You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by SteveKing <st...@gmx.ch> on 2005/04/30 07:27:24 UTC

Analysation of TSVN crashreports

Hi,

with the fixed bug in Subversion 'svn status -u crash w/ moved external' 
I went through all the crashdumps people sent in for TSVN (and which I 
still had in my inbox) again. Karl asked me to send my findings to the 
list, so here they are:

9 crashdumps analyzed
2 crashs occured in TSVN itself
5 crashs because of that 'svn status -u' bug (now that I know what to 
look for, I can assign a crashdump to that specific bug much easier)
2 crashs with the following stacktrace (for 1.1.4):

subversion/libsvn_wc/lock.c, line 621 :
   if (associated->set) // with associated as 0!!
subversion/libsvn_wc/entries.c, line 711
subversion/libsvn_wc/props.c, line 279
subversion/libsvn_client/diff.c, line 1168
subversion/libsvn_client/repos_diff.c, line 817
subversion/libsvn_delta/cancel.c, line 237
subversion/libsvn_ra_svn/editorp.c, line 688
subversion/libsvn_ra_svn/editorp.c, line 794
subversion/libsvn_ra_svn/editor.c, line 730
subversion/libsvn_ra_svn/client.c, line 382
subversion/libsvn_client/diff.c, line 1393
subversion/libsvn_client/diff.c, line 2491

Unfortunately I can't provide much more information about those crashes. 
All I found out until now is that this happens with a dry-run merge 
where not just file contents but property contents changed.
The drawback of the automatic crashdump tool TSVN uses is that *all* 
people can send in those crashdumps. And many of them don't understand 
english (or german) so when I try to ask them about more information 
they can't answer because they don't even understand my question. Or 
they just don't answer my mails (language problems/SPAM filter/...).

But at least the crash which caused most of the sent in crashdumps is 
fixed now so I'm very happy about that!

Stefan

-- 
        ___
   oo  // \\      "De Chelonian Mobile"
  (_,\/ \_/ \     TortoiseSVN
    \ \_/_\_/>    The coolest Interface to (Sub)Version Control
    /_/   \_\     http://tortoisesvn.tigris.org

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

Re: Analysation of TSVN crashreports

Posted by SteveKing <st...@gmx.ch>.
Philip Martin wrote:

>> From what I can see on that issue, I think that's the same bug we're
>> talking about here.
> 
> I don't know about issue 2254, but I think I know how to reproduce
> your bug, can you determine if this produces the stacktrace you are
> seeing:
> 
> svnadmin create repo
> svn mkdir $REPO/foo $REPO/foo/bar $REPO/foo2 -m ""
> touch Makefile
> svn import Makefile $REPO/foo/bar/zig -m ""
> svn co $REPO/foo wc
> svn ps x x wc/bar/zig
> svn ci wc -m ""
> svn sw $REPO/foo2 wc
> svn merge --dry-run -r2:3 $REPO/foo wc
> 
> The merge crashes if I use 1.1.x but it works if I use trunk, I
> suspect it got fixed as a side effect of r12025.

The crash produces the exact same stacktrace I sent you. That means you 
really got it cornered, thanks!

I also tried with TSVN trunk (built against Subversion 1.2RC2). It 
doesn't crash there when I do a dry-run. (also not with a real merge).

Stefan

-- 
        ___
   oo  // \\      "De Chelonian Mobile"
  (_,\/ \_/ \     TortoiseSVN
    \ \_/_\_/>    The coolest Interface to (Sub)Version Control
    /_/   \_\     http://tortoisesvn.tigris.org

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

Re: Analysation of TSVN crashreports

Posted by Philip Martin <ph...@codematters.co.uk>.
SteveKing <st...@gmx.ch> writes:

> kfogel@collab.net wrote:
>> SteveKing <st...@gmx.ch> writes:
>>
>>>subversion/libsvn_wc/lock.c, line 621 :
>>>   if (associated->set) // with associated as 0!!
>>>subversion/libsvn_wc/entries.c, line 711
>>>subversion/libsvn_wc/props.c, line 279
>>>subversion/libsvn_client/diff.c, line 1168
>>>subversion/libsvn_client/repos_diff.c, line 817
>>>subversion/libsvn_delta/cancel.c, line 237
>>>subversion/libsvn_ra_svn/editorp.c, line 688
>>>subversion/libsvn_ra_svn/editorp.c, line 794
>>>subversion/libsvn_ra_svn/editor.c, line 730
>>>subversion/libsvn_ra_svn/client.c, line 382
>>>subversion/libsvn_client/diff.c, line 1393
>>>subversion/libsvn_client/diff.c, line 2491
> [snip]
>> This may be related to
>>    http://subversion.tigris.org/issues/show_bug.cgi?id=2254
>> (Paul Querna, any progress on that reproduction recipe?)
>
> Yes, the stacktrace looks exactly the same!

Really?  The stack trace in issue 2254 makes no sense to me, it
appears to have svn_client_relocate calling svn_client_delete, and
svn_client_delete calling svn_wc_merge_prop_diffs.

>  From what I can see on that issue, I think that's the same bug we're
>  talking about here.

I don't know about issue 2254, but I think I know how to reproduce
your bug, can you determine if this produces the stacktrace you are
seeing:

svnadmin create repo
svn mkdir $REPO/foo $REPO/foo/bar $REPO/foo2 -m ""
touch Makefile
svn import Makefile $REPO/foo/bar/zig -m ""
svn co $REPO/foo wc
svn ps x x wc/bar/zig
svn ci wc -m ""
svn sw $REPO/foo2 wc
svn merge --dry-run -r2:3 $REPO/foo wc

The merge crashes if I use 1.1.x but it works if I use trunk, I
suspect it got fixed as a side effect of r12025.

-- 
Philip Martin

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

Re: Analysation of TSVN crashreports

Posted by SteveKing <st...@gmx.ch>.
kfogel@collab.net wrote:
> SteveKing <st...@gmx.ch> writes:
> 
>>subversion/libsvn_wc/lock.c, line 621 :
>>   if (associated->set) // with associated as 0!!
>>subversion/libsvn_wc/entries.c, line 711
>>subversion/libsvn_wc/props.c, line 279
>>subversion/libsvn_client/diff.c, line 1168
>>subversion/libsvn_client/repos_diff.c, line 817
>>subversion/libsvn_delta/cancel.c, line 237
>>subversion/libsvn_ra_svn/editorp.c, line 688
>>subversion/libsvn_ra_svn/editorp.c, line 794
>>subversion/libsvn_ra_svn/editor.c, line 730
>>subversion/libsvn_ra_svn/client.c, line 382
>>subversion/libsvn_client/diff.c, line 1393
>>subversion/libsvn_client/diff.c, line 2491
[snip]
> 
> This may be related to
> 
>    http://subversion.tigris.org/issues/show_bug.cgi?id=2254
> 
> (Paul Querna, any progress on that reproduction recipe?)

Yes, the stacktrace looks exactly the same!
 From what I can see on that issue, I think that's the same bug we're 
talking about here.

Stefan

-- 
        ___
   oo  // \\      "De Chelonian Mobile"
  (_,\/ \_/ \     TortoiseSVN
    \ \_/_\_/>    The coolest Interface to (Sub)Version Control
    /_/   \_\     http://tortoisesvn.tigris.org

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

Re: Analysation of TSVN crashreports

Posted by kf...@collab.net.
SteveKing <st...@gmx.ch> writes:
> subversion/libsvn_wc/lock.c, line 621 :
>    if (associated->set) // with associated as 0!!
> subversion/libsvn_wc/entries.c, line 711
> subversion/libsvn_wc/props.c, line 279
> subversion/libsvn_client/diff.c, line 1168
> subversion/libsvn_client/repos_diff.c, line 817
> subversion/libsvn_delta/cancel.c, line 237
> subversion/libsvn_ra_svn/editorp.c, line 688
> subversion/libsvn_ra_svn/editorp.c, line 794
> subversion/libsvn_ra_svn/editor.c, line 730
> subversion/libsvn_ra_svn/client.c, line 382
> subversion/libsvn_client/diff.c, line 1393
> subversion/libsvn_client/diff.c, line 2491
> 
> Unfortunately I can't provide much more information about those
> crashes. All I found out until now is that this happens with a dry-run
> merge where not just file contents but property contents changed.
> The drawback of the automatic crashdump tool TSVN uses is that *all*
> people can send in those crashdumps. And many of them don't understand
> english (or german) so when I try to ask them about more information
> they can't answer because they don't even understand my question. Or
> they just don't answer my mails (language problems/SPAM filter/...).

This may be related to

   http://subversion.tigris.org/issues/show_bug.cgi?id=2254

(Paul Querna, any progress on that reproduction recipe?)

-Karl

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