EventLog

[Saveliy Tretiakov]

Hi. I'm implementing an EventLog service now. I want to make it 100% 
compatible with windows. Windows machines will be able to access reactos 
eventlog via rpc.

Undocumented rpc interface is almost reversed now. You can find it in 
attached archive with some tests. Ansi functions work very well, but
I have some problems with unicode ones. When I pass initialized with 
nulls UNICODE_STRING to a function, it works. When I initialize 
structure with some other values, exception is raised on server side 
(1783 Stub received bad data). I don't know why this happens. Advapi32 
initializes structures with RtlInitUnicodeString, nothing special. Any 
ideas?

And I don't know how to compile this with widl and gcc.  SEH is not 
implemented in gcc, right? How rpc exceptions are handled in ROS? Widl 
returns strange errors. Somebody familiar with widl please help me =)

[Eric Kohl]

Hello Saveliy!
> Hi. I'm implementing an EventLog service now. I want to make it 100% 
> compatible with windows. Windows machines will be able to access reactos 
> eventlog via rpc.

That's great! Thanks for your contribution!
> Undocumented rpc interface is almost reversed now. You can find it in 
> attached archive with some tests. Ansi functions work very well, but
> I have some problems with unicode ones. When I pass initialized with 
> nulls UNICODE_STRING to a function, it works. When I initialize 
> structure with some other values, exception is raised on server side 
> (1783 Stub received bad data). I don't know why this happens. Advapi32 
> initializes structures with RtlInitUnicodeString, nothing special. Any 
> ideas?

No, not yet.
> And I don't know how to compile this with widl and gcc.  SEH is not 
> implemented in gcc, right? How rpc exceptions are handled in ROS? Widl 
> returns strange errors. Somebody familiar with widl please help me =)

WIDL is not ready to compile your attached eventlogrpc.idl because a lot 
of required features are not implemented yet. WIDL might crash, return 
strange error messages or generate buggy stub code.

WIDL generates SEH-protected stub code. When gcc compiles the stub code 
the SEH code gets disabled by rpc.h because gcc does not support SEH. 
But we are able to implement support for our ReactOS-specific SEH macros.

Implementing the required features will take several months because I 
have very very little time to work on ReactOS.

Regards
Eric Kohl

[Saveliy Tretiakov]

Ok, I decided to leave EventLog and do something else until required 
widl features will be implemented :)
Are there any undocumented rpc interfaces needed for compatibility with 
windows? If yes, I can try to reverse engeneer them.
I've just compared our svcctl.idl to IDA disassembly of advapi32. GUID 
and stub order are the same(someone already researched it?). Missing 
stubs are:
10 - I_ScSetServiceBitsW
11 - ChangeServiceConfigW
13 - EnumDependentServicesW
14 - EnumServiceGroupW
17 - QueryServiceConfigW
18 - QueryServiceLockStatusW
19 - StartServiceW
20 - GetServiceDisplayNameW
21 - GetServiceKeyNameW
22 - I_ScSetServiceBitsA
23 - ChangeServiceConfigA
24 - CreateServiceA
25 - EnumDependentServicesA
26 - EnumServicesStatusA
29 - QueryServiceConfigA
30 - QueryServiceLockStatusA
31 - StartServiceA
32 - GetServiceDisplayNameA
33 - GetServiceKeyNameA
34 - Unknown
35 - EnumServiceGroupW
36 - ChangeServiceConfig2A
37 - ChangeServiceConfig2W
38 - QueryServiceConfig2A
39 - QueryServiceConfig2W
40 - QueryServiceStatusEx
41 - EnumServicesStatusExA
42 - EnumServicesStatusExW
43 - I_ScSendTSMessage

I didn't looked at parameters yet.

Eric Kohl wrote:
> Hello Saveliy!
> > Hi. I'm implementing an EventLog service now. I want to make it 100% 
> > compatible with windows. Windows machines will be able to access 
> > reactos eventlog via rpc.
>
> That's great! Thanks for your contribution!
> > Undocumented rpc interface is almost reversed now. You can find it in 
> > attached archive with some tests. Ansi functions work very well, but
> > I have some problems with unicode ones. When I pass initialized with 
> > nulls UNICODE_STRING to a function, it works. When I initialize 
> > structure with some other values, exception is raised on server side 
> > (1783 Stub received bad data). I don't know why this happens. Advapi32 
> > initializes structures with RtlInitUnicodeString, nothing special. Any 
> > ideas?
>
> No, not yet.
> > And I don't know how to compile this with widl and gcc.  SEH is not 
> > implemented in gcc, right? How rpc exceptions are handled in ROS? Widl 
> > returns strange errors. Somebody familiar with widl please help me =)
>
> WIDL is not ready to compile your attached eventlogrpc.idl because a lot 
> of required features are not implemented yet. WIDL might crash, return 
> strange error messages or generate buggy stub code.
>
> WIDL generates SEH-protected stub code. When gcc compiles the stub code 
> the SEH code gets disabled by rpc.h because gcc does not support SEH. 
> But we are able to implement support for our ReactOS-specific SEH macros.
>
> Implementing the required features will take several months because I 
> have very very little time to work on ReactOS.
>
> Regards
> Eric Kohl
>
> Ros-dev mailing list
> [e-mail]
> http://reactos.com:8080/mailman/listinfo/ros-dev

[Rick Langschultz]

I think without a service host process I think implementing a service may be
a little to early. Sure you can test the file on Windows 2000 or XP but they
may have some sort of effect on the binary itself, but how will ReactOs
react to the binary? I am sure the service will not start.

-----Original Message-----
From: [e-mail] [mailto:[e-mail]] On
Behalf Of Saveliy Tretiakov
Sent: Sunday, July 31, 2005 8:42 AM
To: ReactOS Development List
Subject: Re: [ros-dev] EventLog

Ok, I decided to leave EventLog and do something else until required 
widl features will be implemented :)
Are there any undocumented rpc interfaces needed for compatibility with 
windows? If yes, I can try to reverse engeneer them.
I've just compared our svcctl.idl to IDA disassembly of advapi32. GUID 
and stub order are the same(someone already researched it?). Missing 
stubs are:
10 - I_ScSetServiceBitsW
11 - ChangeServiceConfigW
13 - EnumDependentServicesW
14 - EnumServiceGroupW
17 - QueryServiceConfigW
18 - QueryServiceLockStatusW
19 - StartServiceW
20 - GetServiceDisplayNameW
21 - GetServiceKeyNameW
22 - I_ScSetServiceBitsA
23 - ChangeServiceConfigA
24 - CreateServiceA
25 - EnumDependentServicesA
26 - EnumServicesStatusA
29 - QueryServiceConfigA
30 - QueryServiceLockStatusA
31 - StartServiceA
32 - GetServiceDisplayNameA
33 - GetServiceKeyNameA
34 - Unknown
35 - EnumServiceGroupW
36 - ChangeServiceConfig2A
37 - ChangeServiceConfig2W
38 - QueryServiceConfig2A
39 - QueryServiceConfig2W
40 - QueryServiceStatusEx
41 - EnumServicesStatusExA
42 - EnumServicesStatusExW
43 - I_ScSendTSMessage

I didn't looked at parameters yet.

Eric Kohl wrote:
> Hello Saveliy!
> > Hi. I'm implementing an EventLog service now. I want to make it 100% 
> > compatible with windows. Windows machines will be able to access 
> > reactos eventlog via rpc.
>
> That's great! Thanks for your contribution!
> > Undocumented rpc interface is almost reversed now. You can find it in 
> > attached archive with some tests. Ansi functions work very well, but
> > I have some problems with unicode ones. When I pass initialized with 
> > nulls UNICODE_STRING to a function, it works. When I initialize 
> > structure with some other values, exception is raised on server side 
> > (1783 Stub received bad data). I don't know why this happens. Advapi32 
> > initializes structures with RtlInitUnicodeString, nothing special. Any 
> > ideas?
>
> No, not yet.
> > And I don't know how to compile this with widl and gcc.  SEH is not 
> > implemented in gcc, right? How rpc exceptions are handled in ROS? Widl 
> > returns strange errors. Somebody familiar with widl please help me =)
>
> WIDL is not ready to compile your attached eventlogrpc.idl because a lot 
> of required features are not implemented yet. WIDL might crash, return 
> strange error messages or generate buggy stub code.
>
> WIDL generates SEH-protected stub code. When gcc compiles the stub code 
> the SEH code gets disabled by rpc.h because gcc does not support SEH. 
> But we are able to implement support for our ReactOS-specific SEH macros.
>
> Implementing the required features will take several months because I 
> have very very little time to work on ReactOS.
>
> Regards
> Eric Kohl
>
> Ros-dev mailing list
> [e-mail]
> http://reactos.com:8080/mailman/listinfo/ros-dev

_______________________________________________
Ros-dev mailing list
[e-mail]
http://reactos.com:8080/mailman/listinfo/ros-dev

[Alex Ionescu]

Saveliy Tretiakov wrote:
> Ok, I decided to leave EventLog and do something else until required 
> widl features will be implemented :)

Hi,

That might not be the best idea... a lot of parts of ReactOS/compiler 
system arne't ready yet, and that would mean sacrificing a lot of work. 
If you like to work on RPC/Services, then by all means, continue to do 
so, even if we won't support it. We always encourage developers to code 
on Windows, and let us handle it later, because:

1) It lets you test your app for compatibility
2) It lets us make sure and know exactly how we should respond, since 
you have the windows behaviour known.

So please continue to do work on any service you'd like...or you could 
try implementing MMC :)

Best regards,
Alex Ionescu

[Saveliy Tretiakov]

I know what I'll do :) I'll temporary modify the interface to make it 
compile with widl and revert it back when needed features will be 
implemented.

Alex Ionescu wrote:
> ...or you could try implementing MMC :)

No, I don't like COM :)