kdesvn segfaults on update

Bug #35720 reported by Bertrand Croq
10
Affects Status Importance Assigned to Milestone
kdesvn (Ubuntu)
Fix Released
Medium
MOTU

Bug Description

I currently run Dapper.

When I run "kdesvn update" in a repository, it segfaults...

dpkg -l kdesvn:
||/ Name Version Description
+++-================-================-================================================
ii kdesvn 0.7.2-2 subversion client with tight KDE integration

I have tried to dpkg-buildpackage kdesvn, using the same version, and it works.

Bertrand Croq (bcroq)
Changed in kdesvn:
assignee: nobody → motu
Revision history for this message
Francois-Denis Gonthier (fdgonthier) wrote :

I don't even need to update. kdesvn crashes after I select an SVN-controled directory in File->Open.

(no debugging symbols found)
Using host libthread_db library "/lib/libthread_db.so.1".
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
[Thread debugging using libthread_db enabled]
[New Thread 46912570371040 (LWP 12343)]
[New Thread 1090525536 (LWP 12346)]
(no debugging symbols found)
(no debugging symbols found)
0x00002aaaac605fb2 in QEventLoop::activateTimers ()
   from /usr/lib/libqt-mt.so.3
#0 0x00002aaaac605fb2 in QEventLoop::activateTimers ()
   from /usr/lib/libqt-mt.so.3
#1 0x00002aaaac5b8b85 in QEventLoop::processEvents ()
   from /usr/lib/libqt-mt.so.3
#2 0x00002aaaac62baa2 in QEventLoop::enterLoop () from /usr/lib/libqt-mt.so.3
#3 0x00002aaaac62b9ab in QEventLoop::exec () from /usr/lib/libqt-mt.so.3
#4 0x00002aaaac612878 in QApplication::exec () from /usr/lib/libqt-mt.so.3
#5 0x000000000040b6ad in ?? ()
#6 0x00002aaaab1c849b in __libc_start_main () from /lib/libc.so.6
#7 0x000000000040b4ba in ?? ()
#8 0x00007fffffc9f968 in ?? ()
#9 0x000000000000001c in ?? ()
#10 0x0000000000000001 in ?? ()
#11 0x00007fffffca0899 in ?? ()
#12 0x0000000000000000 in ?? ()

Changed in kdesvn:
status: Unconfirmed → Confirmed
Revision history for this message
Francois-Denis Gonthier (fdgonthier) wrote :

I just rebuilt the kdesvn package locally and, whether I activate debug options or not, the resulting package *does not crash*. The problem then seem to be with how the package is built.

Revision history for this message
Francois-Denis Gonthier (fdgonthier) wrote :

gdb backtrace

#0 0x0000000000518e20 in ?? ()
#1 0x00002aaab179fc2a in ne_set_request_private () from /usr/lib/libneon.so.25
#2 0x00002aaab17a1795 in ne_request_dispatch () from /usr/lib/libneon.so.25
#3 0x00002aaab17a0e46 in ne_begin_request () from /usr/lib/libneon.so.25
#4 0x00002aaab17a1393 in ne_request_dispatch () from /usr/lib/libneon.so.25
#5 0x00002aaab11b7350 in svn_ra_dav__set_neon_body_provider ()
   from /usr/lib/libsvn_ra_dav-1.so.0
#6 0x00002aaab11b7a28 in svn_ra_dav__parsed_request_compat ()
   from /usr/lib/libsvn_ra_dav-1.so.0
#7 0x00002aaab11b2ea8 in svn_ra_dav__get_props ()
   from /usr/lib/libsvn_ra_dav-1.so.0
#8 0x00002aaab11b2fa6 in svn_ra_dav__get_props_resource ()
   from /usr/lib/libsvn_ra_dav-1.so.0
#9 0x00002aaab11b3206 in svn_ra_dav__search_for_starting_props ()
   from /usr/lib/libsvn_ra_dav-1.so.0
#10 0x00002aaab11b3475 in svn_ra_dav__get_baseline_props ()
   from /usr/lib/libsvn_ra_dav-1.so.0
#11 0x00002aaab11b3a93 in svn_ra_dav__get_baseline_info ()
   from /usr/lib/libsvn_ra_dav-1.so.0
#12 0x00002aaab11b3f69 in svn_ra_dav__do_check_path ()
   from /usr/lib/libsvn_ra_dav-1.so.0
#13 0x00002aaaafe3a36a in svn_client_status2 ()
   from /usr/lib/libsvn_client-1.so.0
#14 0x00002aaaafcabe67 in svn::Client::status ()
   from /usr/lib/kde3/libkdesvnpart.so
#15 0x00002aaaafc84106 in CheckModifiedThread::run ()
   from /usr/lib/kde3/libkdesvnpart.so
#16 0x00002aaaac60a822 in QThreadInstance::start () from /usr/lib/libqt-mt.so.3
#17 0x00002aaaae0b10fa in start_thread () from /lib/libpthread.so.0
#18 0x00002aaaab273ce2 in clone () from /lib/libc.so.6
#19 0x0000000000000000 in ?? ()

Revision history for this message
Francois-Denis Gonthier (fdgonthier) wrote :

The problem lies in how libkdesvnpart.so is built. There is a slight difference between the version produced by a local build an the version in the kdesvn package.

Here is, for proof, the difference between the output of objdump -x for both file, the left hand file being the package's libkdesvnpart.so.

< /usr/lib/kde3/libkdesvnpart.so: file format elf64-x86-64
< /usr/lib/kde3/libkdesvnpart.so
---
> /usr/lib/kde3/libkdesvnpart.so.own: file format elf64-x86-64
> /usr/lib/kde3/libkdesvnpart.so.own
10c10
< filesz 0x0000000000125682 memsz 0x0000000000125682 flags r-x
---
> filesz 0x0000000000125662 memsz 0x0000000000125662 flags r-x
15c15
< EH_FRAME off 0x00000000001041d0 vaddr 0x00000000001041d0 paddr 0x00000000001041d0 align 2**2
---
> EH_FRAME off 0x00000000001041b0 vaddr 0x00000000001041b0 paddr 0x00000000001041b0 align 2**2
106c106
< 11 .rodata 00008f4f 00000000000fb280 00000000000fb280 000fb280 2**5
---
> 11 .rodata 00008f2f 00000000000fb280 00000000000fb280 000fb280 2**5
108c108
< 12 .eh_frame_hdr 00003c64 00000000001041d0 00000000001041d0 001041d0 2**2
---
> 12 .eh_frame_hdr 00003c64 00000000001041b0 00000000001041b0 001041b0 2**2
110c110
< 13 .eh_frame 0000f9b4 0000000000107e38 0000000000107e38 00107e38 2**3
---
> 13 .eh_frame 0000f9b4 0000000000107e18 0000000000107e18 00107e18 2**3
112c112
< 14 .gcc_except_table 0000de96 00000000001177ec 00000000001177ec 001177ec 2**2
---
> 14 .gcc_except_table 0000de96 00000000001177cc 00000000001177cc 001177cc 2**2

Revision history for this message
Francois-Denis Gonthier (fdgonthier) wrote :

I hesitate to put this bug as major since only I have confirmed it. But this package is hardly usable as it is right now.

If a few person could take the time to test it, I'm sure someone could come up with a solution, but this bug is beyond me right now.

Revision history for this message
Bertrand Croq (bcroq) wrote :

The solution is quite simple: just rebuild it and then it won't crash.

Revision history for this message
Francois-Denis Gonthier (fdgonthier) wrote :

After consulting with some people on IRC, it looks that it'll be as simple as rebuilding the package.

Revision history for this message
Barry deFreese (bddebian) wrote :

Sending up for a rebuild. Please let me know if this succeeds. Thank you.

Changed in kdesvn:
status: Confirmed → In Progress
Revision history for this message
Francois-Denis Gonthier (fdgonthier) wrote :

It's now working fine here. I'll assume this problem is solved. Repoter: feel free to reopen this bug if it doesn't work for you. I don't use kdesvn much so I may not have stumbled on the right crash.

Changed in kdesvn:
status: In Progress → Fix Released
Revision history for this message
drx (drx) wrote :

I strongly suggest to port or backport the current 1.x release of kdesvn to hardy. I am geting segfaults again with KDE 3.5.10 when committing a file to the repository and heaps of crazy error dialogs telling me of UTF8 errors. The author Alwin wrote to me he is not supporting versions before 1.0 anymore.

And i think waiting for KDE4 to be ready is not a good option. :) kdesvn is a very handy tool.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.