Re: [ros-svn] [ion] 17811: Major refactoring of the exception handling code + misc fixes:

[Hartmut Birr]

Hi,

calling RaiseException(EXCEPTION_ACCESS_VIOLATION,
EXCEPTION_NONCONTINUABLE, 0, NULL) from user mode does crash reactos.

- Hartmut

(lib\rtl\exception.c:29) RtlRaiseException(Status 0064fea8)
(lib\rtl\exception.c:36) ExceptionAddress 7c801fcc
(ntoskrnl\ke\exception.c:94) KiRaiseException
(ntoskrnl\ke\i386\exp.c:933) KiDispatchException() called
(lib\rtl\exception.c:75) RtlRaiseStatus(Status 0x80000002)
(lib\rtl\i386\exception.c:59) RtlDispatchException(): 87032998, 870326c8
(lib\rtl\i386\exception.c:64) RegistrationFrame is 0x87032D30
(lib\rtl\i386\exception.c:90) Invalid exception frame
(ntoskrnl\ke\exception.c:94) KiRaiseException
(ntoskrnl\ke\i386\exp.c:933) KiDispatchException() called
KeBugCheckWithTf at ntoskrnl\ke\i386\exp.c:993
A problem has been detected and ReactOS has been shut down to prevent
damage to your computer.

The problem seems to be caused by the following file: ntoskrnl.exe

Technical information:

*** STOP: 0x0000001E (0x80000002,0x80072fba,0x00000001,0x870329f0)

***    ntoskrnl.exe - Address 0x80072fba base at 0x80000000, DateStamp 0x0

Divide Error Exception: 0(0)
Processor: 0 CS:EIP 8:80072fba <ntoskrnl.exe:72fba
(ntoskrnl/mm/virtual.c:1002 (ProbeForRead))>
cr2 0 cr3 178f9000 Proc: 81ad12d0 Pid: 15c <t.EXE> Thrd: 81a691d8 Tid: 160
DS 10 ES 10 FS 30 GS 0
EAX: 870326c8   EBX: 81a691d8   ECX: 00000000
EDX: 000002f8   EBP: 81a691d8   ESI: 0064fbc4   ESP: 87032644
EDI: 87032d74   EFLAGS: 00000282 kESP 87032644 kernel stack base 87030000
Frames:
<0>

[Alex Ionescu]

Hartmut Birr wrote:

>  Hi,
>
> calling RaiseException(EXCEPTION_ACCESS_VIOLATION,
> EXCEPTION_NONCONTINUABLE, 0, NULL) from user mode does crash reactos.
>
> - Hartmut
>
> (lib\rtl\exception.c:29) RtlRaiseException(Status 0064fea8)
> (lib\rtl\exception.c:36) ExceptionAddress 7c801fcc
> (ntoskrnl\ke\exception.c:94) KiRaiseException
> (ntoskrnl\ke\i386\exp.c:933) KiDispatchException() called
> (lib\rtl\exception.c:75) RtlRaiseStatus(Status 0x80000002)
>   
>  
^^^ It seems that the user-mode context or exception frame is not 
properly aligned (or the Probe check is broken)... I will test and see 
which of the probe fails.

>  (lib\rtl\i386\exception.c:59) RtlDispatchException():
> 87032998, 870326c8
> (lib\rtl\i386\exception.c:64) RegistrationFrame is 0x87032D30
> (lib\rtl\i386\exception.c:90) Invalid exception frame
>   
>  
^^^ RtlpGetStackLimits should not be shared, because the limits have to 
be taken from the thread's teb, not KPCR. This is my fault, I'll fix it 
right now.

Best regards,
Alex Ionescu

[Hartmut Birr]

Hi,

there exist more problems. Some functions (for instance
KiDispatchException) leave a TRY block with a return statement. Quoting
http://www.reactos.com/wiki/index.php/PSEH: /"Absolutely do not use
*return* inside a try block/: this will lead to crashes, or worse
undetectable corruption, because some per-thread system state will still
refer to local variables of the function you've returned from."

- Hartmut

Alex Ionescu wrote:

>   Hartmut Birr wrote:
>
> >   Hi,
> >
> >  calling RaiseException(EXCEPTION_ACCESS_VIOLATION,
> >  EXCEPTION_NONCONTINUABLE, 0, NULL) from user mode does crash reactos.
> >
> >  - Hartmut
> >
> >  (lib\rtl\exception.c:29) RtlRaiseException(Status 0064fea8)
> >  (lib\rtl\exception.c:36) ExceptionAddress 7c801fcc
> >  (ntoskrnl\ke\exception.c:94) KiRaiseException
> >  (ntoskrnl\ke\i386\exp.c:933) KiDispatchException() called
> >  (lib\rtl\exception.c:75) RtlRaiseStatus(Status 0x80000002)
> >   
> >  
>  ^^^ It seems that the user-mode context or exception frame is not
>  properly aligned (or the Probe check is broken)... I will test and see
>  which of the probe fails.
>
> >   (lib\rtl\i386\exception.c:59)
> > RtlDispatchException(): 87032998, 870326c8
> >  (lib\rtl\i386\exception.c:64) RegistrationFrame is 0x87032D30
> >  (lib\rtl\i386\exception.c:90) Invalid exception frame
> >   
> >  
>  ^^^ RtlpGetStackLimits should not be shared, because the limits have
>  to be taken from the thread's teb, not KPCR. This is my fault, I'll
>  fix it right now.
>
>  Best regards,
>  Alex Ionescu
>  _______________________________________________
>  Ros-dev mailing list
>  Ros-dev(a)reactos.com
>  http://reactos.com:8080/mailman/listinfo/ros-dev
>