[Bug 26373] New: Solidworks 2010 crash ( regression )

http://bugs.winehq.org/show_bug.cgi?id=26373

Summary: Solidworks 2010 crash ( regression )
           Product: Wine
           Version: 1.3.15
          Platform: x86
        OS/Version: Linux
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: -unknown
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]

Solidworks is crashing just before opening its main window. A window pops-up to
send a crash report to solidworks.

Regression test identifies this commit :

c65bcce5899ba81226295303ed3df73a7be86c09 is the first bad commit
commit c65bcce5899ba81226295303ed3df73a7be86c09
Author: Hans Leidekker 
Date:   Wed Mar 2 10:46:12 2011 +0100

server: Map the Unix user id to a local user SID instead of the interactive
SID.

[Bug 26373] Solidworks 2010 crash ( regression )

http://bugs.winehq.org/show_bug.cgi?id=26373

denis bonnenfant  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |regression

[Bug 26373] Solidworks 2010 crash ( regression )

http://bugs.winehq.org/show_bug.cgi?id=26373

Hans Leidekker  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #1 from Hans Leidekker  2011-03-10 14:00:26 CST ---
Please attach terminal output. Did you try starting from a clean prefix?

[Bug 26373] Solidworks 2010 crash ( regression )

http://bugs.winehq.org/show_bug.cgi?id=26373

--- Comment #2 from denis bonnenfant  2011-03-10 15:44:18 CST ---
(In reply to comment #1)
> Please attach terminal output.

There was nothing related. Don't know which channel to enable. Server ?

The problem is caught by SW itself, so from wine's side the program exits
cleanly.
> Did you try starting from a clean prefix?

No, as SW installation is a really long process on my computer. But I
reproduced it on three different installs. But if you think that it is worth to
try, I can do it tomorrow.

[Bug 26373] Solidworks 2010 crash ( regression )

http://bugs.winehq.org/show_bug.cgi?id=26373

--- Comment #3 from Hans Leidekker  2011-03-11 01:55:38 CST ---
(In reply to comment #2)
> > Did you try starting from a clean prefix?
>
> No, as SW installation is a really long process on my computer. But I
> reproduced it on three different installs. But if you think that it is worth to
> try, I can do it tomorrow.

It's likely that your app breaks because the user SID is changed by
this patch, which won't match the one stored in the prefix anymore.

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

Dmitry Timoshkov  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|Solidworks 2010 crash (     |Solidworks 2010 crashes
                   |regression )                |

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

GyB  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #4 from GyB  2011-03-11 09:38:17 CST ---
The same commit caused me problems in certain games which contain Securom
activation: after the commit Securom no longer accepted the existing
registration info (stored in the registry and on the disk, under
/users/your_username/Application Data/Securom directory) and it required a new
activation.
It is quite an annoying problem if the game/app has activation limit (Securom
usually allows 3-5 activations with the same serial).

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

--- Comment #5 from Hans Leidekker  2011-03-11 09:56:36 CST ---
(In reply to comment #4)
> The same commit caused me problems in certain games which contain Securom
> activation: after the commit Securom no longer accepted the existing
> registration info (stored in the registry and on the disk, under
> /users/your_username/Application Data/Securom directory) and it required a new
> activation.
> It is quite an annoying problem if the game/app has activation limit (Securom
> usually allows 3-5 activations with the same serial).

Well, the SID is stable now, it won't change again when you create a new
prefix, which should make such registration data portable across prefixes.

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

--- Comment #6 from GyB  2011-03-11 10:06:12 CST ---
(In reply to comment #5)
> Well, the SID is stable now, it won't change again when you create a new
> prefix, which should make such registration data portable across prefixes.

You answered my untold question: will that part of Wine change frequently in
the future?
Thanks for your reply, that's quite reassuring.

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

--- Comment #7 from Dmitry Timoshkov  2011-03-12 00:06:03 CST ---
Is this basically a WONTFIX then?

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

--- Comment #8 from Hans Leidekker  2011-03-12 02:14:44 CST ---
(In reply to comment #7)
> Is this basically a WONTFIX then?

I'd say INVALID, if the reporter can confirm that starting from a fresh
prefix makes the crash go away.

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

--- Comment #9 from Dmitry Timoshkov  2011-03-12 02:49:24 CST ---
(In reply to comment #8)
> I'd say INVALID, if the reporter can confirm that starting from a fresh
> prefix makes the crash go away.

Is there a way to detect such configurations, so that bug reports with
old configs could be easily identified?

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

--- Comment #10 from Hans Leidekker  2011-03-12 04:50:16 CST ---
(In reply to comment #9)
> Is there a way to detect such configurations, so that bug reports with
> old configs could be easily identified?

Look at user.reg. If the first two lines do not read like this it's
an old prefix:

WINE REGISTRY Version 2
;; All keys relative to \User\S-1-5-21-0-0-0-1000

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

--- Comment #11 from Dmitry Timoshkov  2011-03-14 00:37:45 CDT ---
(In reply to comment #10)
> > Is there a way to detect such configurations, so that bug reports with
> > old configs could be easily identified?
>
> Look at user.reg. If the first two lines do not read like this it's
> an old prefix:
>
> WINE REGISTRY Version 2
> ;; All keys relative to \User\S-1-5-21-0-0-0-1000

I mean at run-time, since obviously there are users with existing installs
and it's going to be a support headache.

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

--- Comment #12 from Hans Leidekker  2011-03-14 03:07:18 CDT ---
(In reply to comment #11)
> I mean at run-time, since obviously there are users with existing installs
> and it's going to be a support headache.

Yeah, a crash might be the only user-visible indicator. A +reg
trace should reveal the user SID, and if +advapi shows calls to
GetTokenInformation(TokenUser) and LookupAccountSid it's likely
that the app depends on the SID <-> username mapping.

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

--- Comment #13 from denis bonnenfant  2011-03-15 11:39:04 CDT ---
(In reply to comment #12)
> (In reply to comment #11)
> > I mean at run-time, since obviously there are users with existing installs
> > and it's going to be a support headache.
>
> Yeah, a crash might be the only user-visible indicator. A +reg
> trace should reveal the user SID, and if +advapi shows calls to
> GetTokenInformation(TokenUser) and LookupAccountSid it's likely
> that the app depends on the SID <-> username mapping.

OK, now I understand. There are some informations related to the old
interactive SID stored in HKLM. They need to be renamed to local user SID.

For Solidworks, it was in 
HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\Installer\UserData

With this hack SolidWorks works perfectly without reinstalling, and I guess
that it would be the same for other apps.

So, basically

sed -i "s/S-1-5-4/S-1-5-21-0-0-0-1000/g"  $WINEPREFIX/system.reg

will do the trick.

Adding it to a winetricks verb, or, better, as a test launched at first 1.3.16
start (when WINEPREFIX is updated) may be a good thing.  I tested it on a few
prefixes and it doesn't break anything.

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

Hans Leidekker  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID

--- Comment #14 from Hans Leidekker  2011-03-19 06:03:54 CDT ---
Thanks for the confirmation.

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

Hans Leidekker  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #15 from Hans Leidekker  2011-03-19 06:04:15 CDT ---
Closing.

[Bug 26373] Solidworks 2010 crashes

http://bugs.winehq.org/show_bug.cgi?id=26373

--- Comment #16 from denis bonnenfant  2011-03-19 06:48:48 CDT ---
(In reply to comment #15)
> Closing.

Adding a warning on SolidWork's Appdb page.