Apache OpenOffice (AOO) Bugzilla – Issue 6253
Chart inside saved spreadsheet cause OO to crash
Last modified: 2013-08-07 15:12:27 UTC
I have reproduced this many times and can submit a file as proof if you need it. I have a spreadsheet with three tabs as follows: Tab1: Data and formulas. Tab2: Formulas refering to tab one data. Tab3: Charts referencing data and formulas on first two tabs. All works great when I create the charts. But, OO dies completely when I try to view the chart tab after I load it from file. Also, editing numbers in the ranges refered to by the chart kills it as well. Thanx, Terry Fehr
Yes, please submit a file so we can reproduce it too. Thanks
Created attachment 2136 [details] sample spreadsheet file showing problems
Hi, sorry, but I can not reproduce the crash. The document works very good. I saw the chart, could change the data range and so on. Sascha
Hi Terry, for me it works too. So I close it . Frank
I've tried to view my charted spreadsheet on a number of different systems. Some worked but most did not. So, I have include the debug info from crash dialogs. Is there any other information that you know of that I could provide to help resolve this issue? Have you tried it on a slower system <= 550Mhz? 1st: SOFFICE caused an exception 10H in module XO641MI.DLL at 016f:019d95e5. Registers: EAX=0093f0c8 CS=016f EIP=019d95e5 EFLGS=00010202 EBX=00000006 SS=0177 ESP=0093f098 EBP=02b0cf4c ECX=00000022 DS=0177 ESI=00000002 FS=2d07 EDX=7fffffff ES=0177 EDI=02fd50e8 GS=0000 Bytes at CS:EIP: dd 18 33 c0 85 c9 0f 94 c0 c3 90 6a ff 68 d0 c4 Stack dump: 01997e1a 0093f0c8 00000022 00000003 02fd50e8 02f9ebdc 02fd04a8 02ff2250 00000002 00000001 02fd5080 02fd04a8 02ff22a8 0093f10c 00596e60 00000003 2nd: SOFFICE caused an invalid page fault in module UCBHELPER1MSC.DLL at 016f:006f0063. Registers: EAX=01e86a90 CS=016f EIP=006f0063 EFLGS=00010206 EBX=026b7488 SS=0177 ESP=0093dc20 EBP=028fbb30 ECX=01e86a90 DS=0177 ESI=028acc78 FS=46b7 EDX=030d0000 ES=0177 EDI=2c4b66a9 GS=0000 Bytes at CS:EIP: 00 94 01 00 00 0b 30 10 30 15 30 1f 30 32 30 3f Stack dump: 1cb315cf 0093dd30 2c4b66a9 00002c4b 028acc78 028fbb30 024066a9 1e81ab04 0093dd30 2c4b66a9 00064dd2 02711540 0093de50 00000000 0093dd94 bff70001
Hi, can you tell us more about the different systems. On which does it work and on which not? I think the most interesting point is the OS and the hardware (memory, diskspace, graphic adapter and so on). Sascha
Created attachment 2146 [details] Zip of DrWatson logs from systems both working & problematic
Hi Sascha, as you have done most of the work now, I assign it to you. Frank
Created attachment 2169 [details] Text versions of Dr Watson logs
Hi, can you try to disable OpenGL in tools->options->Open Office->View->Use OpenGL. It could be, that there is a driver problem with your graphic cards. Thanks Sascha
I tried disabling Open GL as you suggested, but no luck. BTW, I also tried disabling the video hardware acceleration and I tried setting the options->spreadsheet->view->chart to show none, but it still failed. Thanks, Terry Fehr
I tried new version 1.0.1 today and the same problem still exists, I was hoping perhaps some other resolved issue had the same root cause. I have been upgrading the versions quite faithfully and I remember having no problems when I used the old StarOffice 6Beta. I'm not sure, but I believe the first release of OpenOffice also worked, but I don't have it anymore to test. Thot it might help you track it down a little easier. Thanx, Terry Fehr
I finally narrowed it down somewhat. I have found, by trial and error, that having blank cell contents in the chart's datarange cause OOo to crash on my machine. But why it works on some machines and not on others, is still a mystery. Thanx again, Terry Fehr
Hi, sounds like a problem for you. Sascha
I couldn't reproduce it yet with Linux and Solaris. I will check the Windows version. Probably this is a memory overwrite bug. Just as info: Blank cells in the chart are modelled by using DBL_MIN which is about 2e+-308. But I doubt that this makes problems.
I found this in the MS knowledge base (http://support.microsoft.com/default.aspx?scid=kb;en-us;150314) about the exception you get: 16 (10h): Coprocessor Error Fault This interrupt occurs when an unmasked floating-point exception has signaled a previous instruction. (Because the 80386 does not have access to the Floating Point unit, it checks the ERROR\ pin to test for this condition.) This is also triggered by a WAIT instruction if the Emulate Math Coprocessor bit at CR0 is set. This supports at least the thesis that it may have to do with empty cells (DBL_MIN = 2.2250738585072014e-308 is involved). But I still have no idea why this crash occurs. It looks like the systems where the bug occurs (AMD K6, Intel Pentium II) are slightly older than the working systems (AMD Athlon, Pentium III ?)
As we could not reproduce this bug yet, I suppose it only occurs on some older systems with Windows 98. In the new chart implementation scheduled for OOo 2.0 we will no longer use DBL_MIN for empty cells, so maybe this problem also vanishes.
state RESOLVED/LATER is deprecated. However, the bug will be resolved later due to the change to a new data format that no longer uses DBL_MIN.
accepted
fehretic, please try with version 2.0.2 and confirm issue.
Fixed with the reimplementation of the chart. NaN-values are now NaNs and not DBL_MINs, so this issue should no longer occur. ->fehretic, if you still have problems with this in OOo 2.3 (or a preview like a developer snapshot from http://download.openoffice.org/680/index.html ), please re-open.
Closing.