Re: [ros-cvs] CVS Update: reactos

[Alex Ionescu]

Hartmut Birr wrote:
> CVSROOT:	/CVS/ReactOS
> Module name:	reactos
> Repository:	reactos/ntoskrnl/include/internal/i386/
> Changes by:	[e-mail]	04/10/17 03:53:08
>
> Modified files:
>    reactos/ntoskrnl/include/internal/i386/: ps.h
>
> Log message:
> Fixed the definition KPCR_SELF and KeGetCurrentKPCR.
>
> Ros-cvs mailing list
> [e-mail]
> http://reactos.com/mailman/listinfo/ros-cvs

Is that really right? I can swear fs:18h was still used until NT 5.2.
Also, it would be of critical importance to:
    1) Change the hardcoded offsets to FIELD_OFFSET(...,...)
    2) Make the ASM use the defines instead of hardcoding 1Ch/18h

Best regards,
Alex Ionescu

[Hartmut Birr]

> -----Original Message-----
> From: [e-mail] 
> [mailto:[e-mail]] On Behalf Of Alex Ionescu
> Sent: Sunday, October 17, 2004 2:51 PM
> To: [e-mail]; [e-mail]
> Subject: [ros-dev] Re: [ros-cvs] CVS Update: reactos
> > Modified files:
> >    reactos/ntoskrnl/include/internal/i386/: ps.h
> >
> > Log message:
> > Fixed the definition KPCR_SELF and KeGetCurrentKPCR.
>
> Is that really right? I can swear fs:18h was still used until NT 5.2.

I think that is correct. Previous your last changes the offset to KPCR->Self
was 0x18. After your changes the offset is 0x1c. The offset 0x18 does point
to KPCR->Tib.Self. Win2K and WinXP (UP version) doesn't use the value at
offset 0x18. This value is always 0.
> Also, it would be of critical importance to:
>     1) Change the hardcoded offsets to FIELD_OFFSET(...,...)

The assembler doesn't like the C definition for FIELD_OFFSET.
> 2) Make the ASM use the defines instead of hardcoding 1Ch/18h

The inline assembler doesn't like the definition for KPCR_SELF. The
assembler interprets KPCR_SELF as symbol at this point.

I've changed KPCR_SELF from 0x18 to 0x1c, because ros asserts in main.c line
#309.

- Hartmut

[Alex Ionescu]

Hartmut Birr wrote:
> > -----Original Message-----
> > From: [e-mail] 
> > [mailto:[e-mail]] On Behalf Of Alex Ionescu
> > Sent: Sunday, October 17, 2004 2:51 PM
> > To: [e-mail]; [e-mail]
> > Subject: [ros-dev] Re: [ros-cvs] CVS Update: reactos
> > > Modified files:
> > >    reactos/ntoskrnl/include/internal/i386/: ps.h
> > >
> > > Log message:
> > > Fixed the definition KPCR_SELF and KeGetCurrentKPCR.
> >
> > Is that really right? I can swear fs:18h was still used until NT 5.2.
>
> I think that is correct. Previous your last changes the offset to KPCR->Self
> was 0x18. After your changes the offset is 0x1c. The offset 0x18 does point
> to KPCR->Tib.Self. Win2K and WinXP (UP version) doesn't use the value at
> offset 0x18. This value is always 0.

You're right... I was led to believe this only happened in Win2K3. 
Thanks for letting me know.
> > Also, it would be of critical importance to:
> > Change the hardcoded offsets to FIELD_OFFSET(...,...)
>
> The assembler doesn't like the C definition for FIELD_OFFSET.

Is there any way to make it work? My MSVC inline assembler usually 
didn't mind C macros, I don't know what's the case for GNU Assembler files.
> > Make the ASM use the defines instead of hardcoding 1Ch/18h
>
> The inline assembler doesn't like the definition for KPCR_SELF. The
> assembler interprets KPCR_SELF as symbol at this point.
>
> I've changed KPCR_SELF from 0x18 to 0x1c, because ros asserts in main.c line
> #309.

Thanks, I had missed that one too. Once again shows the importance of 
being able to use FIELD_OFFSET.
> Hartmut
>
> Ros-dev mailing list
> [e-mail]
> http://reactos.com:8080/mailman/listinfo/ros-dev

Many thanks again!

Best regards,
Alex Ionescu