Bug: NeoOffice 3.0 crash when the printer is stop

Paolo · Red Pill Joined: Mar 09, 2009 Posts: 8

Mac OS X 10.4.11 PPC
NeoOffice 3.0 Early Access 2 Patch 1 Test 9

How to reproduce this bug:
- System preferences -> Print & Fax -> Select a printer -> Print Queue -> Stop Jobs
- Open this spreadsheet attached and print it.
- After receive the message asking if you want continue also if the printer is stop NeoOffice crash sometimes printing the page sometimes not.

Please find attached the document and the crash log.

I'm able to reproduce this bug only with this document.

Paolo

pluby · The Architect Joined: Jun 16, 2003 Posts: 11949

Can anyone else reproduce this crashing? I don't see any specific NeoOffice code causing the crash so I suspect that Mac OS X is destroying the native print job while NeoOffice is trying to print.

Patrick

OPENSTEP · Posted: Mon Mar 09, 2009 10:10 am Post subject:

When the printer is paused, I get a dialog with three options: "Cancel", "Add to Queue", and "Continue". I'm not able to get a crash with any of these, however. Which one of these options are you using that is leading to your crash?

ed

James3359 · Posted: Mon Mar 09, 2009 10:19 am Post subject:

[ETA NeoOffice 3.0 EA2 Patch 1 iBook 1.33 GHz 512 MB PowerPC G4, OS X 10.4.11]
OK, something funny happens when I try to print this with a stopped print queue. The warning window comes up, but almost instantly greys out as if something else is getting the focus (Picture 1). If I click anywhere in the NeoOffice window I get a beep, except for the title bar of the warning dialog. If I click that it regains focus. If I click on the desktop visible to the side of the window, then NeoOffice sometimes crashes silently. (Crash log attached)

I am also getting the following Console.log:

soffice.bin(402,0x180e600) malloc: *** error for object 0x13819000: incorrect checksum for freed object - object was probably modified after being freed, break at szone_error to debug
soffice.bin(402,0x180e600) malloc: *** set a breakpoint in szone_error to debug
soffice.bin(402,0x180e600) malloc: *** error for object 0x1f2fc00: incorrect checksum for freed object - object was probably modified after being freed, break at szone_error to debug
soffice.bin(402,0x180e600) malloc: *** set a breakpoint in szone_error to debug
soffice.bin(402,0x180e600) malloc: *** error for object 0x1f2fc00: incorrect checksum for freed object - object was probably modified after being freed, break at szone_error to debug
soffice.bin(402,0x180e600) malloc: *** set a breakpoint in szone_error to debug
Mar 9 16:59:53 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: An unexpected Java error has been detected by HotSpot Virtual Machine.
Mar 9 16:59:53 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: If this error is reproducible, please report it with the following information:
Mar 9 16:59:53 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: 1. Provide the steps to reproduce, a test case, and any relevant information
Mar 9 16:59:53 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: 2. The corresponding JavaNativeCrash_pid<num>.crash.log (Java state)
Mar 9 16:59:53 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: 3. The corresponding <name>.crash.log (native state; generated by CrashReporter)
Mar 9 16:59:53 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: 4. This data:
Mar 9 16:59:53 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: Java VM: Java HotSpot(TM) Client VM (1.5.0_16-132 mixed mode, sharing)
Mar 9 16:59:53 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: Bus Error (0xa) at pc=0xfffeff20
Mar 9 16:59:53 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: Process ID: 402, Current Thread: 25646592
Mar 9 16:59:53 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: File report at: http://bugreport.apple.com/
Mar 9 16:59:53 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: An error report file has been written to:
Mar 9 16:59:53 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: /Users/jamesrichards/Library/Logs/Java/JavaNativeCrash_pid402.crash.log
Mar 9 17:00:03 james-richards-ibook-g4 crashdump[415]: soffice.bin crashed
Mar 9 17:00:09 james-richards-ibook-g4 crashdump[415]: crash report written to: /Users/jamesrichards/Library/Logs/CrashReporter/soffice.bin.crash.log
soffice.bin(416,0x180e600) malloc: *** Deallocation of a pointer not malloced: 0x1e60239; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug
Mar 9 17:03:25 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: An unexpected Java error has been detected by HotSpot Virtual Machine.
Mar 9 17:03:25 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: If this error is reproducible, please report it with the following information:
Mar 9 17:03:25 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: 1. Provide the steps to reproduce, a test case, and any relevant information
Mar 9 17:03:25 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: 2. The corresponding JavaNativeCrash_pid<num>.crash.log (Java state)
Mar 9 17:03:25 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: 3. The corresponding <name>.crash.log (native state; generated by CrashReporter)
Mar 9 17:03:25 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: 4. This data:
Mar 9 17:03:25 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: Java VM: Java HotSpot(TM) Client VM (1.5.0_16-132 mixed mode, sharing)
Mar 9 17:03:25 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: Bus Error (0xa) at pc=0xfffeff20
Mar 9 17:03:25 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: Process ID: 416, Current Thread: 25646592
Mar 9 17:03:25 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: File report at: http://bugreport.apple.com/
Mar 9 17:03:25 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: An error report file has been written to:
Mar 9 17:03:25 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: /Users/jamesrichards/Library/Logs/Java/JavaNativeCrash_pid416.crash.log
Mar 9 17:03:29 james-richards-ibook-g4 crashdump[422]: soffice.bin crashed
Mar 9 17:03:31 james-richards-ibook-g4 crashdump[422]: crash report written to: /Users/jamesrichards/Library/Logs/CrashReporter/soffice.bin.crash.log
Mar 9 17:06:10 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: *** WARNING *** extra data equals 100.
Mar 9 17:06:10 james-richards-ibook-g4 /Applications/NeoOffice.app/Contents/MacOS/soffice.bin: *** WARNING *** freeing non-aligned data (2).
soffice.bin(423,0x18d2600) malloc: *** Deallocation of a pointer not malloced: 0x1fa49bc; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug
soffice.bin(423,0x180e600) malloc: *** error for object 0x1f48000: incorrect checksum for freed object - object was probably modified after being freed, break at szone_error to debug
soffice.bin(423,0x180e600) malloc: *** set a breakpoint in szone_error to debug
soffice.bin(423,0x180e600) malloc: *** error for object 0x1f48000: incorrect checksum for freed object - object was probably modified after being freed, break at szone_error to debug
soffice.bin(423,0x180e600) malloc: *** set a breakpoint in szone_error to debug

sardisson · Town Crier Joined: Feb 01, 2004 Posts: 4588

I can't reproduce this with several tries on 10.5.6/Intel, but the printing UI is completely different, so I can't be sure "Pause Printer" is doing the same thing as "Stop Jobs" on 10.5.

Like James, though, I do see that the "Printer is Paused" native dialogue comes up unfocused and that focus remains on the NeoOffice window with the print sheet.

Semi-related: Do we have an existing bug/known issue about the text associated with form controls in the document displaying the "cheap whore" effect while native dialogues are active? I noticed them redrawing repeatedly while the printing sheet and the "Printer is Paused" dialogue were up, and they remain overly done after (you can see that in James's screenshot).

Smokey

pluby · The Architect Joined: Jun 16, 2003 Posts: 11949

pluby · The Architect Joined: Jun 16, 2003 Posts: 11949

FYI. I moved this forum topic to the NeoOffice Early Access Program Support forum so that it does not get buried in the NeoOffice 2.2.5 support requests.

Also, I have created bug 3429 to track this bug.

If you add yourself to the CC: list in that bug, you will get automatically notified when I have a fix available for you to try.

Please note that since this dialog is not our dialog and that dialog appears to be destroying a valid print job in our application, finding a fix for this bug will require some time.

Patrick

pluby · The Architect Joined: Jun 16, 2003 Posts: 11949

I have one more question: can anyone reproduce this bug with Mac OS X 10.4.11 after doing the following steps?:

1. Select the Tools :: Options menu in NeoOffice and in the dialog that appears, click on the NeoOffice :: Java item

2. In the NeoOffice :: Java item, select the unselected Java version and press the OK button

3. Restart NeoOffice and try reproducing the bug

If the crashing does not occur with Java 1.4.x, then I am pretty confident that the bug is caused by special way that Java 1.5.x does printing. Apple's Java 1.5.x code performs printing on a secondary thread instead of on the main thread like other Mac OS X applications.

Patrick

James3359 · Posted: Mon Mar 09, 2009 6:47 pm Post subject:

OK I had a go at this selecting 1.4.x Java and relaunching NO and following the steps in the OP. I still got the odd effect of the dialog appearing and then greying out as it loses focus. I clicked cancel, and then (I think) clicked print again and the dialog reappeared and did the same trick. When it got the focus it was then seemingly unresponsive to a series of clicks on different buttons before a silent crash occurred.

The console log and crash log are attached. As you will see its not a great time of day for me here, so I'll bow out for now as I'm too tired to work with the detail of how this crash is happening (e.g. which button I clicked first on the dialog and stuff like that) so I won't be able to give much help for now on more detailed probing of the problem.

pluby · The Architect Joined: Jun 16, 2003 Posts: 11949

From James3359's crash log, it is clear that this is not caused by Java 1.5.x running printing on a secondary thread. Instead, it confirms my original theory that Mac OS X destroys the native print job and this, in turn, causes Java to crash when it does its usual cleanup after printing of a page is completed.

I am wondering which Mac OS X print function is displaying this dialog. Can anyone take a sample of the NeoOffice process immediately after the native dialog appears? This might give me some idea of what we can do to detect that the dialog has appeared and, hopefully, force Java to abort its current print job at that point.

Patrick

Paolo · Red Pill Joined: Mar 09, 2009 Posts: 8

I confirm to have the same grey warning window followed by a crash also with java 1.4.x.

I found another way to emulate the same (?) bug, if I try to print and leave the first window open for 10/15 seconds without click on "Print" I have a crash.

James3359 · Posted: Tue Mar 10, 2009 8:01 am Post subject:

I'm attaching a sample of the process - I hope it is what you want, Patrick. NeoOffice crashed again while I was collecting it, so I am attaching the crash and console logs.

pluby · The Architect Joined: Jun 16, 2003 Posts: 11949

Thank you everyone. It think I now have enough data so now I will start the slow process of trying to detect the display of the native dialog. If I can do that, then I am hopeful that I can stop any Java print code from executing and, in turn, avoid any crashing.

Patrick

pluby · The Architect Joined: Jun 16, 2003 Posts: 11949

I think that I have fixed this bug and I have posted links to a test patch in my last post in bug 3429.

Can anyone install the test patch in that bug and tell us if this bug is fixed for you?

Patrick

Paolo · Red Pill Joined: Mar 09, 2009 Posts: 8

Thanks, the bug is partially fix.
Now I'm able to print this document also if the printer is paused but I still have a crash when I emulate the situation explained in my previous message.
Please find attached the log of the crash and a screenshot of the window before the crash.

Paolo