ~]$ uname -a
Linux dhcp1-96.pnq.redhat.com 2.6.31-33.fc12.x86_64 #1 SMP Thu Sep 17 15:40:43 EDT 2009 x86_64 x86_64 x86_64 GNU/Linux
[rakesh@test-box ~]$ rpm -qa octave
octave-3.2.2-4.fc12.x86_64

May you paste the bactrace again after installing debuginfo ?

This bractrace does not help much.

Thanks,

Comment 5 Georg Wild 2009-09-21 12:17:01 UTC
Find attached a part of gdb session:

gdb octave
(gdb) run
Starting program: /usr/bin/octave 
[Thread debugging using libthread_db enabled]

Program received signal SIGSEGV, Segmentation fault.
__static_initialization_and_destruction_0 (this=<value optimized out>) at MDiagArray2.cc:51
51	template <class T> MDiagArray2<T> MDiagArray2<T>::nil_array;
Current language:  auto
The current source language is "auto; currently c++".
(gdb) thread apply all bt full

Thread 1 (Thread 0x7f87f86f6720 (LWP 12167)):
#0  0x000000344006920e in __static_initialization_and_destruction_0 (this=<value optimized out>) from /usr/lib64/octave-3.2.2/liboctave.so
No locals.
#1  global constructors keyed to MArray_s.cc(void) (this=<value optimized out>) at MArray-s.cc:48
No locals.
#2  0x0000003440565686 in __do_global_ctors_aux () from /usr/lib64/octave-3.2.2/liboctave.so
No symbol table info available.
#3  0x000000343fd8aefb in _init () from /usr/lib64/octave-3.2.2/liboctave.so
No symbol table info available.
#4  0x00007fff02634f98 in ?? ()
No symbol table info available.
#5  0x000000301c20ebc9 in call_init (l=0x7f87f8743000, argc=1083188016, argv=0x7fff02634f88, env=0x7fff02634f98) at dl-init.c:70
        init_array = <value optimized out>
#6  0x000000301c20ed4f in _dl_init (main_map=0x301c4210e8, argc=1, argv=0x7fff02634f88, env=0x7fff02634f98) at dl-init.c:134
        preinit_array = <value optimized out>
        preinit_array_size = 0x0
        i = 1
#7  0x000000301c200b2a in _dl_start_user () from /lib64/ld-linux-x86-64.so.2
No symbol table info available.
#8  0x0000000000000001 in ?? ()
No symbol table info available.
#9  0x00007fff0263542c in ?? ()
No symbol table info available.
#10 0x0000000000000000 in ?? ()
No symbol table info available.

Comment 6 Alex Lancaster 2009-09-21 20:11:57 UTC
Do you by any chance have a .octaverc file? i.e. in ~/.octaverc ?  It's possible it is executing some code there which makes it hard to reproduce.

Also, do you have octave-forge also installed?  If so, please uninstall it and try to reproduce the crash, that way we may be able to isolate it to octave-forge.

Comment 7 Rakesh Pandit 2009-09-22 03:47:27 UTC
I can reproduce it if I run octave as a root. Is it recommended to do so ?

I don't have octave-forge installed so it an octave issue.

Investigating ...

Comment 8 Rakesh Pandit 2009-09-22 06:38:01 UTC
Well .. I am too much confused. I removed my package .. reinstalled it and now I am not able to reproduce again. I was though able to do it earlier.

Tried building with source again and reproduce but no luck. Back to square one.

@ Georg

May you check what Alex suggested + remove it and reinstall and confirm whether you can re produce ?

Thanks,

Comment 9 Georg Wild 2009-09-22 12:29:36 UTC
After reinstalling I cannot reproduce this any more.

Let me mention that before reinstalling I was able to start octave when I run it within valgrind!

Comment 10 Alex Lancaster 2009-09-22 18:13:03 UTC
(In reply to comment #9)
> After reinstalling I cannot reproduce this any more.

Closing bug.  Feel free to re-open if you can reproduce this.

Comment 11 Georg Wild 2009-10-07 07:52:10 UTC
In the meantime I know sth more about this bug.

I realized that /usr/bin/octave-3.2.2 has been somehow changed.
In /tmp/octave-3.2.2 I have the original, which works, while /usr/bin/octave-3.2.2 doesn't work.
-rwxr-xr-x. 1 georg georg  5664 2009-09-23 09:06 /tmp/octave-3.2.2
-rwxr-xr-x. 1 root  root  28824 2009-09-23 09:06 /usr/bin/octave-3.2.2

I'll attach both /tmp/octave-3.2.2 as well as /usr/bin/octave-3.2.2.

Looking at objdump -x /usr/bin/octave-3.2.2 one observes as a difference for example .gnu.prelink_undo.

Is this bug related to prelinking?

Comment 12 Georg Wild 2009-10-07 07:53:20 UTC
Created attachment 363932 [details]
octave version from /usr/bin, which does not work

Comment 13 Georg Wild 2009-10-07 07:53:56 UTC
Created attachment 363933 [details]
octave version from package, which works

Comment 14 Alex Lancaster 2009-10-07 17:37:03 UTC
I believe prelinking (which runs as a cron job) can modify the original binaries in the package.  Perhaps something happened during prelinking to corrupt the originally shipped binary?

Comment 15 Georg Wild 2009-10-10 10:30:49 UTC
I can now reproduce the bug:

1. sudo yum reinstall octave
2. octave 
     -> works
3. sudo prelink -v /usr/bin/octave-3.2.2  >/tmp/log
4. octave
     -> Segmentation fault (core dumped)

I'll attach /tmp/log!

Comment 16 Georg Wild 2009-10-10 10:32:15 UTC
Created attachment 364338 [details]
/tmp/log from prelink

Comment 17 Michal Schmidt 2009-10-27 17:25:21 UTC
*** Bug 526833 has been marked as a duplicate of this bug. ***

Comment 18 Alex Lancaster 2009-10-27 21:03:54 UTC
Looks like this is a real problem.  We should bring this up with upstream, perhaps other distros (e.g. Debian/Ubuntu) have also had to solve the problem.

Comment 19 Rakesh Pandit 2009-11-10 13:33:37 UTC
Thanks Alex, is this with 3.2.2 only or 3.2.3 also. I could not reproduce on my machine with 3.2.3 (thanks Alex for your work in pushing it ;-).

Comment 20 Rakesh Pandit 2009-11-10 13:36:31 UTC
I am running F11 between .. will try on rawhide.

Comment 21 Michal Schmidt 2009-11-12 12:42:37 UTC
Same problem here on F12 with octave-3.2.3-1.fc12.x86_64. "prelink -u /usr/bin/octave-3.2.3" is a workaround.

Until the real solution is found, would you please add a file /etc/prelink.conf.d/octave.conf with a single line content:
-b /usr/bin/octave-3.2.3
to the package in order to blacklist the octave binary from prelinking?

Comment 22 Alex Lancaster 2009-11-15 08:26:19 UTC
OK, working on a workaround to this along the lines of comment #21.  Need to ultimately investigate this with upstream, so will leave this bug open after updates pushed.

Comment 23 Fedora Update System 2009-11-15 11:10:12 UTC
octave-3.2.3-2.fc12.1 has been submitted as an update for Fedora 12.
http://admin.fedoraproject.org/updates/octave-3.2.3-2.fc12.1

Comment 24 Fedora Update System 2009-11-20 05:29:57 UTC
octave-3.2.3-2.fc12.1 has been pushed to the Fedora 12 stable repository.  If problems still persist, please make note of it in this bug report.

Comment 25 Alex Lancaster 2009-11-24 07:39:54 UTC
*** Bug 540795 has been marked as a duplicate of this bug. ***

Comment 26 Alex Lancaster 2009-12-11 03:43:25 UTC
This should have been fixed by octave-3.2.3-2.fc12.1, so closing.  If you are having problems with crashes that aren't related to the pre-linking issue in this bug, please open a new bug.

Comment 27 Michal Schmidt 2010-02-28 11:20:25 UTC
Created attachment 396857 [details]
always use the correct version in the prelink workaround

The bug is back in F-13 with octave-3.2.4 because the prelink workaround was version-specific. The attached patch fixes it.

Comment 28 Alex Lancaster 2010-02-28 21:44:08 UTC
(In reply to comment #27)
> Created an attachment (id=396857) [details]
> always use the correct version in the prelink workaround
> 
> The bug is back in F-13 with octave-3.2.4 because the prelink workaround was
> version-specific. The attached patch fixes it.    

Thanks for the patch!   I'll look into applying to rawhide and rawhide-branch right away.

Comment 29 Fedora Update System 2010-03-01 00:11:18 UTC
octave-3.2.4-2.fc13 has been submitted as an update for Fedora 13.
http://admin.fedoraproject.org/updates/octave-3.2.4-2.fc13

Comment 30 Alex Lancaster 2010-03-01 00:13:49 UTC
(In reply to comment #27)
> Created an attachment (id=396857) [details]
> always use the correct version in the prelink workaround
> 
> The bug is back in F-13 with octave-3.2.4 because the prelink workaround was
> version-specific. The attached patch fixes it.    

OK, new version ready to be added to branch-rawhide (see comment #29), Michal, please test it, add your +1 karma to the bodhi update and I'll be sure to submit to branched-rawhide stable ASAP.

Comment 31 Fedora Update System 2010-03-01 01:41:12 UTC
octave-3.2.4-2.fc13 has been pushed to the Fedora 13 testing repository.  If problems still persist, please make note of it in this bug report.
 If you want to test the update, you can install it with 
 su -c 'yum --enablerepo=updates-testing update octave'.  You can provide feedback for this update here: http://admin.fedoraproject.org/updates/F13/FEDORA-2010-3188

Comment 32 Michal Schmidt 2010-03-01 09:26:30 UTC
I've added +1 in Bodhi.

Alex, have you discussed the bug with upstream already? I can't find it in octave mailing lists.

Comment 33 Alex Lancaster 2010-03-01 22:41:07 UTC
(In reply to comment #32)
> I've added +1 in Bodhi.

Thanks.

> Alex, have you discussed the bug with upstream already? I can't find it in
> octave mailing lists.    

Not specifically, I haven't had a chance yet.  Feel free to report it to the mailing lists and post a link to the thread here, if I don't get a chance first.

Comment 34 Fedora Update System 2010-03-03 01:48:14 UTC
octave-3.2.4-2.fc13 has been pushed to the Fedora 13 stable repository.  If problems still persist, please make note of it in this bug report.

Comment 35 Jan Kratochvil 2010-05-09 07:16:58 UTC
This bug is no fixed and there is no new bug tracking this problem.
Blacklisting in prelink is only a workaround of some octave bug, not a fix.

Comment 36 Alex Lancaster 2010-07-18 21:02:52 UTC
(In reply to comment #35)
> This bug is no fixed and there is no new bug tracking this problem.
> Blacklisting in prelink is only a workaround of some octave bug, not a fix.    

I filed a bug upstream, and Cc'ed you on it:

https://savannah.gnu.org/bugs/index.php?30478

Comment 37 Alex Lancaster 2010-07-19 05:59:44 UTC
(In reply to comment #35)
> This bug is no fixed and there is no new bug tracking this problem.
> Blacklisting in prelink is only a workaround of some octave bug, not a fix.    

Jan, can you reproduce with a full backtrace and submit that backtrace here, and/or on the bug upstream?  Thanks.

Comment 38 Jan Kratochvil 2010-07-19 08:02:01 UTC
I do not have the problem reproducible.  I found this bug as I have seen /etc/prelink.conf.d/octave.conf which is not a valid solution for a crasher.

Posting some debugging notes to the upstream bug, thanks for filing it.

Comment 39 Bug Zapper 2011-06-02 17:42:54 UTC
This message is a reminder that Fedora 13 is nearing its end of life.
Approximately 30 (thirty) days from now Fedora will stop maintaining
and issuing updates for Fedora 13.  It is Fedora's policy to close all
bug reports from releases that are no longer maintained.  At that time
this bug will be closed as WONTFIX if it remains open with a Fedora 
'version' of '13'.

Package Maintainer: If you wish for this bug to remain open because you
plan to fix it in a currently maintained version, simply change the 'version' 
to a later Fedora version prior to Fedora 13's end of life.

Bug Reporter: Thank you for reporting this issue and we are sorry that 
we may not be able to fix it before Fedora 13 is end of life.  If you 
would still like to see this bug fixed and are able to reproduce it 
against a later version of Fedora please change the 'version' of this 
bug to the applicable version.  If you are unable to change the version, 
please add a comment here and someone will do it for you.

Although we aim to fix as many bugs as possible during every release's 
lifetime, sometimes those efforts are overtaken by events.  Often a 
more recent Fedora release includes newer upstream software that fixes 
bugs or makes them obsolete.

The process we are following is described here: 
http://fedoraproject.org/wiki/BugZappers/HouseKeeping

Comment 40 Georg Wild 2011-06-02 17:51:01 UTC
This bug seems fixed in FC15 GNU Octave, version 3.4.0


Note You need to log in before you can comment on or make changes to this bug.