Bug 441521

Could you please add more information based on the reporting bug part of
http://fedoraproject.org/wiki/AndreasBierfert/Wine.

More info:

wine version: 0.9.58
Selinux mode: enforcing
Window manager: gnome-session with metacity
wine subpackages:

wine-ldap-0.9.58-1.fc9.i386
wine-nas-0.9.58-1.fc9.i386
wine-tools-0.9.58-1.fc9.i386
wine-capi-0.9.58-1.fc9.i386
wine-0.9.58-1.fc9.i386
wine-core-0.9.58-1.fc9.i386
wine-esd-0.9.58-1.fc9.i386
wine-cms-0.9.58-1.fc9.i386
wine-twain-0.9.58-1.fc9.i386
wine-jack-0.9.58-1.fc9.i386

wine works fine with fresh ~/.wine. With ~/.wine that I used in F8 it freezes
the system. This is hard freeze as I cannot ping the system and the following
script that I run in the background:

while true; do
    sleep 1
    date +%T >> $HOME/log.txt
    sync
done

after the reboot shows the time of the crash in ~/log.txt. Also after the crash
there is no disk activity as shown by a notebook indicator. 

So it looks like a kernel bug. Should I change the component? I can also can try
to attach .wine after removing a windows application installed there. 

Created attachment 301664 [details]
system.reg that causes the system freeze

The content of the attached .wine/system.reg causes the freeze. That is,

    rm -rf ~/.wine
    wine notepad.exe 

creates the default ~/.wine and runs notepad.exe normally. But replacing
.wine/system.reg with the attached file triggers the whole system freeze when
running "wine notepad.exe".

Created attachment 301687 [details]
minimal crashing system.reg

The attached system.reg contains single entry:

[System\\CurrentControlSet\\Services\\Spooler] 1181644148
"Description"="Loads files to memory for later printing."
"DisplayName"="Print Spooler"
"ErrorControl"=dword:00000001
"Group"="SpoolerGroup"
"ImagePath"="spoolsv.exe"
"ObjectName"="LocalSystem"
"Start"=dword:00000002
"Type"=dword:00000110

Using the file as ~/.wine/system.reg causes a total system freeze when running
"wine notepad.exe". The crash happens with a wireless networking enabled or
with networking disabled as set with NetworkManager.

Note that a default system.reg generated after 
  
  rm -rf ~/.wine; wine notepad.exe  

does not contain the entry.

Thanks for the good bug report. I will see what I can do about this.

(In reply to comment #5)
> Thanks for the good bug report. I will see what I can do about this.

If necessary I can give an access to the notebook over ssh. 

(In reply to comment #4)
> The attached system.reg contains single entry:
> 
> [System\\CurrentControlSet\\Services\\Spooler] 1181644148
...

Removing that entry from the original system.reg file also fixes the issue and
the Windows application runs normally.

Changing version to '9' as part of upcoming Fedora 9 GA.
More information and reason for this action is here:
http://fedoraproject.org/wiki/BugZappers/HouseKeeping

In reply to Comment #4:

That is not true. A newly generated ~/.wine/system.reg does have that Spooler
stanza.

The difference is that the newly generated functional one has a:
"Start"=dword:00000004
instead of the non-functional old:
"Start"=dword:00000002

Changing that single number (and doing a 'service wine restart' just for fun)
made my wine work again.

I think the wine devs went a little too far in emulating MSWindows behaviour
when a -2 can bring down the entire OS rather than just crash wine ;]

Several people suffer from this bug, a bug was listed at wine:
http://bugs.winehq.org/show_bug.cgi?id=13358
But it was "closed invalid"
Quote: "Wine can't freeze your system, that's most likely a kernel driver bug."

Strangely enough this bug only occurs when wine preforms. Also, this bug occurs
on computers with different architectures, with complete different hardware...
  

Can you add a reference to this Fedora bug number in that upstream bug?
It contains a specific triggering mechanism that might enable the developers to
reproduce it and narrow down what is at fault.

I added the link to this bug in a comment at wine's Bugzilla, wine does not have
a upstream resolution.

Does anyone know already if this is a kernel bug or a x11 bug?

Any way this can get a priority boost?

It's rather insulting to Wine folks that a bug in FC9 that causes
a system crash is rated as priority 'low', severity 'low'.

Wine runs completely in user space, so by definition, anything it does
that causes a system freeze is *not* a Wine bug.

Cheers,

Jeremy

Hello,

same issue here: https://bugzilla.redhat.com/show_bug.cgi?id=443552 ... that one
has at least correct severity ;-)
I added some kernel oopsies with 2.6.25.4-30.fc9.i686 and wine-1.0-rc4 there.

Regards

*** This bug has been marked as a duplicate of 443552 ***