279770 - Sunbird starts and stops repeatedly, console displays *** loading the extensions datasource

Status: VERIFIED WORKSFORME

(Calendar :: Internal Components, defect)

Description

[Dann S. Washko]

User-Agent:       Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.7.5) Gecko/20041217
Build Identifier: Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.7.5) Gecko/20041217

I grabbed Sunbird RC2 for Mac OS X.  I'm running 10.3.7.  I also run Mozilla
1.7.5, Firefox 1.0, and Thunderbird 1.0.  I have the calendar plugin for Mozilla
and Thunderbird and they work fine.

When I first start Sunbird I get the following message:

     Any old extensions that you have installed have been disabled.

The program then crashes with the error:  

-----------------------------
Command: sunbird-bin
Path:    /Applications/Sunbird.app/Contents/MacOS/sunbird-bin
Version: 0.2RC2 (0.2RC2)
PID:     891
Thread:  0

Exception:  EXC_BAD_ACCESS (0x0001)
Codes:      KERN_PROTECTION_FAILURE (0x0002) at 0x00000000

Thread 0 Crashed:
0   org.mozilla.sunbird 	0x002b6fa4
GlobalWindowImpl::SetNewDocument(nsIDOMDocument*, int, int) + 0x4e8
1   org.mozilla.sunbird 	0x00213e3c DocumentViewerImpl::Close() + 0x84
2   org.mozilla.sunbird 	0x0050eef0 nsDocShell::Destroy() + 0xe0
3   org.mozilla.sunbird 	0x004f0eb0 nsWebShell::Destroy() + 0x18
4   org.mozilla.sunbird 	0x0057f2b8 nsXULWindow::Destroy() + 0x180
5   org.mozilla.sunbird 	0x0057de40 nsWebShellWindow::Destroy() + 0x1cc
6   org.mozilla.sunbird 	0x0057c2e0 nsWebShellWindow::Close() + 0x1c
7   org.mozilla.sunbird 	0x00577e4c nsAppShellService::Observe(nsISupports*,
char const*, unsigned short const*) + 0x60
8   libxpcom_core.dylib 	0x00dfc470
nsObserverService::NotifyObservers(nsISupports*, char const*, unsigned short
const*) + 0xf0
9   libxpcom_core.dylib 	0x00df3424 NS_ShutdownXPCOM_P + 0xb4
10  org.mozilla.sunbird 	0x00633ef4 ScopedXPCOMStartup::~ScopedXPCOMStartup
[unified]() + 0x38
11  org.mozilla.sunbird 	0x00633eac ScopedXPCOMStartup::~ScopedXPCOMStartup
[in-charge]() + 0x14
12  org.mozilla.sunbird 	0x0063678c xre_main(int, char**, nsXREAppData const*) +
0xbf0
13  org.mozilla.sunbird 	0x000034ec main + 0xdc
14  org.mozilla.sunbird 	0x00002f00 start + 0x1c8
15  dyld                	0x8fe1a558 _dyld_start + 0x64

---------------

If I try to start Sunbird again it cycles through starting and stopping
repeatedly and must kill the process by opening a terminal and issuing killall
-9 sunbird-bin (the process id keeps changing).  In the log files I can see this
message repeatedly:

***  loading the extensions datasource


This issue seems similar to:  256206 which was reported on the Win32 platform.

At the suggestion of a helpful person on the newsgroup I tried copying over the
Extensions.rdf from my profile to the extensions directory in the Sunbird.app
directory.  This did not correct the problem. 

I can repeat this whole process from the beginning if I delete the Subird
directory in ~/Library/Application Support.  Otherwise, the program just keeps
starting and stopping.

 

Reproducible: Always

Comment 1

[Dann S. Washko]

Release 0.2b of Sunbird seems to work fine.  I pulled this fron the September 8,
2004 nightly build directory.  

I can import my current calendars and they appear to work.

I tried subsequent builds after 0.2b and they all produce the same problem above.

Comment 2

[Dann S. Washko]

After some more testing with version 0.2b and the latest release I have narrowed
it down to one file which appears to be causing the problem:  compatibility.ini.
 I figure that this file is somehow generated in the user's profile when Sunbird
is first executed.  I cannot find it in the actual Application directory.

The latest release writes this to compatiblity.ini:

[Compatibility]
Build ID=2005011113
Components List Changed=1

When Sunbird 0.2b is executed for the first time this is the contents of the
same file in the users Profile directory:

[Compatibility]
Build ID=0000000000
Components List Changed=0 

If I replace the compatibility.ini file in the bad profile with that from the
good profile the latest Sunbird works perfectly.

Comment 3

[Dann S. Washko]

Another bit of information.  If you leave BuildID to all 0's and then change the
value of Components List Changed to 1, it will run.  But then next time you
execute Sunbird it will fail as above.  The compatibility.ini file will be
overwritten and the date Sunbird was executed will appear in the BuildID. 
Components List Changed will be back to 0.  

If you leave both of these values as 0 Sunbird will continue to run without a
problem.  

Comment 4

[timeless]

*** This bug has been marked as a duplicate of 276507 ***

Comment 5

[Mike Shaver (:shaver emeritus)]

No, this is not a dup.  Bug 276507 is about a case where it crashes on first
launch, and then works fine afterwards.  This is about a non-crash looping
restart case.  (That another looping-restart case was dup'd against 276507
doesn't make it a good idea to compound the error with this one, especially
since we have new data here.)

bsmedberg: does the [Compatibility] element from the previous comments ring any
bells for you?

Comment 6

[Benjamin Smedberg]

Not really. mkaply's change to nsExtensionManager.js.in revision 1.81 (r=me,
committed yesterday) may actually fix this problem.

Comment 7

[scott hutinger]

I grabbed the following for those like myself (I have it now) that don't have
moz code.  I'll keep up with the cal code :-)  The following looks like it would
keep down restarts to me.  I'm uncertain if it kills the crash problem under OS
X though.


revision 1.81
date: 2005/01/25 16:07:26;  author: mkaply%us.ibm.com;  state: Exp;  lines: +2
-2
Write components manifests if they are missing, r=bsmedberg

diff -u -r1.80 -r1.81
--- nsExtensionManager.js.in    17 Jan 2005 18:50:16 -0000      1.80
+++ nsExtensionManager.js.in    25 Jan 2005 16:07:26 -0000      1.81
@@ -1656,8 +1656,8 @@
         needsRestart = true;
       }

-      if (aIsDirty)
-        needsRestart = this._finishOperations();
+      if (aIsDirty || needsRestart)
+        needsRestart = this._finishOperations() || needsRestart;

Comment 8

[Charles Herbig]

Attachment: Application crash log.

Comment 9

[Charles Herbig]

I can see Sunbird repeatedly trying to start in the Dock.

Using ps, I can see Sunbird trying to start, but only one copy of it is running:


[saruman:~] chas% ps -ax | grep bird | grep -v grep
 1165  ??  Rs     0:00.46 /Applications/Sunbird.app/Contents/MacOS/sunbird-bin 
[saruman:~] chas% ps -ax | grep bird | grep -v grep
 1171  ??  Rs     0:00.30 /Applications/Sunbird.app/Contents/MacOS/sunbird-bin 
[saruman:~] chas% ps -ax | grep bird | grep -v grep
 1176  ??  Rs     0:00.03 /Applications/Sunbird.app/Contents/MacOS/sunbird-bin 

In system.log, I only see one attempt to write a crash log.

Jan 28 18:34:03 localhost crashdump: Started writing crash report to:
/Users/chas/Library/Logs/CrashReporter/sunbird-bin.crash.log
Jan 28 18:34:04 localhost crashdump: Finished writing crash report to:
/Users/chas/Library/Logs/CrashReporter/sunbird-bin.crash.log

And in console.log, it keeps repeating the same error message over and over:

*** loading the extensions datasource
*** loading the extensions datasource
*** loading the extensions datasource

It appears that the parent forks and dies, and the child either detects this, or
the child code tries to act like a parent and the child then forks and dies.

I am adding my sunbird-bin.crash.log to this bug report.
I can see Sunbird repeatedly trying to start in the Dock.

Using ps, I can see Sunbird trying to start, but only one copy of it is running:


[saruman:~] chas% ps -ax | grep bird | grep -v grep
 1165  ??  Rs     0:00.46 /Applications/Sunbird.app/Contents/MacOS/sunbird-bin 
[saruman:~] chas% ps -ax | grep bird | grep -v grep
 1171  ??  Rs     0:00.30 /Applications/Sunbird.app/Contents/MacOS/sunbird-bin 
[saruman:~] chas% ps -ax | grep bird | grep -v grep
 1176  ??  Rs     0:00.03 /Applications/Sunbird.app/Contents/MacOS/sunbird-bin 

In system.log, I only see one attempt to write a crash log.

Jan 28 18:34:03 localhost crashdump: Started writing crash report to:
/Users/chas/Library/Logs/CrashReporter/sunbird-bin.crash.log
Jan 28 18:34:04 localhost crashdump: Finished writing crash report to:
/Users/chas/Library/Logs/CrashReporter/sunbird-bin.crash.log

And in console.log, it keeps repeating the same error message over and over:

*** loading the extensions datasource
*** loading the extensions datasource
*** loading the extensions datasource

It appears that the parent forks and dies, and the child either detects this, or
the child code tries to act like a parent and the child then forks and dies.

I am adding my sunbird-bin.crash.log to this bug report.

Comment 10

[C. Taylor]

It does this to me without fail thus far. Also, it does NOT allow me to do force
quit, so I have to logout and then get back on OS X. Can anyone else confirm
this, or is it just me?

Comment 11

[Michael]

Also here. OSX 10.3.8, all updates current. 12 inch G4 iBook, 1.2 GHz. 

Comment 12

[Joey Minta]

Can anyone test this with a recent nightly?  I think things are better there.

Comment 13

[Matthew (lilmatt) Willis]

(In reply to comment #12)
> Can anyone test this with a recent nightly?  I think things are better there.

I (unsuccessfully) tried to replicate this using the 2006033110 nightly.

It writes the following to compatibility.ini, unlike the stuff others were seeing:

[Compatibility]
LastVersion=0.3a1+_2006033110/1.9a1_2006033110
LastOSABI=Darwin_ppc-gcc3
LastPlatformDir=/Users/mwillis/Desktop/Sunbird.app/Contents/MacOS

jminta: Could this be a trunk vs. 1.8 thing?

Otherwise, WFM?

Comment 14

[Joey Minta]

(In reply to comment #13)
> (In reply to comment #12)
> > Can anyone test this with a recent nightly?  I think things are better there.
> 
> jminta: Could this be a trunk vs. 1.8 thing?
> 
> Otherwise, WFM?
> 
My guess is that it's a 1.7 vs. 1.8 thing, and WFM is the right response.  Marking as such.  If anyone can reproduce with a pull from 1.8 or trunk, please feel free to reopen.