Apache OpenOffice (AOO) Bugzilla – Full Text Issue Listing
|Summary:||Open Impress Presentation on ATI Crash PC with BluePage!|
|Status:||CLOSED FIXED||QA Contact:||issues@graphics <issues>|
|Priority:||P2||CC:||andre.schnabel, bc.openoffice.org, flibby05, floeff+ooo, issues, Joost.Andrae, jr, julisys, manens, mnagashree, ooo, pagalmes.lists, pavel, r.y.yates-01, rainerbielefeld_ooo_qa, samael, thb, tk, urska|
|Version:||680m95||Keywords:||crash, needhelp, oooqa|
|Target Milestone:||OOo 2.3|
|Issue Type:||DEFECT||Latest Confirmation in:||---|
Description falorca 2005-04-29 14:16:10 UTC
After install the OpenOffice, i tryed to open a presentation and when click on slide show, the presentation is to slow and them crash the pc an appear a ble page from windowsxp telling me that was a problem with ati2vga driver. There's no problem with the driver because i've tried with some programs the openGL, etc and everything is ok! The microsoft office 2003 doens't have this problem so i think there's a problem with openoffice 680m95!
Comment 1 wolframgarten 2005-05-02 08:13:18 UTC
Comment 2 spyresca 2005-05-06 07:23:42 UTC
I have verified this on two separate machines. Windows XP SP2, one with a radeon 9000 128mb and one with an x300 based radeon. On Both machines with build 680m100, certain transitions are not only so slow that they nearly lock the machine, but they also cause blue screens of death on a regular basis. I also have office 2003 on these two machines and powerpoint doesn't exhibit this behaviour. The radeons are using the latest reference catalyst drivers from ATI. This is bad to the point that it's a showstopper for me in terms of using impress for anything. Even on small shows of 2-3 slide with no graphics, I can easily reproduce these blue screens and incredible slowdowns nearly every time.
Comment 3 allroad 2005-05-09 21:45:10 UTC
This issue sounds as though it could be related to issue 45877. Even with build 1.9 m100 presentations can be sluggish - check the Use OpenGL box (see issue 45877) and the problem seems to go away. My PC is a laptop with an ATI 7500 mobile and is using the driver supplied by Sony - the makers of the laptop. Otherwise my box has 2.0 pentium 4 - no HT 256 megs of ram and has Win XP home SP2. I know that my card is much older than those mentioned here - but I have not had any bsod probs.
Comment 4 spyresca 2005-05-10 03:40:00 UTC
Well, I was aware of the other issue and while it may be related, setting the open gl settings (tested on fresh installs of m100 on two different computers) still can reliably reproduce blue screens of death. This is especially easy to do with using any of the "wipe" transitions. I have also tried changing the memory settings and making sure my catalyst drivers (most recent reference drivers from ATI) are set to defaults. Still no help. The wipe transitions put the ATI cards in both computers into some sort of endless loops that kills them with BSOD. Specifically, the error message seen is something along the lines of "ATI2dvag display driver has stopped working". Then the computer either BSOD's or has to be rebooted since the display driver is completely killed leaving 640x480 display modes with totally messed up colors. No other program (office or otherwise) has caused this issue with the ATI display drivers. It seems to be specific to the beta builds of OO 2.0. Most of the Dell's in my office use ati video/drivers and every time I try it on one of them I get the same or a very similar result. With these cards being so popular, this is a deal breaker for me in getting OO accepted in my office at this time.
Comment 5 allroad 2005-05-10 08:36:13 UTC
Glad you knew about the issue 45877. Not sure about the BSoD problem - but do remember OOo 2.0 is still at beta stage so some things still need ironing out. As an aside, I would not try using 2.0 beta to convince others to turn to OpenOffice - it is a little too unstable IMHO and you could put people off for good - especially when it eats some important work (happened to me). A far better bet would be version 1.1.4 - which is excellent too - or wait until 2.0 is at a production/stable level. In the mean time you can keep an eye on this issue. Have you hunted around on the WWW for others with Ati cards and similar problems with open office?
Comment 6 christian.guenther 2005-05-10 10:47:56 UTC
Set to new and change the target.
Comment 7 christian.guenther 2005-05-10 10:51:12 UTC
I can reproduce the bug with my radeon 7200. I can only reproduce it with the slide transition uncover ... (e.g. uncover up) Please have a look.
Comment 8 thb 2005-05-10 11:00:03 UTC
Reproducible, bluescreen tells something about ati2dvag DLL, and "device driver infinite loop".
Comment 9 allroad 2005-05-10 12:43:43 UTC
Confirm also that issue can be reproduced - ATI mobile 7500 - same message re ati2dvag DLL. Did not crash PC but sent graphics driver haywire - I was able to close down applications and re-boot normally. uncover up - transition cause problem - presentation ran very slowly on the first slide and then, before the slide was fully uncovered the graphics diver complained and terminated the presentation - even if it did not close down Impress.
Comment 10 thb 2005-05-11 19:27:43 UTC
Could somebody please verify whether http://www.abxzone.com/forums/archive/index.php/t-28206.html is in any way related to this problem (FWICT, it _seems_ to be about non-Intel chipsets and/or CPUs, which would probably rule out spyresca for his Dells, but who knows)? What would be specifically interesting is whether stress-testing the hardware by a recent 3D game would trigger similar issues, and whether maybe the proposed solution in the forum changes something. Apart from that: is anyone willing to try out ATI's latest beta drivers (Catalyst6.4 or somesuch), just to make sure we've something substantial to report to ATI (because that's what I'm going to do with this issue - crashing a kernel device driver from a user space program MustNotBePossible(tm). Besides that, I'm just innocently moving around some bitmaps via plain ole Windows GDI, when this BSOD happens - dunno how or where to avoid that without insight into the driver internals).
Comment 11 allroad 2005-05-12 17:47:08 UTC
thb - I followed the link and then followed the advice on the forum - re-booted and fired up an impress presentation with uncover up and uncover down transitions and ran the slide presentation. Result: No crash and presentation ran fine, with only a slight slowdown on the 5th slide. Then I undid the change recommended in the forum and re-ran the presentation - with no problems.... Sorry, this is not much help - but it does demonstrate that the bug can be cleared up - don't know how it has sorted itself out though. At least my graphics card seems to cause no problems. Over to those with newer ATI cards and newer drivers.
Comment 12 thb 2005-05-12 18:11:29 UTC
@allroad: Thanks for checking that. CGU has just tried the same, but only succeeded after installing the latest Catalyst driver (5.4 AFAIK). Then, except for some performance problems, all went fine 'til now. @spyresca: Could you try that, too? If this works for three different people, I'd like to resolve this issue as an ATI driver bug (spawning a performance issue, if that persists, of course).
Comment 13 spyresca 2005-05-14 07:19:21 UTC
I'm already running the latest catalyst drivers on my home and work machines (5.4). Tried the fix mentioned in that forums link and also tried turning opengl on/off. Still no joy. Consistently get BSOD when doing a small 4 slide show using the troublesome transitions (wipes, etc.) I installed 1.1.4 and the exact same transitions work smoothly, perfectly without BSOD. I also run many of the videogames that the gentleman was having problems with in that forums thread and have no issues with them at all. The fact that 1.1.4 works great under the same exact conditions on the same exact machines makes me feel this is an issue with 2.0 beta (m100) and not an ati driver issue. I use tons of 3d stuff (games, 3d design using opengl, etc.) and so far only openoffice 2.0 (m100) has caused this issue.
Comment 14 falorca 2005-05-16 20:37:37 UTC
I unistall the ati 5.4 drivers and install previous drivers like ( 5.1, 5.2 and 5.3 without ControlCenter) and the problem stills the same.
Comment 15 thb 2005-05-16 23:25:18 UTC
@spyresca: That's bad news indeed. Of course, at lot has changed since 1.1.4, regarding the way our slideshow is rendered (the slideshow subsystem is, more or less, a complete rewrite). Still, a user-mode program must not be able to crash a display driver - as a matter of fact, we'll have to involve ATI here, either to fix their drivers, or to tell us what _exactly_ to avoid.
Comment 16 thb 2005-05-31 17:13:35 UTC
The easiest way to reproduce this crash, BTW, is to open the 'slide transition' task pane to the right, and quickly thumb through all available slide transitions (typically, already the preview of the effect triggers the crash).
Comment 17 thb 2005-07-27 20:04:36 UTC
*** Issue 50949 has been marked as a duplicate of this issue. ***
Comment 18 bencoman 2005-07-31 12:37:25 UTC
SYSTEM ONE (S1) + Sapphire Radeon 9600 XT (RV360) Catalyst Version 5.5; 2D Version 22.214.171.12446 + AMD Athlon XP2500+ SYSTEM TWO (S2) + Matrox G450 + Intel P4 2.4Ghz BOTH SYSTEMS + Windows XP SP 2 + OpenOffice.org 1.9.118 (installed today all defaults) Trying out the use case given above on S1... + Double-clicked transparents_solucom.ppt to open file in Impress + Clicked <Slide Show> in toolbar > Doesn't crash, but is it usual that the transition "Uncover Down Medium" takes eight seconds to reach half way and then instantaneous for for the remaining half? My use case... + Start > OpenOffice.org 1.9.118 > OpenOffice.org Impress + File > New > Presentation + Select "From template" "Introducing a New Product" <Create> + Right-click "Slide 1" > Slide Transition + Select transition "Uncover Down" and Speed=Medium + Click <Slide Show> in toolbar > "Uncover Down Medium" transition takes eight seconds to make it half way down screen, then instantaneous the rest of the way. + Select transition Speed=Fast > "Uncover Down Fast" transition moves consistently from top to bottom over 5 seconds + Select transition Speed=Slow > "Uncover Down Slow" transition takes 10 seconds to make it half way down screen, then instantaneous the rest of the way. Trying the http://www.abxzone.com fix on S1... + View > Devices By Connection... > shows BEN1 > ACPI Uniprocessor PC > Microsoft ACPI-Compliant System > PCI bus > VIA CPU to AGP Controller > RADEON 9600 Series + Start > My Computer right click > Manage > Device Manager > System Devices + VIA CPU to AGP Controller > Update Driver ? Can Windows connect to Windows Update to search for software ? Select (No, not this time) <Next> + Select (Install from a list of specific location) <Next> + Select (Don't search. I will choose the driver to install) <Next> + Select "PCI standard PCI-to-PCI bridge" <Next> + <Finish> + View > Devices By Connection... > shows BEN1 > ACPI Uniprocessor PC > Microsoft ACPI-Compliant System > PCI bus > PCI Standard PCI-to-PCI bridge > RADEON 9600 Series + Redid My Use Case > same results. Trying My Use Case on S2 + Redid My Use Case > "Uncover Down Medium" transition smoothly moves down screen in two seconds On S1 problem fixed by using plain old VGA with same RADEON card... + Uninstalled ATI Drivers. Rebooted. Redid My Use Case using plain VGA driver. > "Uncover Down Medium" transition smoothly moves down screen in two seconds On S1 try latest RADEON drivers.... + Installed Microsoft .NET Framework 1.1 (pre-requisite for Catalyst 5.7) + Installed Catalyst 5.7 Display Drivers, WDM Drivers and Catalyst Control Center (5-7_xp-2k_dd_ccc_wdm_enu_24085.exe) + Redid My Use Case > "Uncover Down Medium" transition takes eight seconds to make it half way down screen, then instantaneous the rest of the way. > now I've just noticed in the Catalyst SMARGART panel that AGPspeed, FastWrite, AGPwrite, AGPread are all OFF. + Reverted "PCI standard PCI-to-PCI bridge" back to "VIA CPU to AGP bridge" > now AGPspeed=4x, AGPwrite and AGPread are ON, although FastWrite is still OFF. + Installed "VIA 4 in 1 Drivers 4.46Vfor Win98SE/2K/ME/NT4.0/XP" version 4.46 from http://support.asus.com/download/download.aspx?SLanguage=en-us > FastWrite is now also turned ON + View > Devices By Connection... > shows BEN1 > ACPI Uniprocessor PC > Microsoft ACPI-Compliant System > PCI bus > VIA CPU to AGP2.0/AGP3.0 > RADEON 9600 Series + Redid My Use Case > "Uncover Down Medium" transition takes eight seconds to make it half way down screen, then instantaneous the rest of the way. Try the OpenGL fix mentioned above... + Tools > Options > OpenOffice.org > View > 3D View > Use OpenGL = YES + Redid My Use Case > "Uncover Down Medium" transition takes eight seconds to make it half way down screen, then instantaneous the rest of the way. Attached system summaries... 48454-bencoman-RADEON9600XT-system-summary.txt 48454-bencoman-G450-system-summary.txt
Comment 19 bencoman 2005-07-31 12:39:14 UTC
Created attachment 28372 [details] 48454-bencoman-RADEON9600XT-system-summary.txt
Comment 20 bencoman 2005-07-31 12:39:39 UTC
Created attachment 28374 [details] 48454-bencoman-G450-system-summary.txt
Comment 21 thb 2005-08-01 00:01:27 UTC
@bencoman: crap. that must have been frustrating, thanks for pursuing that to the limit! Unfortunately, I haven't heard anyting back from ATI (contacted them, even personally, in early June). Thus, maybe the people affected by this issue (this includes me) should consider going public via the OOo mailing list - might be there's someone listening who is affiliated to ATI.
Comment 22 thb 2005-08-01 00:03:22 UTC
Added bencoman to cc list, IZ did not seem to send a notification mail for my latest update.
Comment 23 jkirchner 2005-08-03 09:23:04 UTC
I have exactly the same problem, I use OOo 1.9.118. I created a presentation on my desktop computer (Intel P4 HT, WinXP) and it worked there without any problems. When I transferred it to my notebook (Intel Centrino Mobile Technology, ATI Mobility Raddeon 9700, Win XP) to show the presentation, all customised animations on a slide work very well, but many of the "next-slide" animations take a lot of time and do not work probably and take a lot of time, especially "Uncover Down Fast". Uncover Down Fast even causes a crash of the graphic card diver (ati2vga driver), so that a system reboot is necessary. I'd like to mention that OOo 1.1.x works without any problems! "uncover Down Fast" is already included in OOo1.1.x and it works on my notebook computer pretty well! So this seems to be a OOo 2 problem!
Comment 24 thb 2005-08-03 10:36:27 UTC
@jkirchner: your conclusion is problematic, since your preposition is wrong (it assumes that only because OOo1.x does not trigger problems in the graphics driver, it must be working correctly). In fact, NO application should EVER be able to crash the display driver. At any rate, as long as I don't get in-depth information from the vendor about where the problem lies, there's little chance for a properfix this on my side. Going back to the way OOo1.x did it is also not a (feasible) option, because that way induced various little bugs (like missing repaints, one-pixel-wide lines left on the screen, etc.).
Comment 25 thb 2005-08-19 12:32:16 UTC
Adding this issue to CWS canvas02, and setting to fixed, because: canvas02 implements various little optimizations, that affect the way the screen is updated during slideshows. On the machine I've tested it, this _seems_ to mask the problems mentioned here. Note: once CWS canvas02 is ready for QA, I'll try to get a build placed somewhere publicly accessible, such that all people that spoke up in this issue could verify for themselves.
Comment 26 thb 2005-09-12 11:09:49 UTC
*** Issue 49999 has been marked as a duplicate of this issue. ***
Comment 27 thb 2005-10-04 11:05:21 UTC
*** Issue 49674 has been marked as a duplicate of this issue. ***
Comment 28 wolframgarten 2005-10-25 10:11:34 UTC
Reassigned. re-open issue and reassign to email@example.com
Comment 30 wolframgarten 2005-10-25 10:12:11 UTC
reset resolution to FIXED
Comment 31 christian.guenther 2005-10-26 15:36:10 UTC
I reopen the issue
Comment 32 christian.guenther 2005-10-26 15:37:27 UTC
I can still reproduce the crash (blue screen) therefore back to thb
Comment 33 thb 2005-10-28 09:29:35 UTC
Some notes on the status: things have improved for some of the slidetransitions, but there are four transitions left that still crash some ATI drivers (the 'uncover' effects). I'm clueless about the root cause of this problem, thus, I'd really need the help of somebody familiar with the internals of Windows display drivers (preferably from ATI, of course).
Comment 34 falorca 2005-10-28 18:19:21 UTC
Today i've tried the slidetransitions, 'uncover' effects, with the official drivers from ATI (Catalyst 5.10) and there's a problem! I don't have the BSOD, but the slide takes too long to change, until middle of the screen, and then is very quick! I'm think this isn't right, but...
Comment 35 ooo 2005-11-04 16:05:24 UTC
problem seems to be not solvable without ATI support; retargeting to OOo 2.0.2
Comment 36 ooo 2005-11-18 10:19:59 UTC
retargeted due to still missing contact at ATI
Comment 37 wolframgarten 2006-02-01 09:52:25 UTC
*** Issue 61073 has been marked as a duplicate of this issue. ***
Comment 38 spyresca 2006-02-01 16:52:16 UTC
I found a sytesms setting change that seems to alleviate this issues underwindows XP (with an ATI card and catalyst drivers), but at cost of often screwing up overall systems performance (not to mention causing issues with other software). I'm hesitant to mention this very kludgey and often detrimental settings change as I fear the devs will say "use this kludge" and not fix the actual problem here. And I think this issue is really important and needs to be properly fixed for a very simple reason. We (and many many other corporations )buy lots and lots of shiny new dell computers for our employees. Many if not most of these machines use some kind of ATI graphics solution. As such, these bright, shiny new machines cannot be used out of the box with Oo Impress. If users try to use even the simplest of transitions, they are faced with unacceptable issues including blue screens of death. As it is, I see lots of "blame ATI!" posts whereupon similar functionality was not a probem under the 1.x branch of Openoffice.org using the same ATI equipment and drivers. I realize the imaging system has likely been changed in 2.x, but that doesn't change the fact that competing products provide similar functionality using the same hardware/driver combo. This tells me this issue can likely be fixed/worked around at some level below the need for users to manually change systems settings in a way that can negatively affect their computers overall performance.
Comment 39 thb 2006-02-02 08:41:50 UTC
@spyres: right - but, I've already tried all 'obvious' differences in rendering between 1.x and 2.0, to no avail. OOo consists of over 7 million lines of code, of which a two-figure amount has been touched for 2.0, so, it's just infeasible to systematically track down the change that causes this problem. Thus, basically, our only way to fix this on the OOo side of things is to revert the slideshow to the 1.x state - which a lot of people will probably hate us for (mind that this is a Windows-only, ATI-only problem). All I need is _somebody_ from ATI to contact us about this, and give us a clue about _what exactly_ to avoid on the GDI API - or simply fix their drivers (it's not that we haven't tried, via various channels, to get a response - maybe you as a hardware customer can push something here, through Dell or directly).
Comment 40 bencoman 2006-02-03 19:11:24 UTC
Hi Spyres, I'm a user with this problem, and at the moment I'm stuck. The problem isn't getting fixed now!, since only so much canbe done about a crash in a driver without the support of the manufacturer. I'd like to see your solution :):), and see for myself how it affects my machine. If its as bad as you say, then OOo on ATI still looks bad, slowing its uptake. I'm sure the devs want a proper solution too.
Comment 41 bencoman 2006-02-19 08:34:29 UTC
(clutching at straws....) I'm not sure if this is relevant and I don't know how to test it, but I just came across this... " I remember once, the magic flag when I have problems with swing apps: -Dsun.java2d.noddraw=true. This works for me. Don't ask me why ... (should be something with DirectDraw and my ATI card?) " ref http://sourceforge.net/forum/forum.php?thread_id=1429056&forum_id=320014
Comment 42 thb 2006-02-20 10:31:22 UTC
@bencoman: sorry, not relevant here. That applied to Java. And even indirectly (from what it disables in Java), this won't change things, since we don't even use DirectDraw in this case.,
Comment 43 christian.guenther 2006-03-06 09:46:54 UTC
*** Issue 60514 has been marked as a duplicate of this issue. ***
Comment 44 andreschnabel 2006-03-09 06:46:02 UTC
raised priority and changed summary.
Comment 45 christian.guenther 2006-03-10 08:29:11 UTC
*** Issue 62947 has been marked as a duplicate of this issue. ***
Comment 46 jr 2006-03-10 09:09:49 UTC
Hi all, I can reproduce this issue too (WinXP, Radeon X300), regardless of with transition I choose. I do not have this affect, if I disable the hardware acceleration. But (may be) good news: I tried to contact someone from ati and had success :-) ATM we are on CeBIT and I try to arrange a meeting with him and tbo. May be that helps. Jacqueline
Comment 47 jr 2006-03-10 09:53:39 UTC
Sorry, of course I meant thb :-)
Comment 48 thb 2006-03-14 13:44:53 UTC
@jrahemipour: any further news on that?
Comment 49 christian.guenther 2006-03-23 16:32:05 UTC
*** Issue 63525 has been marked as a duplicate of this issue. ***
Comment 50 jr 2006-03-30 10:16:08 UTC
Hi all, just for the record, thb and I had a short meeting with Rene from ATI at CeBIT and now he tries to find a solution for our problem. After checking the ATI database he supposes a problem with the video bios. Just now I sent him some more information about my system and wait again for an answer.
Comment 51 christian.guenther 2006-04-05 09:42:29 UTC
*** Issue 64011 has been marked as a duplicate of this issue. ***
Comment 52 jr 2006-05-08 20:49:48 UTC
Another short update: After a request for actual information Rene from ATI answered, that the engineers are working on this issue and he is quite sure, that the problem is caused by the video bios. A few days ago he sent me a new driver for testing, but unfortunately the bluescreen occurs as before. So we have to wait again...
Comment 53 thb 2006-07-11 11:55:37 UTC
@jrahemipour: any news on that?
Comment 54 Rainer Bielefeld 2006-09-02 09:21:32 UTC
*** Issue 68965 has been marked as a duplicate of this issue. ***
Comment 55 christian.guenther 2006-09-13 09:25:56 UTC
*** Issue 69228 has been marked as a duplicate of this issue. ***
Comment 56 tomdkat2 2006-09-13 15:03:37 UTC
Any updates from ATI on the video bios issue? Thanks!
Comment 57 utomo99 2007-01-31 03:08:07 UTC
any update for this issue ? or we close this and say it is ATI problems ?
Comment 58 thb 2007-01-31 10:10:10 UTC
@utomo99: well, it surely is ati's problem when they crash, but I'd still like to fix our behaviour (which somehow triggers this driver bug), if I'd only know how. Let's keep it open for the time being, and I'll see whether I could grab someone from ati/amd this year...
Comment 59 christian.guenther 2007-02-13 12:52:47 UTC
*** Issue 71202 has been marked as a duplicate of this issue. ***
Comment 60 tourla50 2007-03-02 22:03:08 UTC
with the OpenOffice 2.01 and 2.0 impress , after i made a presentation, when click on slide show, the transition of the 1er slide is very slow and them crash the pc an appear a blue page from windowsxp telling me that was a problem with ati2vga driver. this is the same things on an acer computer (at work) and on a hp pavilion at home with some different presentation. (the same thing is arrived last year on the acer PC with oo 1.9) !!!! if you have a solution ....
Comment 61 manens 2007-03-05 10:23:29 UTC
Hi all, I have found a work around for this bug. In windows display property>Settings>Advanced, Troubleshoot, reduce the acceleration by one notch. @jrahemipour: Any update from ATI ?
Comment 62 jr 2007-03-23 22:00:19 UTC
I'm sorry, no news from ATI. The contact from last years CeBIT doesn't respond to my mails anymore. I tried to solve the problem by updating the video bios with the last version, but I get still a bluescreen (with 2.2.0RC4). The only thing that helps is reducing the hardware acceleration.
Comment 63 andreschnabel 2007-04-01 19:39:44 UTC
*** Issue 75961 has been marked as a duplicate of this issue. ***
Comment 64 floeff+ooo 2007-04-17 11:22:46 UTC
I finally had a chance to test it on my IBM T42p. (Fire GL T2) Presentation is horribly slow, and finally crashed on the uncover effect between two slides. Windows (XP) is up to date with all patches and hotfixes, and the machine's drivers and BIOS are also up to date. So, updating my video BIOS is not possible due to no newer drivers. If you google for "ati2dvag 0x000000ea", there are a lot of results with different solutions. Some mentioned software has to be uninstalled, some mentioned tweaking the VGA, AGP and memory settings. thb: If slowing down the hardware acceleration temporarily helps, maybe we can implement a workaround? If user has admin rights (which is true for most cases), then the HW acceleration during the execution of Impress is automatically set down one level.
Comment 65 thb 2007-04-17 12:38:59 UTC
@floeff: does reducing hw accel fix this for you? this might be indeed a temporary solution, given that ATI seems to be somewhat unresponsive...
Comment 66 pagalmes.lists 2007-04-17 12:57:20 UTC
@thb: manens posted on this issue "I have found a work around for this bug. In windows display property>Settings>Advanced, Troubleshoot, reduce the acceleration by one notch." This fix was tested on a bunch of PCs and it worked.
Comment 67 floeff+ooo 2007-04-17 13:07:47 UTC
thb: It does fix it for me. The effect seems to run normal, and no crash occur. Disabling write comnbining does NOT help and requires a reboot. Changing hardware acceleration DOES help in most cases, and does not require a reboot (only blanks/flickers the screen for two seconds or so, but that is not a problem IMHO - better than crashing). Problem is, that this requires admin privileges. One solution is to install a service that does it for OOo (but this is a bit much for a workaround), or rely on users being admins (true for most cases). We also could ask users for admin password if a non-admin user is being detected, and add the possibility to disable the workaround in the users profile. I agree, this is a lot of workload for this bug, but I think this one is real important. "Normal" users just think Impress is shit, because their fancy presentations do not work, and normal users do not want to jiggle with lots of options. :-) Some other solutions that might be helpful: - We could try the "original" MS driver, not the ATI one, maybe that helps (but it does not help our users). - Testing if Vista and 2K have this bug as well, or if it is truly related to XP, so we at least know which systems are affected. - Lowering down other settings in the graphics driver. I think in the ATI drivers, you can adjust most acceleration things, maybe one of these is it. I am willing to help in testing builds with this workaround, as I have an affected system here (my ThinkPad).
Comment 68 christian.guenther 2007-04-20 09:08:50 UTC
*** Issue 75795 has been marked as a duplicate of this issue. ***
Comment 69 pagalmes.lists 2007-05-02 07:37:14 UTC
Some info thb wrote about this issue: http://blogs.sun.com/GullFOSS/entry/windows_display_driver_woes_what
Comment 70 floeff+ooo 2007-05-02 09:07:24 UTC
Thanks for the pointer, that sounds really great! thb, any timeframe on when the new DX backend will be implemented in OOo?
Comment 71 thb 2007-05-03 13:26:56 UTC
Regarding timeframe: no specifics yet, but next feature release (i.e. 2.3) should be doable.
Comment 72 floeff+ooo 2007-05-03 16:25:43 UTC
Sounds great! If you need help in testing, let me know. I have an "affected" machine.
Comment 73 thb 2007-07-23 13:12:33 UTC
DirectX-based implementation now committed to CWS dxliberate01. On track for 2.3 release...
Comment 74 pagalmes.lists 2007-07-23 14:22:48 UTC
Glad to hear about it :-)
Comment 75 thb 2007-07-23 21:58:04 UTC
@cgu: Fixed in CWS dxliberate01, please verify.
Comment 76 thb 2007-07-24 14:42:47 UTC
For those of you wanting to try, there's a preview build of the CWS available here: http://ooo.services.openoffice.org/pub/OpenOffice.org/cws/upload/dxliberate01/OOo_2.3.222_Win32Intel_install.exe Please be aware that this is an untested developer build, so don't use it for production work.
Comment 77 floeff+ooo 2007-07-24 15:33:17 UTC
Just testet it on my IBM T42p with ATI Fire GL T2 under XP. 2.3m222 dxliberate01 works fine :-) Yippieeeeh! Thanks a lot, Thorsten!
Comment 78 thb 2007-07-26 19:08:08 UTC
@floeff: excellent. @all: anybody else tried it? Come on, folks, this needs somewhat wider coverage than a ThinkPad! ;-)
Comment 79 thb 2007-08-03 13:40:48 UTC
Retargetted because surrounding CWS moved to 2.4
Comment 80 floeff+ooo 2007-08-03 17:56:29 UTC
Aaaaah! Who decided that? ;-) If possible by any means, please include the CWS in 2.3 This bug is one of the most annoying ones, actually making OOo look very "unprofessional" on a lot of laptops running Windows. I heard enough people saying they'd switch back to PowerPoint because playing an animated presentation crashes their machine. So, *PLEASE*, I urge you to include this in 2.3
Comment 81 jr 2007-08-03 18:28:09 UTC
I can't believe that. :-( Many of my customers wait for this fix really a long time (and me too). After I told them, that they can switch to 2.3 now - I shall explain, that they have to wait another 6 month?? I checked the cws on my thinkpad and it works fine. What exactly is the reason for retargeting?
Comment 82 pagalmes.lists 2007-08-03 23:03:18 UTC
+1 for including it in OOo 2.3
Comment 83 tomdkat2 2007-08-04 00:53:11 UTC
A friend of mine switched from using OO to MS Office because of this bug as well.
Comment 84 pagalmes.lists 2007-08-04 07:19:55 UTC
Hello thb, first of all, thanks for the work you did on this issue. Wouldn't it possible to set this issue as a P1? It is quite severe to me, as it doesn't only crash OOo, but the whole PC! That means lost data not only for OOo documents but for all the opened applications.
Comment 85 jr 2007-08-05 10:54:20 UTC
As discussed with thb the cws needs to be tested on many different systems, if we want to have a small chance, that the fix will be incluced in 2.3. The germanophone community started with extensive tests today (we have 7 tester atm) and I expect first results today in the evening. So if someone is willing to help, please download the cws from http://ooo.services.openoffice.org/pub/OpenOffice.org/cws/upload/dxliberate01/OOo_2.3.222_Win32Intel_install.exe and give feedback as soon as possible. Please check particulary your existing presentations and check, if the slideshow and the previews work well. Thanks in advance!
Comment 86 andreschnabel 2007-08-05 18:59:42 UTC
I tested the cws build on windows (ATI X11300 and ATI 9600). All tests are fine. (No regresseions, effects are much better than in regular builds, great antialising) Unfortunately there is one regression (see issue 80363)
Comment 87 jr 2007-08-05 19:40:39 UTC
Hi all, here are the (first) results of the germanophone jury... We tried to check as many existing presentations as possible and tested the slideshow and the previews. We tested on Win XP and Win Vista. We found some bugs, but checked them against m222 where the bugs occur, too (for example issue 80357 - crash with custom animation, which occured in several different presentations today, is already fixed or issue 80364 - slide pane scrolling broken). I ran several automated tests (first, topten and all graphics updt tests) without any errors (and yes, no bluescreen :-) ). The slideshow works fine and smooth, only the preview starts on some machines with perceptible delay (AndrÃ© already mentioned it: issue 80363).
Comment 88 thb 2007-08-05 20:29:40 UTC
First off, kudos to jrahemipour and all the other folks that tried out the preview build and provided valuable feedback. Short explanation for the (maybe temporary) move of the CWS to 2.4: we haven't had enough QA resources to finish the CWS in time, and I deemed it too risky to push this into 2.3 literally minutes before code freeze, with only one reported check up to then.
Comment 89 thb 2007-09-27 09:10:26 UTC
As promised, the DX canvas that circumvents the crash is now available as an extension - if you want to help testing, here is the link: http://blogs.sun.com/GullFOSS/entry/directx_canvas_extension_ready_for
Comment 90 pagalmes.lists 2007-10-03 15:37:34 UTC
I tried to reproduce the crash with 2.3, but I can't (even without the extension)! It crashes with OOo 2.2, but not any more with OOo 2.3. It is nice, but is it normal? Could someone confirm this?
Comment 91 thb 2007-10-09 12:23:35 UTC
Comment 92 thb 2007-10-10 19:25:52 UTC
@all: final version of the dxcanvas extension uploaded. A few words here: http://blogs.sun.com/GullFOSS/entry/at_last_directx_slideshow_extension Direct download link: http://ftp.stardiv.de/pub/OpenOffice.org/developer/extensions/dxcanvas.oxt
Comment 93 floeff+ooo 2007-10-10 19:33:38 UTC
Thanks a lot, Thorsten!
Comment 94 pagalmes.lists 2007-10-11 00:04:09 UTC
As I had no adswer, I repost it: Ti looks like this issue is corrected already in OOo 2.3. I tried to reproduce the crash with 2.3, but I can't (even without the extension)! It crashes with OOo 2.2, but not any more with OOo 2.3. It is nice, but is it normal? Could someone confirm this?
Comment 95 floeff+ooo 2007-10-11 00:12:18 UTC
IIRC, the crash should be fixed in 2.3. The extension should fix slow graphics and introduces anti-aliasing.
Comment 96 thb 2007-10-11 08:19:48 UTC
@pagalmes: that would indeed be interesting - I remember one change that slightly modified the way VCL deals with background surfaces, which could have caused this side effect. Funny still. @all: could you please try 2.3 _without_ the extension, switch your ATI acceleration settings back to normal, and check whether you still get this crash?
Comment 97 andreschnabel 2007-10-12 06:54:31 UTC
as I cannot reproduce the crash anymore (Win XP, ATI Radeon 9600, OOo 2.3) I close as suggested.