You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Robert North <aq...@sneakemail.com> on 2002/12/09 15:39:45 UTC

Segmentation fault when not using -m option.

Just downloaded subversion (0.16) to see how it compared to CVS.

No sooner had I started using it, then I got segfaults of the following 
commands:

import, cp, mkdir.

I could avoid the segfaults by adding message with the -m option.

Is this a known bug?
I would expect so, but nothing seems to be up on the Issue tracking part 
of the web site.

Also, might this be fixed by the fix for issue 1027?

Stack trace for the import command follows:
(Note that the backtrace for cp is extremely similar. Other functions 
not debugged, yet).


(gdb) run import file:///usr/local/svnroot/fred
"/home/rob/Downloads/Wine/wine-20021125(with cvs)"
Starting program:
/home/rob/Downloads/subversion/subversion-r3987/subversion/clients/cmdline/.libs/svn
import file:///usr/local/svnroot/fred
"/home/rob/Downloads/Wine/wine-20021125(with cvs)"
[New Thread 1024 (LWP 9114)]

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1024 (LWP 9114)]
0x403b1b0b in strlen () from /lib/libc.so.6
(gdb) bt
#0  0x403b1b0b in strlen () from /lib/libc.so.6
#1  0x0805281d in svn_cl__get_log_message (log_msg=0xbffff14c,
tmp_file=0xbffff120, commit_items=0x80614d0,
    baton=0x8061458, pool=0x8060dd8) at
subversion/clients/cmdline/util.c:455
#2  0x4001cf7c in svn_client_import (commit_info=0xbffff1b0,
notify_func=0x804bfb0 <notify>,
    notify_baton=0x8061440, auth_baton=0x8061200,
    path=0x80613a8 "/home/rob/Downloads/Wine/wine-20021125(with cvs)",
    url=0x8061318 "file:///usr/local/svnroot/fred", new_entry=0x0,
    log_msg_func=0x8052560 <svn_cl__get_log_message>,
log_msg_baton=0x8061458, nonrecursive=0,
    pool=0x8060dd8) at subversion/libsvn_client/commit.c:573
#3  0x0804c74b in svn_cl__import (os=0x8060ef8, baton=0xbffff340,
pool=0x8060dd8)
    at subversion/clients/cmdline/import-cmd.c:122
#4  0x0804ea2d in main (argc=4, argv=0xbffff434) at
subversion/clients/cmdline/main.c:856
#5  0x403559ed in __libc_start_main () from /lib/libc.so.6

Many thanks
    -Rob.



---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: Patch policy for subversion beginner (was: Re: Segmentation fault when not using -m option.)

Posted by Karl Fogel <kf...@newton.ch.collab.net>.
Robert North <aq...@sneakemail.com> writes:
> This brings me to how I should start patching:
> I could either grab the latest subversion trunk from the subversion
> repository,
> Or do ad-hoc patches for each issue I trip over.
> 
> Which method would people suggest is best to maximise subversion stability?

You should always test the latest Subversion trunk to see if a bug is
still present.  If it is, make a patch and/or regression test; see the
HACKING file at the top of the tree for more details.

-K

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: Patch policy for subversion beginner (was: Re: Segmentation fault when not using -m option.)

Posted by ro...@animalcules.com.
On Mon, Dec 09, 2002 at 06:50:58PM +0000, Robert North wrote:

> This brings me to how I should start patching: I could either grab
> the latest subversion trunk from the subversion repository, Or do
> ad-hoc patches for each issue I trip over.

> Which method would people suggest is best to maximise subversion stability?

You should probably always start a bug hunt or a patch from the latest
version.  I can't think of any reasons to do otherwise.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Patch policy for subversion beginner (was: Re: Segmentation fault when not using -m option.)

Posted by Robert North <aq...@sneakemail.com>.
Garrett Rooney rooneg-at-electricjellyfish.net |Subversion list| wrote:

> Robert North wrote:
>
>> Just downloaded subversion (0.16) to see how it compared to CVS.
>>
>> No sooner had I started using it, then I got segfaults of the 
>> following commands:
>>
>> import, cp, mkdir.
>>
>> I could avoid the segfaults by adding message with the -m option.
>>
>> Is this a known bug?
>> I would expect so, but nothing seems to be up on the Issue tracking 
>> part of the web site.
>>
>> Also, might this be fixed by the fix for issue 1027?
>
>
>
> it's quite possible.  i never actually reproduced a crash for that 
> issue, but i imagine it would look somethign like that.  try adding 
> the fix (just change 'char *msg2;' to 'char *msg2 = NULL;' in 
> subversion/clients/cmdline/util.c) and rebuilding.
>
> -garrett

Thanks for the suggestion.
Rotering suggested much the same thing, but fixed it by grabbing a new 
subversion from the repository.

This brings me to how I should start patching:
I could either grab the latest subversion trunk from the subversion 
repository,
Or do ad-hoc patches for each issue I trip over.

Which method would people suggest is best to maximise subversion stability?
    -Rob.








---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: Segmentation fault when not using -m option.

Posted by Garrett Rooney <ro...@electricjellyfish.net>.
Robert North wrote:

> Just downloaded subversion (0.16) to see how it compared to CVS.
>
> No sooner had I started using it, then I got segfaults of the 
> following commands:
>
> import, cp, mkdir.
>
> I could avoid the segfaults by adding message with the -m option.
>
> Is this a known bug?
> I would expect so, but nothing seems to be up on the Issue tracking 
> part of the web site.
>
> Also, might this be fixed by the fix for issue 1027?


it's quite possible.  i never actually reproduced a crash for that 
issue, but i imagine it would look somethign like that.  try adding the 
fix (just change 'char *msg2;' to 'char *msg2 = NULL;' in 
subversion/clients/cmdline/util.c) and rebuilding.

-garrett


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org