Created attachment 110488 [details] Crash stacktrace Hi guys, I'm having this bug for last couple of version. I'm receiving bank statement in RTF format, generated from Oracle Reports, and each time I tried to open it with Writer, it will crash it down. In attachment is a stacktrace I was able to get, however, due the secrecy of the matter I'm not able to provide a document itself. LibreOffice version is 4.3.4, but the crash was present in most of 4.x series. I'll be happy to give you more details if necessary. Thanks!
Forgot to mention, unrtf tool will open it without problems.
Do you reproduce this with a brand new report (so without confidential elements in it)? On which Linux distribution are you? Would it possible you install LO debug package (in Debian and derivates like Ubuntu it's "libreoffice-dbg") and retrieve a more useful backtrace?
Created attachment 110496 [details] Crash stacktrace from debug LO
Yes, I can reproduce this on all reports, including the new one I got this morning. > On which Linux distribution are you? Slackware 14.1 x86_64, but can reproduce it on Ubuntu 14.4 LTS with LO 4.2.7.2. I added stacktrace from libreoffice-dbg package, installed on Ubuntu.
Thank you for your bt, I put it at NEW. If you have some time, it would be interesting to have a bt with symbols with LO 4.3.4 (see https://launchpad.net/~libreoffice/+archive/ubuntu/ppa).
Created attachment 110510 [details] 4.3.4 LO debug stacktrace Sure, here is stacktrace from 4.3.4 debug version.
Please could you try another Oracle report in RTF format like here: bug 59547. With this file I do not have a crash with version 4.4.0.0.beta2+ built at home under Ubuntu 14.10 x86-64. Bets regards. JBF
Just tried it and it is opening without problems. Although after couple of scrolls, it became unresponsive due large CPU usage, just as described in the report.
Thank you Sanel for this new feedback it seems the bt could be here line 4532 (4520 + 12 from d74f3f2a1b98fbcba9e5490b6d790fa40d8b863d, the other commits are after this block): 4529 case RTF_DPPTY: 4530 { 4531 RTFDrawingObject& rDrawingObject = m_aStates.top().aDrawingObject; 4532 if (rDrawingObject.aPolyLinePoints.hasElements()) 4533 { Miklos: could rDrawingObject.aPolyLinePoints or even rDrawingObject be NULL? (and so should be checked) Of course, the root cause must be elsewhere.
I don't think we can do anything about this unless we have an example. If you want to email an example to me I can have a look into it for you.
I'm really sad to see this state: you are faster to make this task invalid than to proceed to *at least* try to solve it. Have you tried to get more information? Do you need more information? Have you tried to contact me in the past without success and decided to mark it as invalid? Do you have some references where is shown that you tried everything, after long discussion with other participants here? I really don't want to be rude here, especially since other devs were really helpful, but I doubt RH would like to see that LO is not able to parse seemingly complex RTF without crashing the whole suite. What you want me to do then? Sit down and debug myself, maybe fix this thing and you receive all praises then? I mean, at least you *could* try to help here... BTW. requesting me to send you confidential document, so *you can look at it*, really tells me how truly you want to solve this thing.
@Sanel, so edit the document to remove the confidential bits, replacing the text with nonsense, but leaving enough stuff so it still crashes LO. Then send that. It's not very much to ask.
(In reply to Sanel Zukan from comment #11) > I'm really sad to see this state: you are faster to make this task invalid > than to proceed to *at least* try to solve it. Sometimes the bt isn't sufficient, we need to reproduce the bug. In this case, it's quite normal to have the document. > Have you tried to get more information? Do you need more information? Have > you tried to contact me in the past without success and decided to mark it > as invalid? Do you have some references where is shown that you tried > everything, after long discussion with other participants here? There are too much bugs badfully and dev time is scarce resource. > ... > What you want me to do then? Sit down and debug myself, maybe fix this thing > and you receive all praises then? I mean, at least you *could* try to help > here... Caolan suggested to email your file in private, that's all. > BTW. requesting me to send you confidential document, so *you can look at > it*, really tells me how truly you want to solve this thing. I think you judge him a little too fast, just take a look at this: http://cgit.freedesktop.org/libreoffice/core/log/?qt=author&q=caolan So if he proposed you to take a look to your file, he meaned it. Now if you want to debug yourself this bug, you can start here https://wiki.documentfoundation.org/Development, then submit a patch and if it's ok, your commit will show your name (so you will get the praises :-))