[bt-devel] [ bibletime-Bugs-1646286 ] malloc/free crash when creating search indices

SourceForge.net noreply at sourceforge.net
Sun Sep 2 23:38:45 MST 2007


Bugs item #1646286, was opened at 2007-01-28 06:11
Message generated for change (Comment added) made by mgruner
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=100954&aid=1646286&group_id=954

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: Frontend / Bookshelf
Group: in progress
>Status: Closed
>Resolution: Fixed
Priority: 8
Private: No
Submitted By: David Ormand (dlormand)
Assigned to: Martin Gruner (mgruner)
Summary: malloc/free crash when creating search indices

Initial Comment:
Program starts fine.  Select "Settings", then "Bookshelf Manager", then "Manager Search Indices", select a Bible text, then "Create Selected Indices".

Sometimes the error is:
*** glibc detected *** malloc(): memory corruption (fast): 0x10547d78 ***
*** BibleTime got signal 6 (Crashing). Trying to save settings.
Alarm clock

Sometimes the error is:
*** glibc detected *** malloc(): memory corruption: 0x1049a8d0 ***
*** BibleTime got signal 6 (Crashing). Trying to save settings.
Alarm clock

(different pointer values every time, of course.)

Doesn't seem to matter if the user is root or not, or where the text modules are located (/usr/local/share/sword or ~/.sword).

malloc failures seem to associate with KJV and MKJV, and free failures seem to associate with RSV and ISV.

BibleTime version 1.6.2
Qt version 3.3.6
running KDE 3.5.4 on PowerMac G4 with 2.6.18.3 kernel.
Slackintosh 11 distro


----------------------------------------------------------------------

>Comment By: Martin Gruner (mgruner)
Date: 2007-09-03 08:38

Message:
Logged In: YES 
user_id=169722
Originator: NO

Closing.

----------------------------------------------------------------------

Comment By: Martin Gruner (mgruner)
Date: 2007-04-19 08:07

Message:
Logged In: YES 
user_id=169722
Originator: NO

Hello everybody.

Can you please try again with the new BibleTime 1.6.4? We fixed what we
believe may have caused the error (unfortunately, we could not reproduce it
ourselves. Please let us know if the bug is fixed now.

Thanks, mg

----------------------------------------------------------------------

Comment By: Joachim Ansorg (joachim)
Date: 2007-03-14 22:35

Message:
Logged In: YES 
user_id=1486
Originator: NO

The current CVS of BibleTime contains a fix for a bug which might be
related to this one.
Could all of you who know how to use CVS test this for us?

Thank you!

----------------------------------------------------------------------

Comment By: Nobody/Anonymous (nobody)
Date: 2007-03-02 21:32

Message:
Logged In: NO 

I have the same problem.  When I try to search something (the KJV, for
example) I get the message that whatever I am trying to search "One or more
modules needs to be indexed. This may take a long time. Do you wish to
continue?" 

When I click "OK" the Bibletime crashes.  I have very few modules
installed, but it does not matter what I try to search, I get the same
message and crashes.

I am running Freespire.

----------------------------------------------------------------------

Comment By: Nobody/Anonymous (nobody)
Date: 2007-03-02 21:31

Message:
Logged In: NO 

I have the same problem.  When I try to search something (the KJV, for
example) I get the message that whatever I am trying to search "One or more
modules needs to be indexed. This may take a long time. Do you wish to
continue?" 

When I click "OK" the Bibletime crashes.  I have very few modules
installed, but it does not matter what I try to search, I get the same
message and crashes.

I am running Freespire.

----------------------------------------------------------------------

Comment By: Martin Gruner (mgruner)
Date: 2007-02-20 18:25

Message:
Logged In: YES 
user_id=169722
Originator: NO

Please keep digging! I can't reproduce this on my machine...
Thanks for all info so far.

----------------------------------------------------------------------

Comment By: David Ormand (dlormand)
Date: 2007-02-20 15:36

Message:
Logged In: YES 
user_id=1703490
Originator: YES

Except for the few apps that I installed from the disk (notably KDE),
everything on my machine is self-compiled.  I've had no troubles at all
with anything else.
Bibletime itself runs fine, it's just the index formation that crashes.
Note that I'm not running "slack", I'm running a slack-like minor
distribution for PowerPC, reportedly built from the same source as Slack
11.
I don't know how backtrace in gdb works, but if there's a limited buffer,
and it's being filled with what looks like recursion in the libc thread
functions, is it possible that the root-cause Blbietime trace is being
pushed out of the buffer?  Or is backtrace dumping the stack of a thread,
instead of Bibletime itself?  Noob furiously guessing...
The problem exists in both 1.6.2 and 1.6.3b on my machine.

----------------------------------------------------------------------

Comment By: Nobody/Anonymous (nobody)
Date: 2007-02-20 05:14

Message:
Logged In: NO 

Linspire 5.1.427. Just upgraded to 1.6.3. Select KJV and attempt to do a
search. Comes up and says that index needs to be built and that this will
take some time. Tell it OK. 4 hours later, TOP shows that BT is taking very
little CPU. Shut it down and restart, and... Same thing, it needs to index
KJV. Basically nothing happens in an hour. I don't think it is a SLACK
problem.

----------------------------------------------------------------------

Comment By: Nobody/Anonymous (nobody)
Date: 2007-02-16 04:33

Message:
Logged In: NO 

Just to reiterate what I said in the other thread, I can create indices
fine with BibleTime 1.6.3b on Slackware 11. I'm running KDE 3.5.4 & kernel
2.6.17.13.
--benjie

----------------------------------------------------------------------

Comment By: Nobody/Anonymous (nobody)
Date: 2007-02-16 02:01

Message:
Logged In: NO 

mgruner: You wrote, "It seems that there are more slack users with this
problem. Might be a
distro issue? [...] This doesn't really look like a BibleTime problem. Do
you have
trouble with other self-compiled software as well?"

I don't know about dmorland, but I myself don't have much trouble with
other self-compiled programs. But considering that creating search indices
works just fine in BT 1.6.2 on my distro (Slack 11), I figure it must be
something to do with whatever changed in BT 1.6.3b that is causing the
crash. If it were a distro problem, wouldn't it also have to crash in all
other BT versions as well?

----------------------------------------------------------------------

Comment By: Martin Gruner (mgruner)
Date: 2007-02-15 18:13

Message:
Logged In: YES 
user_id=169722
Originator: NO

Mumble.... This doesn't really look like a BibleTime problem. Do you have
trouble with other self-compiled software as well? Nothing in your
backtrace pointed to something in BibleTime....

----------------------------------------------------------------------

Comment By: Martin Gruner (mgruner)
Date: 2007-02-15 17:54

Message:
Logged In: YES 
user_id=169722
Originator: NO

It seems that there are more slack users with this problem. Might be a
distro issue?
http://sourceforge.net/tracker/index.php?func=detail&aid=1657055&group_id=954&atid=100954

----------------------------------------------------------------------

Comment By: David Ormand (dlormand)
Date: 2007-02-11 06:19

Message:
Logged In: YES 
user_id=1703490
Originator: YES

running as user.                                                          
     bash-3.1$ bibletime
(BibleTime 1.6.3b) WARNING:
removeRecursive(/home/david/.kde/share/apps/bibletime/indices//MKJV)
(BibleTime 1.6.3b) WARNING:
removeRecursive(/home/david/.kde/share/apps/bibletime/indices/MKJV/standard)
(BibleTime 1.6.3b) WARNING: ASSERT: "windows.count()" in
bibletime_slots.cpp (466)
(BibleTime 1.6.3b) WARNING: QColor::setHsv: HSV parameters out of range
  (several of these messages, more appear as the mouse runs over the
menu)

Invoke crash.  The following occurs:
bc detected *** malloc(): memory corruption: 0x1074ac98 ***
*** BibleTime got signal 6 (Crashing). Trying to save settings.
Alarm clock                                                               
     
Run in gdb.
(gdb) run
Starting program: /opt/kde/bin/bibletime
[Thread debugging using libthread_db enabled]
[New Thread 16384 (LWP 2670)]
(BibleTime 1.6.3b) WARNING:
removeRecursive(/home/david/.kde/share/apps/bibletime/indices//MKJV) 
(BibleTime 1.6.3b) WARNING:
removeRecursive(/home/david/.kde/share/apps/bibletime/indices/MKJV/standard)
(BibleTime 1.6.3b) WARNING: ASSERT: "windows.count()" in
bibletime_slots.cpp (466)
(BibleTime 1.6.3b) WARNING: QColor::setHsv: HSV parameters out of range
(BibleTime 1.6.3b) WARNING: QColor::setHsv: HSV parameters out of range
  (several more of these messages, more appear as mouse runs over the
menu)

Invoke crash.  The following occurs:

. . . .
(BibleTime 1.6.3b) WARNING: QColor::setHsv: HSV parameters out of range
(BibleTime 1.6.3b) WARNING: ASSERT: "installSource" in
cswordsetupmodulelistview.cpp (70)
(BibleTime 1.6.3b) WARNING: QColor::setHsv: HSV parameters out of range
(BibleTime 1.6.3b) WARNING: QColor::setHsv: HSV parameters out of range
. . .

*** glibc detected *** malloc(): memory corruption: 0x101caef8 ***

Program received signal SIGABRT, Aborted.
[Switching to Thread 16384 (LWP 2670)]
0x0d7f2788 in kill () from /lib/libc.so.6

output of "backtrace":
#0  0x0d7f2788 in kill () from /lib/libc.so.6
#1  0x0e149d20 in pthread_kill () from /lib/libpthread.so.0
#2  0x0e14a160 in raise () from /lib/libpthread.so.0
#3  0x0d7f2548 in raise () from /lib/libc.so.6
#4  0x0d7f3cbc in abort () from /lib/libc.so.6
#5  0x0d827b98 in __libc_message () from /lib/libc.so.6
#6  0x0d827b98 in __libc_message () from /lib/libc.so.6
. . . . (duplicate messages)
#55 0x0d827b98 in __libc_message () from /lib/libc.so.6
#56 0x0d827b98 in __libc_message () from /lib/libc.so.6
Previous frame inner to this frame (corrupt stack?)

Under the suspicion that something is going on with
multithreading, I attempted to reconfigure without threading
("./configure --disable-threading").  configure
failed to find Qt!  Has no problem with plain old "./configure"!


----------------------------------------------------------------------

Comment By: Martin Gruner (mgruner)
Date: 2007-02-10 18:06

Message:
Logged In: YES 
user_id=169722
Originator: NO

Hi David, please attach comments to the bug report. I pasted you email.

Well, to generate a stack backtrace the best thing probably to do is to
run bibletime in gdb.
IIRC, you need to execute

gdb /usr/bin/bibletime (or wherever it is)

and then inside gdb

run

and then when the crash occurs

backtrace.

Please look at the documentation, I might well be wrong here.

To see more useful information in the backtrace, you can compile BT
without optimization and with debugging symbols. Use

configure --disable-optimize (not sure about this one) --enable-debug

To configure and then compile BibleTime.

Thanks for your assistance.

mg

----------------------------------------------------------------------

Comment By: Nobody/Anonymous (nobody)
Date: 2007-02-10 18:02

Message:
Logged In: NO 

FYI

----------  Weitergeleitete Nachricht  ----------

Subject: Re: [ bibletime-Bugs-1646286 ] malloc/fr
Date: Samstag, 10. Februar 2007
From: "David Ormand" <dlormand at users.sourceforge.net>
To: mgruner at users.sourceforge.net


Message body follows:

>Can you please test if this also happens with 1.6.3? If so,
please provide
>more details (stack backtrace).

Yep, still does it.  That's 1.6.3b, right?

Can you advise how to generate a stack backtrace?  I figured
it would leave a core file somewheres, but I can't find it!
 I understand such data can be extracted from core files,
but I've never done it personally.

Also, is this the appropriate way to process a bug report
with you, or is there a more "official" method via sourceforge?


--
This message has been sent to you, a registered SourceForge.net user,
by another site user, through the SourceForge.net site.  This message
has been delivered to your SourceForge.net mail alias.  You may reply
to this message using the "Reply" feature of your email client, or
using the messaging facility of SourceForge.net at:
https://sourceforge.net/sendmessage.php?touser=1703490

-------------------------------------------------------

----------------------------------------------------------------------

Comment By: Martin Gruner (mgruner)
Date: 2007-02-09 09:44

Message:
Logged In: YES 
user_id=169722
Originator: NO

Can you please test if this also happens with 1.6.3? If so, please provide
more details (stack backtrace).

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=100954&aid=1646286&group_id=954



More information about the bt-devel mailing list