[Bug 13915] New: Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Summary: Oblivon Script Extender (OBSE) fails to start
                    ("privileged instruction")
           Product: Wine
           Version: 1.0-rc4
          Platform: PC-x86-64
               URL: http://obse.silverlock.org
        OS/Version: Linux
            Status: NEW
          Keywords: download
          Severity: normal
          Priority: P2
         Component: -unknown
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]

Created an attachment (id=14025)
 --> (http://bugs.winehq.org/attachment.cgi?id=14025)
Full terminal crashlog

The Oblivion Script Extender program fails to start, it immediately dies with
an "Unhandled exception: privileged instruction in 32-bit code (0x7ee158d0)."
error although it then continues to load Oblivion which may confuse people as
unless you see the console in the split second before before Oblivion actually
opens, it makes it appear as if it worked.

Full terminal output attached.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Josef Schimke  changed:

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

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #1 from David Goree   2008-07-26 22:48:08 ---
Created an attachment (id=15075)
 --> (http://bugs.winehq.org/attachment.cgi?id=15075)
Backtrace

Still crashing on wine 1.10 on Linux-2.6.26-gentoo-i686. Added an attachment.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #2 from Rob Whalley   2008-07-27 09:32:32 ---
Created an attachment (id=15093)
 --> (http://bugs.winehq.org/attachment.cgi?id=15093)
Terminal output of 'wine obse_loader.exe'

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Rob Whalley  changed:

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

--- Comment #3 from Rob Whalley   2008-07-27 09:33:13 ---
Tested with:

Ubuntu 8.04 64-bit
Wine version v1.1.1 (Git version)
OBSE: v0015
Latest patched version of Oblivion (original retail CD only, no expansion
packs)

Error when using command 'wine obse_loader.exe' to start:

"wine: Unhandled page fault on read access to 0x00000010 at address 0x7ebefd6e
(thread 0017), starting debugger...
Unhandled exception: page fault on read access to 0x00000010 in 32-bit code
(0x7ebefd6e)."

Attached terminal output.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Oliver Dietsch  changed:

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

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Nicolas Joyard  changed:

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

--- Comment #4 from Nicolas Joyard   2008-11-01 22:29:07 ---
According to the UESP wiki, the author of OBSE (Ian Patterson) has been
contacted about this issue. Here's his answer, it might be useful to solve this
bug...

--

This appears to be a discrepancy between the Windows implementation of the DLL
dependency handler/loader and the Wine implementation. Unfortunately I don't
have a linux install to test this on, but my first guess would be that calling
CreateProcess with the CREATE_SUSPENDED flag set leaves the new process' loader
critical section locked until the main thread is resumed.

The loader works in a pretty straightforward way:

- create the new process with CREATE_SUSPENDED
- VirtualAllocEx some executable memory in the new process to store enough code
for "jmp LoadLibraryA" and a null-terminated string pointing to our DLL (in a
perfect world we would just point the thread entry point straight at
LoadLibraryA, but this makes some virus scanners explode, literally)
- use CreateRemoteThread to create a thread in the new process pointing to the
injected code, setting the parameter to the thread function to the DLL path in
the new process (conveniently the calling convention and signature of thread
entry points and LoadLibrary are exactly the same)
- wait for the thread to terminate, signifying that the DLL has been loaded
- clean up

This sequence requires that processes not be holding the loader lock
immediately after creation.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Daniel Rammelt  changed:

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

--- Comment #5 from Daniel Rammelt   2008-11-03 05:43:34 ---
Apparently the issue with CREATE_SUSPENDED is fixed, I emailed a backtrace from
1.1.7 and the developer said the new problem is to do with the CreateMenuEx
function.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #6 from Daniel Rammelt   2008-11-03 05:44:35 ---
Created an attachment (id=17060)
 --> (http://bugs.winehq.org/attachment.cgi?id=17060)
Trace from Obse with Wine 1.1.7

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #7 from Daniel Rammelt   2008-11-03 16:30:30 ---
More specifically this line is where it is crashing in dlls/user32/menu.c
In function CreateMenu.

if (!(hMenu = USER_HEAP_ALLOC( sizeof(POPUPMENU) ))) return 0;

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #8 from Daniel Rammelt   2008-11-03 16:48:45 ---
Placing some extra debug code around the line causing the crash actually allows
OBSE to run perfectly on occasion. It almost seems like a timing issue of some
sort.

I replaced the debug line with a simple sleep(1); statement and it has the same
effect. Probably 1 out of 5 tries will launch OBSE perfectly. The rest it
crashes in FormatMessageW or LocalReAlloc16.

sleep(1);
    if (!(hMenu = USER_HEAP_ALLOC( sizeof(POPUPMENU) ))) return 0;

I am not really sure what this means, but perhaps it will help someone with a
bit more coding skill :)

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

DL  changed:

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

--- Comment #9 from DL   2009-01-06 19:21:35 ---
Still broken with wine-1.1.12.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Edward  changed:

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

--- Comment #10 from Edward   2009-02-10 19:58:23 ---
(In reply to comment #8)
> Placing some extra debug code around the line causing the crash actually allows
> OBSE to run perfectly on occasion. It almost seems like a timing issue of some
> sort.
>
> I replaced the debug line with a simple sleep(1); statement and it has the same
> effect. Probably 1 out of 5 tries will launch OBSE perfectly. The rest it
> crashes in FormatMessageW or LocalReAlloc16.
> sleep(1);
> if (!(hMenu = USER_HEAP_ALLOC( sizeof(POPUPMENU) ))) return 0;
>
> I am not really sure what this means, but perhaps it will help someone with a
> bit more coding skill :)

Still broken with wine 1.1.14

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Samuel Nelson  changed:

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

--- Comment #11 from Samuel Nelson   2009-04-17 02:15:37 ---
With wine 1.1.13, it seems that OBSE will run maybe 1/10 of the time, but it
also seems that Oblivion will subsequently crash while loading and old or
starting a new game (if OBSE mods are in use - I've got Deadly Reflexes
enabled, myself).  Most of the time, a similar memory dump to the previously
posted backtraces happens while attempting to load OBSE, but Oblivion will
still start after OBSE crashes.

I edited the source code and added the previously mentioned sleep(1); to see if
it would cause OBSE to work more frequently, but it didn't seem to change
anything.

I hate vanilla Oblivion, and some of the mods that I feel are essential to make
it an enjoyable game require OBSE, so I, personally, would greatly appreciate
any work done on this - even if I have to start, quit, restart, quit, etc a few
times in order to get OBSE to run correctly.  I don't mind having to work at
starting the game for twenty minutes, so long as I can play with my mods.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Kasperi Kuuskoski  changed:

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

--- Comment #12 from Kasperi Kuuskoski   2009-05-27 17:06:36 ---
It works! It works, at last! The new v0017 beta 2 works perfectly with wine, at
least for me. I use wine 1.1.14, Fedora 9. Havent tested it with mods yet, but
the gameplay works.

Okay, it doesnt work perfectly... it freezes in gameplay few times, but i am
not sure is this because of my computer/mods or because of OBSE. Well, people,
test this, i am eager to see does it work with you.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Ken Sharp  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED

--- Comment #13 from Ken Sharp   2009-05-27 20:39:03 ---
Reported fixed, reopen if not.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #14 from DL   2009-05-27 21:23:27 ---
Bug is not fixed.The difference is that obse still crashes but Oblivion
launches anyway.Mods such as Streamline and Helmet FOV don't work, and oblivion
often crashes while loading a save.Probably because it is trying to use these
mods without OBSE being loaded.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Ken Sharp  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |

--- Comment #15 from Ken Sharp   2009-05-28 13:32:27 ---
Can you attach a new terminal log?

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #16 from [e-mail]  2009-06-13 19:00:16 ---
Created an attachment (id=21780)
 --> (http://bugs.winehq.org/attachment.cgi?id=21780)
Terminal output of 'wine obse_loader.exe'

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

[e-mail] changed:

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

--- Comment #17 from [e-mail]  2009-06-13 19:03:21 ---
(In reply to comment #16)
> Created an attachment (id=21780)

--> (http://bugs.winehq.org/attachment.cgi?id=21780) [details]
> Terminal output of 'wine obse_loader.exe'

Sorry for the double post. Forgot to include:

Distro: Ubuntu 9.04 'Jaunty' x86_64
Uname: Linux core 2.6.28-11-generic #42-Ubuntu SMP Fri Apr 17 01:58:03 UTC 2009
x86_64 GNU/Linux
Wine Version: 1.1.23
Game Version: Oblivion + Shivering Isles v1.2.416

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Rafael Hart  changed:

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

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Jari Vetoniemi  changed:

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

--- Comment #18 from Jari Vetoniemi   2009-08-05 09:46:58 ---
I want to report too that OBSE loads randomly fine and when Oblivion launches
OBSE crashes. Same for FOSE (Fallout 3 Script Extender) I think there is source
code for these programs so maybe someone could take fast look how it injects
into Oblivion exe (or Fallout 3).

It seems its not so far away =)

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Jari Vetoniemi  changed:

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

Philip Nilsson  changed:

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

--- Comment #18 from Jari Vetoniemi   2009-08-05 09:46:58 ---
I want to report too that OBSE loads randomly fine and when Oblivion launches
OBSE crashes. Same for FOSE (Fallout 3 Script Extender) I think there is source
code for these programs so maybe someone could take fast look how it injects
into Oblivion exe (or Fallout 3).

It seems its not so far away =)

--- Comment #19 from Philip Nilsson   2009-08-05 09:54:23 ---
Try ``sudo sysctl kernel.randomize_va_space=0'' before running with a recent
Wine.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Jari Vetoniemi  changed:

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

Philip Nilsson  changed:

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

Adam A. Zajac  changed:

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

--- Comment #18 from Jari Vetoniemi   2009-08-05 09:46:58 ---
I want to report too that OBSE loads randomly fine and when Oblivion launches
OBSE crashes. Same for FOSE (Fallout 3 Script Extender) I think there is source
code for these programs so maybe someone could take fast look how it injects
into Oblivion exe (or Fallout 3).

It seems its not so far away =)

--- Comment #19 from Philip Nilsson   2009-08-05 09:54:23 ---
Try ``sudo sysctl kernel.randomize_va_space=0'' before running with a recent
Wine.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Jari Vetoniemi  changed:

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

Philip Nilsson  changed:

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

Adam A. Zajac  changed:

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

Claudio  changed:

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

--- Comment #18 from Jari Vetoniemi   2009-08-05 09:46:58 ---
I want to report too that OBSE loads randomly fine and when Oblivion launches
OBSE crashes. Same for FOSE (Fallout 3 Script Extender) I think there is source
code for these programs so maybe someone could take fast look how it injects
into Oblivion exe (or Fallout 3).

It seems its not so far away =)

--- Comment #19 from Philip Nilsson   2009-08-05 09:54:23 ---
Try ``sudo sysctl kernel.randomize_va_space=0'' before running with a recent
Wine.

--- Comment #20 from Claudio   2009-09-06 13:54:34 ---
(In reply to comment #19)
> Try ``sudo sysctl kernel.randomize_va_space=0'' before running with a recent
> Wine.

Tried, does not help. Using wine-1.1.28.
Does someone have a somewhat reliable workaround?

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Jari Vetoniemi  changed:

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

Philip Nilsson  changed:

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

Adam A. Zajac  changed:

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

Claudio  changed:

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

Thomas J. Moore  changed:

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

--- Comment #18 from Jari Vetoniemi   2009-08-05 09:46:58 ---
I want to report too that OBSE loads randomly fine and when Oblivion launches
OBSE crashes. Same for FOSE (Fallout 3 Script Extender) I think there is source
code for these programs so maybe someone could take fast look how it injects
into Oblivion exe (or Fallout 3).

It seems its not so far away =)

--- Comment #19 from Philip Nilsson   2009-08-05 09:54:23 ---
Try ``sudo sysctl kernel.randomize_va_space=0'' before running with a recent
Wine.

--- Comment #20 from Claudio   2009-09-06 13:54:34 ---
(In reply to comment #19)
> Try ``sudo sysctl kernel.randomize_va_space=0'' before running with a recent
> Wine.

Tried, does not help. Using wine-1.1.28.
Does someone have a somewhat reliable workaround?

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Robert Myers  changed:

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

--- Comment #21 from Robert Myers   2009-10-28 11:07:33 ---
Yes, the source codes are available, and they are included with the download,
under the "src" directory.  OBSE and it's source codes can be downloaded from
this link:

http://obse.silverlock.org/download/obse_0017b.zip

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Austin English  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |source

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Eddahbi Karim  changed:

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

--- Comment #22 from Eddahbi Karim   2009-11-06 17:58:41 ---
Can't execute the loader with wine 1.1.32 and no extension loaded but the
Nvidia Black Fix bug (which don't make use of this loader).
Here's the error :

err:seh:setup_exception_record stack overflow 2300 bytes in thread 001b eip
7ef756fd esp 00e00a34 stack 0xe00000-0xe01000-0xf00000

Nothing after that.

Same thing happen with 0017 and 0018b version of OBSE.

Using Debian Sid x86-64 packages from the sources stated on the WineHQ's
download page.

Note that obse_loader.log states the following :
crc = 7934C86C
dll = K:\Oblivion\obse_1_2_416.dll
hookBase = 00340000
loadLibraryAAddr = 7ED9BA17
hook thread complete
launching oblivion

Note also that by executing the following command "sysctl
kernel.randomize_va_space=1", the game can be launched after an initial crash,
with the same error that the original poster have on its terminal, but no new
functions are added, floor textures are messed up and the obse_loader.log
'loadLibraryAAddr' line change every time the game can be launched.

I was wondering if that injection can't be replaced by a binary patch.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #23 from Eddahbi Karim   2009-11-10 13:51:33 ---
Just for information, launching the game under Windows Vista with
obse_loader.exe creates a 'obse_loader.log' with the following content :

crc = 7934C86C
dll = C:\Games\Oblivion\obse_1_2_416.dll
hookBase = 00160000
loadLibraryAAddr = 77A59491
hook thread complete
launching oblivion

The game's version is the same since I just did a copy/paste of the files and
registry keys.

BTW, the texture's corruption was unrelated. Still, the loading problem is
still there.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #24 from Dmitry Timoshkov   2009-11-13 04:38:19 ---
(In reply to comment #23)
> The game's version is the same since I just did a copy/paste of the files and
> registry keys.

That configuration is invalid and can't be supported. Install the game
under Wine.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #25 from Eddahbi Karim   2009-11-13 14:28:46 ---
I *did* install the game under Wine normally, I just copy/pasted the game to
install it in *Windows*.
Not the reverse. So this is still valid !

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #26 from Eddahbi Karim   2009-11-19 00:09:53 ---
Tested again with wine 1.1.32 this time. Didn't touch the software files.

I got a little further. First, the log is now always consistent :
crc = 7934C86C
dll = K:\Oblivion\obse_1_2_416.dll
hookBase = 00340000
loadLibraryAAddr = 7B8573D0
hook thread complete
launching oblivion

Then, I got messages from the loader on the console and the same message in the
obse.log. Things I didn't have before. Here's the messages :
OBSE: initialize (version = 17.2 010201A0)
oblivion root = K:\Oblivion\
plugin directory = K:\Oblivion\Data\OBSE\Plugins\
patched

But I can't launch the game with it. It always crash now. The crash isn't
consistent if I do "sysctl kernel.randomize_va_space=1" before. Sometimes a
dialog box, intitled "Microsoft Visual C++ Debug Library", with the following
message :

"Debug Assertion Failed!

Program: K:\Oblivion\obse_loader.exe
File: fclose.c
Line: 47

Expression: (stream != NULL)

For information on how your program can cause an assertion failure, see the
Visual C++ documentation on asserts.

(Press Retry to debug the application)"

Last thing, when I enable the kernel.randomize_va_space option, every crash
brings a wine error with the following pattern :
"wine: Unhandled page fault on read access to 0x0804a3b4 at address 0xf7fXX387
(thread 003e), starting debugger..." where XX is an hexadecimal value

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #27 from Eddahbi Karim   2009-11-19 00:10:47 ---
I meant wine-1.1.33. Sorry for the spam.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Ethan Baldridge  changed:

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

--- Comment #28 from Ethan Baldridge   2009-11-21 22:30:31 ---
OBSE has been working for me since wine 1.1.32. This is on Ubuntu 9.10 with the
wine.budgetdedicated.com build for jaunty. Earlier today with 1.1.33 was the
first time I saw OBSE crash when launching Oblivion. (Oblivion launched but one
of the plugins [Put it in its Place] detected that OBSE wasn't running, popped
up a little message and closed the game) Launching it a second time worked
fine.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Myk Taylor  changed:

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

--- Comment #29 from Myk Taylor   2009-11-30 15:03:38 ---
OBSE seems to load fine for me now.  OBSE 17 and OBSE 18b4 were both
successfully tested.  Gentoo Linux x86, 2.6.31-gentoo-r6 kernel, wine-1.1.33,
GeForce GTX 285, nvidia-drivers-185.18.36.  I've run Oblivion with OBSE
successfully about 20 times without a single failure.  Tested with Streamline,
Gather Ye Rosebuds, and SuperHotKeys, among others.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Markus  changed:

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

--- Comment #30 from Markus   2009-12-07 21:23:04 ---
Oblivion with Obse runs fine under Wine 1.1.33 but in Version 1.1.34 it crashes
at the Loadscreen. Testet under Ubuntu 9.10 Karmic 64 Bit.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Evan Goers  changed:

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

--- Comment #31 from Evan Goers   2009-12-08 00:22:45 ---
(In reply to comment #30)
> Oblivion with Obse runs fine under Wine 1.1.33 but in Version 1.1.34 it crashes
> at the Loadscreen. Testet under Ubuntu 9.10 Karmic 64 Bit.

I still get a page fault in 1.1.33. It seems to launch oblivion but
obse_loader.exe crashes.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #32 from Myk Taylor   2009-12-08 01:19:56 ---
How do your systems differ from mine (see comment #29)?

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #33 from Evan Goers   2009-12-08 01:37:12 ---
(In reply to comment #32)
> How do your systems differ from mine (see comment #29)?

I'm running Debian Testing/Unstable and Wine 1.1.33 from the lamaresh repos.
x86-64, Linux 2.6.31.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #34 from Myk Taylor   2009-12-08 10:23:05 ---
(In reply to comment #33)
> I'm running Debian Testing/Unstable and Wine 1.1.33 from the lamaresh repos.
> x86-64, Linux 2.6.31.

Graphics adapter? Driver version?

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #35 from Markus   2009-12-08 14:53:52 ---
(In reply to comment #32)
> How do your systems differ from mine (see comment #29)?

Ich tried it with Obse 17 on Ubuntu 9.10 64Bit, Linux x86 2.6.31-16-generic.
Wine 1.1.33, Geforce 9800gtx+ with the 185.18.36. Drivers.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #36 from Evan Goers   2009-12-08 17:11:03 ---
(In reply to comment #34)
> (In reply to comment #33)
> > I'm running Debian Testing/Unstable and Wine 1.1.33 from the lamaresh repos.
> > x86-64, Linux 2.6.31.
>
> Graphics adapter? Driver version?

NVIDIA Geforce 8800GTS, nvidia 190.42. I doubt this info is relevant to the
problem though.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Rafael Hart  changed:

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

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #37 from Eddahbi Karim   2009-12-10 18:06:57 ---
It seems to be a problem related to addressing so... Maybe sysctl values can
change that. Like I said, changing some kernel adress randomization, like
stated by someone before my interventions, does change the behaviour. It
changes from "Don't start" to "Start but crash".

So, maybe a list of sysctl values, related to adressing, or the kernel
configuration used would maybe allow to pinpoint the source of the problem
correctly. 
Anyway, if that works, could you give us your game version and the content of
"obse_loader.log", located in the game folder ?

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #38 from Evan Goers   2009-12-10 22:16:28 ---
(In reply to comment #37)
> It seems to be a problem related to addressing so... Maybe sysctl values can
> change that. Like I said, changing some kernel adress randomization, like
> stated by someone before my interventions, does change the behaviour. It
> changes from "Don't start" to "Start but crash".
>
> So, maybe a list of sysctl values, related to adressing, or the kernel
> configuration used would maybe allow to pinpoint the source of the problem
> correctly. 
> Anyway, if that works, could you give us your game version and the content of
> "obse_loader.log", located in the game folder ?

I'm using OBSE 0017b, Oblivion 1.2.0.416(Shivering Isles).

To add to the mystery, if you use OBMM to launch Oblivion, it actually launches
obse_loader.exe, which in turn launches Oblivion.exe. It does *not* crash if
you use OBMM to run Oblivion enhanced with OBSE.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Kelly Miller  changed:

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

--- Comment #39 from Kelly Miller   2010-01-16 18:02:04 ---
When I try to run it, I get this:

bash-4.0$ wine ./obse_loader.exe
wine: Unhandled illegal instruction at address 0x68421e80 (thread 0035),
starting debugger...
Process::InstallHook: waiting for thread = WAIT_TIMEOUT
Couldn't inject dll.
terminating oblivion process
bash-4.0$ wine: Unhandled page fault on read access to 0x0000002a at address
0x73228134 (thread 000b), starting debugger...

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Miro Stoyanov  changed:

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

--- Comment #40 from Miro Stoyanov   2010-01-19 17:29:12 ---
I can confirm comment #28 From  Ethan Baldridge:

I use 64-bit Ubuntu 9.10 Karmic kernel = 2.6.31-17-generic and using wine 1.2
from the official Ubuntu repos refused to work for me. When I used the Jaunty
build from http://wine.budgetdedicated.com/ there was no problem.

I am currently using obse 17b with Toggleable Quantity Prompt mod (in
conjunction with AF Leveling Mod which does not require obse). Some of the
leveling aspects of the game seem to be deliberately designed to torture
players (bad points for Bethesda) so I consider this bug to be very important
as far as Oblivion is concerned.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #41 from Claudio   2010-01-19 18:37:57 ---

(In reply to comment #40)
> I am currently using obse 17b with Toggleable Quantity Prompt mod (in
> conjunction with AF Leveling Mod which does not require obse). Some of the
> leveling aspects of the game seem to be deliberately designed to torture
> players (bad points for Bethesda) so I consider this bug to be very important
> as far as Oblivion is concerned.

This might be a little bit OT, but I could work around the
obvious game design mistakes of Bethesda without any need for OBSE.

I use the latest unofficial patch 3.2.0 +
the OBMM (Omod installer) +
Oscuro Oblivion Overhaul v 1.33 +
Kobus' Character Advancement system v. 2.4.1

It makes the game great for me, without depending on OBSE, which I could never
get to work with wine.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #42 from Myk Taylor   2010-08-18 12:06:54 ---
Created an attachment (id=30211)
 --> (http://bugs.winehq.org/attachment.cgi?id=30211)
crashlog for wine-1.3.0

OBSE was working perfectly for me in previous versions of wine.  But now with
wine-1.3.0 I consistently get page faults.  Like the original poster reports,
Oblivion loads successfully after the "crash", but without OBSE.  This is
actually the first time that Oblivion has loaded for me after an obse_loader
crash.  In previous wine versions (before it started working consistently, back
when it used to work randomly) Oblivion would not start if I got an initial
crash.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #43 from Myk Taylor   2010-08-20 11:51:34 ---
It appears the the crash is happening in the thread created by

HANDLE  thread = CreateRemoteThread(process, NULL, 0,
              (LPTHREAD_START_ROUTINE)hookBase, (void *)(hookBase + 5), 0,
NULL);

in the DoInjectDLL function.  Sometimes I get a page fault on a valid address,
sometimes on NULL, and sometimes it works perfectly, which kinda indicates a
race condition somewhere.

One of the following lines is going wonky:

HANDLE    process = OpenProcess( PROCESS_CREATE_THREAD |
PROCESS_QUERY_INFORMATION | PROCESS_VM_OPERATION | PROCESS_VM_WRITE |
PROCESS_VM_READ, FALSE, info->dwProcessId);

UInt32    hookBase = (UInt32)VirtualAllocEx(process, NULL, 8192, MEM_COMMIT,
PAGE_EXECUTE_READWRITE);

UInt32    loadLibraryAAddr =
(UInt32)GetProcAddress(GetModuleHandle("kernel32.dll"), "LoadLibraryA");

WriteProcessMemory(process, (LPVOID)(hookBase + 5), dllPath, strlen(dllPath) +
1, &bytesWritten);

WriteProcessMemory(process, (LPVOID)(hookBase), hookCode, sizeof(hookCode),
&bytesWritten);

HANDLE    thread = CreateRemoteThread(process, NULL, 0,
(LPTHREAD_START_ROUTINE)hookBase, (void *)(hookBase + 5), 0, NULL);

Of course, it could just be that this line is not actually created a suspended
process:
    bool result = CreateProcess(
        procName,
        NULL,    // no args
        NULL,    // default process security
        NULL,    // default thread security
        TRUE,    // don't inherit handles
        CREATE_SUSPENDED,
        NULL,    // no new environment
        NULL,    // no new cwd
        &startupInfo, &procInfo) != 0;

which would invalidate the assumptions in DoInjectDLL...

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Anastasius Focht  changed:

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

--- Comment #44 from Anastasius Focht   2010-08-20 13:13:24 ---
Hello,

could be anything ... looking at that snippet: make sure you switch address
space randomization off for sure!
If you are really sure that kernel32.dll gets the same base address in child
process the loader creates, provide the following log in case of a crash :

$ WINEDEBUG=+tid,+seh,+loaddll,+module,+process,+relay wine ./loader_foo.exe
> > log.txt 2>&1

(make sure log.txt doesn't exist prior)

Injecting stuff has its limitations. You can't do everything at that point
because the main thread hasn't run to process entry yet.
Imports are fixed but process attach notifications are not yet sent at that
point.
They will get triggered by any injected threads that run _prior_ waking up the
main thread that ought to call it.

Does your injected dll synchronize with the parent (loader) to wake up the
initially suspended main thread?
Be careful if your dll patches stuff from the game in its dllmain, especially
dynamic data structures which might not be fully present yet/or are being
modified in startup phase of game.

Regards

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #45 from Myk Taylor   2010-08-22 01:18:17 ---
Created an attachment (id=30280)
 --> (http://bugs.winehq.org/attachment.cgi?id=30280)
crash log with WINEDEBUG=+tid,+seh,+loaddll,+module,+process,+relay

requested log file (gzipped due to size).  I hit ctrl-C after the crash to
prevent Oblivion from loading anyway (sans OBSE) and cluttering up the log.  no
change in behavior when explicitly turning off address space randomization
(tried both setarch i686 -R and setarch x86_64 -R).  This isn't my code, but
from reading the source (available at http://obse.silverlock.org/) I see that
ResumeThread is called on the suspended process after the DLL is injected into
the remote process' address space.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #46 from Anastasius Focht   2010-08-22 05:04:58 ---
Hello,

I extracted the relevant parts from your log:

--- snip ---
0009:trace:loaddll:load_builtin_dll Loaded L"KERNEL32.dll" at 0x7ed60000:
builtin 
...
0009:Call KERNEL32.CreateFileA(00422430
"C:\oblivion\obse_1_2_416.dll",80000000,00000001,00000000,00000003,00000080,00000000)
ret=00403788
0009:Ret  KERNEL32.CreateFileA() retval=00000040 ret=00403788
...
0009:Call KERNEL32.CreateProcessA(00415834
"Oblivion.exe",00000000,00000000,00000000,00000001,00000004,00000000,00000000,0032fdb8,0032fcf4)
ret=00402fa2
0009:trace:process:CreateProcessW app L"Oblivion.exe" cmdline (null)
0009:trace:process:open_exe_file looking for L"Oblivion.exe"
0009:trace:process:CreateProcessW starting L"Oblivion.exe" as Win32 binary
(0x400000-0xbaf000) 
001b:trace:module:load_builtin_callback loaded ntdll.dll 0x110158 0x7ef60000
...
001b:trace:module:load_dll Found L"ntdll.dll" for L"ntdll.dll" at 0x7ef60000,
count=2
001b:trace:module:load_builtin_callback loaded KERNEL32.dll 0x112628 0x7ed90000
001b:trace:loaddll:load_builtin_dll Loaded L"KERNEL32.dll" at 0x7ed90000:
builtin
001b:Call KERNEL32.__wine_kernel_init() ret=7ef97f83
...
001b:trace:loaddll:load_native_dll Loaded L"C:\oblivion\Oblivion.exe" at
0x400000: native
001b:trace:module:load_dll Loaded module L"C:\oblivion\Oblivion.exe" (native)
at 0x400000 
...
0009:trace:process:CreateProcessW started process pid 001a tid 001b
0009:Ret  KERNEL32.CreateProcessA() retval=00000001 ret=00402fa2
0009:Call KERNEL32.OpenProcess(0000043a,00000000,0000001a) ret=004024d0
0009:Ret  KERNEL32.OpenProcess() retval=00000040 ret=004024d0
0009:Call KERNEL32.VirtualAllocEx(00000040,00000000,00002000,00001000,00000040)
ret=004024f0 
...
0009:Call
KERNEL32.WriteProcessMemory(00000040,00340005,00422430,0000001d,0032fc84)
ret=00402556
0009:Ret  KERNEL32.WriteProcessMemory() retval=00000001 ret=00402556
0009:Call
KERNEL32.WriteProcessMemory(00000040,00340000,0032fc88,00000005,0032fc84)
ret=00402578
0009:Ret  KERNEL32.WriteProcessMemory() retval=00000001 ret=00402578
0009:Call
KERNEL32.CreateRemoteThread(00000040,00000000,00000000,00340000,00340005,00000000,00000000)
ret=00402589
0009:Ret  KERNEL32.CreateRemoteThread() retval=00000050 ret=00402589
0009:Call KERNEL32.WaitForSingleObject(00000050,00007530) ret=004025a4
001c:trace:module:alloc_thread_tls thread 001c idx 0: 440/0 bytes from 0xbac000
to 0x126538
001c:Starting thread proc 0x340000 (arg=0x340005)
001c:trace:seh:raise_exception code=c0000005 flags=0 addr=0x7ed6cf50
ip=7ed6cf50 tid=001c
001c:trace:seh:raise_exception  info[0]=00000000
001c:trace:seh:raise_exception  info[1]=7ed6cf50
001c:trace:seh:raise_exception  eax=00340000 ebx=7efe3ff4 ecx=8ccebec1
edx=ffffffff esi=7ffd4f10 edi=00340000
001c:trace:seh:raise_exception  ebp=00efea78 esp=00efea6c cs=0023 ds=002b
es=002b fs=0063 gs=006b flags=00010206
001c:trace:seh:call_stack_handlers calling handler at 0x7efce930 code=c0000005
flags=0
001c:Call KERNEL32.UnhandledExceptionFilter(00efe5b8) ret=7efce982
wine: Unhandled page fault on read access to 0x7ed6cf50 at address 0x7ed6cf50
(thread 001c), starting debugger...
001c:trace:seh:start_debugger Starting debugger "winedbg --auto 26 24" 
--- snip ---

It seems the kernel32 image load base is still different:

kernel32 base loader: 0x7ed60000
kernel32 base target: 0x7ed90000

The injected thread crashes at 0x7ed6cf50 which is below target kernel32 image
load base. The high word matches the loader (parent) kernel32 image base so the
patched jump trampoline 0x7ed6xxxx matches LoadLibraryA() - in loader address
space!

To cancel out address space randomization problem, we compare the load
addresses of kernel32.dll in all Wine processes:

0009 = initial process = loader

--- snip ---
0009:trace:module:load_builtin_callback loaded KERNEL32.dll 0x112628 0x7ed60000 
...
000b:trace:module:load_builtin_callback loaded KERNEL32.dll 0x112628 0x7ed90000
000b:trace:module:load_dll Loaded module L"C:\windows\system32\wineboot.exe"
(builtin) at 0x7ef30000 
...
000d:trace:module:load_builtin_callback loaded KERNEL32.dll 0x112628 0x7ed90000
000d:trace:module:load_dll Loaded module
L"C:\windows\system32\winemenubuilder.exe" (builtin) at 0x7ef30000 
...
000f:trace:module:load_builtin_callback loaded KERNEL32.dll 0x112628 0x7ed90000
000f:trace:module:load_dll Loaded module L"C:\windows\system32\services.exe"
(builtin) at 0x7ef30000 
...
0012:trace:module:load_builtin_callback loaded KERNEL32.dll 0x112628 0x7ed90000 
0012:trace:module:load_dll Loaded module
L"C:\windows\system32\winedevice.exe" (builtin) at 0x7ef40000
...
0019:trace:module:load_builtin_callback loaded KERNEL32.dll 0x112628 0x7ed90000
0019:trace:module:load_dll Loaded module L"C:\windows\system32\explorer.exe"
(builtin) at 0x7ef30000 
...
001b:trace:module:load_builtin_callback loaded KERNEL32.dll 0x112628 0x7ed90000 
001b:trace:module:load_dll Loaded module L"C:\oblivion\Oblivion.exe" (native)
at 0x400000 
--- snip ---

Interestingly only the _first_ process has a different kernel32.dll load base.
All others match.

Can you try the following: start a "helper" process (cmd.exe, notepad.exe
whatever) in background which keeps wineserver alive and then start the loader
from different terminal.

If it still crashes please give another +relay log exactly as previously
described.
Make sure winedbg can do a backtrace run (in your relay log it starts but never
reaches the point because you CTRL+C too early).

Regards

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Myk Taylor  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #30280|0                           |1
        is obsolete|                            |

--- Comment #47 from Myk Taylor   2010-08-22 21:25:10 ---
Created an attachment (id=30319)
 --> (http://bugs.winehq.org/attachment.cgi?id=30319)
crash log with WINEDEBUG=+tid,+seh,+loaddll,+module,+process,+relay

Requested log file.  I started cmd.exe in another shell, then loaded
obse_loader.exe.  The crash happened as usual.  I left Oblivion to load this
time, but it had not finished loading even after log.txt grew to about 38G in
size, so I eventually ran wineserver -k.  This is the first part of that log
file, up to where the stack traces are.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #48 from Anastasius Focht   2010-08-23 03:07:37 ---
Hello,

--- quote ---
Requested log file.  I started cmd.exe in another shell, then loaded
obse_loader.exe.  The crash happened as usual.  I left Oblivion to load this
time, but it had not finished loading even after log.txt grew to about 38G in
size, so I eventually ran wineserver -k.  This is the first part of that log
file, up to where the stack traces are.
--- quote ---

It seems the kernel32.dll load base is still different hence the crash.

Loader:

--- snip ---
001b:trace:module:load_builtin_callback loaded KERNEL32.dll 0x112628 0x7ed60000
001b:trace:loaddll:load_native_dll Loaded L"C:\oblivion\obse_loader.exe" at
0x400000: native
--- snip ---

vs. target:

--- snip ---
001d:trace:module:load_builtin_callback loaded KERNEL32.dll 0x112628 0x7ed90000
001d:trace:loaddll:load_native_dll Loaded L"C:\oblivion\Oblivion.exe" at
0x400000: native 
--- snip ---

Can you dump the process vm maps at the time of the crash and attach them
(don't CTRL+C before)?

--- snip (bash) ---
$ cat /proc/$(pidof -s 'obse_loader.exe')/maps >loader_mappings.txt

$ cat /proc/$(pidof -s 'Oblivion.exe')/maps >target_mappings.txt
--- snip (bash) ---

If that doesn't work, manually get the pids of the corresponding processes and
replace them as follows: /proc/your_pid_here/maps.

Regards

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #49 from Myk Taylor   2010-08-23 15:02:34 ---
Created an attachment (id=30337)
 --> (http://bugs.winehq.org/attachment.cgi?id=30337)
/proc/$(pidof -s 'obse_loader.exe')/maps

mappings file for obse_loader.exe

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #50 from Myk Taylor   2010-08-23 15:03:05 ---
Created an attachment (id=30338)
 --> (http://bugs.winehq.org/attachment.cgi?id=30338)
/proc/$(pidof -s 'Oblivion.exe')/maps

mappings file for Oblivion.exe

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #51 from Myk Taylor   2010-08-23 15:03:57 ---
I just want to say, thank you very much for helping with this.  This issue has
irked me for years.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #52 from Anastasius Focht   2010-08-23 16:30:47 ---
Hello,

the target (game) has much fewer shared libraries loaded in its early stage as
it was initially created suspended.
In contrast: the loader is fully initialized and has lots of X dependencies at
this point giving addresses spaces a different layout.

Looking at the corresponding kernel32 locations:

Loader:

--- snip ---
7eb28000-7eb48000 rw-p 00000000 00:00 0 
7eb48000-7ed48000 r--p 00000000 08:06 416308                            
/usr/lib64/locale/locale-archive

7ed48000-7ed60000 r-xp 00000000 08:06 154429                            
/usr/lib32/wine/kernel32.dll.so

7ed60000-7ed61000 rw-p 00000000 00:00 0 
7ed61000-7edcd000 r-xp 00019000 08:06 154429                            
/usr/lib32/wine/kernel32.dll.so
7edcd000-7edce000 r--p 00084000 08:06 154429                            
/usr/lib32/wine/kernel32.dll.so
7edce000-7edd6000 rw-p 00085000 08:06 154429                            
/usr/lib32/wine/kernel32.dll.so
7edd6000-7edd8000 rwxp 0008d000 08:06 154429                            
/usr/lib32/wine/kernel32.dll.so
7edd8000-7eebc000 rw-p 0008f000 08:06 154429                            
/usr/lib32/wine/kernel32.dll.so
--- snip ---

vs. target:

--- snip ---
7eb74000-7eb76000 rw-p 00000000 00:00 0 
7eb76000-7ed76000 r--p 00000000 08:06 416308                            
/usr/lib64/locale/locale-archive

7ed76000-7ed90000 r-xp 00000000 08:06 154429                            
/usr/lib32/wine/kernel32.dll.so

7ed90000-7ed91000 rw-p 00000000 00:00 0 
7ed91000-7edfb000 r-xp 0001b000 08:06 154429                            
/usr/lib32/wine/kernel32.dll.so
7edfb000-7edfc000 r--p 00084000 08:06 154429                            
/usr/lib32/wine/kernel32.dll.so
7edfc000-7ee04000 rw-p 00085000 08:06 154429                            
/usr/lib32/wine/kernel32.dll.so
7ee04000-7ee06000 rwxp 0008d000 08:06 154429                            
/usr/lib32/wine/kernel32.dll.so
7ee06000-7eeea000 rw-p 0008f000 08:06 154429                            
/usr/lib32/wine/kernel32.dll.so
--- snip ---

There is not much you can do.
Moving "locale-archive" is most likely not possible as this is glibc internal
(mmap'd) and not a standard ELF binary which could be relocated (using prelink
etc).

Can't you just get rid of the GUI stuff that pulls in the multitude of shared X
libs?
Making loader a console app might possibly fix this, giving both address spaces
a similar layout without having OS loader moving/relocating much stuff around.

Regards

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #53 from Anastasius Focht   2010-08-23 17:02:04 ---
Hello,

well, if you don't want to change much you should use a more reliable method of
getting kernel32 base/address of "LoadLibraryA" entry point.
There exist a several methods - though some don't work in Wine (yet) ;-)

One approach is to use psapi' EnumProcessModules() for getting the module list
(and load base addresses) in the remote process.

You could write a small helper function that uses the API and filters for
kernel32 base. 
To get the address of the "LoadLibraryA" function entry point in remote process
you just need to know the offset to the PE load base.
Subtract the locally retrieved "LoadLibraryA" entry point address from local
kernel32 base -> offset. Use that offset and the remote load base address to
calculate the virtual address for "LoadLibraryA" in remote process.

Regards

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #54 from Alexandre Julliard   2010-08-24 03:43:43 ---
Actually the offset from the base address may not be the same in the other
process, because the PE header needs to be aligned to a 64k boundary. You'd
have to look up the RVA in the remote PE header manually.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #55 from Anastasius Focht   2010-08-24 06:17:12 ---
Hello,

--- quote ---
Actually the offset from the base address may not be the same in the other
process, because the PE header needs to be aligned to a 64k boundary. You'd
have to look up the RVA in the remote PE header manually.
--- quote ---

Sure, populating remote process PE headers would be completely safe ...
Your scenario is more a corner case that applies to executables that are
specially crafted by people that make an art of "optimizing" everything in
headers (e.g. unusual load bases, section alignments) - unlikely to happen for
Wine core libraries (unless someone messed with the build system, explicitly
specifying preferred load bases).

I wanted to give him a "quick" solution without going through the hassle of
parsing PE headers and section tables.

Anyway, do you agree this bug is a WONTFIX?
Wine can't guarantee that core dlls (ntdll, kernel32, user32) will always get
the same load base in all address spaces (with ASLR switched off).

Though I wonder if certain Windows core dlls still have an "immutable" load
address (=preferred base).
It was a requirement in various Windows versions to support all these brain
damaged software that made certain assumptions about system dlls.
If some executable's preferred image base overlaps with core dlls, the infamous
"illegal system dll relocation" would appear and the OS loader refused to
process further.

Regards

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #56 from Alexandre Julliard   2010-08-24 06:33:42 ---
(In reply to comment #55)
> Sure, populating remote process PE headers would be completely safe ...
> Your scenario is more a corner case that applies to executables that are
> specially crafted by people that make an art of "optimizing" everything in
> headers (e.g. unusual load bases, section alignments) - unlikely to happen for
> Wine core libraries (unless someone messed with the build system, explicitly
> specifying preferred load bases).

No, it applies to all builtin libraries. The ELF header is loaded to 4k
alignment by the system loader, but the PE header is 64k aligned, so the entry
point offsets from the PE base will in general differ when the load address
differs.
> Anyway, do you agree this bug is a WONTFIX?
> Wine can't guarantee that core dlls (ntdll, kernel32, user32) will always get
> the same load base in all address spaces (with ASLR switched off).

In general it should work as long as prelink is found at compile time to set
the base address, and as long as the OS doesn't force randomization despite
prelink. But yes, there's nothing more we can do beyond that (though I guess we
could print a nasty warning when prelink isn't found).
> Though I wonder if certain Windows core dlls still have an "immutable" load
> address (=preferred base).
> It was a requirement in various Windows versions to support all these brain
> damaged software that made certain assumptions about system dlls.

That's certainly still true, and it's one of the reasons we set a fixed base
address for kernel32. Some copy protection schemes explicitly check that
kernel32 is within a certain range. And of course there are a million apps that
assume that core function addresses are identical across processes.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #57 from Myk Taylor   2010-08-24 10:15:51 ---
I have engaged the original developer (Ian Patterson) regarding this issue and
pointed him to Anastasius's comments.  Here is what he had to say:

"I've tried EnumProcessModules (plus various other psapi calls),
CreateToolhelp32Snapshot, plus just some basic stuff like querying the main
thread's current eip, but none of them work. A process created with
CREATE_SUSPENDED set is halted before DLLs are loaded, so there's no way of
knowing where anything will end up unless the load address is guaranteed to be
the same for every process (true for a few core DLLs including kernel32).
http://www.nynaeve.net/?p=198 has some more details."

Do you have any tips for him?

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #58 from Myk Taylor   2010-08-24 19:28:48 ---
(In reply to comment #56)
> In general it should work as long as prelink is found at compile time to set
> the base address, and as long as the OS doesn't force randomization despite
> prelink.

<snip>
> That's certainly still true, and it's one of the reasons we set a fixed base
> address for kernel32.

on my system, prelink is installed, and I have set
  kernel.randomize_va_space = 0
in /etc/sysctl.conf, yet it still seems /usr/lib32/wine/kernel32.dll.so is
loaded at a different address in the two processes.  While I pursue a better
solution, What else should I do to ensure that it works?

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

paulo  changed:

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

--- Comment #59 from paulo  2010-08-30 13:18:02 CDT ---
Morrowind script extender is also not working.

By not working i mean, it loads of fine, and thinks it is attached. However the
scripts that need MWSE behave very erroneously, and invariably crash/hang. 
The keyring mod duplicates the inventory keyring whenever you reload a
savegame.
The fleeing fletchers mod hangs the game when you cast a spell
The alchemy storage helper mod places duplicate bags whenever you are looking
at.
etc, basically no mod using MWSE works.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #60 from Myk Taylor  2010-09-07 20:54:17 CDT ---
was work done on this in 1.3.2?  it is working for me now on both my amd64
machine and my x86 machine.  Of course, it has randomly worked for me in the
past with certain combinations of wine version and gcc version...

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #61 from paulo  2010-09-11 02:34:10 CDT ---
I can confirm that MWSE also appears to be fixed in wine 1.3.2

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #62 from paulo  2010-09-19 02:05:30 CDT ---
Sorry for the spam, but the info i gave above is wrong, actually MWSE is not
working. The symptoms just changed (eventual freeze in keyring, other things in
other mods). I have a intel x86-64 if it is relevant.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

Macarthur  changed:

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

--- Comment #63 from Macarthur  2010-10-21 06:48:28 CDT ---
still there in the latest wine unstable. I believe that my other bug is the
same thing as this one so i guess it's going to have to be merged with it.
There's no need to upload the backtrace since i believe it's the same thing as
what everyone else has said with a few minor changes ie the first line.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #64 from Juan Lang  2010-10-21 11:19:13 CDT ---
*** Bug 24839 has been marked as a duplicate of this bug. ***

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

MassteR  changed:

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

--- Comment #65 from MassteR  2010-11-02 04:59:04 CDT ---
Looks like resolved in stable WINE (1.2.1). obse_loader.exe starts normally
without any workaround.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start ("privileged instruction")

http://bugs.winehq.org/show_bug.cgi?id=13915

--- Comment #66 from madx  2012-01-17 03:18:30 CST ---
OBSE works perfectly fine for me, I even finished the game with it
Tested on wine 1.3.34-37, OBSE v0020, Oblivion 1.2.0416

[Bug 13915] Oblivon Script Extender (OBSE) fails to start (needs address space layout randomization (ASLR) disabled and prelink enabled)

http://bugs.winehq.org/show_bug.cgi?id=13915

Anastasius Focht  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |FIXED
            Summary|Oblivon Script Extender     |Oblivon Script Extender
                   |(OBSE) fails to start       |(OBSE) fails to start
                   |("privileged instruction")  |(needs address space layout
                   |                            |randomization (ASLR)
                   |                            |disabled and prelink
                   |                            |enabled)

--- Comment #67 from Anastasius Focht  2012-01-17 09:33:30 CST ---
Hello,

reported working twice.

With address space layout randomization (ASLR) disabled and prelink enabled it
should work in general.
There might be rare cases where non-Wine shared libraries might already occupy
virtual address space locations that ought the be reserved for Wine shared
libs, forcing them to relocate.
There is nothing more Wine can do.

Regards

[Bug 13915] Oblivon Script Extender (OBSE) fails to start (needs address space layout randomization (ASLR) disabled and prelink enabled)

http://bugs.winehq.org/show_bug.cgi?id=13915

Alexandre Julliard  changed:

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

--- Comment #68 from Alexandre Julliard  2012-01-27 14:18:09 CST ---
Closing bugs fixed in 1.4-rc1.

[Bug 13915] Oblivon Script Extender (OBSE) fails to start (needs address space layout randomization (ASLR) disabled and prelink enabled)

http://bugs.winehq.org/show_bug.cgi?id=13915

Anastasius Focht  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |obfuscation