Bug 83910 - Loading DOCX causes hanging (loop in frmform.cc)
Summary: Loading DOCX causes hanging (loop in frmform.cc)
Status: RESOLVED FIXED
Alias: None
Product: LibreOffice
Classification: Unclassified
Component: Writer (show other bugs)
Version:
(earliest affected)
Inherited From OOo
Hardware: All All
: high critical
Assignee: Tobias Lippert
URL:
Whiteboard: target:5.1.0
Keywords:
Depends on:
Blocks:
 
Reported: 2014-09-16 00:15 UTC by Peter Silva
Modified: 2022-05-13 20:15 UTC (History)
4 users (show)

See Also:
Crash report or crash signature:


Attachments
Sample crasher document, many more at URL listed (179.13 KB, application/msword)
2014-09-16 00:15 UTC, Peter Silva
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Peter Silva 2014-09-16 00:15:46 UTC
Created attachment 106348 [details]
Sample crasher document, many more at URL listed

http://cbs-ext2014.wmo.int/documents-english
roughly two thirds of the documents there, when loaded, cause libreoffice to hang.  For example, this one: CBS-Ext(2014)-d03-2(1)-WIS-Approved_en.docx
(Attached)

tested with abiword, and it hangs also.  But Kingsoft WPS opens it OK, it gives an error message about font loading, but presents the document anyways.
Comment 1 Peter Silva 2014-09-16 00:19:27 UTC
libreoffice installed on ubuntu 14.04. claims to be 4.2.6.3 (choice not offerred in pull-downs.)  reproduced all the time, on two different machines with multiple documents.
Comment 2 penttila 2014-09-16 14:22:22 UTC
I can confirm this issue. When opening the attached file, the first page displays but writer doesn't response any more!

LinuxMint 17 Cinnamon (based on Ubuntu 14.04)
LO  4.2.6.3 Build id.: 420m0(Build:3)
Comment 3 Terrence Enger 2014-09-16 18:56:20 UTC
Digging around in gdb, I see a pretty busy loop, possibly *the* busy
loop, in SwTxtFrm::_Format, source file
sw/source/core/text/frmform.cxx, from line 1427 to line 1526.

One point in one iteration of the loop is in SwTxtIter::GetLineNr,
source sw/source/core/text/itrtxt.hxx:87, where rLine.GetlineNr() is
1, which is less than nOrhpLines, which is 2, which makes the function
return false, which is the exit condition of the loop in
SwTxtFrm::_Format.

These observations are on master commit f74a633, fetched 2014-08-23, configured:
    --enable-option-checking=fatal --enable-dbgutil --enable-crashdump
    --without-system-postgresql --without-myspell-dicts
    --with-extra-buildid --without-doxygen
    --with-external-tar=/home/terry/lo_hacking/git/src
    --enable-online-update
built and running on debian-wheezy 64-bit.
Comment 4 Tobias Lippert 2015-06-17 19:08:15 UTC
I can confirm the bug on current LO on Fedora 22. (4.4.3.2)
I will look into it.
@Peter - Thanks for providing the test document.
Comment 5 Tobias Lippert 2015-06-17 19:10:46 UTC
The bug also exists on the current master
Comment 6 Tobias Lippert 2015-06-17 20:20:57 UTC
I did a callgrind analysis.
The problem seems to be
SwTxtFrm::FormatLine which gets called quite often on line
frmform.cxx:1484
I will try to find out what the code is supposed to do, and how to optimize this.
Comment 7 Marek Dolezel 2015-07-12 20:13:45 UTC
Is this linux only? Repro on Ubuntu 15.04 x86-64 

version: 4.4.2.2
build ID: 40m0(Build:2)
locale: cs_CZ
Comment 8 Terrence Enger 2015-07-12 21:51:55 UTC
I still see the loop in daily dbgutil bibisect repo version
2015-07-10.

On Windows Vista, LibreOffice ...
    Version: 5.1.0.0.alpha1+
    Build ID: d3b6f3790953bdfeaeebcd3ba9ec370d94ca4ebf
    TinderBox: Win-x86@39, Branch:master, Time: 2015-07-09_00:11:56
    Locale: en-CA (en_CA)
renders the top windowful of the document then loops.  I guess that
this is the same loop, so am setting O/S All.
Comment 9 Tobias Lippert 2015-08-06 20:06:02 UTC
It seems as if the loop starting in frmform.cc:1427 does not terminate.
(if I force to terminate it after, e.g., 100 iterations, the document loads correctly)

To fix the bug, I need to understand the logic of the aforementioned loop which used to identify line breaks. It is quite complicated, but I am working on it.
Comment 10 Tobias Lippert 2015-09-27 19:35:27 UTC
I have added a fix which is currently waiting for review at  https://gerrit.libreoffice.org/18896
With the fix, the document loads correctly
Comment 11 Tobias Lippert 2015-11-02 20:12:13 UTC
The fix is still waiting for review. I will post an update when it has been merged.
Comment 12 Commit Notification 2015-11-04 14:31:05 UTC
Tobias Lippert committed a patch related to this issue.
It has been pushed to "master":

http://cgit.freedesktop.org/libreoffice/core/commit/?id=20538f233fe120b33a23d594458d4639b0c9670e

tdf#83910 Formatting of lines which consist of a single dummy line only

It will be available in 5.1.0.

The patch should be included in the daily builds available at
http://dev-builds.libreoffice.org/daily/ in the next 24-48 hours. More
information about daily builds can be found at:
http://wiki.documentfoundation.org/Testing_Daily_Builds

Affected users are encouraged to test the fix and report feedback.
Comment 13 Timur 2015-11-12 17:08:06 UTC
I guest it's fixed, I can open the document.