SDL causing BSODs on Vista

I’m using Code::Blocks with the bundled compiler (mingw), and everytime I try running a game I made using SDL in Vista it just causes a BSOD.
I don’t know if it has anything to do with it, but I compiled the game on Windows 7, but running the game on Windows XP doesn’t crash the computer.

Are the same drivers on both systems?On Sun, Sep 20, 2009 at 2:54 PM, MikeDomingues wrote:

I’m using Code::Blocks with the bundled compiler (mingw), and everytime I
try running a game I made using SDL in Vista it just causes a BSOD.
I don’t know if it has anything to do with it, but I compiled the game on
Windows 7, but running the game on Windows XP doesn’t crash the computer.


SDL mailing list
SDL at lists.libsdl.org
http://lists.libsdl.org/listinfo.cgi/sdl-libsdl.org

Of course they aren’t. Vista uses a completely different driver model than XP does, especially for video drivers.________________________________

From: pjmlp@progtools.org (Paulo Pinto)
Subject: Re: [SDL] SDL causing BSODs on Vista

Are the same drivers on both systems?

On Sun, Sep 20, 2009 at 2:54 PM, MikeDomingues wrote:

I’m using Code::Blocks with the bundled compiler (mingw), and everytime I try running a game I made using SDL in Vista it just causes a BSOD.

I don’t know if it has anything to do with it, but I compiled the game on Windows 7, but running the game on Windows XP doesn’t crash the computer.


SDL mailing list
SDL at lists.libsdl.org
http://lists.libsdl.org/listinfo.cgi/sdl-libsdl.org

Well, speaking for NVida drivers, even if the underlying code is a bit
different, the versions tend to be the same across platforms.

Anyway the first step to solve such problems is always make sure if one is
using the latest driver versions.–
Paulo

On Sun, Sep 20, 2009 at 10:25 PM, Mason Wheeler wrote:

Of course they aren’t. Vista uses a completely different driver model than
XP does, especially for video drivers.


From: Paulo Pinto <@Paulo_Pinto>
**Subject: Re: [SDL] SDL causing BSODs on Vista

Are the same drivers on both systems?

On Sun, Sep 20, 2009 at 2:54 PM, MikeDomingues wrote:

I’m using Code::Blocks with the bundled compiler (mingw), and everytime
I try running a game I made using SDL in Vista it just causes a BSOD.
I don’t know if it has anything to do with it, but I compiled the game on
Windows 7, but running the game on Windows XP doesn’t crash the computer.


SDL mailing list
SDL at lists.libsdl.org
http://lists.libsdl.org/listinfo.cgi/sdl-libsdl.org


SDL mailing list
SDL at lists.libsdl.org
http://lists.libsdl.org/listinfo.cgi/sdl-libsdl.org

Thanks for the replies, however I thought SDL wasn’t hardware accelerated and therefore didn’t need a graphic card.

This is a serious problem with mingw (maybe triggered by SDL?) that
nobody is investigating. I already posted in this list and at c::b and
mingw about the problems that I had using this combo C::B + mingw+ SDL

  • no solution yet. Anyway, I tried to debug it using WinDBG but the
    problem must be related to WinSxS and/or CRT (or OpenGL loading).
    Do a test, try to compile the same project using Vistual Studio 2008
    Express and then, run it on that machine, after that, run your mingw
    version, you will see it will run fine.On Tue, Sep 22, 2009 at 1:47 PM, MikeDomingues wrote:

Thanks for the replies, however I thought SDL wasn’t hardware accelerated
and therefore didn’t need a graphic card.


SDL mailing list
SDL at lists.libsdl.org
http://lists.libsdl.org/listinfo.cgi/sdl-libsdl.org


Animal Liberation Front
http://www.animal-liberation.com/

Hello !

This is a serious problem with mingw (maybe triggered by SDL?) that
nobody is investigating. I already posted in this list and at c::b and
mingw about the problems that I had using this combo C::B + mingw+ SDL

  • no solution yet. Anyway, I tried to debug it using WinDBG but the
    problem must be related to WinSxS and/or CRT (or OpenGL loading).
    Do a test, try to compile the same project using Vistual Studio 2008
    Express and then, run it on that machine, after that, run your mingw
    version, you will see it will run fine.

Do you have a minimal example that shows the problem ?
I am coding with CYGWIN ( MinGW Mode ) + SDL on Vista
and don’t see any crashes.

What about compiling the programm with -g and using GDB to run it ?

CU

I haven’t any here right now. But you can follow my previous mail
here, there (I hope) you can find some tests.

http://lists.libsdl.org/pipermail/sdl-libsdl.org/2009-July/071312.htmlOn Wed, Sep 23, 2009 at 1:34 PM, Torsten Giebl wrote:

Hello !

This is a serious problem with mingw (maybe triggered by SDL?) that
nobody is investigating. I already posted in this list and at c::b and
mingw about the problems that I had using this combo C::B + mingw+ SDL

  • no solution yet. Anyway, I tried to debug it using WinDBG but the
    problem must be related to WinSxS and/or CRT (or OpenGL loading).
    Do a test, try to compile the same project using Vistual Studio 2008
    Express and then, run it on that machine, after that, run your mingw
    version, you will see it will run fine.

Do you have a minimal example that shows the problem ?
I am coding with CYGWIN ( MinGW Mode ) + SDL on Vista
and don’t see any crashes.

What about compiling the programm with -g and using GDB to run it ?

CU


SDL mailing list
SDL at lists.libsdl.org
http://lists.libsdl.org/listinfo.cgi/sdl-libsdl.org


Animal Liberation Front
http://www.animal-liberation.com/

I haven’t any here right now. But you can follow my previous mail
here, there (I hope) you can find some tests.

http://lists.libsdl.org/pipermail/sdl-libsdl.org/2009-July/071312.html

Hello !

This is a serious problem with mingw (maybe triggered by SDL?) that
nobody is investigating. I already posted in this list and at c::b and
mingw about the problems that I had using this combo C::B + mingw+ SDL

  • no solution yet. Anyway, I tried to debug it using WinDBG but the
    problem must be related to WinSxS and/or CRT (or OpenGL loading).
    Do a test, try to compile the same project using Vistual Studio 2008
    Express and then, run it on that machine, after that, run your mingw
    version, you will see it will run fine.

This is clearly an OS bug. Seriously, if Windows crashes because of
something wrong in a compiler, well … Really, if the compiler can
crash the OS, then the OS is seriously broken. So, take it over to
Microsoft and complain to them. They are the only ones who should be
working to fix the problem.

Bob PendletonOn Wed, Sep 23, 2009 at 12:00 PM, fungos wrote:

On Wed, Sep 23, 2009 at 1:34 PM, Torsten Giebl wrote:

Do you have a minimal example that shows the problem ?
I am coding with CYGWIN ( MinGW Mode ) + SDL on Vista
and don’t see any crashes.

What about compiling the programm with -g and using GDB to run it ?

CU


SDL mailing list
SDL at lists.libsdl.org
http://lists.libsdl.org/listinfo.cgi/sdl-libsdl.org


Animal Liberation Front
http://www.animal-liberation.com/


SDL mailing list
SDL at lists.libsdl.org
http://lists.libsdl.org/listinfo.cgi/sdl-libsdl.org


±----------------------------------------------------------

Tautology, Bob. Windows is an OS bug. That’s why I don’t use it. :slight_smile:

JeffOn Wednesday 23 September 2009 20:18, Bob Pendleton wrote:

This is clearly an OS bug. Seriously, if Windows crashes because of
something wrong in a compiler, well … Really, if the compiler can
crash the OS, then the OS is seriously broken. So, take it over to
Microsoft and complain to them. They are the only ones who should be
working to fix the problem.

Bob Pendleton

This is clearly an OS bug. Seriously, if Windows crashes because of
something wrong in a compiler, well … Really, if the compiler can
crash the OS, then the OS is seriously broken. So, take it over to
Microsoft and complain to them. They are the only ones who should be
working to fix the problem.

That’s a rather ridiculous statement. The compiler isn’t crashing the OS,
the code produced by the compiler is. A compiler is a program that
produces executable code, nothing more, nothing less. Your statement,
therefore, is logically equivalent to "any OS on which it’s possible to
produce executable code that can crash the OS is seriously broken."
Now, doesn’t Linux have a “kernel panic” feature that’s functionally
equivalent to Windows’s BSOD? Doesn’t every modern OS have its
own equivalent?>----- Original Message ----

From: Bob Pendleton
Subject: Re: [SDL] SDL causing BSODs on Vista

This is clearly an OS bug. Seriously, if Windows crashes because of
something wrong in a compiler, well … Really, if the compiler can
crash the OS, then the OS is seriously broken. So, take it over to
Microsoft and complain to them. They are the only ones who should be
working to fix the problem.

That’s a rather ridiculous statement. The compiler isn’t crashing
the OS,
the code produced by the compiler is. A compiler is a program that
produces executable code, nothing more, nothing less. Your statement,
therefore, is logically equivalent to "any OS on which it’s possible
to
produce executable code that can crash the OS is seriously broken."
Now, doesn’t Linux have a “kernel panic” feature that’s functionally
equivalent to Windows’s BSOD? Doesn’t every modern OS have its
own equivalent?

A kernel panic can only be triggered by a bug in kernel mode code. I
would consider any full system crash caused by a user land program,
on any OS (be it Windows, Linux, Mac OS X, *BSD, …) a bug in the OS
– well, or in one of the drivers. Typically, the video card drivers
tend to have buggy crashy versions that can tear down the whole system.

The fact that it doesn’t crash on one machine but not on the other
further supports this. Really, any bug in SDL or in an SDL using app
shouldn’t crash Windows (or any of the others). Whenever a BSOD is
involved, the first thought should be “faulty driver or faulty
hardware”.

Bye,
MaxAm 24.09.2009 um 05:56 schrieb Mason Wheeler:

----- Original Message ----
From: Bob Pendleton
Subject: Re: [SDL] SDL causing BSODs on Vista

I know it is just another OS bug, a very hidden one by the way,
because it is really intrinsic within the way the SO loads a (managed)
PE and it’s dependency. This is clearly not SDL fault. But I can’t
guarantee that it is not fault (in part - just the trigger part) of
cigwin. It is not building a “correct” binary to be loaded in Vista in
certain cases (when related to later loading of opengl.dll as SDL
does?).

My only problem here is Marketing and/or Politics. I’m trying REALLY
HARD to introduce foss in the company I’m working at. We develop
casual games (www.techfront.com.br) and sadly, our clients only want
games for Windows and sometimes a MacOS or iPhone port. What I’m
trying to do is to get foss accepted by some developers here and to
try to create a Linux port for any game developed. But this little bug
caused too much trouble for me. Everything that goes wrong is because
foss doesn’t work, or because foss is for amateurs or either, because
foss is just totally flawed. No matter what, for other developers and
for our bosses and clients the Vista is not bugged and the fault is
entirely of the open source tools. So, I really don’t care if this is
a windows bug, and neither you should because at the end, “Microsoft
is not amateur to create buggy software”. Sorry, but we must do things
works in a buggy OS no matter what when that buggy OS is the
dominating one.

Anyway, we are on schedule and I will do everything possible to create
a linux port of our current game, even if I need add it hidden in the
windows version. But sadly any other developer here is using Visual
Studio 2008 Express and they are almost begging to buy Visual Studio
2008 Professional because it just works on Vista and any other windows
version.

ps: I developed (most at home) a little framework for games, like
Popcap, Playground, HGE, etc. that runs in Windows, MacOS, Linux,
iPhone, Xbla, Zune and Nintendo Wii (soon porting to PSP/PS3/PSN) and
this will enable me to build a instantly linux port of any game
developed here (but when other devs uses Visual Studio, always has
something to fix!).

ps2.: this is a very interesting opportunity for these Security
Experts out there to find a interesting exploit on Vista ;)On Thu, Sep 24, 2009 at 6:52 AM, Max Horn wrote:

Am 24.09.2009 um 05:56 schrieb Mason Wheeler:

----- Original Message ----

From: Bob Pendleton
Subject: Re: [SDL] SDL causing BSODs on Vista

This is clearly an OS bug. Seriously, if Windows crashes because of
something wrong in a compiler, well … ?Really, if the compiler can
crash the OS, then the OS is seriously broken. So, take it over to
Microsoft and complain to them. They are the only ones who should be
working to fix the problem.

That’s a rather ridiculous statement. ?The compiler isn’t crashing the OS,
the code produced by the compiler is. ?A compiler is a program that
produces executable code, nothing more, nothing less. ?Your statement,
therefore, is logically equivalent to "any OS on which it’s possible to
produce executable code that can crash the OS is seriously broken."
Now, doesn’t Linux have a “kernel panic” feature that’s functionally
equivalent to Windows’s BSOD? ?Doesn’t every modern OS have its
own equivalent?

A kernel panic can only be triggered by a bug in kernel mode code. I would
consider any full system crash caused by a user land program, on any OS
(be it Windows, Linux, Mac OS X, *BSD, …) a bug in the OS – well, or in
one of the drivers. Typically, the video card drivers tend to have buggy
crashy versions that can tear down the whole system.

The fact that it doesn’t crash on one machine but not on the other further
supports this. Really, any bug in SDL or in an SDL using app shouldn’t crash
Windows (or any of the others). Whenever a BSOD is involved, the first
thought should be “faulty driver or faulty hardware”.

Bye,
Max


SDL mailing list
SDL at lists.libsdl.org
http://lists.libsdl.org/listinfo.cgi/sdl-libsdl.org


Animal Liberation Front
http://www.animal-liberation.com/

That sounds like a perfectly reasonable statement to me. At least when
talking about non-privileged usermode code, which is what we’re
talking about here.On Wed, Sep 23, 2009 at 23:56, Mason Wheeler wrote:

That’s a rather ridiculous statement. ?The compiler isn’t crashing the OS,
the code produced by the compiler is. ?A compiler is a program that
produces executable code, nothing more, nothing less. ?Your statement,
therefore, is logically equivalent to “any OS on which it’s possible to
produce executable code that can crash the OS is seriously broken.”

That’s a rather ridiculous statement. The compiler isn’t crashing the OS,
the code produced by the compiler is. A compiler is a program that
produces executable code, nothing more, nothing less. Your statement,
therefore, is logically equivalent to “any OS on which it’s possible to
produce executable code that can crash the OS is seriously broken.”

That sounds like a perfectly reasonable statement to me. At least when
talking about non-privileged usermode code, which is what we’re
talking about here.

Not completely. We’re talking about a multimedia library that speaks to
graphics drivers, which aren’t “non-privileged usermode code” and are
notoriously touchy.

I hate Windows as much as anyone, but you’ve gotta be objective about
these sorts of statements.>----- Original Message ----

From: Simon Roby <simon.roby at gmail.com>
Subject: Re: [SDL] SDL causing BSODs on Vista
On Wed, Sep 23, 2009 at 23:56, Mason Wheeler <@Mason_Wheeler> wrote:

That’s a rather ridiculous statement. ?The compiler isn’t crashing the OS,
the code produced by the compiler is. ?A compiler is a program that
produces executable code, nothing more, nothing less. ?Your statement,
therefore, is logically equivalent to “any OS on which it’s possible to
produce executable code that can crash the OS is seriously broken.”

That sounds like a perfectly reasonable statement to me. At least when
talking about non-privileged usermode code, which is what we’re
talking about here.

Not completely. ?We’re talking about a multimedia library that speaks to
graphics drivers, which aren’t “non-privileged usermode code” and are
notoriously touchy.

I hate Windows as much as anyone, but you’ve gotta be objective about
these sorts of statements.

But that’s still not the compiler’s code itself that’s at fault.On Thu, Sep 24, 2009 at 14:15, Mason Wheeler wrote:

----- Original Message ----
From: Simon Roby <@Simon_Roby>
Subject: Re: [SDL] SDL causing BSODs on Vista
On Wed, Sep 23, 2009 at 23:56, Mason Wheeler wrote:

Simon Roby wrote:

But that’s still not the compiler’s code itself that’s at fault.

If an incorrect executable causes a system crash, the OS / video driver
is at fault for allowing the crash, but the compiler is still at fault
for generating the incorrect executable in the first place.–
Rainer Deyke - rainerd at eldwood.com

Definition: crashme: Stress tests operating system stability crashme generates
strings of random bytes and then attempts to execute them. Used to test
kernel stability.

I don’t recall who it was, but someone required his Linux distribution to
survive crash me tests for 24 hours without crashing the system before he
would release a new version.

So while technically you can complain about the compiler, the fault lies in
the OS. If you try hard enough, you can crash Linux, but crashing Windows is
obnoxiously easy.

JeffOn Friday 25 September 2009 14:59, Rainer Deyke wrote:

Simon Roby wrote:

But that’s still not the compiler’s code itself that’s at fault.

If an incorrect executable causes a system crash, the OS / video driver
is at fault for allowing the crash, but the compiler is still at fault
for generating the incorrect executable in the first place.

From http://linux.about.com

Jeff Post wrote:

So while technically you can complain about the compiler, the fault lies in
the OS. If you try hard enough, you can crash Linux, but crashing Windows is
obnoxiously easy.

The OS may be at fail for not properly containing the crash, but no OS
in the world can cause an incorrect executable to execute correctly.
The compiler is still at fault for generating an incorrect executable.–
Rainer Deyke - rainerd at eldwood.com

Um… folks, can we stop this now? My original point was that the
problem was not with SDL and so the discussion should not be on the
SDL mail list.

Jeff Post wrote:

So while technically you can complain about the compiler, the fault lies in
the OS. If you try hard enough, you can crash Linux, but crashing Windows is
obnoxiously easy.

The OS may be at fail for not properly containing the crash, but no OS
in the world can cause an incorrect executable to execute correctly.
The compiler is still at fault for generating an incorrect executable.

Clearly, if a compiler generates incorrect code it has a bug. Just as
clearly, if the OS crashes because of the compiler bug the OS has a
bug.

Bob PendletonOn Fri, Sep 25, 2009 at 6:27 PM, Rainer Deyke wrote:


Rainer Deyke - rainerd at eldwood.com


SDL mailing list
SDL at lists.libsdl.org
http://lists.libsdl.org/listinfo.cgi/sdl-libsdl.org


±----------------------------------------------------------