Heroes of Might and Magic 5 crash

gedgeck · Post by **gedgeck** » Sat May 31, 2008 8:12 am

Hi, all.
Since wine 0.9.59 (before I used 0.9.50 and all was O.K., but from 0.9.59 till 1.0-rc2 appear described problem) I am find strange behaviour of heroes 5 at wine. After starting application it begin intensively put on console output string like that:

fixme:win:SetLayeredWindowAttributes (0x30024,0x00000000,0,2): stub!

but that is not so critical as I know. Realy worst things begins when application stop show a heroes 5 logo and actualy must start a game... Here it crashed with appearing "Failed to initialize Direct3D9" window.
At console output last string is:

fixme:win:EnumDisplayDevicesW ((null),0,0x32f808,0x00000000), stub!

I have tried install DirectX (Nov 2007 and Mar 2008), same problem. At other games, for example Warcraft 3:TFT all is O.K.

I have tried completely remove wine and $HOME/.wine and install some various versions - it works only at wine-0.9.50 (I didn't download 0.9.51-0.9.58). Also I tried specify VideoRam at my xorg.conf by hand and add 'VideoMemorySize' string to my regestry at wine. Any changes.

Any suggestion how to beat this sad error?

Added: rc3 affected too.

austin987 · Post by **austin987** » Sun Jun 01, 2008 6:39 pm

On Sat, May 31, 2008 at 8:12 AM, gedgeck <[email protected]> wrote:

Hi, all.
Since wine 0.9.59 (before I used 0.9.50 and all was O.K., but from 0.9.59 till 1.0-rc2 appear described problem) I am find strange behaviour of heroes 5 at wine. After starting application it begin intensively put on console output string like that:

fixme:win:SetLayeredWindowAttributes (0x30024,0x00000000,0,2): stub!

but that is not so critical as I know. Realy worst things begins when application stop show a heroes 5 logo and actualy must start a game... Here it crashed with appearing "Failed to initialize Direct3D9" window.
At console output last string is:

fixme:win:EnumDisplayDevicesW ((null),0,0x32f808,0x00000000), stub!

I have tried install DirectX (Nov 2007 and Mar 2008), same problem. At other games, for example Warcraft 3:TFT all is O.K.

I have tried completely remove wine and $HOME/.wine and install some various versions - it works only at wine-0.9.50 (I didn't download 0.9.51-0.9.58). Also I tried specify VideoRam at my xorg.conf by hand and add 'VideoMemorySize' string to my regestry at wine. Any changes.

Any suggestion how to beat this sad error?

Please run a regression test:
http://wiki.winehq.org/RegressionTesting

decoo · Post by **decoo** » Wed Jun 04, 2008 1:29 pm

All I saw was the splash screen...
Thats backtrace from console:
Backtrace:
=>1 0x7c0db5d0 (0x0032eff0)
2 0x7e40c76c ActivateContext+0x4fc() in wined3d (0x0032f070)
3 0x7e44129b drawPrimitive+0x17b() in wined3d (0x0032f380)
4 0x7e418feb in wined3d (+0x28feb) (0x0032f3f0)
5 0x7e4f2c5e in d3d9 (+0x12c5e) (0x0032f420)
6 0x007de72b in h5_game123 (+0x3de72b) (0x000027d9)
7 0x00000000 (0x00000000)

(wine installed from *.deb pack..)
And second try:
- compiled wine from source (for 32-bit using 32 bit libra...)
and now i get more information, but still I just don't know what its mean...:
Backtrace:
=>1 0x7c0da120 (0x018eebc0)
2 0x7e6df7fc ActivateContext+0x4fc(This=0x14f740, target=0x2cde6e8, usage=0x2) [/home/decmik/Programy/wine-git/dlls/wined3d/context.c:1091] in wined3d (0x018eec40)
3 0x7e71434b drawPrimitive+0x17b(iface=0x14f740, PrimitiveType=0x4, NumPrimitives=0x2, StartVertexIndex=0x0, numberOfVertices=0x200, StartIdx=0x0, idxSize=0x4, idxData=0x2da00c0, minIndex=0x0) [/home/decmik/Programy/wine-git/dlls/wined3d/drawprim.c:1046] in wined3d (0x018eef50)
4 0x7e6ec07b IWineD3DDeviceImpl_DrawIndexedPrimitive+0xeb(iface=0x14f740, PrimitiveType=0x4, minIndex=0x0, NumVertices=0x200, startIndex=0x0, primCount=0x2) [/home/decmik/Programy/wine-git/dlls/wined3d/device.c:5239] in wined3d (0x018eefc0)
5 0x7e7c6c6e IDirect3DDevice9Impl_DrawIndexedPrimitive+0xae(iface=, PrimitiveType=0x4, BaseVertexIndex=0x0, MinVertexIndex=0x0, NumVertices=0x200, startIndex=0x0, primCount=0x2) [/home/decmik/Programy/wine-git/dlls/d3d9/device.c:1204] in d3d9 (0x018eeff0)
sock_set_error: Success
fixme:winmm:MMDRV_Exit Closing while ll-driver open

I have no idea whats its mean...
Anyone?

austin987 · Post by **austin987** » Wed Jun 04, 2008 1:36 pm

On Wed, Jun 4, 2008 at 1:29 PM, decoo <[email protected]> wrote:

I have no idea whats its mean...
Anyone?

Please run the regression test.

decoo · Post by **decoo** » Wed Jun 04, 2008 11:06 pm

OK, can you tell me what number is "good" wine, and what is "bad", it doesn't work properly in wine 0.9.56, 0.9.50, and rc3,
and - im not using compiz/beryl- is this can be problem?

jeffz · Post by **jeffz** » Wed Jun 04, 2008 11:20 pm

decoo wrote:OK, can you tell me what number is "good" wine, and what is "bad", it doesn't work properly in wine 0.9.56, 0.9.50, and rc3,
and - im not using compiz/beryl- is this can be problem?

the good value is the version which it worked in last.
bad is the first version it stopped working in.

http://wiki.winehq.org/FAQ#head-f1ef104 ... 5c9271e671

decoo · Post by **decoo** » Thu Jun 05, 2008 10:34 am

And that's the problem, it never works good...

Daily_Lama · Post by **Daily_Lama** » Fri Jun 06, 2008 1:04 pm

Just tried it with my selfcompiled wine.10-rc3 and rc4 - both seem to work fine (seem because i only waited until it reached the main menu).

I'd try googling for the last line in your backtrace: "fixme:winmm:MMDRV_Exit Closing while ll-driver open", although I don't know what "driver" that is and if you could do anything to "close" it, but that seems to be the problem if you ask me.

gedgeck · Post by **gedgeck** » Sat Jun 07, 2008 3:17 pm

Please run a regression test:
http://wiki.winehq.org/RegressionTesting

Well, I am done regression test. I used wine-0.9.50 as "good" and wine-0.9.59 as "bad". All of ten tests show that bug is at first half of patches and finally I get message that says:

Code: Select all

Bisecting: 0 revisions left to test after this

and any strings like "a460a2df43aa8eae10b1db028c6020829b009c54 is first bad commit". From RegresionTesting How to I'm not understand what I need to do next and what means that result. So please tell me my next step.

May be it is significant. When I use stable wine version (i.e. wine-0.9.59) I don't get message with stack status, only window popup, that inform me about problem with identifiing Direct3D 9. When I use git wine always get this stack.

austin987 · Post by **austin987** » Sat Jun 07, 2008 3:36 pm

On Sat, Jun 7, 2008 at 3:17 PM, gedgeck <[email protected]> wrote:

Please run a regression test:
http://wiki.winehq.org/RegressionTesting
Well, I am done regression test. I used wine-0.9.50 as "good" and wine-0.9.59 as "bad". All of ten tests show that bug is at first half of patches and finally I get message that:

Bisecting: 0 revisions left to test after this

and any strings like "a460a2df43aa8eae10b1db028c6020829b009c54 is first bad commit". From RegresionTesting How to I'm not understand what I need to do next and what means that result. So please tell me my next step.

It should tell you the first bad commit. Once it does, post that
information here.

gedgeck · Post by **gedgeck** » Sat Jun 07, 2008 3:49 pm

austin987 wrote:On Sat, Jun 7, 2008 at 3:17 PM, gedgeck <[email protected]> wrote:

Please run a regression test:
http://wiki.winehq.org/RegressionTesting
Well, I am done regression test. I used wine-0.9.50 as "good" and wine-0.9.59 as "bad". All of ten tests show that bug is at first half of patches and finally I get message that:

Bisecting: 0 revisions left to test after this

and any strings like "a460a2df43aa8eae10b1db028c6020829b009c54 is first bad commit". From RegresionTesting How to I'm not understand what I need to do next and what means that result. So please tell me my next step.

It should tell you the first bad commit. Once it does, post that
information here.

Sorry, after message about none revisions left to test... I type again 'git bisect bad' and got next:
Code:

Code: Select all

29d66b48ef1f3ed3519fb020260086c8d3aba1a6 is first bad commit
commit 29d66b48ef1f3ed3519fb020260086c8d3aba1a6
Author: Jacek Caban <[email protected]>
Date:   Fri Nov 30 13:32:56 2007 +0100

    mshtml: Added IHTMLElement2::get_dir implementation.

:040000 040000 08f95dcc0c71f8ec9e80970d166ead59f19e8977 801bf15bc9323ff88d3c4942896b1855a88a2430 M    dlls

Is it positive result?

austin987 · Post by **austin987** » Sat Jun 07, 2008 4:40 pm

On Sat, Jun 7, 2008 at 3:49 PM, gedgeck <[email protected]> wrote:

austin987 wrote:
On Sat, Jun 7, 2008 at 3:17 PM, gedgeck <[email protected]> wrote:

Well, I am done regression test. I used wine-0.9.50 as "good" and wine-0.9.59 as "bad". All of ten tests show that bug is at first half of patches and finally I get message that:

Bisecting: 0 revisions left to test after this

and any strings like "a460a2df43aa8eae10b1db028c6020829b009c54 is first bad commit". From RegresionTesting How to I'm not understand what I need to do next and what means that result. So please tell me my next step.

It should tell you the first bad commit. Once it does, post that
information here.
Sorry, after message about none revisions left to test... I type again 'git bisect bad' and got next:
Code:

29d66b48ef1f3ed3519fb020260086c8d3aba1a6 is first bad commit
commit 29d66b48ef1f3ed3519fb020260086c8d3aba1a6
Author: Jacek Caban <[email protected]>
Date: Fri Nov 30 13:32:56 2007 +0100

mshtml: Added IHTMLElement2::get_dir implementation.

:040000 040000 08f95dcc0c71f8ec9e80970d166ead59f19e8977 801bf15bc9323ff88d3c4942896b1855a88a2430 M dlls

Is it positive result?

Doesn't sound like that patch should affect it. Try running the test again.

gedgeck · Post by **gedgeck** » Sun Jun 08, 2008 1:47 pm

Hi to all.
I do test again. Now I download almost all versions of wine and find out that bug appear between 0.9.51 and 0.9.52. So I reset wine-git and set "bad" to 0.9.52 and "good" to .51. Result is different one but also last (it means that no commits stay to test). Here it is:

Code: Select all

[ged@ged ~/wine_git/wine-git]$ git bisect bad
964a0303c129d59ea227fcce68f647d2e76ddc56 is first bad commit
commit 964a0303c129d59ea227fcce68f647d2e76ddc56
Author: Marcus Meissner <[email protected]>
Date:   Fri Dec 14 15:20:04 2007 +0100

    msi: Fixed buffer overflow in number parsing.

:040000 040000 d6d38f65c20a2d7bbf4dd6c819f5f39ed839d8f2 c6db2b0caaf3cb98f1a83c30f32e1e1c81b075a6 M    dlls

If it is wrong too, than I'm wait following instructions and want do suggestion: may be this bug can't located by git and depend on my local machine\OS\programs features or so...
In any case, thanks for attention and patience (and of course for your work).

austin987 · Post by **austin987** » Sun Jun 08, 2008 3:57 pm

On Sun, Jun 8, 2008 at 1:47 PM, gedgeck <[email protected]> wrote:

Hi to all.
I do test again. Now I download almost all versions of wine and find out that bug appear between 0.9.51 and 0.9.52. So I reset wine-git and set "bad" to 0.9.52 and "good" to .51. Result is different one but also last (it means that no commits stay to test). Here it is:

Code:

[ged@ged ~/wine_git/wine-git]$ git bisect bad
964a0303c129d59ea227fcce68f647d2e76ddc56 is first bad commit
commit 964a0303c129d59ea227fcce68f647d2e76ddc56
Author: Marcus Meissner <[email protected]>
Date: Fri Dec 14 15:20:04 2007 +0100

msi: Fixed buffer overflow in number parsing.

:040000 040000 d6d38f65c20a2d7bbf4dd6c819f5f39ed839d8f2 c6db2b0caaf3cb98f1a83c30f32e1e1c81b075a6 M dlls

If it is wrong too, than I'm wait following instructions and want do suggestion: may be this bug can't located by git and depend on my local machine\OS\programs features or so...
In any case, thanks for attention and patience (and of course for your work).

Try compiling wine 0.9.51 and see if it works, then do the same for
0.9.52. It may be a problem in the precompiled packages you're using.

gedgeck · Post by **gedgeck** » Mon Jun 09, 2008 3:45 am

austin987 wrote:On Sun, Jun 8, 2008 at 1:47 PM, gedgeck <[email protected]> wrote:

Try compiling wine 0.9.51 and see if it works, then do the same for
0.9.52. It may be a problem in the precompiled packages you're using.

Thing is that I am always compile wine myself and next do packages for slackware, but now I also try do it "from scratch": only do './configure --prefix=/usr && make && make install'. At 0.9.52 I get such error message at console:

Code: Select all

ERROR: Failed to open file with type descriptions (types.xml)
wine: Unhandled page fault on read access to 0x0000004c at address 0x515602 (thread 0009), starting debugger...
Unhandled exception: page fault on read access to 0x0000004c in 32-bit code (0x00515602).

And dump.

austin987 · Post by **austin987** » Mon Jun 09, 2008 4:00 am

On Mon, Jun 9, 2008 at 3:45 AM, gedgeck <[email protected]> wrote:

austin987 wrote:
On Sun, Jun 8, 2008 at 1:47 PM, gedgeck <[email protected]> wrote:

Try compiling wine 0.9.51 and see if it works, then do the same for
0.9.52. It may be a problem in the precompiled packages you're using.
Thing is that I am always compile wine myself and next do packages for slackware, but now I also try do it "from scratch": only do './configure --prefix=/usr && make && make install'. At 0.9.52 I get such error message at console:

Code:
ERROR: Failed to open file with type descriptions (types.xml)
wine: Unhandled page fault on read access to 0x0000004c at address 0x515602 (thread 0009), starting debugger...
Unhandled exception: page fault on read access to 0x0000004c in 32-bit code (0x00515602).

And dump.

But when you compile 0.9.51 that way, does it work?

gedgeck · Post by **gedgeck** » Mon Jun 09, 2008 4:54 am

austin987 wrote:On Mon, Jun 9, 2008 at 3:45 AM, gedgeck <[email protected]> wrote:
But when you compile 0.9.51 that way, does it work?

It is strange to me but not. It doesn't work :\

austin987 · Post by **austin987** » Mon Jun 09, 2008 5:24 am

On Mon, Jun 9, 2008 at 4:54 AM, gedgeck <[email protected]> wrote:

austin987 wrote:On Mon, Jun 9, 2008 at 3:45 AM, gedgeck <[email protected]> wrote:

But when you compile 0.9.51 that way, does it work?
It is strange to me but not. It doesn't work :\

That's why your regression testing isn't working. Something's fishy
with your compile.

gedgeck · Post by **gedgeck** » Mon Jun 09, 2008 6:48 am

austin987 wrote:On Mon, Jun 9, 2008 at 4:54 AM, gedgeck <[email protected]> wrote:
austin987 wrote:On Mon, Jun 9, 2008 at 3:45 AM, gedgeck <[email protected]> wrote:
It is strange to me but not. It doesn't work :\

That's why your regression testing isn't working. Something's fishy
with your compile.

It is unbelievably but if ./configure without --prefix (that is to say to /usr/local instead of /usr) than wine-0.9.51 work. What it may be?

austin987 · Post by **austin987** » Mon Jun 09, 2008 10:36 am

On Mon, Jun 9, 2008 at 6:48 AM, gedgeck <[email protected]> wrote:

austin987 wrote:On Mon, Jun 9, 2008 at 4:54 AM, gedgeck <[email protected]> wrote:

austin987 wrote: That's why your regression testing isn't working. Something's fishy
with your compile.
It is unbelievably but if ./configure without --prefix (that is to say to /usr/local instead of /usr) than wine-0.9.51 work. What it may be?

You may be running the unintended version. You don't have to do a make
install each time, just run it from the build directory.

gedgeck · Post by **gedgeck** » Mon Jun 09, 2008 2:25 pm

austin987 wrote: You may be running the unintended version. You don't have to do a make
install each time, just run it from the build directory.

Here is log of wine after run from build directory without actually install:

Code: Select all

fixme:win:SetLayeredWindowAttributes (0x10024,0x00000000,252,2): stub!
fixme:win:EnumDisplayDevicesW ((null),0,0x33f840,0x00000000), stub!
fixme:d3d:IWineD3DImpl_CheckDeviceMultiSampleType Quality levels unsupported at presen
t
fixme:d3d:IWineD3DDeviceImpl_CreateQuery (0x1576f8) Unhandled query type 4
fixme:d3d9:IDirect3DDevice9Impl_CreateQuery (0x151030) call to IWineD3DDevice_CreateQu
ery failed
fixme:xrandr:X11DRV_XRandR_SetCurrentMode Cannot change screen BPP from 16 to 32
fixme:d3d:IWineD3DDeviceImpl_CreateQuery (0x1576f8) Unhandled query type 4
fixme:d3d9:IDirect3DDevice9Impl_CreateQuery (0x1513f8) call to IWineD3DDevice_CreateQu
ery failed
fixme:wininet:InternetSetOptionExA Flags 04000000 ignored
fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_CONNECT_TIMEOUT (1000): STUB
fixme:wininet:InternetSetOptionExA Flags 04000000 ignored
fixme:wininet:InternetSetOptionW INTERNET_OPTION_SEND/RECEIVE_TIMEOUT not supported on
 protocol 1
No arguments found, but required
No arguments found, but required
sock_set_error: Победа
fixme:wininet:InternetSetOptionExA Flags 04000000 ignored
fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_CONNECT_TIMEOUT (1000): STUB
fixme:wininet:InternetSetOptionExA Flags 04000000 ignored
fixme:wininet:InternetSetOptionW INTERNET_OPTION_SEND/RECEIVE_TIMEOUT not supported on
 protocol 1
sock_set_error: Победа
fixme:wininet:InternetSetOptionExA Flags 04000000 ignored
fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_CONNECT_TIMEOUT (1000): STUB
fixme:wininet:InternetSetOptionExA Flags 04000000 ignored
fixme:wininet:InternetSetOptionW INTERNET_OPTION_SEND/RECEIVE_TIMEOUT not supported on
 protocol 1
sock_set_error: Победа
fixme:wininet:InternetSetOptionExA Flags 04000000 ignored
fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_CONNECT_TIMEOUT (1000): STUB
fixme:wininet:InternetSetOptionExA Flags 04000000 ignored
fixme:wininet:InternetSetOptionW INTERNET_OPTION_SEND/RECEIVE_TIMEOUT not supported on
 protocol 1
sock_set_error: Победа

At beginning there were many strings like first from this listing, but I cut them to minimize listing. Game still not work.

P.S. "Победа" is "Victory" in Russian.

austin987 · Post by **austin987** » Mon Jun 09, 2008 2:59 pm

On Mon, Jun 9, 2008 at 2:25 PM, gedgeck <[email protected]> wrote:

austin987 wrote:
You may be running the unintended version. You don't have to do a make
install each time, just run it from the build directory.
Here is log of wine after run from build directory without actually install:

Seems we're chasing invisible bugs here. Is there a demo/download
showing the problem? I'll run the regression test...

The regression test is only useful if you have problems in some self
built versions, but not others. If they're all broken or good on your
compile, then the problem is somewhere else.

gedgeck · Post by **gedgeck** » Mon Jun 09, 2008 3:27 pm

austin987 wrote: Seems we're chasing invisible bugs here. Is there a demo/download
showing the problem? I'll run the regression test...

The regression test is only useful if you have problems in some self
built versions, but not others. If they're all broken or good on your
compile, then the problem is somewhere else.

If you mean game demo than it may be downloaded with this link:
http://dodownload.filefront.com/4981162 ... d6e349d402
If you want something else please explain in detail, because I'm may be not understand 'demo/download showing the problem' phrase.

austin987 · Post by **austin987** » Mon Jun 09, 2008 3:32 pm

On Mon, Jun 9, 2008 at 3:27 PM, gedgeck <[email protected]> wrote:

If you mean game demo than it may be downloaded with this link:
http://dodownload.filefront.com/4981162 ... d6e349d402
If you want something else please explain in detail, because I'm may be not understand 'demo/download showing the problem' phrase.

Please test the demo to see if it has the same problem. If so, I'll
test on my desktop and if I can reproduce, I'll run the regression
test.

gedgeck · Post by **gedgeck** » Tue Jun 10, 2008 3:13 am

austin987 wrote:On Mon, Jun 9, 2008 at 3:27 PM, gedgeck <[email protected]> wrote:
If you mean game demo than it may be downloaded with this link:
http://dodownload.filefront.com/4981162 ... d6e349d402
If you want something else please explain in detail, because I'm may be not understand 'demo/download showing the problem' phrase.

Please test the demo to see if it has the same problem. If so, I'll
test on my desktop and if I can reproduce, I'll run the regression
test.

I have tested demo. It has that problem too.