[Bug 11584] New: Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

Summary: Recent patch breaks Call of Duty 4
           Product: Wine
           Version: CVS/GIT
          Platform: Other
        OS/Version: other
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: directx-d3d
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]
                CC: [e-mail]

Recent patch inbetween .55 and now breaks playing call of duty 4 demo.  Git
bisect returned the following :

(Gentoo x86 32 bit, Geforce 7900GS with 100.14.19 driver.)

ba90a740beb9ce9a839cc843db8d87f5a37becdd is first bad commit
commit ba90a740beb9ce9a839cc843db8d87f5a37becdd
Author: Roderick Colenbrander 
Date:   Sun Feb 10 22:20:15 2008 +0100

wined3d: Add read_from_framebuffer_texture which combines code from
read_from_framebuffer (drawpixels) and LoadLocation.

This makes the code easier to read and the pieces borrowed from
    read_from_framebuffer are more correct than the code in LoadLocation.

:040000 040000 74e4bdc73e367c8f38cd3d0818df0fc86eb788bf
3e54409be7c9d2964efbf3d3c2f3d3b84a267047 M      dlls

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

Zach Goldberg  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://files.filefront.com/C
                   |                            |all+of+Duty+4+Modern+Warfare
                   |                            |+Demo/;8774322;/fileinfo.htm
                   |                            |l
           Keywords|                            |download

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

James Hawkins  changed:

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

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #1 from Vitaliy Margolen   2008-02-13 23:39:49 ---
Duplicate of bug 11545?

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #2 from Zach Goldberg   2008-02-13 23:42:12 ---
I don't think so.  Game works just fine in .55.  This patch is one on the first
day of commits after 55 i think.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

Rico  changed:

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

--- Comment #3 from Rico   2008-02-14 15:04:07 ---
Confirming this bug.

This is NOT a dupe of bug 11545.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #4 from Austin English   2008-02-16 19:11:04 ---
How exactly does it 'break'? When I try to run it, I get an error saying:

Error during initialization:
Video card or driver doesn't support separate alpha blend, glow will be
disabled."

Then have this log in the client:
----- Initializing Renderer ----
----- Client Initialization Complete -----
Attempting 44 kHz 16 bit [Windows default] sound
----- R_Init -----
Getting Direct3D 9 interface...
Pixel shader version is 2.0
Vertex shader version is 2.0
Video card or driver doesn't accelerate dynamic textures.
Video card or driver doesn't support separate alpha blend, glow will be
disabled.

Error during initialization:
Video card or driver doesn't support separate alpha blend, glow will be
disabled.

And COD refuses to run. This is current git, haven't tried wine 0.9.55.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #5 from Rico   2008-02-17 04:00:59 ---
You need to apply this patch http://bugs.winehq.org/attachment.cgi?id=10811 .

This makes the game happy.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #6 from Gregor Münch   2008-02-17 08:30:37 ---
Which more or less proves that bug #10010 is still valid...

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #7 from Gregor Münch   2008-02-17 11:05:56 ---
Game runs just fine for me.
Retail v1.0, current git and 8600 GT (nvidia drivers 169.09)

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #8 from Roderick Colenbrander   2008-02-17 16:22:06 ---
In what way does it crash with my patch in? As the patch moves code around and
cleans some stuff up.

Note that the game can behave differently on different videocards and drivers.
For instance it crashes when there isn't enough video memory and the 3d
capabilities change between different cards (e.g. the game launches a different
backend when we report GeforceFX-like capabilities or Geforce8).

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #9 from Rico   2008-02-18 13:05:27 ---
Created an attachment (id=10835)
 --> (http://bugs.winehq.org/attachment.cgi?id=10835)
Picture where the game hangs

the console output (last lines):
fixme:d3d_shader:shader_glsl_color_correction Add color fixup for vertex
texture WINED3DSIO_TEXLDL
fixme:d3d:IWineD3DVolumeImpl_LockBox (0xeee04d0) : pBox=(nil) stub
fixme:d3d:IWineD3DVolumeImpl_LockBox (0xeee04d0) : pBox=(nil) stub
fixme:d3d:IWineD3DVolumeImpl_LockBox (0xd8dcb08) : pBox=(nil) stub
err:seh:setup_exception stack overflow 44 bytes in thread 001c eip 7e9f59f4 esp
7cedbfd4 stack 0x7cedc000-0x7cfdc000
err:ntdll:RtlpWaitForCriticalSection section 0x7e4b7980 "x11drv_main.c:
X11DRV_CritSection" wait timed out in thread 0014, blocked by 001c, retrying
(60 sec)

My specs:
Fedora7 x86_64, 8800GTS(640MB), proprtary driver 100.14.23

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #10 from Rico   2008-02-18 13:11:06 ---
Created an attachment (id=10836)
 --> (http://bugs.winehq.org/attachment.cgi?id=10836)
Full crash log

This is an addition to the previous post,  the full crash log.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #11 from Zach Goldberg   2008-02-18 13:22:11 ---
Just to confirm I'm witnessing identical behavior to Rico.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #12 from Roderick Colenbrander   2008-02-18 14:19:46 ---
Are you sure it is just my framebuffer patch which did this? You haven't
changed any registry settings? (e.g. video memory size) It sounds impossible
(really). Try to undo the patch.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #13 from Rico   2008-02-19 15:38:07 ---
Created an attachment (id=10852)
 --> (http://bugs.winehq.org/attachment.cgi?id=10852)
patch to revert the problematic things

If I apply this patch (which just reverts your patch) the game starts and runs
fine.
I used wine-0.9.55-276-g2472e81

I haven't changed any registry settings. I undo your patch with mine (I
couldn't undo yours directly).

Just applying my patch (reverting yours) -> make -> game work
Reverting my patch (applying yours) -> make -> game crash

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #14 from Roderick Colenbrander   2008-02-19 16:04:40 ---
As I can't reproduce the crash you should do some more testing on your own.
What I would recommend as a first step is to copy all the code which you now
moved below  'if (This->Flags & SFLAG_INDRAWABLE) {' to
read_from_framebuffer_texture and disable the code which is now inside that
function. Further these few lines should be at the start of
read_from_framebuffer_texture and be removed from the code before
'if(This->Flags & SFLAG_INDRAWABLE)
+        d3dfmt_get_conv(This, TRUE /* We need color keying */, TRUE /* We will
use textures */, &format, &internal, &type, &convert, &bpp, This->srgb);
+        ActivateContext(device, device->lastActiveRenderTarget,
CTXUSAGE_RESOURCELOAD);
+        surface_bind_and_dirtify(This);

Further the 'else' portion should again look like (the way I changed it in the
patch)
        } else { /* Upload from system memory */
            d3dfmt_get_conv(This, TRUE /* We need color keying */, TRUE /* We
will use textures */, &format, &internal, &type, &convert, &bpp, This->srgb);

ActivateContext(device, device->lastActiveRenderTarget,
CTXUSAGE_RESOURCELOAD);
            surface_bind_and_dirtify(This);
            ENTER_GL();

After these changes it should in theory work as before. If this is the case we
can look further.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

Rico  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #10852|0                           |1
        is obsolete|                            |

--- Comment #15 from Rico   2008-02-20 15:08:46 ---
Created an attachment (id=10867)
 --> (http://bugs.winehq.org/attachment.cgi?id=10867)
this patch copys the single steps in the function read_from_framebuffer_texture

I hope I've done all right and I've attached a patch which makes the changes (I
hope so) you suggested. With this patch the game runs without problems. And I
think we can look further ;-).

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #16 from Roderick Colenbrander   2008-02-20 15:14:06 ---
The main difference between the original code and my code is the read buffer
handling. Try to make it look like the new code and see if it still works..

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

Rico  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #10867|0                           |1
        is obsolete|                            |

--- Comment #17 from Rico   2008-02-20 15:15:10 ---
Created an attachment (id=10868)
 --> (http://bugs.winehq.org/attachment.cgi?id=10868)
the main problem located

Ok, I think I've located the problem. 
Could you have a look at the attached patch?

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #18 from Roderick Colenbrander   2008-02-20 16:09:55 ---
The patch is likely correct but I'm not an expert on the context switching
stuff (stefan wrote that piece). Likely (99% certain) it is correct. Submit the
patch and if it isn't correct, Stefan will tell.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

Stefan Dösinger  changed:

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

--- Comment #18 from Roderick Colenbrander   2008-02-20 16:09:55 ---
The patch is likely correct but I'm not an expert on the context switching
stuff (stefan wrote that piece). Likely (99% certain) it is correct. Submit the
patch and if it isn't correct, Stefan will tell.

--- Comment #19 from Stefan Dösinger   2008-02-22 16:20:21 ---
It isn't correct I think as explained on wine-devel. There are problems when
the app tries to read the back buffer or front buffer while an offscreen render
target is active

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #20 from Bobby Gill   2008-02-23 19:46:00 ---
(In reply to comment #5)
> You need to apply this patch http://bugs.winehq.org/attachment.cgi?id=10811 .
>
> This makes the game happy.

I am getting the exact same error as Austin, using wine-git. How do I apply
this patch you have linked??

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #21 from Austin English   2008-02-23 20:19:26 ---
(In reply to comment #20)
> (In reply to comment #5)
> > You need to apply this patch http://bugs.winehq.org/attachment.cgi?id=10811 .
> >
> > This makes the game happy.
>
> I am getting the exact same error as Austin, using wine-git. How do I apply
> this patch you have linked??

Download the patch.
$ cd ~/wine-git/
$ patch -p1 < /path/to/patch/here/
$ ./configure && make clean && make depend && make

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #22 from Rico   2008-02-24 07:23:47 ---
I have some new stuff:

The game runs without problems (without any patch from me) when the
OffscreenRenderingMode is set to FBO.

Just for me that I know where I am ;-):
If the OffscreenRenderingMode != FBO (it is pbuffer or backbuffer) the game
crash in IWineD3DSurface_PreLoad(target) in context.c

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

Rico  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #10868|0                           |1
        is obsolete|                            |

--- Comment #23 from Rico   2008-02-24 07:56:24 ---
Created an attachment (id=10936)
 --> (http://bugs.winehq.org/attachment.cgi?id=10936)
pointed out where the game hangs and breaks short before that

This is a patch which breaks shortly before the game crash. I post a back trace
soon.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #24 from Rico   2008-02-24 08:02:32 ---
Created an attachment (id=10937)
 --> (http://bugs.winehq.org/attachment.cgi?id=10937)
simple back trace

This is the back trace which I get after the debugger is started with the
previous patch. This looks like a function is calling it self recursivly
without returning.

Is there a chance to get the full back trace? It looks like that there is a
limit of 200 lines.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #25 from Stefan Dösinger   2008-02-24 08:39:04 ---
The more deeply rooted problem is that PreLoad itself can call ActivateContext
to find a GL context. It should be filtered out, but its not really a clear
design.

I think what needs to be done (with the current scheme) is to set
lastActiveRenderTarget before calling this PreLoad call, to make sure that
there isn't an infinite recursion.

However, the PreLoad is wrong at the current place. FindContext must not do
anything that leads to opengl calls, because it's FindContexts job to find a GL
context to use, and it can't use it itself because it is activated later on.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #26 from Rico   2008-03-19 15:30:20 ---
Created an attachment (id=11503)
 --> (http://bugs.winehq.org/attachment.cgi?id=11503)
sample patch
> However, the PreLoad is wrong at the current place.

Is this the right patch?

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

byteframe  changed:

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

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #27 from Tiemen Schreuder   2008-04-09 16:53:48 ---
Created an attachment (id=12024)
 --> (http://bugs.winehq.org/attachment.cgi?id=12024)
Possible wined3d CoD4 stack overflow ActivateContext fix

First of all, I'm new to this, so bare with me!

Confirming this bug in Call of Duty 4 [1.0] after level loaded (stack overflow,
last calls to PreLoad, read_from_framebuffer_texture, etc).

The attached patch fixes the problem for me, and makes sure the
lastActiveRenderTarget and lastThread is set before calling a function
(FindContext) that might start indirect recursion. Patch fixes that in
ActivateContext, which fixed it for me, but perhaps this happens in other
places as well. Reading through the posts this seems similar to what Stefan
suggested and a more permanent fix would probably indeed be to prevent this
recursion from happening at all.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #28 from Stefan Dösinger   2008-04-09 17:47:32 ---
We should make sure that reading back the texture from the backbuffer or
pbuffer doesn't call ActivateContext again at all. However, that's a dirty
business because it will have to call LoadLocation, and LoadLocation calls
ActivateContext on its own in some situations, and in some it doesn't, etc. The
whole ruleset when LoadLocation calls ActivateContext and vice versa needs a
review :-/

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #29 from Stefan Dösinger   2008-04-09 18:22:38 ---
Which PreLoad is making the problem here in FindContext()? The one in the
readTexture block preloading "This->lastActiveRenderTarget", or the one above
PreLoad "target"? If it is the former one, can you try to change the if
statement above that in the following way:

if (wined3d_settings.offscreen_rendering_mode != ORM_FBO) {

===>

if (wined3d_settings.offscreen_rendering_mode != ORM_FBO && target !=
This->lastActiveRenderTarget) {

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #30 from Tiemen Schreuder   2008-04-10 03:41:00 ---
Created an attachment (id=12039)
 --> (http://bugs.winehq.org/attachment.cgi?id=12039)
Selection of recursion stack overflow, CoD4

I've attached a stack dump by winedbg, the PreLoad causing the problem is
indeed the "target" one. This Preload calls: LoadTexture > LoadLocation >
read_from_framebuffer_texture > ActivateContext > FindContext > Preload, etc.

Shouldn't this call chain happen at least once on a new context, to read back
the texture, but stop once it tries to ActivateContext again?

I'm currently recompiling with the change you suggested (in combination with my
patch). I will also try without the patch and just your change and let you know
the results.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

Stefan Dösinger  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1

--- Comment #31 from Stefan Dösinger   2008-04-10 04:35:41 ---
As far as I can see the problem will occur when the app switches threads while
rendering to an offscreen surface. In that case it will try to alloc a texture,
which calls activatecontext with lastActiveRenderTarget. Thus it works fine on
render target switches, but on thread switches it fails because the new thread
number isn't assigned yet.

This PreLoad call should be removed at all. The other one is needed, but this
one here is at the wrong place. I am not sure why I added it there. Maybe it
was a workaround for a bug elsewhere then we should track it down. If it is
really needed it should probably be moved to SetRenderTarget rather than
ActivateContext, or the recursion has to be prevented.

The other PreLoad needs some safeguards as well. For reading back on a target
switch it works because the readback will activate the context for the
old(current) render target, and since the new render target is set after
FindContext it works as intended. If there is a thread switch we wouldn't have
to read back at all, so we should make sure readTexture is only set on a target
switch. However, there's still the issue if the thread and target are switched
at the same time. In this case we have to activate the old target for the new
thread, read back, then activate the new target for the new thread.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

Tiemen Schreuder  changed:

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

--- Comment #32 from Tiemen Schreuder   2008-04-10 05:25:35 ---
Tested the four cases and the combination of the patch and your suggested
change ('check') seems to work alright too. The patch is required to prevent
stack overflow tho, the check should increases performance slightly, although
not noticeable for me in game.

In the long run, making sure FindContext never leads to (re-)activation seems a
more elegant solution (but in effect similar to current patches). Unfortunately
I know very little about DirectX, but looking through the code the only reason
ActivateContext is called from read_from_framebuffer_texture is seemingly to
set the 'usage' to something else, not to actually activate it. How about (yes,
this is probably a bit rash) a seperate function for that? So:

ActivateContext(..,usage)
 -..stuff (find context, preload, read_from_framebuffer)..
 -SetContextUsage(.,usage) [instead of switch statement]

read_from_framebuffer_texture
 -..stuff..
 -SetContextUsage(.,BLIT) [instead of ActivateContext]
 -..stuff.

SetContextUsage(.,usage) [new function]
 <current code from ActivateContext usage switch statement>

I can test if it works if you want, however considering the size of the change
(not in time to apply, but in possible undesirable side-effects) I rather wait
till you've given it a quick glance in case there are any obvious errors.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

Tiemen Schreuder  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #12024|Possible wined3d CoD4 stack |[introduces errors in other
        description|overflow ActivateContext fix|d3d apps!] wined3d CoD4
                   |                            |stack overflow
                   |                            |ActivateContext fix
  Attachment #12024|0                           |1
        is obsolete|                            |

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #33 from Tiemen Schreuder   2008-04-10 18:33:20 ---
Created an attachment (id=12057)
 --> (http://bugs.winehq.org/attachment.cgi?id=12057)
fix for stackoverflow CoD4

I've tried your suggestion, simply removing the whole Preload target call and
that works for me! Currently no more problems in CoD4 (besides it being slow).

Simple patch included.

However, like you said, that code was probably there for a reason, so hopefully
it won't break anything else.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

Tiemen Schreuder  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #12057|fix for stackoverflow CoD4  |<<see rico's patch>>
        description|                            |

--- Comment #34 from Tiemen Schreuder   2008-04-10 18:42:44 ---
(From update of attachment 12057)
<<see rico's patch>>

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

Vitaliy Margolen  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]
             Blocks|                            |11630
           Severity|normal                      |major
         OS/Version|other                       |Linux
           Platform|Other                       |PC
            Version|CVS/GIT                     |0.9.56.

--- Comment #35 from Vitaliy Margolen   2008-04-22 00:28:59 ---
Setting to major - several programs affected in default configuration. However
this is not the only one place. As indicated in bug 11630 fixing one preload is
not enough.

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #36 from Vitaliy Margolen   2008-04-27 11:31:17 ---
*** Bug 11630 has been marked as a duplicate of this bug. ***

[Bug 11584] Recent patch breaks Call of Duty 4

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #37 from Vitaliy Margolen   2008-04-27 11:32:03 ---
*** Bug 12488 has been marked as a duplicate of this bug. ***

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Vitaliy Margolen  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|Recent patch breaks Call of |Multiple games crash with
                   |Duty 4                      |stack overflow error

--- Comment #38 from Vitaliy Margolen   2008-04-27 11:42:06 ---
Making subject more generic.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Wouter Cox  changed:

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

--- Comment #39 from Wouter Cox   2008-05-01 13:01:10 ---
This bug also seems present in Prince of Persia 2, using wine 0.9.60

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #40 from Wouter Cox   2008-05-01 13:04:57 ---
Created an attachment (id=12624)
 --> (http://bugs.winehq.org/attachment.cgi?id=12624)
Terminal Output showing where Prince of Persia 2 'hangs' (stack overflow)

I am not sure if this will be of any use, but I tested using OpenSUSE 10.3
32-bit and used the 1-cd version from the 'Prince of Persia Triple Pack';
another person tested it using OpenSUSE 10.3 64-bit version (not sure which
retail version he used) and says he has no problems running it.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #41 from Vitaliy Margolen   2008-05-01 13:22:19 ---
(In reply to comment #40)
The problem present with "backbuffer" and "pbuffer" "OffscreenRenderingMode"s.
It does not happen with "fbo".

Here is the snipped of the code in question:

if (wined3d_settings.offscreen_rendering_mode != ORM_FBO) {
   IWineD3DSurface_PreLoad(target);
}

As you see it's not being executed if you selected fbo off-screen rendering
mode.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Roderick Colenbrander  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |1.0.0

--- Comment #42 from Roderick Colenbrander   2008-05-01 16:02:41 ---
This is a very severe bug in the wined3d surface / texture. One of the patches
I submitted a while ago to fix other bugs exposed this one. It might require a
lot of patches to fix it. We will enter a code freeze soon but I think this
issue is quite severe and should be fixed as it affects a lot of programs.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #43 from Stefan Dösinger   2008-05-02 09:10:47 ---
Ya, with fbos we do not have to read back the framebuffer into the texture
since OpenGL does it for us. Thus the entire issue is bypassed.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #44 from Raul   2008-05-06 20:28:04 ---
Created an attachment (id=12786)
 --> (http://bugs.winehq.org/attachment.cgi?id=12786)
Fifa 2006 world cup logs

Fifa 2006 world cup also is affected with this bug

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Julian Rüger  changed:

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

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #45 from Jérôme Gardou   2008-05-10 10:11:49 ---
Created an attachment (id=12884)
 --> (http://bugs.winehq.org/attachment.cgi?id=12884)
+d3d9,+d3d debug log

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Jérôme Gardou  changed:

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

--- Comment #46 from Jérôme Gardou   2008-05-10 10:13:52 ---
I was trying to give a shot to a similar problem in Supreme Commander. Those
seem very related.

The problem seems to be that sometimes wined3d and/or d3d9 is using getparent
when the parent queried has already been released -> crash !

See the previous trace (I had to add a TRACE to IWineD3DResourceImpl_GetParent)
: parent 0xa83e400 is queried whereas it has been released just before.

This one was made when changing video settings of SupCom.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #47 from Stefan Dösinger   2008-05-10 18:43:02 ---
The object releasing is a different issue

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Jumper  changed:

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

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Roman Nykyforchyn  changed:

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

--- Comment #48 from Roman Nykyforchyn   2008-05-16 11:48:47 ---
I have faced and fixed similar problem with stack overflow with Oblivion game.

You may refer to comment http://bugs.winehq.org/show_bug.cgi?id=8868#c12
of bug 8868 for details.

Hope it will help to resolve this issue.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Ian Schwarz  changed:

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

--- Comment #49 from Ian Schwarz   2008-05-22 11:20:09 ---
Okay, so I have this error in Supreme Commander too, right. I'm using Wine
1.0-rc1 on a 64-bit openSuSE and I've encountered the following:
- Supreme Commander first refused to run because of missing DLL files.
- After adding them, the game started and was playable fine (I only started a
skirmish to test), but without sound. I didn't have to patch the game as the
copy protection seemed to work (with CD in drive, without CD no workie).
- In order to get sound, I followed the instructions on its AppDB page
(regsvr32 xactengine2_*.dll).
- After this, the game refuses to start even and gives the following exception:

backflip@eden:/media/philips/Supreme Commander/Supreme Commander/bin> wine
SupremeCommander.exe
fixme:ntoskrnl:KeInitializeTimerEx 0x1110b8 0
fixme:system:SystemParametersInfoW Unimplemented action: 59 (SPI_SETSTICKYKEYS)
fixme:system:SystemParametersInfoW Unimplemented action: 53 (SPI_SETTOGGLEKEYS)
fixme:system:SystemParametersInfoW Unimplemented action: 51 (SPI_SETFILTERKEYS)
wine: Call from 0x2689d71 to unimplemented function msvcrt.dll._aligned_malloc,
aborting
err:seh:setup_exception_record stack overflow 908 bytes in thread 0009 eip
f7d7c2cf esp 00240fa4 stack 0x240000-0x241000-0x340000

- Patching the game doesn't help.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #50 from Roman Nykyforchyn   2008-05-22 11:28:43 ---
Created an attachment (id=13249)
 --> (http://bugs.winehq.org/attachment.cgi?id=13249)
Fix for stack owerflow issue

Try to apply this patch and test on games.

It for sure fixes Oblivion, The Punisher and probably all other games that have
this stack overflow issue.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Austin English  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #51 from Stefan Dösinger   2008-05-22 13:10:51 ---
Roman Nykyforchyn's patch is wrong. It can break texture readback in some
situations.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #52 from Vitaliy Margolen   2008-05-22 13:26:03 ---
(In reply to comment #49)
> wine: Call from 0x2689d71 to unimplemented function msvcrt.dll._aligned_malloc,
> aborting

It has nothing to do with this bug. You have broken / old Wine. That function
is implemented in Wine's msvcrt.dll

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #53 from Ian Schwarz   2008-05-22 13:55:10 ---
Created an attachment (id=13252)
 --> (http://bugs.winehq.org/attachment.cgi?id=13252)
console output

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #54 from Ian Schwarz   2008-05-22 14:03:02 ---
Supreme Commander keeps crashing on startup even if I remove my ~/.wine folder
(then without complaining about msvcrt though). I'm using Wine 1.0-rc1.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #55 from Vitaliy Margolen   2008-05-22 15:55:59 ---
(In reply to comment #53)
> Created an attachment (id=13252)

--> (http://bugs.winehq.org/attachment.cgi?id=13252) [details]
> console output

err:x11settings:X11DRV_ChangeDisplaySettingsEx No matching mode found
1024x768x32 @60! (XRandR)
Again wrong stuff - has nothing to do with this bug. Also see Comment  #46. 
Stop posting here! There is already enough information here. Unless you want to
propose a patch.

If you need help with using Wine - refer to forum.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Vitaliy Margolen  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #13252|0                           |1
        is obsolete|                            |

--- Comment #56 from Vitaliy Margolen   2008-05-22 15:56:19 ---
(From update of attachment 13252)
Not related to bug

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Vitaliy Margolen  changed:

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

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Vitaliy Margolen  changed:

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

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Julian Rüger  changed:

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

--- Comment #57 from Julian Rüger   2008-05-23 07:44:23 ---
@Vitaliy: Hey, why did you remove my CC? Anything wrong?

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #58 from Vitaliy Margolen   2008-05-23 08:16:42 ---
(In reply to comment #57)
> @Vitaliy: Hey, why did you remove my CC? Anything wrong?

Oops sorry - the wrong e-mail.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Ben P.  changed:

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

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Vitaliy Margolen  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Blocks|                            |9269

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Vitaliy Margolen  changed:

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

--- Comment #59 from Vitaliy Margolen   2008-06-01 23:48:40 ---
Actually first Preload comes from this commit:
bd5c02e24504470e0853096a2ef50120b4050b50

Too bad the patch has no explanation about why it was necessary.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #60 from Roderick Colenbrander   2008-06-02 01:08:24 ---
I don't have a proper git, so I had to use git web. From what I saw it is my
D24S8 patch and that one is really needed to fix z-buffer issues.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #61 from Stefan Dösinger   2008-06-02 02:00:58 ---
I don't see a preLoad added in this patch. Wrong SHA1?

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #62 from Vitaliy Margolen   2008-06-02 07:22:56 ---
Oops have to check what I have selected. Here is the commit:

commit 2d0016c5bc872afb562727278cfd341cea182600
Author: Felix Nawothnig 
Date:   Sun Apr 8 02:46:40 2007 +0200

wined3d: Preload target in ActivateContext() for
ORM_BACKBUFFER/ORM_PBUFFER.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Dan Kegel  changed:

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

--- Comment #63 from Dan Kegel   2008-06-06 09:28:17 ---
Roderick/Stefan, ok to defer?

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #64 from Austin English   2008-06-06 11:03:00 ---
(In reply to comment #63)
> Roderick/Stefan, ok to defer?

Considering it's a major regression affecting several games/apps, I don't think
we should defer it just yet.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #65 from Dan Kegel   2008-06-06 11:47:56 ---
Got an ETA on a fix?  Time is short...

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #66 from Dan Kegel   2008-06-06 11:48:15 ---
Got an ETA on a fix?  Time is short...

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #67 from Dan Kegel   2008-06-06 12:16:16 ---
And on the topic of reverting it: Any progress since
Alexandre reported a problem?
http://winehq.org/pipermail/wine-devel/2008-June/066175.html

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #68 from Stefan Dösinger   2008-06-06 13:43:26 ---
My opinion is to defer it. This is a bug hiding at least one, possibly more
other bugs. The reverting patch Vitaliy sent was correct.

It may be worth a try adding a newrendertarget != oldrendertarget checks to
both PreLoad calls; However, that would also be fixing the symptoms, not the
bug, for a part at least.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #69 from Vitaliy Margolen   2008-06-06 23:16:57 ---
(In reply to comment #68)
> It may be worth a try adding a newrendertarget != oldrendertarget checks to

You mean if (target != This->lastActiveRenderTarget) ? That won't work since
lastActiveRenderTarget is changed in ActivateContext - after the find context.
> both PreLoad calls; However, that would also be fixing the symptoms, not the
> bug, for a part at least.

Even fixing a symptom still worth it - this can fix lots of games. We can
remove it the first version after 1.0 is out.

And it really needs fixing. One more bug 13079 caused by another Preload. Seems
like there is something fundamentally wrong about preload.

IMHO this bug is a blocker for 1.0. There is no point knowingly releasing
something that does not work.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Luke Bratch  changed:

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

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #70 from Stefan Dösinger   2008-06-07 06:00:03 ---
> You mean if (target != This->lastActiveRenderTarget) ? That won't work since
> lastActiveRenderTarget is changed in ActivateContext - after the find context.

It will avoid the PreLoad call if the game switches threads during offscreen
rendering, which I think is the problem here. Can you give it a try to see how
it works out?

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #71 from Vitaliy Margolen   2008-06-07 09:45:20 ---
Created an attachment (id=13782)
 --> (http://bugs.winehq.org/attachment.cgi?id=13782)
Possible patch

(In reply to comment #70)
> > You mean if (target != This->lastActiveRenderTarget) ? That won't work since
> > lastActiveRenderTarget is changed in ActivateContext - after the find context.
>
> It will avoid the PreLoad call if the game switches threads during offscreen
> rendering, which I think is the problem here. Can you give it a try to see how
> it works out?

I did already try it - doesn't work. There is nothing to break the loop. Even
rearranging things like this don't work.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #72 from Stefan Dösinger   2008-06-08 08:24:44 ---
I looked a bit at the code, and tried to find a reason for the failure when
Vitaliy tried to revert the patch. The situation the reverting causes is that
IWineD3DTexture::PreLoad is called without an existing texture.

PreLoad calls BindTexture, which finds that no texture exists. It creates a gl
texture and calls SetGlTextureDesc on the surface.

SetGlTextureDesc calls ModifyLocation(INTEXTURE, FALSE), which is correct. This
makes sure the texture location is reloaded when something else asks to load
it.

SetGlTextureDesc calls AddDirtyRect, supposdately to set the dirty rectangle.
AddDirtyRect calls ModifyLocation(INSYSMEM, TRUE). This will make PreLoad load
the surface from the sysmem copy, rather than the drawable. Not good. I think
the AddDirtyRect call in this case is wrong, and a leftover from older times.

SetGlTextureDesc updates the texture addressing flags(GL_TEXTURE_RECTANGLE_ARB
vs normal textures), removes the gl-texture-is-allocated flag. All OK here.

Back to PreLoad: The palettized change returns FALSE - palettized surfaces
aren't renderable. Let's assume no sRGB change for now, to keep the matter
simple. I *think* that an sRGB change works out correctly as well, once this
bug is fixed. I am not even sure if I can construct a situation where an sRGB
flag change occurs the first time the texture is loaded...

Surface::LoadTexture is called. It does color keying checks and then calls
LoadLocation(INTEXTURE).

LoadLocation should run into the read_from_framebuffer call because the most up
to date location is the drawable, and we want to read the contents from there
into the texture. However, due to the above AddDirtyRect call it doesn't. But
let's assume it would.

read_from_framebuffer calls ActivateContext(dest surface). This is OK - the
surface is already active, so ActivateContext returns immediately. It doesn't
though, if there is a thread switch at the same time, that is the problem with
the other PreLoad call.

After that, it checks if a glTexImage2D call is needed to allocate the mipmap
sublevel - yes, because setGLtextureDesc unset the flag as it is supposed to.

After that the content is read back.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #73 from Dan Kegel   2008-06-08 22:13:21 ---
Fun factoid: with current wine source, I see a Valgrind error
in the tests involving IWineD3DTextureImpl_PreLoad:
http://kegel.com/wine/valgrind/logs-2008-06-08/vg-d3d8_visual.txt
In d3d9, I don't see that same error, but I do see a stack overflow:
http://kegel.com/wine/valgrind/logs-2008-06-08/vg-d3d9_visual.txt

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #74 from Dan Kegel   2008-06-08 22:16:35 ---
And I use a lot of suppressions for libGL bugs; without them, one
sees way more valgrind warnings, e.g.
http://kegel.com/wine/valgrind/logs-2008-06-04/vg-d3d9_visual.txt

Valgrind seems to expose a whole mess o' problems in Nvidia's drivers.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #75 from Vitaliy Margolen   2008-06-08 23:16:28 ---
(In reply to comment #74)
> SetGlTextureDesc calls AddDirtyRect, supposdately to set the dirty rectangle.
> AddDirtyRect calls ModifyLocation(INSYSMEM, TRUE). This will make PreLoad load
> the surface from the sysmem copy, rather than the drawable. Not good. I think
> the AddDirtyRect call in this case is wrong, and a leftover from older times.

Removing call to AddDirtyRect() doesn't make any difference. Maybe because it's
already been called before in PreLoad()?

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #76 from Stefan Dösinger   2008-06-09 05:23:38 ---
Dan: I don't think Valgrind exposes bugs in the Nvidia driver. Since the driver
essentially communicates with a remote machine(the GPU), it may be a similar
situation like the one with the uninitialized values in send() in wineserver.

Vitaliy: Just to clarify, did you remove both the PreLoad call in LoadLocation
as well as the AddDirtyRect? Do you see any other AddDirtyRect I missed?

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Richie  changed:

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

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Dan Kegel  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|1.0.0                       |1.0.1

--- Comment #77 from Dan Kegel   2008-06-11 16:29:30 ---
1.0 is breathing down our necks.  Deferring to 1.0.1 (and possibly later).

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #78 from Austin English   2008-06-11 16:47:51 ---
(In reply to comment #77)
> 1.0 is breathing down our necks.  Deferring to 1.0.1 (and possibly later).

While it may not be possible, we really should try to get this fixed. Breaking
a ton of apps in a 1.0 release that used to work is IMHO a really bad idea.

*ducks and hides*

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #79 from Austin English   2008-06-12 16:11:16 ---
Stefan posted 3 patches today that he thinks will fix this. Please test:

http://www.winehq.org/pipermail/wine-patches/2008-June/055783.html
http://www.winehq.org/pipermail/wine-patches/2008-June/055785.html
http://www.winehq.org/pipermail/wine-patches/2008-June/055787.html

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #80 from Ben P.   2008-06-12 18:33:25 ---
(In reply to comment #79)
> Stefan posted 3 patches today that he thinks will fix this. Please test:
>
> http://www.winehq.org/pipermail/wine-patches/2008-June/055783.html
> http://www.winehq.org/pipermail/wine-patches/2008-June/055785.html
> http://www.winehq.org/pipermail/wine-patches/2008-June/055787.html

Patches seem to make Call of Duty 4 work without issues.

Thanks
BennyP

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #81 from Vitaliy Margolen   2008-06-12 20:08:14 ---
Of course the second patch "WineD3D: Do not PreLoad the new render target"
fixes the problem with Psychonauts. Other 2 have no noticeable affect.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #82 from Stefan Dösinger   2008-06-13 04:56:53 ---
The first one prevents the 2nd from breaking the test, and the 3rd one fixes
the endless recursion in the other PreLoad.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Alexandre Julliard  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|1.0.1                       |1.0.0

--- Comment #83 from Alexandre Julliard   2008-06-13 05:55:33 ---
The patches are committed.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Alexandre Julliard  changed:

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

--- Comment #84 from Alexandre Julliard   2008-06-13 10:45:17 ---
Closing bugs fixed in 1.0-rc5.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #85 from Austin English   2008-06-13 11:05:50 ---
(In reply to comment #82)
> The first one prevents the 2nd from breaking the test, and the 3rd one fixes
> the endless recursion in the other PreLoad.

Great work Stefan! Thanks for getting it into RC5.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Raul  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #12786|0                           |1
        is obsolete|                            |

--- Comment #86 from Raul   2008-06-15 13:05:40 ---
Created an attachment (id=14071)
 --> (http://bugs.winehq.org/attachment.cgi?id=14071)
Fifa 2006 wc

i'm still having the same problem with Fifa 06 world cup using wine-1.0-rc5

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Raul  changed:

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

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #87 from Stefan Dösinger   2008-06-15 13:52:07 ---
Can you attach a +d3d,+d3d_surface trace to check if it is really the same
problem? There can be other sources for stack overflow exceptions

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #88 from Roderick Colenbrander   2008-06-15 13:55:01 ---
+d3d,+d3d_surface,+seh is a little more useful ;)

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #89 from Raul   2008-06-15 16:23:52 ---
the log file size with +d3d,+d3d_surface,+seh is 79mb, should i upload all? O.o

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #90 from Dan Kegel   2008-06-15 17:14:41 ---
compress it with rzip first

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

Raul  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #14071|0                           |1
        is obsolete|                            |

--- Comment #91 from Raul   2008-06-15 18:02:58 ---
Created an attachment (id=14076)
 --> (http://bugs.winehq.org/attachment.cgi?id=14076)
Fifa 2006 world cup +d3d,+d3d_surface,+seh logs

Fifa 2006 world cup '+d3d,+d3d_surface,+seh' log

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #92 from Stefan Dösinger   2008-06-15 19:27:47 ---
This is a different crash. The last ActivateContext is far away from the
exception. It seems that the stack overflow itself occurs in the exception
handler, so it is a symptom of the crash, not than the cause.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

emkay  changed:

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

--- Comment #93 from emkay   2008-06-17 04:21:58 ---
i still get this error output with RC5:

wine SupremeCommander.exe /novalidate
ALSA lib pcm_dmix.c:935:(snd_pcm_dmix_open) The dmix plugin supports only
playback stream
fixme:mixer:ALSA_MixerInit No master control found on MPU-401 UART, disabling
mixer
fixme:system:SystemParametersInfoW Unimplemented action: 59 (SPI_SETSTICKYKEYS)
fixme:system:SystemParametersInfoW Unimplemented action: 53 (SPI_SETTOGGLEKEYS)
fixme:system:SystemParametersInfoW Unimplemented action: 51 (SPI_SETFILTERKEYS)
fixme:faultrep:ReportFault 0xe2f224 0x0 stub
fixme:dbghelp:dump_system_info fill in CPU vendorID and feature set
fixme:faultrep:ReportFault 0x7d77cc64 0x0 stub
fixme:faultrep:ReportFault 0x7d77be20 0x0 stub
fixme:faultrep:ReportFault 0x7d77afdc 0x0 stub
err:seh:setup_exception_record stack overflow 848 bytes in thread 001e eip
7ef95d93 esp 7d56efe0 stack 0x7d56e000-0x7d56f000-0x7d77f000

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #94 from emkay   2008-06-18 04:17:46 ---
very strange got the whole game up and running but only with one game version
all the other ones won't start i got v.3217 with a no-cd-patch.

[Bug 11584] Multiple games crash with stack overflow error

http://bugs.winehq.org/show_bug.cgi?id=11584

--- Comment #94 from emkay  2008-06-18 04:17:46 CDT ---
very strange got the whole game up and running but only with one game version
all the other ones won't start i got v.3217 with a no-cd-patch.