Re: [ros-dev] [ros-diffs] [greatlrd] 37343: Bugfix : Do not link to LPDDRAWI_DIRECTDRAW_INT->lpLink when we do not have any LPDDRAWI_DIRECTDRAW_INT->lpLcl

[gedmurphy]

Is it not time to start getting rid of IsBadWritePtr and friends?
I don't see a good reason for these in our own code, especially when we can just use SEH directly.

-----Original Message-----
From: [e-mail] [mailto:[e-mail]] On Behalf Of [e-mail]
Sent: 13 November 2008 19:52
To: [e-mail]
Subject: [ros-diffs] [greatlrd] 37343: Bugfix : Do not link to LPDDRAWI_DIRECTDRAW_INT->lpLink when we do not have any LPDDRAWI_DIRECTDRAW_INT->lpLcl

Author: greatlrd
Date: Thu Nov 13 13:52:20 2008
New Revision: 37343

URL: http://svn.reactos.org/svn/reactos?rev=37343&view=rev
Log:
Bugfix : Do not link to LPDDRAWI_DIRECTDRAW_INT->lpLink when we do not have any LPDDRAWI_DIRECTDRAW_INT->lpLcl

Modified:
    branches/reactx/reactos/dll/directx/ddraw/startup.c

Modified: branches/reactx/reactos/dll/directx/ddraw/startup.c
URL: http://svn.reactos.org/svn/reactos/branches/reactx/reactos/dll/directx/ddraw...
==============================================================================
--- branches/reactx/reactos/dll/directx/ddraw/startup.c [iso-8859-1] (original)
+++ branches/reactx/reactos/dll/directx/ddraw/startup.c [iso-8859-1] Thu Nov 13 13:52:20 2008
@@ -33,8 +33,8 @@

This = (LPDDRAWI_DIRECTDRAW_INT)*pIface;

-    /* fixme linking too second link when we shall not doing it */
-    if (IsBadReadPtr(This,sizeof(LPDIRECTDRAW)))
+    if ( (IsBadWritePtr(This,sizeof(LPDDRAWI_DIRECTDRAW_INT)) != 0) || 
+         (IsBadWritePtr(This->lpLcl,sizeof(LPDDRAWI_DIRECTDRAW_LCL)) != 0) )
     {
         /* We do not have a DirectDraw interface, we need alloc it*/
         LPDDRAWI_DIRECTDRAW_INT memThis;

[James Tabor]

Yes, I agree, we have SEH and we should just use that. Keep
IsBadXxxXxx for compatibility reasons only.

On Thu, Nov 13, 2008 at 2:59 PM, gedmurphy [e-mail] wrote:
> Is it not time to start getting rid of IsBadWritePtr and friends?
> I don't see a good reason for these in our own code, especially when we can just use SEH directly.

[Magnus Olsen]

1. if u warp code with PSEH it can not be optimze with the compiler (acroing 
KJK when he told me try avoid PSEH, long time ago).

2. This is old code I had lying around, the ddraw.dll was devloper frist in 
MSVC then port to GCC and reactos build system.

3. If u look how ms ddraw.dll  doing their call they using thuse api.

I can rewrite the code later to use PSEH, but I do not known when I have 
time todo that.
at momment I cleaing up my hd with old dx code. I hope within 1-2 month I 
commit eveything I have lying around for ddraw.dll. I have a girl friend 
now. so do not expect equal many commit as I use todo, or fast bugfix. Now I 
will only focuse on reactx when I got some time over.

----- Original Message ----- 
From: "James Tabor" [e-mail]
To: "ReactOS Development List" [e-mail]
Sent: Friday, November 14, 2008 10:18 AM
Subject: Re: [ros-dev] [ros-diffs] [greatlrd] 37343: Bugfix : Do not link 
toLPDDRAWI_DIRECTDRAW_INT->lpLink when we do not have 
anyLPDDRAWI_DIRECTDRAW_INT->lpLcl
> Yes, I agree, we have SEH and we should just use that. Keep
> IsBadXxxXxx for compatibility reasons only.
>
> On Thu, Nov 13, 2008 at 2:59 PM, gedmurphy [e-mail] wrote:
> > Is it not time to start getting rid of IsBadWritePtr and friends?
> > I don't see a good reason for these in our own code, especially when we 
> > can just use SEH directly.
>
> Ros-dev mailing list
> [e-mail]
> http://www.reactos.org/mailman/listinfo/ros-dev

[Gregor Schneider]

GBrunmar posted a link to a nice article about IsBadXXXPointer some time
ago.
Check it out, it's worth reading:
"IsBadXxxPtr should really be called CrashProgramRandomly"
http://blogs.msdn.com/oldnewthing/archive/2006/09/27/773741.aspx

Best regards,
Gregor Schneider

[KJK::Hyperion]

Magnus Olsen wrote:
> if u warp code with PSEH it can not be optimze with the compiler (acroing
>
> KJK when he told me try avoid PSEH, long time ago).

PSEH 2.0 doesn't seem to have this issue (which makes it a little risky
to use, maybe?)
> I can rewrite the code later to use PSEH, but I do not known when I have 
> time todo that.

just remove the call to IsBadWritePtr. Don't replace it with anything

[Magnus Olsen]

Only for DirectX api big issue
In allot case in dxg.sys or win32k and the graphice drv.
They does not vaildate if it vaild pointer or not been pass down
it is up to the usermode part vaildate them inside ddraw.dll
if I pass down wrong value whole windows xp will crash to
black screeen and some part being powerdown. then the comptuer
complete froze with nice black screen. How can we vaildate the
pointer right in usermode without using IsBadWritePtr  and friends
I do not known who to solv this issue without using  IsBadWritePtr
the PSEH or _SEH can not pickup this crash at usermode when it is kmode
in the drv or dxg.sys that blows. I have try it today,

The graphic manufactor are aware of this issue but they does not check for 
all case.

I will not leave detail in public how to make whole xp goes complete dead to 
nice
black screen and some part in the comptuer have been powerdown.  It is
imposble for me see exacly what crash. I already suffer 10 of this type of 
crash
today.

Yes some parth maybe can be change to SEH
the whole HEL layer in ddraw will be use SEH
the HAL part in ddraw will not, untill I known a good
idea how this issue can be solv on XP.

This Test was done on XP SP3.

----- Original Message ----- 
From: "KJK::Hyperion" [e-mail]
To: "ReactOS Development List" [e-mail]
Sent: Sunday, November 16, 2008 7:32 PM
Subject: Re: [ros-dev] [ros-diffs] [greatlrd] 37343: Bugfix : Do not link 
toLPDDRAWI_DIRECTDRAW_INT->lpLink when we do not have 
anyLPDDRAWI_DIRECTDRAW_INT->lpLcl
> Magnus Olsen wrote:
> > if u warp code with PSEH it can not be optimze with the compiler
> >
> > (acroing
> > KJK when he told me try avoid PSEH, long time ago).
>
> PSEH 2.0 doesn't seem to have this issue (which makes it a little risky
> to use, maybe?)
> > I can rewrite the code later to use PSEH, but I do not known when I have
> > time todo that.
>
> just remove the call to IsBadWritePtr. Don't replace it with anything
> _______________________________________________
> Ros-dev mailing list
> [e-mail]
> http://www.reactos.org/mailman/listinfo/ros-dev